Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

SPE 115769

The Relationship Between Fracture Complexity, Reservoir Properties, and


Fracture Treatment Design
C.L. Cipolla, N.R. Warpinski, M.J. Mayerhofer, and E.P. Lolon, Pinnacle Technologies, and M.C. Vincent,
Carbo Ceramics

Copyright 2008, Society of Petroleum Engineers

This paper was prepared for presentation at the 2008 SPE Annual Technical Conference and Exhibition held in Denver, Colorado, USA, 21–24 September 2008.

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
In many reservoirs fracture growth may be complex due to the interaction of the hydraulic fracture with natural fractures,
fissures, and other geologic heterogeneities. The decision whether to control or exploit fracture complexity has significant
impact on fracture design and well performance. This paper investigates fracture treatment design issues as they relate to
various degrees and types of fracture complexity (i.e., simple planar fractures, complex planar fractures, and network fracture
behavior), including the effect of fracture fluid viscosity on fracture complexity, proppant distribution in complex fractures,
and fracture conductivity requirements for complex fractures. The impact of reservoir properties (including permeability,
stress and modulus) on treatment design is also evaluated. The paper includes general guidelines for treatment design when
fracture growth is complex. This includes criteria for the application of water-fracs, hybrid fracs, and crosslinked fluids.
The paper begins with an evaluation of microseismic fracture mapping data that illustrates how fracture complexity can
be maximized using low viscosity fluids, which includes an example of how microseismic data can be used to estimate the
permeability and spacing of secondary or network fractures. The effect of proppant distribution on gas well performance is
also examined for cases when fracture growth is complex, assuming that proppant was either concentrated in a primary planar
fracture or evenly distributed in a fracture network. Examples are presented that show when fracture growth is complex the
average proppant concentration will likely be too low to materially impact well performance if proppant is evenly distributed
in the fracture network and un-propped fracture conductivity will control gas production. This paper also extends published
conductivity data for un-propped fractures and embedment predictions for partially propped fractures to lower modulus rock
to provide insights into fracture design decisions. Exploiting fracture complexity may not possible when Young’s modulus is
2 x 106 psi or lower due to insufficient network conductivity resulting from asperity deformation and proppant embedment.
Fracture conductivity requirements are examined for a wide range of reservoir permeability and fracture complexity.
Reservoir simulations illustrate that the network fracture conductivity required to maximize production is proportional to the
square-root of fracture spacing, indicating that increasing fracture complexity will reduce conductivity requirements. The
reservoir simulations show that fracture conductivity requirements are proportional k1/2 for small networks and k1/4 for
large networks, indicating much higher conductivity requirements for low permeability reservoirs than would be predicted
using classical dimensionless conductivity calculations (Fcd) where conductivity requirements are proportionate to reservoir
permeability (k). The results show that when fracture growth is complex, proppant distribution will have a significant impact
on network conductivity requirement and well performance. If an infinite conductivity primary fracture can be created,
network fracture conductivity requirements are reduced by a factor of 10 to 100 depending on the size of the network. The
decision to exploit or control fracture complexity depends on reservoir permeability, the degree of fracture complexity, and
un-propped fracture conductivity.
The paper also examines the effect of fluid leakoff on maximum fracture area, illustrating potential limits for fracture
complexity as reservoir permeability increases. Although the expected range of un-propped fracture conductivity is
controlled by Young’s modulus and closure stress, in many reservoirs it can be beneficial to exploit fracture complexity when
the permeability is on order 0.0001 mD by generating large fracture networks using low viscosity fluids (water-fracs). As
reservoir permeability approaches 0.01 mD, fluid efficiency decreases and fracture conductivity requirements increase,
fracture designs can be tailored to generate small networks with improved conductivity using medium viscosity or multiple
fluids (hybrid fracs). Fracture complexity should be controlled using high viscosity fluids and fracture conductivity
optimized for moderate permeability reservoirs, on order 1 mD.
2 SPE 115769

Introduction
Complex growth of hydraulic fractures has been documented in mine-back experiments, providing direct observations of
hydraulic fracture complexity in a variety of environments including tight sandstones and coalbed methane reservoirs.1-6
Complexity is frequently associated with the interaction of the hydraulic fracture with a pre-existing rock fabric such as
natural fractures, fissures, or cleats. Unfortunately data from mine-back experiments are very limited, requiring other
methods to diagnose hydraulic fracture complexity. Until recently, fracture pressure analysis was the only diagnostic
available to estimate complexity.7-21 Fracture pressure analysis has been used to estimate both near-wellbore7,22 and far-field
fracture complexity; the focus of this paper is far-field fracture complexity. In most cases far-field fracture complexity is
deemed detrimental due to excessive fluid leakoff and/or reduced fracture width that can result in early screenouts.23,24 In
many cases, fracture complexity is reduced by adding particulates that likely plug secondary fractures and/or fissures.20,25,26
In some reservoirs, however, maximizing fracture complexity is the goal of the treatment design. During the past ten years,
thousands of hydraulic fracture treatments have been characterized using microseismic and tiltmeter fracture mapping
technologies. These measurements have shown a surprising diversity in hydraulic fracture growth, ranging from simple
planar fractures to very complex fracture systems to extreme fracture height confinement (that are not explained by variations
in rock properties and stress).27-44 The occurrence of complex fracture growth is much more common than initially anticipated
and is becoming more prevalent with the increased development of unconventional reservoirs. The nature and degree of the
fracture complexity must be clearly understood to select the best stimulation strategy. This paper focuses on the relationship
between fracture complexity, reservoir properties, and stimulation design.
Hydraulic fracture growth can be
Planar-decoupled divided into four categories:
- Planar-coupled growth
Planar
- Planar-decoupled growth or
fissure opening
- Complex growth
No
communication • Non-communicating
Complex Planar • Communicating
- Network growth
communication
Figure 1 illustrates the various types
ΔXs of fracture growth. The depictions are
simplistic, but will serve as a reference
for calculations and modeling presented
in subsequent sections. Examples of
each fracture growth category have been
Xn Network documented with direct fracture
geometry measurements using tiltmeter
and/or microseismic frac mapping or
inferred from fracture pressure analyses.
Large-scale fracture complexity can be
2 Xf
measured using microseismic and/or
tiltmeter fracture mapping, allowing
Figure 1 - Fracture growth and complexity scenarios direct detection of network fracture
growth and large-scale decoupling
(fissure opening); however, it is difficult to identify small-scale fracture complexity such as complex-planar growth or
limited planar-decoupled growth using hydraulic fracture mapping technologies as the resolution is typically not sufficient to
distinguish these features. Therefore, fracture pressure analysis is used to determine small-scale complexity.

Effect of Viscosity on Fracture Complexity


Fracture complexity is typically reduced when fluid viscosity is increased. Laboratory tests have shown that both near-
wellbore and far-field fracture complexity may be reduced using high viscosity fluids.22,45 Field data have also indicated that
high viscosity fluids can reduce fracture complexity.8,18-20 It is more difficult for high viscosity fluids to penetrate intersected
natural fractures or fissures.45 In addition, the penetration distance into these intersected natural fractures or fissures may be
significantly reduced as viscosity increases. For gas reservoirs, the penetration distance can be approximated using a limiting
estimate discussed by Warpinski, et al.46 Given the expected behavior for fluid flow into essentially evacuated fractures, fluid
movement is expected to obey a law similar to that for viscosity-dominated leakoff and the distance y that fluid will penetrate
is estimated by Equation 1.

2 k ΔP t
y= = g t ………………….. (1)
ϕμ
SPE 115769 3

Where k is the fracture permeability, ΔP is the pressure drop inside the fracture, t is time, φ is fracture porosity, μ is fluid
viscosity, and g is a constant. Additional details are provided in Appendix 1.

Evaluation of Microseismic Data to Estimate Fluid Diffusion


A proper application of the relative penetration distance (Equation 1) requires some assumptions or inferences regarding
the pressure drop and permeability for the fractures created with low and high viscosity fluids. These inferences will be
evaluated in more detail by comparing the behavior of low and high viscosity fracture treatments. In relatively simple
environments, the development of micro-seismicity may provide important information about the reservoir. Shapiro has
shown that the induced seismicity can be used to extract information regarding the diffusion of fluid in a geothermal reservoir
and in some cases for fracturing applications, such as a network fracture.48 This approach is applied here to network fractures
to explore viscosity relationships, but particularly for cases with a well-defined geometry where the configuration is
amenable for evaluation of diffusion effects. An example of an appropriate case would be a fracture in a vertical well in a
reservoir like the Barnett Shale, where the primary fracture azimuth is known and the diffusion orthogonal to the fracture can
be evaluated from the extending micro-seismicity. In the Barnett Shale, the composite permeability normal to the principal
fracture at fracturing conditions will be a function of the opening of pre-existing healed natural fractures. In general,
diffusion effects from horizontal wells in the Barnett would not be appropriate, since multiple origination points are possible
(several perforation clusters per stage) and the diffusion behavior would be difficult to sort through; however, several
horizontal Barnett Shale wells have been oriented so that the primary created fractures are longitudinal. In such cases, the
diffusion approach is warranted since flow orthogonal to the wellbore/fracture would be indicative of the behavior of interest.
Consider the fractures mapped in Warpinski, et al. (2005), with dual monitoring arrays to determine microseismicity.35
Fracture treatments were performed with a cross-linked gel treatment and, because of poor performance, re-stimulated several
months later with a water-frac. The locations of the micro-seismicity orthogonal to the wellbore/fracture system can be
displayed as a function of elapsed treatment time (Figure 2). The microseismic clouds clearly expand in lateral extent with
time and the two treatments (gel and water) do so at a different rate. It should also be noted that the very early time data is
probably not useful since the general width of a microseismic cloud around the hydraulic fracture itself may not be indicative
of diffusion if the mechanism is not associated with fluid flow (e.g., stress effects around the fracture tip and location errors).
A square-root-of-time function can be fit to the microseismic envelope to estimate the diffusion according to Equation 1.
For the two cases shown in Figure 2, the values of g given by the dashed lines are 50 ft/√min for the water-frac and 24
ft/√min for the gel frac. Converting to units where k is in Darcies, ΔP is in psi, t is in minutes, and μ is in cp, the behavior
can be used to solve Equation 1 for the
average orthogonal system permeability,
1400 if the other parameters are known. Table
♦ - Water-frac 1 shows best-chosen values; note that ΔP
1200 ▲- Cross-linked Gel Frac generally denotes the difference between
the fracturing pressure and the reservoir
Orthogonal Distance (ft)

1000 pressure, but since the fractures will not


open unless the pressure exceeds the in
800
situ stress, it is probably more appropriate
to use the fracturing pressure minus the
maximum horizontal stress. If the net
600 pressure in a water-frac treatment is 500
psi, then the effective ΔP for flow through
400 the cross fractures is more likely going to
be ~300 psi (since it is acting against the
200 maximum horizontal stress and assuming
a differential of 200 psi between
0 minimum and maximum horizontal
0 100 200 300 400 500 600 stress). Solving Equation 1 yields:
Time (min) g2μ
k = 114 ……………………. (2)
ΔP
Before computing results, it is also
Figure 2 – Micro-seismicity orthogonal to wellbore and fracture azimuth likely that the fluid flow orthogonal to the
hydraulic fracture is not going to be a
direct path. Based upon the complex microseismic development observed in water-fracs, the process could be approximated
with a random walk pattern. In a random walk, it is equally likely that the extension of natural fractures could be orthogonal,
sub parallel left, or sub parallel right, then the total distance traveled could be as much as three times the orthogonal distance.
This is likely a maximum value, but a reasonable range appears to be 1 to 3. It is expected that gel fractures would be closer
to unity (based on microseismic data) and water-fracs would be closer to the upper range of 3.
4 SPE 115769

Using the “direct path” orthogonal distance, the average system orthogonal permeability for the water-frac is 950 Darcies
while pumping. This is about what one would expect to get fluid movement 1,000 ft from the fracture. For the gel fracture,
the value is 25,000 Darcies while pumping. This value also makes sense as higher pressures and wider fractures are required
to move the much higher viscosity gel any significant distance. Although it might appear advantageous to pump gel and
induce wider fractures, the orthogonal network will be much more limited in size and following fracture closure, residual gel
damage must be considered. If we now add the “random” path to the water-frac calculations (using a maximum value of 3),
g becomes 150 and the permeability becomes 8,550 Darcies. Table 1 summarizes the permeability estimates and
assumptions used in the calculation.
Table 1 - Input Data and Estimated Permeability
of Dilated Fracture (While Pumping)
Note that this calculation is independent of the
ΔP μ Permeability
(Darcies) volume pumped, as long as it is sufficient for the
(psi) (centipoise)
treatment to actually start developing diffusional
Water-frac 300 1 950-8550
characteristics. For the water-fracs, the diffusion is
Cross-linked Gel Frac 800 100-300 8300-25,000
probably considerably underestimated because the
fluid from the orthogonal fractures must fill the sub parallel fractures as well. This is not likely to be the case for the gel
treatments, where it is not clear if any real network develops. Taken one step farther, this analysis can be used to estimate the
fracture widths from the parallel plate analogy using Equation 3:

w = 12k , in units of feet and Darcies, w = 1.1x10 −5 k ……………………………… (3)

The water-frac orthogonal widths during the treatment would then be 0.00035 ft (direct path) and 0.001 ft (random path),
while the gel widths could range from 0.001 ft (100 cp) to 0.0017 ft (300 cp). It is tempting to now calculate volumetrics, but
the large leakoff area (particularly for the water-frac) suggests that there could be a very significant percentage of leakoff.
Ignoring leakoff, the volume of any one orthogonal fracture would be 105 ft3 (water-frac, direct), 300 ft3 (water-frac,
random), 90 ft3 (gel frac-100 cp) and 153 ft3 (gel frac-300 cp) for the two cases, assuming the orthogonal fractures are 300 ft
high and 1,000 ft long in the water-frac and 300 ft long in the gel frac (microseismic width). With no leakoff, two (bi-wing)
fractures, and tiltmeter information suggesting that 40% of the fluid goes into the orthogonal fractures28, it would require 650
(direct), 226 (random), 145 (gel-100cp) and 85 (gel-300 cp) fractures to account for a treatment volume of 60,000 bbl. This
fracture density equates to a spacing of 4 to 12 ft for the water-frac and 19 to 32 ft for the gel frac. The fracture spacing
would be proportional to the fluid efficiency and would increase by a factor of two if fluid efficiency were 50%.

