Professional Documents
Culture Documents
Injection Shut-in
tion and expensive instrumentation currently limit 9000
Fracture Transient reservoir
widespread application. closing pressure near the wellbore
wmax d
σmin wmax
σmin
pf
4520
0.6
4530
interval
4550 Micro-fracture 0.4
test
Depth (ft)
4560
Perforations
4570
0.2
4580
4600 0
0 50 100 150 200 250 3000 3200 3400 3600
Gamma ray (API units) Bottomhole pressure (psi)
Figure 9A-2. Gamma ray log showing the micro- Figure 9A-4. Normalized inclinometer response versus
fracture test location and perforated interval. bottomhole pressure.
the height. Consequently, the 2D model is valid in as is the case for a horizontal fracture in a vertical
cases where the fracture length is either relatively wellbore, or as an intermediate condition between
small or large in comparison with the height. In the two limiting cases of the 2D models.
practice, these models are applicable when the • Correction for fluid pressure gradient
dimensions differ by a factor of about 3 or more.
The fundamental elastic relations (Eqs. 9-10
The radial model is most appropriate when the
through 9-13) assume that the pressure in the frac-
total length 2L (2R in Fig. 9-3) is approximately
ture is constant. The fluid flow relation (Eq. 9-2),
equal to the height. This condition occurs for frac-
however, indicates that a pressure gradient exists
ture propagation from a point source of injection,
p 2n + 2
1 ( 2n + 2 )
K Ic −apparent =
2
260 psi
(38 ft × 12 ft / in.) = 4430 psi / in.1 2.
= ∆p µ + 1
tip
p net , (9B-2) π 2
∆p µ
This estimate for KIc-apparent is higher than the critical stress
where the fracture half-length L > h/2 and ∆pµ is the PKN pres- intensity factor KIc commonly measured with laboratory tests
sure contribution from the tip to the wellbore but without the (see Section 3-4.6). The larger value can be attributed to any
rock resistance to propagation. An analytical relation for ∆pµ of the tip mechanisms described in Section 6-7 and results in
for the PKN fracture model beyond the tip region is derived as a correspondingly higher resistance to fracture propagation.
This discussion applies exclusively to an elongated fracture
1 ( 2n + 2 )
E ′ Kh q i geometry, as approximated by the PKN fracture model. In con-
n
∆p µ ≅ p net ,PKN ≅ 1.5 (L − Lt ) , (9B-3) trast, both ptip and ∆pµ decrease with continued injection for the
h E ′ h radial fracture model. As a result, it is less clear which mecha-
nism dominates the pressure response for radial fractures.
where E ′ is the plane strain modulus, K is the fluid consistency
coefficient, qi is the fluid injection rate, and Lt is the length of
the tip region. Equation 5-19 provides a more specific relation- 1000
ship for pnet with a Newtonian fluid.
The relative contributions of ∆pµ and ptip determine whether Toughness dominated Viscous dominated
the fracture growth is viscosity dominated or tip dominated,
respectively. Usually one phenomena dominates; a nearly
equal contribution from both mechanisms is only rarely
observed. Although ptip generally shows little variation during
pnet
2
After shut-in
Radial,
0.5 β = ∆p f/pnet where L is the fracture half-length.
βp = 1/1.5 = 0.67 (before)
0 βs = 1/1.3 = 0.77 (after) Martins and Harper (1985) derived the compli-
ance for a fracture that grows as a series of con-
focal ellipses. In this case, the fracture width
Flow rate in fracture
1.0
depends on the elliptic integral of the fracture
Injection rate
Before shut-in
After shut-in aspect ratio. This analysis is applicable during
0.5
Decreasing during closure
fracture growth in an unbounded fashion following
0
initiation from a perforated interval that is shorter
Well 1/2 Tip than the fracture height.
Distance into fracture
KGD w ∝ i n (9-27) Equation 9-31 indicates that for typical fracturing flu-
E ′ h f
e
ids (i.e., n ≈ 0.4–1.0), the fracture pressure during
Kqin 2−n e
[R ] .
Radial w ∝ injection is only nominally sensitive to the reduced
E′ time t* or fluid efficiency η. Consequently, the effi-
These fracturing pressure and width relations indicate ciency, or alternatively the fluid-leakoff coefficient,
that their dependence on the fluid rheology parame- cannot be determined by analyzing pressure during
ters K and n, wellbore injection rate qi and plane fluid injection exclusively.
strain modulus E′ is the same for all the models. Equations 9-30 and 9-31 also show that the net
Their dependence on the fracture extension L or R pressure and fracture width for any efficiency η can
and height hf differs. In addition, Eqs. 9-25 and 9-27 be approximated by their values for the case of no
also show that pnet and w, respectively, have a weak fluid loss, if the time is scaled by ηt. This time scal-
dependence on qi and that for increasing penetration ing is illustrated for the PKN fracture model in
L or R, pnet increases for the PKN model but Fig. 9-5, which shows the net pressure corresponding
decreases for the KGD and radial models. to no fluid loss (i.e., η = 1) and to an efficiency η =
The time dependence of the fracture width and 0.2 at a time of 50 min. The latter case corresponds
pressure is developed using the definition of η from to a reduced time of t* = ηt = 50 × 0.2 = 10 min.
Eq. 9-4 at a constant injection rate qi (i.e., Vi = qit): Figure 9-5 illustrates that the net pressure at a time of
50 min for the fluid-loss case is equal to the net pres-
Efficiency, η
400 0.4 are similarly outlined in the Appendix:
pnet (psi)
pnet ∝ t 1/ 4 ( n+1) η→ 0
PKN (9-35)
200 0.2 pnet ∝ t 1 ( 2 n+3) η→1
pnet ∝ t − n /2 ( n+1)
Net pressure
100
Efficiency
0.1
η→ 0
KGD (9-36)
2 5 10 20 50 pnet ∝ t − n ( n+2 ) η→1
Injection time (min)
pnet ∝ t −3n /8( n+1) η→ 0
Figure 9-5. Reduced time illustrated for PKN fracture Radial (9-37)
geometry (Nolte, 1991). pnet ∝ t − n ( n+2 ) η → 1.
