Sei sulla pagina 1di 9

Interpretation of Fracturing Pressures

Kenneth G. Nolte, SPE, Amoco Production Co.


Michael B. Smith, SPE, Amoco Production Co.

Summary
This paper presents a basis for interpreting fracturetreating pressures that permits identification of
periods of confined-height extension, uncontrolled
height growth, and, more importantly, a critical
pressure. When a treatment reaches the critical
pressure, fracture extension is reduced significantly
and a pressure (or screenout) condition or undesired
fracture height can follow. Example applications for
data from five treatments are presented along with
potential explanations for, and implications of, the
critical pressure.

Introduction
This paper is limited to the discussion of
hydraulically created fractures that are in the vertical
plane. In addition, if these fractures are desired to be
deeply penetrating, they must have confined or
limited height growth. Therefore, the paper is limited
further to fractures having a horizontal penetration
appreciably larger than their vertical height.
There are two fundamentally different concepts
for the propagation of a constant-height vertical
fracture that lead to very different results. One
concept 1 is that the fracture width is constant across
the height of the fracture. This requires the
assumption that the formation bed being fractured is
independent of the beds above and below - i.e., the
beds can slip freely, independent of one another, at
their boundaries. This assumption leads to the
conclusion that the fluid pressure required to extend
the fracture decreases with time.
The other concept, presented by Perkins and
Kern,2 assumes that there is no, or negligible, slip of
boundaries along the horizontal planes that confine
the fracture height. This assumption leads to the
01492136/!XJ09.8297$OO.25
Copyright 1981 Society of Petroleum Engineers of AIME

SEPTEMBER 1981

conclusion that the fluid pressure required to extend


the fracture increases with time. This concept, as
refined by Nordgren,3 predicts that for a Newtonian
fluid creating a confined-height fracture at a constant
injection rate, the well bore pressure increases
proportionally to time raised to an exponent:
p(t) ex t e ,

118 <e< 115 . ................. (1)

The larger value is for the assumption of a relatively


small fluid loss rate, relative to the injection rate,
while the smaller value of the exponent is for a
relatively large fluid loss rate. For Eq. 1 and all other
cases in this paper, reference to pressure implies
pressure above the fracture closure pressure.
Fig. 1 shows the wellbore treating pressure, above
the in-situ closure pressure, vs. time for three massive
treatments. The data shown in this paper were
collected by a tubing/annulus wellbore configuration
having no packer. For this configuration the
treatment can be pumped down the tubing or annulus, with the surface pressure on the static line
giving the bottomhole pressure after a correction for
hydrostatic head. The closure pressures for the
formations were determined by a pump-in/flowback
procedure. Briefly, this procedure uses the same
tubing/annulus configuration and consists of injecting a volume of fluid at a sufficient rate to initiate
or open a fracture in the formation. After the injection, the well is backflowed at an appropriate
constant rate (e.g., through a surface choke) that
varies for different formations. In the desired range
of flowback rates (e.g., one-quarter of the injection
rate), a plot of pressure vs. flowback time will exhibit
a characteristic reversal of curvature (Le., increasing
rate of decline) when the fracture closes. The initial
portions of the data are not shown in Fig. 1 because
they contained periods of significant variations of
1767

._Lockup

!::I

4000

III

(j

.,>

2000

III

-1-+111 ..17
'
... ..........
~

., .

.Q

<

.,

1000

0.13
Slope: 0.25

...-:::

,(.

40

.j-II-f. '"
IV

1.0

Negative

2000

............, Case

II
Mode: I
*. Stert Proppent

20

III
":::" Cese 2

.. , . ;

III

Cese 1

., .......
.......... -....

A.

4000

60
200
100
Time. Minutes

1000

10400

well bore in terms of both the fluid flow and the


continuity equations. The wellqore pressure above
closure for a power-law fluid and a constant height
was given by Perkins and Kern 2 for constant flow
rate down the fracture (not physically possible due to
leakoff and storage) and constant viscosity
properties. This pressure relationship can be shown 4
to be similar for the more realistic case in which the
flow rate and viscosity increase from the fracture tip
to the well bore and can be expressed as the
proportionality
Kq7L )2n~2 .................. (4)
'
P ex ( h n C 2n + 1

where

Fig. 1 - Examples with the different characteristic slopes.

C= hlp

injection rate or fluid viscosity that are typical of


fracture treatments. The figures also show the time at
which proppant was introduced into the wellbore.
The data in Fig. 1 are shown on a log-log plot, for
which the slope of a straight line defines the exponent
e given in Eq. 1. Although the figure presents
pressure vs. time, a cumulative injected fluid volume
could be substituted for time with the same results
since a constant injection rate is assumed. The figure
shows that the slope of the initial portion of each
treatment falls within the Nordgren bounds given in
Eq. 1. Therefore, the positive slope portion of each
treatment can be interpreted as confined height
extension governed by the no-slip concept of Perkins
and Kern. The portion of the treatments that deviate
from this behavior are discussed in later sections.
The bounds on the time exponent in Eq. 1 are for a
Newtonian fluid and can be generalized 4 for a
power-law fluid, with a power-law exponent n, as
1

- - <e<-- ..................... (2)