Implications of Barnett Shale Fluid Diffusion Evaluation


The fluid diffusion evaluation indicates that high viscosity fluids reduce fracture complexity, as evidenced by the much
less complex microseismic pattern that is developed, smaller penetration distance, and likelihood that high viscosity fluids
create fewer secondary fractures. In the example presented, the cross-linked gel frac resulted in a penetration distance of
about 300 ft, while the water-frac resulted in a penetration distance of 1,000 ft (as interpreted from the microseismic event
development, Figure 2). The relative penetration distance for the cross-linked gel is much greater than predicted from
Equation 1, which would indicate that the penetration for a 100 to 300 cp fluid would be 7 to 11% of the penetration distance
of water (includes a correction for injection time and pressure drop) – or about 70 to 110 ft. The actual penetration distance
of 300 ft for the high viscosity fluid seems to indicate higher permeability (i.e., wider) secondary fractures were created with
the cross-linked gel, as calculated from the diffusion behavior. The overall implications from the diffusion evaluation are
that high viscosity fluid reduces fracture complexity and penetration distance (in the Barnett Shale). In addition, the
application of microseismic data to estimate permeability and width of secondary (orthogonal) fractures combined with tilt
mapping measurements of directional fluid movement may provide important information to estimate fracture spacing.

Proppant Distribution in Complex Fractures


Proppant transport in simple, planar, perfectly vertical fractures with smooth faces has been well studied and appears to be
reasonably well understood; however, proppant transport when fracture growth is complex, particularly in cases where a
fracture network is developed, is not as clear. This paper will evaluate several potential arrangements of proppant within
fracture networks. As previously shown, network fractures are more commonly created with low viscosity fracturing fluids;
unfortunately, these low viscosity fluids provide correspondingly poor proppant transport properties. For these low viscosity
fluids the settling rate46 for most proppant types used in the industry exceeds five feet per minute and it is expected that most
proppant particles will settle into a proppant bank at the bottom of the fracture or in notches or offsets at bed boundaries even
with simple planar fracture growth. In cases of extreme fracture complexity with induced fractures only slightly wider than
the proppant grains that may capture grains prior to settling or with neutrally buoyant materials78, proppant particles may
arguably be sparsely distributed through the fracture network. It is also possible that large portions of the fractures will only
have the residual conductivity of the closed un-propped fracture.
SPE 115769 5

ΔXs Due to the inability to accurately model


proppant transport when fracture growth is
ΔXs
complex, proppant placement will be
Xn Evenly Distributed estimated using three limiting scenarios:
(Case 1)
(Case 1) the proppant is evenly distributed
throughout the complex fracture system,
2 Xf (Case 2) the proppant is concentrated in a
dominant planar fracture with an un-
propped complex fracture system accepting
fluid only and (Case 3) the proppant settles
Concentrated in a
dominant fracture and forms “pillars” that are evenly
(Case 2) distributed within the complex fracture
system. Figure 3 and Figure 4 illustrate
the three proppant distribution scenarios.
In some cases, moderately complex fracture
Pillar Distribution
growth is approximated using multiple
(Case 3) parallel (planar) fractures in fracture
models. In these cases proppant will be
distributed evenly between each planar
Figure 3 - Proppant transport scenarios (plan view) fracture. The additional concerns of
proppant settling and failing to effectively
prop the entire pay interval (Figure 4) in complex fracture systems are not addressed in this work and proppant is assumed to
be evenly distributed over the fracture height. The effects of un-propped fracture height for simple planar fracture growth
have been addressed by Britt, et al.69 This work indicated that in reservoirs in which kv/kh > 0.1, well productivity may not
be negatively affected if 50% or more of the pay zone is propped or if un-propped fracture conductivity provides sufficient
fracture flow capacity. The general conclusions of Britt likely apply to cases were fracture growth is complex, and may bode
well for naturally fissured reservoirs with
Proppant in Dominant Proppant in Complex enhanced vertical flow. Appendix 2 provides
additional details on proppant transport and
Planar Fracture Fracture System
equations for estimating average proppant
concentration.
.... ..... ....
Pumping
.. .. .. Example Calculations of Proppant
.. .. .. Concentration in Complex Fractures
.. .. .. .. .. ..
.. .. .. Two examples are presented to illustrate
Closed

.. .. ..
Fracture
the effect of fracture complexity on proppant
.. ... ...... ......
distribution and the range of proppant
.. .. .. concentrations that may result when
. . .
.. . .... ....
. complexity is high. The effect of fracture
.. .. .. conductivity on well productivity and the
... ... ... relationship between fracture complexity,
.. .. .. fracture conductivity, and well productivity
will be examined later in the paper.
Barnett Shale. The Barnett Shale
high viscosity Fluid low viscosity Fluid Pillar Distribution Evenly Distributed
stimulation treatment discussed previously, a
water-frac or slick-water treatment, provided
Figure 4 - Proppant transport scenarios (side view) microseismic measurements of fracture length,
height, and network width.35 The diffusion
analysis for this horizontal well stimulation
indicated a fracture spacing of ~10 ft for the water-frac, while previous reservoir simulation studies have indicated effective
fracture spacing of 100 to 300 ft for vertical well stimulations using a matrix permeability of 0.0001 mD.47 The difference
between the diffusion analysis and reservoir modeling could be due to limitations in the diffusion theory to accurately
describe fluid flow in these complex fracture systems, differences between “effective” fracture area from production
modeling and “created” area, and/or over-estimation of matrix permeability which would result in over-predicting fracture
spacing from production history matching. However, the two approaches provide a reasonable bound for fracture spacing
that can be used to estimate fracture area and proppant concentration. It may be possible to estimate the relative fracture
spacing or density using advanced microseismic analyses, which indicates that the microseismic moment density could be a
crude measure of fracture density.79 This treatment consisted of 60,000 bbl of water and 385,000 lbm of proppant and
resulted in a complex fracture with a half-length of 1,500 ft, height of 300 ft, and network width of 2,000 ft. Assuming a fluid
efficiency of 75% (required only for Case 3-pillar distribution), the average proppant concentrations for fracture spacing of
6 SPE 115769

10, 50, and 150 ft are provided in Table 2. The average proppant concentration for Case 1 (evenly distributed) is very low,
ranging from 0.001 to 0.015 lbm/ft2. This seems to indicate that an even distribution of proppant throughout this very
complex network would result in concentrations that are probably too low to be effective and un-propped fracture
conductivity would likely dominate well productivity. It should be noted that the width at end of pumping for 10-ft fracture
spacing is too small to accept 40/70-mesh proppant. If the proppant is concentrated in a primary planar fracture (Case 2), the
average proppant concentration would be 0.43 lbm/ft2 (independent of network fracture spacing). In Case 3, where the
proppant is assumed to be evenly distributed in pillars, the average proppant concentration of the pillars would range from
0.07 to 1.0 lbm/ft2 depending on fracture spacing; however, only 1.5% of the fracture area would be propped. The very small
percentage of the total fracture area that is propped in Case 3 would likely be insufficient to support the closure stress and
resist fracture closure. Cases 1 and 3 do not appear to provide effective propped fractures. Therefore, in this example well
productivity may be dominated by the un-propped fracture conductivity. If proppant transport is best described by Case 2,
then a much higher conductivity connection to the wellbore will result compared to Cases 1 and 3 which could significantly
impact productivity.

Cotton Valley Sand. Mayerhofer, et al.37,


Table 2 - Proppant Concentration, Typical Barnett Shale Treatment presented fracture mapping results from a
Cotton Valley hybrid stimulation
Fracture Spacing
consisting of 10,000 bbl of slick-water
10 ft 50 ft 150 ft
2
Afc(ft ) 361,500,000 73,500,000 25,500,000
and 20 lbm/1000 gal linear gel containing
Wfc(inches) - at end of pumping 0.008 0.041 0.120
300,000 lbm of proppant (equal amounts
2
Cp1(lbm/ft ) 0.001 0.005 0.015
of 40/70-mesh sand and 20/40-mesh
2
Cp2(lbm/ft ) 0.43 0.43 0.43
resin-coated sand). The microseismic
2
Cp3(lbm/ft ) 0.07 0.34 0.99
fracture mapping indicated a fracture half-
% Propped (case 3) 1.5% 1.5% 1.5% length of 1,550 ft and a height of 250 ft.
The width of the microseismic event
cloud was ~200 ft, indicating relatively planar fracture growth, with no evidence of network fracture growth. Although there
may be small-scale fracture complexity, the microseismic data cannot provide sufficient detail to distinguish between simple
planar growth and complex planar growth. Therefore, proppant concentrations will be estimated for planar growth and
complex planar growth (low degree of complexity). Fracture modeling indicated a fluid efficiency of ~70%. Table 3 shows
the proppant concentrations for a simple planar fracture, three parallel fractures (Figure 1, complex planar – no
communication), and a narrow 200-ft wide network with 100-ft fracture spacing (Figure 1, complex planar –
communication). The results indicate that a planar fracture would have an average proppant concentration of 0.39 lbm/ft2 if
proppant is evenly distributed over the fracture area. However, average proppant concentration is reduced by a factor of three
(0.13 lbm/ft2) if three parallel fractures are used to approximate moderate complexity and to 0.076 lbm/ft2 if a small network
was created with 100-ft fracture
spacing. For all cases, only 7.6% of the Table 3 - Proppant Concentration, Typical Cotton Valley Hybrid Treatment
fracture area is propped if the proppant
settles into pillars (or to the bottom of Fracture Spacing
the fracture system). In the Cotton Planar 3-parallel 100 ft network
2
Valley, an even distribution of proppant Afc(ft ) 775,000 2,325,000 3,925,000
may result in sufficient proppant W fc (inches) - at end of pumping 0.609 0.203 0.120
2
concentration if fracture complexity is C p1 (lbm/ft ) 0.39 0.13 0.076
low (simple planar fracture). However, Cp2(lbm/ft2) 0.39 0.13 0.39
2
a modest degree of fracture complexity Cp3(lbm/ft ) 5.1 1.69 1.00
could result in a partial monolayer % Propped (case 3) 7.6% 7.6% 7.6%
distribution.
The Barnett Shale and Cotton Valley examples illustrate how the degree of fracture complexity can impact proppant
concentration. In many cases, fracture complexity can result in sparsely distributed partial monolayer concentrations (Case
1) or insufficient propping of the fracture area (Case 3), and the subsequent well productivity my be dominated by the un-
propped fracture conductivity. Even if proppant is concentrated in a single planar fracture, the un-propped fracture
conductivity may be an important consideration when selecting stimulation fluids and optimizing design. Therefore,
estimating the conductivity of un-propped and partially propped fractures will be an important component of stimulation
design in complex environments.

Conductivity of Partially Propped and Un-Propped Fractures


The surface area of the fracture increases as fracture complexity increases, which can result in the distribution of proppant
over a much larger area compared to a simple, single planar fracture (Figure 1). The conductivity of propped fractures has
been extensively studied; especially for higher concentrations of proppant in the fracture.58-65 However, there is less
understanding of the actual particle arrangement and the corresponding conductivity of a partial monolayer. The conductivity
of partially propped and un-propped fractures is much more difficult to determine without accurate descriptions of the
SPE 115769 7

fracture face and any asperities. The importance of fracture conductivity for a partial monolayer and un-propped fracture has
been emphasized with the re-introduction of water-fracs, where very low proppant concentrations are pumped.66-69 The
conductivity of partially propped and un-propped fractures has been studied in the laboratory by Fredd, et al., providing
important information for fracture design.70 There are very few direct field measurements of un-propped fracture conductivity
to confirm the lab results presented by Fredd, but pre- and post-fracture pressure buildup results from a pure CO2 stimulation
in the Ozona field71,72 indicated an un-propped fracture conductivity of 0.6 mD-ft (5,500 psi closure stress and E=5x106 psi),
which is consistent with the laboratory results of Fredd. Reservoir modeling47 in the Barnett Shale has indicated the
conductivity for partially propped and/or un-propped fractures could be 1 to 5 mD-ft (4,500 psi closure stress and E=6x106
psi), also within the range predicted using the Fredd data.
The Fredd data from Cotton Valley cores and the field examples noted above focus on relatively high modulus rock and
moderate closure stresses. However, a relationship between closure stress and modulus is needed to extend Fredd’s work and
develop guidelines for fracture design. Although attempting to extrapolate such limited data that exhibits a wide range of
uncertainty is problematic, the intent is to provide general relationships to approximate fracture conductivity for un-propped
fractures. The case of residual conductivity in a fracture that has been opened and then closed is one of asperity contact that
has been studied extensively in the literature. Fredd provided experimental results, but earlier work by Walsh and others laid
the theoretical groundwork for the expected behavior of the fracture, particularly with regard to stress.73,74 Gall, et al.,
studied damage to such fractures caused by leakoff fluids.75 Using a Walsh type model, the general form of the permeability
is given by:
3
⎧⎪ ⎡ DE ⎤ ⎫⎪ 1
k = C ⎨ln ⎢ ⎥⎬ , which is equivalent to the familiar k 3 = A − B ln(σ ) , ……………………………(4)
⎪⎩ ⎣ (1 − ν 2 )σ ⎦ ⎪⎭

Where A, B, C and D are material/crack parameters, E is Young’s modulus, σ is the net normal stress on the fracture, and
ν is Poisson’s ratio. This type of relationship has been found to match several types of experimental data (e.g., Jones, Kranz,
et al., Gall, et al., and Warpinski).2,75-77 From the form of Equation 4 on the left side, it is clear that high modulus formations
and low effective stresses are more amenable to retaining conductivity. However, because of the log function, this equation
does not lend itself to generalized conclusions about relative behavior without knowing the material parameters.
Nevertheless, it is possible to fit data and
then use that information to make an
assessment of other parameters. An
example of the fit to Fredd’s data for an un-
propped displaced fracture is shown in
Figure 5.
Assuming the Walsh model is
representative of the behavior of un-
propped fractures, it can now be used to
assess the effect of the modulus of the
formation on the conductivity of the
fractures as a function of stress level.
Figure 6 shows the same data (E=6x106
psi) and extrapolations of those data for
lower modulus. The conductivity drop off
is dramatic with lower modulus materials.
The extrapolation indicates that un-propped
fractures, even when displaced and
supported by asperities, will essentially
Figure 5 – Fracture conductivity for un-propped and partially propped close when modulus is 2 x106 psi or lower
fractures (reproduced from Fredd, et al., SPE Journal Sept. 2001) and stress exceeds 4,000 psi.

