Documente Academic
Documente Profesional
Documente Cultură
com
◆
Basics of
Hydraulic Fracturing
M. B. Smith, NSI Technologies, Inc.
J. W. Shlyapobersky,† Shell E&P Technology Co.
Figure 5-1. Internal pressure breaking a vertical wellbore. Figure 5-2. Cross-sectional view of a propagating fracture.
nonvertical wells. However, in general, the wellbore open and maintain a conductive flow path for the
breaks—i.e., the rock fractures—owing to the action increased formation flow area during production.
of the hydraulic fluid pressure, and a “hydraulic” The propping agent is generally sand or a high-
fracture is created. Because most wells are vertical strength, granular substitute for sand (see Section 7-7).
and the smallest stress is the minimum horizontal Alternatively, for carbonate rocks, the hydraulic fluid
stress, the initial splitting (or breakdown) results in may consist of acid that dissolves some of the forma-
a vertical, planar parting in the earth. tion, leaving behind acid-etched channels extending
The breakdown and early fracture growth expose into the reservoir.
new formation area to the injected fluid, and thus the After the breakdown, the fracture propagation rate
rate of fluid leaking off into the formation starts to and fluid flow rate inside the fracture become impor-
increase. However, if the pumping rate is maintained tant. They are dominated by fluid-loss behavior. As
at a rate higher than the fluid-loss rate, then the introduced by Carter (1957) and discussed in the fol-
newly created fracture must continue to propagate lowing (and in Chapters 6 and 9), the fluid-loss rate
and grow (Fig. 5-2). This growth continues to open qL from a fracture can be expressed as
more formation area. However, although the
hydraulic fracture tremendously increases the forma- 2 CL A
qL ≈ , (5-1)
tion flow area while pumping, once pumping stops t−τ
and the injected fluids leak off, the fracture will close
and the new formation area will not be available for where CL is the fluid-loss coefficient, A is an element
production. To prevent this, measures must be taken of the fracture area (i.e., increased inflow area), t is
to maintain the conductive channel. This normally time measured from the start of pumping, and τ is
involves adding a propping agent to the hydraulic the time when each small area element of a fracture
fluid to be transported into the fracture. When pump- is created or opened. As a direct consequence of this
ing stops and fluid flows back from the well, the relation, the highest rate of fluid loss is always at the
propping agent remains in place to keep the fracture fracture tip. Newly created fracture area exists at that
point (t – τ = 0 in the denominator), making qL in- Thus, proppant concentration (i.e., volume fraction
stantly infinite. of solid proppant) increases as the slurry stages dehy-
Initially, fracture penetration is limited, and hence drate. The pump schedule, or proppant addition
fluid loss is high near the wellbore. For that reason, schedule, must be engineered much like handicapping
the first part of a hydraulic fracture treatment con- horse races, but with no single winner. Rather, all
sists of fluid only (no proppant); this is termed the stages should finish at the right place, at the right
pad. The purpose of a pad is to break down the well- time, with the right final proppant concentration. The
bore and initiate the fracture. Also, the pad provides pad should be completely lost to the formation, and
fluid to produce sufficient penetration and width to the first proppant stage should be right at the fracture
allow proppant-laden fluid stages to later enter the tip (which should be at the design length).
fracture and thus avoid high fluid loss near the frac- As the proppant slurry stages move down the frac-
ture tip. After the pad, proppant-laden stages are ture, they dehydrate and concentrate. Slurry stages
pumped to transport propping agent into the fracture. pumped later in the treatment are pumped at a higher
This chapter describes the process for propped frac- concentration. These stages are not in the fracture for
ture treatments; acid fracture treatments are dis- long prior to the treatment end (i.e., prior to shut-
cussed in Section 10-6. down) and are thus exposed to less fluid loss and less
However, because fluid loss to the formation is still dehydration. Ideally, the first proppant stage pumped
occurring, even near the well, the first proppant is reaches the fracture tip just as the last of the pad
added to the fluid at low concentrations. The prop- fluid is lost into the formation (a correctly handi-
pant-laden slurry enters the fracture at the well and capped race), and this first stage has concentrated
flows toward the fracture tip (Fig. 5-3). At this point, from its low concentration to some preselected,
two phenomena begin. First, because of the higher higher final design concentration. Meanwhile, the
fluid loss at the fracture tip, slurry flows through the slurry concentration being pumped is steadily
fracture faster than the tip propagates, and the prop- increased to the same final design concentration.
;
pant-laden slurry eventually overtakes the fracture tip. At treatment end, the entire fracture is filled with the
Next, because of fluid loss, the proppant-laden slurry design concentration slurry. Design considerations
stages lose fluid (but not proppant) to the formation. for the final concentration are discussed later in this
section and in detail in Section 10-4.
The preceding description might be termed a “nor-
;; ;;
; ;
mal” design, where the entire fracture is filled with a
uniform, preselected, design proppant concentration
just as the treatment ends. If pumping continues past
that point, there would be little additional fracture
extension because the pad is 100% depleted. Con-
tinued pumping forces the fracture to become wider
;;;;;;
(and forces the pressure to increase) because the
increased volume simply acts like blowing up a bal-
loon. In some cases the additional propped width
that results may be desirable, and this procedure is
used purposely. This is termed tip-screenout (TSO)
fracturing.
At the conclusion of the treatment, the final flush
stage is pumped. This segment of a treatment con-
sists of one wellbore volume of fluid only and is
intended to sweep the wellbore clean of proppant
(Fig. 5-4). The well is generally then shut-in for
some period to allow fluid to leak off such that the
fracture closes on and stresses the proppant pack.
Shut-in also allows temperature (and chemical
Figure 5-3. Introducing proppant into the fracture. breakers added to the fluid while pumping) to reduce
;;;
;
• Damage bypass
Near-wellbore damage reduces well productivity.
This damage can occur from several sources,
including drilling-induced damage resulting from
fines invasion into the formation while drilling
;;;
;
and chemical incompatibility between drilling flu-
ids and the formation. The damage can also be
due to natural reservoir processes such as satura-
tion changes resulting from low reservoir pressure
near a well, formation fines movement or scale
deposition. Whatever the cause, the result is unde-
sirable. Matrix treatments (discussed in Chapters
13 through 20) are usually used to remove the
damage chemically, restoring a well to its natural
productivity. In some instances, chemical proce-
dures may not be effective or appropriate, and
hydraulic fracture operations are used to bypass
the damage. This is achieved by producing a high-
conductivity path through the damage region to
Figure 5-4. Flushing the wellbore to leave a propped frac- restore wellbore contact with undamaged rock.
ture. • Improved productivity
the viscosity of the fracturing fluid (see Section 7-6.2). Unlike matrix stimulation procedures, hydraulic
Ideally, this process leaves a proppant-filled fracture fracturing operations can extend a conductive
with a productive fracture length (or half-length xf), channel deep into the reservoir and actually stimu-
propped fracture height and propped fracture width late productivity beyond the natural level.
(which determines the fracture conductivity kfw). All reservoir exploitation practices are subject
Here, xf is the productive fracture half-length, which to Darcy’s law:
may be less than the created half-length L or less kh ∆p A
than the propped length. q≈ , (5-2)
µ ∆x h
;;;
;;
;;
; ;
massive stimulation treatments in tight gas forma-
tions (see Appendix to this chapter). Although out-
wardly a traditional application of fracturing to
poorer quality reservoirs, these treatments repre-
sented the first engineering attempts to alter reser-
;;;
;;
voir flow in the horizontal plane and the methodol-
ogy for well placement (e.g., Smith, 1979).
Fracturing for vertical inflow conformance (i.e.,
reservoir management) was successfully used in
the Gullfaks field (Bale et al., 1994), where selec-
tive perforating and fracturing were used to opti-
mize reserve recovery and control sand production
while maintaining (but not necessarily increasing)
the required production rates. This is illustrated in
Fig. 5-6, where the bottom, low-permeability
Rannoch-1 zone was perforated to create a
propped fracture that extends up and into the
high-permeability (>1000-md) Rannoch-3 zone.