4n+4
2n+3
Since n is generally between O.S and 1 (Newtonian
fluid n = 1) for actual fluids currently used for
fracturing, the bounds in Eq. 1 differ slightly for a
power-law fluid with n < L For n =O. 7S and O.S,
these bounds increase to

............................. (S)

and h is the height and L the length of the fracture. K


and qi are wellbore values of the power-law coefficient (proportional to apparent viscosity) and flow
rate, and C is the fracture compliance, which is
defined in this paper as the ratio of the average width
h to the pressure p. Eq. 4 is taken from Eq. 16,
Appendix A, of Ref. 4 in which the inverse of C was
defined as the stiffness. For a fracture in a
homogeneous, infinitely elastic formation, 2 the
compliance is
C= 7rhl (2'),

........................ (6)

with E' being the plane strain elastic modulus. As a


result, C is proportional to height for this idealized
case.
The continuity equation for a fracture 3 can be
expressed as

dV

q i = q + dt' ......................... (7)


which states that the flow rate q i into the fracture is
equal to the sum of the fluid-loss rate q f to the
formation and the rate of change of fracture volume
V with respect to time. The volume can be expressed
as the product of length and the average values of the
fracture width and height, using Eq. S as
V=LpCh, ............................ (8)

0.14<e<0.22, n=0.7S,
and

0.17<e<0.2S, n=O.S . . "., ... ,.".", .. (3)

Diagnostic Descriptions
In this section, potential causes are given for
deviations of the pressure data from the confinedheight extension mode, denoted as Mode I in Fig. 1.
The purpose of this section, and for the paper as a
whole, is not to present all possible explanations or to
provide definitive claims for any presented but to
demonstrate the potential utility of the procedure.
Two relationships are used: the pressure at the
1768

for which all variables denote average values over the


fracture length. Substituting Eq. 8 into Eq. 7 and
expressing the result with respect to incremental
changes during time tit yields
AL t:.p I1C
q;=qf+LpCh ( T+
+C+

11h)
1
h tit'

.(9)

For a constant injection rate q;, the right side of Eq.


9 is a constant and places a restriction on the permissible changes in the variables with respect to each
other. As a consequence, any positive increase in p,
C, or h results in a decrease in the potential rate of
length extension. In addition to Eq. 9, Eq. 4 also
places a restriction on the permissible changes in the
variables.
JOURNAL OF PETROLEUM TECHNOLOOY

In the following, these restrictions on the changes


in the variables are investigated as a tool to interpret
the pressure data in Fig. 1. This interpretation
assumes that the flow rate and fluid injected (i.e.,
viscosity characteristics) do not change significantly
during the treatment, as such changes could affect
pressure (Eq. 4). The discussion centers on the four
characteristic slopes of the plots shown in Fig. 1:
small positive slope (Mode I), zero slope (Mode II),
unit slope (Mode III), and negative slope (Mode IV).
To facilitate this discussion, the order of presentation
is Modes I, III, IV, and II.

Small Positive Slope (Mode I)


The initial portion of the data for each case in Fig. I,
denoted as Mode I, indicates that the log-log slope is
between the bounds given by Cases 1 and 3, and as a
result, the interpretation can be made that the
fracture is propagating under the assumption for
these slope bounds. These assumptions are those of
Perkins and Kern with confined height, constant
compliance, and unrestricted extension.

Unit Slope (Mode III)


The data sets for Cases 1 and 2 in Fig. 1 have a region
near the end with a unit positive slope, denoted as
Mode III. A unit log-log slope implies that the
pressure is proportional to time or, more importantly, that the incremental pressure change is
proportional to the incremental injected-fluid
volume. In the following, a basis is given for the
interpretation that a unit slope implies that a
significant flow restriction has formed in the fracture
(e.g., proppant screenout). Case 1 did sand out, and
pumping even at a very low rate could not be
achieved. The treatments for all other cases presented
in this paper ended on schedule.
Since the pressure is increasing at a significantly
greater rate for this condition than for the small
positive slope mode (Mode I), it follows that the rate
of change of one or more of the other variables in
. Eq. 9 must be significantly less. Since there is no
plausible basis, under the condition of increasing.
pressure, for height, compliance, or fluid loss to
decrease significantly over a major portion of the
fracture, it follows that the rate of extension must
have decreased significantly relative to the Mode I
condition.
The implication of this is that the only variable
changing significantly during the unit slope mode
(Mode III) is pressure, and since length extension has
been reduced significantly, relatively little fluid is
reaching the fracture tip (e.g., fracture screened out
with proppant at some point in the fracture). As a
result, the fluid being pumped into the fracture
essentially is being stored by increasing width or,
from Eq. 5, increasing pressure. With respect to Eq.
9, the rates of change of length, compliance, and
height are small compared with pressure, and
pressure is increasing at a rate proportional to the net
injection rate (qj -q,).
The treatment for Case 2 in Fig. 1 was believed to
be for a tip screenout because of a long unscheduled
SEPTEMBER 1981

shutdown during the pad that was not compensated


for by sufficient additional pad. This can be verified
by substituting C from Eq. 6 into Eq. 9, assuming
changes in length, compliance, and height are zero
and rewriting as
Ap _ 2(qj-q,)E' < 2q jE'

llt -

1rh2 L

(10)

1rh2 L . ....... .