Partially Propped Fractures


Predicting the conductivity of partially propped fractures may be necessary when fracture complexity is high and fracture
area is large. Fredd presented data for partially propped fractures with a proppant concentration of 0.1 lbm/ft2 (Figure 5). It
would be beneficial to extrapolate these results to lower modulus rock and lower proppant concentrations to aid in fracture
design when lab data are not available. Although extrapolations of such data are difficult and clearly not rigorous, the
following discussion and estimations provide general relationships that should be useful when evaluating fracture treatments
in complex environments.
Fredd’s work illustrated the effect of particle strength on fracture conductivity. Figure 5 shows a significant difference
between the conductivity of fractures that are partially propped with sand and bauxite. When proppants are sparsely arranged
8 SPE 115769

within a fracture, closure stresses are concentrated on fewer particles, and bauxite was measured to provide 100 times the
flow capacity of white sand, even at a modest stress of 4,000 psi. Due to the higher modulus Cotton Valley core used for the
tests, it is likely that the conductivity profile is dominated by particle crushing, not embedment. However, as modulus
decreases, embedment effects will increase, reducing the conductivity of the partially propped fracture.
It may be possible to extend Fredd’s data
1000
by evaluating the two extreme cases: (1)
E=6E+6 psi conductivity is dominated by embedment
E=4E+6 psi and (2) conductivity is dominated by
Conductivity (mD-ft)

100
E=2E+6 psi
proppant crushing. When conductivity is
dominated by crushing and the fracture is
E=1E+6 psi
partially propped, higher strength particles
10
will improve fracture conductivity. This
improvement can be seen in Figure 5 (Fredd
1 data) by comparing the bauxite conductivity
to the sand conductivity. The effect of
changing the size of the particles is more
0.1 complicated. Based on work by Huitt and
others using a Hertz cap model, it can be
0.01 observed that for the same proppant
0 2000 4000 6000 8000
concentration in the fracture, smaller
particles will be more resistant to
Stress (psi) crushing.88-90 Even though small particles are
not as strong as larger grains, a specified
Figure 6 - Effect of modulus on conductivity of un-propped fractures mass concentration (e.g., 0.1 lb/sq ft)
with shear offset, extrapolation of Fredd data using Walsh model composed of small particles will provide
more individual grains per surface area and
more contact points to distribute the applied load. This is the reason that the reported crush diminishes with small diameter
proppants. However, because of the reduced permeability provided by smaller particles, it is not certain whether the overall
fracture conductivity is improved or decreased by changing particle size. Smaller proppant diameters have a distinct
advantage in proppant transport characteristics along the primary fracture and potentially into the network fractures.
Leonard, et al.,56 and other anecdotal sources demonstrate that 40/70 and 100-mesh proppant can be carried distances on the
order of 1,000 ft in a slickwater fracture. The small diameter of these proppants also enhances their ability to penetrate into
network fractures that may be significantly narrower than the primary hydraulic fracture.
In the case where embedment dominates fracture conductivity behavior for sparsely distributed proppant, it may be
possible to utilize proppant embedment calculations to estimate the effects of decreasing modulus. The predicted embedment
for moderate to hard rock is shown in Figure 7.61 The embedment estimates shown in Figure 7 were developed for multi-
layers of proppant and are not valid for partially
propped fractures where the stress on each
particle will be significantly higher. Therefore, 1.60
1E+6 psi
Embedment (grain diameters)

using these embedment estimates for partially 1.40 2E+6 psi


propped fractures is clearly a very optimistic 4E+6 psi
assumption, as proppant crushing is ignored and 1.20 6E+6 psi
particle stress could be many times higher. 1.00
Figure 7 provides an upper bound of stress and
modulus for the application of low proppant 0.80
concentrations. In the absence of crushing, it 0.60
would be expected that partially propped
fractures would close when embedment exceeds 0.40
0.5 grain diameters into each face of the 0.20
fracture.
It may also be possible to extrapolate 0.00
Fredd’s data to proppant concentrations other 0 2000 4000 6000 8000 10000 12000
than 0.1 lbm/ft2 by using an effective stress that Stress (psi)
is calculated as 0.1/Cp. This approach may be a Figure 7 - Proppant embedment for moderate to hard rock
reasonable approximation for a limited range of
proppant concentrations. For example, assuming that the percentage of the total load carried by the rock is constant, the
effective stress for a proppant concentration of 0.05 lbm/ft2 would be twice that of 0.1 lbm/ft2. The conductivity for a partially
propped fracture could then be estimated using the concentration corrected stress. This approach would indicate that ultra
low proppant concentrations would likely behave similar to un-propped fractures, especially at higher stresses.
SPE 115769 9

Application of Conductivity Estimates


Although quantifying actual fracture conductivity for un-propped and partially propped fractures is not possible with such
a limited amount of data, it may be possible to use the available data and subsequent extrapolations to provide guidelines for
fracture design when fracture complexity exists. It may be possible to approximate reasonable bounds for fracture
conductivity given estimates of fracture complexity, closure stress, Young’s modulus, location and concentration of the
proppant. While the effect of non-Darcy, multiphase flow, and other damage factors have been described63,64 in fully packed
fractures containing multiple layers of proppant, additional guidance is needed to apply these reductions to partially or
irregularly propped fractures. The fracture conductivity extrapolations for un-propped fractures shown in Figure 6 and the
embedment calculations shown in Figure 7 indicate that fracture conductivity decreases rapidly for lower modulus rock,
especially at higher closure stress. For example, un-propped fracture conductivity for a 2x106 psi modulus rock at 6,000 psi
stress is estimated to be less than 0.01 mD-ft and embedment would be 0.5 grain diameters into each face of the fracture for a
partially propped fracture (indicating a closed fracture).

Relationship Between Fracture Conductivity and Well Productivity for Complex Fractures
A series of reservoir simulations was performed to evaluate the
relationship between fracture conductivity and well productivity when Table 4 - Reservoir Simulation Inputs
fracture growth is complex (vertical well). The simulations focused on Formation depth, D 7000 ft
low permeability gas reservoirs where water-fracs and hybrid fracs are Reservoir permeability, k 0.0001, 0.01, 1 mD
commonly applied. A low-range permeability of 0.0001 mD was selected Net pay thickness, h 300 ft
to represent many shale reservoirs and mid-range permeability of 0.01 Porosity, φ 0.03 (k = 1.e-4 mD)
0.09 (k = 1.e-2 mD)


mD to represent tight gas sands. Simulations were also performed using a 0.18 (k = 1 mD) –
permeability of 1 mD to represent moderate permeability reservoirs. The
basic input parameters for the reservoir modeling are shown in Table 4. Initial pore pressure, pi 3000 psi
Water saturation, Sw 0.3 –
A fracture or network length of 1,000 ft was used for all the simulations.
Reservoir temperature, T 180 ºF
The reservoir modeling covered a wide range of fracture complexity, from Rock compressibility, cf 3.00E-06 1/psi
simple planar fractures to large networks. In each case, a wide range of Gas viscosity, μg 0.019 cp
fracture conductivity was evaluated. When simulating medium to large Gas gravity, γg 0.6 –

Table 5 – Summary of Reservoir Modeling: Maximum fracture networks, two scenarios were used to evaluate
Achievable 1st Year Production (MMCF) the effect of network conductivity on well productivity:
(1) uniform fracture conductivity throughout the network
Case k = 0.0001 mD k = 0.01 mD k = 1 mD and (2) an infinite conductivity primary fracture that is
Large Network 1000 x 500 ft Uniform Conductivity connected to a fracture network. In all cases the one-
Dx (Frac Spacing) = 100 ft 1867 9520 43710 year cumulative production was normalized by dividing
Dx (Frac Spacing) = 50 ft 2113 9525 43718
Dx (Frac Spacing) = 25 ft 2152 9542 43777
the gas production (Gp) by the infinite conductivity gas
Large Network 1000 x 500 ft Infinite Conductivity production or maximum theoretical gas production
Dx (Frac Spacing) = 100 ft 1867 9522 43731 (Gpmax). For reference, the maximum achievable first-
Dx (Frac Spacing) = 50 ft 2113 9529 43744 year production for each case is summarized in Table 5.
Dx (Frac Spacing) = 25 ft 2152 9543 43777
The normalized production allows cases with widely
Small Network 1000 x 150 ft Uniform Conductivity
varying productivity to be easily compared. First-year
Dx (Frac Spacing) = 50 ft 736 4674 39803
production values were selected to minimize the
Dx (Frac Spacing) = 25 ft 742 4683 39867
Small Network 1000 x 150 ft Infinite Conductivity
influence of boundary affects while still providing a time
Dx (Frac Spacing) = 50 ft 736 4676 39845
frame of economic relevance. Correlating parameters
Dx (Frac Spacing) = 25 ft 742 4685 39885
were identified to “collapse” the normalized production
Small Network 1000 x 10 ft Communicating Fractures
curves for each set of simulations into a single curve to
Dx (Frac Spacing) = 5 ft 180 2682 35948 emphasize the relationship between reservoir
Small Network 1000 x 10 ft Non-Communicating Fractures permeability, fracture complexity, fracture conductivity
Three-Parallel Fracs 179 2669 35832 and well productivity. The reservoir simulations are
No Fracture Network
presented in order of complexity, from simple planar
Single Frac (Xf = 1,000 ft) 139 2581 53391
fractures to large networks.

Simple Planar Fracture


As a base line, the production from a simple planar fracture for various fracture conductivities was evaluated. Figure 8
compares the normalized production for permeability of 0.0001, 0.01, and 1.0 mD. The reservoir geometry is 10,000 ft x
10,000 ft with a 1,000-ft fracture half-length. Fracture conductivity divided by the square-root of reservoir permeability
(denoted FcDc) was plotted on the x-axis and appears to collapse the various permeability ranges fairly well. The simulations
indicate that to achieve near maximum (+98%) one-year production requires 30,000 mD-ft, 3,000 mD-ft, and 300 mD-ft for
permeability of 1 mD, 0.01 mD, and 0.0001 mD, respectively (FcDc of about 30,000 in Figure 8). A dimensionless fracture
conductivity (Fcd = kfwf/kxf) of 10 to 30 is normally considered optimum; however, the optimum Fcd depends on the time
selected for comparison and the reservoir permeability.83 In very tight reservoirs, conventional Fcd applications may not be
10 SPE 115769

appropriate. In this case, an Fcd of 30


would indicate fracture conductivity
requirements of 30,000 mD (k=1.0 mD),
300 mD-ft (k=0.01 mD), and 3 mD-ft
(k=0.0001 mD). For the 1 mD case, the
fracture conductivity estimated using an
Fcd of 30 is consistent with the near
maximum one-year production.
However, the Fcd based value of 300 mD-
ft for the 0.01 mD case would generate an
FcDc of 3,000 and result in about 15% less
production in the first year (Figure 8).
The difference becomes even greater for
the very low permeability case (k=0.0001
mD), where designing a fracture to
achieve an Fcd of 30 would result in about
40% less production in the first year
compared to what could be achieved with
a fracture conductivity 100 times higher
(300 mD-ft or Fcd=3000). These results
illustrate the limitations of conventional
Fcd based estimates of optimum fracture
conductivity and the benefits of reservoir
Figure 8 - Normalized Gp for the single fracture case (one year) simulation when evaluating fracture
conductivity requirements.

Complex Planar Fractures


The effect of small-scale fracture complexity was
(a) No Communication (b) Communication
evaluated for the case of five parallel fractures. Two
scenarios were modeled – one with no 1000 ft 1000 ft
communication between the fractures and one with
10 ft
an orthogonal set of fractures. Figure 9 illustrates 10 ft

the reservoir model used to simulate complex planar


fractures (quarter symmetry).
Figure 10 summarizes the results for the 3 Parallel Fractures Fracture Network
simulations, illustrating the effect of fracture 1000 ft 1000 ft

conductivity on first-year production. To collapse


the curves, the normalized gas production is graphed
as a function of fracture conductivity divided by the 2000 ft 2000 ft

square-root of reservoir permeability, consistent with


the single planar fracture case above. Figure 9 - Reservoir Model for Complex Planar Fractures
The simulations indicate that near maximum
one-year production is achieved with an FcDc of about 15,000, half that of the single planar case above. The results for the
“communication” and “no communication” cases are the same. This appears to indicate that only the outside fractures are
contributing to production, even when the fractures are communicating. Therefore, in the presence of small-scale fracture
complexity, fracture conductivity requirements may be higher than that for a single planar fracture. For example, if typical
FCD estimates of required fracture conductivity are used and single planar growth is assumed, then the design fracture
conductivity would be 300 mD-ft to achieve an Fcd of 30 for k=0.01 mD and xf=1000 ft. For the fracture complexity used in
this case (Figure 9), fracture conductivity may be reduced by a factor of 5 to 9 compared to a single planar fracture because
the proppant would be spread over a much larger area. The corresponding FcDc values in Figure 9 for k=0.01 mD would
range from 333 (33.3 mD-ft) to 600 (60 mD-ft) and result in 15 to 20% lower gas production if fracture complexity were
present – but not anticipated in the design.