Each of these bounding expressions for the net
sure derived for the particular case of no fluid loss at pressure is a power law relation. Consequently, the
a time of 10 min. log-log graph of net pressure versus time should yield
This observation is significant because simple ana- a straight line with a slope equal to the respective
lytical expressions for the three basic models are exponent: positive for PKN behavior and negative for
readily available when η → 1. The various fracture KGD and radial behavior. In particular for PKN
parameters for any generalized value of η can then behavior, the log-log slope for commonly used frac-
be obtained from this limiting conditions merely by turing fluids (i.e., n ≅ 0.5) is typically less than 1⁄4 and
scaling the time by a factor of 1/η. decreases as the efficiency decreases. The log-log
plot of the net pressure versus time during injection,
commonly known as the Nolte-Smith plot, forms the
9-4.1. Time variation for limiting fundamental basis for the interpretation of pressure
fluid efficiencies profiles during fracturing and is analogous to the log-
log diagnostic plot for reservoir flow, as discussed in
Approximations for the time dependency of the frac- Chapter 2.
ture penetration and pressure can be derived from the
equations presented in the previous section for the
two extreme values of the fluid efficiency η. These 9-4.2. Inference of fracture geometry
limiting cases are for very high and low fluid effi- from pressure
ciencies, approaching 1 and 0, respectively. This sim-
plification provides bounding expressions for the The primary reservoir interval is bounded on both
fracture penetration and related pressure. A similar sides by shale formations in the majority of fracturing
approach is used in Section 9-5 to derive relations for applications. Shale zones are generally at higher
analyzing pressure decline during the shut-in period. stress and provide the primary barrier to fracture
Following the mathematical derivations outlined in height growth, particularly during the initial stage
“Mathematical relations for fluid loss” in the Appen- of fracture propagation. The restriction of fracture
dix to this chapter, it can be shown that the fracture height growth is important in low- to moderate-per-
penetration is bounded in the following fashion: meability formations, where relatively long fractures
are required for effective stimulation. Figure 9-6
shows the evolution of the fracture geometry and the
PKN L ∝ t 1/ 2 η→ 0 (9-32)
corresponding wellbore pressure for fracture propaga-
L ∝ t ( 2 n + 2 ) ( 2 n +3 ) η→1 tion under these conditions.
The initial character of fracture propagation, labeled
KGD L ∝ t 1/ 2 η→ 0 (9-33)
as stage 1 on the figure, depends on the length of the
L ∝ t ( n+1) ( n+2 ) η→1 perforation interval providing fluid entry into the frac-
ture relative to the reservoir thickness. Two limiting
+ + + ++ +
Radial
For either the radial or elliptical propagation mode
model KGD model during stage 1, the net pressure decreases with con-
+
3
1
⁄4 and 1⁄8, once again depending on n and η (Eq. 9-35).
2
Confined fracture height with its characteristic
1
positive log-log slope can be expected until the frac-
turing net pressure approaches a value that is approxi-
log (time) mately one-half of the stress difference ∆σ to which-
ever stress barrier bounding the fracture has the lower
Figure 9-6. Evolution of fracture geometry and pressure stress value. At this magnitude of the net pressure,
during pumping.
the fracture begins to penetrate in a restricted, or con-
trolled, fashion into the adjacent barrier layer with the
cases are described: a limited fluid entry interval and lower stress value. The fracturing pressure continues
one where fluid entry occurs over the complete reser- to increase with penetration, although at a rate that is
voir thickness. Short fluid entry intervals (i.e., limited progressively less than for the PKN model. This con-
perforation intervals) may be desired in vertical well- dition of fracture propagation is indicated as stage 3
bores to mitigate the occurrence of near-wellbore in Fig. 9-6.
problems (see Section 11-3.2). They also occur in hor- If one of the formation barriers is absent (i.e., ∆σ
izontally oriented fractures, during the placement of = 0), height growth into the higher stress barrier is
transverse hydraulic fractures in a horizontal well or arrested. The fracture height, however, continues to
in wellbores that are inclined with respect to the plane grow essentially in a radial-like fashion along the
of σmin. The limited fluid entry into the fracture is direction where the barrier is absent and exhibits a
approximated by a point source. As shown in Fig. 9-6, continuously decreasing pressure (stage 1). This frac-
the fracture area increases in a circular shape for a ture height growth pattern could also occur when
point-source fluid entry and hence is best described fractures are deliberately initiated from zones at
by the radial geometry model. Fluid entry over the higher stress and propagated into bounding layers
complete reservoir thickness is approximated by a line at lower stress, as during an indirect vertical fracture
completion (IVFC; see Section 5-1.2).
9400 20
9-4.4. Examples of injection 9200
Calibration test 18 Step rate
Bottomhole pressure (psi)
pressure analysis 16
Injection rate (bbl/min)
Tip screenout
dp w
t = Abt b . (9C-3) Begin proppant
dt
1000
pnet
Defining the left side of Eq. 9C-3 as the pressure deriva-
tive, it follows that
• Pressure derivative versus time exhibits the same log-log 500
slope as the net pressure.
• Net pressure and pressure derivative are separated by a
factor of 1/b on a log-log plot.
The pressure derivative is independent of the particular 200
choice of closure pressure and is thus unaffected by errors Pressure derivative
in its determination. For typical PKN, KGD and radial fracture
behavior, the closure pressure can thus be inferred from the
injection pressure by selecting a value that makes the net 100
pressure response parallel to the pressure derivative on a log- 5 10 20 50 100
log plot (Fig. 9C-1). This feature of pressure derivative analy- Injection time (min)
sis was applied to the calibration test in Fig. 9-8 to confirm the
closure pressure magnitude (Fig. 9-9).