Without fracturing, the entire zone can be perfo-
rated, and a low drawdown allows a significant
production rate on the order of 20,000 STB/D,
Figure 5-5. Increased flow area resulting from a fracture. sand free. However, sand production is triggered
by water breakthrough in the high-permeability
flow shape that is altered, as discussed in detail in zone (from downdip water injection). The result-
Chapter 1.) ing wellbore enlargement caused by sand produc-
tion acts to stimulate production from the high-
This is the classic use of fracturing, to increase permeability zone. To stop sand production, draw-
the producing rate by bypassing near-wellbore
formation damage or by increasing exposure of
the formation area and thus stimulating well per- Stress (psi)
formance beyond that for no damage. For a single 4500 5500
well, treatment design concentrates on creating Rannoch-3
1820
the required formation flow area to yield increased
production at minimal cost. More formally, the Rannoch-3
design should optimize economic return on the
1840
basis of increased productivity and treatment cost. Rannoch-3
• Reservoir management
Along with improving well productivity, fractures Rannoch-2
1860
also provide a powerful tool for altering reservoir
flow. In combination with the other parts of field Rannoch-1
development, the fracture becomes a reservoir
management tool. For example, creating long 1880
fractures in tight rock (k < 0.1 md) enables field
development with fewer wells. However, even 0 20 40 60 80
fewer wells are required if the fracture azimuth TVD Fracture penetration (m)
(m below
is known and the wells are located appropriately sea level)
(e.g., not on a regulatory-required square pattern).
The actual philosophy shift for fracturing, from Figure 5-6. Fracturing for vertical inflow conformance.
down must be reduced even more. The production (as compared with a simple completion in which
is then essentially 100% water coming from the the entire zone is perforated).
stimulated high-permeability zone, and the well Another example of reservoir management is
must be abandoned. This further diminishes pro- waterflood development utilizing fractures and a
duction from the large reserves found in the “line drive” flood pattern (i.e., one-dimensional
deeper zones with lower permeability. [1D] or linear flow from injection fractures to pro-
Open- or cased hole gravel packing could be duction fractures). Knowledge of the fracture
used to eliminate the sand production. However, azimuth, combined with conductive fractures (or
such completions are less than satisfactory for two correctly controlled injection greater than the frac-
reasons. First, the deeper, lower permeability ture pressure) results in improved sweep efficiency
zones can significantly benefit from stimulation. and enables more efficient field development.
Second, significant scaling occurs with water
breakthrough and quickly plugs the gravel pack.
The fracturing tool selected to manage the 5-1.3. Design considerations and
Gullfaks field is termed an indirect vertical frac- primary variables
ture completion (IVFC). The IVFC accomplishes This section introduces the primary variables for frac-
several goals: ture design. Sidebar 5A summarizes how the design
– Some (although choked) production is achieved variables originate from treatment design goals.
from the main zone to enable the well to reach As mentioned previously, fracturing was con-
minimum productivity standards. trolled historically by operational considerations.
– Production from the lower, moderate-perme- This limited its application because fracturing is
ability zone is stimulated, maximizing reserves dominantly a reservoir process, and hence why a
from this zone. reservoir is fractured and what type of fracture is
– Greater drawdown is allowed because the weak required should be dominated by reservoir engineer-
high-permeability rock is separated from the ing considerations. The permeability k becomes the
perforations, and greater drawdown increases primary reservoir variable for fracturing and all
the total rate and significantly increases recov- reservoir considerations. Other, so-called normal
ery from the lower zones. reservoir parameters such as net pay and porosity
dominate the economics and control the ultimate
– If the upper high-permeability zone has sand viability of a project but do not directly impact how
production tendencies (as is typically the case), the fracturing tool is employed. As discussed in
then producing this zone via the fracture totally Chapter 12, postfracture productivity is also gov-
avoids the need for sand control. erned by a combination of the fracture conductivity
– Any potential for water breakthrough in the kf w and xf, where kf is the permeability of the prop-
high-permeability zone is retarded, and post- pant in the fracture, w is the propped fracture width,
water-breakthrough oil production is signifi- and xf is the fracture penetration or half-length.
cantly increased. These variables are controlled by fracturing and
To achieve these goals, fracture conductivity therefore identify the goals for treatment design.
must be tailored by synergy between the reservoir The productive fracture half-length xf may be less
and fracture models. Too much conductivity than the created (or the created and propped) half-
accelerates production and the time to water length L because of many factors (see Section 12-3).
breakthrough from the high-permeability main For example, the fracture width near the tip of a
zone. Also, too much conductivity, because of fracture may be too narrow to allow adequate
surface or tubular limits for the production rate, propped width. As another example, vertical varia-
restricts drawdown on the lower zones, and the tions in formation permeability, or layering, can
desired, more uniform vertical production profile cause the apparent productive length xf to be less
is not achieved. The fracture design goal is not to than the actual propped length (Bennett et al., 1986).
simply accelerate the rate but to achieve maxi- Similarly, this also makes the fracture height hf
mum reserves recovery with no sacrifice of rate important in several ways (Fig. 5-7):
This discussion briefly summarizes the design goals of hydraulic fracturing that provide a road map for the major design variables.
Design goals
Design goals result from Darcy’s law (Eq. 5-2), in which the dimensionless term A/(∆xh) is defined by flow conditions and equals
ln(re /rw´ ) for steady-state flow (as discussed in Chapter 1). For steady-state flow, Prats (1961) showed that a fracture affects produc-
tivity through the equivalent wellbore radius rw´ and that rw´ is related to the fracture half-length or penetration xf by the dimension-
less fracture conductivity (CfD = kfw/kxf). Cinco-Ley et al. (1978) extended these concepts for transient flow with the relation among
xf, rw´ and CfD shown in Fig. 5-11 for pseudoradial flow (where the pressure-depletion region >> xf but is not affected by external
boundaries). Thus, the primary design goals are fracture half-length or penetration and the fracture conductivity kfw, with their rela-
tive values defined by CfD.
Design variables
Design variables result from material balance, rock mechanics and fluid mechanics considerations.
The material balance is (Eqs. 5-10 through 5-12)
where CL and Sp are fluid-loss parameters that can be determined by the results of a fluid-loss test (Fig. 5A-1) for which the filtrate
volume divided by the exposed area VL /A = Sp + 2CL√t . Combining the relations in Eq. 5A-1 gives Eq. 5-13:
q it p
L ≈ ,
6CLhL t p + 4hLS p + 2whf
where fracture penetration L is related to pump rate, fluid loss, height, width, etc.
Next is the elasticity equation (Eq. 5-14):
2p net d
w max = ,
E'
where pnet = pf – σc, and width is related to net pressure as a function of modulus and geometry and the pressure required to propa-
gate the fracture (Eq. 5-21):
where d is the characteristic fracture dimension and generally is the smaller dimension between hf and L.
Third is the fluid flow equation (Eqs. 5-15 through 5-19), in which Eq. 5-15 (dpnet /dx = 12µq/hfw3) is combined with the width equation:
1/ 4
E' 3
p net ≈ 4 {κµq i L} + p net
4
tip
, (5A-3)
hf
where the pressure drop down the fracture is related to viscosity, pump rate, fracture length (and thus to fluid loss), etc. The net
pressure distribution gives the fracture width distribution and thus the final propped fracture width (i.e., kfw). Hence the primary
design variables are CL, hL, Sp, hf, E ,́ KIc-apparent, qi , µ and σc .
Optimum design
The optimum design results from maximizing
Volume lost/area, VL /A
2Cw
Sp
would have been produced through the entire well- that many of these interactions will be contradictory
bore circumference (had there been no damage). The or incompatible. This is discussed later, but an exam-
fracture conductivity kf w must be greater than 2πrwk, ple is as follows. Consider a case where reservoir
where rw is the wellbore radius. For higher perme- goals require a long fracture. With deep penetration
ability formations that can deliver high rates with into the pay zone, getting good proppant transport
sufficient fracture permeability, fracture width and down a long fracture clearly requires high fluid vis-
any variables that affect width become important. As cosity. However, high viscosity increases the net
discussed in the following and in Section 6-2, width pressure inside the fracture. This reacts with the
is controlled by the fracture dimensions (hf and L), stress difference between the pay and the overlying
net pressure inside the fracture acting to open and and underlying shales and causes height growth,
propagate the fracture, and another property, the resulting in less penetration than desired, and thus
modulus or stiffness of the rock. less viscosity is required.
