Professional Documents
Culture Documents
Summary. This paper presents the basis for determining the parameters that quantify the creation of a
hydraulic fracture from the fracturing pressure decline after a fracture treatment. The analyses are generally
applicable to the standard models used in fracture design-i.e.; Perkins-Kern (PK), Khristianovitch-Geertsmade Klerk (KGD), and the penny or radial models. The parameters that can be determined, as shown in an
example, include the fracture's dimensions, fluid-loss coefficients, fluid efficiency, and fracture closure time.
Introduction
Hydraulic fracturing, like other drilling, completion, and
reservoir behavior, is complicated because the processes
cannot be observed directly. This deficiency has been
overcome for description of reservoir processes by the
development, over the past 50 years, of analyses based
on the wellbore pressure and flow rate. During the past
5 years, similar analyses for fracturing have been
introduced 1.2 and successfully applied. 3-7 Recently, field
application has been significantly facilitated by the introduction of computerized vans specially equipped to
record and to analyze fracturing data. 7-9 The analysis of
fracturing pressure during injection 2 provides qualitative
information on the nature of the fracture's growth-i.e.,
height confinement or growth, excessive fluid loss, or restricted extension-and quantitative information by comparing recorded pressure data with pressures simulated
by a computer model. 6 Analysis of the pressure decline
after fracturing provides information on the fluid-loss
characteristics, fracture dimensions, fluid efficiency, and
fracture closure time. 1 This analysis was initially limited to the PK model for fracture geometry. 10 However,
this analysis can be extended to the other standard
models-the KGD model ll - 13 and penny or radial
model. 10,12 Schematics of these models are shown in Fig.
1. Recently the analysis'has been appliedto'a fractUre with
an elliptical area. 14 The derivations of the more general
equations, considerations for their applications, and an
example are given in the following sections.
Assumptions
The assumptions made and the effect on specific equations are summarized below.
1. A continuous rate of fracture growth (Eq. 1) that
would require a constant injection rate, no accelerated
growth as a result of penetration of a zone having significantly lower closure pressure, and no retarded growth
caused by proppant bridging are assumed.
2. A continuous change of the ratio of fracture to fluidloss area would not be'violated by the subsequent penetration of an impermeable and higher-stress zone, generally
the case for shale, for which both width and fluid loss
would be relatively negligible; conversely, it would be
violated for the subsequent penetration of either a lowerstress or permeable zone.
3. Spurt loss, the opening and closing of natural fractures, and pressure-dependent loss are not explicitly considered. However, Eqs . 34 and 35 give corrections for
these effects.
4. The assumption of constant fracture area from shutin of injection to fracture closure would be strictly violated by either additional extension or partial closure. The
practical consequences of both are unknown but may be
unimportant. The amount of additional extension is likely to be relatively small, and if the closure results in mismatches of the fracture faces or propping caused by rock
fragments or proppant, the closed section would still be
in communication with the open section and would contribute to fluid loss.
5. The assumption that the fracture closes without interference of proppant requires either a calibration treatment without proppant or, for an actual stimulation with
proppant,tfie use orEg. 2Zfor fracture cfosure time and
pressure matching (Eq. 27) before the fracture closes on
the proppant. Closure on the proppant is generally indicated by an accelerated rate of pressure decline. 6
6. For relationships based on pressure (p* or p s given
in Eqs. 26 through 31), closure-pressure changes during
shut-in must be assumed negligible compared with p s or
p* if the well bore pressure is used directly. If the change
in closure pressure is significant and can be predicted or
estimated, these equations may be applied with appropriate corrections.
Derivations
Area/Fluid-Loss Relation. Proceeding in a similar but
more general manner than for the previous derivations 1
for pressure decline, the fracture's rate of areal growth
can be bounded 1,15 by either the assumption of negligi571
Well
Perkins-Kern
(Ref. 10)
~I------L--------~I
------~~------
T
hf
Khristianovic.
Geerstma-de Klerk
(Refs. 11. 12)
Well
~Profiles
~ .