The pressure derivative magnifies fracturing events Figure 9C-2. Pressure derivative analysis for a TSO
because of its enhanced sensitivity. This characteristic of the response (Nolte, 1991).
pressure derivative is used to quantify fracture height growth
into higher stress bounding zones, as for the examples in
Section 9-4.4. The occurrence of a tip screenout (TSO) is also
magnified and can be detected earlier in time. This is noted on
Height
Radial PKN growth
pnet or pressure derivative (psi)
100
10
pnet
1 Pressure derivative
0 1 10
Time (min)
1 d ( pnet ) 2500 0
≈ 0.1
∆σ dt
t (9-38) 25 75 125 175
Time (min)
b
log pnet
a c
Net pressure, pnet (psi)
log t
pnet
100
0 1 10
Time (min) hi a b c a b c
1 hf/hi
Figure 9-11. Log-log net pressure analysis for the calibra-
tion treatment in Fig. 9-10.
20
30
10
3500 20
10
2500 0
50 100 150
Treatment time (min)
(b)
5400
5600
Depth (ft)
5800
6000
3500 5000 –0.6 0.6 0 500 1000 1500
Stress (psi) Width (ft) Fracture half-length (ft)
Figure 9-13. Pressure response for accelerated height growth. (a) Bottomhole pressure match plot. Calculated pressure is
from surface pressure. (b) Fracture profile at the end of injection.
trolled vertical growth beyond a pinch point, as in al., 1988). The horizontal fracture component
Fig. 9-12. Although uncontrolled vertical growth increases the area available for fluid loss and
commences at a pressure less than that of the stress decreases the treatment efficiency. In addition, the
barrier, T-shaped fracture growth occurs at a pres- horizontal component readily accepts fluid but pre-
sure slightly larger than the vertical stress. vents proppant from entering because of its limited
The width of the horizontal fracture component width. Both effects can excessively dehydrate the
is narrow and has twin pinch points at the juncture slurry in the vertical component, which could lead
with the vertical component because of the elastic to premature screenout.
interaction of the two components (Vandamme et
b c pw ≈ overburden
100
a
10 100
Time (min)
log t
Figure 9-15. Calibration treatment pressure response for a
Overburden T-shaped fracture.
100
Injection
Permeabiltiy (normalized)
CL,fissure/CL
well is drawn down to extract stimulation fluids and the natural
fissure permeability decreases. Thus, stimulation fluids are 1.5
injected under wide-open fissure conditions but produced
under clamped fissure conditions, making it difficult to clean √∆p
up the reservoir. 1.0
The Walsh and other models can be incorporated into a
fluid-loss equation to represent changing leakoff conditions
0.5
(Warpinski, 1991). Figure 9D-3 shows the calculated pres-
sure-sensitive leakoff of a Mesaverde fissure system com-
pared with the normal pressure sensitivity of a conventional 0
pore space (recall that the filtrate leakoff coefficient is propor- 0 200 400 600 800 1000 1200 1400
tional to the square root of the fracturing pressure minus the
Net treatment pressure (psi)
reservoir pressure). Generally, the pressure sensitivity of con-
ventional reservoirs is ignored because the changes are
small; in this case, it varied from 1.0 to 1.4. However, the Figure 9D-3. Pressure-sensitive leakoff of a Mesa-
pressure sensitivity of the fissures greatly exceeded this
change, and the leakoff of the fissures reached about 3 times
verde fissure system.
the conventional leakoff. Even at low fracturing pressures, the
leakoff of the fissures was greater than conventional leakoff
decreasing slope with decreasing pressure. Fissure-sensitive
because the fracturing pressure was large relative to the
fluid loss can be recognized in the injection pressure behav-
reservoir pressure and the fissure pores had much less stress
ior, but it is difficult to identify because it looks much like
closing them. This example is based on a tight Mesaverde
height growth during the injection (i.e., a nearly constant pres-
coastal zone reservoir with a base leakoff coefficient CL of
sure derivative). Fissure dilation is usually followed by flatten-
0.0004–0.0006 ft/min1⁄2 (measured during pressure declines)
ing of the fracturing pressure, and screenout most likely
and a fracture-calibrated leakoff coefficient CL,fissure of
occurs relatively fast, depending on the injected proppant
0.0015–0.0019 ft/min1⁄2 during injections.
concentration (Fig. 9-16).
Pressure-sensitive fissure behavior is best recognized in
the pressure decline where the G-plot shows a continuously
600
the magnitude triggering complex fracture growth.
400
5 10 20 50 100 200
2.0
pumping, as discussed in Section 9-5. The only dif- ∆tso
∆tDso = =2
tso
ference, however, is that although no additional fluid 1.5
volume is introduced during the decline, slurry injec-
tion continues after the onset of the screenout and
1.0
should be included in the fundamental relations pre- 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
sented in Section 9-3. Efficiency at screenout, ηso
Applying these modifications to the material-bal-
ance relations, it can be shown (see Appendix Eq. 51) Figure 9-21. Log-log slope after a TSO for ∆tDso = 2.
for the commonly used polymer fracturing fluids that
the log-log slope of the net pressure plot after a screen-
out and in the absence of spurt as β approaches 1 is
9-4.8. Fracture diagnostics from log-log
π
− sin −1 (1 + ∆t Dso )
−1/ 2
plot slopes
2
π
t dpnet 2 − sin (1 + ∆t Dso ) − ∆t Dso (1 + ∆t Dso )
−1 −1/ 2 The previous sections provide conceptual analyses
= of the different types of log-log slopes of net pressure
pnet dt ηso → 0 versus time. The slopes are characteristic of various
1.0 ηso → 1 , types of fracture geometries and modes of propaga-
tion. Therefore, the log-log plot, its associated slopes
(9-41)
1
⁄4 for high efficiency.