As implied by the term hydraulic fracturing, fluid Inherent contradictions controlling fluid selection
mechanics is an important element in fracturing. The abound:
two dominant fluid mechanics variables, injection
• Good viscosity is required to provide good prop-
(pump) rate qi and fluid viscosity µ, affect net pres-
pant transport, but minimal pipe friction is also
sure inside the fracture (and thus width) and largely
desirable to reduce surface pump pressure.
control transport and the final placement of proppant
in the fracture. These variables also have a role in • The fluid system is expected to control fluid loss,
controlling the volume of fluid lost to the formation but without damage to the formation or fracture
during pumping. For example, high pump rates permeability.
reduce the total fluid loss because for a given volume • Performance at high temperature, for long periods
pumped there is less time for fluid loss to occur. of time, is required from a fluid system that does
Another key factor of a good design is selection not cost much.
of the fluid and proppant systems with performance
characteristics (e.g., µ, CL, kf) that best meet the
requirements for the fracture treatment (i.e., material 5-2. In-situ stress
selection). In addition, the performance variables for In-situ stress, in particular the minimum in-situ stress
the materials must be properly characterized. Fluids (termed the fracture closure pressure for nonhomoge-
and proppants are addressed in Chapter 7, and their neous zones, as discussed earlier) is the dominant
performance is discussed in Chapter 8. parameter controlling fracture geometry. It is dis-
Finally, all the design parameters must be molded cussed in detail in Chapter 3. For relaxed geologic
to be compatible with existing well conditions (i.e., environments, the minimum in-situ stress is gener-
operational considerations). For example, it does lit- ally horizontal; thus a vertical fracture that formed
tle good to complain that the detailed design and when a vertical wellbore broke remains vertical and
analysis done in planning a treatment for an existing is perpendicular to this minimum stress. Hydraulic
well call for a high pump rate of 60 bbl/min when fractures are always perpendicular to the minimum
the wellbore conditions limit the maximum allow- stress, except in some complex cases, and even for
able pump rate to one-half that rate. Clearly, for new those cases any significant departure is only at the
wells the operational considerations (detailed in well. This occurs simply because that is the least
Chapter 11) should be an integral part of planning resistant path. Opening a fracture in any other direc-
for the drilling and completion process (e.g., well tion requires higher pressure and more energy.
trajectory for extended reach wells) (Martins et al., The minimum stress controls many aspects of
1992c). fracturing:
• At very shallow depths or under unusual condi-
5-1.4. Variable interaction tions of tectonic stress and/or high reservoir pres-
sure, the weight of the overburden may be the
It is clear that with major design considerations com-
minimum stress and the orientation of the
ing from multiple disciplines, the variables will
hydraulic fractures will be horizontal; for more
react, interact and interconnect in multiple ways and
normal cases, the minimum stress is generally hor- sure, the closure stress is typically between 0.6 and
izontal and the maximum horizontal stress direc- 0.7 psi/ft of depth (true vertical depth, TVD). More
tion determines whether the vertical fracture will generally, as discussed in Chapter 3, the minimum
run north–south, east–west, etc. stress is related to depth and reservoir pressure by
σ c ≅ Ko (σ v − pr ) + pr + T ,
• Stress differences between different geologic lay-
(5-4)
ers are the primary control over the important
parameter of height growth (Fig. 5-9). where Ko is a proportionality constant related to the
• Through its magnitude, the stress has a large bear- rock properties of the formations (possibly to both
ing on material requirements, pumping equipment, the elastic properties and the faulting or failure prop-
etc., required for a treatment. Because the bottom- erties), σv is the vertical stress from the weight of
hole pressure must exceed the in-situ stress for the overburden, pr is the reservoir pore pressure, and
fracture propagation, stress controls the required T accounts for any tectonic effects on the stress (for
pumping pressure that well tubulars must with- a relaxed, normal fault geology, T is typically small).
stand and also controls the hydraulic horsepower Ko is typically about 1⁄3. For fracture design, better
(hhp) required for the treatment. After fracturing, values are required than can be provided by such a
high stresses tend to crush the proppant and reduce simple relation, and methods of measuring or infer-
kf ; thus, the stress magnitude dominates the selec- ring the in-situ stress are discussed in Chapters 3 and
tion of proppant type and largely controls postfrac- 4. For preliminary design and evaluation, using
ture conductivity. Eq. 5-4 with Ko = 1⁄3 is usually sufficient.
Therefore, the detailed design of hydraulic fracture
treatments requires detailed information on in-situ 5-3. Reservoir engineering
stresses. An engineer must know the magnitude of
the minimum in-situ stress for the pay zone and As previously mentioned, because the ultimate goal
over- and underlying zones and in some cases must of fracturing is to alter fluid flow in a reservoir,
know the direction for the three principal stresses. reservoir engineering must provide the goals for a
For a simple, relaxed geology with normal pore pres- design. In addition, reservoir variables may impact
the fluid loss.
∆σ
σv
σh
0.9
0.8
0.7
0.6
pnet/∆σ
Shale 0.5
pnet
hf hfo 0.4
3 4 σH 0.3
2
0.2
0.1
0
1 1 2 3
h f /h fo
Shale
Figure 5-9. Fracture height growth. (a) Idealized fracture profile of the relation of fracture geometry to in-situ stresses.
σh = minimum horizontal stress, σH = maximum horizontal stress. (b) Typical fracture vertical cross section illustrating the
relation of the total fracture height hf to the “original” fracture height hfo. (c) Theoretical relation among hf /hfo, pnet and the
in-situ stress difference ∆σ (Simonson et al., 1978).
5-3.1. Design goals fracture and also assumes infinite conductivity. Prats
correctly accounted for the pressure distribution
Historically, the emphasis in fracturing low-perme- around a fracture and provided a general relation
ability reservoirs was on the productive fracture between dimensionless conductivity and rw´ for
length xf. For higher permeability reservoirs, the con- steady-state conditions (see Chapter 1). The relation
ductivity kfw is equally or more important, and the shows that for infinite-conductivity fractures, the
two are balanced by the formation permeability k. upper limit on rw´ is slightly less than that from the
This critical balance was first discussed by Prats flow area balance in Eq. 5-6. For infinite kfw, Prats
(1961), more than 10 years after the introduction of found
fracturing, with the important concept of dimension-
less fracture conductivity CfD: rw′ = 0.5 x f . (5-7)
kw
C fD = f . (5-5)
k xf Cinco-Ley et al. (1978) later integrated this into a
This dimensionless conductivity is the ratio of the full description of reservoir response, including tran-
ability of the fracture to carry flow divided by the
ability of the formation to feed the fracture. In gen- 5B. Highway analogy for dimensionless
eral, these two production characteristics should be fracture conductivity
in balance. In fact, for a fixed volume of proppant,
maximum production is achieved for a value of CfD A simplistic analogy for dimensionless fracture conductivity
CfD is a highway system. The numerator of this dimension-
between 1 and 2, as discussed in Chapters 1 and 10, less variable is kfw, which is the capacity of the highway or
with an analogy to highway design in Sidebar 5B. the ability of the highway to carry traffic. The denominator is
kxf ; this is the ability of the feeder roads to supply traffic to
Prats also introduced another critical concept, the the highway.
idea of the effective wellbore radius rw´. As shown in The famous old U.S. highway known as Route 66 ran, for
much of its length, across sparsely populated areas where
Fig. 5-10, a simple balancing of flow areas between feeder roads were few, narrow and far between. The ability
a wellbore and a fracture gives the equivalent value of the feeder road network to supply traffic to the highway
was limited (similar to the conditions existing when a prop-
of rw´ for a propped fracture (qualitative relation ped hydraulic fracture is placed in a formation with very low
only): permeability). In this case, the width, or flow capacity, of the
highway is not an issue (kfw does not have to be large).
2 What is needed (and was eventually built) is a long, narrow
rw′ ≈ xf . (5-6) (low-conductivity) highway.
π As a comparison, consider Loop 610, the “superhighway”
surrounding the city of Houston. The feeder system is locat-
However, this simple flow area equivalence ed in a densely populated area, and the feeder roads are
ignores the altered pore pressure field around a linear numerous and wide. Here, the width, or flow capacity, of the
highway is critical. Making this highway longer has no effect
;;;
on traffic flow, and the only way to increase traffic flow is to
widen (i.e., increase the conductivity of) the road. This is
obviously analogous to placing a fracture in a higher perme-
. . Flow area = 2πrwh
. .. . .. . .. . . ability formation, with the postfracture production limited by
.. .. .... . . . .. .. . . .. . the fracture width (or, more accurately, limited by kfw).
. .. . ..
