You are on page 1of 63

SPE-177277-MS

Microseismic 101: Monitoring and Evaluating Hydraulic Fracturing to


Improve the Efficiency of Oil and Gas Recovery from Unconventional
Reservoirs
Adam Yousefzadeh, Qi Li, and Roberto Aguilera, Schulich School of Engineering, University of Calgary

Copyright 2015, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Latin American and Caribbean Petroleum Engineering Conference held in Quito, Ecuador, 18 –20 November
2015.

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

Abstract
Hydraulic fracturing will be main drive force in oil and gas production from tight hydrocarbon reservoirs
in the future. We present a comprehensive literature review on applications, advancement, and limitations
of geomechanics and microseismic monitoring of hydraulic fracturing treatments for exploitation of
unconventional reservoirs. Microseismic method is the sate-of-the-art technology to monitor the hydraulic
fracturing process. Visualization of microseismic events determines the spatial extent of hydraulic
fracturing at first glance. Advance analysis of microseismic measurements provides more detailed
information about the fracturing mode and fractures geometry. It can even be used for determination of
the state of stress in the reservoir and for helping in reservoir characterization at a large scale.
Microseismic imaging has successfully been used on stimulation design and control, reservoir character-
ization and simulation of unconventional reservoirs particularly in North America. Geomechanical,
petrophysical, and geophysical mechanisms of hydraulic fracturing and associated seismicity which are
not fully understood are topics of ongoing research.
The goal of this study is to provide a comprehensive guideline and a study reference for geoscientists
and engineers who would like to get familiar with the theoretical and practical aspects of microseismic
monitoring. Starting with a brief history of exploitation of tight reservoirs and microseismic monitoring,
the mechanism of hydraulic fracturing and microseismicity are described. Applications, processing,
interpretation, and limitations of the methods are explained as wells as how microseismic technology has
responded to some of the public concerns about environmental aspects and safety of the hydraulic
fracturing process. In closing, four case studies are reviewed to provide some insights into the practical
application and limitations of microseismic monitoring.

Introduction
Microseismic methods were initially developed based on the known earthquake seismology theories and
methods. History of analyzing the man-made weak seismic signals returns back to the monitoring of
mining induced seismicity at the beginning of 19th century (Maxwell, 2014). Using microseismic
monitoring to evaluate fluid injection related seismicity was probably begun by monitoring of enhanced
2 SPE-177277-MS

geothermal systems in the 80s. Geothermal exploitation includes circulation of water through usually
more than 5km depth and hot (above 200°C) rocks to transfer heat as a source of energy to the surface
(Dyer et al., 2008). Stimulation of natural fractures increased the efficiency of water flow and heat
transfer. Microseismic monitoring successfully enabled the evaluation and monitoring of fracture growth
due to fluid injection into the hot formations.
Tight oil and tight gas reservoirs became economically attractive mostly after development of
technology for drilling long horizontal wells and the introduction of multistage ⬙hydraulic fracturing⬙ (also
known as fracturing, fracking, or fracing) methods. Hydraulic fracturing was introduced to the oil and gas
industry in the 1930s when Dow Chemical Company successfully fracked a reservoir by injection of high
pressure fluid into a borehole during an acid stimulation program (Grebe and Stoesser, 1935). Although
the hydraulic fracturing in shale reservoirs started previous to 1965, and by 1980 there were thousands of
hydraulically fractured vertical shale wells producing natural gas commercially from the Appalachian
basin in eastern United States (Gatens et al, 1989), its successful employment in the Barnet Shale in 1997
expedites the growth of the fracking methods in shale plays of the USA. The shale gas industry in the
United Sates experienced 45% grow from 2005 to 2010 (The Economist, 2012). Shale gas production was
2.12 trillion cubic feet only in 2008 which was 71% more than the production in 2007 in the USA. The
shale gas production increased to 11.42 trillion cubic feet in 2013 which represents 540% growth since
2008 (EIA, 2015). By 2013 more than 1.1 million fracturing jobs have been completed in the United State
and about 90% of the onshore wells became economically feasible using hydraulic fracturing stimulation
(Halliburton, 2015). Fig. 1 illustrates the main shale plays including Barnett Shale in the North America
by 2011.
SPE-177277-MS 3

Figure 1—North American shale plays as May 2011 (EIA, 2011)

Hydraulic fracturing is currently the most viable and economical method of exploitation and comple-
tion of reservoirs with very low permeability which is leading the United State to become independent
from foreign oil and gas. Without hydraulic fracturing method, the United States would have 30% less oil
and gas resources and would produce 600 trillion cubic feet (tcf) of natural gas and 7 billion barrel of
crude oil less than the amount that was produced by 2010 (Fisher, 2010).
Microseismic as a tool to monitor hydraulic fracturing stimulation in the oil and gas industry did not
begin until the late 90s. Early microseismic data were recorded using borehole instruments only. First
surface microseismic data was gathered in 2003 and the first buried array of receivers were used to record
microseismic data in 2008 (Eisner, 2014).
The commercial use of microseismic methods has grown rapidly during the last two decades. This
development is mostly due to increasing demand from companies exploiting and producing tight/
unconventional hydrocarbon resources including oil sandstone, tight sandstone, and shale reservoirs.
Microseismic is the best practice for evaluation and monitoring of the hydraulic fracturing. Observation
of fracturing processes is the most common application of the newly developed microseismic methods
(Maxwell, 2010). About 10% of fracturing stages were monitored using microseismic methods by 2010
(Duncan, 2010). The number of hydraulic fracturing operations which are mapped under some type of
microseismic monitoring is growing by the new advances in microseismic data acquisition, analyses, and
interpretation technologies, as well as the reduction in operational expenses.
4 SPE-177277-MS

With the economical gas exploitation from shale gas reservoirs at the Barnett Shale in Fort Worth Basin
of Texas, microseismic monitoring technology started a new era of development (Warpinski et al., 1998;
King, 2010), particularly with the introduction of stimulated reservoir volume (SRV) calculation and
concept (Warpinski et al., 2005).
Real time mapping of hydraulic fractures and fluid flow has been successfully used for more efficient
well placement, stimulation strategies, and to prevent the potential environmental hazards (Mueller,
2013). Microseismic methods may also be used for monitoring of CO2 storage, waste water injection, and
injection for enhanced oil recovery methods.
In addition to many new service companies that are specialized in the microseismic data acquisition
design, operation, processing, software development, and interpretation, many pioneer oil service com-
panies are currently offering services related to microseismic monitoring to the oil companies around the
world.
One may add to these efforts, the growing number of research groups and individuals at universities
around the world performing research focused on topics related to geomechanics and microseismic
monitoring of hydraulic fracturing.
Other methods of monitoring or evaluation of hydraulic fracturing and related events include surface
and downhole pressure monitoring, measuring temperature, pressure, and sound using fiber optics,
tiltmeters, tracer tagger proppant method, using chemical tracers in fracking fluid, temperature and
production logging, pressure buildup evaluation and rate transient analysis (King, 2012). Fiber optic
measurements, temperature and production logging, and tracer proppant show the fracturing location at
the wellbore.
However, among all these methods, microseismic monitoring and to some degree tiltmeters are only
methods which are able to show the fracture direction, height, and length in an almost real time (King,
2012). In the following sections we review concepts and applications of microseismic monitoring and
geomechanics of hydraulic fracturing, uncertainties and limitations associated with the method, and the
role of microseismic to protect the environment from potential fracturing hazards.
Hydraulic fracturing
Commercial hydraulic fracturing was introduced 66 years ago by Halliburton in Stephens County,
Oklahoma. Since then more than 1 million jobs have been conducted in it the United States and more than
1 million jobs in the rest of the world.
Until recently hydraulic fracturing had been used without any controversy in all types of reservoirs.
During the last few years it has been used, with some controversy, as an effective method to enhance well
productivity in hydrocarbon reservoirs with low and ultra-low permeability.
Fracking fluid, usually water based gels, is injected into the desired formation at a designed rate and
very high pressure for a period of time to break the formation rocks. Breaking starts when the injection
pressure overcomes the formation rocks breakdown pressure. Injecting and fracturing may continue from
a few hours up to a few days.
Numerical modeling and laboratory experiments show that the fluid injection into homogenous solids
generally creates a planar fracture which is perpendicular to the direction of minimum stress. In reservoir
rocks the behavior is more complex due to geomechanical and petrophysical heterogeneity, and presence
of natural fractures, bedding, and flaws (Warpinski, 2014; Ma et al., 2015).
However, as shown by microseismic measurements, hydraulic fractures in most sandstone and
limestone reservoirs will have a general planar shape if the there is a significant difference between the
local minimum and maximum in-situ stresses (Warpinski, 2014).
Fracturing continues until the fluid’s leakoff into the formation, pre-existing fractures, or faults drops
the pressure at the fracture tip down to the pressure below the breaking pressure. Fracture growth may also
terminate when the fracture contacts another formation or layer with different lithology or strength. After
SPE-177277-MS 5

fracture growth ends, the injection terminates and ideally, we end up with a network of permeable
fractures in the reservoir.
Fractures are generally about 2 to 3 mm wide and may continue a few hundred feet away from the
wellbore (King, 2012). Fractures have the tendency to close due to in-situ stresses after removing the
injection pressure. Adding properly selected sand or hard ceramic proppant to the fracking fluid during
injection, forces the cracks to remain open for longer time after injection terminates.
It should also be noted that hydraulic fracturing increases the connection between the wellbore and the
reservoir without significant change in the formation permeability (Cipolla and Wallace, 2014). The
improvement in connectivity of a reservoir to the injection well by tensile fracture seems obvious.
However, permeable shear fractures might need a proper justification. Fracture surfaces are not generally
smooth planes due to small scale heterogeneities in rocks as shown by Ma et al. (2015). The relatively
coarse topography on the fracture surface causes the shear slippages on the surface to create small
openings for fluid to flow (Fredd et al., 2001; Maxwell and Cipolla, 2011).
However, in order to explain (1) the large volume of fluid injected to tight reservoirs in a short time,
(2) the small portion of the energy released in the form of shear wave microseismic data in comparison
to total induced energy, and (3) the evidence of fluid flow due to fracture closure after injection, it is
necessary to consider opening to be the dominant mechanism for the volume created during a hydraulic
fracturing job (Maxwell and Cipolla, 2011).
Refracturing, repeating the hydraulic fracturing treatment in the fractured formation using same
wellbore, might be performed when the hydrocarbon production declines, usually a few months after
initial hydraulic fracturing. Refracturing is a cost-effective method in comparison to drilling new wells.
Refracturing proved to be very effective in increasing production especially in those wells where no
proppant had been injected during initial fracturing. Refracturing was able to double the production in
some wells (King, 2012). Refracturing may produce a new network of fractures and connectivity, and may
or may not follow the same direction and trend of the initial fractures. This could be due to change in stress
state, permeability, and fluid flow regime during or after the first fracking.
The best practical design for hydraulic fracturing, especially in horizontal wells, is multistage fractur-
ing. Multistage fracturing divides a horizontal (or deviated or vertical, in some occasions) well into a few
sections (stages). In the case of cased wells each section is perforated and the surrounding rocks are
fractured, separately. Each stage might be perforated in a few clusters to better distribute the stimulation
fluid and proppant along the wellbore.
Multistage fracturing is more efficient than fracking the entire well simultaneously since it allows more
control on the fracture propagation and fluid injection and provides more focused pressure. Multistage
fracturing generally starts with the perforation and fracking at the tow, which corresponds to the deepest
section of the well and is usually numbered as stage 1. At the next stage, with a proper packing system
(second deep part of the well), stage 2 is perforated and hydraulically fractured. This procedure continues
to the heel of the well or until it is necessary.
In some cases a method called Zipperfrac (Vermylen and Zoback, 2011) or Sequential fracking (Nagel
et al., 2013) is used. With multiple horizontal wells, perforation at each stage of one well (well 1 for
instance) will be followed by the same stage at the nearby well (well 2) before going to next stage (stage
2) in well 1. This alternate fracturing continues until the last stages in both wells are fractured. Zipperfrac
method reduces the time required for the multiwell and multistage fracturing.
Another method of fracturing in multiple wells is the Simulfrac method. In Simulfrac method, stages
1 of both wells are fracked simultaneously before synchronized fracturing of stage 2. Stage 2 of the two
wells is stimulated after finishing the stage 1 fracturing and the procedure continues for all stages. This
may increase the total connectivity between two wells and the total SRV (Vermylen and Zoback, 2011).
6 SPE-177277-MS

Geomechanics of hydraulic fracturing


A brief explanation of elasticity terms and relations is explained next to better understand the mechanism
of hydraulic fracturing and faulting for sliding along a weak plane. Stress is the main player and an
important concept in rock mechanics and hydraulic fracturing. Stress is defined as the force per unit area
of a surface:
(1)

where F is force, A is unit surface area, and ␴ is the stress in Pascal (1 Pa ⫽ 1 N/m2) in SI. For our
purpose, stress is considered to be positive for compression and negative for tension. If force is normal
to the surface, stress is known as a normal stress. If force is tangential to the surface, stress is a shear stress.
Considering a small cube inside a stressed rock, stresses that act on each surface of the cube can be
resolved into one stress normal to the surface and two stress vectors perpendicular to each other and
tangential to the surface. The commonly accepted convention is that ␴ij shows the stress component in the
i direction on a surface with the normal in the direction of j as seen in in 3D in Fig. 2. As an example,
␴xx shows the normal stress on the plane orthogonal to the x axis (Fig. 2). Stress is a second order tensor
and can be shown by a 3 ⫻ 3 matrix ␴ as:
(2)
SPE-177277-MS 7

Figure 2—Components of stress in acting on a surface of cube three dimensions.

For a material in static equilibrium, stresses that act on one surface of a cube must be balanced with
stresses from the opposite side of the cube. Therefore, for any balanced material the ␴ij ⫽ ␴ji relation is
valid. This symmetry relation reduces stress components down to only 6 independent elements as ␴xx, ␴yy,
␴zz, ␴xy(⫽ ␴yx), ␴xz(⫽ ␴zx), and ␴yz(⫽ ␴zy).
Strain, another second order tensor, is the relative change in the shape or dimension of the material
under stress. Normal strain shows the relative increase in length,
(3)

where u is the displacement in x direction (Fig. 3). Similar definition can be used to explain ␧yy and
␧zz considering v and w as the displacements in y and z directions, respectively, as shown in Fig. 3 for a
2D scenario. Shear strain measures changes in the shape of a body under stress (Sheriff and Geldart,
1995). For example shear strain on a x-y plane is:
(4)
8 SPE-177277-MS

Figure 3—Strain components in 2 dimensions. Black square changed to the red diamond under stress (Sheriff and Geldart, 1995).