As the fracture pressure increases, it can reach the
1
pressure capacity of the formation. This leads to a
1 regulator effect, resulting in nearly constant pressure
because of accelerated fluid loss primarily near the
wellbore (see Section 9-4.5). A nearly constant pres-
100
sure measurement that is equal to the overburden
1 10 100 stress indicates a T-shaped fracture. Controlled frac-
Time (min) ture height growth into a barrier is characterized by
a gradually decreasing log-log net pressure slope and
(b) a constant pressure derivative. Net pressures steadily
Field Data
1000 decrease if uncontrolled fracture height growth
beyond a pinch point occurs. Fissure-dominated
fluid-loss behavior regulates the pressure to a con-
Net pressure, pnet (psi)
4100
End of injection
Fluid injection 4000 Proppant
at perforations
Casing 3800
3700
3600
Proppant injection
3500
3400
3300
0 50 100 150 200
Treatment time (min)
4100
4000
End of injection
Bottomhole pressure (psi)
3900
15° reorientation
Fracture 3800
Wellbore plane
at infinity 3700
3600 5° reorientation
3500
Fracture
reorientation 3400 Aligned fracture
3300
0 10 20 30 40 50
Treatment time (min)
Restriction area
Wellbore
Fracture Fracture
Channel to
fracture wings
Perforation
A dual-curve-fit algorithm can be used to determine the best- result, several proppant slugs were planned for as early
fit values of kpf and knear wellbore, thus defining the respective as possible during the pad period of the propped treatment.
values of ∆ppf and ∆pnear wellbore for any given rate (Fig. 9E-2). Before the slugs reached the perforations, the tortuosity
continued to increase, which limited the injection rate so
Limitations using surface pressure measurements that the surface pressure could be maintained below the
Rate step-down analysis using the surface treating pressure acceptable value of 6000 psi. However, when the proppant
can be difficult if the tubular friction is large compared with slugs arrived at the perforations, the tortuosity was signifi-
∆pentry because the wellbore friction can vary unpredictably cantly reduced, which enabled increasing the injection rate.
from published or expected values. The friction of water can The propped fracture treatment was successfully placed
be significantly reduced by small amounts of gel contamination with 6-ppg maximum proppant loading.
in the wellbore fluids or from a gel hydration unit. Fluid friction • Poor perforation effectiveness and excessive pad volume
with crosslinked gel may vary with small variations in the fluid
composition. Foam friction behavior is extremely unpredictable The rate step-down test in Fig. 9E-4 was performed after
and variable, and the analysis is further complicated by the first KCl injection. The test clearly shows the dominance
changes in hydrostatic pressure. With turbulent flow (i.e., of ∆ppf (i.e., near-wellbore pressure losses that relate to the
essentially all water injections), the friction is functionally closer injection rate to the 1.94 exponent). The estimated value of
to ∆ppf; with laminar flow (i.e., low injection rates with viscosi- ∆ppf was about 4500 psi at 18 bbl/min, the equivalent of
fied fluids), the friction is functionally closer to ∆pnear wellbore. only 4 of 60 holes open. This condition would not allow fluid
injection at the planned rate of 30 bbl/min. The near-well-
Field example
bore fracture tortuosity was low, at about 50 psi at 18 bbl/min.
The following example illustrates the usefulness of rate step- Additional KCl breakdown injections and surging did not
down test analysis resulting from the ability to understand and improve the low injectivity.
apply the information contained in fracture pressure behavior. The well was reperforated with larger holes, which
reduced ∆ppf to 1500 psi at 18 bbl/min. Following a
• Severe near-wellbore fracture tortuosity crosslinked gel calibration treatment (including a 20-bbl,
The rate step-down test following a second injection of 4-ppg proppant slug), the equivalent of 20 holes was open,
potassium chloride (KCl) water showed that the near-well- which provided an acceptable value of ∆ppf of 500 psi at
bore fracture tortuosity in a naturally fractured dolomite for- 30 bbl/min. The value of ∆pnear wellbore remained at less than
mation was extremely high, at 1900 psi (Fig. 9E-3). As a 50 psi, with C equal to approximately 0.6 for fluid injection.
∆p1
∆p1 Total entry friction
Fracture entry friction
∆p2
∆p2
∆p3 ∆p3
∆pnear wellbore ≅ qi1⁄2
∆qi1
∆qi2 ∆p4
ISIP ∆p4
∆qi3
∆qi4 ∆ppf ≅ qi2
∆p1
Fracture entry friction
∆p2
∆p2
∆qi1 ∆p3
∆qi2 ∆p3 ∆ppf ≅ qi2
∆p4
∆qi3 ISIP
∆p4
∆qi4 ∆pnear wellbore ≅ qi1⁄2
Figure 9E-2. The step-down test is conducted to measure ∆ppf and ∆pnear wellbore. ISIP = instantaneous shut-in pressure.
Conclusions
Rate step-down tests are simple to implement and can provide
key insights into the nature of the near-wellbore connection
between the wellbore and the far-field hydraulic fracture.
Although rate step-down tests have limitations, they can pro-
vide the rare combination of critical information at minimal
additional cost.
6000 20
1300-psi reduction Surface pressure limitation = 6000 psi
(first slug)
Injection rate (bbl/min)
3600 12
Step-down 2: Zero tortuosity at
300-psi tortuosity end of pumping
2400 8
Increased maximum
proppant loading
from 4 to 6 ppg
1200 4
0 0
0 30 60 90 120 150
Time (min)
Figure 9E-3. Propped treatment example with severe near-wellbore friction (i.e., fracture tortuosity) that was mitigated
by pumping two proppant slugs.
9-5. Analysis during fracture closure ties as those observed with the interpretation of con-
ventional well test data.