. .. . .. . .. .. .. .. . If CfD is the ratio of the ability of a highway to carry traffic
. . .. . .. .. .. ... . .. .. . to the ability of the feeder system to supply that traffic to the
. . . . .. . .
highway, clearly a highway should be engineered to approxi-
mately balance these conditions. That is, a CfD value > 50 is
;;;
seldom warranted, because a highway would not be con-
structed to carry 50 times more traffic than the feeder system
Flow area = could supply. In the same way, a value of 0.1 makes little
2πrw´h 2
rw´ = x sense. Why construct a highway that can only carry 10% of
π f
the available traffic? In general, an ideal value for CfD would
be expected to be about 1 to result in a balanced, well-
engineered highway system.
A balance of about 1 is certainly attractive for steady-flow
traffic conditions that may exist through most of the day.
Flow area = 4xfh However, during peak traffic periods the feeder system may
supply more traffic than normal, and if this rush hour or tran-
sient traffic period is a major consideration, then a larger ratio
of CfD may be desirable. Thus, a CfD of 10 may be desirable
for peak flow (transient) periods, as opposed to a CfD value of
approximately 1 for steady-state traffic conditions.
Figure 5-10. Equivalent wellbore radius rw′.
sient flow. For pseudoradial flow, Cinco-Ley et al. ex- 5-3.2. Complicating factors
pressed rw´ as a function of length and CfD (Fig. 5-11).
The chart in Fig. 5-11 (equivalent to Prats) can be These principal concepts give a straightforward
used (when pseudoradial flow is appropriate) as a method for predicting postfracture production; how-
powerful reservoir engineering tool to assess possi- ever, complications can reduce postfracture produc-
ble postfracture productivity benefits from propped tivity below the levels expected or give better pro-
fracturing. For example, the folds of increase (FOI) ductivity than that calculated. The major complica-
for steady-state flow can be defined as the postfrac- tions include non-Darcy (or turbulent) flow, transient
ture increase in well productivity compared with flow regimes, layered reservoirs and horizontal per-
prefracture productivity calculated from meability anisotropy (particularly any natural fissure
permeability).
ln (re / rw ) + s For high-rate wells, non-Darcy or turbulent flow
FOI = , (5-8)
ln (re / rw′ ) can be an important factor that causes an increased
pressure drop along the fracture. This creates an
where re is the well drainage or reservoir radius, rw apparent conductivity that is less than the equivalent
is the normal wellbore radius, and s is any prefrac- laminar flow conductivity. The apparent CfD is also
ture skin effect resulting from wellbore damage, reduced and productivity is less than that expected.
scale buildup, etc. An equivalent skin effect sf result- Another complicating effect that can reduce produc-
ing from a fracture is tivity from expected levels is formation layering,
s f = − ln (rw′ / rw ) (5-9) where a fracture is in multiple layers with signifi-
cantly different values for porosity, permeability or
for use in reservoir models or other productivity cal- both. Unlike radial flow into a wellbore, average val-
culations. Equation 5-8 provides the long-term FOI. ues of permeability and porosity do not apply, and
Many wells, particularly in low-permeability reser- for layered formations, postfracture performance
voirs, may exhibit much higher (but declining) early- falls below simple calculations based on average per-
time, transient FOI. The preceding relations are for meability (Bennett et al., 1986). These and other
transient pseudoradial flow before any reservoir effects are discussed in Section 12-3.
boundary effects; the case for boundary effects is For lower permeability formations and for some
discussed in Section 12-2.6. time period, postfracture performance is dominated
by transient flow (also called flush production) as
discussed by Cinco-Ley et al. (1978). For transient
conditions, reservoir flow has not developed into
0.5 pseudoradial flow patterns, and the simple rw´ rela-
tions are not applicable. In the example in Fig. 5-12,
Effective well radius rw ´
100
12 24 36 48 60 72 84 96 108
Time (months)
productivity was better than that predicted from the where qi is the total injection rate and tp is the pump-
pseudoradial flow rw´. The duration of the transient ing time for a treatment. Equally simple, the fracture
flow period is a function of permeability, CfD and xf 2 volume created during a treatment can be idealized
such that for moderate- to high-permeability wells as
the period is too short to have practical significance
for fracture design. However, it may be important for Vf = h f × w × 2 L = η × Vi , (5-11)
postfracture well test analysis. For low-permeability –
wells with long fractures, transient flow may domi- where hf is an average, gross fracture height, w is the
nate most of the productive well life. average fracture width, L is the fracture half-length
or penetration, and η is the fluid efficiency. Finally,
as discussed by Harrington et al. (1973) and Nolte
5-3.3. Reservoir effects on fluid loss (1979), the volume lost while a hydraulic fracture
treatment is being pumped can be approximated by
Reservoir properties such as permeability to reservoir
fluid, relative permeability to the fracturing fluid fil-
trate, total system compressibility, porosity, reservoir VLp ≅ 6CL hL L t p + 4 LhL Sp , (5-12)
fluid viscosity and reservoir pressure all play a role in
where CL is the fluid-loss coefficient (typically from
fluid loss while pumping (see Section 6-4). Thus, cer-
0.0005 to 0.05 ft/min1/2), hL is the permeable or
tain reservoir information is required for treatment
fluid-loss height, and Sp is the spurt loss (typically
design, as well as for specifying design goals.
from 0 to 50 gal/100 ft2). Because material balance
must be conserved, Vi must equal VLp plus Vf, and
Eqs. 5-10 through 5-12 can be rearranged to yield
5-4. Rock and fluid mechanics
qi t p
Rock and fluid mechanics (along with fluid loss) L≅ , (5-13)
6CL hL t p + 4hL Sp + 2 wh f
considerations control the created fracture dimen-
sions and geometry (i.e., fracture height hf, length showing a general relation between several impor-
L and width w). These considerations all revolve tant fracture variables and design goals.
around the net pressure pnet given by Eq. 5-3. Modeling of hydraulic fracture propagation in
However, pnet, which controls hf and L, is itself a low- to medium-permeability formations typically
function of hf and L, and the various physical behav- shows an average width of about 0.25 in. (±50%)
iors connecting height, net pressure, width, etc., over a fairly wide range of conditions (e.g., Abou-
interact in many ways. This makes simple statements Sayed, 1984). Using this value, the effect of the pri-
about the relative importance of variables difficult or mary variables height hf and fluid-loss coefficient
impossible. However, the basic physical phenomena CL on fracture penetration L are investigated in
controlling fracture growth are understood and are Fig. 5-13. This is for a simple case of a constant
well established. 0.25-in. fracture width. Figure 5-13a shows length
as a strong, nearly linear function of hf; e.g., dou-
bling hf cuts fracture penetration by 50%. For similar
5-4.1. Material balance conditions, Fig. 5-13b shows that the fluid-loss coef-
The major equation for fracturing is material bal- ficient is not as important; e.g., doubling CL reduces
ance. This simply says that during pumping a certain L by only about 20%. However, with fracturing, such
volume is pumped into the earth, some part of that is simple relations are never fixed. As seen in Fig.
lost to the formation during pumping, and the 5-13c, for a higher loss case, doubling CL from 0.005
remainder creates fracture volume (length, width and to 0.01 reveals a nearly linear relation between CL
height). It is the role of fracture models to predict and L, just as for height in Fig. 5-13a. Basically, for
how the volume is divided among these three dimen- Figs. 5-13a and 5-13b, the loss term (first term in the
sions. The volume pumped is simply denominator of Eq. 5-13) is small compared with the
fracture volume term (third term in the denominator).
Vi = qi × t p , (5-10) Therefore, the fluid loss is relatively low and fracture
(a)
fluid efficiency (η, as defined in Eq. 5-11) is high. In
Fig. 5-13c, the loss term is large compared with the
1600 volume term (high loss and low efficiency), and the
CL = 0.001 ft/min1/2
1400 qi = 30 bbl/min loss coefficient becomes the dominant variable, with
–
w = 0.25 in. hf = hL = 100 ft L less sensitive to variations in hf or equivalently w if
1200 Sp = 0
it varies from the fixed value of 0.25 in.
1000
2:1
L (ft)
800
600 5-4.2. Fracture height
400 hf = hL = 200 ft Equation 5-13 demonstrates that fracture height hf
200 and fluid-loss height hL are important parameters for
0 fracture design. Loss height is controlled by in-situ
0 20 40 60 80 100 variations of porosity and permeability. Fracture
Time (min) height is controlled by the in-situ stresses, in particu-
lar by differences in the magnitude or level of stress
(b) between various geologic layers. More formally,
height is controlled by the ratio of net pressure to
1600
hf = hL = 200 ft stress differences ∆σ, as illustrated in Fig. 5-9, where
1400 q i = 30 bbl/min
w = 0.25 in.