Network Fracture Productivity


Network fracture productivity was evaluated for small and large fracture networks. In the case of small and large fracture
networks, two scenarios were evaluated: (1) uniform fracture conductivity throughout the network and (2) an infinite
conductivity primary fracture that is connected to a finite conductivity network. The two cases are meant to describe the two
extremes for proppant transport in complex fractures, where proppant is evenly distributed throughout the fracture network or
concentrated in a primary planar fracture. Partially propped and un-propped fractures may be important when evaluating the
effects of fracture conductivity on well productivity for network fracture growth.
SPE 115769 11

Small Network Fracture Productivity


Productivity for a small fracture network was
studied using a 300-ft network width (xn) and
fracture spacing (Δxs or Dx) of 25 and 50 ft. For
reference, the total fracture area for the 25-ft and
50-ft spacing is 13 to 25 times larger than the
single planar fracture (xf = 1000 ft for all cases).
The quarter symmetry reservoir simulation model
is shown in Figure 11. The reservoir simulation
results for a small network are summarized in
Figure 12. “Infinite FC” denotes the results for an

1000 ft
150 ft

Network is ON
1000 ft

2000 ft

Figure 11 - Small fracture network model


Figure 10 - Normalized Gp for complex planar fracture growth
infinite conductivity primary fracture, while
“Uniform FC” denotes the results when conductivity is uniform throughout the fracture network. For the uniform
conductivity case, the normalized production will have a minimum that will equate to the one-year cumulative production for
the un-stimulated reservoir. If an infinite conductivity primary fracture is created, the normalized production will have a
minimum that will equate to the one-year cumulative production for an infinite conductivity single planar fracture. The
benefits of exploiting fracture complexity in low permeability reservoirs is illustrated in Figure 12 by comparing the
Normalized Gp for an infinite conductivity single fracture to the Normalized Gp for an infinite conductivity small network
(discussed in more detail later).
In this case the normalized production results are graphed as a function of kfwf/Δxs0.5k0.5, which reasonably collapses the
results for permeability that ranges from 0.0001 to 1.0 mD. The FcDc in this case indicates that the required network fracture
conductivity is proportional to the square-root of fracture spacing and permeability. The fracture conductivity required to
achieve near maximum production is summarized in Table 6.
If an infinite conductivity primary fracture is created, near maximum one-year cumulative production is achieved when
fracture conductivity reaches 150 to 212 mD-ft
(k=1.0 mD), 15 to 21 mD-ft (k=0.01 mD) and 1.5
to 2.0 mD-ft (k=0.0001 mD) (Table 6).
Compared to a small network with uniform
fracture conductivity, the network fracture p G : Infinite Conductivity
conductivity required to maximize first year Single Planar Fracture

production is 100 times lower when an infinite


conductivity primary fracture is present (Table 6).
This indicates that production could be
significantly impacted by proppant transport
behavior when fracture growth is complex.
The presence of an infinite conductivity
primary fracture dominates well performance as
permeability approaches 1 mD. For example,
when reservoir permeability is 1 mD, production
for an infinite conductivity network and a very
low conductivity network differ by only 10%
(Figure 12, Infinite FC, 1 mD) when an infinite
conductivity primary fracture is present. This G : Un-stimulated Reservoir
p

would suggest that controlling fracture Figure 12 - Normalized Gp, small fracture network
complexity should be the design goal when
permeability is on order 1.0 mD. However, for low permeability reservoirs, significant production enhancement above that
of a single infinite conductivity fracture can be realized if sufficient network conductivity can be achieved. For example, an
infinite conductivity single planar fracture in a 0.01 mD reservoir will result in a normalized Gp of about 0.55 or 55% of the
maximum that could theoretically be realized from a small fracture network with 25 to 50 ft fracture spacing. Referring to
12 SPE 115769

Table 6, if a network fracture conductivity of 15 to 21 mD-ft could be achieved in addition to an infinite conductivity primary
fracture or a uniform fracture network with a conductivity of 1,500 to 2,121 mD-ft was created, production could almost be
doubled.

Tight Gas Applications. The Barnett Shale47 and Cotton Valley37 Table 6 – Small Network Conductivity Requirements
treatment sizes are representative of many tight gas stimulation
treatments, with about 1,250 lbm of proppant pumped per foot of Uniform Fracture Conductivity
fracture height. For a hypothetical tight gas stimulation where Conductivity (mD-ft) to Achieve Near-Max Gp
Fracture Spacing
1,250 lbm/ft of proppant is pumped per foot of fracture height and
k (mD) 50 ft 25 ft
a small fracture network is created with a fracture spacing of 50 1 21213 15000
ft and a 1,000-ft half-length (as described above), there would be 0.01 2121 1500
2
26,300 ft of fracture area per foot of fracture height. The 0.0001 212 150
average proppant concentration would be ~0.05 lbm/ft2 if the
proppant were evenly distributed in the fracture network. If the Infinite Conductivity Primary Fracture
proppant were distributed in pillars throughout the network, the Conductivity (mD-ft) to Achieve Near-Max Gp
proppant concentration at each pillar would be much higher, but Fracture Spacing
only a small percentage of the total fracture area would be k (mD) 50 ft 25 ft
1 212 150
propped by the pillars – as was illustrated previously in the 0.01 21 15
Barnett Shale and Cotton Valley examples. 0.0001 2 1.5
Assuming a closure stress of 4,000 psi, sand proppant, and a
Young’s modulus of 5x106 psi, a range of possible fracture conductivity can be estimated by extrapolating the laboratory
results of Fredd, as previously discussed. The stress for a proppant concentration of 0.05 lbm/ft2 would be twice that of the
0.10 lbm/ft2 concentration used in Fredd’s tests, resulting in an effective stress of 8,000 psi on the proppant. Although the
maximum stress in Fredd’s laboratory work is 7,000 psi, it is clear from Figure 5 that the conductivity for a partially propped
fracture at 8,000 psi stress is less than 10 mD-ft and could be as low as 1 mD-ft if the fracture network behaves as an un-
propped displaced fracture – which is possible for such low proppant concentrations.
A permeability range that covers many tight gas reservoirs is 0.0001 to 0.01 mD; 71,84 the corresponding FcDc in Figure 12
for this permeability range and a fracture conductivity of 10 mD-ft is 14 (k=0.01 mD) and 140 (k=0.0001 mD). This would
indicate that when sand is distributed throughout the network (Case 1), the fracture conductivity is insufficient to maximize
first year gas production, resulting in normalized one-year gas production of about 40% (k=0.01 mD) to 80% (k=0.0001 mD)
of the maximum. If the fracture network behaves as a displaced un-propped fracture (kfwf=1 mD-ft), then the corresponding
FcDc in Figure 12 would be ten times lower and the resulting production would be 10% (k=0.01 mD) to 40% (k=0.0001 mD)
of the maximum attainable. If a stronger proppant (bauxite) provides 10 to 100 times the conductivity of sand as suggested
by Figure 5, significant production gains would be achieved. It is important to note that the normalized gas production for an
infinite conductivity single planar fracture is 55% (k=0.01 mD) and 18% (k=0.0001 mD) of the maximum (Figure 12) and
should be used as a baseline for design decisions. Given estimates of the achievable network fracture conductivity and
proppant distribution, if the network production does not exceed that of an infinite conductivity single planar fracture, then
the design should be tailored to minimize fracture complexity. Therefore, network fracture growth could be detrimental to
production when reservoir permeability is 0.01 mD if the proppant is evenly distributed throughout the fracture network, as
sufficient network conductivity may not be achieved with partially sand-propped fractures or displaced un-propped fractures
to surpass the production from an infinite conductivity single planar fracture. However, when permeability is 0.0001 mD,
un-propped fracture conductivity may be sufficient to provide significant production enhancement above that of an infinite
conductivity single planar fracture if some degree of shear-offset can be obtained in the fracture network. If proppant is
selected to accommodate the higher effective stresses with irregularly propped fractures, it may be possible to exploit small-
scale fracture complexity when permeability is 0.01 mD, even if the proppant is distributed evenly throughout the fracture
network.
If all of the proppant were concentrated in a primary fracture (Case 2), the average proppant concentration would be 0.625
lbm/ft2. The resulting reference63 fracture conductivity for 40/70-mesh proppant in this primary fracture would be about 75 to
500 mD-ft for a closure stress range of 4,000 to 6,000 psi, depending on the type of proppant pumped (sand or ceramic). The
realistic retained conductivity (after damage) would be significantly lower.63,64 The reference conductivity for 20/40-mesh
sand would be about 250 to 750 mD-ft, while a 20/40-mesh lightweight ceramic would provide 500 to 1,500 mD-ft. Typical
reductions63,64 for gel damage, cyclic stress, non-Darcy and multiphase flow generally reduce these values by over 90%.
Using the results above for a single fracture (Figure 8), it would require 300 to 3,000 mD-ft to achieve near maximum first
year production for a permeability of 0.0001 to 0.01 mD, respectively. Therefore, it may be possible that a near infinite
conductivity primary fracture could be achieved in many tight gas reservoirs with complex fracture growth if the proppant
were concentrated in the primary fracture (i.e., cannot enter into secondary fractures), proppant concentration or quality were
maximized, and if damage factors were minimized by use of slick-water to reduce gel damage, etc.
Significant production enhancement could be achieved from small fracture networks with effective conductivities of 2 to
20 mD-ft for reservoir permeabilities of 0.0001 to 0.01 mD, respectively, if an infinite conductivity primary fracture were
created (Figure 12). However, the network fractures would be un-propped. Referring to Figure 5 and Figure 6, the expected
SPE 115769 13

conductivity for a displaced un-propped fracture at 4,000 to 6,000 psi closure stress would be 1 to 10 mD-ft for E=6x106 psi.
Therefore, for very low permeability reservoirs like the Barnett Shale, it may be possible to maximize production without
transporting proppant into the fracture network if the network fractures have shear offsets (displaced fracture case). If the
network fractures do not have shear offsets (Figure 5, aligned fracture case) then sufficient conductivity cannot be obtained
with un-propped fractures. For lower modulus rock, it is unlikely sufficient network conductivity can be obtained with un-
propped fractures (Figure 6), even when permeability is 0.0001 mD, as fracture conductivity would likely be less than 0.1
mD-ft at stresses of 4,000 to 6,000 psi.

Moderate Permeability Applications. In the case of a uniform conductivity fracture network in higher permeability
reservoirs, achieving maximum production may not be possible due to limitations in network fracture conductivity. For
example, the network fracture conductivity required to maximize first year production for a 1 mD reservoir is in excess of
15,000 mD-ft (Table 6). For reference, it would require about 2 to 10 lbm/ft2 of 12/18-mesh ceramic proppant to achieve a
conductivity of 15,000 mD-ft at modest closure (4,000 to 6,000 psi) and typical retained conductivity values.61 In many cases,
moderate permeability reservoirs exhibit lower Young’s modulus than their low permeability counterparts, often ranging
from 1 to 3x106 psi. 19,86,87 In this modulus range it is unlikely that sufficient un-propped fracture conductivity could be
retained to exploit fracture complexity (Figure 6). Embedment effects would likely render a partially propped fracture
completely ineffective (Figure 7). Therefore, the only possible way to capitalize on fracture complexity would be a fully
propped fracture, which is very difficult to achieve when complexity is severe.10,11,18,20,21 In this example, a 50-ft network
fracture spacing would result in a fracture area of 26,300 ft2 per foot of fracture height. Given the calculation above for
12/18-mesh ceramic proppant (2 to 10 lbm/ft2), it would require 48,000 to 240,000 lb of proppant per foot of fracture height
to maximize one-year production if fracture complexity were present. Even for modest fracture heights, this is an excessive
amount of proppant and it is probably impossible to place such large quantities at concentrations sufficient to achieve 2 to 10
lbm/ft2 when moderate to high fracture complexity exists.
If an infinite conductivity single planar fracture is created, network fracture conductivity requirements decrease by a
factor of 100 to 150-212 mD-ft (Table 6). However, in a 1 mD
reservoir, achieving significant production improvements by 1000 ft
exploiting fracture complexity is not likely. Given the
conditions of the simulations, an infinite conductivity planar
fracture in a 1 mD reservoir will result in a normalized first 500 ft
year production of 90%. Therefore, even if sufficient network 1000 ft
conductivity could be achieved, first year gas production would
only increase by about 10%; controlling fracture complexity is Fracture Network
the key in moderate permeability reservoirs. Although it is
unlikely that a treatment would be designed to create a 1,000-ft
fracture half-length in most 1 mD reservoirs, fracture half- 2000 ft
lengths of 200 to 300 ft have been reported for this
Figure 13 – Reservoir model for large network
permeability range in chalks and sandstone reservoirs and the
19,86,87
same general conclusions probably apply.

Large Network Fracture Productivity


The reservoir model used to evaluate conductivity requirements for large network fracture productivity is shown in
Figure 13 (¼ reservoir symmetry), consisting of a 1,000-ft network width and a fracture half-length of 1,000 ft. Fracture
spacings of 25, 50, and 100 ft were modeled. The results of the simulations are summarized in Figure 14. With the large
fracture network, the correlating parameter (FcDc) that provides the best collapse of the results is fracture conductivity divided
by the square-root of the fracture spacing and the fourth-root of the reservoir permeability (kfwf/Δxs0.5k0.25). The fourth-root
of permeability may reflect the bilinear nature of flow in the large fracture network, compared to more linear flow that was
represented by the square-root of permeability in the correlating parameter for the small fracture network. The network
conductivity required to achieve near maximum one-year production is summarized in Table 7, indicating very high
conductivity requirements for the lower permeability reservoirs relative to the 1 mD case. For example, the conductivity
requirements for the 0.0001 mD case are only ten times lower than the 1 mD cases, while permeability differs by a factor of
10,000 (Table 7). This is contrary to classical fracture design theory based on dimensionless fracture conductivity (FCD),
where conductivity requirements are directly proportional to reservoir permeability (given the same fracture length).
Near maximum first year production is obtained when FcDc reaches about 1,000 for the large fracture network with
uniform fracture conductivity. If an infinite conductivity primary fracture is created (Case 2), near maximum production is
obtained when FcDc reaches about 100, indicating that ten times less conductivity is required in the network. As with the
small network case, as fracture network conductivity becomes very low and fracture network permeability approaches
reservoir permeability, the normalized production will reach a minimum that is equal to that of an infinite conductivity single
planar fracture (or an un-stimulated reservoir for the uniform conductivity case). The resulting network conductivity
requirements are shown in Table 7, indicating fracture conductivity of 500 to 1,000 mD-ft is required to achieve near
maximum production when reservoir permeability is 0.0001 mD and proppant is uniformly distributed. The conductivity
14 SPE 115769

requirements for a large network are ten times lower when an infinite conductivity primary fracture is created as compared to
a uniform conductivity network (Table 7); however, the production benefits of optimizing fracture network conductivity are
dependent on reservoir permeability. Figure 14 shows that an infinite conductivity single planar fracture (no network) will
result in a normalized one-year gas production of less than 10% of the maximum when permeability is 0.0001 mD, about
26% of the maximum for 0.01 mD, and 83% of the maximum for 1 mD. As with the small network case, the benefits of
exploiting fracture complexity are mostly limited to low permeability reservoirs (0.01 to 0.0001 mD).
Comparing the conductivity requirements for
large (Table 7) and small (Table 6) networks
shows that much higher network fracture
conductivity is required when reservoir
permeability is lower for the large network to
maximize first year production. For example, a G : Infinite Conductivity
p

large network with 50 ft spacing in a 0.0001 mD Single Planar Fracture

reservoir would require 707 mD-ft to achieve


near maximum first year production, while only
212 mD-ft are required when the network is
small. The near maximum one-year production
for the large network will be much higher than
the small network, emphasizing the importance
of achieving large fracture networks with
adequate conductivity (if possible) in very low
permeability reservoirs.

Tight Gas Applications. Suppose the same


treatment that was pumped in the small network G : Un-stimulated Reservoir
p

evaluation actually develops a large network.


