SPE

SPE 23434

Fracturing Net Pressure Analysis for Variable-Rate Injection

O.K. Poulsen, Halliburton Services
SPE Member

Copyright 1991, Society of Petroleum Engineers, Inc.

This paper was prepared for presentation at the SPE Eastern Regional Meeting held in Lexington, Kentucky, October 22-25, 1991.

This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(sl. Contents of the paper,
as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(sl. The material, as presented, does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees ot the Society
of Petroleum Engineers. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledg-
ment of where and by whom the paper is presented. Wrne Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836 U.S.A. Telex, 730989 SPEDAL.

ABSTRACT instantaneously reach the design value when pumping

Using various graphical techniques to observe net fracturing
pressure behavior can be invaluable in diagnosing fracture
growth behavior, often providing the engineer with
information that allows him to take corrective action as the
treatment is proceeding. Techniques previously presented
for plotting net pressures and methods for interpreting the
plots were developed under the implicit assumption of
constant injection rate. New, more general graphing
techniques are shown to be less affected by variations in
injection rate, extending net pressure analysis to a wider
variety of situations.
IMTRODUCTION
The log-log plot of net fracturing pressure versus injection
time was introduced as a tool for diagnosing fracture growth
behavior in 1981 1 and has been widely and successfully
used since then. Various guidelines have been presented1-7
for interpreting these graphs and additional, related plotting
techniques have since been developed.7·8
One assumption implicit in most, if not all, of the guidelines
given for explaining the curves on these graphs is tha~ the
injection rate is constant. Typically, slurry injection rates
are held more or less constant throughout hydraulic
fracturing treatments. In a strict sense, though, injections
rates are never constant. For example, rate does not
References and illustrations at end of paper.
starts, but increases continuously, albeit quickly, from zero
to the design rate with fracture growth initiating before the
design rate is attained. Although this most especially
affects the early-time behavior on Iog(Ap)-log(t) plots, it may
also affect the interpretation of the entire plot. In addition,
brief shut-in periods for calibrating perforation frictions, etc.,
may be inserted into the treatment schedule. Rate
variations may also occur as equipment is brought off and
on line dUring the course of a treatment. And even when
the rate is held constant, mechanically induced variations in
the injection rate exist.
It is of interest to develop methods for interpreting net
fracturing pressures that are less dependent on variations
in injection rate. This paper presents and demonstrates
new, more general methods of graphing net pressures
based on injected volume rather than injection time and
shows that most guidelines previously developed for time-
based plots may be applied to these volume-based plots.
Under the condition of constant injection rate, many of the
developed methods, in fact, reduce to previously presented
techniques.
MATHEMATICAL DEVELOPMENT
For constant height fracture models such as those based
on the Khristianovic and Zhelto~11 or Sneddon12-14 width
equations (the former of which will be referred to as KZ-
type models and the latter as PK-type models), the volume
of the fracture is simply the product of the total length, the

185
2 FRACTURING NET PRESSURE ANALYSIS FOR VARIABLE RATE INJECTION SPE 23434

height, and the average width; for a radial fracture,12,13 the

volume is proportional to the square of the radius times the
width.

(KZ-typ6) (6)
(PK-typ6) (1)
(radls!}

From the geometry of the fraclure, a proportionality

constant between the maximum width of the fracture at the From Eq. 6, we car1 see that for fracture length or radius to
wellbore and the average fracture width can be found for remain constant (restricted extension). which may be
each of the geometries. indicative of proppant bridging, the net pressure must
remain in proportion to fracture volume, assuming, of
7r course, that fracture height is unchanging for the KZ and
4 PK models. This translates to a slope of 1 on a logarithmic
w: (2) plot of Ap versus V,. Since, by definition
w- :!!l2n+2)
4l2n+3 "*
2 V.
TI - -..!. (7)
3 It,
Substituting Eq. 2 into Eq. 1 gives the relationships between
fracture volume and maximum fracture width: and, for a constant injection rate,

It, '" at (8)