Fracture behavior during shut-in and prior to closure Basic decline analysis, outlined initially in this sec-
is governed by the fluid-loss characteristics and the tion, follows derivations presented by Nolte (1979,
material-balance relation (Fig. 9-2). A mathematical 1986b). A generalization of the technique using ana-
description of the pressure during the fracture closing lytical extensions to address nonideal conditions
period can be developed by also incorporating the (Nolte et al., 1993) is also presented.
fracture compliance relation (Eq. 9-21). These two
relations and that describing fluid loss are combined
to develop the specialized G-plot, which describes the 9-5.1. Fluid efficiency
pressure response during shut-in. Application of the
“Mathematical relations for fluid loss” in the
G-plot is analogous to the Horner analysis used for
Appendix to this chapter derives the fundamental
conventional well tests. The selection of an applica-
relations for fluid loss at the end of pumping VLp
ble slope for the G-plot also has the same uncertain-
(Appendix Eq. 23) and during the subsequent shut-in
(a)
12,000 50 6000 600
Injection rate (bbl/min) Perforation friction (psi)
2000 0 0 0
5.0 5.5 6.0 6.5 7.0 7.5 0 4 8 12 16 20
Time (min) Pumping rate (bbl/min)
(b)
50 10,000 100
Surface pressure limitation = 10,000 psi
Step-downs 1 and 2 show severe Proppant loading (ppg)
perforation friction (only 4 out of 60 Surface pressure (psi)
perforations open) and low tortuosity Slurry rate (bbl/min)
40 8000 80
Step-downs 3 and 4 show
lower perforation friction and
small tortuosity; No reaction
Proppant loading (ppg)
on proppant slug
20 4000 40
Cut pad percentage
from 50% to 25%
Reperforate
10 2000 20
0 0 0
0 60 120 180 240 300
Time (min)
Figure 9E-4. The step-down test (a) diagnosed high ∆ppf that was remedied by reperforating (b).
period VLs (Appendix Eq. 27). These equations are are not routinely used during field practice. Simple
in terms of the dimensionless volume-loss function analytical approximations, however, can be derived
g(∆tD) and its value at shut-in g0. In this context, ∆tD for certain values of α. These values of α are valid for
is referred to as the dimensionless time and is defined the commonly used crosslinked fluids that develop
in Appendix Eq. 15 as the ratio of the shut-in time ∆t a polymer filter cake along the fracture walls. These
to the injection (or pumping) time tp. fluids are the focus of the remainder of this section.
The general expressions for g(∆tD) and g0 in Corresponding relations for the non-wall-building
Appendix Eqs. 17 and 20, respectively, are based on fluids, such as linear gels or viscoelastic surfactant–
the assumption of a monotonically increasing fracture based fluids, are also discussed in “Mathematical rela-
area that is defined by a power law expression with tions for fluid loss” in the Appendix to this chapter.
an exponent α (Appendix Eq. 2). The generalized The value of the area exponent α can be explicitly
expressions are relatively complicated and therefore determined for two limiting cases of fracture growth.
g(∆tD)
and in this case the area increases approximately lin- 3
early with time. (1 + ∆tD)sin (1 + ∆tD)–1/2 + ∆tD1/2
–1
α = α 0 + η(α1 − α 0 ) = + η α1 − ,
1 1 Substituting Appendix Eqs. 22 and 26 into Eq. 9-46
(9-44)
2 2 produces the following expression for the efficiency:
where α1 is from Appendix Eq. 40. For typical effi- g( ∆tcD ) − g0
ciency values of 0.3 < η < 0.6, α is equal to approxi- η= , (9-47)
mately 0.6. g( ∆tcD ) + (κ − 1)g0
A much simplified expression for the fluid-loss where the spurt factor κ is defined in Appendix Eq. 24.
volume function g(∆tD) and its initial value g0 can κ denotes the ratio of fluid loss for a case with spurt
then be developed by using the bounding values of to that without spurt. For the propagation period, κ =
α (Eq. 9-43). These are given by Appendix Eqs. 29 1 for no spurt; more generally,
and 30, respectively, and graphically presented in
κ −1 spurt volume loss
Fig. 9-27. Throughout this development, the differ- = . (9-48)
ence between the upper and lower bounds of g(∆tD) is κ total leakoff volume loss
nominal, as illustrated on the figure. This observation
Fluid efficiency, η
are independent of the geometry model and enable 0.6
defining the efficiency for proppant scheduling (see
Sidebar 6L). 0.4
The fracture penetration is determined using the
following equation, which is obtained by combining 0.2
Eq. 9-6 and Appendix Eq. 22:
Af =
(1 − η)Vi , (9-49)
0
0 0.05 0.1 0.2 0.5 1.0 2.0 5.0 10.0
2 rp κCL t p g0 Dimensionless closure time, ∆tcD
where the fracture surface area Af for the three basic Figure 9-28. Relationship between efficiency and closure
models is from Eq. 9-29, rp is the ratio of permeable time (Nolte, 1986a).
(or fluid-loss) area to fracture area, and CL is the
fluid-loss coefficient. The average fracture width
is then obtained by using Eqs. 9-4 and 9-49: 9-5.2. Basic pressure decline analysis
2 ηrp κCL t p g0 The previous section presents relations for the fracture
V
w = fp = . (9-50) geometry parameters from simple considerations of
Af 1− η material balance and the assumption of power-law-
Finally, the maximum fracture width immediately based fracture area growth. The basic decline analysis
after shut-in for the three basic models is obtained as assumes that the end of injection marks the termina-
tion of additional fracture extension and that the
4 ( πβ s ) PKN change in the fracture volume during shut-in is attrib-
wmax,si = w 4 π KGD (9-51) uted entirely to the change in the average fracture
3 2 Radial, width during this period. The latter assumption also
implies that the fracture area is invariant throughout
where 〈w– 〉 is from Eq. 9-50 and β is from Eq. 9-20.
s the shut-in period. The fracture geometry models
The maximum width at the end of pumping is relate the fracture width to the net pressure through
obtained similarly: their compliance. Consequently, the combination of
the compliance and fracture geometry equations
β s β p PKN
enables determination of the fluid-leakoff coefficient
wmax,p = wmax,si ≈ 1 KGD (9-52) on the basis of the rate of pressure decline during
≈ 1
Radial. shut-in.