Assuming a fracture spacing of 50 ft and 1,250 Figure 14 - Normalized G , large network
p
lbm of proppant pumped per foot of fracture
2
height, the area of the large network would be 83,000 ft per
foot of fracture height, resulting in an average proppant Table 7 – Large Network Conductivity Requirements
concentration of 0.015 lbm/ft2 (Case 1) if the proppant were
evenly distributed in the fracture network. Assuming a Uniform Fracture Conductivity
closure stress range of 4,000 to 6,000 psi and hard rock Conductivity (mD-ft) to Achieve Near-Max Gp
6 Fracture Spacing
(E=5x10 psi), fracture conductivity can be estimated from
k (mD) 100 ft 50 ft 25 ft
Figure 5 using an effective stress on the proppant grains that is
2 2 1 10000 7071 5000
6.6 times (0.1 lbm/ft /0.015 lbm/ft ) larger than was 0.01 3162 2236 1581
represented in Fredd’s lab tests. Using this approximation of 0.0001 1000 707 500
effective stress, the stress on 0.015 lbm/ft2 appears to be
excessive and it is unlikely that the proppant would contribute Infinite Conductivity Primary Fracture
materially to overall fracture conductivity. Therefore, the best Conductivity (mD-ft) to Achieve Near-Max Gp
estimate of the network conductivity would be the un-propped Fracture Spacing
conductivity at 4,000 to 6,000 psi in Figure 5, which is about k (mD) 100 ft 50 ft 25 ft
1 to 10 mD-ft for an un-propped displaced fracture. As 1 1000 707 500
discussed previously, the distribution of the proppant in 0.01 316 224 158
pillars (Case 3) would result in a very small percentage of the 0.0001 100 71 50
overall fracture area propped (less than 2% for most large
networks) and may also behave as an un-propped fracture. Using a typical permeability range in tight and unconventional
gas reservoirs, from Table 7 the required fracture conductivity to achieve near maximum first-year production is about 700
mD-ft (k=0.0001 mD) to 2,250 mD-ft (k=0.01 mD). Therefore, near maximum first-year production cannot be achieved with
a uniform distribution of proppant (Case 1) or pillar distribution (Case 3).
Assuming a network fracture conductivity of 1 to 10 mD-ft, the corresponding “large network” FcDc values would be 0.44
to 4.4 (k=0.01 mD) and 1.4 to 14 (k=0.0001 mD). Using Figure 14, the resulting normalized first-year production would be
about 5 to 25% of the maximum if reservoir permeability is 0.01 mD and 10 to 50% of the maximum for a reservoir
permeability of 0.0001 mD. In the case of a large network with evenly distributed proppant that behaves like displaced un-
propped fractures, stimulation effectiveness would likely be moderate to good for very low permeability reservoirs (i.e.,
shale) compared to an infinite conductivity simple planar fracture, which would result in a normalized first year gas
production of about 0.10 or 10% (Figure 14). However, for tight gas reservoirs with a permeability of 0.01 mD, stimulation
effectiveness would be poor, as the expected normalized first-year gas production for an infinite conductivity planar fracture
would be 0.26 or 26% (Figure 14) - the same production as a large network with a uniform fracture conductivity of 10 mD-ft.
Therefore, when reservoir permeability is around 0.01 mD, large fracture networks with uniform conductivity are likely
SPE 115769 15

detrimental to production, while for much lower permeability reservoirs they are probably beneficial to production, based on
the expected conductivity of 1 to 10 mD-ft.
If shear offset is not achieved and the conductivity is not represented by Fredd’s displaced un-propped fracture data, the
network fracture conductivity would be much less than 0.1 mD-ft (aligned fracture, Figure 5), resulting in normalized first
year gas production of 2% or lower (Figure 14). In the case of evenly distributed proppant that behaves like aligned un-
propped fractures, stimulation effectiveness would be very low and fracture complexity would be detrimental – suggesting
that controlling fracture complexity and creating a more planar hydraulic fracture would be desirable. In addition, for lower
modulus rock, even displaced fractures will likely provide insufficient network fracture conductivity (Figure 6) and fracture
complexity should be minimized in these cases.
Now let’s consider the case where an infinite conductivity primary fracture is created and proppant presumably is
prohibited entry into the distant network. In this case, network conductivity will be dominated by un-propped fractures,
either displaced or aligned. Un-propped displaced fractures could exhibit conductivities of 1 to 10 mD-ft. For very low
permeability reservoirs like the Barnett Shale, the FcDc values would range from 1.4 to 14 (k=0.0001 mD, 50 ft fracture
spacing, 1 to 10 mD-ft un-propped network conductivity) and result in first year production that is 75 to 95% of the
maximum (Figure 14). When reservoir permeability is 0.01 mD, the corresponding FcDc values for 50 ft spacing and 1 to 10
mD-ft un-propped network conductivity are 0.45 to 4.5 and result in production that is 50 to 80% of the first year maximum.
Therefore, if shear offsets are present, providing un-propped fracture conductivity of 1 to 10 mD-ft, it may be possible to
significantly improve production if large un-propped fracture networks can be created in addition to an infinite conductivity
primary fracture. As discussed previously, in the absence of shear offsets (displaced fractures), aligned un-propped fracture
conductivity would be less than 0.1 mD-ft and result in insufficient conductivity to achieve even modest production benefits
from network fracture growth.
Moderate Permeability Applications. As discussed for small networks, with reservoir permeability of 1 mD or higher,
achieving significant production improvements above the infinite conductivity single planar fracture is not likely. Given the
conditions of the simulations, an infinite conductivity planar fracture in a 1 mD reservoir will result in a normalized first year
production of 83% of what could be achieved with a large fracture network (Figure 14). Even if an infinite conductivity
primary fracture and a large network with sufficient network conductivity could be achieved (500 to 1,000 mD-ft), first year
gas production would only increase by about 17%. Although this may equate to a large amount of gas, it is probably not
possible to achieve both the desired primary fracture and network conductivity. Fluid loss will likely limit the ability to
create large networks in moderate permeability reservoirs, as will the high viscosity fluids needed to place the proppant
concentrations required to achieve the desired conductivity. Therefore, controlling fracture complexity is again the key in
moderate permeability reservoirs.

Fluid Leakoff and Fracture Area


As fracture area and reservoir permeability increase, fluid loss rates will increase dramatically. The low viscosity fluids
specified to create large fracture networks generally omit gelling and wall-building components, thus fluid loss is dominated
by reservoir properties and fracturing conditions. The ability to achieve large fracture networks will be limited by fluid loss
behavior and there will be practical limits to the maximum fracture area that can be created. Figure 15 illustrates the effect
of fluid loss on approximate maximum
1.00E+09 fracture area for a wide range of fluid loss
48,000 bbl
24,000 bbl
coefficients, assuming an injection rate of
12,000 bbl 100 BPM. The maximum fracture area
Q=100 BPM was estimated by calculating the area that
1.00E+08 resulted in a leakoff rate equal to the
Maximum Fracture Area (ft2)

injection rate of 100 BPM. The likely


range of fluid loss coefficients that would
be expected for a reservoir permeability
k ~0.0001 mD
1.00E+07 of 0.0001 mD and 0.01 mD are shown in
the figure69, allowing an approximation of
the maximum area that could be achieved
for injection volumes of 12,000, 24,000,
1.00E+06
k ~0.01 mD and 48,000 bbl.
When permeability is very low,
fracture networks in excess of 100 million
square feet might be possible with large
1.00E+05
injection volumes. Referring to the
0.00001 0.00010 0.00100 0.01000 Barnett Shale example that was
Ct (ft/min1/2) previously presented, network fracture
areas of 24 to 360 million square feet
were estimated depending on fracture
Figure 15 - Maximum fracture area as a function leakoff coefficient
spacing for a typical Barnett Shale
16 SPE 115769

fracture treatment (Table 2). This is consistent with the approximate range of fracture area from Figure 15 and the expected
permeability (k ~ 0.0001 mD) and injected volume (60,000 bbl).
As permeability increases to 0.01 mD, the expected maximum fracture area for a slick-water treatment decreases to about
2 to 20 million square feet. Fracture area ranged from 0.775 to 3.1 million square feet, depending fracture complexity, in the
Cotton Valley example discussed previously (Table 3) for a 10,000-bbl hybrid treatment. As permeability approaches 1 mD,
fluid loss control and increased viscosity are required to reduce leakoff and improve proppant transport to allow successful
placement of sufficient proppant concentrations to achieve adequate fracture conductivity. In moderate permeability
reservoirs (~1 mD), generating even modest network fracture growth is unlikely with low viscosity fluids due to excessive
fluid leakoff.
For reference, the fracture area for the small network example with a 300-ft fracture height would be 8 to 15 million
square feet (25 and 50 ft fracture spacing). For the same fracture height and fracture spacing, the fracture area would be 25 to
49 million square feet for the large network example. Figure 15 illustrates that generating large fracture networks with slick-
water may be possible only when reservoir permeability is less than 0.001 mD. As permeability approaches 0.01 mD, fluid
loss will control maximum fracture area and limit treatments designs to small network growth.

Diagnosing Fracture Complexity


Although it is difficult to uniquely characterize fracture complexity, it may be helpful to evaluate the degree of fracture
complexity for various geologic environments. A simple approximation of fracture complexity is given by Equation 5.
xn
FCI = ...................(5)
2x f
Figure 16 is a graph of published microseismic fracture mapping data showing the total width of the microseismic cloud
versus the total length of the cloud. Several slopes are shown for reference. Network complexity appears to occur in the 0.25
to 0.5 range, but it is not very clear when the fracture is small because of normal microseismic activity that occurs away from
the fracture, as well as uncertainty in event
location. Some of the network width in the
1600
FCI = 1 0.5 smaller fractures is due to interaction with faults.
1400 Note that there is some interpretation required, as
0.25 some cases do not image the entire fracture
1200 length, in which case the total fracture length was
assumed to be twice the length of the observed
Total Width (ft)

1000
Bossier
wing. Figure 16 illustrates that many geologic
Grand Valley
800
Rulison
environments may exhibit some degree of
600 West Tavaputs fracture complexity, with the shale plays
Barnett generally exhibiting the highest degree of
FCI = 0.1 Devonian Sands
400 complexity and the tight gas sands low to
Canyon Sands moderate complexity. Unfortunately, a
200 Cotton Valley
microseismic event “cloud” will typically have a
Lance/Mesaverde
0 minimum width of 100 to 200 ft. Therefore, for
0 1000 2000 3000 4000 5000 short hydraulic fractures (i.e., less than 1,000 ft)
Total Length (ft) FCI may over-predict the degree of fracture
complexity. An alternate approach may be to
simply evaluate the width of the microseismic
Figure 16 - Fracture Complexity Index (FCI) for various geologic
event pattern; when the width of the event pattern
environments
exceeds 200 ft, fracture complexity and/or
network growth may be present.

Net Pressure & Fracture Complexity


Net pressure data may provide qualitative insights into fracture complexity when microseismic data are not available or
the microseismic event pattern cannot distinguish between simple planar and complex planar fracture growth. Nolte and
Smith80 provided a simple estimate of fissure opening pressure based on the net pressure in the fracture and the difference
between the minimum and maximum horizontal stress and Poisson’s ratio (Equation 6).
Δσ h
Po = ...................(6)
(1 − 2v)
Warpinski2 showed fissure opening at net pressures of 850 to 1,050 psi in tight gas sands in the Piceance Basin. The
actual fissure opening pressures were somewhat lower that predicted by Equation 6, but still in reasonable agreement.
Therefore, in cases where a pre-existing rock fabric is present, the potential for fracture complexity may be estimated from
net pressure data if the maximum horizontal stress has been determined. The analysis of minifrac pressure decline data can
also provide estimates of fissure opening pressures and insights into fracture complexity.81,82
SPE 115769 17

Fracture complexity can also be estimated by evaluating the level of net pressure using fracture modeling techniques.
Although current fracture models cannot rigorously model fracture complexity, it may be possible to approximate fracture
complexity by modeling a system of interacting parallel fractures.9,19 This approach provides estimates of proppant
distribution that can be used to evaluate fracture conductivity requirements and the impact of fracture complexity on well
productivity. In the event fracture complexity cannot be determined from fracture mapping measurements, the application of
fracture modeling and net pressure analyses may provide insights into fracture growth that are critical when selecting the
appropriate treatment design.

Summary
The tendency for complex fracture growth is most likely a function of the pre-existing rock fabric and the stress regime.
Reservoirs with a high concentration of natural fractures or fissures are more prone to complex fracture growth. Complexity
may be encouraged by dilation or shear failure depending on the interaction between the stress regime, natural fractures, and
hydraulic fracture. Fracture complexity can be reduced by pumping high viscosity fluids and maximized by pumping low
viscosity fluids. The decision whether to control or enhance fracture complexity will depend on reservoir permeability,
proppant distribution, and desired fracture conductivity.
The fracture conductivity required to achieve near
Table 8 - Conductivity Requirements to Achieve Near
maximum first year gas production varies considerably
Maximum First Year Gas Production
depending on reservoir permeability, the degree of fracture
Reservoir Permeability complexity, fracture spacing, and network width. Table 8
Case
0.0001 mD 0.01 mD 1 mD shows the fracture conductivity required to achieve near
Planar 300 3000 30000 maximum first year gas production for the cases evaluated,
Complex Planar 150 1500 15000 assuming 50-ft fracture spacing for network growth. The
Small Network (U)
1
212 2100 21200 table shows that conductivity requirements do not follow
Small Network (I) 1
2 21 212
simple dimensionless conductivity design guidelines83 where
1 conductivity requirements are directly proportional to
Large Network (U) 707 2240 7070
1 reservoir permeability (for a given fracture length). When
Large Network (I) 71 224 707
fracture complexity is modest, conductivity requirements are
U = uniform conductivity proportional to k1/2. For large fracture networks,
I = infinire conductivity primary fracture conductivity requirements are proportional to k1/4. In the
Note 1 - 50-ft fracture spacing cases case of network fracture growth, conductivity requirements
are also proportional to xs1/2, indicating that less conductivity
is required as fracture spacing decreases.
When fracture growth is complex but still planar in nature (complex planar), fracture conductivity requirements are half
that of a single planar fracture; however, if the proppant is distributed evenly within fractures, the average proppant
concentration and fracture conductivity would decrease as complexity increases. Therefore, to compensate for the lower
average proppant concentration, the treatment design will require more focus on delivering fracture conductivity. For
network fracture growth, if the proppant is concentrated in a single planar fracture with infinite conductivity, then the
network conductivity requirements decrease by a factor of 100 for small networks (300 ft wide) and by a factor of 10 for
large networks (1,000 ft wide). Arguably, this may be a plausible scenario for proppant transport (Case 2) in water-fracs and
low-viscosity hybrid fracture treatments, which implies that un-propped fracture conductivity may play a critical role in well
productivity.
To achieve 90% of the maximum first year production,
Table 9 - Conductivity Requirements to Achieve 90% of
fracture conductivity requirements are significantly less than
Maximum First Year Gas Production
that required to achieve near maximum first-year production
(Table 9) and may provide additional insights for fracture Network Conductivity Requirments to Obtain 90% of
design. For a large fracture network with an infinite Maximum 1st-year production
conductivity primary fracture, 90% of the maximum first-
Reservoir Permeability
year gas production can be achieved with a network Case
0.0001 mD 0.01 mD 1 mD
conductivity 2.83 to 22.4 mD for low permeability
reservoirs (0.0001 to 0.01 mD). When fracture growth is Planar 20 500 10000
Complex Planar 10 100 1000
less complex (small network), the required network 1
conductivity to achieve 90% of the maximum first year Small Network (U) 25 212 1768
1
production is only 0.35 to 3.5 mD-ft for the same Small Network (I) 0.35 3.5 NA
1
permeability range (Table 9). Therefore, it may be possible Large Network (U) 71 224 1414
1
to exploit fracture complexity in low permeability reservoirs Large Network (I) 2.83 22.4 50
with an infinite conductivity primary fracture if un-propped U = uniform conductivity
fracture conductivity is 1 to 20 mD-ft. Un-propped fracture I = infinire conductivity primary fracture
conductivity of 1 to 10 mD-ft may be possible in many tight Note 1 - 50-ft fracture spacing cases
gas environments if shear offsets result in displaced
fractures (Figure 5). However, if an infinite conductivity
18 SPE 115769

primary fracture is not present and the fracture conductivity is uniform throughout the network, the fracture conductivity
required to achieve 90% of the maximum first-year production is increased by a factor of 10 to 100 depending on network
size. In the case of uniform network conductivity, it is unlikely that fracture conductivity will be sufficient to fully exploit
fracture complexity – with the exception of small networks in very low permeability reservoirs. However, even when
network conductivity is significantly lower than the values shown in Table 9, significant production enhancement may still be
possible by exploiting fracture complexity in many low permeability reservoirs.