:!£HL
2
it can be seen that the interpretation is the same as that for
v, = 7r(.J!!!...)HL
2n+3
w:"* (3)
a slope of 1 on a Iog(Ap)-Iog(t) graph generated under the
27r R2 condition of constant rate of increase in fracture volume.
3 011.1-3.8
The fracture width equations for the various geometries Also from Eq. 6, it can be seen that for fracture length or
are9,12 radius to decrease as fracture volume increases (i.e.• for
length to be a monotonically decreasing function of V,).
either the Iog(Ap)-log(V,) slope must be greater than 1 or
(4) the fracture height must increase. This reveals that a slope
greater than 1 on such a plot may indicate that the fracture
length is decreasing. Because the rock cannot heal, this
may be best described as an effective decrease in length
Solving the fracture width equations (4) for Ap. we obtain or radius, caused most likely by proppant packing off the
fracture increasingly nearer to the wellbore. The analogous
1 guideline for a Iog(Ap)-Iog(t) plot was presented by
2L Poulsen.8
E' 1 (5)
IJ.p - -w: H Nolte3 claimed that a slope of 2 on a Iog(Ap)-log(t) plot
2
indicates that "half the prior fracture area [becomes]
- 7r restricted to flow [possibly resulting] from one wing of the
4R
fracture becoming blocked to flow." Poulsen8 showed this
Solving Eq. 3 for W~ and substituting the result into Eq. 5 interpretation to be in error and using an analogous
shows that argument here, we can see that complete blockage of flow
into one fracture wing would result in a doubling of flow into
the remaining wing. assuming injection rate remained
constant. The doubled rate into the single wing would
result in a correspondingly higher pressure. such as would

186
SPE 23434 DON K. POULSEN
be seen if the injection rate into two unrestricted wings had
been doubled. This would result in an increase in slope on
a cartesian plot of L\p versus V, or -reduced time; 11t,7 but,
on a log-log graph of either sort, the curve would simply
exhibit a vertical shift similar to that shown on Fig. 1. In
other words, slopes on 1og(L\p)-log(V,) graphs or those on
Iog(L\p)-log(11t) graphs when rate is constant are
independent of the value of rate for well-eonfined and radial
fractures. This is contrary to the contention that blockage
of flow into one of the fracture's wings (presumably with
growth of the other wing restricted) would result in a log-log
slope of 2.
As illustrated in Fig. 1, under most circumstances a flow
restriction would not occur instantaneously, but would result
from a gradual packing of the fracture. For certain
circumstances, the degree of shift accompanying a flow
restriction or a change in rate may be determined through
relationships presented in the Fluid Mechanics
Considerations section below.
As pointed out above, a slope greater than 1 could indicate
a continuous, but possibly rapid, blocking process of one or
both fracture wings. By substituting 11Vj for V, in Eq. 6, we
can see, very similarly to what Martins et al.5 pointed out for
a Iog(L\p)-log(t) plot, that after tip screenout has occurred,
the increase in fluid efficiency resulting from having a
constant or decreasing fracture area will result in an even
larger slope on a log-log net pressure versus injected
volume plot, assuming injection rate is constant, or at least
not decreasing rapidly enough to counteract the decrease
in fluid-loss rate.
To obtain further guidelines, we can solve Eq. 1 for length
or radius and substitute the result into Eq. 6 to get
2WZH
V,
Ap= EI(2n+3~-
-- (9)
n n+1 H
3(nW)312
16~fl
From Eq. 9, we can see that for fracture width to remain
constant, l1p must be inversely proportional to V, for KZ
geometry, constant for PK geometry, and inversely
proportional to the square root of V, for radial geometry.
Stated in terms of slope, £y, on a logarithmic plot of L\p
versus V" .
3
-1
o (10)
EV'" 1
2
Slopes lower than these would indicate that the fracture
width is narrowing. This could be indicative of less
restricted height growth resulting from penetration into a
zone of lower least principle stress. It could also indicate
fracture penetration, vertically or horizontally, into an area
of higher fluid-loss rate. Again, this is in agreement with
guidelines developed for Iog(L\p)-log(t) graphs.1.3.6
Besides being less dependent on injection rate, the
advantage of plotting Iog(l1p) versus log(V,) is that such a
plot would be, to a large extent, independent of fluid
efficiency as will be demonstrated in the next section. This
should make the plot applicable to shut-in and flowback
conditions as well as injection. For constant rate injection,
using the logarithm of reduced time, 11t,7 as the abscissa
variable can be seen from Eqs. 7 and 8 to be equivalent to
using log(V,) .
Of course, since it is not possible, using current technology,
to determine fracture volume, and thus fluid efficiency, with
any accuracy, developing guidelines for interpreting 10g(L\p)
vs. log(V,) or 109(11t) plots is, to some extent, just an
intellectual exercise. To bring things to a more practical
level, the relation of Eq. 7 allows us to substitute 11Vj for V,
in the above equations. If we can then assume that fluid
efficiency is constant, or nearly so, as was implicitly done in
earlier works1.U , then the derived guidelines remain valid
when we create a 10g(L\p) vs. Iog(VJ graph or one of the
alternate forms discussed in the Plotting Techniques section
below. This is not an unreasonable assumption for, when
there is no fluid loss the efficiency is identically 1, and as
has been shown elsewhere 15 for high fluid loss, constant
injection rate, and normal fracture growth
n
t- 2(n+1)
1+2n (11)
11 .. t - 4(n+1)
2+5n
t - 8(n+1)
Equation 11 shows that for a Newtonian fluid (n=1),
4 FRACTURING NET PRESSURE ANALYSIS FOR VARIABLE RATE INJECTION SPE 23434

n
H~ 2(""')
'I,..,. (12) 3 3n+2 1
(15)
lip • Q 4("..') H 2("..1) ~4(""1)