Radial or Penny
(Refs. 10. 11)
I
Fig. 1-Schematlc of fracture models.
where C is the conductance or fluid-loss coefficient. Integrating Eq. 4 after substituting Eq . 2 gives
a=aIA=TII
and
a=aIA=(Tlt)'h,
.... . .. .. . . .. : .. .. .. . . . . . (lb)
or
2CA
ra
vI
0 vl-a e
----r- J
da
2C:
[2.(1 ~~)]
vI
sm 1 a
. ............. (5)
where e= 1 for the upper bound and e=2 for the lower
bound, A and 1 are the current area and time. and a and
T are area and time at some other time (see. Fig. 2). The
upper bound assumes not only negligible fluid loss but
also constant fracture width. Because the width generally increases and stores a portion of the injected fluid , the
upper-bound exponent in Eq. 1 is actually less than
unity-i.e., lie < 1. The actual exponent for this bound
with the PK model was derived in Ref. I. and can be found
by similar methods for the KGD and penny or radial
models. The values for the PK, KGD , and radial models.
respectively, are
_ 2CA k
v 10
2Jtito
sin - I a
(l-~) ]
'
. . . .. . (6)
where the upper expression in the brackets is for the upper bound (e= 1) and the lower expression in the brackets is for the lower bound (e=2) . By substitution of
a=Ako the final fluid loss or porous area, and T=lo into
Eqs: 5 and 1. Eq. 6 follows. The subscript 0 denotes the
end of pumping (beginning of shut-in) .
The total area for one face of the fracture, A f. can be
exp,ressed by Akand the following expre.ssion can be substituted into the above relations .
.
and
and
2COO
q,(I,OO)= Jt-T(a) ,
572
. .. . . . .. .. . ........... (4)
Eq. l'
2 0.05
... -
0.1
0.5
0.2
1.0
"/2
~~
It
Or-------------------------~
Bounds,
cu
-~
-o
- - Lowe<
4
:>
..~:
..
-~,.
...-/
,,/2. _ _4/3 ..
I~~~~--~--~~~~~~--~--~I
log TIt
o 0.05
0.2
0.1
0.5
1.0
2.0
5.0
10.
where K(K ~ 1) is a multiple to C that accounts for additional fluid loss only during pumping (e.g . spurt loss or
opening of natural fractures only during injection). These
effects, along with the definition of K, will be discussed
later. Eq. 10 can be expressed as
q,(t,A k )=
r
v 10
sin -I ...rr;;it
1f'>
2CfpAf
JIo
and
ID=(t-tO)lto
v,
KCfpAfJlo
10%).
[4/3[(1
+tb)3I2 -I D312]
1
(l + I D )sin - (1
I/,
"J'
I L '
+I D) - +t D "
(12a)
results in
where the upper and lower bounds for g(ID) are given
in the brackets of Eq. 12a. The lower-bound integration
was given in Ref, 5. Fig. 4 shows these bounds, and as
for !(ID)', the bounds are very close.
573
or
go =g(O) = 4/3 ]
[ 1(12
..................... (13b)
!x=VflV"
Vj=2C!pA f../'to (g e ~ go),
............................ (17a)
and
y.t =2KC!pAf../'to go ,
..................... (17b)
and
or
and
ef=VlV j ; l- ef=V t IV j ,
574
1.0
5.
-- - ., - 1
::D
e,-1-110~
p-~1110 -1
",-
,;.
...
1.5
",-1 Upper
e,-O Lower
~
i4: 04
<
0
.6 .
0.4 C
3
!!'
0.2
<:
i
1.0 0
0.8 r0
0.6 l!!
Bounds:
w
==
2.
2-
0.6
.!.!
.,- 0 ('--Bound)
(Upper Bound)
aci"
0.8
0.2
-b
1.
.5
p.
1e
0.5
0.2
0.5
1.0
2.0
5.0
0
10.0
02
.
110
0.1
/
0.05 .
0 05
0.1
to
0.2
0.5
1.
2.
DimenSionless Tme. I D
F~g. 6-Dlmenslonless difference functlon--master for
curve ' matching.
where!vp is the volume fraction of proppant pumped (including theproppant porosity, 4 relative to the total slurry
pumped, W is the proppant weight. and w is the srcific .
weight of the proppant material-e.g . 165 lbm/ft ,2.65
g/cm 3 [2643 kg/m 3 , 0.957 Ibm/in. 3 ] for sand.
for the average width at shut-in with!x from Eq. 18a for
no proppant or Eq. 22a with proppant.