Using the known values of q j =30 bbllmin = 168 cu


ft/min (0.08 m 3 Is), E' =6x 106 psi (4.1 x 104 MPa),
and h = 150 ft (46 m), neglecting fluid loss, and
estimating L (tip to tip) =5,000 ft (1640 m) gives
Aplllt=5.7 psi/min (0.65 kPals) , whereas Fig. 1
shows 8 psi/min (0.92 kPals). This approximate
agreement indicates that the unit slope for Case 2
appears likely to result from a tip screenout and gives
additional credence to the claim that a unit slope
implies a significant flow restriction at some point in
the fracture and that the fluid being injected is being
stored by increasing width.
A simple diagnostic tool can be derived from Eq.
10 for the distance from the wellbore (i.e., L12) that
a screenout occurred using the rate of pressure increase during the screenout.
q-E'
Ds. 2'
...................... (11)
1rh (ApI llt)
If ApI llt is small (i.e., D is large), the screenout
probably occurred near the tip (i.e., pad depletion)
and Eq. 11 can be used to estimate the propped
penetration. If ApI llt is large (i.e., D is small), the
screenout likely occurred near the wellbore (e.g.,
abnormal fluid loss near the wellbore).

Negative Slope (Mode IV)


The negative slope at the end of the treatment for
Case 2 in Fig. 1 was interpreted as rapid height
growth. In this section, a basis is developed to
support the premise that any significant decrease in a
fracture's pressure probably results from unstable
height growth.
From Eq. 9, a significant decrease in pressure must
be accompanied by a significant increase in one or
more of the other variables. A significant increase in
fluid loss is possible from opening new fractures or
fissures but is not likely with decreasing pressure, as
discussed in a later section. Also, from Eq. 4, a
significant increase in length is not compatible with a
decrease in pressure; however, a decrease in length is
compatible but not necessary.
From both Eqs. 4 and 9, an increase in height or
compliance is compatible with a pressure decrease.
However, as discussed in Appendix B, there appears
to be no physical basis, under the condition of
decreasing pressure, for a significant increase of
compliance over a major portion of the fracture
unless it results from a significant increase in height
(e.g., Eq. 6). Thus, the most probable cause of a
significant pressure decrease is a significant increase
in height. The relationship of height and pressure is
discussed in Appendix A.

Constant Pressure (Mode II)


The last characteristic of the plots in Fig. 1 to be
1769

discussed is the constant-pressure portion exhibited


by Cases 1 and 3 and denoted as Mode II. This
portion of the curve is the most difficult to provide a
definitive physical description, as discussed in the
following; however, this portion is potentially the
most significant. As can be seen for Case 1 in Fig. 1
and Case 4 in Fig. 2, the constant-pressure region
preceded rapid increases in pressure (potentially a
pressure or screenout condition) or an undesirable
height growth as for Case 3.
A region of constant pressure, IIp =0, places no
stringent requirement on the other variables in the
continuity equation (Eq. 9) or the flow equation (Eq.
4); however, Eq. 9 does imply that since the rate of
pressure increase is less than for Mode I (extension
under confmed height), the other variables must

..!

assumed process

Do

2600
::I 2000
8
C3 1600

::I

, r/H

8
- .. ' .~, .J
LII __

~
r

1000
20

40 60

= .075

Ci

"0

2000L:c
1600
ase 6

!
:I
!
A.

II -0.7, ,_( zone. low coefficient


0.1 ~ --. of friction (II)

; -..../ 1700
cas.e. ~. , "',"-III 1111

II

c(

increase at a larger rate. A rate of change of length


equal to or greater than that for the Mode I condition
is possible from Eq. 9 if the other variables do not
change but is not compatible with Eq. 4. Therefore,
the rate of extension is probably less than that for
Mode I extension.
Consequently, the constant-pressure region must
result from larger increases in fluid loss, height, or
compliance than with respect to the desired small
positive slope mode. As a result of the relative increases in one or more of these variables, more fluid
is lost to the formation, stored in additional height or
additional width. As discussed in the following, there
are creditable reasons why any of these variables can
increase at essentially constant pressure.
The potential for height increase at constant

:;1.1
0
~ E

::

.. u

C3

100
200 300 400
Time. Min.

1.0
E = 5 x 10 psi
v= .2

.....

.5

,g.

.4

H= 100 ft .
Normal stress on boundary= 5000 psi

oe

Fig. 2 - Similar critical pressures in same field.

.3

:j;n;,;..on ,

.2

w max

U)

"0

..".
.~

E
0

500
I1b- I1c

"b

t--,\

-ooj

1000

1500

Net Fracture Pressure. p/Tc


Z

3
!.

l1a- "c

leu

Q.

,,

...

!:::I

ct
!:::I

.::
z

..

1.0

,,

Fig. 4 - Effect of Slip on pressure and compliance.

j'

5". Sy: Far Fiald Horizontal Str...

Q.

,,

.....,

.r". "y. Local Strs at Fractur. Wa..

"2i'

.r = 5. = Closur. Pr sur.

..... .....

.9

'

-.. ... -.. -

.... ,--.,.
~

.........