_n_ .-!!L..
Q I(n+1) ~ 8(""1)
and for the (unattainable) lower bound on the flow behavior
index (n=O), Although Eqs. 14 and 15 were derived using certain
assumptions about rate behavior, several interesting
observations can be made from these equations.
(13)
Equation 14 and Appendix A show that, if logarithm of net
pressure could be plotted versus logarithm of fracture
volume for a true constant height or radial fracture, the
slope of the resulting curve would be largely independent of
demonstrating that, in most instances, fluid efficiency is not the amount of fluid loss. In addition, for KZ-type or radial
a strong function of time. fractures, the predicted slopes on this type of plot are
identical.
Fluid Mechanics Considerations
Another observation is that the predicted slopes for a log-
By considering fluid mechanics, it is possible to derive log net pressure vs. injected volume plot are identical to
ranges on the Iog(~)-log(V) slope in a manner similar to those for a log-log net pressure vs. time plot 1,3 under the
that of Ref. 3. constant rate condition. The reason for this can be seen by
substituting the product at for Vi in Eqs. 14 and 15 to get
If we assume the functions gL and gR' defined in Appendix the net pressure-time relationships.
A as the ratio of flow rate at a given position to the injection
rate, are approximately independent of injection rate, then And, although the behavior of data plotted on a Iog(Ap)-
Eqs. A-10, A-11, and A-21 of Appendix A show that, Iog(Vj) graph is not completely independent of variations in
independently of fluid-loss rate injection rate, with the possible exception of KZ geometry
at low efficiencies, it is, in many cases, affected less by any
n .-!!- such variations. This can be seen by comparing the
Q "..2 \', n+2
magnitudes of the exponents on a in Eqs. 14 and 15 to
n 3n+3 1
(14) those in the relationships obtained when at is substituted
lip· Q 2n+3 H- 2n+3 v,2n+3 for Vj' Using the limiting n values of 0 and 1, it can be seen
-!!- .-!!- in ten of the resulting twelve comparisons that the absolute
Q"..2V, n+2 values of the rate exponents in the volume relationships are
less than or equal to those in the time relationships.
Relationships between net pressure and injected volume for
minimal fluid-loss conditions may be obtained by The slope on a 10g(Ap)-log(Vj) graph and that on a 10g(Ap)-
substituting Vj for VI in Eq. 14. Iog(t) graph are related by

