Beruflich Dokumente
Kultur Dokumente
SPE 14379
Fracture Stimulation Design and Evaluation
by P. A. Warembourg,* Dowe// Schlumberger, E. A. Klingensmith, F/opetro/ Johnston Schlumberger,
I K
w. k.
untima.
, , ~,
.1.
J. ,
ITnwdl
-.. ,,
Sphltfmharnar
, ,, ,,,- r~.., ,,
and
-., -
.1
-.
F
-.
Frrik
-. -.-,
Flnnetrnl
..vr - -------
Inhnstcm
. . -----
~~hi~rnberger
Member SPE
ThiQ
. ... namw
r--- was nrm-y~~~ for n~fisantation
.. -.-r r --- a! the LMh AMUM Technical Conference and Exhibition of the So@ety of Petroleum Engineers held in Lea
Vegas, NV September 22-25, 1985.
This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the
author(s). Contents of the paper, as praeertted, have not been reviewed by tha Sodety of Petroleum Engineers and are subject to correction by the
author(s). The material, as presentad, does not neceeeerily reflect any position of the Seciefy of Petroleum Engineera, ita officers, or members. Papem
presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Permieeien to copy k
reatricfed to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgment of where
and by whom the paper is presented. Write Publications Manager, SPE, P.O. Sox 833836, Richardson, TX 750S3-3836. Telex, 730989 SPEDAL,
The Industry spends millions of dollars each Optimization is possible with proper identifi-
year for fracture stimulation treatments. The cation and utilization of accurate critical vari-
majority of these treatments cannot optimize produc- ables in the design process.
tion because realistic values for the critical vari-
ables used in the treatment design process have not
been properly identified. Critical variables are CRITICAL VARIABLES
defined as those variables that have the largest
impact on the production obtained from any stimula- The critical variables needed to optimize stim-
tion treatment. Obtaining an accurate value for ulation results can be grouped into three cate-
gories. Each category contains variables that are
needed to determine specific design criteria. The
three categories are (1) reservoir and producing
system variables that determine production response
RQfWfMCgS and illustrations at end of oaDer=
~r from the well; (2) stimulation design variables that
------------ .... . . . . . . ------ . . .
2 I-KACWRE STIMULAr IUN D IGN AND EVALUATION SPE 14379
determine achievable fracture geometry; and (3) fects of these variables are determined by using
economic variables that determine the optimum treat- available models to describe fluid flow through
ment. The specific variables are listed below. perforations, tubulars, flowlines, chokes, separa-
tors and artificial lift systems.23Q Combining
- mesenuol
D--..-..-*-
. ..A nA.._4.._ #?. . ..4. ~~e au! Iauc
~,fi~~~ ~~Gre plulllullly
....-63.-... ..1 , .. ..l.a-.. ,.s=,..+.
ellcul,s
.w.f~~ ~~e
w am rJ-wucmny ayau.um
creating the desired fracture geometry. (For exam- pret, especially in determining fracture growth
ple: if 100,000 gal (378,500 L) of fracturing fluid below the perforations. Temperature surveys should
was used in a treatment and at the end of the treat- have several runs with sufficient time between runs
ment only 40,000 gal (151,400 L) remained in the to pick up the subtle changes necessary for accurate
fracture, then the fluid efficiency would be 40%. interpretation. Results from tracer and temperature
The other 60,000 gal (227,100 L) of fracturing fluid surveys assume that all of the fracture height is ifl
was lost through leakoff.) Fluid efficiency 1s contact wtn the wellbore.
dependent on fracturing fluid properties, injection
rate, reservoir characteristics, area and any fluid- The last variable dealing with stimulation
10SS additives used. The term that is probably most design is the fracture geometr{ mo$el. Basically,
often used to describe leakoff, and is more specific there are two design models131 wi h many modifica-
to the fluid, is fluid-loss coefficient. The tions to handle fluid loss and other variables. The
larger the value of the fluid-loss coefficient, the models are also well documented in the literature,
less efficient the fracturing fluid system will be, but generally for an identical set of conditions the
i.e., a fluid-loss coefficient of 1.5 E-3 is more Khristianovich and Zheltov model (KZ) calculates
efficient than a coefficient of 3 E-3. Fluid-loss less fracture length than the Perkins and Kern model
coefficients are usually obtained from laboratory (PK). Choosing the correct model for designing an
testing on cores or other permeable media. Fluid- optimum fracture stimulation treatment is very ire-.