Strain matrix includes all longitudinal and shear strain components and is expressed by:
(5)

Strain involved in seismic and microseismic events is very small (in the order of about 10– 8).
Therefore, the relation between stress and strain in exploration seismology is linear and can be expressed
by Hook’s law (Sheriff and Geldart, 1995). Hook’s law in a matrix form may be expressed as:
(6)

where ␴ and ␧ are stress and strain matrices, respectively, and C is the stiffness matrix with 36
independent components. Stiffness components are either zero or some combination of Lamé first elastic
constants or fluid incompressibility, ␭, and the second elastic constant or shear modulus, ␮. Shear
modulus or rigidity is the ratio of the shear stress to the shear strain and is always positive.
(7)

Since strain is dimensionless, the shear modulus has the same unit as the stress unit. Rigidity is zero
for a non-viscose fluid, ␮ ⫽ 0. For a material under uni-axial tensile stress, ␴xx for example, dimension
parallel to x axis increases and dimensions perpendicular to x axis decreases. Poisson’s ratio, v, is the ratio
of the relative changes in the orthogonal directions:
SPE-177277-MS 9

(8)

Poisson’s ratio is always positive and varies between about 0.05 for extremely hard rocks and very
close to, but less than 0.45 for very soft rocks. The value of Poisson’s ratio is a maximum for a nonviscose
fluid and is equal to 0.5. Table 1 shows the typical values of E and v for some common rocks. As an
example Poisson’s ratio values equal to about 0.1 and 0.3 are considered low and high, respectively, in
the Montney formation. Young’s modulus shows the extensional deformation as the ratio of the normal
stress and strain along the normal stress axis:
(9)

Table 1—Example of magnitudes of elastic properties in rocks (Sheriff and Geldart, 1995; Jaeger et al., 2007).
Rock Young’s modulus, E, ⴛ 109 Pa Poisson’s ratio, v Density, ␳,gr/cm3 P Velocity, VP, km/s S Velocity, Vs, km/s

Sandstone 16 034 2.1–2.4 2–.35 0.8– 1.8


Limestone 54 0.25 2.4–2.7 3.5–6 2–3.3
Granite 50 0.20 2.5 – 2.7 4.5–6 2.5– 3.3
Shale 20 0.35 – 0.45 2.0–2.4 1.1–2.5 0.2– 0.8

The relationships between Lamé constants and Poisson’s ratio and Young’s modulus can be shown for
instance by
(10)

Brittle rocks have higher Young’s modulus and plastic rocks have higher Poisson’s ratios. Hydraulic
fracturing in a formation with high E and low v results in many cracks which are not necessarily open.
Wider cracks can be expected to occur when fracking rocks with low v and E. Microseismic events tend
to grow toward areas with lower v and stress (Cipolla et al., 2012).
If there is no tangential stress on a surface and all stresses are normal, they are called principal stresse:
and are represented by ␴1, ␴2, and ␴3. In general, ␴1 is considered as the major stress, ␴2 as th⬍
intermediate and ␴3 as the minor principal stress, and ␴1 ⱖ ␴2ⱖ ␴3.
State of stress in the crust is considered by three principal stresses, one in the vertical direction, an(two
in horizontal directions. Since we are interested in the stress regime inside the earth’s crust, w⬍ generally
consider ␴v to be the vertical stress due to overburden pressure (integration of density times the gravity
acceleration of the overburden layers), and ␴h and aH to be minimum and maximum horizonta stresses,
respectively, which are perpendicular to each other and to the vertical stress. As an example, ii a tensional
tectonic regime where the normal faulting is more frequent; ␴1, ␴2, and ␴3 correspond to ␴v ⬍H, and ␴h,
respectively. Reverse faulting happens when aH ⱖ ␴h ⱖ ␴v is the state of stresses in the earth Strike slip
faulting occurs when aH ⱖ ␴v ⱖ ah.
At the scale of hydrocarbon reservoirs, fluid trapped in the porous rocks plays as a negative force to
the confining stresses on the pore walls. Pore pressure is related to hydrostatic pressure which increase!
with the rate of about 10 Mpa/km (0.44 psi/ft) with depth (Zoback, 2007). Pore pressure is equal to the
hydrostatic pressure as long as pores are connected to the surface. However, pore pressure may exceed the
hydrostatic pressure in confined reservoirs when the formation is hydraulically isolated from the surface
10 SPE-177277-MS

(Zoback, 2007). Pore pressure decreases the effect of confining pressure. The effect of pore pressure on
normal stresses in a saturated rock is expressed by,
(11)

where, ij is normal effective stress, P is the pore fluid pressure, and ␣ is Biot-Willis coefficient
(Terzaghi, 1923; Biot, 1941).
Fig. 4 schematically illustrates changes in pore pressure and its effect on decreasing the effective stress.
As illustrated in Fig. 4 the depth level with high pore pressure is called the over-pressure zone. The
underpressure zone has lower pore pressure. Slip of pre-existing fractures is more likely to occur in the
overpressure depth where the effective normal stress is lower.

Figure 4 —Schematic illustration of the effect of pore pressure in reducing the lithostatic pressure (Schmitt, 2014).

The effect of porosity on fracture extension due to pore pressure increment is very important due to its
effects on the Biot-Willis coefficient (␣). The value of ␣ as suggested by Geertsma (1957) and Skempton
(1960) and derived by Nur and Byerlee (1971) may be calculated using,
SPE-177277-MS 11

(12)

where K and Kg are the effective and grains bulk modulus, respectively. Therefore, ␣ is a characteristic
of the porosity and the pore shape of the rocks. Bulk modulus, K, is low for low porosity tight sedimentary
or igneous rocks and is large for porous rocks (Walsh, 1965). The value of Biot-Willis coefficient for
reservoir rocks varies between 0.5 and 0.8 (Detournay et al., 1989). However, ␣ can be as low as 0.19 for
some low porosity rocks such as Tennessee Marble (Rice and Cleary, 1976). Since hydraulic fracturing
treatment is performed mostly in tight reservoirs with low and ultra-low permeability, the fluid injection
and propagation might be explained by assuming the presence of permeable natural fractures and flaws
in the reservoir formations. Considering the effect of pore pressure, the effective principal stress in the
earth is shown by 1, 2, and 3, where i ⫽ ␴i — ␣P.
Knowing the magnitude of the principal stresses, it is possible to find the stress on any arbitrary plane
in the crust. The relation between ␴n and ␶, normal and shear stress on an arbitrary plane, and the effective
principle stresses in two dimensions (␴2 ⫽ ␴3) are,
(13)

and
(14)

where ␴1 and ␴3 are the largest and minimum principal stresses ␤ and is the angle between the normal
to the arbitrary plane and the direction of maximum stress, ␴1 (Jaeger and Cook, 1979). These two
equations define the Mohr circle parameters in 2D. Although these mathematical equations provide a
meaningful method to analyze the stresses on any arbitrary surface, Mohr circles and Mohr failure
envelope provide a simple method to visualize the rock failure due to changes in the stress regime, too.
Considering the normal stress, ␴n, on horizontal axis and shear stress, ␶, on the vertical axis, one can draw
a Mohr circle considering the (␴n ⫽ 0.5(␴1 ⫹ ␴3),␶ ⫽ 0) as the center and 0.5(␴1 — ␴3) as its radius. A
line drawn from the pole (␴3 on horizontal axis) and with the angle ␤ in the anticlockwise direction
intersects the Mohr circle on a point with the ␴n␤ and ␶␤ coordinates (Fig. 5).

Figure 5—Mohr circle is used to find the normal and shear stresses on an arbitrary plane with the angle ␤ to the maximum stress, for
a uni-axial stress experiment.
12 SPE-177277-MS

Mohr failure envelope shows all combinations of ␴1and ␴3 that a rock sample can tolerate before
fracturing begins. Fracturing initiates when a circle defining ␴1 and ␴3 intersects the Mohr failure
envelope (Fig. 6). The Mohr failure envelope may be linearized and expressed by the following equation
(Jaeger and Cook, 1979),
(15)

Figure 6 —Mohr failure envelope and its linear approximation in a uni-axial compression test.

where S0 is called cohesive strength, cohesion, or unconfined compressive strength (USC), ␴n and ␶ are
the normal and shear stress on the plan, respectively, and ␮i ⫽ tan␾ is the internal friction coefficient and
␾ is the angle of internal friction. This is the Coulomb criteria which state that cohesion of the material
and the normal stress across a plane times a constant causes a rock to resist failure under a shear stress
on the plane. This is the criterion for slip on a weak plane under compression and is very similar to what
mostly happens during a hydraulic fracturing process. If the value of ␶ exceeds the amount specified by
equation (15), slip on the plane occurs. Stronger rocks have higher cohesion. The value of ␮i is usually
between 0.5 and 2 with a median value in the order of 1.2. Shale formations have a value as low as 0.2
(Zoback, 2007).
If the Mohr circle intersects with the failure curve at the quadrant with positive stress, it causes a shear
failure and results in mode II or III cracks. For a rock under uni-axial compression test, increasing either
␴1 or ␴3 or the change in the pore pressure causes the Mohr circle to touch the failure curve and fracturing
initiates. Increasing pore pressure decreases the effective stress without any change in shear stress. This
phenomenon shifts the Mohr circle to the left which causes the initiation of slippage on the weak plane
and fracture propagation (Fig. 7). For this reason fracturing in over-pressure zones happens more than in
under-pressure intervals. This is also the mechanism for the activation of fault planes and induced
microseismicity events related to the fluid injection for enhanced oil recovery or hydraulic fracturing. It
is notable that production from a reservoir has the opposite effect. Decreasing the pore pressure, increases
the normal stresses and shifts the Mohr circle to the right, farther from the Mohr failure envelope.
However, reservoir depletion may cause slide on the weak plans by either change in stress fields and
increasing the radius of Mohr circle or change in slope of Mohr failure envelop (Fig. 7).
SPE-177277-MS 13

Figure 7—Effect of pore pressure and change in the stress field on the Mohr circle. The initial stable Mohr circle (␴1, ␴3) touches the
failure envelope by the effect of pore pressure increase ( 1, 3) or changes in the stress field (␴’1, ␴3).

The difference in stresses that causes the slippage on a plane is a function of ␤ and the least principal
stress, ␴3, as shown by Jaeger and Cook (1979):
(16)

This equation shows that if a weak plane is in the direction of major principal stress, ␤ ⫽ 90°, then
␴1 — ␴3 must approach infinity for this slippage to occur (Jaeger and Cook, 1979). Another extreme case
is when ␤ ¡ ␾. Therefore, the solutions to the above equation are limited to the cases where ␤ ⬍ ␾ ⬍
90°. It can be shown that when tan 2␤ ⫽ ⫺1/␮ a minimum value of ␴1 for slippage is required which is
equal to (Jaeger et al., 2007),
(17)

For a given values of ␴1, ␴3, and ␮, sliding occur only on the plans between ␤1and ␤2 defined by
(18)

and
(19)

In the above equations ␴m is the mean normal stress, ␴m ⫽ 0.5(␴1 ⫹ ␴2), and ␶m is the maximum shear
stress, ␶m ⫽ 0.5(␴1 — ␴2).
Considering a planar fracture surface, the fracture plane has a dip, the maximum angle between the
plane and the horizon, a strike, the angle between a horizontal line on the fault surface and the
geographical north, and a rake, showing the direction of relative movement of the hanging wall block. In
normal and reverse faulting the relative movement is in the dip direction. Positive rank shows that the
hanging wall moved up and the faulting is reverse or thrust and negative rank indicates the normal faulting
in which the hanging wall moved down relative to the other fault block. In strike-slip faulting the
movement is along the fault plane strike direction. In addition to strike, dip and rake, which are important
factors in the definition of faults in structural geology, fracture height, width, and length, are character-
istics that influence the rate of hydrocarbon productivity in a hydraulic fracturing treatment. These
concepts are commonly used with the planar fractures. Height and length are the vertical and horizontal
extent of a planar fracture, respectively. Length, usually expressed by the half-length, is the maximum
distance that fracture propagates from the stimulated well and is generally about a few hundred feet.
14 SPE-177277-MS

Height is ideally restricted to the targeted layer. If height grows beyond the targeted layer the horizontal
extent of the fracture decreases significantly. Width is the opening of the fracture. Usually the maximum
width is at the well location and decreases with distance from perforation point.
Fracturing in rocks may be divided into three different modes. Mode I is an opening mode in which
a tensile stress normal to the crack plane causes the fracture growth. Mode II is a sliding mode where a
shear stress acts parallel to the crack plane and perpendicular to the crack front. Mode III or shear
dislocation with tearing mode is where the shear stress is parallel to the crack front (Fig. 8). Modes II and
III cracks are similar to strike slip faulting. The goal of moment tensor inversion is to classify fractures
into the abovementioned crack modes and fault types.

Figure 8 —Three types of fracturing modes; Left: Mode 1 or tensile opening, Middle: Mode II or In plane shear fracturing, Right: Mode
3 or out-of-plane shear fracturing.

Tensile and shear fracturing mechanisms and microseismicity


Maxwell and Cipolla (2011) considered creep deformation for the events with low frequencies which are
outside of bandwidth of seismic recording instruments. Tensile opening of the reservoir rocks is believed
to be either an aseismic (creep) deformation (Cipolla and Wallace, 2014) or a fracturing process that
releases seismic pulses which are mainly too weak to be detectable by the current recording instruments
or to be separable from noise by the current processing methods (Mueller, 2013).
This could be due to the low tensile strength in the reservoir rocks which causes the slow opening of
fractures or fracturing without significant release of energy in a short time. Rarely found rocks with high
tensile strength, may produce enough energy in form of microseismic signal to be detectable during
fracturing (Maxwell and Cipolla, 2011).
Microseismic signals are typically being related to shear (Albright and Pearson, 1982; Rutledge et al.,
2003; Maxwell, 2014) or hybrid shear-tensile failure (Maxwell and Cipolla, 2011; Busetti et al., 2014) on
the fracture plane.
Most of the energy induced to the reservoir by the fluid injection, which might be at the range of 10s
of gigajoules (GJ), is released by either tensile opening or fluid friction. Only a very small portion of the
induced energy is converted to microseismic signals by the shear failure (Maxwell and Cipolla, 2011;
Warpinski et al., 2013; Cipolla and Wallace, 2014).
Maxwell and Cipolla (2011) presented an example of hydraulic fracturing in which the total energy of
microseismic events was about 0.5 ⫻ 109 time smaller than the total injected energy. Warpinski (2014)
reported an example in which the total microseismic energy was 1 KJ where the total induced energy was
540 MJ. Table 2 from Cipolla et al. (2012) shows an example fracturing treatment where not only the
SPE-177277-MS 15

microseismic energy is very small fraction of the total fracking energy, but also the fracturing volume and
the fracturing area computed from microseismic measurements are many times smaller than the fracking
volume and area in the calculated stages.

Table 2—Comparison of microseismic area, volume, and energy with hydraulic fracture treatment energy and geometry (Cipolla et
al., 2012).
Stage 1 2 3 4

MS volume (bbl) 0.82 0.15 0.075 0.30


Frac. volume (bbl) 25,300 25,300 25,300 25,300
% 0.0033 0.00059 0.0003 0.0012
MS area (ft2) 19375 4018 3533 9040
Frac. area (ft2) 7,000,000 - 6,200,000 -
% 0.28 - 0.057 -
MS energy (J) 62,500 11,300 5,700 22,950
Frac. Energy (J) 149,040,000,000 149,040,000,000 149,040,000,000 149,040,000,000
% 0.0042 0.00076 0.00038 0.0015
MS: Microseismic, Frac.: Hydraulic fracturing

Microseismic events considered to be due to tensile openings are used in volume calculations by
Cipolla et al. (2012). With this assumption, microseismic volumes are about 10 million times smaller than
the SRV volume which can be about 10,000,000 m3 (Cipolla et al., 2012).
Not all aseismic deformations are related to tensile openings. The source mechanism analysis of shear
failures might show the slip area that is 1000s times smaller than the real slip area of the fault (Maxwell
and Cipolla, 2011). Therefore, the shear slippage can also be mostly slow and without detectable seismic
sound emission.
Assuming that the reservoir rocks are brittle and fracking is relatively quick, sudden crack opening or
slippage failure due to injection or stress field changes can create a series of very weak seismic pulses
known as (induced) microseismic signal. The frequency range of microseismic signal is between 1 and
1000 Hz which is higher than the frequency of injection pressure which is below 1 Hz (Maxwell and
Cipolla, 2011).
Emitted weak signals travel in all directions underground and are being recorded with many specific
high sensitive receivers, one-component or mostly three-component geophones or seismometers with high
natural frequency. The detected events are considered to be the result of breakage or slippage occurrence
by change in the pore pressure or in-situ stress field due to either injection of a large volume of high
pressure fluid, usually water with or without proppants, or from reservoir depletion during production
(Mueller, 2013).
Microseismic is referred as to the events with the Richter moment magnitude less than 0 or 1 and more
than about ⫺3 (Maxwell, 2014; King, 2012). Signals with magnitudes less than ⫺3 Richter are difficult
to detect or separate from noise.
Due to their wide dynamic range, moment magnitudes are expressed on a logarithmic scale. One unit
increase in magnitude is equal to 10 times increase in the signal amplitude and 32 times increase in the
released energy. Therefore, the relative amplitude of microseismic data has a broad range between 1 and
10000 or more. This wide range of amplitudes makes the proper recording, detection, and analysis of the
microseismic data to be much more difficult than that of the seismic data.
Since hydrocarbon reservoir formations are naturally fractured in general (Aguilera, 1980, 2003), the
effect of natural fractures should be considered in any hydraulic fracturing modeling. Maxwell and Cipolla
(2011) presented two different models to describe the microseismicity of hydraulic fracturing. Both
models consider the presence of natural fractures in the reservoir and a tensile mechanism, mostly
16 SPE-177277-MS

aseismic, of hydraulic fractures and microseismic signals as the result of shear movement of pre-existing
natural fractures. Since the change in the state of stress due to fluid injection is not sufficient to create a
new fracture in consolidated rock, shearing in a hydraulic fracturing should only occurs at the weak planes
of preexisting natural fractures or bedding (Warpinski, 2014).
In the first model, natural fractures are parallel or sub-parallel to hydraulic fractures and the hydraulic
fractures are orthogonal to the direction of minimum stress field (Fig. 9). Growth of hydraulic fractures
reduces the normal stress on the preexisting fracture and causes the shear slip of natural fractures.