:::I

_-

n
n

--Fracture

"-

.8

:J:
II

:J:

1~--------~2--------~3~--------~4

fTX=P=p+"

Slope= 1
........ ~~.ry=Sy-2'" p
~;

Slope = 2"11

...=5. -+'O'-----LPo---

P<Po"P+"'<"y:
Fissur. Clo
P>PO"P+">"'y:
Fissure Open
P=Po".-::t,y

Sy=.r+(1.2v)po

N.t FrllCtur. Pr ure. P


Normalized Height. H/H 1

Fig. 3 - Effect of height on pressure and compliance.


1770

or

Sy . 5"

po"~

Fig. 5 - Pressure to open fissures.

JOURNAL OF PETROLEUM TECHNOLOGY

pressure is shown in Fig. 3 and discussed in Appendix


A. Fig. 3 shows that as the fracture height penetrates
a section with higher stresses, a condition is reached
of increasing height at essentially constant pressure.
Also, Fig. 3 shows that the initial increases in height
are accompanied by decreases in compliance which
would tend initially to obscure the effect of the height
growth on the pressure (Le., Eqs. 4 and 9).
The potential for increasing compliance at constant height, discussed in Appendix B, is shown in
Fig. 4, which illustrates boundary slip and/or
material yielding at the top and/or bottom of the
fracture. If a compliance change was the major cause
of the constant-pressure regions, the long duration of
these regions would suggest an approximate doubling
of the average width or compliance. Fig. 4 indicates
that such changes in compliance generally cannot be
supported by boundary slip in the presence of
friction resulting from overburden pressures. An
exception would be a horizontal fracture along the
boundary. This would eliminate the friction;
however, the horizontal fracture would occur only
over the necessary distances if the fracture pressure
exceeded the overburden pressure. This was not the
case for the wells shown. Therefore, in these cases an
increase in compliance alone cannot account for the
constant-pressure regions.
The remaining variable that can increase at
constant pressure without violating Eq. 4 or 9 is the
fluid loss. Possible conditions are the initiation of
additional fractures from the well bore or opening up
natural fissures crossing the primary fracture. For
both these possibilities, the pressure required would
be slightly greater than the in-situ rock stress in the
direction parallel to the primary fracture, as
illustrated in Fig. 5 and discussed in Appendix C.
Both systems would tend to act as pressure regulators
and give rise to essentially a constant-pressure
condition. Since the width of these auxiliary fractures
would be proportional to the incremental pressure
above the opening pressure, and since the incremental pressure remains small because of the
regulator effect, the width of the fractures would be
relatively narrow and restrict proppant entry into
them. As a result, the auxiliary fractures would
provide a mechanism to screen proppant out of the
fluid entering these fractures and concentrate this
proppant in the main fracture. Therefore, auxiliary
fractures would tend to lead to proppant screenouts.
Although the physical mechanism that causes the
constant-pressure portions of the data in Figs. 1 and
2 cannot be determined directly with any certainty,
what occurs following these portions can give clues.
In all cases studied by the authors, the prolonged
constant-pressure portion was followed by either a
sharp increase or decrease in pressure and never a
return to the desired Mode I extension.
When the constant pressure is followed by a
decrease in pressure, the most likely cause is height
growth as discussed in the Negative Slope section.
This is supported further by the analysis in Appendix
A and is shown in Fig. 6 for Case 3 of Fig. 1.
Increases in fluid loss, height, or, to a lesser
degree, compliance could have occurred for cases in
SEPTEMBER 1981

which the pressure rises on a unit slope after the


constant pressure region and indicates a significant
flow restriction in the fracture. This follows because
an increase in any or all of these variables would slow
the fluid velocity in the fracture and accelerate
proppant deposition on the bottom of the fracture.
However, if the treatment was designed correctly,
excessive slurry concentration or dehydration (in
contrast to deposition on the bottom) would result
only for the case of accelerated fluid loss produced
by auxiliary fractures. Therefore, if the pressure
increase led to a proppant screenout all the way to the
wellbore (completely locked up as for Case 1) or a
rapid increase in pressure (i.e., Eq. 11), opening of
auxiliary fractures is a creditable cause for the
preceding constant-pressure region.
In summary, the prolonged constant-pressure
portion of the data in Figs. 1 and 2 potentially can be
caused by increases in height, fluid loss, and, to a
lesser degree, compliance. If the subsequent
pressures decrease, the most likely cause is height. If
the subsequent pressures increase, the cause could be
any of the three, except height in areas where experience indicates heights are essentially the same for
all treatments. If the subsequent pressure increase
rate is high and/or leads to a locked-up fracture with
proppant packed to the wellbore, the potential cause
is auxiliary fractures.

Application
The most important portion of the pressure curve to
identify is the constant-pressure region, since it
Log Derived
Stress. S. psi

Depth. ft.
r---_~-

9000

S=6055
S=6845

1~~~~~:9470
9495

S = 6550

""'-Io-t---.j

9525
U"c=6150

686~5E=~

1000,

800
Increase Effective Stress
by 50% in Shale Zones (I,II,III,V)
to Correlate with Treating Pressures
(Case 3. Fig.1)

..