Dividing Eq. 20 by Eq. 21 yields
Vpr
(g~ -go) .
1--=
.
VI
I(golx
Fracture Area and Penetration. Combining the continuity equation, Vi = V,+ VI' and Eqs. 13a and 16a gives the
following for the fracture area.
Vi =:: v;+ VP= Vi(r+!~)
or
I-fyp
or for
.
Ix =
1(=
I,
g~/go
1,-!vp
-I
7fr/
. . ...
(24b)
in which
!vp = VprlV j
=[~~f],
=W/[wV (I-4],
j
and
and
. . . . . . . . . . . . .. . . . . (25b)
for the three models, where b is the average width, cf
is the fracture's compliance for width, E' is the planestrain elastic modulus of the formation, and {3 is the ratio
of the average net pressure in the fracture to the wellbore
value. The term "average" implies the value of p" that,
if distributed uniformly over the fracture area, would
produce the same average width as the actual nonuniform
pressure. The expressions for cf are found from Refs.
10 and 12. Values of {3 appropriate for the different models
are discussed later.
Fluid-Loss Coefficient. The continuity equation implies
that after shut-in and before the fracture closes, the rate
of fluid-loss volume is equal to the rate of change in the
fracture's volume:
dVf
db
- - = -Af-=qr(t,Ad .
dt
dt
dp"
-AJCf-=
dl
2CfpAf
or
AI'
'Kps
V,
4ICgop
1,,=--=
~[311'/16]
tep
112
'II 10
or
dp"
---=
dt
576
2Clp
Cf'llto
Q.
800
::J
II)
II)
Qj
ct
IV
'5
en
600
1000 .....- - - - - - - - - - - - . 4 6 0 0
CO)
::J
4400
800
- 7 5 0 - -....... 4350
100
10
150
12
. . . , . ....... . . . .. . (30)
CI)
'0
200
14
4200
16
6008~--~---~----L~~
en
1
Shut-In. Min.
for the three models, or substituting the last expression
in Eqs. 16 and 29,
Q.
ci
.s=gE
(5
Application
Closure Pressure. There are two considerations for
closure pressure: the definition of Ix in terms of the time
for ~e fracture to reach closure pressure (time to close
in Eq. 18) and the definition of p. in terms of the decline
in the wellbore pressure (directly measurable) if the
closure pressure is constant instead of the more general
case of decline in net pressure (Eqs. 25 and 27). For liquid reservoirs, closure pressure can change significantlys
during and after a treatment or for subsequent injections.
I
577
iii
1.5
1.3
II)
;,
i.e&: 1.0
j ~ 0.5
C>
II
II>
':.
'"
.r;
..
-"II:
Well
112
Tip
.!! 1.0
!!
:t:
...
Q)
-...
:)
0.5
( .)
1'0
~"=-
u.: .
Ii:
1:
en
'4)
1~
0.1
______
__
~~~~~
__
~~
0.2
Net Pressure, Pn11w
Generally, however, closure pressure changes are insignificant for gas reservoirs because of their relatively high
compressibility that retards the reservoir pressure transient from significantly penetrating the reservoir. For this
case, closure pressure can be detennined before the fracture treatment and used for subsequent analysis of closure
time and net shut-in pressure, P s .
Closure pressure can generally be found from the pumpin/flowback procedure. 5 .6 This procedure consists of
pumping a small quantity of fluid (e.g., 20 bbl [31. 7 m 3 )
of water at 5 bbl/min [0.79 m 3/min)) and immediately
flowing back at a constant rate of about or less than onequarter the injection rate. Closure pressure is exhibited
by a change in curvature of the pressure plot from positive to negative curvature (Fig. 8). For the decline analyses, this procedure would be most applicable before the
treatment, when closure pressure changes are negligible.
This same technique can be used for liquid reservoirs,
but the value found before the treatment generally will
not be valid to infer the net shut-in pressure, P s' or to
determine when the . fracture closes. Generally, for liquid (and gas) reservoirs, the fracture's closure can be
detennined from a plot of shut-in pressure vs. the square
root of time. 6 Closure pressure is interpreted as the pressure where the slope changes on the plot (Fig. 8).