Figure 9 —Two mechanisms for fracture growth. Top: natural fractures are sub-parallel to hydraulic fracture, Bottom: natural fractures
are orthogonal to hydraulic fracture (Modified from Maxwell and Cipolla, 2011).

Since the natural fractures are almost parallel to hydraulic fractures, when the hydraulic fracture
reaches the natural fracture, dilation in natural fracture begins and it becomes a hybrid fracture producing
both tensile and shears microseismic components. With continuing fluid injection, dilation of both natural
fracture and hydraulic fracture continues as aseismic deformation (Maxwell and Cipolla, 2011). This
model is not able to explain the complex fracture network and cloud of microseismic events which are
observed in most shale reservoirs such as Barnett Shale. The microseismic events behind the leading edge
of the events cannot be explained using this model either (Maxwell and Cipolla, 2011).
SPE-177277-MS 17

The second model considers the natural fractures to be mostly orthogonal to the hydraulic fractures and
consequently parallel or sub-parallel to the direction of the minimum stress (Fig. 9). Extension of
hydraulic fracture changes the stress regime and induces shear deformation on the preexisting natural
fractures which produces shear microseismic signals.
However, the natural fractures do not dilate significantly due to the direction of maximum stress which
tends to force the natural fractures to close. Natural fractures still act as a permeable path to transfer fluid
and hydraulic pressure to another location which might be an explanation for complex fracture network
seen in shale reservoirs. This model is able to explain the microseismic events behind the edges (Maxwell
and Cipolla, 2011).
In the second model, tensile deformations generally happen in the direction of minimum stress and
shear fractures mostly occur along the preexisting natural fractures and perpendicular to the direction of
maximum stress. However, in addition to the orientation and size of natural fractures and the stress
regime, the pattern of microseismic events also depends on rock properties and fracture treatment design
(Cipolla et al., 2011a). Fig. 10 illustrates the expected different microseismic patterns due to deformation,
leakoff, and activation of preexisting fractures in low permeability gas reservoirs and in an oil or water
reservoir with moderate permeability. More diversity of event patterns is observed when the effect of pore
pressure is added to the model.

Figure 10 —Schematic general patterns of microseismic in a low permeability gas reservoir; Different patterns are hypothetically
related to a) deformation and leakoff, b) limited activation of natural fractures, c) extensive natural fracture activation, d) deformation
and leakoff in oil/water reservoirs with moderate permeability (Cipolla et al., 2011a).

Nolte and Smith (1981) suggested a slightly different model for the areas with low anisotropy in
horizontal stresses. Natural fractures which are significantly more permeable than the surrounding
18 SPE-177277-MS

reservoir matrix allow a relatively deep penetration of injected fluid from the injection point. Natural
fractures open when the fluid pressure exceeds the closing stress according to the following condition
(Nolte and Smith, 1981; Warpinski, 2014):
(20)

where Pf and Pt are the net pressure in the fracture and bottom hole pressure, respectively, ␴min and
␴max are the minimum and maximum horizontal in situ stress, respectively, and v is the Poisson’s ratio.
Warpinski (2014) considers this mechanism to be responsible for the complex fracture network found in
the Barnett Shale.

Magnitude and frequency of microseismic events as an interpretation aid


The Richter scale has been used for many decades for measuring the naturally occurred earthquake
magnitudes with the use of the following equation,
(21)

where A is the largest amplitude recorded by Wood-Anderson seismograph, ⌬ is the distance between
earthquake and seismograph, and ML is the magnitude in Richter scale. This definition was initially used
for earthquakes in California, USA, which were recorded by Wood-Anderson seismograph. Events
recorded with other seismographs around the world needs to be calibrated to the Wood-Anderson
seismograph mechanical specifications. Seismic moment, M0, of a specific earthquake is related to the
geometry of a fault by,
(22)

relation, where ␮ is the shear modulus at the fault location, D is average fault slip, and A is total slipped
area of the fault surface. Assuming that the displacement is mostly a tensile fracture opening, multipli-
cation of the fracture surface, A, and displacement, D, is very important since it is a measure of
(maximum) fracture volume associated with the microseismic event (Cipolla et al., 2012). Microseismic
moment can be calculated using the following equation (Cipolla et al., 2011a):
(23)

where ␳ is density, R is the offset between source and sensor, C is the wave speed, ␻ is the low velocity
level of displacement for P-waves, and F is the radiation pattern or directionality factor. For shear waves
␻ ⫽ (␻2SV ⫹ ␻2SH)0.5 where SV and SH stand for the vertical and horizontal components of shear waves,
respectively. The radiation pattern factor can be considered as 0.5 for P waves and 0.63 for S waves.
Moment magnitude, Mw, is the logarithmic scale of M0:
(24)

Most recorded earthquakes have an Mw less than 9.6. As an example the large Chile earthquake in 1960
had 2.5 ⫻ 1030 dyn.cm moment and Mw ⫽ 9.57. Moment and Richter magnitudes are scalars showing the
amount of energy released during earthquakes without considering the faulting mechanism or the
environmental damage.
It is notable to mention that the amount of damage to structures on the earth’s surface depends on other
important factors in addition to the magnitude of the earthquakes. Depth of the earthquake hypocenter, the
distance of the observation point to the epicenter, the faulting mechanism, and the radiation pattern of the
S-waves are the main factors affecting the degree of damage to earth surface structures. Damages to
human-made structures are described by the modified Mercalli intensity scale. Modified Mercalli intensity
SPE-177277-MS 19

scale varies between I for very weak earthquakes to XII for extreme earthquakes with total damage to
almost all human-made structures.
Seismic (and microseismic) moment tensor, M, represents the source mechanism and might be related
to M0 by,
(25)

in the Cartesian coordinate system, where Mij shows a force couple in the ⫾i direction acting in two
closely spaced points on an arm in the j direction as seen in Fig. 11. M is a symmetric matrix with 6
independent elements. The Mxx, Myy, and Mzz components are linear dipoles without moment and arm and
force are at the same directions. The off-diagonal components have the arm perpendicular to the force and
are couples with moment (Udias, 1999). The shape of a seismic moment matrix depends on the source
strength and the fault orientation. For example an explosive source can be represented by the following
matrix:
(26)
20 SPE-177277-MS

Figure 11—The nine elements composing a seismic moment tensor.

For an isotropic volumetric change, the diagonal elements are equal. This explosive source does not
produce any S-wave and the generated P-wave signals, with equal polarity and amplitude, propagate in
all directions. In shear faulting which is the main mechanism in the majority of earthquakes, the force and
arm are perpendicular to each other (Fig. 12). Shear source and compensated linear vector dipole (CLVD)
source moment tensors can be shown for example by,
(27)

and,
SPE-177277-MS 21

Figure 12—Left: A Double Couple source mechanism for a pure shear events and the P wave radiation pattern. Green arrows show the
relative slip of the strike-slip fault planes. Right: the corresponding beach ball diagram.

(28)

Any real earthquake or microseismic moment tensor can be decomposed into the isotropic and
deviatoric matrices for analyzing the fracturing mechanisms. Deviatoric moment tensor can further
decompose into simpler moment matrices such as double couple (DC) and CLVD. DC is the result of
shear slippage on pre-existing fracture planes as a result of change in the stress field or pore pressure and
generally results in modes II and III cracks. Fig. 12 illustrates an example of DC source mechanism with
its P-wave radiation pattern. In Fig. 12 the DC is created by two elements, Mxy and Myx. The fracture
happens in x-z plane and the ellipses show the radiation pattern of P-waves. The area in the first and third
quarter is compressed and the resulted P-wave shows a positive (first motion is up) amplitude. Conse-
quently, the areas in other two quarters are dilated and create negative P-wave amplitude (first motion is
down).
Beach ball is a graphical mean of moment tensor and fracturing focal mechanism. Beach balls have
been widely used to show faulting mechanisms and tectonic regimes on tectonic maps. A beach ball is a
stereographic projection of a fault surface passing through the center of the hemisphere on the lower
hemisphere. Earthquake hypocenter is considered to be at the center of the hemisphere. Beach balls have
up to two intersection curves which divide the whole area into up to four sections. Conventionally the
darker areas of the beach ball shows the compression parts and the white parts show the tension regions
based on the first arrival of seismic P-waves (Fig. 12). Therefore, an explosion which creates only
compressional waves propagating in all directions is shown by a completely dark circle (Fig. 13). A DC
mechanism may be shown by a crisscross shape for mode II crack, or a circle with half dark and half white
for mode III cracking. Fig. 13 shows most common types of beach ball presentations and associated
moment tensors. In Fig. 13 the first row (gray circles) represents an isotropic source which is either
22 SPE-177277-MS

explosive (compressional) or implosive (tensional). Different DC source mechanisms are shown in rows
2 to 4 of Fig. 13, where figures in the second row (red circles) correspond to the Mode II cracking and
figure in the third and fourth rows (blue circles) are for a Mode III cracking. Finally last two rows (black
circles) of Fig. 13 are representations of CLVDs.

Figure 13—Moment tensors and associated focal mechanism beach ball diagrams (Modified from Stein and Wysession, 2003).

Gutenberg and Richter (1944) found the following linear relationship between the logarithm of the
number of earthquakes in California and the earthquake magnitudes in Richter scale,
(29)

where M is the magnitude, N is the number of earthquakes with magnitude larger than or equal to M,
and a and b are empirical constants. Gutenberg-Richter relation, also known as ⬙power law⬙, simply states
that smaller events occur many times more than the larger ones. For instance when an earthquake with the
magnitude of 8 or higher happens about once a year, earthquakes with magnitude about 6 — 6.9 happens
SPE-177277-MS 23

about 130 times a year, and one earthquake with the magnitude 2 — 2.9 happens about every 4 seconds
globally. Slop of this line in a semi-log graph called b-value and is about 1.0 for most earthquakes
happening in the crust. Equation (29) shows that there is 10 fold increase in event occurrence for each unit
decrease in magnitude. However, it is found that the Gutenberg-Richter relationship may not be valid for
smaller earthquakes and fewer events may be recorded with the magnitude less than 3 (Aki, 1987). In
earthquake seismology, this could be due to limited ability of detection of weak events. Constant a, the
intercept, is considered to be the level of seismic activity of a particular area in earthquake seismology
(Vermylen and Zoback, 2011). Higher a corresponds to an area with a higher number of earthquakes.
The concept of power law and the meaning of a and b-values in earthquake seismology is extended to
analyze of microseismic measurements. Urbancic et al. (1999) used the frequency-magnitude analysis on
microseismic data to estimate the seismicity rate and fracture density in hydraulic fracturing treatments.
Empirical analysis of microseismic measurements showed that the b-values are generally greater than 1
and about 2 for events resulted from hydraulic fracturing. Consequently the usual assumption is that when
two distinct b-values are obtained from microseismic data, events with b-value of 2 or larger are generally
considered to be the result of hydraulic fracturing and events with magnitude of about 1 are believed to
be the result of activation of pre-existing natural fractures and faults. Fig. 14 illustrates an example of
microseismic events recorded during and after a hydraulic fracturing treatment. These data clearly show
two distinct trends with b-values equal to about 2 and 1 which can be related to the hydraulic fracturing
and post-fracturing fault activation, respectively. A reliable application of extending the GutenbergRichter
relation to the microseismic monitoring is the ability to isolate and distinguish hydraulic fractures from
the activation of preexisting fractures. This separation is very important for creating reliable models to
simulate hydraulically fractured reservoirs and to analyze faulting mechanisms.

Figure 14 —Frequency-magnitude relationship for the events recorded during a fracturing treatment. Brown: events with the b-value ~2
indicating the hydraulic fracturing. Orange: events with the b-value ~1 indicating the fault activation after fracturing (Modified from
Maxwell et al., 2009).
24 SPE-177277-MS

Analysis of the a value or activity level can be used to remove the effect of detection distance bias as
shown by Vermylen and Zoback (2011). Events from the same fracturing treatment may have different
detection limit in far and nearby observation wells. However, it is expected that both measurements have
the same a values regardless of the offset as illustrated in Fig. 15 (left). Fig. 15 (left) illustrates same
fracturing data recorded with two arrays of receivers, one (blue line) closer and one (red line) farther from
events. Blue line shows events with the magnitude as low as ⫺3.0. The detection limit for red line events
is ⫺2.5. However, both lines have same a and b-values suggesting a method to isolate the sensor effects
(Vermylen and Zoback, 2011). Fig. 15 (right) shows the frequency of events versus magnitude for two
fracturing stages at the same sensor array. Both lines have same b-value but different activity levels.
Purple line in Fig. 15 indicates that more events recorded at closer distance. However, events showed by
green events are more robust since they show a higher level of activity (Vermylen and Zoback, 2011).

Figure 15—Frequency of events versus event’s magnitude; Left: Same events measured from two close (blue) and far (red) wells; Right:
Data from two fracturing stages recorded at the same receiver array show same b-value but different activity levels. (Vermylen and
Zoback, 2011).

Maxwell et al. (2006) and Cipolla et al. (2012) recommend avoiding number of events (quakes) as a
measure of activity level since many small events may represent very small amount of total released
energy in comparison to a few larger events. Cipolla et al. (2012) indicate that total source strength or
cumulative seismic moment is less sensitive to the variations in detection distance and signal to noise
(S/N) ratio and are thus better quantities to measure microseismic activities.
Microseismic monitoring
Single, or commonly three component geophones, are firmly placed on the ground surface, in the same
manner as in a surface 3D seismic data acquisition, buried at a shallow depth, or positioned in 100 to 300
m depth intervals in a well at 8 to more than 40 locations lower or frequently higher than the injection
depth (Mueller, 2013). Geophones buried up to a few meters below the surface help to attenuate the
ambient noise and avoid the problems associated with the low velocity weathered near-surface layers.
Combination of the abovementioned designs is possible and increases the quality of measurements and
reduces the uncertainties by recording emitted signal from different azimuths.
SPE-177277-MS 25

Microseismic data recorded using surfaces receiver arrays have the advantages of providing better
coverage and more precise information about the horizontal location of the events, and azimuth and length
of the fractures. Surface measurements can be used for detection of event location using P-wave arrival
times only. P-waves are practically less problematic than the S-waves to analyze (Duncan, 2010). Surface
measurements are less sensitive to uncertainties in the horizontal variations of the velocity model. There
is no limitation on the number, extent, and type (1C, 3C) of receivers on the surface which makes
monitoring of the reservoir depletion and stress change during the production more feasible (Duncan and
Eisner, 2010). Another benefit of surface surveying is that it is not required to specify a production well
for observation. However, largely extended surface surveys are much more expensive than the borehole
acquisitions. Surface recorded data offer relatively lower vertical resolution and may also suffer from the
high environmental noise, mostly from pumping engines, other drilling activities in the area, fluid flow
friction, and other surface activities. Surface monitoring may also leave some environmental footprints,
land may not be accessible for the crew, and the method is not easily extendable to offshore reservoirs.
Downhole microseismic methods, at the other hand, which are typically performed in a single or a few
vertical, deviated, and/or horizontal wells with three component geophones at certain depth levels or
horizontal distances, are less costly, and provide enhanced depth resolution and better estimation of
fracture height. These measurements are less contaminated with the surface noise, too. However,
minimum one well must be considered as the observation well. Higher quality of data is achieved when
monitoring wells that are in the vicinity of the well to be perforated and hydraulically fractured in multiple
stages. However, there are limitations on the number of receivers which can be placed at each well and
borehole data have lower horizontal accuracy on the event location. The poor vertical and horizontal
resolution in the velocity model is a challenge for precise determination of the hypocenters with the
downhole measurements.
Due to technical and economic efficiency, single-well downhole monitoring of microseismic data is
currently the most frequently method used to monitor hydraulic fracturing. Downhole instruments are able
to detect signal with magnitude as low as about ⫺3 Richter where the average low detection limit for the
surface recordings is about ⫺2.5 Richter (Mueller, 2013). An ideal data gathering plan is to have
microseismic data recorded in more than one nearby well, surrounding the hydraulic fracturing area, at
different depth levels. Recording data in multiple wells significantly reduces the uncertainties and
increases the precise determination of event location and makes the moment tensor inversion possible.
Recording data in multiple horizontal and vertical wells provide an accurate velocity model which is very
critical in determination of events location, SRV calculation, and discrete fracture network (DFN)
creation.
Processing and interpretation of microseismic data
The correct amplitude of the P- or S- wave arrival times (event detection), determination of event location
(finding the hypocenter location of microcosmic source), fracturing time, and events’ magnitude are the
primary information to get extracted from processing and interpreting microseismic data. Quality control
of data and the acquisition geometry, an estimate of the S/N ratio, and evaluation of the available data for
velocity model building are necessary prior to any data analysis including the determination of event
location.
Event detection is the first and probably the most important and time consuming step in the processing
and interpretation of microseismic data. Since every further analysis and use of microseismic measure-
ments strongly depends on the accurate recognition of events, event detection must be performed with
high accuracy. Assuming that the velocities are exactly known, each millisecond error in determination
of arrival time may introduce more than 4 m error in event position relative to the sensor. The accuracy
of event detection depends on the S/N ratio. Event detection may be facilitated by visualizing data from
all components and receivers on a single computer screen. Comparing arrived and recorded signals on
26 SPE-177277-MS