Fracture Top

InZ~

-111-~~~~--11--..
~ I~.--I

OIl

70

100

150

:::...--..

200

Fracture. Height. ft.

Fig. 6 - Height VS. pressure for multizone section.


1771

f!'= Closure Pressure

> IT+P

= Net Fracturing
Pressura
V = Poisson's Ratio
P

IT

> u+p

0.10
0.15
0.20

(1-211 )
0.8
0.7
0.6

>IT+(1-2V)p

Fig. 7 - A sufficient stress condition for confined fracture


extension,

indicates a reduction in fracture extension rate and


subsequent undesirable consequences. As a result,
this pressure is termed the "critical pressure." Since
the critical pressure is the fracturing pressure
capacity of the formation and is associated with
physical phenomena related to the in-situ stress
environment surrounding the fracture (Le., height
extension or opening auxiliary fractures), it should be
expected that the pressure capacity of the formation
and, hence, the critical pressure would be the same
for offset wells in the formation that have the same
lithology and tectonic histories. Thus, the critical
pressure, in general, should be similar for a relatively
large area of the same pay zone. This apparently is
the case for the Wattenberg field of the Denver basin,
as shown in Fig. 2, where two wells from this field
indicate a critical pressure of approximately 1,700 psi
(11. 7 MPa) above the in-situ closure pressure. Other
well data for this field, not shown, also support this
approximate value of a critical pressure. Case 2 (Fig.
1) is from this field; however, as indicated earlier,
this well is suspected of a tip screenout before
reaching a pronounced constant-pressure phase.
However, for this well, a slight indication of a
constant-pressure region is found at about 1,700 psi
(11.7 MPa) and preceding the unit slope.
For areas where treatments reach a critical pressure
with the resulting excessive pressure increases or
heights, treatments should be altered to reduce
pressures. The flow equation Eq. 4 implies that this
can be achieved by reducing flow rate, viscosity, or
length. It is not desirable to reduce length; reduced
flow rate would reduce fluid efficiency and increase
required volumes; and reduced viscosity may
decrease proppant transport and the propped
fracture length. Thus, the critical pressure can place a
real constraint on the propped length which can be
achieved in a cost-effective manner. It is easier to
conclude what not to do. Historically, when treating
pressures became excessive, a common response was
to conclude that the primary cause was poor sand
transport, and consequent changes were to increase
viscosity and reduce proppant. Recognizing that the
criticl:l1 pressure results from the limitations of the insitu rock stresses and not directly from proppant,
that response may be incorrect. Increasing viscosity
1772

results in higher pressure and reaching the critical


pressure sooner in a given treatment. Also, reducing
proppant would reduce the fracture capacity for
subsequent production.
Recognizing that additional length extension
probably is reduced significantly after the critical
pressure is reached, the optimal treatment is one in
which the pressures are limited to below the critical
pressure or reached only at the end of the treatment.
To achieve this result may require a reduction in the
treatment volume, viscosity, and/or injection rate
and the acceptance that a previously desired propped
length may not be feasible. Sufficient, and not excessive, viscosity to transport the proppant to the
fracture tip should be used in conjunction with a
proppant schedule that provides a sufficient pad and
a high proppant concentration for all parts of the
fracture. To achieve this, a fracture simulator is
required that can track slurry elements from the
entrance to their ultimate position and account for
the fluid loss and subsequent concentrating of the
proppant. In addition, "sufficient viscosity" implies,
for large treatments in hot zones, the reduction of the
viscosity of the fluid entering the fracture throughout
the treatment. The fluids entering later do not require
relatively high initial viscosity because they are exposed to relatively less time and temperature
degradation. Eq. 4 implies that any excess viscosity
(Le., K) above the minimum required to transport
proppant will reduce proportionally the penetration
(i.e., L) achieved before the pressure reaches the
critical value (Le., the capacity of the formation).
That is, twice the required viscosity will result in onehalf of the potential penetration when the critical
pressure is reached.

Implications
The identification of a critical pressure, related to the
in-situ rock stresses, raises a question concerning the
adequacy of the in-situ stresses to permit a fracturing
pressure large enough to produce the desired fracture
penetration.
Figs. 1 and 2 indicate Mode I net pressures in the
range of 1,000 to 2,400 psi (6.9 to 16.6 MPa) for the
three areas represented by these wells. For height
confinement, a horizontal stress difference greater
than the net pressure generally will be required
between the pay zone and confining zones above and
below (Figs. 3 and 6). Also, to avoid opening
auxiliary fractures and the resulting large fluid loss,
the principal stresses in the horizontal plane must
differ by about 7011,10 of the net pressure (Figs. 5 and
7). These two requirements are significant since these
differences usually are neglected in most analyses of
rock stresses. In light of the potentially large pressure
required to create long, propped fractures, these insitu stress requirements are significantly severe that
they should not be anticipated to exist for all, or even
most, potential zones that may require long fractures
and massive hydraulic fracturing treatments.
Recognizing the severity of these in-situ stress
requirements, we should give serious consideration to
determining the magnitude of the critical pressureand the associated limitations on fracturing - from a
JOURNAL OF PETROLEUM TECHNOLOGY

calibration treatment in areas where the potential


exists for large drilling and treatment expenditures.
An ideal calibration treatment would consist of three
stages: (1) a breakdown/flowback procedure, (2) a
relatively large treatment, and (3) a long-term
pressure decline after the treatment. The first stage
would provide a determination of the fracture
closure pressure, which is required as the datum
pressure for interpretive plots such as in Figs. 1 and
2. The second stage would provide the data, similar
to Figs. 1 and 2, for determining the critical pressure,
and the final stage would provide data on the fluid
loss characteristics 4 required for designing proppant
schedules for future treatments. Ideally, the second
phase would use the fluid type anticipated to be used
commonly in the area and would not contain
proppant, or less than the optimal proppant. The
reduction of proppant is to ensure that the fracture
does -not screen out before reaching the critical
pressure and to enhance the interpretation of the
decline data.