10SS coefficients obtained in the laboratory will portant. Using accurate input data (critical vari-
usually be a different number from the one actually ables) with previous history and evaluation allow
representative of the reservoir. Laboratory tests verification or modifications that are applicable to
simply cannot account for the total leakoff charac- a specific formation in a specific location or
teristics in the reservoir, such as natural frac- field. Figure 6 illustrates the sensitivity of
tures, faults, joints or reservoir permeability propped fracture length and fracturing fluid volume
variations. The fluid-loss coefficient is a criti- to geometry models.
cal variable and should be the total fluid-loss
coefficient for design purposes. Figure 5 illus- Accurate identification of critical variables
trates the relationship between fluid loss, fracture allows a realistic economic optimization of the
length and volume of fracturing fluid required. fracture stimulation, plumbing system and operating
condition design. The economic optimization is the
Obtaining Variables culmination of the totai design process. Economic
optimization compares the revenue potential to the
Closure pressure can be determined from an cost of the stimulation treatment, plumbing system
instantaneous shut-in pressure by assuming that the and operating condition options.
closure is equal to bottom-hole fracture pressure.
This can also be obtained from a step-rate test.l Production system modeling techniques provide a
The step-rate test data are obtained by pumping a production forecast that accounts for the effect
thin, ungelled fluid (such as water) at matrix con- each design option will have on the total system
ditions. The rate is gradually increased in steps production potentiai. As the design options change
until the formation fractures and the pressure sta- (i.e., fracture length, fracture conductivity, arti-
bilizes. ficial lift methods, etc.), the total system produc-
tion potential will also change. Production fore-
Monitoring pressure decline (caused by fluid casts are synonymous with revenue potential, if the
leakoff) after injection can be used to determine net revenue per hydrocarbon unit is known. The
the minimum stress in the formation. A pump-in present value of the revenue forecast can be calcu-
flowback procedure can also be used in formations lated provided a revenue discount factor is deter-
where leakoff is extremely low and pressure decline mined. It is a simple process to calculate net
would require too much time. Using the flowback present value after establishing the cost of each
procedure, the well is opened up through a special stimulation treatment, plumbing system and operating
valve and the pressure decline is plotted. condition design option (Net Present Value = Dis-
counted Revenue - Investment). The economic optimum
The total fluid-loss coefficient and fluid design for the total system is the one that provides
efficiency are obtained in situ by performing a very the largest net present value. Evaluating the net
small injection test using the same fracturing fluid present value of each design option uses very simple
and injection rate proposed for the actual propped economic engineering practices and the end result is
fracture treatment. This small treatment does not the most cost-effective design.
use any propping agent because proppant can inter-
fere with the pressure decline and affect the calcu-
lation for the fluid-loss coefficient. Three other variables that affect the economic
optimum design are (1) the duration of the produc-
The volume of fluid required for this test is tion forecast from which net present value is calcu-
usually determined from previous experience. Injec- lated, (2) the net discounted production revenue,
tion should continue until pressure data indicate and (3) the amount of investment required to achieve
that gross fracture height is stable. It is recom- the design option. Figures 7, 8 and 9 show the
mended that the fluid used in the injection treat- effect each of these variables has on optimization.
ment contain radioactive tracer material other than
sand : Fract~rQ heiaht l~gs are also a valuable
.._.=.._ - - tool
for gross fracture height prediction.11 The optimization process assumes the design
parameters will be achieved. There is always a
Gross fracture height can be determined b possibility that the desired design will not be
running a ganwna ray survey. Temperature surveysl3 achieved; however, better values for the critical
can also be used but are more difficult to inter- variables minimize the risk.
4 FRACTURE STIMULATION D IGN AND EVALUATION SPE 14379
M3HOD FOR CPTINIM FRACTURE STIMULATION DESIGN Step 4. Pump and Monitor the Desired Fracture Stim-
ulation Treatment
Step 1. Forecast Production for Various Fracture
Stimulation, Pl~bing System and Operating Condition Procedure
Designs
. Use a treatment monitor vehicle (on location) to
procedure assure the job is pumped as designed (ail data
stored for future reference and analysis).