Using the material-balance relation during shut-in
The values for the PKN fracture geometry model from Eq. 9-7, it follows for a constant area that
are from Nolte (1979), whereas those for the KGD
and radial models assume an elliptical width profile dVf ( ∆t ) d w
− = − Af = qL . (9-53)
(Eqs. 9-11 and 9-13, respectively) and no change in d∆t d∆t
the fracture volume immediately before and after For an assumed constant fracture compliance that
shut-in. The value of wmax,p for the radial model from is ensured by a constant area, differentiation of
Eq. 9-52 is valid only for a line-source fluid entry Eq. 9-21 and substitution in Eq. 9-53, with the
condition (Fig. 9-6). The high entry velocity and expression for the fluid leakoff rate qL from Appendix
pressure gradient during pumping for the point- Eq. 11, results in
source case produce a nonelliptical width profile
dpnet 2 rp CL A f
(Geertsma and de Klerk, 1969). − Af c f = f ( ∆t D ) , (9-54)
d∆t tp
Slope, p*
• Fracturing fluid is incompressible.
pnet,si
• Formation closure pressure is constant.
pc All these assumptions are seldom met in routine
G(∆tcD) field practice. A departure from any of them produces
a G-plot with a continuous curve (i.e., not a straight
G(∆tD) line with a constant slope). In such cases, the applica-
tion of basic pressure decline analysis generally pre-
Figure 9-29. G-plot of the G-function response approxi- dicts optimistic estimates of the fluid-leakoff coeffi-
mating idealized fracture propagation conditions. cient and treatment efficiency. However, a rigorous
Length extension
Length recession
pnet
pnet
pnet,si pnet,si
Height recession
Length recession
pnet ≈ 3/4pnet,si
pnet ≈ 0.4∆σ
G(∆tD) G(∆tD)
Figure 9-31. Conceptual G-plot response for nonideal fracture behavior during shut-in for (a) fracture extension and (b)
height growth.
Complete height recession from the bounding Height growth into a higher stress barrier
zones occurs when the wellbore net pressure requires a previously increasing net pressure
equals about 0.4 times the stress difference ∆σ. response. It therefore is not consistent with the
Also from Fig. 9-7, the net pressure at shut-in is basic requirements of the radial and KGD mod-
about 0.8∆σ when significant height growth els, both of which exhibit decreasing net pres-
occurs. Therefore, complete height recession sures. Height growth should be anticipated only
from the bounding zones occurs when the well- for PKN-type behavior, with an increasing and
bore net pressure reduces to a value that is relatively large net pressure during injection. As
approximately one-half of that at shut-in. a result, for the PKN model with significant
Following fracture withdrawal into the primary height growth, the correction mGc is required to
reservoir zone, the pressure subsequently the slope of the G-plot following the termination
declines faster than the initially reduced rate and of height recession and prior to fracture closure.
as for a fracture geometry where no height The corrected slope mG′ that accounts for length
growth occurs. Thus, fracture height growth into recession during this latter phase can be inferred
the bounding zones changes the otherwise from numerical simulations and the material-
straight-line G-plot into a convex-upward curve balance relation during shut-in (Nolte, 1991):
(Fig. 9-31b).
(1 + ∆t ) f (∆t ) ,
1
From this discussion, it is clear that when mG′ ≈ mGc (9-62)
βs
cD D cD
height growth occurs, the fracture is still reced-
ing from its bounding zones when without where fD(∆tD) is the dimensionless fluid-loss rate
height recession the fracture area is momentarily function and is given in Appendix Eq. 28. The
stationary (i.e., pnet/pnet,si = 3⁄4). Therefore, the correction in Eq. 9-62 for fracture height reces-
fluid-leakoff coefficient is underestimated by sion complements Eq. 9-61 for length change to
using the slope at the 3⁄4 point in Eq. 9-58. The account for PKN-type behavior during the shut-
equation also assumes constant compliance and in process.
hence is valid only after complete height reces-
sion into the primary reservoir zone. Conse- • Variable fluid-loss coefficient
quently, the decline analysis requires that the Basic pressure decline analysis assumes that the
G-function slope be evaluated after the period fluid-loss volume is defined by a constant leakoff
of height recession. coefficient. This assumption has been shown to
Cc correction factor, Kc
0.9
cient Cv) or reservoir properties (i.e., compressibil- η = 0.2
η = 0.3
– Reservoir-controlled leakoff
0.7
Reservoir-controlled leakoff commonly occurs η = 0.4
when reservoirs with highly viscous oil are frac- mǴ = γ Kcm3/4
0.6
tured. It can also occur in low-permeability
η = 0.6
reservoirs that exhibit a high water saturation.
In the case of reservoir-controlled leakoff, the 0.5
slope of the G-plot at the 3⁄4 point can be modi- 0 1 2 3
Dimensionless fluid-loss pressure, pDLs
fied to include the pressure dependency on fluid
leakoff (Nolte et al., 1993):
Figure 9-32. G-plot slope correction factor Kc for reservoir-
mG′ = γKc m3/ 4 , (9-63) controlled leakoff (Nolte et al., 1993).