Conclusions
This paper has demonstrated that a log-log plot of
fracture treating pressure, above closure stress, vs.
treating time (or equivalently the accumulative fluid
volume) provides a potentially powerful tool to
interpret the fracturing process for vertical fractures
with penetrations appreciably greater than the
height. In particular, identification can be made of
(1) the phase during which the desired confinedheight maximum extension rate occurred (Mode I)
and (2) a critical pressure (Mode II) that subsequently
will lead to either an undesirable pressure increase
due to restricted penetration (Mode III) or accelerated height growth (Mode IV).
One consequence of the critical pressure is that it
cannot be assumed that all, or even most, potential
zones will have the necessary rock stress conditions to
provide the fracturing pressure capacity required for
deep penetrating fractures.
The critical pressure should be similar throughout
a significant area of a given pay zone and, after being
quantified, can be used to provide more costeffective treatments in this area.
The data presented were interpreted to be consistent with the fracturing assumption of Perkins and
Kern 2 (Mode I) during the initial portions of the
treatments with subsequent variations from these
idealized assumptions during the latter portions of
the treatments - i.e., variations from constant height
(Modes II and IV), normal matrix fluid loss (Mode
II), and unrestricted extension (Mode III).

Nomenclature

= fracture width
= fracture compliance
= distance from well
e = time exponent

b
C
D
E'
h
K
L

= plane strain modulus


= fracture height
= power-law coefficient
= fracture length

SEPTEMBER 1981

n = power-law exponent
= injection rate
q f = fluid-loss rate
p = pressure above closure
t = time
V = fracture volume
..:l = difference
oc = proportional

qj

References
1. Khristianovic, S.A. and Zheltov, Y.P.: "Formation of Vertical
Fractures by Means of Highly Viscous Liquid," Proc., Fourth
World Pet. Cong., Rome (1955).
2. Perkins, T.K. and Kern, L.R.: "Widths of Hydraulic Fractures," J. Pet. Tech. (Sept. 1961) 937-949; Trans., AIME, 222.
3. Nordgren, R.P.: "Propagation of a Vertical Hydraulic
Fracture," Soc. Pet. Eng. J. (Aug. 1972) 306-314; Trans.,
AIME,2S3.
4. Nolte, K.G.: "Determination of Fracture Parameters from
Fracturing Pressure Decline," paper SPE 8341 presented at the
SPE 54th Annual Technical Conference and Exhibition, Las
Vegas, Sept. 23-26, 1979.
5. Harrison, E., Kieschnich, W.F. Jr., and McGuire, W.J.: "The
Mechanics of Fracture Induction and Extension," Trans.,
AIME (1954) 201, 252-263.
6. Treatise on Fracture, H. Liebowitz (ed.), Academic Press Inc.,
Washington, DC (1968) 2.
7. Simonson, E.R., Abou-Sayed, A.S., and Clifton, R.J.:
"Containment of Massive Hydraulic Fractures," Soc. Pet.
Eng. J. (Feb. 1978) 27-32.
8. Rosepiler, J.M.: "Determination of Principal Stresses and
Confinement of Hydraulic Fractures in Cotton Valley," paper
SPE 8405 presented at the SPE 54th Annual Technical Conference and Exhibition, Las Vegas, Sept. 23-26, 1979.
9. Ruina, A.L.: "Influence of Coupled Deformation-Diffusion
on Retardation of Hydraulic Fracture," Proc., 19th U.S.
Symposium on Rock Mechanics, Y.S. Kim (ed.), U. of NevadaReno (May 1978).
10. Tueffel, L. W.: "An Experimental Study of Hydraulic
Fracture Propagation in Layered Rock," PhD thesis, Texas
A&M U., College Station (Aug. 1979).
11. Daneshy, A.A. and Conrad, N.: "Fluid Pressure Variations
During Hydraulic Fracturing," Proc., 19th U.S. Symposium
on Rock Mechanics, Y.S. Kim (ed.), U. of Nevada-Reno (May
1978).
12. Sneddon, LN. and Lowengrub, M.: Crack Problems in the
Classical Theory of Elasticity, J. Wiley and Sons Inc., New
York City (1969).