When closure pressure is expected or is found to change ';
significantly,5 the above equations that contain pressure
(i.e ., p. or PSt Eqs . 26 through 31) cannot be used unless the change in closure pressure can be predicted 5,19
ties of the fracture. This flow generates a pressure gradient until the fluid is completely lost to the formation
or trapped in the proppant when the fracture closes. It
has been implicitly and erroneously assumed that after
shut-in the fluid flow ceases, the fluid friction is zero,
and the pressure is equal to the closure pressure. Fig. 9
shows the simulated pressure and flow rate in a fracture
both before and directly after shut-in. A computer
algorithm 6 based on the PK model and including the continuity equation 15 was used for the simulation. The figure
shows that (3 before shut-in (f3p) is about 0.67 and directly
after shut-in (3$) is about 0.77. The difference is caused
by the reduction in both the flow rate and resulting pressure gradient in the entrance area of the fracture. This
reduction in pressure at shut-in is generally and erroneously considered part of the pressure drop through the
perforations. Because the average width (Le., volume)
of the fracture is essentially constant directly before and
after shut-in, f3 p /(3s is equal to the ratio of net pressure
directly after and before shut-in. The fluid flow is approximately linear from the wellbore to the tip before shutin, and after shut-in it approaches a uniform value over
the central part of the fracture, tapering to zero at the welJbore and tip.
Expressions for (3p and (3 s based on the PK model were
previously .derived 1 on the assumption that the fluid's apparent viscosity and flow rate varied as a power of distance from the tip to the wellbore-i.e ., exponents equal
o and b, respectively. A value of 0=0 would represent
uniform viscosity, whereas 0 = 1 would represent linearly decreasing viscosity from the wellbore to the tip. The
exponent for flow rate, b, was assumed to be unity (linear
variation) while pumping and zero (constant) for shut-in,
and gavel
.
.
and
f3 s =(2n+2)/(2n+3+0), .................. (32b)
13=[6~~+2)/(2n+3+a)J
.................. (33)
3?l'2/32
Cc<C(I+
Ps
Pc-PR
)e, ...................
(34)
1600~------~------~------~-------,
'0;
0-
"
II)
II)
l
...
't:
en
"
QI
1000
800
5MPa
600~
____
~~
50
______
______
100
Shut-in lime. Min,
______
150
200
proximately the perforated interval with limited penetration,14 or it assumes that slip 20 occurs between the
formations with no transfer of shear stresses. As a result,
height growth for this model would not occur. Vertical
height change or growth, however, is possible with the
PK model. This model (see Fig. 1) assumes that the width
is zero at the top and bottom of the fracture because the
elastic solution on which the model is based considers the
shear stresses between vertical sections. The tendency for
limitC'd height growth is caused by higher stresses in the
bounoary formations. 21 For this case, the width is significantly narrower 6 in the boundary formations that are
penetrated by height growth. As a result, the appropriate
fracture height, hf' to use 6 in the pressure and dimension relationships is the height of the lower-stress
formation-generally the gross reservoir section-without
including any additional growth of height. Fig. 10 shows
the change in the assumed constant compliance, Cf, of
Eq. 25 for an idealized case 6 of height limitation because
of stress differences in the formations. The figure shows
that Cf changes by less than 10 % for net pressures, P n ,
less than 0.4 Au (Au=stress difference in formations) or
equivalently for height growth of less than 20% of the
initial height. The compliance andheightarenormalized
in Fig. 10 by their initial values at essentially zero-i.e.,
0+ -net pressure. Also the figure shows that the net
pressure is limited to about 0.9 Au: max Pn or
Ps =0.9Au. For cases in which height growth occurs (if
the pressure-decline analysis is performed after the net
pressure has declined to about one-half of Ps), cf will remain essentially constant until closure and the analysis
will be valid. This restriction is not very limiting because
height growth does not affect the time-ta-close expressions (equations before Eq. 25) and the pressure match,
Eq. 27, can be performed during the latter portion of
closure instead of the initial portion. Height growth would
appear to violate the assumption of constant!p =hplhf,
Eq. 7. This is not the case if the bounding beds are impermeable (generally shales, with no fluid loss) and have
higher stress (generally the case for shales 6 ). For this
case, both the volume of fluid lost and that stored in the
579
narrow width of the boundary formations would be relatively insignificant, and the effective fracture height would
not include the height in these formations.