different components provides better separation between P- and S- wave arrivals, and the environmental
noise. Computer algorithms exist to detect and separate events such as short-term average to long-term
average ratio (STA/LTA ratio), and to calculate power spectral density (PSD) and peak Eigen values
(PEV). However, the reliability of these algorithms in event detection depends on the data S/N ratio. For
data which are highly contaminated with environmental noise, the amplitudes and times determined by an
experienced data analyst may be more reliable than using a sophisticated algorithm. This is a time
consuming procedure and may not be efficient when real time monitoring of hydraulic fracturing is the
main concern.
Events location might be simply determined by the Triangulation, a classical method that seismologists
use to locate earthquake epicenters if the signal is recorded in more than two stations which may not be
located in a straight line with the fault location. In Triangulation method, an earthquake epicenter is found
by measuring the distance of the event to the known receiver locations. Detecting only P- or S-wave
arrival times at three locations is enough to find the event location precisely. A method to locate events
using P- and S- data with a single receiver is also available. In this method the distance of an event from
the sensor can be calculated using the difference in P- and S-wave travel times. Suppose TP and Ts are the
arrival times of the P- and S- waves, respectively. The difference between these two arrival times, ⌬TPS,
is ⌬TPS ⫽ TP — Ts. The distance between the event and receiver, x, is calculated by:
(30)

where P and s are the average speed of P- and S- waves in the medium between the event and the
receiver. The event is located somewhere on a sphere with the radius of calculated distance, x, and sensor
location as the center. The azimuth of the event relative to the receiver is determined using a P-wave
hodogram analysis. Notice that in calculating the distance using the above equation, there is no need to
know the time at which fracturing occurred. More advanced signal processing and inversion techniques
may be implemented for finding a more realistic location. Special considerations in regard to sensor
orientation is required when locating the events using data recorded in deviated or horizontal wells.
Precise determination of the location is very important since any further evaluation of the fracturing
extension and dynamic of fracturing process and reservoir simulation is strongly depended on the fracture
network constructed based on the microseismic event location.
Moment tensor inversion, waveform spectral analysis, and analysis of the ratio of shear to compression
wave amplitudes lead to a comprehensive analysis of source strain tensor and fracturing mechanisms.
With high quality well-sampled microseismic wavefield, usually from multiple wells surrounding the
fracturing area or from wide azimuth and long aperture surface acquisition covering the entire injection
area, the fracturing mode, fracture geometry including orientation, dip, length, height, and width; and
details of stress axes may be extracted by proper data analysis and inversion of the moment tensors (Eisner
et al., 2010). Moment tensor inversion may be used for identifying the fracture opening from fracture
closing in addition to distinguishing the tensile failure from shear failure. However, due to relatively low
energy of tensile opening and closing, this interpretation is currently uncertain (Cipolla et al., 2012).
Unlike in regular surface seismic surveying, processing and interpretation of microseismic data are
coupled together and it would be better if these two procedures are performed in the same office.
Appropriate processing of data requires some information from the interpretation side, and interpretation
may require re-processing of data. Re-evaluation of the velocity model, for instance, is a processing step
which requires some geological and geophysical interpretation as well as primary interpretation of
microseismic records.
SPE-177277-MS 27

First step in interpretation of microseismic data


Evaluation of uncertainties in the event location is the first step in interpretation of microseismic data
(Cipolla et al., 2011a). Location uncertainty is directional, meaning for instance that there may be more
confidence in the depth of an event than in the horizontal location of the event. Uncertainty in location
can be shown by a 3D error ellipsoid at the event location (Fig. 16). For a large error in a given direction,
the error ellipsoid has longer axes on that direction. It is important to filter out events with high degree
of uncertainty previous to further use of data. However, there is not a definitive role in assigning a cut-off
value to accept a microseismic event as a fracture or to reject it as noise with a view to discard the
corresponding data. The overall quality of data, the S/N ratio, and the complexity of fracturing patterns
are some important factors to consider before filtering microseismic measurements. Uncertainty in
velocity model is coupled with the ambiguity in the events location and it could be considered during the
calculation of error ellipsoids (Cipolla et al., 2011a).

Figure 16 —Example of error ellipsoids in a section view. Top figure: Events location; Bottom figure: Error ellipsoids (Modified from
Cipolla et al., 2011a).

Second step in interpretation of microseismic data


The second step in the interpretation of microseismic data is evaluation and remediation of observation
biases. Sources of observation bias include the effect of viewing distance (this called offset in surface
seismic surveys), viewing azimuth, and the possible effect of nodal planes (Cipolla et al., 2011a). Low
magnitude events become less detectable with increasing the distance between events and the sensor.
Calculating the lowest magnitude event farthest from receiver on a magnitude versus distance graph helps
to identify the detection limit of the microseismic data for a certain treatment stage. Assessment of the
28 SPE-177277-MS

event magnitudes versus distance determines if the fracturing area is completely covered by the acqui-
sition design or if there is a requirement to account for the viewing distance bias. The fracturing is
sufficiently mapped by the data gathering if there are many higher magnitude events beyond the detection
limit. Normalizing the events density by distance is necessary for removing the possible distance bias.
This can be accomplished by discarding data with magnitudes lower than the detectable magnitude in the
farthest distance from receiver. Distance bias becomes more problematic when the number of events in
a specific volume (event density) is used in the interpretation (Cipolla et al., 2011a). Calculation of SRV
based on the number of events in an unit area or volume is a good example of this problem.
Due to hodogram variations, the azimuth (relative to the sensor location) uncertainty is larger than the
distance uncertainty. Azimuthal uncertainty increases with increasing distance to the observation point.
Azimuthal uncertainty in the farther stages might result in wrong estimation of fracture length. The
viewing azimuth bias can be visualized using error ellipsoids (Cipolla et al., 2011a). Finally, the nodal
plane bias may be evaluated by the ratio of P-wave to horizontal S-wave amplitudes (P/SH ratio).
Localized very low P/SH values correspond to the P-wave nodal areas and S-wave nodal areas have high
P/SH ratio. The radiation pattern correction is important in areas where slippage is dominant in one
direction as strike-slip faults can be activated. However, this correction may be unnecessary or impossible
if the fracture network is complex (as in areas where the SRV concept is applicable). This is the case of
many tight shale gas reservoirs (Zimmer, 2011); also light oil and condensate shale reservoirs.
Anomalies in the event patterns can be useful in identification of pre-existing faults and natural
fractures. Cipolla et al. (2012) consider any asymmetry in the pattern of microseismic events around the
perforation point as an artifact if it is not due to geology or interaction with the depleted formations. Fault
activation may be identified by anomalous high magnitude localized events on a magnitude versus
distance graph, a group of events that are not well-connected to the main network of events and may show
deferent azimuth, and/or presence of many high magnitude events when fluid injection terminated
(Cipolla et al., 2011a).
Interpretation of microseismic data may include a comparative interpretation, comparing microseismic
responses from two adjacent fracturing treatments at the same formation by assessing the difference in
fracture geometry or induced microseismic deformations (Cipolla et al., 2012). Interpretation is performed
by comparing the SRV, number of events, or the event’s patterns.
Comparative interpretation is more convenient than interpretation of single fracturing as both fractur-
ing treatments have a comparable velocity model and generally similar degrees of uncertainty or
confidence. Consistency of the acquisition pattern and fracking strategy is very important for a valid
comparative interpretation. In theory this is excellent. In practice this might be challenging as in many
instances procedures must be changes during operations in the field. Special considerations are necessary
when interpreting the difference in geometry of two cases since the observation distance or direction may
create a bias in comparing interpretations. As shown in an example by Cipolla et al. (2012), larger
fracturing volume observed farther from the observation well could be due to geometric directional error
which increases from the observation point. Comparison interpretation is a very valuable tool in
evaluating the fracturing overlaps from adjacent stages in order to economically improve completion
strategies (Cipolla et al., 2012). In absence of an accurate velocity model, real time monitoring can still
help in fracture diversion and refracturing by comparative interpretation of data. Cipolla et al. (2012)
presented an example of refracturing using real time microseismic monitoring where the well productivity
doubled and long term productivity increased by 60%.
Special caution must be taken when the number of events is counted for comparative interpretation of
microseismic data. Fig. 17 (left) from Cipolla et al. (2012) shows the magnitude verses distance of two
fracking stages A and B. Stage B is closer to the observation well and shows a higher number of events
than the stage A which is farther from receivers. Considering the number of events as the activity rate may
lead to misinterpretation of data as mentioned earlier.
SPE-177277-MS 29

Figure 17—Magnitude verses distance (left) and cumulative number of events versus magnitude (right) of two stages A and B. (Cipolla
et al., 2012).

Table 3 shows the number of events and cumulative seismic moments for all events, for events with
S/N⬎3, and events with the magnitude larger than ⫺1.7 for both stages. As seen in this table, number of
events and events with the S/N⬎3 recorded from stage B, closer to the observation well, is about three
times larger than the event corresponding to stage A. The cumulative seismic moment for events related
to stage A is only slightly larger than the cumulative moment for the stage B, which suggests the
possibility of higher activity rate in stage A. Cumulative number of events vs. magnitude for this two
stages are shown in Fig. 17 (right). The green line seen in this graph is the ⬙magnitude of completeness⬙
equal to ⫺ 1.7. Magnitude of completeness is the magnitude above which all events from all stages have
been detected (Cipolla et al., 2012). Both, number of events and the cumulative seismic moment with
magnitude larger than magnitude of completeness show that the activity rate at stage A is almost 3 times
larger than in stage B. This is an excellent example of possible false interpretation in a comparative
interpretation and shows the importance of proper filtering and normalization of events before any
comparative interpretation of microseismic data.

Table 3—Number of events and cumulative seismic moment for two stages in Fig. 17 (Cipolla et al., 2012)
# of events / Cumulative moment (MN-m) Stage A (Farther from observation) Stage B (Closer to observation)

All events 397 / 1148 1191 / 993


S/N ⬎ 3 397 / 1148 1129 / 981
Magnitude ⬎ ⫺1.7 98 / 683 34 / 263

Microseismic applications can be categorized for use in reservoir characterization and reservoir
development applications. Cipolla et al. (2012) divided the applications of microseismic data for reservoir
development further into three categories: (1) Real time control of fracturing, (2) Completion effective-
ness, and (3) Field development applications. Cipolla et al. (2012) suggested workflows for proper
applications of microseismic data for these three purposes.
Real time monitoring of hydraulic fracturing by microseismic measurements helps with refracturing
and diversion, prevention of potential geohazards, and treatment modification and optimization. Real time
monitoring is mainly based on simple analysis such as data visualization, event patterns, b-value
calculations, event histograms, and SRV extension. As an example, lack of microseismic activity in shale
30 SPE-177277-MS

gas may represent non-producing zones and the requirement in increasing the number of fracking stages.
A basic knowledge about the geology and velocity model is required in real time monitoring. Applications
that are related to completion design and field development require microseismic modeling in addition to
visualization to help with staging, perforation and completion strategy, well placement, drainage pattern
and recovery optimization (Cipolla et al., 2012).
After compensation for possible observation biases, integration of microseismic interpretation with the
geophysical, geological, and engineering data including stress regime, rock properties, seismic data, the
natural fractures network and mass balance (volume of injected of fluid, proppant, and pressure at the end
of treatment) is necessary for evaluating and constraining the interpretation of microseismic measure-
ments (Cipolla et al., 2012).
Proper analyze, integration, and visualization of microseismic data may result in extraction of some or
all of the following interpretations and information (Cipolla et al., 2011a):
● Temporal and spatial evaluation of event patterns,
● Identification of fracture growth using event histogram and events’ count,
● Separation of the hydraulic fractures from fault activation by analyzing magnitude-frequency
relationship,
● Evaluation of the failure orientation using P/SH graphs,
● Interpreting the anomalous behaviors in the events pattern,
● Analysis of aspect ratio versus S/N ratio,
● Analysis of fracture complexity index (FCI) for vertical wells,
● Calculation of SRV,
● Synchronization and integration of microseismic measurements with treatment data and mass
balance.
Fracture mechanics equations and modeling are also beneficial for the estimation of the created volume
for calibration of the interpretation results. More advanced interpretation may be available only if high
quality data from multiple observer wells are available. Cipolla et al. (2012) consider the following items
as the common sources of misinterpretations and misapplications in microseismic monitoring:
● Ignoring the fact the energy associated with microseismic data represents only a small friction of
the total energy of the hydraulic fracturing treatment,
● Non-uniqueness of the solutions to moment tensor inversion,
● Assuming that results from moment tensor inversion necessarily represent the mechanism of
hydraulic fracturing and not the microseismic events,
● Ignoring the complexities in relationship between events source mechanism, deformations, and
fracture modeling.
● Considering data with high uncertainty or with low S/N ratio in SRV calculation and also SRV as
the indicator of well performance, since production from hydraulic fracture reservoirs depend on
the fracture surface area, connectivity and conductivity,
● Apparent complexity due to ignoring the uncertainty in the event location,
● Ignoring mass balance and fracture mechanics, geology, stress regime, and rock properties,
● Misinterpretation of overlaps between adjacent stages or wells.

Uncertainties
The fracture locations determined by microseismic data may not be precise due to limited information on
the velocity model, human mistakes in event selection or lack of reliable algorithms for picking the correct
P- and S- wave arrival times on noisy data. Evaluating and quantifying uncertainties in the determination
of event location which are controlled by many factors such as S/N ratio, observation distance, acquisition
SPE-177277-MS 31

geometry, uncertainty in the velocity model, and the implemented imaging methods, is a challenge in
microseismic monitoring of hydraulic fracturing (Muller, 2013; Thornton and Eisner, 2011; Zimmer,
2011). Among the above parameters proper acquisition geometry and accuracy of the velocity model play
important roles in finding the accurate event location. A good distribution of many sensors around and
close to the fractures reduces the uncertainties about the event location. The location of microseismic
event is determined by calculation of the travel time and the path between the source and observation
points, which is inversely related to the medium velocity.
Velocity model building usually starts with the velocity information extracted from sonic logs. It is
important to consider that sonic logs represent the velocity of high frequency seismic waves in the
direction of well logging, usually vertical. The microseismic waves have lower frequencies and propagate
mostly in horizontal direction previous to arrival at the borehole sensors. There is usually anisotropy
between the vertical and horizontal seismic velocities in sedimentary layers due to bedding, different
lithology, or different compaction. Using geological knowledge about the anisotropy, seismic velocities,
or checkshots and offset vertical seismic profiling (VSP) data may assist to calibrate the sonic log
velocities and account for the anisotropy. Signals created and recorded during well perforation are good
indicators of the average velocity since the perforation time is usually well-known if a clear S-wave is
identified (Cipolla et al., 2012).
Availability of sonic logs from multiple wells in the area, calibration with signals from perforation
shots, and using seismic velocities and velocity inversion methods improve the estimation of the velocity
model. Confidence on the velocity model is increased by integrating it with the geologic model.
Uncertainty in event location due to errors in velocity model is less for deeper sensors (Cipolla et al.,
2011a). This could be considered during microseismic data acquisition planning in the areas with a
complex velocity model to further reduce location uncertainties. Uncertainty in the velocity model may
be evaluated in combination with the uncertainty in the event location. Complex velocity model or
variable anisotropy in velocity model shifts the location of events and result in artifacts in the event
location. Monte Carlo simulation may be used for evaluation of the combined uncertainties (Cipolla et al.,
2011a).
Uncertainty of the event locations may be schematically presented by either a tri-axial ellipsoid, where
the relative axis lengths correspond to the relative uncertainty in the axis direction, a three perpendicular
error bar in a 3D space, or stated by a 3D probability density function. The location uncertainties can be
introduced in SRV calculations by determining different SRVs. Maximum and minimum SRVs are
calculated based on the maximum and minimum relocation of the individual events from the injection well
within a certain confidence level, respectively. The most-likely SRV value corresponds to the most- likely
location of the individual events. Since the location uncertainly is directional, minimum and maximum
SRVs are affected by the direction of the ellipsoid error in addition to the error values (Zimmer, 2011).
In general, surface microseismic data have less uncertainty on the horizontal location and borehole data
have more confidence on the vertical location of the events.
Microseismic signals, similar to seismic waves, lose their strength (amplitude) when propagating
through the earth layers as a result of absorption, geometrical spreading, scattering, and reflection and
transmission through subsurface layers. Depending on the receiver location relative to the shear fracture
plane, radiation pattern also affects the amplitude. Assuming that the travel path and receivers are in the
same layer, the amplitude still decreases with distance from the source due to attenuation and geometrical
spreading. By the time a seismic pulse arrives to the receiver, its attenuated amplitude could be smaller
than the amplitudes of the environmental noise. Unlike surface seismic data, methods to attenuate noise
or split them from signal in microseismic records are very limited.
The common solution is to discard any event, signal or noise, with a magnitude lower than a threshold
value. This threshold value increases with the offset (distance between source and receiver). Conse-
quently, the contribution of far events in SRV calculation, for instance, will be discarded. Due to the
32 SPE-177277-MS

nonlinear relation between amplitude and distance between source and observation point, the minimum
detection limit versus offset is a nonlinear curve as shown in Fig. 18. Minimum S/N ratio of the events
to be considered in the calculation of SRV is about 1.4 (Zimmer, 2011). S/N ratio larger than 2 or 3 may
help to distinguish a real planar fracture network from a biased complex fracture network as seen in Fig.
19, as an example. S/N ratio may be quantified by analyzing the amplitudes when the pumping engine
stopped functioning during operation or before fracturing begin.