Pressure (psi)
Fracture height recession Gdp/dG 6000
Pressure
1200
Height recession during shut-in from high-stress bounding lay- 5500
ers results in changes in the fracture compliance and total Fracture
800 closure
fracture surface area relative to the leakoff (i.e., permeable) 5000
area (Figs. 9-7 and 9-36). This behavior causes several obvi-
ous signatures on the pressure and derivative plots (Fig. 9F-1). 400 dp/dG 4500
The pressure versus G-function curve shows a distinct down- Fissure closure
ward bend as height recession progresses, as discussed in 0 4000
Section 9-5.3. This behavior results in an increasing magni- 0 0.5 1.0 1.5 2.0 2.5 3.0
tude of the dp/dG curve and the superposition Gdp/dG curve. G-function
Figure 9F-1 elaborates this diagnostic using shut-in pres-
sure measurements acquired during a water injection test in
a carbonate formation. Injection was confined to a relatively Figure 9F-2. Pressure derivative analysis for fissure-
small perforated interval surrounded by several hundred feet dominated leakoff.
of similar lithology. The continuously increasing values of
dp/dG and Gdp/dG indicate continuous height recession dur-
ing closure. The figure also suggests that complete fracture Pressure-dependent leakoff
closure has not occurred by the end of the shut-in period.
As discussed in Section 9-5.3, the value of dp/dG can contin-
uously decrease during the preclosure shut-in period. This
behavior results because of change in the fracture penetration
500 1500
during shut-in, and it may occur in spite of a constant value of
Bottomhole pressure (psi)
Pressure (psi)
( pISI − pc ) = pc + pnet ,si
3 3
1200 9700 p3/ 4 = pc +
Pressure
4 4
9300 3
= 8910 + (9012 − 8910) = 8987 psi. (9-74)
800 Gdp/dG
400
dp/dG
8900
4
8500
The G-function slope m3⁄4 at this 3⁄4 point is esti-
0
0 0.5 1.0 1.5 2.0 2.5 mated to be 136 psi in Fig. 9-33. For the PKN
G-function fracture geometry model, the slope prior to closure
mGc is inferred to be 123 psi. This slope is cor-
Figure 9F-3. Pressure derivative analysis for filtrate- or rected for the length recession following the termi-
reservoir-controlled leakoff.
nation of height growth using Eq. 9-62:
dependent leakoff. One ambiguity exists, however. The same 1
pattern can be generated by extension of a preexisting nat- mG′ = 123 1 + 0.58 × 0.93 = 194 .
ural fissure set oriented nearly parallel to the hydraulic frac- 0.74
ture or leakoff into a swarm of parallel fractures created by
the fracturing process.
9040
Physical length, L
Apparent length, xfa
Figure 9-35. Fluid-loss volume and rate distribution (a) in the absence of spurt and (b) for a spurt-dominated treatment.
after-closure response can then be characterized in 9-6.1. Why linear and radial flow
terms of an equivalent and spatially uniform fluid-loss after fracture closure?
flux over an apparent half-length xfa that is generally
less than L, the physical half-length of a propagating A reservoir is disturbed by fluid-loss invasion during
fracture. Consequently, the value of dimensionless fracture propagation and closing. The fluid-loss rate
time T that is expressed in terms of L for after-closure changes with time as well as over the fracture length
analysis can be different from that of tD, the standard (i.e., has temporal and spatial components). This con-
dimensionless time (see Chapter 12) based on the cept is further explained on Fig. 9-36, which illus-
apparent stationary fracture half-length xfa inferred trates the change in fracture length during a calibra-
from the reservoir response. This distinction between tion test. At a time t = 0, the fracture length is zero.
T and tD is consistent with the notation for the fracture The fracture length increases over the injection time
lengths L and xf employed in this volume. tp, and it finally reaches a maximum value shortly
The after-closure period provides information that after the end of injection. Fluid leakoff at a hypotheti-
is traditionally determined by a standard well test cal location in the reservoir (labeled point a) is initi-
(i.e., transmissibility and reservoir pressure). It com- ated at the time ta when the fracture arrives at this
pletes a chain of fracture pressure analysis that pro-
vides a continuum of increasing data for developing
a unique characterization of the fracturing process.
At optimum conditions, it objectively determines L
parameters that either cannot be otherwise obtained
(e.g., spurt) or exhibit considerable uncertainty when
estimated with conventional pressure decline analyses
Distance
(a) (b)
Fracture Fracture
1 0.5re re 1 0.5re re
0 0
0.5re 0.5re
re re
Figure 9-37. Pressure distribution in the reservoir shortly after closure for (a) small and (b) large values of dimensionless
time. re = reservoir radius.
difference
response:
and pressure derivative
• Fluid injection tests characterized by a low effi- T = 0.005 T = 0.05 T = 0.5 T = 5.0
0.7
ciency can be devoid of linear flow in spite of the Linear Extended
creation of a fracture. 0.6
linear
As shown on Fig. 9-38b, linear flow appears ini- also emphasizes the “knee” formed by the intersec-
tially and lasts until T exceeds 0.005 and exhibits the tion of the slopes for the linear and radial flow peri-
anticipated half-slope for the pressure difference and ods. This intersection defines a unique dimensionless
derivative. Reservoir radial flow, characterized by a time that can be used to infer the fracture length
unit log-log slope, is the terminal flow regime that achieved during propagation, as subsequently dis-
occurs only for T > 5. The intermediate transitional cussed in Section 9-6.7.
period lasts over a time period spanning a factor of Figure 9-38 also illustrates a significant shortcom-
approximately 1000 times beyond the end of linear ing for extracting the reservoir transmissibility from
flow. The plot of the pressure difference in Fig. 9-38a the after-closure response in low-permeability reser-
Knee time
equal to approximately 3tc.
Normalized pressure
0.1
fied from the diagnostic log-log plot (Fig. 9-41)
2×
for estimating either the closure time or spurt.
1 Obtaining representative values of these param-
2
eters generally requires a good prior estimate of
0.01 the initial, undisturbed reservoir pressure.