Appendix B shows that under conditions of high fluid loss

and constant injection rate, net pressure is related to the
injected volume through
SPE 23434 DON K. POULSEN
From this we can see that if the instantaneous injection rate
is greater than the average injection rate up to the point
under consideration, then the slope on the time graph will
be larger than that on the volume graph and vice versa. It
also reveals the greater dependency of f, on the prior rate
history.
Plotting Techniques
As has been mentioned, one method for facilitating the
interpretation of net f~acturing pressures is to plot the
logarithm of net pressure versus the logarithm of injected
volume, from which the slopes exhibited by the data provide
the primary source for data interpretation. Using this
method, however, requires some measurement and
calculation if precise slope values are desired. A more
direct, and thus quicker, determination of fracture behavior
can be made from a plot having the log-log slope as the
ordinate value. The slope is calculated as
E ~ t(Iog(AP)] ~ ~ dAp ~ ~ dp. (17)
v c(log(~1 lAp dV, lAp dV,
Several numerical techniques exist for calculating the
derivative values.
The abscissa value may be any of several variables;
however, to allow direct comparison with the 10g(Ap) versus
log(V) graph, plotting log(Vi) on this axis is the most
practical choice. The use of a graph of 1og(l1p)-Iog(t) slope
vs. log(t) has been previously introduced?
As can be noted from Eqs. 14 and 15, the log-log slope
varies with fluid efficiency; thus, at least theoretically, a
relative indication of fluid efficiency may be obtained by
plotting a normalized slope,
(18)
as the ordinate value, where
n similar to that presented earlier, this implies that plots of
2(n+1)
1 (19)
Eo"" 4(n+1)
3n
8(n+1)
and
189
5
n
n+2
1 (20)
E1 ""
2n+3
n
n+2
If the fracture behaves as predicted by the assumed model,
the normalized slope will have a value of 1 when the
efficiency is 1 and a value of 0 when the efficiency is O. Be
cautioned, however, that although the normalized slope and
the fluid efficiency correspond at these two values, the
relationship between slope and efficiency is not necessarily
linear. An empirical relationship between these values has
been developed for PK-type geometries. 18 Of course, to
use the normalized slope plot, one must assume that the
fracture behaves according to a particular fracture growth
model. One must also assume the value of the fluid's flow
behavior index, n.
A more practical alternative to creating an actual
"normalized slope" graph is to plot horizontal lines
corresponding to the limits given in Eqs. 19 and 20 on the
derivative plot. In doing so, the actual slope values are
retained and comparisons to the behavior predicted by each
of the models can be realized. This is illustrated in the
examples below.
Recently, Ayoub at al.8 introduced the technique of plotting
Iog(tldPJdt]) vs. log(t). This graph will yield, as is noted in
that paper, the same slope as the Iog(Ap) vs. log(t) graph,
but only when a power-law relation holds between l1p and
t; i.e., when the Iog(Ap)-log(t) slope is constant.
From Eq. 17, it can be seen that
~~ ... E.,AP (21)
and thus, relationships between Vi(dPJdV) and Vi may be
obtained by multiplying both sides of those equations (6,
14, and 15) relating Ap and Vi by £yo Through reasoning
Iog(Vi[dPJdVJ) vs.Iog(V) will be less sensitive to variations
in Injection rate than will the plots of Ayoub and therefore
be applicable over a greater range of conditions.
A major shortcoming of both of these plots Is that they
cannot handle constant or decreasing pressures. A partial
remedy Is to plot pressure decreases on a graph having
Iog(-VJdpJdVJ) on the ordinate. Unfortunately since
6 FRACTURING NET PRESSURE ANALYSIS FOR VARIABLE RATE INJECTION
pressure increases and decreases commonly occur within
the same treatment, this will require two graphs or two
distinct curves on a graph with a IoglVldPJdVJI ordinate
scale.
Most any other valid guidelines for interpreting graphs
based on injection time at constant injection rate can be
shown to be valid for interpreting graphs based on injected
volumes. These include, but are not limited to, guidelines
for creation of multiple parallel fractures," intersection with
natural fractures, and intersection with bounding faults. 8
LIMITATIONS
One instance in which using a graph based on injected
volume does not provide the information that a graph based
on fracture volume would is during a period when the
fracture is shut in. By definition, Vi does not change when
a well is shut in, but the volume that has leaked-off, and
thus the fluid efficiency continues to change, invalidating the
assumption of near proportionality between injected and
fracture volumes. More simply put, because Vi is not
changing during a shut-in period but Ap is, a Ap versus Vi
plot will display a vertical line and a -00 value will be
calculated for a slope or derivative plot. However, in
reference to a Ap versus t plot, little is lost because no
value has been demonstrated for the shut-in portion of
these graphs.
EXAMPLES
Example 1-SFE3 Mlnlfrac
The first test example is Mini-Frac No. 2 of GRI's Staged
Field Experiment NO.3 (SFE3).17 Downhole pressure data
were obtained through a 9001 ft dead string containing fluid
of specific gravity 1.077 with hydrostatic pressure calculated
to a depth to top of perforations of 9225 ft. Perforation
frictions, determined by comparing surface pressures
immediately before and after sudden changes in rate (e.g.,
shut-ins), gave an average perforation discharge coefficient
of 0.685 for the 72 0.330 in. perforations (which is very
comparable to the 52 open perforations with Cd=0.95
assumed in Ref. 17). A reported closure pressure (amiJ
value of 5250 psi was used. As shown in Fig. 2, in this
example, where total injection into the fracture lasted
approximately 371h min, the injection rate followed a
somewhat erratic upward trend for about the first 111h min
of the treatment until stabilizing at approximately 48.4
bbVmin. The treatment was performed with a 40 Ib/1000
gal linear CMHPG gel with a reported n value of 0.56.
190
SPE 23434
Figure 3 shows the Iog(Ap)-log(VJ plot of the data and Fig.
4 shows the Iog(Ap)-log(t) plot. It can be seen from the
graphs and Table 1 that the slopes on the injected volume
plot are somewhat more shallow than those on the time"
plot. As will be verified with the derivative plots, they are
smoother as well.
As discussed previously, after shut-in, the curve on the time
plot decreases gradually and that on the injected volume
plot becomes vertical. As also noted above, because no
diagnostic value has been placed on this portion of the
curve on a log-log plot, this presents no problem.
Figures 5 and 6 are the corresponding derivative plots. In
addition to the derivative curves, these graphs contain
horizontal lines indicating the maximum and minimum
slopes predicted by each of the three fracture geometries
(Eqs. 14 and 15). From the top down, these lines are (1)
high efficiency PK geometry, (2) low efficiency PK
geometry, (3) low efficiency radial geometry, (4) low
efficiency KZ geometry, and (5) high efficiency KZ and
radial geometries.
Examining the derivative curves and Table 1 shows that
there is less variation in the slopes on the injected volume
plot than on the time plot and that the slopes are, in most
cases, noticeably shallower.
Noting the relationship of the curves to the slopes predicted
by the three fracture models, they fall within the predicted
ranges of any of the models for only brief durations.
Although these lines were drawn using a given value of n,
using a different n value might increase the amount of data
falling within the range of a given model, but would still
leave much of the data outside that range. This implies
that actual fracture behavior falls, for the most part, outside
that assumed in devising any of these models.
Figure 7 is a graph of Iog(VldpJdVJ) vs.Iog(Vj). This figure
illustrates that in an instance such as this, where for even
short periods log-log slopes become very shallow or even
negative, this type of graph may appear very erratic and be
difficult to interpret. Its sensitivity to changes in slope also
serves to illustrate the earlier point that the log-log slope
must be very nearly constant for this type of graph to
clearly show growth trends.
Example 2
The second example uses data from a fracturing treatment
performed in the san Andres formation of west Texas.
Pressure was measured through a live annulus. As can be
seen in Fig. 8, excepting the very early portion of the job
SPE 23434 DON K. POULSEN 7