Figure 18 —Magnitude versus distance of microseismic events shows the minimum detection limit and maximum detection distance
(Modified from Zimmer, 2011).
SPE-177277-MS 33

Figure 19 —Plan view of same microseismic data with S/N ratio of > 2. 5 (left) and > 5 (right). Left shows a complex pattern of fracturing
while using data with less uncertainty shows a planar fracturing pattern (Modified from Maxwell et al., 2010).

It is important to consider that the fracture propagation may continue after injection due to gradual
changes in stress field or pore pressure after injection. Ignoring weaker events which are farther from the
source is equivalent to putting more weight on closer events. With a single downhole observation, more
events are detected on the observer side of the injection well even if the spread of fracturing is symmetric
around the perforation well (Fig. 20). Appropriate corrections must be performed to distinguish unreal
asymmetry observed on data from real asymmetry of the fracturing area before any further analysis for
SRV calculation or fracture modeling.
34 SPE-177277-MS

Figure 20 —Map view of microseismic events from a single adjacent horizontal well. More events recorded in the side closer to the
monitoring well (at NW) is a distance bias (Modified from Bazan et al., 2010)

In addition to P- and S- wave arrival times, location, and S/N ratio, other attributes that are inherent
in microseismic data, such as magnitude, coherency, P to S wave amplitude ratio, and phase provide better
estimation of uncertainties in the microseismic data (Mueller, 2013). Since environmental noise is mostly
random, statistical methods may become helpful in quantifying uncertainties. However, the number of
events or events magnitude is not sufficient to be considered as factors of stimulation efficiency (Zhang
et al., 2014). Only attributes from microseismic data that are related to rock permeability such as fracture
size and aperture must be accounted while considering stimulation efficiency.

Microseismic for SRV calculations and reservoir simulation


Earlier microseismic data in hydraulic fracturing of typical tight gas sands and conventional resources
mostly demonstrated planar fracture models which were initiated from the injection point perpendicular
to the direction of the minimum horizontal stress (Mayerhofer et al., 2008). Estimation of the fracture
geometry (strike, dip, height, length, and width) of these planar fractures was the main result of early
microseismic measurements. Data from shale reservoirs, in the Barnett Shale in particular, illustrated
multiple plans of fractures and a cloud of complex pattern of microseismic events. Successful enhance-
ment in rate of hydrocarbon production at the Barnet Shale suggested significant increase in the
connectivity of shale reservoir to the production well by the observed complex network of microseismic
events. Consequently, existence of a cloud of complex fracture networks was considered to be a sign of
excellent productivity in tight shale reservoirs. The concept of SRV which is also known as stimulated
volume (SV), estimated stimulated volume (ESV), or stimulated reservoir area (in 2D), was developed as
SPE-177277-MS 35

a quantitative evaluation of the stimulation volume/area in the Barnett Shale and similar formations based
on the spatial extent of microseismic clouds (Fisher et al., 2002; Maxwell et al., 2002; Fisher et al., 2004;
Mayerhofer et al., 2008; Zimmer, 2011; Cipolla and Wallace, 2014). Mayerhofer et al. (2008) considered
SRV as a part of the reservoir volume which is affected by the stimulation regardless of the productivity
of the corresponding fracture network.
Since its introduction to the oil and gas industry, SRV has been used extensively for evaluation of
fracturing performance and production prediction (Mayerhofer et al., 2008; Zimmer, 2011; Cipolla and
Wallace, 2014). The concept of SRV has been used to qualitatively describe the shape, extension, size,
and connectivity of the fracture network to the injection well, drainage area, stage overlaps, and well
production prediction resulting from hydraulic fracturing of reservoirs with low and ultra-low permea-
bility (Nagel et al., 2012; Zimmer, 2011). In principle SRV as a reasonable descriptor of fracturing
performance since productivity of tight reservoirs depends primarily on the extent and permeability of the
fracture network, and the volume of the reservoir that is connected to the production wells. This volume
can be used for optimum well placement by reducing the cost of unnecessary drilling, perforations, and
overstimulation and by avoiding unnecessary overlaps between fracture stages or wells. The general
assumption is that the larger the SRV as a quantity describing the fractured 3D volume size, the higher
is the cumulative production. Even a rough estimate of the reservoir volume stimulated by hydraulic
fracturing will be very valuable for more efficient reservoir management, simulation, and well completion
(Fisher et al., 2004; Zimmer, 2011). However, applications of SRV are mostly qualitative rather than
quantitative (Cipolla et al., 2012).
Work by Fisher et al. (2004) showed a linear correlation between production in Barnett Shale and the
calculated SRVs. Fig. 21 from Fisher et al. (2004) illustrates the correlation between SRV and average
production rate from 7 vertical wells and 11 horizontal wells (5 cemented and 6 un-cemented) in the
Barnet Shale. The coefficient of determination, R2, is 0.6108 which is close to the R2 value achieved by
Zimmer (2011). Zimmer (2011) showed the following linear relationship between SRV and 6 months
average production in 25 hydraulic fracturing jobs the in Barnett Shale:
(31)

where Q is the 6-month average production in Mcf/D and VSRV is the SRV in MMcf. The average-size
SRV volume is in the order of one billion cubic feet. Any specific correlation between SRV and well or
stage productivity is often limited to the geographical region and treatment design (Cipolla et al., 2012).
36 SPE-177277-MS

Figure 21—Correlation between stimulated reservoir volume (SRV) connected by fractures and improved well productivity (Fisher at
al., 2004).

The SRV concept is considered to be applicable mostly in operations which result in a complex fracture
network shown by microseismic events. Such networks are observed mostly in tight shale reservoirs. The
SRV calculated from microseismic measurements at reservoirs with simpler planar hydraulic fractures and
without microseismic cloud, for instance in the Montney and Cardium formations, may not correlate with
production data (Zimmer, 2011).
Calculation of SRV value may be achieved by binning the volume around the well and counting the
number of events at each defined grid. The volumes of bins with a certain minimum number of events are
added to calculate the total SRV. The minimum number of events at each bin dictates if the bin is fractured
enough to be counted as a part of the SRV. Outliers, events that are believed not to be connected to the
main fracture network, and stress induced events must be discarded from SRV calculations (Cipolla et al.,
2012).
There are a few important difficulties with the event selection for SRV calculations including the
threshold magnitude of an event to be considered as indicator of an effective fracture, the minimum
number of events inside each grid bin to count in the calculations, and the location uncertainties. SRV may
be calculated by defining a shrink-wrap volume around microseismic events (Zimmer, 2011). Even using
the shrink-wrap method, uncertainties still exist when defining the SRV boundaries (Zimmer, 2011).
Cipolla et al. (2008) defined the fracture complexity index (FCI) as the ratio of the microseismic image
width to the image length. FCI concept can be used to distinguish the planar fractures from complex
fractures. Cipolla et al. (2008) recommended that fractures with an FCI ⬍ 0.25 should be considered to
be planar fractures. If the fracture network is planar, the SRV concept does not apply (Cipolla et al., 2012).
Extent of the fracture network usually shows the boundary that the injection fluid has reached. Since
shale reservoirs have very low permeability, fluid flow or changes in pore pressure are not enough to
explain the microseismic events in these types of reservoirs. Considering pre-existing permeable natural
fractures or activation of natural fractures as path for fluid flow and microseismic events as the result of
shear failure in the vicinity of activated natural fractures seems to be a reasonable explanation for
microseismic signals emission (Mayerhofer et al., 2008). Lack of a clear distinction between events that
SPE-177277-MS 37

are related to the fluid flow and outlier events that are due to changes in far field stresses, make the
estimated SRV to be significantly larger than real stimulated volume. Other source of overestimation is
consideration of events with high uncertainty and with low S/N ratio (Maxwell et al., 2010).
Contrary to the results shown in Fig. 21, other authors (Zimmer, 2011; Cipolla and Wallace, 2014)
indicate that the SRV should not be considered as a quantitative measure of well performance and
hydraulic fracturing accomplishment. They indicate that well performance must be separated from the
SRV concept as production is controlled by many important factors such as fracture length and spacing
(density), fracture conductivity, reservoir (shale) matrix permeability, stress regime and elastic properties
of the reservoir rocks, and natural fractures that are activated by fluid injection. As a result using only
microseismic data for production estimation from SRV could lead to misleading results. The main source
of error is the fact that the microseismic data only cover a very small portion of hydraulic fracturing
deformations as discussed earlier (Maxwell and Cipolla, 2011; Maxwell et al., 2013). One may add to this
shortcoming, the fact that the microseismic measurements are not able to provide any direct information
about the connection of fractures and their total area and volume or the proppant segregation (Cipolla et
al., 2011a; Cipolla et al., 2012). As mentioned earlier, fractures are generally considered to consist of two
orthogonal sets, one resulted from hydraulic fracturing and the other set made out of activated natural
fractures. Total fracture length, LT, can be calculated with the equation (Mayerhofer et al., 2008),
(32)

where xf is the fracture network half-length, ⌬xs is the fracture spacing, and xn is the fracture width.
Decreasing fracture spacing increases the total fracture length and the cumulative hydrocarbon production.
Numerical simulation shows more increase in gas production rate is achievable if fracture spacing is
decreased by the same rate in larger SRVs (Mayerhofer et al., 2008).
Proper simulation of tight reservoirs requires a good estimation of DFN, hydraulic fractures, and SRV
value. Fracture intensity, complexity, opening, and connectivity to the DFN and injection well must be
coupled to the SRV calculation for a reliable reservoir simulation. Quantitative calibration of modeled
deformation energy with the injected energy and microseismic energy provides information about the
strain release during hydraulic fracturing (Maxwell and Cipolla, 2011). A calibrated model is necessary
for reservoir simulation and improvement of hydraulic future fracturing planning. Hydraulic fracture
modeling requires as a minimum information about the natural fractures network, stress regime, physical
and mechanical properties of reservoir rocks, injection pressure profile and fracture conductivity (Max-
well and Cipolla, 2011). Lack of a detailed model of natural fractures in the formation, DFN, geome-
chanical properties, in-situ stress regime including the intermediate stress component which affects the
general direction of fracture propagation and an all-inclusive fracture modeling software package (Adachi
et al., 2007), makes the hydraulic fracturing modeling to be practically complicated (Cipolla and Wallace,
2014).
Microseismic data facilitate the modeling by finding the location, orientation, and slip of activated
preexisting fractures and stress orientation. However, available techniques for microseismic data acqui-
sition and interpretation cannot estimate and characterize the hydraulic fracturing conductivity using only
microseismic results according to Cipolla and Wallace (2014). They indicate that using SRV for rate
transient analysis (RTA) methods for permeability and fracture length estimation in order to use for
completion optimization is inappropriate since RTA is not capable of considering the fracture propagation
complexity. They further indicate that Microseismic data are not able to answer all questions that
engineering is looking for in a reservoir simulation. Fiber optic distributed temperature sensing (DTS) and
distributed acoustic sensing (DAS) provide valuable real time information about perforation and the area
close to borehole. Integration of microseismic data with the surface and borehole tiltmeters, DTS, DAS
may provide more reliable information on the details of fracturing (Warpinski, 2014). In any hydraulic
38 SPE-177277-MS

fracturing planning, number of perforation points and stages, injection pressure, and proppant size and
volume at each well placement are very important factors to be considered. Probably the current best
practice is extending the analysis of the successful fracture treatment and monitoring to the fracturing
process of similar nearby formations.
The methodologies discussed above are the most rigorous available at the present time as they attempt
to represent the correct physics of hydraulic fracturing, which include tensile hydraulic fracture propa-
gation, fluid leakoff at fracture tips and normal to the fracture planes, and elastic fracture opening and
closing. But it is important not to forget the high levels of uncertainty, also discussed above, associated
with the interpretations.
Use of the Diffusivity Equation
Although not as rigorous as the methodologies presented above another school of thought for microseis-
mic interpretation indicates that the diffusivity equation can be used as a fair approximation for evaluation
and planning of hydraulic fracturing jobs. For example, Shapiro et al. (1997) developed an analytical
method by setting up a diffusion equation of pore pressure in a homogeneous isotropic poroelastic
medium, and used it to describe the development of hydraulic fractures. They assumed that initially the
state of stress in a reservoir was in equilibrium until a small increase of pore pressure caused by fluid
injection to change the effective normal stress, triggering the microseismic events. Once the minimum
pressure increase required to trigger microseismic events at the front of the pressure propagation was
defined, the corresponding front of the microseismic event cloud could be located. The solution to that
equation shows that the hydraulic diffusivity coefficient is related to the distance from the injection point
to the triggering pressure front and the injection time. Thus, it can be determined from the spatiotemporal
distribution of the microseismic-event cloud. This method was used to estimate the hydraulic fracturing
permeability (Grechka et al., 2010, Shapiro and Dinske, 2009).
The idea of using the diffusivity equation has been challenged on the grounds that it does not represent
the correct physics of hydraulic fracturing as described throughout this paper. In practice, however,
solutions to the diffusivity equation with proper initial and boundary conditions can mimic very well the
growth of a microseismic cloud while at the same time allowing an estimate of the hydraulic diffusivity
in the x, y and z directions. As the hydraulic diffusivity is a function of permeability, porosity,
compressibility and viscosity, matching of the microseismic cloud growth can provide valuable informa-
tion for reservoir forecasting.
This has been shown by Yu and Aguilera (2012) who developed a 3D analytical method based on the
concept of pore pressure diffusion mentioned above assuming a homogeneous anisotropic poroelastic
medium. The diffusivity coefficients in 3 dimensions are determined from the analysis to match the
growth of microseismic-events in 3D vs. time. The key assumption is that the microseismic event cloud
and its growth are a fair representation of the SRV growth with time. This assumption has been debated
for some time within different professional societies of the oil and gas industry. The advantage of the 3D
analytical model is simplicity as it can be reproduced easily and rapidly in a spread sheet. And nothing
precludes calibrating it with results of the more rigorous physics-based methods described previously in
this paper. Case study 4 near the end of the paper illustrates the use of this method with microseismic data
collected in the Marcellus shale.
The diffusivity equation has also been used for building seismicity based reservoir characterization
(SBRC) models when combined with Mohr-Coulomb failure criteria (Hosseini and Aminzadeh, 2013) and
Mogi empirical rock failure relationship (Qi et al., 2015).
Respond to the public concerns about hydraulic fracturing
As the employment of hydraulic fracturing methods has expanded, a growing number of questions and
concerns regarding the safety and environmental hazards associated with hydraulic fracturing has been
SPE-177277-MS 39

generated. Major public concerns are the possible contamination of the underground drinking water
aquifers with the fracking fluid or hydrocarbon, fracking-related earthquakes, using fresh water resources
for fracking, waste water and drilling fluid management, and the effect of fracking fluid composition and
gas emissions on human body and the surrounding environment. Microseismic measurements provide
very exclusive information to address the public concerns on relationship between earthquakes and
fracturing and the possible contamination with the drinking water aquifers. Microseismic monitoring may
also prevent the environmental hazards by real time imaging of the fracture growth.
Fisher (2010) and Fisher and Warpinski (2012) attempted to respond to the concern of the connectivity
of hydraulic fractures to the fresh water aquifers in tight reservoirs using microseismic and tiltmeters data.
They gathered fracking data from thousands of fracturing stages and data from water wells in Barnett,
Marcellus, Eagle Ford, and Woodford, and plotted the maximum water depth and recorded fracturing top
and bottom at each location (Fig. 22 for Barnett Shale). Their plots showed that in 22 fracturing projects
in the Barnet Shale, for instance, the maximum upward growth of the fractures from perforation top was
around 2000 ft which was about 4000 ft deeper than the deepest water well level at the same location
(Fisher and Warpinski, 2012). Similar conclusion obtained by plotting the data from the Marcellus Shale
and other reservoirs. The result of their study is summarized in Table 4 (King, 2012). As Fisher (2010)
mentioned, there is not a single documented report on the pollution of fresh water aquifer due to hydraulic
fracturing from more than two million fracturing treatment that has been employed between 1949 and
2010 in the United States. It is commonly accepted idea that the hydraulic fracturing does not provide a
pathway to the shallow water aquifers unless for very shallow hydrocarbon reservoirs. However, a safe
practice and an intelligent recommendation would be to avoid hydraulic fracturing of very shallow
reservoirs and in the faulted regions.