Fluid Loss From Spurt and Opening Natural Fractures. In the derivation section, a distinction was made
between the fluid-loss conditions during and after pumping, KC and C, respectively. This distinction was made
by the variable K multiplying the loss coefficient for expressions based on fluid loss while pumping. Two primary
effects that can cause increased loss during pumping are
spurt loss and the opening of natural fractures at a specific value of fluid pressure. 2 If the decline analysis for
p. is performed below this opening/closing value of pressure, the inferred conditions for fluid loss would not
reflect the increase in fluid-loss area, denoted as Aoj,
when the fractures were open. The actual loss for this case
should be based on the area Ale +Aof (Aoj for only one
side of the fracture face).
Fluid loss to spurt occurs when a portion of the formation is first fractured and is generally assumed to occur
essentially instantaneously. Therefore, spurt would occur
only during pumping and not after shut-in when the fracture penetration is assumed to stop. The loss caused by
spurt is not included in the time-dependent loss defined
by the loss coefficient. Spurt is defined by volume lost
per unit area and has the dimension of width-i.e., spurt
width, which will be denoted as b s ' The amount of loss
volume caused by spurt during a treatment is 2 b sA Ie. and
A/c=/pA j . From inspection ofEq. 13 for the loss volume
caused by time-dependent fluid loss, and including the effect of natural fractures opening while pumping, Aoj' it
follows that
2a
where
a=gdf _gdu'
b=gdu _2 g dt,
and
c=g1-1.
Eq. 37 was used for the interpolated curve relating efficiency to time to close in Fig. 5.
Example
In this section, expressions derived in the previous sections will be applied to a field case from the Wattenberg
tight-gas field in the Denver basin. The formation is generally about 8,200 ft [2500 m] deep and has a permeability
of less than 5 x 10 -3 md and a temperature of 265F
[130C]. The data were collected from a calibration treatment, without proppant, to determine the fluid-loss coefficient before a massive hydraulic fracture treatment on the
well. The calibration treatment consisted of 500 bbl [80
m 3 ] at a rate of 5 bbllmin [0.79 m 3 /min], with a pump
time, 10, of 100 minutes. The fluid was an emulsion consisting of two-thirds condensate and one-third water-based
poIymer~ Forthis fluid and formation, spurt loss is negligi- "
ble. Also, the pressures during pumping were less than
the natural-fracturing opening pressure of the formation,
interpreted 6 to be Pn = 1,700 psi [11.7 MPa]. Thus, there
is no correction for these effects and K (Eq. 35) is unity .
The pressure-decline curve is shown in Fig. 11 with
a surface shut-in pressure of about 1,550 psi [10.7 MPa].
Because of the formation's low permeability and the relatively high compressibility of gas, significant changes in
closure pressure would not be expected. This was the case,
as Fig. 8 shows that both the pump-in/flow-back test prior to the calibration treatment and a decline pressure vs.
square root of shut-in time after the treatment give essentially the same surface-reference value of 750 psi [5.2
MPa] for the closure pressure. Based on this value of
closure pressure, the time for the fracture to close, Ie,
was 150 minutes after shut-in 'see Fig. 8 or 11) and the
ne~ shut-in pressure, PSI was 800 psi [5.5 MPa].
SPE Formation Evaluation. December 1986
SOOO.2 .
0.5
1.0
D ol
6I)(t t
2.0
AA
-p (t01 - p (tOI
~'l-A
a.
~.
100
0.5
1.0
(3701
200
<I
..
0.2
~r-------~----~~~~~
~o
:..~O
.#
...Q
00
100
502O=~----~50~--~~----~~
100
200
Shut-in Tme. Min.,
50
20
50
Shut-in TIM. MIn.