Fracture Design Guidelines When Fracture Growth Is Complex


The primary decision when designing fracture treatments in reservoirs that exhibit complex fracture growth is whether to
control or exploit the complexity. This decision is primarily a function of reservoir permeability and achievable fracture
conductivity, but also depends on Young’s modulus, the degree of complexity, and the likely proppant transport scenario. In
general, fracture complexity should be minimized in moderate permeability reservoirs (k > 1 mD), as there is very little
potential benefit from exploiting complexity and the fracture conductivity requirements are high. However, for low
permeability reservoirs (k < 0.01 mD), exploiting fracture complexity may result in significant production enhancements.
Fracture network size may be limited by excessive fluid loss, even when permeability is low (k~0.01 mD). The effects of
fluid loss and network conductivity requirements will probably dictate that small networks are optimal when permeability is
on order 0.01 mD and large networks are optimal when permeability is on order 0.0001 mD.
The following very general guidelines are provided for treatment design in reservoirs that exhibit complex fracture
growth. However, it should be emphasized that designing treatments when fracture growth is complex is a difficult process
and not easily defined using simple guidelines.

• Determine the reservoir permeability, Young’s modulus, and tendency for complex fracture growth. Fracture
mapping can distinguish between network and planar fracture growth and also provide direct measurements of
fracture length and height; however, fracture pressure analysis will likely be the primary method to distinguish
between simple planar and complex planar growth. Fracture spacing can be estimated using reservoir modeling
given the Stimulated Reservoir Volume (SRV) from fracture mapping and estimates of matrix permeability.29,30,47
The degree of complex planar fracture growth can be estimated using net pressure history matching, given fracture
length and height from fracture mapping44 or less reliably using net pressure matching alone.9,18-21 It may be
necessary to evaluate the effect of fluid viscosity on fracture complexity.
• If permeability is 1 mD or greater, minimize fracture complexity by pumping viscous fluids and particulate slugs.
Optimize fracture conductivity based on reservoir permeability and fracture complexity. When pumping viscous
fluids and particulate slugs it is unlikely that network fracture growth will be achieved and fracture conductivity
requirements can be estimated based on fracture modeling estimates of proppant concentration assuming complex
planar growth (i.e., number of equivalent multiple fractures).
• If permeability is 0.01 mD or less and shear offsets are possible, estimate un-propped fracture conductivity based on
closure stress and Young’s modulus (Figure 6 or lab measurements on actual core). Using the un-propped
conductivity, determine whether the design should target small or large network growth based on the expected
production. Evaluate the production for a uniform conductivity network and a network with an infinite conductivity
primary fracture to determine the appropriate proppant transport goal (although this may be very difficult to control).
If un-propped fracture conductivity is insufficient to effectively exploit network fracture growth OR shear offset is
unlikely, determine if sufficient propped or partially propped fracture conductivity can be achieved to exploit small
network growth based on closure stress and embedment (Figure 5, Figure 7). It is assumed that sufficient proppant
cannot be placed to effectively prop a large network. If sufficient propped or partially propped conductivity cannot
be achieved for small network growth, then attempt to restrict fracture complexity and target the design for complex
planar growth.
• Generating large fracture networks (water-fracs). Pump very large volumes of low viscosity fluid (slickwater) to
maximize fracture complexity and generate large networks. In addition, low viscosity fluid will provide the best
cleanup behavior. In most cases it is desirable to target an infinite conductivity primary fracture to reduce network
conductivity requirements. Achieving an infinite conductivity primary fracture may require pumping relatively
large amounts of proppant. Proppant type, concentration, and size will be dictated by closure stress, conductivity
requirements, and proppant placement efficiency. Smaller proppants will likely be transported further in the primary
fracture and possibly modest distances into the fracture network, but higher permeability proppants may be required
to achieve an effectively infinite conductivity primary fracture.
• Generating small fracture networks (hybrid fracs). Pump large volumes of moderate viscosity fluid (linear gel,
lower viscosity cross-linked fluid, or both) to moderate fracture complexity and improve proppant transport
characteristics. Linear gel may provide adequate cleanup behavior in small networks. Ensure that an infinite
conductivity primary fracture is obtained by pumping larger or higher quality proppants at greater concentrations
compared to water-frac designs. In cases where obtaining an infinite conductivity in the primary fracture is difficult,
SPE 115769 19

consider using cross-linked gel in the later proppant stages, which should allow higher proppant concentrations and
improve proppant transport – while also promoting primary fracture propagation.
• Completion schemes to increase complexity & fracture intensity. In many cases complexity can be enhanced by
stimulating multiple perforation clusters in a horizontal well. Microseismic mapping data often shows a much wider
zone of activity in such cases, although whether it is due to multiple parallel fractures or some network development
is often difficult to discern. Similarly, simultaneous fracturing, sequential fracturing, fracturing of closely spaced
stages, and other strategies to cause interaction between different fracture treatments is another method used to
intensify the degree of fracturing. These procedures may help to divert fluid into new fractures that they might not
otherwise have opened or penetrated, and mapping does appear to support that premise. The application of small
diameter proppant particles, such as 40/70 or 100-mesh, may also help intensify fracturing by bridging existing
fractures and forcing new ones to open.
• Obtaining sufficient network fracture conductivity. In cases where sufficient fracture conductivity cannot be
obtained from un-propped fractures, consider pumping small diameter, higher strength proppants such as 70/140-
mesh ceramic proppant. The higher strength proppant, combined with the much smaller diameter (compare to
40/70-mesh), may allow partial propping of network fractures further away from the wellbore.

Conclusions
The following conclusions are based on a set of reservoir simulations that cover a wide range of reservoir permeability and
fracture conductivity, but only a limited range of fracture growth and reservoir geometry (i.e., fixed fracture half-length and
reservoir size). In addition, the evaluations focused on comparisons of first year cumulative production. However, the trends
identified in this work should be applicable to a wide range of situations, although the quantitative results will vary
depending of the specifics of each case.

1. Fracture complexity may be affected by fluid viscosity. In reservoirs prone to complex fracture growth, complexity
can be reduced by pumping high viscosity fluid or maximized by pumping low viscosity fluid.
2. For any given treatment, as fracture complexity increases, average proppant concentration will decrease resulting in
lower fracture conductivity, increased embedment effects, and less efficient cleanup.
3. With modest stresses and relatively hard formations, proppant characteristics (strength, size, resistance to
embedment) have been shown to be key to retaining flow capacity with low proppant concentrations.
4. When low proppant concentrations are placed in high stress or relatively soft formations, stress concentration,
particle crushing, and embedment may render such fractures only marginally more effective than un-propped
fractures.
5. If proppant is evenly distributed throughout fracture networks, the resulting proppant concentrations are probably
inadequate to materially contribute to network fracture conductivity. The amount of proppant pumped in typical
slick-water and hybrid fracture treatments will result in average proppant concentrations << 0.1 lbm/ft2 or a propped
area << 10%. In this scenario, it is probable that the majority of the created fracture area has residual conductivity
similar to an un-propped fracture.
6. Although un-propped fracturing treatments are successful in some reservoirs, the production results from a wide
variety of formations69,85,91-100 indicate that stimulation treatments containing proppant generally outperform
treatments with no proppant. This could be evidence that, if fracture growth is complex, the proppant is
concentrated in a primary fracture or is restricted to limited penetration distances into the fracture network.
However, proppant could be providing other benefits (erosion, diversion, etc.) not addressed herein.
7. Un-propped fracture conductivity will likely be insufficient to exploit fracture complexity in the absence of shear
offsets (displaced fractures). In addition, even with shear offsets, un-propped fracture conductivity will likely be
insufficient for Young’s modulus of 2 x 106 psi or lower, as deformation of asperities will be excessive and likely
result in closure of un-propped fractures.
8. Fracture complexity should normally be controlled when Young’s modulus is less than 2 x 106 psi. As Young’s
modulus decreases, the ability to exploit fracture complexity becomes more challenging due to the inability to
generate sufficient un-propped or partially propped fracture conductivity due to proppant embedment effects and
asperity deformation. It is likely that complex fracture growth cannot be exploited when Young’s modulus is less
than 2 x 106 psi and may be detrimental to stimulation effectiveness due to excessive leakoff and proppant
placement problems.
9. Partially propped fracture conductivity may play an important role in production enhancement when fracture
complexity is moderate if proppant can be transported into secondary fractures. The application of smaller, higher
strength proppants may be the key to exploiting moderately complex fracture growth in low permeability reservoirs.
However, proppant embedment will likely be excessive (i.e., greater than one particle diameter) when Young’s
modulus is ≤ 2 x 106 psi for partial monolayer proppant distributions, resulting essentially in un-propped fractures.
20 SPE 115769

10. When fracture complexity is moderate to high (network growth) and proppant is not or cannot be transported into
the network, creating a high conductivity primary fracture will significantly reduce network conductivity
requirements. If the primary fracture exhibits infinite conductivity behavior, network conductivity requirements
decrease by a factor of 10 (large networks) to 100 (small networks).
11. In moderate permeability reservoirs (~ 1 mD), fracture complexity should typically be minimized.
12. In tight reservoirs (~0.01 mD) with relatively hard rock (E=4-6 x 106 psi), significant production enhancements may
be possible by exploiting moderately complex fracture growth (small networks). In addition, when permeability is
~0.01 mD, excessive fluid loss will likely limit fracture area and make it impractical or impossible to create very
large networks.
13. In very tight reservoirs (~0.0001 mD) with relatively hard rock (E=4-6 x 106 psi), significant production
enhancements may be possible by exploiting very complex fracture growth (large networks).
14. Increasing fracture complexity by generating a more dense fracture network (smaller fracture spacing) reduces
fracture conductivity requirements, but challenges our capability to effectively place proppant throughout the
fracture.

Acknowledgments
The authors would like to thank Pinnacle Technologies and Carbo Ceramics for supporting the publication of this work.

Nomenclature
Af = fracture area, one face (subscripts denote: c=complex, p=planar, propped=fraction of fracture area propped), L2
Cp = Proppant concentration (Case denoted by subscript 1, 2, or 3) M/L2
d = particle diameter, L
E = Young’s modulus
FCD = dimensionless conductivity
FcDc = complex fracture conductivity correlating function
FCI = fracture complexity index
g = gravity, L/T2
hf = fracture height, L
k = permeability, L2
khf = hydraulic fracture permeability, L2
Mp = Proppant pumped, M
Po = Fissure opening pressure, M/LT2
t = time, T
Vfl = Volume of fluid pumped, L3
VS = settling velocity, L/T
wf = fracture width (subscripts denote: p=planar, c=complex), L
xf = hydraulic fracture wing or half-length, L
xn = hydraulic fracture network width (from microseismic event pattern), L
Δxs,Dx = orthogonal fracture spacing, L
y = distance fluid penetrates into a natural fracture or fissure, L
ε = fluid efficiency
ΔP = pressure drop, specifically treating pressure minus reservoir pressure, M/LT2
Δσh = Difference between minimum and maximum horizontal stress, M/LT2
v = Poisons’ ratio
μ = fluid viscosity, M/LT
ρf = fluid density, M/L3
ρp = particle density, M/L3
ϕ = porosity
σ = stress, M/LT2

SI Metric Conversion Factors


acre x 4.046 873 e+03 = m2
bbl x 1.589 874 e-01 = m3
cp x 1.0 e-03 = Pa.s
ft x 3.048 e-01 = m
o
F (oF – 32)/1.8 = o
C
lbm/gal x 1.198 264 e+02 = kg/cm2
psi x 6.894 757 e+00 = kPa
SPE 115769 21