Figure 22—Water/aquifer depths and the measured fracture heights in Barnett Shale fracturing treatments. (Fisher and Warpinski,
2012).
40 SPE-177277-MS

Table 4 —Comparisons between fracture height growth and water depth in 4 major shale reservoirs (King, 2012). Depths are in feet.
Number of Depth range of Maximum Average distance between fracture Minimum distance between fracture
Reservoir stages studied fracking zone water depth top and deepest water level top and deepest water level

Barnett 3000 ⫹ 4700–8000 1200 4800 2800


Eagle Ford 300 ⫹ 8000 – 13000 400 7000 6000
Marcellus 300 ⫹ 5000–8500 1000 3800 3800
Woodford 200 ⫹ 4400 – 10000 600 7500 4000

Another important public concern about hydraulic fracturing is the occurrence of earthquake activities
and induced seismicity which might be related to the hydraulic fracturing and fluid injection. Induced
seismicity refers to the seismic events which result from human activities. The main reasons for induced
seismicity in hydrocarbon reservoirs are changes in pore pressure and the stress fields around faults, or
in the formations with rocks under critical stress conditions. Alteration of pore pressure due to fluid
injection or extraction changes the stress field and causes the slip of faults which were initially under
stable conditions. Seismicity occurs if the slipping is fast and a large amount of energy is released in a
short period of time. A shallow seismic event must have a magnitude of about 2.0 or larger to be felt by
a human being.
About 14,450 earthquakes with magnitudes larger than 4.0 are recorded globally every year (National
Research Council, 2013). Only a very small portion of these events might be related to human activities
including geothermal exploitation, nuclear tests, hydrocarbon production, waste liquid disposal into the
underground formations, surface water reservoirs and dams, and mining activities. Probably the magni-
tude 5–5.5 Richter earthquakes in 1960s in the Rocky Mountain Arsenal in Colorado, USA, which
occurred due to the waste liquid injection into the impermeable basement rocks were one of the earliest
documented induced seismicity phenomena in the USA which was related to fluid injection.
As seen in Fig. 23, there is a clear direct relation between the amounts of injection with the occurrence
frequency of seismicity events in the Rocky Mountain Arsenal area. Increasing the water injection raises
the frequency of seismic events. However, there is about one month delay between the time where
injection volume peaks and the peak in the number of seismic activities. Another notable feature is the
occurrence of microseismic activities when injection stopped between October 1963 and August 1964.
These phenomena are very similar to the microseismicity during a hydraulic fracturing program. A
research by the National Research Council (2013) showed that 43% of 156 documented induced
seismicity events around the world were due to oil and gas extraction, secondary recovery methods, or
hydraulic fracturing. Probably due to more oil and gas activities in the USA, this number raises to about
64% in this country. However, National Research Council’s (2013) researchers showed that there are only
two induced seismicity events in the world that could be related to the hydraulic fracturing treatments, one
in the Eola field, Oklahoma, USA, with the magnitude of 2.8 (Holland, 2011) and the other one at the
Blackpool, UK, with the magnitude about ML ⫽ 2.3 (de Pater and Baisch, 2011).
SPE-177277-MS 41

Figure 23—Relation between volume of waste injection (top) and the frequency of earthquakes (bottom) in Rocky Mountain Arsenal well
between 1962 and 1965 (National Research council, 2013).

Another notable parameter is the very low magnitude of induced seismicity events. A comprehensive
study by the National Research Council (2013) showed that the maximum magnitude of felt induced
seismicity in 35,000 hydraulic fracturing wells in the USA was about 2.8 (Holland, 2011) while there were
7 events with magnitude larger than 4.0 during waste water injection in 30,000 wells. This number is 3
for 108,000 water-flooding wells in secondary recovery projects. Earthquakes with magnitudes less than
2 are related to the faulting events with less than 1 cm total slippage (Fig. 24). Events with magnitude less
than 3 Richter are considered to have the intensity of I in the modified Mercalli scale, meaning these
events are not able to create any damage to man-made constructions. Generally, there is no damage to the
human life by earthquakes with intensity of less than III. The biggest microseismic magnitude recorded
in the Barnet Shale fracturing is ⫺1. The energy that is released during this event is equal to the explosion
of 30 mg explosive charge. This is extremely small when compared to the energy of very weak
earthquakes. The amount of energy release for a weak earthquake with the moment magnitude of 3, is
equivalent to explosion 30,000 kg of explosive charge.
42 SPE-177277-MS

Figure 24 —The relation between the earthquake magnitude and fault and slippage size (Zoback and Gorelick, 2012).

Since the amount of fluid being injected during a hydraulic fracturing treatment is relatively small and
the induced pore pressure is below the minimum in-situ stress, the seismicity events related to hydraulic
fracturing are weak and considered as microseismic events rather than as earthquake events (National
Research Council, 2013). From any perspective, based on the frequency of induced seismicity events
occurrence and the events’ magnitude, the available statistical information shows that hydraulic fracturing
is generally a safe treatment for oil and gas exploitation (Fig. 25).
SPE-177277-MS 43

Figure 25—Maximum magnitudes of the documented induced seismicity for different oil and gas activities, wastewater disposal and
geothermal activities. Events emanating from hydraulic fracturing activities are relatively rare and weak (National Research council,
2013).

Case study 1, Montney formation


The study by Maxwell and Norton (2012) on hydraulic fracturing of the Montney formation is an excellent
example of the application and integrated interpretation of microseismic measurements for enhancing
shale gas reservoir characterization and evaluation of a hydraulic fracturing treatment. The Triassic
Montney formation is located in NE of British Columbia in Canada (Fig. 1). The lithology of Montney
begins with the shallow-water shoreface sand to the east and ends with deep water mudstone in the west.
Conventional gas has been extracted from the sandy part of the Montney for a long time. There are two
hybrid shale gas potentials in sandy silty shale in Lower Montney and tight sand in Upper Montney to be
exploited using unconventional methods. Montney is estimated to have between 80 to 700 tcf of gas in
place (BC Ministry of Energy, Mine and Petroleum Resources, 2006).
Maxwell and Norton (2012) showed the enhancement of shale gas reservoir characterization using
microseismic data in the Upper Montney formation. The Montney formation has between 3% and 6%
porosity and 0.001 to 0.05 mD permeability in the area of the study. Fracturing is performed at
approximately 1750 m TVD with 6 to 8 stages in a 1500 m horizontal well. Injection rate was 10 m3/min
and microseismic data was recorded with eight 3-component borehole geophones. Maximum offset was
about 1600 m.
Fig. 26 shows a map view of the recorded microseismic events from all stages in the wells A, B, and
C, where the radius of each circle represents the events’ magnitude. As seen in Fig. 26, there is a large
number of events near wells C and B with a general trend toward the SW, and lack of almost any event
in the NE. This could be a detection bias. Well A shows planar events that are different from events in
wells B and C. In order to find an acceptable explanation, Maxwell and Norton (2012) performed an
integrated study on this project. Using surface seismic data and AVO inversion, Norton et al. (2010)
created a map of Poisson’s ratio to find any changes in brittleness of the formation. Further analysis and
44 SPE-177277-MS

data include attenuation of detection-distance bias by eliminating the events with magnitudes lower than
-2.2, mapping the seismic moment density, calculating the b value to distinguish the hydraulic fractures
from activated faults, computation of composite focal mechanisms for estimating the fracturing mecha-
nism, and evaluation of pressure and production measurements (Fig. 27).

Figure 26 —Microseismic events from all stages in Montney hydraulic fracturing jobs; Left: All events, Right events with magnitude
larger than ⴚ2.2. Events are scaled by magnitude (Maxwell and Norton, 2012).
SPE-177277-MS 45

Figure 27—Left: b-value map in the background of microseismic events, right: b-value diagram. b-value is high around well A
suggesting the hydraulic fracturing in this region and low around wells B and C suggesting activation of preexisting natural fractures
(Maxwell and Norton, 2012).

The integrated study enabled Maxwell and Norton (2012) to evaluate the characteristic of the fracturing
process, microseismicity, and well performance. They concluded that the larger events seen near wells B
and C are the result of activation of pre-existing natural fractures which are mostly in the NW-SE
direction. The detected pre-existing fractures were small and below the resolution detectability of the
surface seismic data. Events closer to well A are the result of a planar hydraulic fracture with a NE-SW
direction and orthogonal to the minimum stress direction in the study area (Maxwell and Norton, 2012).
It is also concluded that the lack of microseismic events in the NE is not a detection-distance bias; rather
it is due to lack of fracturing in that area which is the result of interaction with existing faults as they act
as barriers for further propagation of hydraulic fractures. Analysis of production data suggested interfer-
ence between well A and well B that is also seen by the overlap of microseismic data in that region. The
difference between the fracture geometry near well A and wells B and C is due to the presence of natural
fractures around wells B and C.
Case study 2, Barnett Shale
The Mississippian Barnett Shale which is located in the Fort Worth Basin, Texas, is probably the largest
producible natural gas field in the United Sates (Fig. 1). Rapid growth of gas production from the Barnett
Shale begun in 1997 with hydraulic fracturing that utilized slick-water as fracturing fluid. Reservoir rock
consists of organic-rich shale and mudstone with high silica and low clay content. Reservoir depth varies
between 1200 m in the west and 2600 m in the east. Thickness varies between 200 and 500 m.
Vermylen and Zoback (2011) performed an integrated study for evaluation of hydraulic fracturing,
microseismic data, and stress field in the Barnett Shale. Five horizontal wells (named well A to well E)
were horizontally drilled and hydraulically fractured with 51 fracturing stages. In the area of study,
thickness of the Barnett formation is about 100 m and depth to top of the reservoir is 1700 m. Wells are
about 900 m long and the distance between wells is in the order of 150 m.
Only well C was logged including FMI and sonic logs. Fracturing in the cased holes was done by 6
perforations, 15.2 m spaced. Fluid and sand as proppant were injected at rates of 50 — 60 bpm. About
400,000 pounds of water and sand was injected in 150 minutes for each stage.
46 SPE-177277-MS

Wells A and B were fractured using the Simulfrac method. Then wells D and E were fractured using
the Zipperfrac method and well C was fractured using the conventional method. Well C was used for
monitoring of Simulfrac and Zipperfrac stimulations and well B was used for monitoring the treatment in
well C. A total of 4485 events were detected. Fig. 28 shows all recorded events. Data with magnitude
larger than ⫺2.5 is shown in Fig. 29.

Figure 28 —Microseismic events in Barnett Shale for all stages in wells A to E (Vermylen and Zoback, 2011).
SPE-177277-MS 47

Figure 29 —Microseismic events in Barnett Shale for all stages in wells A to E, only events with magnitude larger than ⴚ2.5 are shown
(Vermylen, 2011).

Vermylen (2011) determined the following geomechanical properties for the study area: Pi ⫽ 0.48
psi/ft (Pi is the initial pore pressure), ␴v ⫽ 1.1 psi/ft, ␴H ⫽ 0.73 psi/ft, and ␴h ⫽ 0.63 psi/ft. It is notable
that the maximum horizontal stress is only 12% higher than the minimum horizontal stress. Vermylen and
Zoback (2011) observed that fracturing was more successful in the final stages than in the initial stages.
They used microseismic magnitude scaling to compare the successfulness of different fracturing geom-
etries independent of the monitoring geometry.
High b-values of data show that events are mostly related to hydraulic fracturing and not the result of
natural earthquakes or fault activation, which are generally characterized by b values close to 1. Fig. 30
shows the commutative Gutenberg-Richter plot for wells A to B fractures. The curves show the changes
between activity levels for different stages. Earlier stages which are closer to the well toe show lower
activity level than the later stages which are closer to hill. Vermylen and Zoback (2011) concluded that
Simulfrac method had poorer performance than the conventional or Zipperfrac methods.
48 SPE-177277-MS

Figure 30 —Cumulative Gutenberg-Richter plot for wells A_B tracking data shows a wide range of variability between stages (Vermylen
and Zoback, 2011).

In order to explain the differences in efficiency of fracturing stages and methods, Vermylen and Zoback
(2011) analyzed the changes in instantaneous shut-in pressure (ISIP) values for each well. ISIP data are
used to determine the magnitude of the least principal stress. Fig. 31 shows the measured ISIPs for all
wells. There is a general increase in ISIPs with increasing distance from well toe. However, the last stages
in wells A, B, and C show a decrease in ISIP which is against the general trend. Vermylen and Zoback
(2011) considered this to be a stress shadow and the product of sequential propped planar fractures. By
plotting the microseismic activity level versus ISIP for each stage, they found two different correlation
trends for wells A-B and wells D-E. The Simulfrac wells show higher ISIP but lower microseismic
activity than the Zipperfrac wells. Therefore, it seems that the fracturing method has a significant effect
on microseismicity and fracturing (Vermylen and Zoback, 2011). Using the mentioned analysis and
investigating the effect of poroelasticity, Vermylen and Zoback (2011) were not able to analytically
explain the dependency relation between stress, fluid pressure, and stimulation, and suggested the
requirement for coupling geomechanics and fluid flow for evaluation of stress changes and propagation
of the permeable fracture network.
SPE-177277-MS 49

Figure 31—ISIPs vs. distance from well toe for all wells shows different trends for Zipperfrac and Simulfrac (Vermylen and Zoback,
2011).

Case study 3, Integrated study for shale gas characterization


Alzate and Devegowda (2013) performed an integrated study using 3D surface seismic, microseismic, and
production data including temperature, pressure, deviation survey, and multi-spinner flow-meter logs from
four wells (Well A to Well D) on the Newark East Field in the Barnett Shale formation. Production from
unconventional organic-rich shale reservoirs are related to reservoir quality including porosity and total
organic carbon, the extent and connectivity of fracture propagation and the time that fractures remain open
after fracking (Cipolla et al., 2011b).
Alzate and Devegowda (2013) utilized seismic inversion methods to extrapolate the petrophysical
characterizations at the well locations to the whole area of study which was covered by 3D seismic. They
further created P- and S- impedance volumes and density. Using these data, Alzate and Devegowda (2013)
managed to calculate the ␭ and ␮, and based on the work by Refunjol et al. (2012) utilized the ␭␳/⫽␮␳
crossplots to divide the Barnett Shale at the Newark East Field into four different rock types.
Fig. 32 shows a typical ␭␳/⫽␮␳ cross-plot for rocks with different ratios of quartz/clay content and
porosity range between 0% and 20%. It is notable from Fig. 32 that increasing the quartz content enhances
the brittleness of the rock and recovery factor (RF), but decreases the original gas in place (OGIP).
Recovery factor is a function of porosity (and richness of the rocks) and rock brittleness. Fractures in
ductile rocks (with more clay content) tend to close faster. Therefore, ductile shales have lower recovery
factors.
50 SPE-177277-MS

Figure 32—The ␮␳ versus ␭␳ cross-plot for different compositions of quartz and clay and porosity and relation to the brittleness (Alzate
and Devegowda, 2013).