Vi' ft3
E'. psi
h" ft
PI' psi
fp
K
p., psi
to. minutes
te. minutes
te/t o
2.800
4x 10 6
60
800
0.53
1.00
370
100
150
1.5
Conclusions
Theoretical relationships based on the pressure decline after fracturing were derived for the inference of fracture
penetration, average width, fracture closure time, loss
coefficient, and fluid efficiency. These relationships were
found for the three commonly applied fracture modelsthe PK, KGD, and the radial or penny models. Fluid efficiency can ~ found without the assumption of a partic582
Nomenclature
a, A = general variables for area, ft2 [m 2]
gc
g~
gd
gt
g.,
go
G(I D ,I
= g(lc/lo)
= gc
=
=
=
=
with proppant
gc/go
hI
hie
k
L
n
P
Pc
Pfl
PR
= vertical fracture
height, ft [m]
permeability
fracture length; tip-to-tip, it [m]
power-law exponent for fluid
pressure., psi [kPa]
closure pressure, psi [kPa]
net pressure, Pw -Pc' psi [kPa]
= reservoir pressure, psi [kPa]
P s = net pressure at shut-in, psi [kPa]
Pw = pressure in wellbore, psi [kPa]
p. = match decline pressure, psi [kPa]
qt = fluid-loss rate, ft3/min [m 3 /min]
RI = fracture radius, ft [m]
s = spacing between polymer elements, ft
[m]
I = time since start of pumping, minutes
Ic = closure time, minutes
.
I D = dimensionless 'time, I/to-l
I D= reference value of I D
SPE Fonnation Evaluation. December 1986
tD'
t Dc
to
VI = fracture volume, end of pumping. ft3
[m 3 ]
Vi
ft3 [m 3]
Vt,si = loss volume after pumping ends, ft3
Vt
[m 3 ]
Vpr
ft3 [m 3]
Reference.
1. Nolte, K.G.: "Detennination of Fracture Parameters from
Fracturing Pressure Decline," paper SPE 8341 presented at the 1979
SPE Annual Technical Conference and Exhibition, Las Vegas, Sept.
23-26.
2. Nolle, K.G. and Smith, M.B.: "Interpretation of Fracturing
Pre~ures," JPT (Sept. 1981) 1767-75.
3. Veatch, R.W. and Crowell, R.F.: "Joint Research Operations
Programs Accelerate Massive Hydraulic Fracturing Technology,"
JPT(Dec. 1982) 2763-75.
4. Dobkins, T.A.: "Procedures, Results and Benefits of Detailed
Fracture Treatment Analysis," paper SPE 10130 presented at the
1981 SPE Annual Technical Conference and Exhibition, San
Antonio, Oct. 4-7.
5. Smith, M.B.: "Stimulation Design for Sliort, Precise Hydraulic
Fractures-MHF," SPEJ (June 1985) 371-79.
6. Nolte, K.G.: "Fracture Design Considerations Based on Pressure
Analysis," paper SPE 10911 presented at the 1982 SPE Cotton
Valley Symposium, Tyler, May 20.
7. Elbel, I.L. et al.: "Stimulation Study of Cottage Grove Formation," JPT (Iuly 1984) 1199-1205.
8. Cooper. G.D. et al.: "Improved Fracture Design Through the Use
of an On-Site Computer System." paper SPE 12063 presented at
the 1983 SPE Annual Technical Conference and Exhibition, San
Francisco, Sept. 5-8.
9. Hannah, R.R. et al.: "The Real Time Calculation of Accurate
Bortomhole Fracturing Pressures, " paper SPE 12062 presented at
the 1983 SPE Annual Technical Conference and Exhibition, San
Francisco, Sept. 5-8.
10. Perkins, T.K. and Kern, L.R.: "Widths of Hydraulic Fractures,"
JPT(Sept.I96I) 937-49; Trans., AlME, 222.
11. Khristianovitch, S.A. and Zheltov, Y .P.: "Formation of Vertical
Fractures by Means of Highly Viscous Liquid," Proc., Fourth
World Pet. Cong., Rome (19.55).
E -01
m3
E+OO = lcPa
SPEFE
Original manuscript (SPE 12941) received in the Society 01 Petroleum EngiMeIS office
Feb. 'n, 1984, Paper accepted lor publication May IS, 1986. R8viaed manuscript receiYed March 24, 1986.
583