∆σ is the difference between stress in the boundary
1200 Sp = 0 shales and stress in the pay zone. Ignoring any pres-
1000 sure drop caused by vertical fluid flow, the relation
CL = 0.0005 among fracture height, initial fracture height, pnet
L (ft)
800
and ∆σ can be calculated as demonstrated by
600
Simonson et al. (1978). This relation is included
400 CL = 0.001 in Fig. 5-9c.
200 For cases when pnet is relatively small compared
0 with the existing stress differences (e.g., less than
0 20 40 60 80 100 50% of ∆σ), there is little vertical fracture growth
Time (min) and the hydraulic fracture is essentially perfectly
confined. This gives a simple fracture geometry
(c)
(Fig. 5-14a) and increasing net pressure (Fig. 5-14b).
For cases when pnet is much larger than the existing
600
hf = hL = 100 ft stress differences, vertical fracture height growth is
qi = 30 bbl/min
500 w = 0.25 in. essentially unrestrained. Again, the geometry is a
Sp = 0
CL = 0.005 fairly simple radial or circular fracture (Fig. 5-14c)
400
and declining net pressure (Fig. 5-14b).
For more complex cases when pnet is about equal
L (ft)
300
1.8:1 to ∆σ, fracture geometry becomes more difficult to
200 predict, and significant increases in height can occur
CL = 0.01 for small changes in net pressure. Also, for this case,
100
the viscous pressure drop from vertical flow retards
0 fracture height growth (see Weng, 1991), and the
0 20 40 60 80 100 equilibrium height calculations in Fig. 5-9 are no
Time (min) longer applicable.
Figure 5-13. Effect of hf and CL on L.
(a) Depth
detail in Chapter 6), the slit opens into an elliptical
(ft) shape, with a maximum width
4900
2 pnet d
wmax = , (5-14)
∆σ = 1500 psi
E′
5000
where E′ is the plane strain modulus (E′ = E/(1 –
ν2), ν is Poisson’s ratio and typically equals about
5100 0.2), and d is the least dimension of the fracture.
pnet < 1/3∆σ
For a confined-height fracture with a tip-to-tip length
5200
greater than hf, d equals hf. This shows a direct rela-
3500 5000 100 200 300
tion between net pressure and width and introduces
Stress (psi) Fracture penetration (ft) an important material property, the plane strain mod-
ulus. However, because typically ν2 < 0.1, the plane
(b) strain modulus seldom differs from Young’s modulus
E by a significant amount.
2000
CL = 0.002
1000 qi = 20 bbl/min
∆σ = 1500 psi hfo = 100 ft
500 5-4.4. Fluid mechanics and fluid flow
Net pressure (psi)
E = 4E+6 psi
µ = 200 cp
200 The major fluid flow parameters are the fluid viscos-
∆σ = 50 psi
100 ity (resistance to flow) µ and injection rate qi. The
50 rate also effects the pump time and hence is impor-
tant to fluid-loss and material-balance considerations,
20
Shut-in as discussed previously. Both parameters are critical
10
1 2 5 10 20 50 100 200
for proppant transport, and both parameters also
Pump time (min) affect net pressure and thus affect fracture height
and width.
As an example, consider a Newtonian fluid flowing
(c)
Depth laterally through a narrow, vertical slit (i.e., fracture)
(ft) (Fig. 5-15). For laminar flow (the general case for
4800 flow inside hydraulic fractures), the pressure drop
∆σ = 50 psi
4900 along some length ∆x of the slit is
∆pnet 12µq
5000
= . (5-15)
5100
∆x hf w3
q = qi ⁄ 2
– q
5-4.3. Fracture width w v =
hfw
Eq. 5-15 is integrated from the fracture tip back to tion in Eq. 5-16 assumes no net pressure at the frac-
the wellbore to give ture tip; i.e., fracture tip effects or fracture propaga-
tion effects are ignored. When tip effects are taken
6µqi L
pnet = . (5-16) into account, the fracture width is affected by both
hf w3 fluid viscosity and tip effects (Shlyapobersky et al.,
1988a, 1988b). As shown by Nolte (1991), tip
For this long, confined-height fracture, hf is the
effects can be approximated by considering the net
minimum fracture dimension for Eq. 5-14, and the
pressure within the tip region to equal ptip (as
fracture width and net pressure are related by
opposed to zero) in Eq. 5-16. For a positive tip pres-
E ′w sure, the net pressure equation becomes
pnet = . (5-17)
2h f 1/ 4
E′3
Combining the two equations gives the proportionality pnet ≈ 4 {κµqi L} + ptip
4
, (5-20)
h f
µq L
1/ 4
carbonates and consider confining pressures from phenomena. Tip phenomena are discussed in
0 to 5000 psi. detail in Chapters 3 and 6.
From elastic fracture mechanics, for a simple Measured values for tip extension pressure
radial or circular fracture geometry with a pene- that are higher than predicted from laboratory-
tration of L, the fracture tip extension pressure is measured rock toughness KIc can be accounted
for in hydraulic fracture calculations through the
π
ptip = K Ic , (5-21) use of the effective, or apparent, fracture tough-
48 L ness KIc-apparent (Shlyapobersky, 1985). In practice,
because KIc-apparent is not a material constant, the
and it decreases as the fracture extends. For even
tip effects should be defined or calibrated by frac-
a small fracture penetration of 25 ft, this gives a
turing pressure data for a particular situation (see
tip extension pressure of 29 psi, whereas viscous
Sidebar 9B).
pressures (Eq. 5-19) are typically 10 or more
times larger. Thus normal linear elastic fracture
mechanics considerations indicate that fracture 5-4.6. Fluid loss
mechanics, or the tip extension pressure, generally
plays a negligible role for hydraulic fracturing. As seen from the material balance (Eq. 5-13), fluid
• Apparent fracture toughness loss is a major fracture design variable characterized
by a fluid-loss coefficient CL and a spurt-loss coeffi-
Field data typically show fracture extension pres- cient Sp. Spurt loss occurs only for wall-building flu-
sure to be greater than that given by Eq. 5-21, ids and only until the filter cake is developed. For
with 100 to 300 psi as typical values and even most hydraulic fracturing cases, the lateral (and ver-
higher values possible. This difference is due to tical) extent of the fracture is much greater than the
several behaviors not included in elastic fracture invasion depth (perpendicular to the planar fracture)
mechanics calculations. One important (and long- of fluid loss into the formation. In these cases, the
recognized) consideration is that the fracturing behavior of the fluid loss into the formation is linear
fluid never quite reaches the fracture tip; i.e., there (1D) flow, and the rate of fluid flow for linear flow
is a “fluid lag” region at the tip that increases the behavior is represented by Eq. 5-1.
apparent toughness and tip pressure (Fig. 5-16). This assumption of linear flow fluid loss giving the
In other cases, tip pressure may be even greater. CL ⁄√t relation has been successfully used for fractur-
Other tip phenomena include nonelastic rock ing since its introduction by Carter (1957). The rela-
deformation near the fracture tip and tip plugging tion indicates that at any point along the fracture, the
with fines, with these mechanisms acting alone or rate of fluid loss decreases with time, and anything
in conjunction with the fluid flow and/or fluid lag that violates this assumption can cause severe prob-
lems in treatment design. For example, fluid loss to
Closure stress natural fissures can result in deep filtrate invasion
into the fissures, and the linear flow assumption may
no longer be valid. In fact, for the case of natural fis-
sures if net pressure increases with time, the fluid-
loss rate can increase, and treatment pumping behav-
ior may be quite different from that predicted. The
Fluid
lag total fluid loss from the fracture is controlled by the
total fluid-loss coefficient CL, which Howard and
Fast (1957) decomposed into the three separate
p1 p2
mechanisms illustrated in Fig. 5-17 and discussed
in Section 6-4.