and subsequent minor fluctuations, the rate was held near C "" fluid-loss coefficient
12 bbVmin.
Cd "" perforation discharge coefficient
Figure 9 displays the 1og(L\p)-Iog(Vi) graph and Fig. 10 the
Iog(L\p)-log(t) graph. As can be seen from these two DH "" hydraulic diameter
graphs, slopes are more shallow on the volume plot, most
especially in the early portion of the job when the rate is E "" Young's modulus
changing most dramatically. Although not very obvious on
these graphs, it is slightly more noticeable on the slope E' "" plane strain modulus "" EI(111~
graphs of Figs. 11 and 12 that there is some moderation to
the slopes in the near-constant rate portion of the H = fracture height
treatment. The horizontal lines on these figures represent,
as in the Figs. 5 and 6, the maximum and minimum slopes K = power-law consistency index
predicted by the various models, but for n equal to 0.57.
L "" fracture half length
The moderation in slope is brought out even more clearly
in Table 2, which presents the results of a linear regression n "" power-law flow behavior index
on the 10g(L\p)-log(Vi) and 1og(L\p)-log(t) data for the time A

span from 2 to 10 min of injection, during which no unusual P "" wetted perimeter
pressure behavior was noted.
p "" pressure
CONCLUSIONS
Pw = pressure adjacent to the wellbore
1. Methods of plotting net fracturing pressure based
on injected volume have been shown to be less dependent L\p "" net fracturing pressure "" Pw - a min
on variations in injection rate than those based on injection
time. The effects of changing rate are not eliminated, but Q = injection rate
reduced, giving more correct slope values.
a .. average injection rate
2. Most guidelines previously presented in the
literature for interpreting injection time-based graphs may q "" flow rate
be used to interpret the injected volume-based graphs.
When injection rate is constant or very nearty so, the R = fracture radius
Iog(L\p) vs. log(t) plot is equivalent to a 10g(L\p) vs. Iog(VJ
plot, with corresponding equivalencies when t is substituted r "" radial distance from wellbore
for Vi in any of the other plots.
t "" injection time
3. More ideal than plots based on time or injected
volume would be plots based on fracture volume; however, VI = fracture volume
such plots cannot, at present, be accurately created
because current technology does not allow direct and Vi .. slurry volume injected
accurate determination of the fracture dimensions or of the
total rate at which fluid is lost to the formation. v • average velocity

4. An equation, more theoretically sound than that VI .. apparent fluid-loss velocity

used by Perkins and Kem13 or Nordgren,14 for calculating
pressure drop through a fracture of elliptic cross section has W "" fracture width at a given length or radius from the
been developed and is presented in Appendix C. wellbore