References
1. Warpinski, N.R., and Teufel, T.W.: “Influence of Geologic Discontinuities on Hydraulic Fracture Propagation,” JPT (February
1987), 209.
2. Warpinski, N.R.: “Hydraulic Fracturing in Tight, Fissured Media,” JPT (February 1991) 146.
3. Warpinski, N.R., et al.: “Examination of a Cored Hydraulic Fracture in a Deep Gas Well,” SPEPF (August 1993), 150.
4. Fast, R.E., Murer, A.S., and Timmer, R.S.: “Description and Analysis of Cored Hydraulic Fractures, Lost Hills Field, Kern
County, California,” paper SPE 24853 presented at the 1992 SPE Annual Technical Conference and Exhibition, Washington, DC,
October 4-7.
5. Steidl, P.F.: “Observations of Induced Fracture Intercepted by Mining in the Warrior Basin, Alabama, Rock Creek Methane from
Multiple Coal Seems Completion Project,” Report No. GRI-91/0327 (1991).
6. Branagan, P.T., et al.: “Characterization of Remotely Intersected Set of Hydraulic Fractures: Results of Intersection Well No. 1-B,
GRI/DOE Multi-site Project,” paper SPE 36452 presented at the 1996 SPE Annual Technical Conference and Exhibition, Denver,
Colorado, October 6-9.
7. Wright, C.A.: “On-Site Step-Down Test Analysis Diagnoses Problems and Improves Fracture Treatment Success,” Hart’s
Petroleum Engineer International (1997).
8. Cleary, M.P., et al.: “Field Implementation of Proppant Slugs to Avoid Premature Screenout of Hydraulic Fractures with Adequate
Proppant Concentration,” paper SPE 25982 presented at the Joint Rocky Mountain Regional Meeting and Low Permeability
Reservoirs Symposium, Denver, Colorado, April 1993.
9. Sato, K., Wright, C.A., and Ichikawa, M.: “Post-Fracture Analyses Indicating Multiple Fractures Created in a Volcanic
Formation,” SPEPF (November 1999) 284.
10. Weijers, L., et al.: “The First Successful Fracture Treatment Campaign Conducted in Japan: Stimulation Challenges in a Deep,
Naturally Fractured Volcanic Rock,” paper SPE 77678 presented at the 2002 Annual Technical Conference and Exhibition, San
Antonio, Texas, September 29 – October 2.
11. Weijers, L., Wright, C.A., Sugiyama, H., Sata, K., and Zhigang, L.: “Simultaneous Propagation of Multiple Hydraulic Fractures –
Evidence, Impact and Modeling Implications,” paper SPE 64772 presented at the 2000 International Oil and Gas Conference and
Exhibition, Beijing, November 7-10.
12. Shlyapobersky, J., Wong, G.K., and Walhaug, W.W.: “Overpressure Calibrated Fracture Design in Hydraulic Fracture
Stimulations,” paper SPE 18194 presented at the 1988 Annual Technical Conference and Exhibition, Houston, October 2-5.
13. Gas Research Institute Technical Summary GRI-95/0258 on “Multiple Hydraulic Fractures” (1995).
14. Wright, C.A., et al.: “Real-Time and Post-Frac 3-D Analysis of Hydraulic Fracture Treatments in Geothermal Reservoirs,”
presented at the 1994 Stanford Geothermal Conference, Stanford U., Palo Alto, California, January.
15. Stadulis, J.M.: “Development of a Completion Design to Control Screenouts Caused by Multiple Near-Wellbore Fractures,” paper
SPE 29594 presented at the 1995 Rocky Mountain Regional/Low Permeability Reservoirs Symposium and Exhibition, Denver,
Colorado, March 19-22.
16. Wright, C.A., Tanigawa, J., Shixin, M., and Zhigang, L.: “Enhanced Hydraulic Fracture Technology for a Coal Seam Reservoir in
Central China,” paper SPE 29989 presented at the 1995 International Oil and Gas Exhibition, Beijing, November 14-17.
17. Minner, W.A., Molesworth, G.R., Wright, C.A., and Wood, W.D.: “Real-Data Fracture Analysis Enables Successful Hydraulic
Fracturing in the Point of Rock Formation, Kern County, California,” paper SPE 38326 presented at the 1997 Western Regional
Meeting, Long Beach, June 25-27.
18. Cipolla, C.L., Jensen, L., Ginty, W., and de Pater, C.J.: “Complex Hydraulic Fracture Behavior in Horizontal Wells, South Arne
Field, Danish North Sea,” paper SPE 62888 presented at the 2000 SPE Annual Technical Conference, Dallas, Texas, 1–4 October.
19. Cipolla, C.L., Berntsen, B.A., Moos, H., Ginty, W.R., and Jensen, L.: “Case Study of Hydraulic Fracture Completions in
Horizontal Wells, South Arne Field Danish North Sea,” paper SPE 64383 presented at the SPE Asia Pacific Oil & Gas Conference
and Exhibition, Brisbane, Australia, 16-18 October.
20. Cipolla, C.L., et al.: “Fracture Treatment Design in Low Porosity Chalk Reservoirs,” paper SPE 86485 presented at the SPE
International Symposium and Exhibition on Formation Damage Control held in Lafayette, Louisiana, 18-20 February 2004.
21. Cipolla, C.L., et al.: “Evolution of FracPack Design, Modeling, and Execution in the Ceiba Field, Equatorial Guinea,” paper SPE
95514 presented at the 2005 SPE Annual Technical Conference, Dallas, Texas 9-12 October.
22. Weijers, L.: The Near-Wellbore Geometry of Hydraulic Fractures Initiated from Horizontal and Deviated Wells, Ph.D. thesis Delft
UT, Delft University Press (February 1995) 189.
23. Warpinski, N.R., Branagan, P.T., Sattler, A.R., Cipolla, C.L., Lorenz, J.C., and Thorne, B.J.: “A Case Study of a Stimulation
Experiment in a Fluvial, Tight, Sandstone Gas Reservoir,” SPE Production Engineering, Vol. 4, pp. 403-410, November 1990.
24. Warpinski, N.R.: “Dual Leakoff Behavior in Hydraulic Fracturing of Tight, Lenticular Gas Sands,” SPE Production Engineering,
Vol. 4, pp. 243-252, August 1990.
25. Britt, L.K., Hager, C.J., and Thompson, J.W.: “Hydraulic Fracturing in a Naturally Fractured Reservoir,” SPE 28717 presented at
the SPE International Petroleum Conference & Exhibition of Mexico, Veracruz, Mexico, 10-13 October 1994.
26. van Batendburg, D.W. and Hellman, T.J.: “Guidelines for the Design of Fracture Treatments for Naturally Fractured Formations,”
SPE 78320 presented at the SPE 13th European Petroleum Conference, Aberdeen, Scotland, 29-31 October 2002.
27. Cipolla, C.L., et al.: “State-of-the-Art in Fracture Diagnostics,” paper SPE 64434 presented at the 2000 SPE Asia Pacific Oil and
Gas Conference and Exhibition, Brisbane, Australia, 16-18 October.
28. Fisher, M.K., Davidson, B.M., Goodwin, A.K., Fielder, E.O., Buckler, W.S., and Steinberger, N.P.: “Integrating Fracture
Mapping Technologies to Optimize Stimulations in the Barnett Shale,” paper SPE 77411 presented at the 2002 SPE Annual
Technical Conference and Exhibition, San Antonio, Texas, September 29-October 2.
29. Maxwell, S.C., Urbancik, T.I., Steinsberger, N.P., and Zinno, R.: “Microseismic Imaging of Hydraulic Fracture Complexity in the
Barnett Shale,” paper SPE 77440 presented at the 2002 SPE Annual Technical Conference and Exhibition, San Antonio, Texas,
September 29-October 2.
22 SPE 115769

30. Fisher, M.K., Heinze, J.R., Harris, C.D., Davidson, B.M, Wright, C.A., and Dunn, K.P.: “Optimizing Horizontal Completion
Techniques in the Barnett Shale Using Microseismic Fracture Mapping,” paper SPE 90051 presented at the 2004 SPE Annual
Technical Conference and Exhibition, Houston, Texas, September 26-29.
31. Warpinski, N.R., Wright, T.B., Uhl, J.E, Engler, B.P., Drozda, P.M., Peterson, R.E., and Branagan, P.T.: “Microseismic
Monitoring of the B-Sand Hydraulic Fracture Experiment at the DOE/GRI Multi-Site Project,” paper SPE 36450 presented at the
1996 SPE Annual Technical Conference and Exhibition, Denver, Colorado, October 6-9.
32. Warpinski, N.R., Branagan, P.T., Peterson, R.E., Fix, J.E., Uhl, J.E., Engler, B.P., and Wilmer, R.: “Microseismic and
Deformation Imaging of Hydraulic Fracture Growth and Geometry in the C Sand Interval, GRI/DOE M-Site Project,” paper SPE
38573 presented at the 1997 SPE Annual Technical Conference and Exhibition, San Antonio, Texas, October 5-8.
33. Rutledge, J.T. and Phillips, W.S.: “Hydraulic Stimulation of Natural Fractures as Revealed by Induced Microearthquakes,
Carthage Cotton Valley Gas Field, East Texas,” Geophysics, 68, Issue 2, (March-April, 2003) 441.
34. Wolhart, S.L., Odegard, C.E., Warpinski, N.R., Waltman, C.K., and Machovoe, S.R.: “Microseismic Fracture Mapping Optimizes
Development of Low Permeability Sands of the Williams Fork Formation in the Piceance Basin,” paper SPE 95637 presented at
the 2005 SPE Annual Technical Conference and Exhibition, Dallas, Texas, October 9-12.
35. Warpinski, N.R., Kramm, R.C., Heinze, J.R., and Waltman, C.K.: “Comparison of Single- and Dual-Array Microseismic Mapping
Techniques in the Barnett Shale,” paper SPE 95568 presented at the 2005 SPE Annual Technical Conference and Exhibition,
Dallas, Texas, October 9-12.
36. Cipolla, C.L., Peterman, F., Creegan, T., McCarley, D., and Nevels, H.: “Effect of Well Placement on Production and Frac Design
in a Mature Tight Gas Field,” paper SPE 95337 presented at the 2005 SPE Annual Technical Conference and Exhibition, Dallas,
Texas, October 9-12.
37. Mayerhofer, M.J., Bolander, J.L., Williams, L.I., Pavy, A., and Wolhart, S.L.: “Integration of Microseismic Fracture Mapping,
Fracture and Production Analysis with Well Interference Data to Optimize Fracture Treatments in the Overton Field, East Texas,”
paper SPE 95508 presented at the 2005 SPE Annual Technical Conference and Exhibition, Dallas, Texas, October 9-12.
38. Wolhart, S.L., Harting, T.A., Dahlem, J.E., Young, T.J., Mayerhofer, M.J., and Lolon, E.P.: “Hydraulic Fracture Diagnostics
Used to Optimize Development in the Jonah Field,” paper SPE 102528 presented at the 2006 SPE Annual Technical Conference
and Exhibition, San Antonio, Texas, September 24-27.
39. Griffin, L.G., Sullivan, R.B., Wolhart, S.L., Waltman, C.K., Wright, C.A., Weijers, L., and Warpinski, N.R.: “Hydraulic Fracture
Mapping of the High-Temperature, High-Pressure Bossier Sands in East Texas,” paper SPE 84489 presented at the 2003 SPE
Annual Technical Conference and Exhibition, Denver, Colorado, October 5 – 8.
40. Wright, C.A., et al.: “Surface Tiltmeter Fracture Mapping Reaches New Depth: 10,000 Feet and Beyond,” paper SPE 39919
presented at the 1998 SPE Rocky Mountain Regional Conference, Denver, 6-8 April.
41. Wright, C.A., et al.: “Downhole Tiltmeter Fracture Mapping: A New Tool for Directly Measuring Hydraulic Fracture
Dimensions,” paper SPE 49193 presented at the 1998 ATCE, New Orleans, 27-30 September.
42. Wright, C.A., et al.: “Real-Time Fracture Mapping from the ‘Live’ Treatment Well,” paper SPE 71648 presented at the 2001
ATCE, New Orleans, 30 September – 3 October.
43. Griffin, L.G., C.A. Wright, E.J. Davis, S.L. Wolhart, and Z.A. Moschovidis: “Surface and Downhole Tiltmeter Mapping: An
Effective Tool for Monitoring Downhole Drill Cuttings Disposal,” SPE paper 63032.
44. Weijers, L., et al.: “Developing Calibrated Fracture Growth Models for Various Formations and Regions Across the United
States,” paper SPE 96080 presented at the 2005 SPE Annual technical Conference and Exhibition held, Dallas, Texas, October 9-
12.
45. Beugelsdijk, L.J.L., de Pater, C.J., and Sato, K.: “Experimental Hydraulic Fracture Propagation in a Multi-Fractured Media,”
paper SPE 59419 presented at the 2000 SPE Asia Pacific Conference on Integrated Modeling for Asset Management, Yokohama,
Japan, April 25-26.
46. Warpinski, N.R., Mayerhofer, M.J., Vincent, M.C., Cipolla, C.L., and Lolon, E.P.: “Stimulating Unconventional Reservoirs:
Maximizing Network Growth while Optimizing Fracture Conductivity,” paper SPE 114173 presented at the 2008 SPE
Unconventional Reservoirs Conference, Keystone, Colorado, February 10-12.
47. Mayerhofer, M.J., Lolon, E.P., Youngblood, J.E., and Heinze, J.R.: “Integration of Microseismic Fracture Mapping Results with
Numerical Fracture Network Production Modeling in the Barnett Shale,” paper SPE 102103 presented at the 2006 SPE Annual
Technical Conference and Exhibition, San Antonio, Texas, September 24-27.
48. Shapiro, S.A., Rothert, E., Rath, V., and Rindschwentner, J.: “Characterization of fluid transport properties of reservoirs using
induced seismicity,” Geophysics, 67, 212-220, 2002.
49. Sharma, M. M.: “Advanced Fracturing Technology for Tight Gas: An East Texas Field Demonstration,” DOE Report DE-FC26-
01NT41326 (March 2005).
50. Chien, S.: “Settling Velocity of Irregularly Shaped Particles,” SPEDC (December 1994).
51. Kern, L.R., Perkins, T.K., and Wyant, R.E.: “The Mechanics of Sand Movement in Fracturing,” Petroleum Transactions, AIME,
216, (1959) 403.
52. Blot, M.A. and Medlin, W.L.: “Theory of Sand Transport in Thin Fluids,” paper SPE 14468 presented at the 1985 Annual
Technical Conference, Las Vegas, Nevada, September 22-26.
53. Patankar, N.A., Joseph, D.D., Wang, J., Barree, R. Conway, M. and Asadi, M.: “Power Law Correlations for Sediment Transport
in Pressure Driven Channel Flows,” International Journal of Multiphase Flow, 28, (2002) 1269.
54. Woodworth, T.R. and Miskimins, J.L.: “Extrapolation of Laboratory Proppant Placement Behavior to the Field in Slickwater
Fracturing Applications,” paper SPE 106089 presented at the 2007 Hydraulic Fracturing Technology Conference, College Station,
Texas, January 29-31.
55. Besler, M.R., Steele, J.W., Egan, T., and Wagner, J.: “Improving Well Productivity and Profitability in the Bakken – A Summary of
Our Experiences Drilling, Stimulating, and Operating Horizontal Wells,” paper SPE 110679 presented at the 2007 SPE Annual
Technical Conference, Anaheim, California, November 11-14.
SPE 115769 23