After analyzing Young modulus and Poisson’s ratio in the ␭␳/␮␳ crossplots, Alzate and Devegowda
(2013) defined four rock types in the area of study: Group 0 for brittle and rich rocks, Group 1 for rich
and ductile rocks, Group 2 for brittle and poor rocks, and Group 3 for ductile and poor rocks as seen in
Fig. 33. Adding the gas production data from perforation points into the aforementioned cross-plot
revealed that the perforations in Group 0 (brittle and rich rocks) leads to the highest production rate as seen
in Fig. 34. Perforations in Group 0 belong to well B which has shown the maximum production rate in
the study area. Lowest production rate corresponds to Group 3 (ductile and poor rocks) which is mostly
related to perforations in well D. Fig. 34 clearly shows the fracture productivity and the geomechanical
properties of the rocks in the volume surrounding each perforation point (Alzate and Devegowda, 2013).
SPE-177277-MS 51

Figure 33—Classification of reservoir rocks at Lower Barnett Shale. Four groups are identified based on the brittleness and TOC (Alzate
and Devegowda, 2013).
52 SPE-177277-MS

Figure 34 —Gas production rate at each perforation in the rock type plot shown in Fig. 33 (Alzate and Devegowda, 2013).

Using the microseismic measurements Alzate and Devegowda (2013) analyzed the SRV and patterns
of fracture growth for all wells and observed that fracture events are mostly clustered in the brittle rock
groups. Adding the microseismic events into the ␭␳/␮␳ crossplots shows that fracturing is dominant at the
locations identified previously as brittle zones (Group 0 and Group 2). For instance more than 70% of
microseismic events of well B perforation are relevant to the brittle areas as seen in Fig. 35. The study
shows an excellent correlation between classifications of rock types based on inversion of surface seismic
data, production data, and microseismic events. The created cross-plot might be used for future fracking
treatment design in this area or modified for fracking in similar shale plays.
SPE-177277-MS 53

Figure 35—The ␮␳ versus ␭␳ cross-plot showing microseismic events for well B. About 70% of events are related to hydraulic fracturing
in the brittle areas. (Alzate and Devegowda, 2013).

Case study 4, Analytical 3D analysis of microseismic cloud in Marcellus shale


This case has been described by Yu and Aguilera (2012). It considers a vertical hydraulic fracture created in
a horizontal gas well drilled in the Marcellus shale. The microseismic monitoring data for hydraulic fracturing
stage 5 was published by Hulsey et al. (2010). However there were 6 hydraulic fracturing stages. The well is
located in the Pittsburgh Low Plateau of the Appalachian Basin with its lateral section along the top of the
Marcellus Shale formation. The hydraulic fluid is injected from the perforation point at the horizontal well and
the SRV grows into a 3D space with no apparent confinement; so a 3D analytical model is used for the analysis.
A Cartesian coordinate system is set up to visualize matching of the microseismic cloud as a function
of time. Since this is a vertical fracture, z axis is made parallel to the direction in which the microseismic
cloud grows the fastest vertically. In this case, the hydraulic fracturing has a strike at N45°W and a dip
of 45° towards southeast. The x axis is set parallel to the direction in which SRV grows the fastest
horizontally. In this case, the injection pressure is estimated at about 45MPa based on knowledge of the
surface injection pressure and the fluid injection rate. Finally, the 3D model is built as shown in Fig. 36
(A), (B), and (C). Fig. 36 (A) shows a x-y map view of actual events and calculated SRV growth. Fig.
36 (B) shows a cross-sectional z-x view. Fig. 36 (C) shows a cross-sectional z-y view. It is noticed that
a large SRV is developed within a short period of time (30 min). This is caused by natural fractures in
the reservoir, which allow the injected fluid to open the natural fractures rather easily. For details on the
quantitative 3D analytical methodology the reader is referred to Yu and Aguilera (2012).
54 SPE-177277-MS

Figure 36 —Microseismic events (dots) and SRV presented by the 3D model (continuous curves). (A) map view (zⴝ0m). (B) side view
(yⴝ0m). (C) side view (xⴝ0). (Yu and Aguilera, 2012. Microseismic data from Hulsey et al., 2010).
SPE-177277-MS 55

Conclusions
Hydraulic fracturing boosted hydrocarbon production from tight reservoirs and will probably remain the
best economical method for exploitation of unconventional resources for many years. Hydraulic fracturing
increases connectivity of low and ultra-low permeability reservoirs to the production well by tensile
fracturing and activation of pre-existing natural fractures. Mechanisms of hydraulic fracturing and related
microseismic events are very complicated. Many, mostly unknown, parameters and variables that control
the behavior of hydraulic fracturing cause fracture modeling to be very complicated. However, micro-
seismic monitoring provides valuable information on the geometry, effectiveness, and simulation of
hydraulic fracturing as well as the network of natural fractures. Microseismic events recorded usually with
3-compoenet geophones in the observation boreholes show the spatial extension of fracture propagation.
Advanced analysis of microseismic measurements, including processing and inversion, provide infor-
mation about the fracturing mechanism and a baseline for creating more reliable DFN for proper reservoir
simulation, economical well placements and fracture treatments. A comprehensive analysis of velocity
model and event location uncertainties is recommended before any interpretation of data. Events must be
corrected for any bias from recording distance, azimuth, and observation wells. Integration with the local
geology, stress regime, mechanical properties of reservoir rocks, seismic data, fluids, treatment data, and
mass balance increases the reliability of the interpretations and the definition of DFN for reservoir
simulation and well planning.
The diffusivity equation has been used to develop an easy-to-reproduce analytical model for analyzing
the orientation and geometry of hydraulic fracturing SRV in 3D. A spread sheet suffices for the analysis
that allows matching the microseismic cloud growth with time in 3D. The key assumption is that the
microseismic cloud represents the SRV. This has been debated for some time within different professional
organizations of the oil industry.

Reading suggestions
The following literature is recommended for further study of microseismic methods for monitoring
hydraulic fracturing. Most of these materials have been used in this manuscript.
● Hydraulic fracturing:
National Research council, 2013; King, 2012.
● Seismic and earthquake seismology:
Stein and Wysession, 2003; Sheriff and Geldart, 1995; Gibowicz and Kijko, 1994.
● Microseismic monitoring:
Maxwell, 2014; Cipolla and Wallace, 2014; Akram, 2014; Warpinski et al., 2013; Cipolla et al.,
2012; Cipolla et al., 2011a; Zimmer, 2011.
● Geomechanics and fracture modeling:
Zhuang et al., 2014; Vermylen, 2011; Adachi et al., 2007; Jaeger et al., 2007; Zoback, 2007;
Griffith, 1920.
● Poroelasticity:
Wang, 2000; Biot, 1941.

Acknowledgement
The support CNOOC Limited and Nexen, the Schulich School of Engineering at the University of Calgary
and Servipetrol Ltd. of Calgary, Canada is gratefully acknowledged. We also thank the GFREE research
team [GFREE refers to an integrated research program including Geoscience (G); Formation Evaluation
(F); Reservoir Drilling, Completion, and Stimulation (R); Reservoir Engineering (RE); and Economics
and Externalities (EE)] at the University of Calgary for their continued help and support.
56 SPE-177277-MS

Nomenclature
°C ⫽ Degree Celsius
F ⫽ Force
A ⫽ Unit surface area
␴ ⫽ Stress
␴ij ⫽ Stress components
␴i, i⫽ 1,2,3 ⫽ Principal stresses
␴V ⫽ Vertical stress
␴h ⫽ Minimum horizontal stress
␴H ⫽ Maximum horizontal stress
ij ⫽ Effective stress components
i, i⫽ 1,2,3 ⫽ Normal effective stresses
␴n ⫽ Normal stress on an arbitrary plane
␶ ⫽ Shear stress on an arbitrary plane
␴m ⫽ Mean normal stress
␶m ⫽ Maximum shear stress
␧ ⫽ Strain
␧ij ⫽ Strain components
u, v, w ⫽ Displacement in x,y, z directions
C ⫽ Stiffness matrix
␭ ⫽ Lamé first elastic constants or fluid incompressibility
␮ ⫽ Second elastic constant or shear modulus
v ⫽ Poisson’s ratio
E ⫽ Young’s modulus
P ⫽ Pore fluid pressure
␣ ⫽ Biot-Willis coefficient
K ⫽ Effective bulk modulus
Kg ⫽ Grains bulk modulus
S0 ⫽ Cohesive strength, cohesion, or unconfined compressive strength (USC)
␮i ⫽ Internal friction coefficient
␾ ⫽ Angle of internal friction
Pf ⫽ Net pressure in the fracture
Pt ⫽ Bottomhole pressure
␴min ⫽ Minimum horizontal (in situ) stress
␴max ⫽ Maximum horizontal (in situ) stress
ML ⫽ Magnitude in Richter scale
A ⫽ Largest amplitude recorded by Wood-Anderson seismograph
⌬ ⫽ Distance between earthquake and seismograph
Mo ⫽ Seismic moment
D ⫽ Average fault slip
A ⫽ Total slipped area of the fault surface
␳ ⫽ Density
R ⫽ Offset between source and sensor
C ⫽ Wave speed
␻ ⫽ Low velocity level of displacement for P-waves
F ⫽ Radiation pattern factor
MW ⫽ Moment magnitude
SPE-177277-MS 57

M ⫽ Microseismic moment tensor


Mij ⫽ Microseismic moment tensor components
TP ⫽ P-wave arrival time
Ts ⫽ S-wave arrival time
⌬TPS ⫽ Difference between the P- and S- waves arrival times
x ⫽ Distance between the event and receiver
P ⫽ Average speed of P-wave
s ⫽ Average speed of S- waves
Q ⫽ Average production
VSRV ⫽ SRV volume
LT ⫽ Total fracture length
xf ⫽ Fracture network half-length
⌬xs ⫽ Fracture spacing
xn ⫽ Fracture width

Acronyms
1C ⫽ One Component
2D ⫽ Two Dimensional
3C ⫽ Three Component
3D ⫽ Three Dimensional
AVO ⫽ Amplitude Verses Offset
CLVD ⫽ Compensated Linear Vector Dipole
DAS ⫽ Distributed Acoustic Sensing
DC ⫽ Double Couple
DFN ⫽ Discrete Fracture Network
DST ⫽ Distributed Temperature Sensing
ESV ⫽ Estimated Stimulated Volume
FCI ⫽ Fracture Complexity Index
GJ ⫽ Gigajoules
ISIP ⫽ Instantaneous Shut-In Pressure
LTA ⫽ Long-Term Average
P- ⫽ Primary (Wave)
PEV ⫽ Peak Eigen Value
PSD ⫽ Power Spectral Density
RTA ⫽ Rate Transient Analysis
S/N ⫽ Signal to Noise
S- ⫽ Secondary (Wave)
SH ⫽ Horizontal Secondary Wave
SRV ⫽ Stimulated Reservoir Volume
STA ⫽ Short-Term Average
SV ⫽ Stimulated Volume
TOC ⫽ Total Organic Carbon
TVD ⫽ True Vertical Depth
SC ⫽ Unconfined Compressive Strength
USC ⫽ Unconfined Compressive Strength
VSP ⫽ Vertical Seismic Profiling
58 SPE-177277-MS

References
Adachi, J., Siebrits, E., Peirce, A., and Desroches, J. 2007. Computer simulation of hydraulic fractures.
International Journal of Rock Mechanics and Mining Sciences 44 (5): 739 –757. 10.1016/
j.ijrmms.2006.11.006.
Aguilera, R. 1980. Naturally fractured reservoirs, First Edition. Tulsa, Oklahoma, USA: PennWell
Books.
Aguilera, R. 2003. Geologic and engineering aspects of naturally fractures reservoirs. Canadian
Society of Exploration Geophysicists, Recorder. February, 2003: 44 –49.
Aki, K. 1987. Magnitude-frequency relation for small earthquakes: A clue to the origin of f-max of
large earthquakes. Journal of Geophysical Research. 92 (B2): 1349 –1355. 10.1029/
JB092iB02p01349.
Akram, J. 2014. Downhole microseismic monitoring: processing, algorithms and error analysis. PhD
thesis dissertation, University of Calgary, Calgary, Alberta, Canada.
Albright, J. N. and Pearson, C. F. 1982. Acoustic emission as a tool for hydraulic fracture location:
Experience at the Fenton Hill hot dry rock site. SPE J. 22: 523–530. SPE-9509-PA. 10.2118/
9509-PA.
Alzate, J. H. and Devegowda, D. 2013. Integration of surface seismic, microseismic, and production
logs for shale gas characterization: methodology and field application. Society of Exploration
Geophysicists, Interpretation 1 (2): SB37-SB49. 10.1190/INT-2013– 0025.1.
Bazan, L. W., Larkin, S. D., Lattibeaudiere, M. G., and Palisch, T. T. 2010. Improving production in
the Eagle Ford shale with fracture modeling, increased fracture conductivity, and optimized stage
and cluster spacing along the horizontal wellbore. Presented at the SPE Tight Gas Completions
Conference, San Antonio, Texas, USA, 2–3 November. SPE-138425-MS. 10.2118/138425-MS.
Biot, M. A. 1941. General theory of three dimensional consolidation. Journal of Applied Physics, 12:
155. 10.1063/L1712886.
British Columbia Ministry of Energy, Mines, and Petroleum Resources. 2006, Regional ⬙Shale Gas⬙
Potential of the Triassic Doig and Montney Formations, Northeastern British Columbia. Petroleum
Geology Open File 2006 – 02.
Busetti, S., Wenjie, J., and Ze’ev, R. 2014. Geomechanics of hydraulic fracturing microseismicity:
Part 1. Shear, hybrid, and tensile events. AAPG Bulletin 98, (11): 2439 –2457. http://dx.doi.org/
10.1306/05141413123.
Cipolla, C. L. and Wallace, J. 2014. Stimulated Reservoir Volume: A Misapplied Concept? Presented
at the SPE Hydraulic Fracturing Technology Conference, The Woodlands, Texas, USA, 4 – 6
February. SPE-168596-MS. 10.2118/168596-MS.
Cipolla, C. L., Lewis, R. E., Maxwell, S. C., and Mack, M. G. 2011b. Appraising unconventional
resource plays: Separating reservoir quality from completion effectiveness. Presented at the
International Petroleum Technology Conference, Bangkok, Thailand, 15–17 November. IPTC-
14677-MS. 10.2523/14677-MS.
Cipolla, C. L., Maxwell, S. C., and Mack, M. G. 2012. Engineering guide to the application of
microseismic interpretations. Presented at the SPE Hydraulic Fracturing Technology Conference,
The Woodlands, Texas, USA, 6 – 8 February. SPE-152165-MS. 10.2118/152165-MS.
Cipolla, C. L., Maxwell, S. C., Mack, M. G., and Downie, R. 2011a. A practical guide to interpreting
microseismic measurements. Presented at the SPE North American Unconventional Gas Confer-
ence and Exhibition, Woodland, Texas, USA, June 14 –16. SPE-144067-MS. 10.2118/144067-
MS.
SPE-177277-MS 59