APPENDIX A

Effect of Fracture Height Variation


on Pressure Behavior
The height of a vertical fracture has a major in- .
fluence on the wellbore pressure (Eq. 4). The
magnitude can be large (Fig. 1) and considerably
more than the minimum pressure required to extend
the fracture tip. This gives rise to two basic
questions.
1. What natural mechanisms or forces can contain
these pressures and confine height?
2. What will be the effect on the pressure if the
barrier is overcome and height growth occurs?
Several physical mechanisms have been proposed
as potential barriers to height extension, and some of
these are discussed here as they pertain to these
questions. It long has been recognized that variations
of the in-situ stress between formations can provide
substantial barriers to vertical growth. 5 This effect
can be calculated in a straightforward manner for
linear, elastic materials, 6 as quantified in terms of
1773

hydraulic fracturing in Ref. 7. The details are not


repeated here, although a couple of brief examples
are considered.
In general, a stress boundary will provide a
relatively stable boundary for initial pressure increases; however, vertical growth then will accelerate
until the fracture is growing vertically at very nearly
constant pressure, as shown by the illustrative
example in Fig. 3 (which, along with the example in
Fig. 6, was calculated using elastic finite-element
techniques with an energy-release-rate crack-growth
criterion). If the high-stress formation has a finite
thickness, then fracture growth could become unstable following penetration through the top of this
formation as seen in Fig. 6. Unfortunately, the
difficulties in obtaining accurate data for these
stresses can be considerable. Many methods have
been proposed, but hydraulic fracturing remains the
only viable technique for measurement of in-situ
stresses at any depth, and even this may not be as
simple as previously treated. Analysis is complicated
further for cased and perforated completions. Also
for the case of finite-thickness bounding beds, the
use of mini-fractures for determining formation
stresses could be detrimental to the ultimate
stimulation success. Another proposal for the indirect determination of in-situ stress has been the use
of sonic logs to determine the in-situ moduli of the
formations and, subsequently, the use of these data
to calculate stresses by assuming boundary conditions. Calculations along these lines (where an
increase in Poisson's ratio indicates increasing stress)
have been successful in predicting stresses in some
deep sandstone formations 8 and, in some cases, do
predict higher stresses in surrounding shales.
This procedure is attractive from an operational
view but has several pitfalls, including proper log
interpretation and the fact that any elastic analysis
will provide only a lower bound for the stresses. This
can be visualized easily for the normal case where the
overburden applies the maximum stress and produces
shear stresses in the reservoir. Since real materials
will tend to behave somewhere between an elastic
solid and a viscous fluid, these shear stresses will tend
to relax over time and result in higher in-situ stresses
in the horizontal plane than would be predicted from
an elastic solution. Qualitative indications for such
high stresses have been seen by one of the coauthors
in strain relaxation measurements on Second
Frontier shale samples (sands showed no relaxation)
and by others in strain-gauge overcoring results on
sand and shale samples from the Muddy-J formation. For Case 3 (Fig. 1), it was possible to use
lithological correlations and sonic-log-inferred
Poisson ratios from an offset well to predict stresses
for a well where bottomhole pressures were recorded
during fracturing. These calculated stresses for
several zones surrounding the target formation are
shown in Fig. 6 along with the associated
pressure/height relationship. However, a 50070 increase (as shown in the figure) in the effective stresses
for the shale sections was required to match correctly
the treating pressure behavior for this case and an
1774

offset well. This analysis implies considerable stress


relaxation in the shale sections.
The effect of an increased fracture height on
treating pressures will be a general lowering or at
least a decline in the rate of pressure increases. This
effect can be separated into three distinct categories
on the basis of the nature of the physical barrier that
is controlling height. These are material property
confinement, slow stable growth, and blunting.
For material property confinement, such as an
increase in fracture toughness or moduli, the crack
initially will arrest at the formation interface and the
pressure will behave as predicted by basic theory.
However, the pressure can increase only to a fairly
moderate level before the formation interface is
penetrated and vertical growth becomes unstable
with vertical propagation continuing with decreasing
pressure.
Other confining mechanisms can create slow,
stable height growth with the length extension greatly
exceeding any vertical growth. Such a situation can
be created by in-situ stress changes or by a permeability (or effective stress) barrier. 9 The former
was discussed in the preceding, while the latter is a
result of the dilatation ahead of the crack tip. This
dilatation causes a decrease in pore pressure in this
region and results in high effective stresses that
prevent rapid fracture growth. The general results
will be similar for either case as shown in Fig. 3. The
initial low rate of height growth will create behavior
as if height were constant. As pressure and height
continue to increase with time, the vertical growth
will accelerate and will result in a lowering of the rate
of well bore pressure increases.
Experiments concerning the behavior of fractures
at a frictional interface have shown 10 that under
certain conditions of confining pressure and overburden load (which provides the normal force along
the friction boundary), a crack will arrest and then be
reinitiated at an increased pressure and continue to
propagate as seen by the illustrative example in Fig.
4. However, such reinitiation is difficult to predict
since an analysis of the shear stresses on the interface
still predict a stress singularity; therefore, a "stress"
analysis for failure is not appropriate.
APPENDIXB