NOMENCLATURE W. =average fracture width at a given length or radius

from the wellbore
A • cross-sectional area
Wmu = maximum fracture width

191
8 FRACTURING NET PRESSURE ANALYSIS FOR VARIABlE RATE INJECTION
W ... average fracture width
x ... distance from wellbore
Et ... 1og(~)-1og(t) slope
£y ... Iog(~)-Iog(VJ or Iog(~)-log(Vf) slope
11 ... fluid efficiency
A ... dimensionless distance along fracture length
~ = Poisson's ratio
G min ... least principle stress
ACKNOWLEDGEMENTS
I thank Halliburton Services for the opportunity to present
this paper.
REFERENCES
1. Nolte, K. G. and Smith, M. B.: "Interpretation of
Fracturing Pressures," JPT(Sept. 1981) 1767-75.
2. Conway, M.W., et a/.: "Prediction of Formation
Response From Fracture Pressure Behavior,"
paper SPE 14263 presented at the 1985 SPE
Annual Technical Conference and Exhibition, Las
Vegas, Sept. 22-25.
3. Nolte, K. G.: "Fracturing-Pressure Analysis," Recent
Advances in Hydrau/ic Fracturing, J. Gidley et a/.
(eds.), Monograph series, SPE, Richardson, TX
(1989), Chap. 14.
4. Nolte, K. G.: "Discussion of Influence of Geologic
Discontinuities on Hydraulic Fracture Propagation,"
JPT(Feb. 1987) 998.
5. Martins, J. P. et a/.: "Tip SCreenout Fracturing
Applied to the Ravenspurn South Gas Field
Development," paper SPE 19766 presented at the
1989 SPE Annual Technical Conference and
Exhibition, San Antonio, Oct. 8-11.
6. Poulsen, D. K.: "Net Pressure Fracture Design,"
paper CIMISPE 90-42 presented at the Petroleum
Society of CIMISPE International Technical
Meeting, Calgary, June 10-13, 1990.
192
SPE 23434
7. Nolte, K. G.: "Fracturing-Pressure Analysis for
Nonideal Behavior," JPT (Feb. 1991) 210-218.
8. Ayoub, J. A., Brown, J. E., Barree, R. D., and
Elphick, J.: "Diagnosis and Evaluation of Fracturing
Treatments," paper SPE 20581 presented at the
1990 SPE Annual Technical Conference and
Exhibition, New Orleans, Sept. 23-26.
9. Khrlstlanovlc, S. A. and Zheltov, Y. P.: "Formation
of Vertical Fractures by Means of Highly Viscous
Liquid," Proc., Fourth World Pet. Cong. (1955) II,
579-86.
10. Geertsma, J. and de Klerk, F.: "A Rapid Method of
Predicting Width and Extent of Hydraulically
Induced Fractures," JPT(Dec. 1969) 1571-1581.
11. Daneshy, A. A.: "On the Design of Vertical
Hydraulic Fractures," JPT (Jan. 1973) 83-93.
12. Sneddon, I. N.: "The Distribution of Stress in the
Neighborhood of a Crack in an Elastic Solid," Proc.,
Roy. Soc. (1946) A, 187, 229.
13. Perkins, T. K. and Kern,l. R.: "Widths of Hydraulic
Fractures," JPT(Sept. 1961) 937-949.
14. Nordgren, R. P.: "Propagation of a Vertical
Hydraulic Fracture," Soc. Pet. Eng. J. (Aug. 1972)
306-314; Trans., AIME, 253.
15. Nolte, K. G.: "Determination of Proppant and Fluid
SChedules From Fracturing-Pressure Decline,"
SPEPE (July 1986) 255-265.
16. Nolte, K. G.: "Fracturing Pressure Analysis:
Deviations From Ideal Assumptions," paper SPE
20704 presented at the 1990 SPE Annual
Technical Conference and Exhibition, New Orleans,
Sept. 23-26.
17. "Staged Field Experiment No.3: Application of
Advanced Technologies in Tight Gas Sandstones-
Travis Peak and Cotton Valley Formation, Waskom
Field, Harrison County, Texas Reservoirs," Report
No. GRI-91/0048 (Feb. 1991).
18. Cleary, M. P.: "Comprehensive Design Formulae
for Hydraulic Fracturing," paper SPE 9259
presented at the 1980 SPE Annual Technical
Conference and Exhibition, Dallas, Sept. 21-24.
SPE 23434 DON K. POULSEN 9

19. Meyer, B. R.: "Design Formulae for 2-D and 3-D

Vertical Hydraulic Fractures: Model Comparison
and Parametric Studies," paper SPE 15240
presented at the 1986 SPE Unconventional Gas (A-7)
Recovery Symposium, Louisville, May 18-21.