Cipolla, C. L., Warpinski, N. R., Mayerhofer, M. J., and Lolon, E. P. 2008. The Relationship between
fracture complexity, reservoir properties, and fracture treatment design. Presented at the SPE
Annual Technical Conference and Exhibition, Denver, Colorado, USA, September 21–24. SPE-
115769-PA. 10.2118/115769-PA.
de Pater, C. J. and Baisch, S. 2011. Geomechanical study of Bowland shale seismicity. Cuadrilla
Limited open report.
Detournay, E., Cheng, A. H. D., Roegiers, J. C., and McLennan, J. D. 1989. Poroelasticity consid-
erations in in situ stress determination by hydraulic fracturing. Int. J. Rock Mech. Min. Sci. &
Geomech. Abstr. 26 (6): 507–513. 10.1016/0148 –9062(89)91428 –9.
Duncan, P. M. 2010. Microseismic monitoring - Technology state of play. Presented at the SPE
Unconventional Gas Conference, Pittsburgh, Pennsylvania, USA, 23–25 February. SPE-131777-
MS. 10.2118/131777-MS.
Duncan, P. M. and Eisner, L. 2010. Reservoir characterization using surface microseismic monitor-
ing. Geophysics, 75 (5): 75A139 –75A146. 10.1190/1.3467760.
Dyer, B. C., Schanz, U., Ladner, F., Haring, M. O., and Spillman, T. 2008. Microseismic imaging of
a geothermal reservoir stimulation. The Leading Edge 27 (6): 856 –869. 10.n90A.2954024.
EIA. 2011. North American shale plays map. US Energy Information Administration, 9 May 2011,
Web. 1 April 2015, ⬍ http://www.eia.gov/pub/oil gas/natural gas/analysis publications/maps/
maps.htm⬎.
EIA. 2015. ⬙Shale Gas Production⬙. Web. 13 August 2015, ⬍ http://www.eia.gov/dnav/ng/
ng_prod_shalegas_s1_a.htm⬎.
Eisner, L. 2014. Microseismic monitoring in oil and gas reservoir. Canadian Society of Exploration
Geophysicists, Doodle Train course notes, Calgary, Canada, November 3–4.
Eisner, L., Williams_Stroud, S., Hill, A., Duncan P., and Thornton, M. 2010. Beyond the dots in the
box: Microseismicity constrained fracture models for reservoir simulation. The Leading Edge, 29
(3): 326333. 10.1190/L3353730.
Fisher, M. K. 2010. Data confirm safety of well fracturing. American Oil and Gas Reporters, July
2010.
Fisher, M. K. and Warpinski, N. R. 2012. Hydraulic-fracture-height growth: Real data. SPE Produc-
tion and Operations. 27 (1): 8 –16. SPE-145949-PA. 10.2118/145949-PA.
Fisher, M. K., Heinze, J. R., Harris, C. D., Davidson, B. M., Wright, C. A., and Dunn, K. P. 2004.
Optimizing horizontal completion techniques in the Barnett Shale using microseismic fracture
mapping. Presented at the SPE Annual Technical Conference and Exhibition, Houston, Texas,
USA, September 26 –29. SPE-90051-MS. 10.2118/90051-MS.
Fisher, M. K., Wright, C. A., Davidson, B. M., Goodwin, A. K., Fielder, E. O., Buckler, W. S., and
Steinberger, N. P. 2002. Integrating fracture mapping technologies to optimize stimulations in
Barnett Shale. Presented at the SPE Annual Technical Conference and Exhibition, San Antonio,
Texas, USA, September 29-October 2. SPE-77441-MS. 10.2118/77441-MS.
Fredd, C. N., McConnell, S. B., Boney, C. L., and England, K. W. 2001. Experimental study of
fracture conductivity for water-fracturing and conventional fracturing applications. SPE J. 6 (3):
288 –298. SPE-74138-PA. 10.2118/74138-PA.
Gatens III J. M., Lee, W. J., Lane, H. S., Watson, A. T., Stanley, D. K., and Lancaster D. E. 1989.
Analysis of eastern Devonian gas shales production data. J. Pet. Technol. 41 (5): 519 –525.
SPE-17059-PA. 10.2118/17059-PA.
Geertsma, J. 1957. The effect of fluid pressure decline on volumetric changes of porous rocks. Trans.
AIME: 331.
Gibowicz, S. J. and Kijko, A. 1994. An introduction to mining seismology, First Edition: Academic
Press.
60 SPE-177277-MS

Grebe, J. J. and Stoesser, M. 1935. Increasing crude production 20,000,000 bbl. from established
fields, World Petroleum J, (August): 473–482.
Grechka, V., Mazumdar, P., and Shapiro, S. A. 2010. Case history: predicting permeability and gas
production of hydraulically fractured tight sands from microseismic data. Geophysics 75 (1):
B1-B10. 10.1190/L3278724.
Griffith A. A. 1920. The phenomena of rupture and flow in solids. Phil. Trans. R. Soc. A221: 163–198.
Gutenberg, B. and Richter, C. F. 1944. Frequency of earthquakes in California. Bulletin of the
Seismological Society of America, 34: 185–188.
Halliburton.2015. Hydraulic Fracturing 101. ⬍ http://www.halliburton.com/public/projects/pubsdata/
Hydraulic Fracturing/fracturing 101.html⬎. (August 13, 2015).
Holland, A. 2011. Examination of possibly induced seismicity from hydraulic fracturing in the Eola
Field, Garvin County, Oklahoma. Oklahoma Geological Survey, Open-File Report OF1–2011.
Hosseini, S. M. and Aminzadeh, F. 2013. A new model for geomechanical seismicity based reservoir
characterization including reservoir discontinuity orientations. Presented at the SPE Annual
Technical Conference and Exhibition held in New Orleans, Louisiana, USA, 30 September-2
October 2013. SPE-166485-MS. 10.2118/166485-MS.
Hulsey, B. J., Cornette, B., and Pratt, D. 2010. Surface microseismic mapping reveals details of the
Marcellus shale. Presented at the SPE Eastern Regional Meeting held in Morgantown, West
Virginia, USA, 13–15 October. SPE-138806-MS. 10.2118/138806-MS.
Jaeger, J. C. and Cook, N. G. W. 1979. Fundamentals of rock mechanics, Third Edition. London, UK:
Chapman and Hall Ltd.
Jaeger, J. C., Cook, N. G. W., and Zimmerman, R. W. 2007. Fundamentals of rock mechanics, Forth
Edition. USA: Blackwell Publishing Ltd.
King, G. E. 2010. Thirty years of gas shale fracturing: What we have learned? Presented at the SPE
Annual Technical Conference and Exhibition, Florence, Italy, 19 –22 September. SPE-133456-
MS. 10.2118/133456-MS.
King, G. E. 2012. Hydraulic fracturing 101: What every representative, environmentalist, regulator,
reporter, investor, university researcher, neighbor, and engineer should know about hydraulic
fracturing risk and improving frac performance in unconventional gas and oil wells. Presented at
the SPE Hydraulic Fracturing Technology Conference, Woodland, Texas, USA, February 6 – 8.
SPE- 152596-MS. 10.2118/152596-MS.
Li, Q., Yousefzadeh, A., and Aguilera, R. 2015. New seismic base reservoir characterization (SBRC)
geomechanical model based on Mogi empirical rock failure relationship. Presented at the SPE
Latin American and Caribbean Petroleum Engineering Conference, Quito, Ecuador, 18 –20 No-
vember. SPE- 177265-MS. 10.2118/177265-MS.
Ma, S., Guo, J., Li, L., Tham, L. G., Xia, Y., and Tang, C. 2015. Influence of pore pressure on tensile
fracture growth in rocks: a new explanation based on numerical testing. Front. Earth Sci. 9 (3):
412426. 10.1007/s11707– 014 – 0481– 4
Maxwell, S. C. 2010. Microseismic: Growth born from success. The Leading Edge 29 (3): 338 –343,
10.1190/L3353732.
Maxwell, S. C. 2014. Microseismic imaging of hydraulic fracturing: improved engineering of
unconventional shale reservoirs, Distinguished Instructor Short Course, Distinguished Instructor
Series, No. 17, First Edition. USA: Society of Exploration Geophysicists.
Maxwell, S. C. and Cipolla, C. 2011. What does microseismicity tell us about hydraulic fracturing?
Presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, USA, 30
October-2 November. SPE-146932-MS. 10.2118/146932-MS.
Maxwell, S. C. and Norton, M. 2012. Enhancing shale gas reservoir characterization using hydraulic
fracture microseismic data. European Association of Geoscientists & Engineers, First Break 30,
SPE-177277-MS 61

(2): 95–101.
Maxwell, S. C., Jones, M., Parker, R., Miong, S., Leaney, S., Dorval, D., D’Amico, D., Logel, J.,
Anderson, E., and Hammermaster, K. 2009. Fault activation during hydraulic fracturing. 79th
Annual International Meeting, Society of Exploration Geophysicists, Expanded Abstracts, 1552–
1556, 10.1190/L3255145.
Maxwell, S. C., Kresse, O., and Rutledge, J. 2013. Modeling microseismic hydraulic fracturing
deformation. Presented at the SPE Annual Technical Conference and Exhibition, New Orlean, LA,
30 September-2 October. SPE-166312-MS. 10.2118/166312-MS.
Maxwell, S. C., Underhill, W. B., Bennett, L., Woerpel, C., and Martinez, A. 2010. Key criteria for
successful microseismic project. Presented at the SPE Annual Technical Conference and Exhibi-
tion, Florence, Italy, 19 –22 September. SPE-134695-MS. 10.2118/134695-MS.
Maxwell, S. C., Urbancik, T. I., Steinberger, N. P., and Zinno, R. 2002. Microseismic imaging of
hydraulic fracture complexity in Barnett Shale. Presented at SPE Annual Technology Conference
and Exhibition, San Antonio, Texas, USA, 29 September - 2 October. SPE-77440-MS. 10.2118/
77440-MS.
Maxwell, S. C., Waltman, C., Warpinski, N. R., Mayerhofer, M. J., and Boroumand, N. 2006. Imaging
Seismic Deformation Induced by Hydraulic Fracture Complexity. Presented at the SPE Annual
Technology Conference and Exhibition, San Antonio, Texas, USA, 24 –27 September. SPE-
102801- MS. 10.2118/102801-MS.
Mayerhofer, M. J., Lolon, E. P., Warpinski, N. R., Cipolla, C. L., Walser, D., and Rightmire, C. M.
2008. What is stimulated reservoir volume? SPE Production & Operations 25 (1): 89 –98,
SPE-119890-PA. 10.2118/119890-PA.
Mueller, M. 2013. Meeting the challenge of uncertainty in surface microseismic monitoring. European
Association of Geoscientists & Engineers, First Break 31 (7), 89 –95.
Nagel, N. B., Garcia, X., Sanchez, M. A., and Lee, B. 2012. Understanding ⬙SRV⬙: A numerical
investigation of ⬙wet⬙ vs. ⬙dry⬙ microseismicity during hydraulic fracturing. Presented at the SPE
Annual Technical Conference and Exhibition, San Antonio, Texas, USA, 8 –10 October. SPE-
159791- MS. 10.2118/159791-MS.
Nagel, N. B., Zhang, F., Sanchez-Nagel, M., Garcia, X., and Lee, B. 2013. Quantitative evaluation of
completion techniques on influencing shale fracture complexity. Presented at ISRM International
Conference for Effective and Sustainable Hydraulic Fracturing, Brisbane, Australia, 20 –22 May.
National Research council. 2013. Induced seismicity potential in energy technologies. National
Academies Press, USA.
Norton, M., Hovdebo, W., Cho, D., Jones, M., and Maxwell, S. C. 2010. Surface seismic to
microseismic: An integrated case study from exploration to completion in the Montney Shale, NE
British Colombia, Canada. 80th Society of Exploration Geophysicists, Annual Meeting, Expanded
Abstract. 10.1190/1.3513258.
Notle, K. G. and Smith, M. B. 1981. Interpretation of fracturing pressure. Journal of Petroleum
Technology 33 (9): 1767–1775.SPE-8297-PA. 10.2118/8297-PA.
Nur, A. and Byerlee, J. D. 1971. An exact effective stress law for elastic deformation of rock with
fluids. Journal of Geophysical Research 76 (26): 6414 –6419.
Refunjol, X. E., Keranen, K. M., LeCalvez, J. H., and Marfurt, K. J. 2012. Integration of hydraulically
induced microseismic events location with active seismic attributes: A North Texas Barnett Shale
case study. Geophysics 77 (3): KS1-KS12. 10.1190/geo2011– 0032.1.
Rice, J. R. and Cleary, M. P. 1976. Some basic stress diffusion solutions for fluid-saturated elastic
porous media with compressible constituents. Review of Geophysics and Space Physics 14 (2):
227–241. 10.1029/RG014i002p00227.
Rutledge, J. T. and Phillips, W. S. 2003. Hydraulic stimulation of natural fractures as revealed by
62 SPE-177277-MS

induced microearthquakes, Carthage Cotton Valley gas filed, East Texas. Geophysics 68 (2):
441–452. 10.1190/1.1567214.
Schmitt, D. R. 2014. Basic geomechanics for induced seismicity: A tutorial. Canadian Society of
Exploration Geophysicists, Recorder, November 2014, 24 –29.
Shapiro, S. A. and Dinske, C. 2009. Fluid-induced seismicity: pressure diffusion and hydraulic
fracturing, European Association of Geoscientists & Engineers, Geophysical Prospecting 57 (2):
301–310. 10.1111/j.1365–2478.2008.00770.x.
Shapiro, S. A., Huenges, E., and Borm, G. 1997. Estimating the crust permeability from fluid-
injection- induced seismic emission at the KTB site, Geophysical Journal International 131 (2):
F15-F18. 10.1111/j.1365–246X.1997.tb01215.x.
Sheriff, R. E. and Geldart, L. P. 1995. Exploration seismology, Second Edition. New York, NY, USA:
Cambridge University Press.
Skempton, A. W. 1960. Effective stress in soils, concrete and rocks: Pore Pressure and Suction in
Soils, P. 1, Butterworth, London.
Stein, S. and Wysession, M. 2003. An introduction to seismology earthquakes and earth structure,
First Edition. USA: Blackwell Publishing Ltd.
Terzaghi, K. 1923. Die berechnung der durchlassigkeitzifer des tones aus dem verlauf der hydrody-
namischen spannungserscheinungen. Sitzungsber, Akad. Wiss. Wien Math-Naturwiss. Kl. Abt. IIA,
132:105–124.
The Economist. 2012. Shale of the century. June 2, 2012. http://www.economi st.com/node/21556242
(August 13, 2015).
Thornton, M. and Eisner, L. 2011. Uncertainty in surface microseismic monitoring, 81st Society of
Exploration Geophysicists, Annual Meeting, Expanded Abstract: 1524 –1528. 10.1190/L3627492.
Udias, A. 1999. Principles of seismology, First Edition: Cambridge University Press.
Urbancic, T. I., Shumila, V., Rutledge, J. T., and Zinno, R. J. 1999. Determining hydraulic fracture
behavior using microseismicity. Vail Rocks 1999, The 37th US Symposium on Rock Mechanics
(USRMS). 990991.
Vermylen, J. P. 2011. Geomechanical studies of the Barnet Shale, Texas, USA. PhD thesis disserta-
tion, Stanford University, California, USA.
Vermylen, J. P. and Zoback, M. D. 2011. Hydraulic fracturing, microseismic magnitudes, and stress
evaluation in Barnet Shale, Texas, USA. Presented at the SPE Hydraulic Fracturing Technology
Conference and Exhibition, Woodland, Texas, USA, January 24 –26. SPE-140507-MS. 10.2118/
140507-MS.
Walsh, J. B. 1965. The effect of cracks on the compressibility of rock. J. Geophys. Res. 70 (2):
381–389. 10.1029/JZ070i002p00381.
Wang, H. F. 2000. Theory of linear poroelasticity with application to geomechanics and hydrogeol-
ogy. Princeton University Press, USA.
Warpinski, N. R. 2014. Microseismic monitoring-The key is integration. The Leading Edge 33 (10):
10981100. October 2014. 10.1190/tle33101098.1.
Warpinski, N. R., Branagan, P. T., Peterson, R. E., Wolhart, S. L., and Uhl, J. E. 1998. Mapping
hydraulic fracture growth and geometry using microseismic events detected by wireline retrievable
accelerometer array, Presented at the SPE Gas Technology Symposium, Calgary, Alberta, Canada,
15–18 March. SPE-40014-MS. 10.2118/40014-MS.
Warpinski, N. R., Kramm, R. C., Heinze, J. R., and Waltman, C. K. 2005. Comparison of Single-and
DualArray Microseismic Mapping Techniques in the Barnett Shale. Presented at the SPE Annual
Technical Conference and Exhibition, Dallas, Texas, 9 –12 October. SPE-95568-MS. 10.2118/
95568-MS.
Warpinski, N. R., Mayerhofer, M. J., Agarwal, K., and Du, J. 2013. Hydraulic fracture geomechanics
SPE-177277-MS 63

and microseismic source mechanisms. SPE J. 18 (4): 766 –780. SPE-158935-PA, 10.2118/158935-
PA.
Yu, G. and Aguilera, R., 2012. 3D Analytical modeling of hydraulic fracturing stimulated reservoir
volume. Presented at the SPE Latin American and Caribbean Petroleum Engineering Conference,
Mexico City, Mexico, 16 –18 April. SPE-153486-PP. 10.2118/153486-MS.
Zhang, X., Holland, M., van der Zee, W., and Moos, D. 2014. Using microseismic to estimate
stimulation efficiency, application and drawbacks. Presented at the Beijing 2014 International
Geophysical Conference & Exposition, Beijing, China, 21–24 April: 1351–1354. 10.1190/IGC-
Beijing2014 –341.
Zhuang, Z., Liu, Z., Cheng, B., and liao, J. 2014. Extended finite elements method, Tsinghua
University Press Computational Mechanics Series, First Edition: Academic Press.
Zimmer, U. 2011. Calculating stimulated reservoir volume (SRV) with consideration of uncertainties
in microseismic event locations. Presented at the Canadian Unconventional Resources Conference,
Calgary, Alberta, Canada, 15–17 November. SPE-148610-MS. 10.2118/148610-MS.
Zoback, M. D. 2007. Reservoir geomechanics, First Edition. Cambridge, UK: Cambridge University
Press.
Zoback, M. D. and Gorelick, S. M. 2012. Earthquake triggering and large-scale geologic storage of
carbon dioxide. Proceeding of the National Academy of Science (PNAS) 109 (26): 10164 –10168,
10.1073/pnas.1202473109.

You might also like