Design, conduct and interpret a pressure tran-
sient test to determine critical reservoir vari- . Monitor and plot real-time fracture growth data.
ables (transmissibility, mobility, permeability,
reservoir pressure). Step 5. Evaluate the Fracture stimulation Results
Perfoming as Forecast
Step 2. Detensine Achievable Fracture Stimulation
Iksion
--- .=.. htiinnc
~- -----
Dr.nt-ad,,wfi
, , ..=, .=
Step 3. Economically l@timize Fracture Stimulation Use various models and simulators to construct a
Treatment Lksign new fracture geometry model, if necessary.
e Use the production system forecast to determine Three case histories are presented to demon-
cumulative production vs time for each of the strate same of the key points presented in this
tentative fracture stimulation designs. paper. Case No. 1 illustrates a situation that
frequently occurs when all of the critical variables
o Calculate the present value of the production used in the fracture design are not accurately de-
using net revenue per hydrocarbon unit and an termined. Case No. 2 illustrates a fracture stimu-
appropriate discount factor. lation treatment that follows the procedure pre-
viously listed in Methad for Optimum Fracture Stim-
Determine the investment required to achieve each ulation Design. Case No. 3 shows the results of
of the realistic fracture stimulation design using the data from Case No. 2 on an offset well.
options.
Case No. 1
e Calculate the net present value for each of the
design options (Net Discounted Revenue - Invest- A pressure transient test was performed to
ment). determine how the well would respond to fracture
stimulation. Production system modeling techniques
The economic, optimum fracture, stimulation were used to provide the production forecast shown
treatment design has the largest net present in Figure 10. Table 1 lists the reservoir and pro-
value. ducing system data used in the forecast.
A fracture half-length of 1,000 ft (304 m) was Using the production forecast data and the new
chosen, based on the data provided in the fore- fracture geometry, the economic optimum fracture
cast. A fracture stimulation treatment was then half-length was determined to be 500 ft (152 m).
designed for this 1,000 ft (304 m). However, the Figure 16 illustrates fracture half-length vs net
values for several critical variables (gross frac- present value.
ture height, fluid efficiency, closure pressure)
necessary for realistic stimulation design were not The fracture stimulation treatment, designed
accurately determined. A gross fracture height of for 500 ft (152 m) of fracture half-length and a
60 ft (18 m) was assumed for the stimulation de- conductivity of 485 md-ft (145 ~m2-m), was performed
sign. Figure 11 shows the sensitivity of fracture and monitored in real time. The proposed treatment
half-length and stimulation job cost to gross frac- schedule vs the actual is illustrated in Figure
ture height for this well. The effect of increasing 17. The treatment design schedule was adjusted
the assumed gross fracture height by 20 ft (6 m) is during the actual treatment because the calculated
also shown. bottom-hole treating pressure plot indicated a po-
tential screenout at the tip of the fracture. How-
The treatment was designed to use 52,000 gal ever, flush was completed and no proppant was left
(196,800 L) of fracturing fluid to place approxi- in the tubulars. Using calculated bottom-hole frac-
mately 100,000 lb (45,360 kg) of 20/40-mesh sand. ture pressure to make these decisions vs actual .
Estimated time to pump the treatment was 28 min. bottom-hole fracture pressure (from a static tubing
The treatment did not go to completion and had to be string or a bottom-hole gauge), is debatable.
terminated after only 14 min of pumping because of a
screenout. Figure 12 shows this treatment as de- Modeling the treatment as actually pumped gave
signed and as pumped. a calculated fracture conductivity of 300 md-ft (90
um2-m). This conductivity was used to construct a
Approximately 19,000 lb (8,620 kg) of proppant new production forecast. Actual production rates
was in the fracture at the time of screenout. The did not match the forecast production. A post-frac-
cause of this screenout is difficult to evaluate ture pressure transient test was conducted to deter-
without knowing gross fracture height and in-situ mine the effective fracture half-length and conduc-
fluid loss. A post-fracture stimulation pressure tivity. A restriction was found in the tubing when
transient test was conducted to determine the effec- the electric line gauge was run. No restrictions
tive fracture half-length and conductivity. The were encountered during prefracture electric line
production forecast was updated to reflect the post- runs. This pressure transient test was successfully
fracture half-length, conductivity and the operating conducted, however, and the calculated effective
conditions of the well. Post-fracture production fracture half-length was 490 ft (149 m) with a con-
matches the updated production forecast. Figure 13 ductivity of 346 md-ft (104 ~m2-m). Figure 18 shows
shows the updated discounted revenue forecast in the actual production from this well vs the original
comparison to the original forecast and revenue forecast (500 ft Xf, 485 md-ft). Also shown is the
potential lost after one year from not creating the forecast with the restriction. The operator will
originally desired fracture length. remove this restriction in the near future.