The first mechanism is the wall-building character-
p1 = fracture pressure istics of the fracturing fluid, defined by the wall-
p2 ≤ reservoir pressure building coefficient Cw. This is a fluid property that
helps control fluid loss in many cases. For most frac-
Figure 5-16. Unwetted fracture tip (fluid lag). turing fluid systems, in many formations as fluid loss
pnet (psi)
(e.g., formations such as unconsolidated sands 200
with E ≤ 0.5 × 106 psi). For treatments using low-
100
viscosity fluid or pumping at very low rates, the Nolte-Smith plot
50 ∆σ = 1000 psi
viscous term of the net pressure equation becomes
qi = 15 bbl/min
small, and fracture toughness becomes a dominant 20 hf = 25 ft
E = 5 × 106 psi
parameter. Although many cases fall into one of
0
these extremes, neither effect should be over- 0 0.2 0.5 1 2 5 10 20 50 100 200 500
looked for the prudent application of fracturing. Pump time (min)
The magnitude of net pressure may also be con-
(b)
trolled by in-situ stress differences between the
Depth
pay and the bounding layers. Consider a case (ft)
where barrier zones (e.g., formations with higher
closure stress) surround the pay zone (Fig. 5-9)
and further assume that because of either viscous
4900
or toughness effects, pnet increases to the level of
the stress differences. Massive height growth then Perforated
begins, and only very small increases in the net interval
(a)
2000
µ = 200 cp
1000 2000 Nolte-Smith plot
Confined height
500 1000 Tip-dominated behavior
µ = 50 cp
pnet (psi)
200 500
µ = 50 cp
pnet (psi)
100 200
∆σ = 1000 psi
50 q = 15 bbl/min 100
i
hf = 50 ft
20 E = 5 × 106 psi 50
∆σ = 1000 psi
0 q = 15 bbl/min
20 hi = 25 ft
0 0.2 0.5 1 2 5 10 20 50 100200 500 f
E = 5 × 106 psi
Pump time (min)
0.2 0.5 1 2 5 10 20 50 100 200 500
Figure 5-19. Height growth example in a thicker hard-rock Pump time (min)
formation. (b)
Depth
(ft)
height growth occurs. For a more viscous (200-cp)
fluid, net pressure again approaches the stress dif-
ference of 1000 psi, and again extensive height 4900
growth occurs. These examples show that fracture Perforated
interval
height is a function of fracture height. 5000
Finally, consider the original (h = 25 ft) case
again, but assume this is a soft-rock (unconsoli-
dated sand with E < 0.5 × 106 psi) zone. Further 5100
sible. It also includes material selection and on-site 5-5.2. Pad volume
operational considerations as discussed in Section
5-6. The product of this process is a treatment pump For a treatment using viscous fluid, the fluid carries
schedule. This includes the pad volume necessary the proppant out to the fracture tip. For these cases
to create the desired fracture penetration, along with the pad volume determines how much fracture pene-
acid or proppant scheduling to achieve the desired tration can be achieved before proppant reaches the
postfracture conductivity. For propped fracturing, tip and stops penetration in the pay zone. Once the
pump scheduling includes fluid selection, proppant pad is depleted, a fracture may continue to propagate
selection, pad volume, maximum proppant concen- into impermeable layers until the proppant bridges in
tration to be used and a proppant addition schedule. low-width areas. Thus, pumping sufficient pad to
After the design goals and variables are defined, the create the selected length is critical. For treatments
proppant addition schedule is usually obtained by using very low viscosity fluid (i.e., “banking”-type
using a fracture simulator, although for many cases treatments), proppant settles out of the fluid and
analytical calculations based on fluid efficiency are essentially replenishes the pad. The pad volume must
also easily implemented. Chapter 10 provides addi- only be sufficient to open enough fracture width for
tional detail for treatment design. proppant admittance, and the carrying capacity of the
fluid, as opposed to the pad volume, determines the
final propped length.
5-5.1. Fluid and proppant selection On the other hand, too much pad can in some
instances be even more harmful, particularly for
Fracturing materials are discussed in Chapter 7, and cases requiring high fracture conductivity. The frac-
their performance characterization is discussed in ture tip continues to propagate after pumping stops,
Chapter 8. The major considerations for fluid selec- leaving a large, unpropped region near the fracture
tion are usually viscosity (for width, proppant trans- tip. Significant afterflow can then occur in the frac-
port or fluid-loss control) and cleanliness (after flow- ture, carrying proppant toward the tip and leaving
back) to produce maximum postfracture conduct- a poor final proppant distribution. This afterflow
ivity. Other considerations that may be major for occurs because the widest section of the fracture is
particular cases include near the wellbore at shut-in, and most of the prop-
• compatibility with reservoir fluids and reservoir rock pant pumped is stored there. However, the highest
• compatibility with reservoir pressure (e.g., foams fluid-loss rates are near the fracture tip. Thus, prop-
to aid flowback in low-pressure reservoirs) pant-laden slurry continues to flow toward the tip of
the fracture. Afterflow continues until either the frac-
• surface pump pressure or pipe friction considerations ture closes on the proppant, stopping proppant move-
• cost ment, or until proppant-laden slurry reaches the frac-
• compatibility with other materials (e.g., resin- ture tip. At that point the slurry dehydrates and stops
coated proppant) any additional fracture propagation. Ideally, of
• safety and environmental concerns (see Chapter 11). course, it is better to have the proppant at the frac-
ture tip at shut-in and thus minimize afterflow.
Proppant selection must consider conductivity at An ideal schedule for a normal treatment (as
in-situ stress conditions (i.e., the effect of stress on opposed to subsequently discussed TSO designs) is
proppant permeability kf). Proppant size must also be one where the pad depletes and proppant reaches the
considered. In general, bigger proppant yields better fracture tip just as the desired fracture penetration is
conductivity, but size must be checked against prop- achieved and also just as pumping stops. This is the
pant admittance criteria, both through the perforations sequence in Figs. 5-2, 5-3 and 5-4.
and inside the fracture (see Section 5-5.4). Finally, The critical parameter of the pad volume or pad
the maximum in-situ proppant concentration at shut- fraction fpad is related directly to the fluid efficiency
in must be selected, as it determines how much of the for a treatment (Nolte, 1986b). This relation from
hydraulic width created by the fracture treatment will Sidebar 6L gives the pad volume expressed as
be retained as propped width once the fracture closes. a fraction of the entire treatment volume:
1− η
f pad ≈ . (5-22) 1.0
1+ η % of fracture filled by proppant pack
Pack porosity = 0.35
0.8
That is, a treatment with an expected efficiency η
of 50% would require a pad fraction of about 1⁄3. As
Fill fraction, F
0.6
discussed in Chapter 9, the efficiency for a specific Proppant s.g. = 2.65
(sand)
formation and fluid system can be determined by a 0.4
calibration treatment. Proppant s.g. = 3.2
(intermediate strength)
This discussion of pad volume has so far concen- 0.2
trated on the fluid-loss aspects of the pad volume;
i.e., the pad is pumped first to serve as a sacrificial 0.0
0 4 8 12 16 20 24
stage of the treatment to enable the fracture to pene-
Proppant concentration (lbm/gal)
trate into permeable formations. This important effect
of the pad volume may be the critical aspect govern- Figure 5-21. Fill fraction versus proppant concentration.
ing the size of the pad for most applications. How-
ever, hydraulic fracturing is complicated, in that most
things are done for at least two reasons, which where F is the fill fraction (Fig. 5-21), the con-
applies to pad volume specification. The second pur- stant 8.33 converts the units to lbm/gal, γprop is
pose of the pad volume is to create sufficient fracture the specific gravity (s.g.) of the proppant, C is the
width to allow proppant to enter the fracture (see final in-situ proppant concentration at shut-in
Section 5-5.4 on proppant admittance). Even for a expressed as pounds of proppant per fluid gallon
case of very low fluid loss, some minimum pad vol- (ppg), and φ is the porosity of the proppant pack,
ume is required. Both of these aspects of the pad vol- typically about 0.35.
ume must always be considered for treatment design. Increasing the concentration from 8 (F ≈ 0.4)
• Propped width to 16 ppg (F ≈ 0.6) significantly increases the
propped fracture width (50% increase in the fill
A major design goal is fracture conductivity kf w,
fraction). However, this large increase in propped
which consists of proppant pack permeability and
width is accomplished at the expense of additional
propped fracture width. Proppant permeability kf
risk to the job and to the well, because of either
is a function of the proppant selected, in-situ
surface mechanical problems or an unexpected
stress and residual damage from fluid additives
total screenout somewhere in the fracture or in the
(see Chapter 8). Propped width is controlled by
near-wellbore region between the well and the far-
the treatment design.
field fracture (see the discussion of tortuosity in
The effective propped width wp-eff is a function
Section 6-6). In practice, most treatments use a
of the average fracture width wf at shutdown (i.e.,
maximum concentration of about 8 to 14 ppg,
hydraulic width at the end of pumping a treat-
although concentrations of 20 ppg have been
ment), proppant concentration C in the fracture at
pumped.
that time (i.e., giving the ideal propped width wp)
Another manner of increasing propped width is
and the volume of proppant wlost that is lost on the
to increase fracture width. Theoretical and numer-
faces of the fracture to embedment, gel residue,
ical models generally show that the fracture
etc. (usually expressed as lbm/ft2 “lost”). In terms
width, while the fracture is growing, is relatively
of these parameters, the effective propped width
insensitive to the controllable job variables of
can be expressed as
pump rate and fluid viscosity. For a simple frac-
w p−eff = w p − wlost = w f × F − wlost (5-23) ture geometry, width is proportional to rate and
viscosity raised to a small power. For Eq. 5-18
with the exponent 1⁄4, doubling the pump rate
C
F= increases fracture width by only 18%, at the
( )
,
8.33 × γ prop + C × (1 − φ)
(5-24)
expense of significant pipe friction and surface
pressure increases. Viscosity is easily increased by ture (see Section 6-5). First is what may be termed
an order of magnitude (e.g., 10 times increase in µ simple or single-particle settling. Behavior of this
increases the width by 77%), but only at the type is governed by Stokes law, in which the veloc-
expense of using more fluid additives and with ity of a single particle falling through a liquid medi-
additional conductivity damage potentially negat- um is
ing the extra width.