In the absence of spurt, the closure time can
be estimated as the smallest value of tc that, when
used with the independently derived reservoir
1
1
pressure estimate, provides the linear flow diag-
0.001
1 10 100 nostic discussed in the preceding “Flow regime
1/ {FL(t/tc)}2
identification” step. If spurt is expected, the value
of tc should be independently determined using the
Figure 9-41. Identification of linear and radial flow using procedures outlined in “Estimating closure pres-
the diagnostic log-log plot. sure” in the Appendix to this chapter. The after-
closure linear flow analysis can then be used to
estimate the spurt, with the linear flow period ini-
factor of 2 and each has a half-slope. Radial flow
tially identified from the diagnostic log-log plot.
is characterized by a log-log slope that approaches
If linear flow is identified, a Cartesian plot of the
unity as well as an approximate overlying of both
pressure versus FL(t/tc) is constructed to determine
the pressure difference and pressure derivative
the slope mlf. Equation 9-94 is used to obtain the
curves.
spurt factor κ. The reservoir mobility k/µ required
• Reservoir parameters determination by this equation is derived from transitional or
The testing sequence should preferentially include radial flow analysis and CL from the decline analy-
a mini-falloff test to characterize the reservoir sis. The spurt coefficient Sp can then be obtained
parameters. Shut-in pressures monitored during the from Appendix Eq. 24.
mini-falloff test should initially be investigated for • Crossvalidation
radial flow using the log-log diagnostic described
Equation 9-101 is applied to compare the fracture
in the preceding step. If radial flow can be identi-
length predicted by the reservoir response to that
fied, a Cartesian plot of the pressure response dur-
derived from decline analysis (see Section 9-5) for
ing radial flow versus {FL(t/tc)}2 is constructed.
the calibration test. The radial flow slope mrf is
The slope of the straight-line portion on this graph,
anticipated from Eq. 9-93, with the reservoir trans-
or mrf, can be used with Eq. 9-93 to estimate the
missibility as previously determined from the
reservoir transmissibility. The pressure intercept of
mini-falloff test and Vi and tc defined as the vol-
the straight-line portion is an estimate of the initial
ume injected and closure time for the calibration
reservoir pressure. This procedure is analogous to
test, respectively. Agreement between the two
the conventional Horner analysis used for well
independently inferred fracture lengths indicates
testing.
correct evaluation of the calibration treatment.
In the absence of radial flow, the transitional
flow-based analysis described in “Comparison
of fixed-length and propagating fractures” in the 9-6.9. Field example
Appendix to this chapter should be applied to the
mini-falloff test. This type-curve matching proce- Application of the after-closure analysis methodology
dure estimates the initial reservoir pressure and its is illustrated by analyzing the pressure monitored
transmissibility based on the transitional flow during the field calibration tests in Fig. 9-10. Analysis
period. of the injection pressure for the calibration test
π2 1.02
faL = = 0.86.
100
2×
(1 − 0.16) π 2 × (1.02 − 1) + 2 5.6 4.6
1
Pressure/injection rate
Step rate/
treatment that exhibits PKN behavior, before penetra- Mini-falloff flowback/rebound Calibration
tion of the bounding layers occurs. Alternatively, the
value of E could conform to core tests or sonic logs
of the reservoir (see Chapter 4), and the primary zone
thickness could be calibrated with the pressure match
and verified against well logs. Similarly, variation in
the zone stresses must be based on their trends as pre-
dicted by stress logs. In the absence of stress logs, Time
lithology logs can be used to develop an approxima-
tion for particular geologic environments (Smith et
Figure 9-45. Formation calibration testing sequence.
al., 1989). Furthermore, the simulated fracture
dimensions could be compared against independent
measurements, such as radioactive tracers or micro- process based on these tests.
seismic measurements, to validate the outcome of the 1. Review existing information pertinent to fracturing
pressure-matching exercise (Gulrajani et al., 1998). A properly calibrated sonic log (see Chapter 4)
The inversion solution is confirmed by applying the provides an excellent basis for assessing the for-
numerical simulator based on these calibrated fracture mation stress profile. It can also be used to identify
parameters to predict the fracture behavior during the lithologic aggregates for developing an under-
propped treatment or for offset wells. An objective standing of the zone layers. In its absence, a stan-
pressure evaluation is indicated by agreement between dard gamma ray or spontaneous potential log
the simulated pressure response and actual field mea- should be applied to define the depth and thickness
surements. Conformity also establishes the ability of of the layers. The permeable zone thickness (i.e.,
the calibrated numerical simulator to develop the thickness over which fluid loss occurs) can be
improved designs for future fracture treatments. identified using specialized magnetic resonance
logs that directly measure permeability. This infor-
mation can also be obtained from the separation
9-8. Comprehensive calibration of various resistivity measurements, which indi-
test sequence cates drilling fluid invasion and therefore a finite
permeability.
An optimized hydraulic fracturing program integrates
An estimate of the formation stresses can be
the analyses and diagnostic methods discussed in this
derived by history matching prior fracture treat-
chapter with the proper design process (see Chapters
ments in offset wells. This information can be used
5 and 10) and execution aspects (see Chapter 11) of
to characterize the fluid leakoff behavior as well
the stimulation procedure. These considerations
as calibrate either sonic or gamma ray logs (Smith
include identifying candidate wells that will benefit
et al., 1989) to obtain the stress variation over the
from the stimulation treatment, determining the
depth of interest. Where available, laboratory core
appropriate injection rate and proppant addition
tests should be used to obtain related information
sequence as well as assessing the risks involved with
(e.g., mechanical properties, existence of natural
treatment execution. The techniques described in the
fissures, fluid-loss behavior).
previous sections provide a basis for estimating the
fracture and reservoir parameters required during the 2. Determine reservoir production parameters
optimization process. To ensure completeness and The reservoir pressure and permeability are best
consistency in the evaluation process, fracturing pres- determined from a well test. In its absence, the
sure data should be obtained through a planned information can be obtained using the mini-falloff
sequence of calibration tests, as shown in Fig. 9-45. test. The mini-falloff test attempts to propagate a
The following steps describe the fracture evaluation short, inefficient fracture to attain the required