APPENDIX A

Pressure Response from Idealized Fractures Substituting into Eqs. A-1 and A-2 and rearranging gives

KZ- and PK-Type Fractures Ap. -fO tip. xK(4)>(O)n_L _ f, n

[gL(1.0l cA
(A-8)
For laminar flow of a power-law fluid between parallel 6p H W:;' 0(W:)2n+1
plates, it is known that

where
_El!. = J4n+2)n-KJ.!l.)n (A-1)
ax \ n wz,,+1lH
(A-9)
For an elliptical fracture with an eccentricity of 1, Appendix
C shows that

_El!. = (.!.)2n+1(3n+1)n-KJ!l..)n (A-2)

ax 1t 4n~1lH By considering Eq. A-8, the width equation (Eq. 4), and Eq.
3, it can be shown that
Somewhat similarly ~o what Nordgren14 presented, the
continuity equation for incompressible flow in a fracture can
be expressed as

(A-3) for KZ geometry, and

which, when reformulated in integral terms, becomes Ap = [...,.K4,.f,..("'fr(~)]..:.. Yo":" (A-11)

aw)
a at
q(1.0 = q(0.0 -HL v,+_a cA
o
(A-4)
for PK geometry where
7f.H3n+S '

n
where f. Irf} = f' [gL(1.0l cA (A-12)
KZ\ 2n+1
x
1 =- (A-S) 0(1-12)-2-
L
and
Applying a no-flow boundary condition at the fracture tip,
gives
(A-13)

o = q(1.0
naW)
= q(O.O-HL v,+_a
o at
cA (A-6)

Radial Fractures
Taking the ratio of q(A.,t) (Eq. A-4) to q(O,t) (Eq. A-6) gives
For radial flow, the pressure drop due to the flow of a

193
10 FRACTURING NET PRESSURE ANALYSIS FOR VARIABLE RATE INJECTION SPE 23434

power-law fluid between parallel plates is given by

_ W~ = K(4n+2--!l-)n (A-14)
2 ar n 1EWZr
where
The continuity equation for incompressible flow in a fracture
can be expressed as
(A-22)

an
.:::..3. + 21Etv, +
aw .. 0
21Er-- (A-15)
ar at
APPENDIX B
which, when reformulated this in integral terms, becomes
Injected Volume-Pressure Relationship for High Fluid
Loss

It has been shown 10,14 that under conditions of high fluid

loss and constant injection rate
where
(B-1)
r
A. = - (A-17)
R
for KZ and PK geometries, and
Applying a no-flow boundary condition at the fracture tip,
=.!~ Q 2t
gives 2
R (B-2)
1E 0

for radial geometry. Solving Eqs. B-1 and B-2 for t and
substituting into Eq. 8, which says that, for a constant
Taking the ratio of q{A.,t) (Eq. A-16) to q{O,t) as determined injection rate the injected volume is the product of the
from Eq. A-18 gives injection rate and the pumping time, gives

(B-3)

Considering the fracture width relationship of Eq. 4 and

Eqs. A-8 for KZ and PK geometry and Eq. A-20 for radial
Substituting into Eq. A-14 and rearranging gives geometry, it can be easily shown that
o
6.p = - [dp (B-4)
lop (A-20)
2K(4n+2 ~(t»)n R1-n [1 [gRCA..t)]n dA.
i
z

n 1E W:;1 0 A. ';:)2n+1
or

By considering the width equation for radial geometry (Eq.

4) and Eq. 3, It can be shown that

194
SPE 23434 DON K. POULSEN 11

__n_
HV, 2(n+1)
_El!. '" 32(4n+2)n---.!S..-J.!l..)n (C-7)
ax 3n n ~lH
S Sn+2 _,_
I1p. Q 4(n+1) H - 2(n+1) v,4(n+1)
(8-5)
and which was obtained by modifying the equation for
n .--!!!- power-law flow between parallel plates (Eq. A-1) with a
Q I(n+1) V, 1(n+1)
correction factor for Newtonian flow through an ellipse of
near-zero eccentricity (H-W). As well as being used in
most implementations of PK-type geometry, Eq. C-7 has,
APPENDIX C with the incorporation of additional factors, been used in
other, subsequently developed models.18•19 Nordgren14
Pressure Drop for Flow Through an Elliptical Fracture used the Newtonian form of Eq. C-7.

It is well known that flow of a power-law fluid through a pipe

can be modeled by the equation

The average velocity for flow through an elliptical fracture

is

V"'.!"'~
A nWH

To generalize Eq. C-1 for other than perfectly circular

condUits, it is necessary to substitute the hydraulic
diameter, DH, for D. Hydraulic diameter is defined as

(C-S)

For an elliptical fracture with an eccentricity approaching 1

(H>W),

and thus

DH •
i nWH
4
2H
) '" ~W
2
(C-5)

Substituting Eqs. C-2 and C-5 into Eq. C-1 gives

_El!. '" (!)2n+1(Sn+1)n---.!S..-J.!)n (C-6)

ax n 4n ~+1~H

Compare this to the equation used by Perkins and Kem,13

which may be written as

195
 Table 1
Linear Regression Results for Example 1

Time Interval Average Slope Coefficient of

(min) Determination
Ap VS. VI Ap vs. t Ap vs. VI Ap vs. t

3.1 - 11.5 0.054 0.080 0.9699 0.9696

11.5 - 37.4 0.146 0.187 0.9898 0.9792

Table 2
Linear Regression Results for Example 2
Time Interval Average Slope Coefficient of
(min) Determination
Ap vs. VI Ap vs. t Ap vs. VI Ap vs. t