Thus, the hydraulic fracture width is fairly fixed v fall = 1.15 × 10 3
d 2prop
µ
( )
γ prop − γ fluid , (5-25)
(±50%, at least in terms of the treatment’s control-
lable parameters), and the proppant fill fraction has where vfall is the settling rate in ft/s, dprop is the aver-
a practical limit of about 0.5 (±0.1). Without TSO age proppant particle diameter in in., µ is the fluid
designs (discussed in the following) the final, viscosity in cp, and γprop and γfluid are the specific
effective propped width is almost fixed by nature. gravity of the proppant and the fluid, respectively.
The goal for a normal fracture design is then to The settling rate, and thus the efficiency with which
achieve a required kf w within these limits, with proppant can be transported into the fracture, is
proppant concentration, proppant selection and directly related to the fluid viscosity. This is usually
fluid selection allowing a large range of values. the main consideration for how much viscosity is
• Tip-screenout designs required for a fracture treatment. However, there are
As mentioned previously, as long as a fracture is additional considerations for calculating settling fol-
free to propagate, the hydraulic fracture width is lowing Stokes law. At low proppant concentrations
relatively insensitive to the controllable treatment (e.g., less than 1 or 2 ppg) particles may clump, pro-
parameters of fluid viscosity and pump rate. If ducing an apparent diameter greater than the actual
more conductivity is required than can be particle diameter and accelerating settling. Higher
achieved from a normal design, the only effective particle concentrations act to increase the slurry vis-
manner to increase the propped width is to stop cosity and retard settling (also known as hindered
the fracture from propagating but to continue to settling). The pump rate is also an important param-
pump. This technique has come to be called TSO eter controlling proppant transport for simple settling
fracturing (Smith et al., 1984). by Stokes law.
For a normal treatment, the pad volume is As shown in Fig. 5-22, for a Newtonian fluid the
designed to deplete just as pumping stops. What distance D a proppant particle is transported into a
would happen if pumping simply continued fracture, before that particle can fall from the top of
beyond that time? If the pad is depleted, then
proppant-laden slurry will be located everywhere
around the fracture periphery. If there is fluid loss,
then this slurry will dehydrate and leave packed
proppant around the periphery. Even with no fluid v1
loss, the proppant may bridge in the narrow frac- h
ture width around the periphery, particularly in vfall
places where the width is extremely narrow as a
result of the fracture penetrating a boundary layer.
In either case, any additional propagation is D
restricted and further pumping causes an increase
D ⁄ h = v1 /v 2
of net pressure and thus an increase of fracture
width. TSO designs are discussed in detail in v1 = fluid velocity
Chapter 10. ∝ qi/hw ∝ qi/h(µqi)1/4
∝ qi3/4/hµ1/4
vfall = fall rate ∝ 1/µ
5-5.3. Proppant transport D/h ∝ (qiµ)3/4/h
D is independent of h
Several modes of proppant settling can occur while
proppant is being transported into a hydraulic frac- Figure 5-22. Stokes law.
the fracture to the bottom, is related to (qiµ)3/4. This fluid (which describes most fracturing fluid systems)
distance is independent of the fracture height and, flowing down a channel imparts a normal force to
more significantly, shows that for some given trans- particles entrained in the fluid such that the particles
port distance, less viscosity can be used at higher tend to migrate to and concentrate in the center of
pump rates. This relation can be important for higher the channel. For low average concentrations, this can
temperature applications, where fluid viscosity can result in a center core of high-proppant-concentration
degrade significantly with time. At higher rates (and slurry, with a region of essentially clean fluid on
hence shorter pump times), less viscosity is required either side. This heavier core of concentrated slurry
for proppant transport. Also, the fluid is exposed to tends to fall owing to its greater density, carrying the
the high formation temperature for less time, so the entrained proppant toward the bottom of the fracture
fluid system maintains better viscosity. In general, at a faster rate than for a dispersed slurry (Nolte,
considering how fluid viscosity degrades down a 1988b).
fracture, including the effect of proppant concentra- Finally, any calculations for proppant settling must
tion increasing the effective slurry viscosity, and consider geologic reality. Detailed examinations of
considering the non-Newtonian nature of most hydraulic fractures both at the wellbore using televi-
fracturing fluids, if a fracturing fluid retains 50- to sion cameras (Smith et al., 1982) or away from wells
100-cp viscosity (at reservoir temperature and at a in mineback tests (see Warpinski, 1985) show some-
shear rate of 170 s–1) at the end of the fracture treat- thing other than the smooth fracture walls assumed
ment, it will provide essentially perfect proppant for settling calculations. Small shifts and jogs of the
transport (Nolte, 1982). fracture probably have no significant impact on fluid
The next mode of proppant settling is termed con- flow or on lateral proppant transport into the frac-
vection, and it was probably first included in fracture ture. However, these small irregularities could signif-
modeling in the context of a fully three-dimensional icantly impact settling. Calculations for proppant set-
(3D) planar model by Clifton and Wang (1988). This tling that ignore these effects will be a worst-case
type of settling is controlled by density differences scenario.
(i.e., buoyancy) between two fluids. For example,
a proppant-laden fluid stage with an 8-ppg concen-
tration has a slurry density of 11.9 lbm/gal (s.g. = 5-5.4. Proppant admittance
1.44). If this slurry is placed directly next to a clean Proppant admittance is critical to hydraulic fracturing
fluid stage with a density of 8.5 lbm/gal (s.g. = 1.02), in two forms: entrance to the fracture through perfora-
the heavier slurry will tend to sink and underride the tions and entrance of proppant into the fracture itself.
lighter clean fluid, simply carrying the proppant These effects were recognized early, and the original
toward the bottom of the fracture. However, a treat- fracture width models were used primarily for deter-
ment does not normally follow clean pad fluid with mining a pad volume that would allow admittance by
a heavy 8-ppg slurry. Rather, the treatment increases generating a fracture width greater than 2.5dprop,
proppant concentration slowly to account for fluid- where dprop is the average proppant particle diameter.
loss effects and mitigate convection effects. Only Before these models, operators were reluctant to
near the end of pumping (when the need for trans- pump significant volumes of pad as it was considered
port decreases), when the initial proppant stages expensive and potentially damaging.
have undergone significant dehydration, can a signif- The laboratory data in Fig. 5-23 (Gruesbeck and
icant density difference begin to develop. In general, Collins, 1978) illustrate two important ideas:
rigorous numerical modeling of this phenomena
shows convection is not a major factor during pump- • A minimum perforation diameter is required for
ing (Smith et al., 1997). If excessive pad is used, proppant to flow through the perforations.
such that a large unpropped region of the fracture • Minimum perforation diameter is a function of the
exists after shut-in, convection can occur during the slurry concentration.
shut-in after flow, with potentially significant adverse At low concentrations (e.g., 1 ppg), the perforation
effects on the final proppant placement. hole diameter must be only slightly greater than that
The third effect on proppant transport is termed of the proppant particles. The required hole diameter
migration (see Chapter 6). In brief, a viscoelastic increases with concentration until at about 6 ppg
6
Experimental Correlation
5 Bridge Formation‡ Bridge
100-cp HEC solution
4 0.5 to 2 1.8 1.15 to 2.0
Tap water
3
2 to 5 2.2 2.0 to 3.0
2
5 to 8 2.6 3.0
1 † Sand as proppant
‡ Data from van der Vlis et al. (1975)
0
0 5 10 15 20 25 30 35
Sand concentration
(lbm sand/gal fluid)
6 to 8 ppg, the required average fracture width
Figure 5-23. Proppant admittance through perforations increases to 3dprop.