56. Leonard, R., Woodroof, R., Bullard, K., Middlebrook, M. and Wilson, R.: “Barnett Shale Completions: A Method for Assessing
New Completion Strategies,” paper SPE 110809 presented at the 2007 SPE Annual Technical Conference and Exhibition,
Anaheim, California, November 11-14.
57. Diamond, W.P., and Oyler, D.C.: “Effects of Stimulation Treatments on Coalbeds and Surrounding Strata – Evidence from
Underground Observations,” Bureau of Mines Report of Investigations #9083 (1987).
58. Dedurin, A.V., Majar, V.A., Voronkov, A.A., Zagurenko, A.G., Zakharov, A.Y., Palisch, T., and Vincent, M.C.: “Designing
Hydraulic Fractures in Russian Oil and Gas Fields to Accommodate Non-Darcy and Multiphase Flow – Theory and Field
Examples,” paper SPE 101821 presented at the 2006 Russian Oil and Gas Conference, Moscow, October 3-6.
59. Vincent, M.C.: “Proving It – A Review of 80 Published Field Studies Demonstrating the Importance of Increased Fracture
Conductivity,” paper SPE 77675 presented at the 2002 SPE Annual Technical Conference and Exhibition, San Antonio, Texas,
September 29-October 2.
60. Vincent, M.C.: “Field Trial Design and Analyses of Production Data from a Tight Gas Reservoir: Detailed Production
Comparisons from the Pinedale Anticline,” paper SPE 106151 presented at the 2007 Hydraulic Fracturing Technology Conference,
College Station, Texas, January 29-31.
61. Stim-Lab Proppant and Fluids Consortia (1986-2007).
62. American Petroleum Institute: “Recommended Practices for Testing Sand Used in Hydraulic Fracturing Operations,” API RP 56,
December 1995.
63. Palisch, T., Duenckel, R., Bazan, L., Heidt, H.J., and Turk, G.: “Determining Realistic Fracture Conductivity and Understanding
its Impact on Well Performance – Theory and Field Examples,” SPE 106301 presented at the 2007 Hydraulic Fracturing
Technology Conference, College Station, Texas, January 29-31.
64. Barree, R.D., et al.: “Realistic Assessment of Fracture Conductivity for Material Selection,” paper SPE 84306 presented at the
2003 SPE Annual Technical Conference & Exhibition, Denver, Colorado, October 5-8.
65. Barree, R.D., et al.: “Closing the Gap: Fracture Half-length from Design, Buildup, and Production Analysis,” paper SPE 84491
presented at 2003 SPE Annual Technical Conference & Exhibition, Denver, Colorado, October 5-8.
66. Mayerhofer, M.J., Richardson, M.F., Walker R.N. Jr., Meehan D.N., Oehler M.W., and Browning R.R., Jr.: “Are Proppants Really
Necessary?” Journal of Petroleum Technology (March 1998).
67. Mayerhofer, M.J., Richardson, M.F., Walker, R.N. Jr., Meehan, D.N., Oehler, M.W., and Browning, R.R. Jr.: “Proppants? We
Don’t Need No Proppants,” SPE 38611 presented at the 1997 ATC 97 in San Antonio, Texas, October.
68. Mayerhofer, M.: “Water-fracs – Results from 50 Cotton Valley Wells,” SPE 49104 presented at the 1998 SPE Annual Technical
Conference and Exhibition, New Orleans, Louisiana, 27-30 September.
69. Britt, L.K., et al.: “Water-Fracs: We Do Need Proppant After All,” paper SPE 102227 presented at the 2006 SPE Annual Technical
Conference and Exhibition, San Antonio, Texas, September 24-27.
70. Fredd, C.N., McConnell, S.B., Boney, C.L., and England, K.W.: “Experimental Study of Fracture Conductivity for Water-
Fracturing and Conventional Fracturing Applications,” SPE Journal (September 2001).
71. Cipolla, C.L. and Mayerhofer, M.J.: “Understanding Fracture Performance and Optimizing Treatment Designs using Well Test
Data and Fracture Modeling,” paper SPE 49044 prepared for the 1998 SPE Annual Technical Conference & Exhibition, New
Orleans, Louisiana, 27-30 September. Also published in the JPT (March 1999).
72. Cipolla, C.L.: “Hydraulic Fracture Technology in the Ozona Canyon and Penn Sands,” paper SPE 35196, presented at the 1996
Permian Basin Oil & Gas Recovery Conference, Midland, Texas, March 27-29.
73. Barton, N., Bandis, S., and Bakhtar, K.: “Strength Deformation and Conductivity Coupling of Rock Joints,” International Journal of
Rock Mechanics, Mining Sciences & Geomechanical Abstracts, 22 (1985) 121.
74. Walsh, J.B.: “Effect of Pore Pressure and Confining Pressure on Fracture Permeability,” International Journal of Rock
Mechanics, Mining Sciences & Geomechanical Abstracts, 18 (1981) 429.
75. Gall, B.L, Sattler, A.R., Maloney, D.R., and Raible, C.J., 1988: “Permeability Damage to Natural Fractures Caused by Fracturing
Fluid Polymers,” paper SPE 17542, presented at SPE Rocky Mountain Regional Meeting, Casper, Wyoming, May 11-13.
76. Jones, F.O.: “A Laboratory Study of the Effects of Confining Pressure on Fracture Flow and Storage Capacity in Carbonate
Rocks,” JPT, 21-27, January 1976.
77. Kranz, R.L., Frankel, S.D., Engelder, T. and Scholz, C.H.: “The Permeability of Whole and Jointed Barre Granite,” International
Journal of Rock Mechanics, Mining Sciences, and Geomechanical Abstracts, 16, 225-234, 1979.
78. Brannon, H.D., et al.: “Maximizing Fracture Conductivity with Proppant Partial Monolayers: Theoretical Curiosity or Highly
Productive Reality?” paper SPE 90698 presented at the 2004 Annual Technical Conference and Exhibition, Houston, Texas,
September 26-29.
79. Maxwell, S.C., Waltman, C.K., Warpinski, N.R., Mayerhofer, M.J. and Boroumand, N.: “Imaging Seismic Deformation Induced by
Hydraulic Fracture Complexity,” paper SPE 102801 presented at the 2006 SPE Annual Technical Conference and Exhibition, San
Antonio, Texas, September 24-27.
80. Nolte, K.G. and Smith, M.B.: “Interpretation of Fracturing Pressures,” JPT (September 1981) 1767.
81. Barree R.D.: “Application of Pre-Frac Injection/Falloff Tests in Fissured Reservoirs –Field Examples,” paper SPE 39932 presented
at the Rocky Mountain Regional/Low-Permeability Reservoirs Symposium and Exhibition, Denver, Colorado, April 1998.
82. Barree, R.D.: “Determination of Pressure Dependent Leakoff and Its Effect on Fracture Geometry,” SPE 36424 presented at the
71st Annual Technical Conference, Denver, Colorado, 6-9 October 1998.
83. Reservoir Stimulation Second Addition, M.J. Economides and K.G. Nolte, Schlumberger Educational Services, 1989, pp 9-5.
84. Morrow, N.R., Buckley, J.S., Cather, M.E., Brower, K.R., Graham, M., Ma, S., and Zhang, X.: “Rock Matrix and Fracture
Analysis of Flow in Western Tight Gas Sands,” US DOE Report DOE/MC/21179-2853, New Mexico Institute of Mining and
Technology, (February 1990).
85. Coulter, G.R., Benton, E.G., and Thomson, C.L.: “Water Fracs and Sand Quality: A Barnett Shale Example,” SPE 90051, SPE
ATCE, Houston, Texas, September 26-29, 2004.
24 SPE 115769

86. M. D. Murrey, et al.: “Fracture Design, Execution, and Evaluation in Retrograde Condensate Reservoirs: Case History of the
Angsi Field, Offshore Malaysia,” paper SPE 84395 presented at the 2003 SPE Annual Technical Conference and Exhibition,
Denver, Colorado, U.S.A., 5-8 October.
87. X. Liu, et al.: “Case History of Fracture Performance in a Channel Reservoir,” paper SPE 84396 presented at the 2003 SPE
Annual Technical Conference and Exhibition, Denver, Colorado, U.S.A., 5-8 October.
88. Huitt, J.L., McGlothlin, B.B., and McDonald, J.F.: “The Propping of Fractures in Formations in which Propping Sand Crushes,
Drilling and Production Practices,” API, 120, 1959.
89. McGlothlin, B.B. and Huitt, J.L.: “Relation of Formation Rock Strength to Propping Agent Strength in Hydraulic Fracturing,”
JPT, 377-384, 1966.
90. Huitt, J.L. and McGlothlin, B.B.: “The Propping in Formations Susceptible to Propping Sand Embedment, Drilling and
Production Practices”, API, 115, 1958.
91. Howard, G.C. and Fast, C.R. 1970: Hydraulic Fracturing AIME Monograph #2, New York, 6: 59-60.
92. Ghauri, W.K.“Results of Well Stimulation by Hydraulic Fracturing and High Rate Oil Backflushing,” paper SPE 1382 JPT June
1980.
93. Lambert, S.W., Trevits, M.A., “The Feasibility of No-proppant Stimulation to Enhance Removal of Methane from the Mary Lee
Coalbed,” US DOE Report of Investigation, April, 1980.
94. Palmer, I.D.: “Review of Coalbed Methane Well Stimulation,” paper SPE 22395 presented at the 1992 International Meeting on
Petroleum Engineering, Beijing, China, March 24-27.
95. MacDonald, R.J. et al. “Comparing Production Responses from Devonian Shale and Berea Wells Stimulated with Nitrogen Foam
and Proppant vs. Nitrogen Only, Pike Co., Kentucky,” paper SPE 77464 presented at the 2002 Annual Technical Conference, Sep
29 – Oct 2.
96. Harper, T. R., et al. “Fracturing Without Proppant,” paper SPE 13858 presented at the 1985 Low Permeability Gas Reservoirs
Symposium, Denver, March 19-22.
97. Gottschling, J.C. et al. “Nitrogen Gas and Sand: A New Technique for Stimulation of Devonian Shale,” paper SPE 12313, JPT
May 1985.
98. Rieb, B.A. and Leshchyshyn, T. T. “Production Success of Proppant Stimulation on Horseshoe Canyon Coalbed Methane,” paper
SPE 96864 presented at the 2005 Annual Technical Conference, Dallas, Oct 9-12.
99. Crow, W.L., “Mini Massive Frac – A Stimulation Technique for Low-Porosity, Low-Permeability Zones,” paper SPE 6627
presented at the 1977 AIME Eastern Regional Meeting, Pittsburgh, Oct 27-28.
100. Grieser, B. et al. “The Rocket Science Behind Water Frac Design,” paper SPE 80933 presented at the 2003 SPE Production and
Operations Symposium, Oklahoma City, Oklahoma, March 23-25.

Appendix 1 – Fluid Viscosity Supplement


Figure 17 shows the relative penetration
Viscosity (cp)
distance compared to water (1 cp) as function of
1 10 100 1000 10000 100000
viscosity. The comparison assumes that all other 1
factors are equal; however, the pressure drop and
fracture permeability may differ for high viscosity
fluids. The figure illustrates that even moderate
viscosity fluids such a linear gels (10 to 50 cp)
could reduce penetration distance by a factor of 3 to 0.1
Relative Distance

7. High viscosity cross-linked fluids (100 to 1000


cp) would reduce penetration distance by a factor
of 10 to 30 compared to water. Therefore, high
viscosity fluids should reduce the degree of fracture 0.01
complexity, while pumping low viscosity fluids
should maximize fracture complexity. The Barnett
Shale is an example of a reservoir where
maximizing fracture complexity using low
viscosity fluid (water) is the key to increasing well 0.001

productivity47, while many North Sea Chalk


reservoirs are examples where high viscosity fluids Figure 17 - Effect of fluid viscosity on penetration
(cross-linked gels) which minimize fracture
complexity are key to success.20

Appendix 2 – Proppant Transport & Distribution Supplement


The settling of proppant particles is often predicted using the relatively simple relationship shown in Equation A1
(Stokes’ Law).
(ρ p − ρ f )gd p
2

Vs = ………………….(A1)
18μ
Where Vs is the particle settling velocity, ρp is the density of the particle, ρf is the density of the fluid, g is the acceleration of
gravity, μ is the fluid viscosity, and dp is the diameter of the particle.
SPE 115769 25

This form of the equation has many simplifying assumptions which are violated in actual slickwater treatments. Stokes’
Law in this form predicts the terminal settling velocity of spherical particles in stagnant Newtonian liquid without wall
effects and without particle interaction. In actual treatments, realistic power-law fluids are highly turbulent near-wellbore,
irregularly shaped particles settle at a velocity adequate to generate wakes requiring corrections for inertia, particles
agglomerate (draft) and interact (hinder), while rough and/or inclined fracture faces complicate settling. Although Stokes’
Law has been deemed to be “grossly inadequate”49 for describing proppant placement, the general relationships are valid:
• with low viscosity (μ) fluids – proppant settling will be rapid,
• settling can be slowed with reduced particle density (ρp), and most poorly recognized,
• when particle diameter is reduced, settling velocity reduces exponentially.
Although more sophisticated mathematical treatments have been proposed to describe proppant settling,49,50 a predictive
tool is not available to provide a precise description of the created fracture geometry, and the hindrance of particle movement
by rough, inclined, stair-stepping fractures. However, the maximum distance that particles will be transported with slick-
water is not exclusively a function of suspension time. Instead, in slick-water fractures, the main mode of lateral transport is
actually saltation, or bed transport as proposed by Kern, Perkins, and Wyant51, Blot52, Patankar53 and shown in recent
laboratory videos by Stim-Lab61 and others54. Field validation of proppant transport is somewhat limited, but there is
evidence of proppant being transported significant distances – even with low viscosity fluids.56,57

Estimating Proppant Concentration and Propped Fracture Area


With the application of fracture mapping technologies it is now possible to directly measure hydraulic fracture length,
height, and network width. Therefore, proppant concentrations can be directly calculated for the three scenarios presented if
reasonable ranges of fracture complexity can be estimated. Fracture area (one fracture face) for a simple planar fracture is
given by Equation B1 and the average fracture width for a simple planar fracture is given by Equation B2.

εV fl
A fp = 2 x f h f …………...……………….. (B1), w fp = .........................(B2)
A fp
Fracture area for a complex fracture system is given by Equation B3 and the average width is given by Equation B4.
4 x f xn h f εV fl
A fc = + 2 h f x f + h f x n ......(B3), w fc = .........................(B4)
Δxs A fc
For simplicity equation B3 assumes that the fracture network consists of evenly spaced blocks (Δxs). The proppant
concentrations for the three limiting scenarios presented above can be calculated using equations B1, B3 and B4, the fluid
volume pumped, an estimate of the fluid efficiency, and the amount of proppant pumped using the follow equations.
Mp Mp
Case 1: C p1 = ...................(B5), Case 2: C p 2 = ..............(B6), Case 3: C p 3 = w fc ρ pb ...............(B7)
A fc A fp
The percentage of the fracture that is propped for Case 3 can be calculated using Equation B8.

Mp
Apropped = ...................(B8)
C p 3 A fc

You might also like