2.0 - 10.0 0.080 0.093 0.8958 0.8760

196
 ~ -=~I--------""jI;P=====: ......:---11
o

-
Q.
S0) --_/
c """"
:t~
...0
...o~

-o cJ
Q

o I
I I I I I I I I II I I II I I I I I I
1
10° 10 102
10g(Vt) or 109(11t) t, min
Fig. 1 - Net pressure behavior for arrested Fig. 2 - Injection rate vs. log(time), Example 1
growth in one wing

~ 104 ::)------------------
I

"iii 3
0.. lO-
ci
<l

1d-\ I I i I I I III I I I I i 1III I I I i I I III

~ ~ ~ 1d
V.,I bbl
Fig. 3 - Log-log plot of net pressure vs. injected
volume, Example 1
 4
10 r------
I I

U1
c:i

.[ 103 - >
ci.
iN 0 I \\ r/ 'v-"": \CIlIlV
'f v ' ' ' 1111 I
<]
lf1
o
J ::::::·::::::::\~2LD[~~~;;~~:2~J7-7:::::::::::::::·:::l::::.:::-:_::
\ ,.,,"oc, ~y
:-:_::_._-
-------------'"''
I HIGH EFFIcIENCY KZ AND RADIAl

1d I I I I I I I III
1
I I j I I I III II 1 ,11111111 I IIIIII11 1111111"
10° 10 1cf 2 3 4
10 10 10 10
t, min
V.,I bbl
Fig. 4 - Log-log plot of net pressure vs.
injection time, Example 1 Fig. 5 - Injected volume log-log slope plot, Example 1

U1
c:i

iN... 0 I \ II v VlJY, • I 11/ I

::::::.:::.. . \. J::-:_:::-:------ ----
~J ~::~:~:::,~37="E:=,=
I HIGH EFFICIENCY KZ AND RADIAl

II 11111"" 111111111
2
10° 1d 10
t, min
Fig. 6 - Injection time log-log slope plot,
, Example 1
 3
10
=Jr------ I ~--,

~
I

>0-
\ c
$2

~
°E co

%
~ 10
>0-
2

'"
...Q
...0(0
a
""""
N

101 ~ I I I I I I f II I I I I I I III I I I j I I I I I I 0-1 - I I I I I I III I I I I I iii I I I I \I I I Ii I

~ ~ ~ ~
1 1
10 10° 10
VI bbl
o '

t, min
Fig. 7 - Log-log plot of VI dPw/dVI vs. injected volume,
Example 1 Fig. 8 - Injection rate vs. log(time), Example 2
CD
CD 4
10 =l I

OR 103
ci:.
<I

102 II I I I I I I I II I I I I f I I II I I I IIi I I
I
~ ~ ~ ~
VJ bbl
o'

Fig. 9 - Log-log plot of net pressure vs. injected

volume, Example 2
 4
10 3"""1- - - - - - I N I .1 II.
l{)
~-

~-

l{) HIGH EFFICIENCY PI(

.~ 103 > ci :=:=:=:=:=~~=~~~~:~;~~:=:=._._:=:=:=:=:_.=:

ci
<1
'V 0
l{)
o-_-=::-_-_-=::-_~-:=~-""'===::====::::::====:====-:====:====-:===:-:====
1---------- " "lDw EFFICIENCY PJIoIAL
ci l
, I
lDw EFFICIENCY KZ
HIGH EFFICIENCY KZ AND PJIoIAL

,-
LI1

1cf I I' I I I I I II I I I , I Iii I I I I I I I I II NI I I I I I I I I II I I 1 I I II II I II I I I I I I I

~ ~ ~ rl 10° 10
1
102 103
t, min Y.,I bbl
Fig. 10 - Log-log plot of net pressure vs.
Fig. 11 - Injected volume log-log slope plot, Example 2
injection time, Example 2
g
N I I, L I
~

ci ~ ._._._:=:=~~~~~~~~~~=:=: .
'0- 0 ::-_-=::__-_-=:=:.-_-~-~=;;:::===:-:===:-:::- :===:-:====-:====:===:-:=:
l{) l----------~"
,,~ EFFICIENCY PJIoIAL
ci LDw EFFICIENCY KZ

,-
I HIGH EFFIcIENCY KZ AND PJIoIAL

,-
LI1

NI II 1111\1111 111111111 11111111 I

1 1 2
10- 10° 10 10
t, min
Fig. 12 - Injection time log-log slope plot, Example 2