(Gruesbeck and Collins, 1978). This critical width is important to the hydraulic
fracturing process. Should proppant enter a part of the
fracture where sufficient width does not exist, the
(solid volume fraction of about 0.20), the perforation proppant will bridge and no longer flow down the
hole diameter must be 6 times the average particle fracture. Additional slurry flowing in this direction
diameter. will cause proppant to pile up, dehydrate and block
This same trend applies for slurry flow down a nar- that part of the fracture. Should this occur near the
row fracture. An approximate proppant bridging or wellbore, possibly as a result of some form of near-
proppant admittance criteria can be derived by calcu- wellbore width restriction (see tortuosity discussion in
lating an equivalent hydraulic radius for a narrow Section 6-8), a total screenout can result with serious
slot, rhyd = w/2, where w is the average width of the consequences for the success of the fracture treatment.
fracture. For a round perforation hole, the hydraulic
radius is d/4, where d is the perforation hole diameter.
Equating the two hydraulic radius values shows that 5-5.5. Fracture models
2w is equivalent to the diameter of a round hole.
Using this along with two lines fitting the data Clearly, developing a final treatment pump schedule
of Gruesbeck and Collins leads to an approximate must consider many options. The interactive roles of
admittance criteria for a hydraulic fracture: the various major variables (hf, E, CL, KIc-apparent, µ
and qi) must be considered along with the various
• For a proppant solid volume fraction fv less than roles of fluid viscosity for net pressure, width, prop-
0.17, the average width must be greater than pant transport and fluid loss. In addition, the design
(1 + 2fv /0.17) × dprop. must consider the various roles of the pad volume
• For fv greater than 0.17, the average width must be concerning fluid loss and creating fracture width.
greater than 3dprop (i.e., a width greater than three Fracture simulators, or fracture placement models,
proppant grain diameters). provide the means to handle this complexity and to
consider the interaction of the multitude of variables.
This approximate correlation also compares well
For this reason, a final schedule is generally devel-
with other experimental data from proppant-laden
oped using a fracture geometry model. However, as
slurry flowed through a narrow slot (van der Vlis et
discussed in Section 5-5.2, Sidebar 6L and Section
al., 1975), although the correlation may be optimistic
10-4, in many instances an acceptable pump schedule
for low proppant concentrations. As shown in Table
can be developed more simply for a treatment on the
5-1, the behavior for bridging in a fracture is similar
basis of the expected fluid efficiency (as determined
to bridging in perforation holes. At low proppant
from a calibration treatment). The use of a properly
concentrations, the average fracture width must be
calibrated fracture geometry model also enables the
only slightly greater than the average particle diame-
consideration of multiple scenarios for designing the
ter. As the proppant concentration increases toward
optimum treatment for a specific application. This with a large hydraulic fracture may drain the reser-
approach is briefly discussed in Section 5-6.1. voir much faster, making the economics much more
attractive despite the additional cost of the treatment.
Carrying this forward, 2, 10 or 100 or more wells
5-6. Economics and operational could be drilled and/or fractured. Between these
considerations extremes is the optimum plan, which is the number
of wells, number of fractured wells or both that max-
The preceding discussion covers most of the techni-
imize the economic value of the production com-
cal aspects of hydraulic fracturing (reservoir engi-
pared with the development capital costs and the
neering, fluid mechanics, rock mechanics, etc.) and
ongoing operating costs.
reviews the complex interactions that exist between
As a simple example, the process (at least for a
the various, often competing design variables.
single well) could proceed as pictured in Fig. 5-24
However, to complicate things further, hydraulic
(Veatch, 1986). First, reservoir engineering calcula-
fracturing and treatment design are generally gov-
tions provide a production forecast for various com-
erned by—or are at least sensitive to—two final
binations of fracture half-length xf and conductivity
considerations: economics and field operations.
kf w (including the case of no fracture at all). Based
on some future price forecast, this allows calculation
5-6.1. Economics of a present value, which is the future revenue from
the production less future operating costs and dis-
At the most basic level, hydraulic fracturing is about counted back to the present. Hydraulic fracturing
time and money: “economics.” Given reasonable calculations based on fluid loss, fracture height, etc.,
geologic continuity, a single well would, given suffi- are used to determine the treatment volumes required
cient time, drain an entire reservoir. However, the to generate various combinations of fracture length
operating costs of maintaining a well over the and propped fracture width, and these calculations
decades required to accomplish this drainage would are easily converted into estimated treatment costs.
probably make the entire operation unattractive from Some form of net revenue economic analysis is then
a commercial viewpoint. Alternatively, a single well used to determine the best type of proppant, desired
Discounted revenue ($)
Cumulative production
x f = x f2
kfw = kfw2 kfw = kfw2
x f = x f1
kfw = kfw1 kfw = kfw1
No fracture
Revenue less cost ($)
kfw = kfw2
Treatment volume
xf
Cost ($)
kfw = kfw2
kfw = kfw1
xf xf
fracture length and other requirements for the opti- to the hydrostatic head of the fluid in the wellbore
mum treatment. phead and the pipe friction ppipe friction:
There are, of course, many variations of this basic
process. For example, full-cycle economics includes psurf = σ c + pnet + p pipe friction − phead (5-26)
drilling and other completion costs, along with frac-
ture treatment costs, in determining the optimum
fracture design. This type of analysis is usually hhp ∝ qi × psurf . (5-27)
appropriate in any case involving multiple wells
(e.g., should a resource be developed using 10 wells Pipe friction is a major term, and thus the size
with huge fractures or 20 wells with smaller or no of the well tubulars has a strong influence on
fracture treatments?). Point-forward analysis, on the allowable pump rates (because pipe friction is typ-
other hand, considers only the fracture treatment ically related to ve, where v = qi/A is the flow
costs (because drilling and other completion costs velocity down the tubing, and e is typically about
are already expended) and is most appropriate for 1.1 to 1.7). Also, the strength and condition of the
working over existing wells. tubulars (along with the associated wellhead
equipment) set an allowable surface pressure and
thus indirectly set an allowable injection rate. In
5-6.2. Operations addition, the size, type and condition of the well-
As discussed in the preceding section, economics bore tubulars may limit (or prohibit) future work-
provides the final design consideration for hydraulic over and recompletion opportunities.
fracturing, whereas field conditions provide the prac- A critical aspect of wellbore considerations is a
tical limits within which the design must fit. Even good cement job around the casing or liner to pro-
beyond defining these limiting conditions, however, vide zonal isolation. In general, a fracture grows
any design is only as good as its execution; thus the where nature dictates, and the engineer has little
treatment must be pumped as designed. Field opera- control over fracture height growth. The only con-
tions and operational considerations impact trol possible is the ability to specify where the
hydraulic fracturing in two ways: perforations are placed and the fracture initiates.
If that ability is compromised by a poor cement
• prefracture condition of the wellbore, quality of sheath around the casing that allows the perfora-
the cement job, perforations, pressure limits, etc., tions to communicate directly with an undesired
with these considerations defining practical limits interval, then even this minimal level of control is
that the design must meet lost, and the hydraulic fracture treatment may be
• quality assurance and quality control (QA/QC) seriously compromised.
before and during the actual treatment. Another important consideration is the perfora-
These operational considerations are discussed in tions that allow the fluid to leave the wellbore and
Chapters 7 and 11, with some of the major items create the fracture. The number and size of the
highlighted in the following. perforation holes are important for proppant
admittance, as discussed briefly in Section 5-5.4
• Wellbore considerations and in detail in Section 11-3.
Some of the major wellbore considerations for • Quality assurance and quality control
hydraulic fracturing include
Quality issues are critical for hydraulic fracturing.
– size and condition of wellbore tubulars After proppant pumping starts, a treatment cannot
– quality of the cement job for zonal isolation be stopped because of problems without signifi-
– perforations cantly compromising the design goals. For this
time period, everything must work, including
– wellbore deviation.
the wellbore equipment, pumping and blending
During a hydraulic fracture treatment, the pre- equipment and chemicals (i.e., the fluid system).
dicted surface pressure psurf and the hydraulic To cite a simple example, if a treatment uses
horsepower required for a treatment are related 10 tanks of batch-mixed fluid, and one of the