Effect of Compliance Changes


on Fracturing Pressure
Changes in fracture compliance can result from two
basic mechanisms. The first is the penetration of
formations with differing moduli or in-situ stresses
(Fig. 3). The second mechanism is nonlinear behavior
such as blunting or ductile deformation. This type of
behavior could have extreme effects as illustrated by
the discussion in the Introduction on the two different theories for fracture geometry.
The penetration of various formations can cause
large changes in compliance; however, with the
exception of increasing stresses, such penetration
normally will result in unstable height growth. For
this reason, only the effect of compliance changes
with increasing stresses is discussed. Normally,
JOURNAL OF PETROLEUM TECHNOLOGY

compliance will increase linearly with height, but if


this growth is into formations with higher stresses (or
moduli), the width will be less than calculated for a
single zone. This can be accounted for by means of
an increase in effective modulus or a decrease in
compliance. Fig. 3 pictures the changing compliance
associated with height growth into highly stressed
formations. Viewed in the context of the continuity
equation (Eq. 9), the initial decreases in compliance
will tend to offset the increases in height growth.
Only when the compliance changes become relatively
small with subsequent height extension will large
deviations begin to occur in the pressure behavior.
No simple analytical solutions are possible for
fracture blunting, but such behavior is subject to
numerical simulation. An example of this is seen in
Fig. 4 for the case of a friction boundary or interface
slip. This example was calculated using the finiteelement technique with slip elements. As shown, only
a minor increase in compliance resulted. These
changes are typical for this type of analysis, and the
unrestrained slip required by one theory of fracture
geometry I does not appear to be possible except for
very shallow depths or perhaps highly overpressured
formations.

APPENDIXC

Auxiliary Fractures
As the fracturing pressure increases, the potential
exists for auxiliary fractures to be created in addition
to the primary fracture. These fractures can result at
the wellbore or by opening natural fissures. II Even a
closed natural fissure intersecting the primary
fracture in a low-permeability reservoir will have
significantly more permeability than the surrounding
matrix. As a result, fluid with pressure comparable to
that in the primary fracture will penetrate relatively
deep into the fissure. The fissure will open when the
fluid pressure inside the fissure exceeds the rock
stress tending to close the fissure. This condition is
illustrated in Fig. S. The illustration shows a rock
element with dimensions that are small compared
with the fracture's height and that is along the wall
and near midheight of the fracture. The rock
pressure tending to close the fissure will increase as
the fracture pressure increases. For the assumption
of plane strain in the direction parallel to the primary
fracture, the closing pressure (denoted by (]y in Fig.
5) will increase by an amount depending on the net
fracture pressure p and Poisson's ratio v, while the
vertical stress l2 will increase by p. For this assumption, the increase in the closing stress is 2vp. Since the
closing stress (]y increases at a rate less than p, there
will be a value of p for which the fluid in the fissure
exceeds the closing stress and forces the fissure open.
This pressure is denoted as Po in the figure. For the
condition when the fissure opens,p=p", a condition

SEPTEMBER 1981

of isotropic horizontal stresses exists in the rock


element with (]x = (])' = total fluid pressure in the
fracture and fissure. This condition implies that the
normal stress is the same on any vertical plane
through the element. Therefore, the pressure
required to force the fissure open is independent of
the fissure's orientation. As shown in Fig. 5, the
pressure above the closure pressure Po to open the
fissure is equal to the difference in the far field
principal stresses divided by one minus twice
Poisson's ratio.
An auxiliary fracture in a permeable zone will tend
to act as a pressure regulator at the opening pressure
Po' This occurs because at pressures p above Po' the
fissure opens with a width proportional to p - P 0 and
the flow rate into the fissure depends on the width
raised to the 2n + 1 power (n = power-law exponent
of the fluid; 2n + 1 generally> 2). Thus, the flow rate
from the primary fracture is a nonlinear increasing
function of P-Po' which, from the continuity expression (Eq. 9), reduces the rate of pressure increase
proportionally. A doubling of p - p 0 would result in
at least a fourfold increase in fluid loss from the
fracture. The fluid flowing into the fissure is
dissipated by fluid loss to the newly exposed matrix
surrounding the fissure. As a result, the open fissure
regulates the pressure in the fracture to values slightly
above Po' If the pressure were to drop below Po' the
fissure would close, permitting the pressure to increase again above Po' This mechanism can continue
until the entrance to the fissure(s) becomes plugged
with proppant and permits the pressure to increase
significantly above Po'
For an auxiliary fracture to propagate from the
wellbore, similar pressure conditions would be
required once the fracture penetrated beyond the
influence of the wellbore (Le., several wellbore
diameters) and perforations around the wellbore. As
an auxiliary fracture propagates away from the
influence of the primary fracture (Le., on the order
of the fracture height), it will tend to align itself
normal to the minimum principal stress or parallel to
the primary fracture. As a result, the auxiliary
fracture will not add significantly to the drainage
area or to the long-term productivity of the fracturing system, even if the auxiliary fracture is
propped.

SI Metric Conversion Factors


ft x 3.048*
psi x 6.894 757
Conversion factor is exact.

E-Ol
E + 00

m
kPa

JPT

Original manuscript received in Society of Petroleum Engineers office July


25, 1979. Paper accepted for publication May 2, 1980. Revised manuscript
received Aug. 12, 1981. Paper (SPE 8297) first presented at the SPE 54th Annual
Technical Conference and Exhibition, held In las Vegas, Sept. 2326, 1979.

1775

Potrebbero piacerti anche