Production system modeling was used to deter- Reservoir and treatment design data obtained
mine how the well would respond to fracture stimula- from Case No. 2 were used to design a fracture stim-
tion. Reservoir variables were obtained from pres- ulation treatment for an offsetting well. The frac-
sure transient testing and analysis. Figure 14 ture stimulation treatment design on the offsetting
shows cumulative production vs fracture half-length well was identical (i.e., 500 ft (152 m) and 485 md-
and time. Table 2 lists the variables and data used ft (145 um2-m)). The treatment was successfully
in the forecast. Fracture half-lengths of 300 to pumped to completion. Figure 19 illustrates the
1,000 ft (92 to 304 m) provided approximately the prefracture production forecast vs the actual post-
same cumulative production after one year. The fracture production for the offset well.
fracture half-length contemplated by the operator
prior to this production system modeling was 2,100
ft (640 m) (80% of the drainage radius). CONCLUSIONS
Step rate tests, a small nonpropped fracture . A successful fracture stimulation treatment
treatment, and surveys were performed. These tests should be defined as one that provides the pro-
determined closure pressure, fluid-loss coefficient duction predicted by the design process.
(fluid efficiency) and gross fracture height. The
originally estimated gross fracture height, prior to o The critical variables interact, and accuracy is
testing, was 50 ft (15 m). The actual gross frac- extremely important for production forecasting,
ture height was 358 ft (109 m), created by pumping stimulation design and economic optimization.
crosslinked elled water at an injection rate of 15
~p~ (~.~g ~! /mj~)e The fp~~.~~~ nruna+ P mrwial c * ?~~ ..
met ~0 rletenn<ne... . . . .
--. rn~~~ ~CCgT~~~ y~~~~S for
YGwrgl--wJ --vu-v-
were rerun using the new variables obtained from the critical variables is usually a very small
testing and surveys. Figure 15 shows the relation- percent of the total well cost. Early optimiza-
ship between fracture half-length, proppant volume tion will prevent repetitive errors from occur-
and two gross fracture heights (50 and 358 ft (15 ring in the designs for future wells.
and 109 m)). Fracture geometr simulations indicate
Gross Height = 50 ft
that if the original design t . The economic optimum can be determined prior to
(15 m)) would have been pumped, a screenout would any expenditure for a fracture stimulation treat-
6 FRACTURE STIMULATION 01 IGN AND EVALUATION SPE 14379
The success of a fracture stimulation treatment the 1984 SPE Deep Drilling and Production Sym-
can be quantified in exact terms, regardless of posium, Amarillo, April 1-3.
the volume of fluid and proppant actually pumped.
6. Holditch, S. A.: Criteria of Propping Agent
Selection, prepared for the Norton Company,
Much of the technology used in optimizing frac- Dallas (Jan. 1979). .
ture stimulation is not new. Using all of the
tools available, in a logical engineering method, 7. Elbel, J. L.: Considerations for Optimum Frac-
will decrease the risk associated with anv stimu- ture Geometry D~h&gn, paper SPE/DOE
lation treatment and the production resul~s. 13866presented at 1985 SPE/DOE Joint
Symposium on Low-Permeability Reservoirs,
-1-1.. R______ ..... ,A ma
e Ille 11-.,-..-4----
cape!-lellceS.-A--
Iact.ur.,
II Wl[nnun[y
--..-1 ..--4 c--
Useu rur de- waver, may LY-LL.
signing fracture stimulation treatments for fu-
ture wells, will be accelerated and enhanced. 8. Veatch, R. W. Jr.: Current Hydraulic Fractur-
ing Treatment and Design Technology, paper SPE
10039 presented at the 1982 SPE International
ACKNOULE06MENT Petroleum Exhibition and Technical Symposium,
Beijing, China, March 18-26.
The authors express their ameciation to the
management of Dowell Schlumberger and Flopetrol 9. Agarwal, R. G., Carter, R. D., and Pollock,
Johnston Schlumberger for permission to publish this C. B.: Evaluation and Performance of Low-Per-
paper. meability Gas Wells Stimulated by Massive Hy-
draulic Fracturing, J. Pet. Tech. (March 1979)
362-372.
REFERENCES
10. Nolte, K. G.: Fracture Design Considerations
1. Erdle, J. C.: Current Drillstem Testing Prac- based on Pressure Analysis, paper SPE 10911
tices: Design, Conduct and Interpretation, presented at the 1982 SPE Cotton Valley Sympo-
paper SPE 13182 presented at the 1984 SPE sium, Tyler$ May$ and 1983 SPE Production Oper-
Annual Technical Conference and Exhibition, ations Symposium, Oklahoma City, Feb.
Houston, Sept. 16-19.
11. Newberry, B. M., Nelson, R. F., and Ahmed, U.:
2. Meng, H. Z., Proano, E. A., Buhidma, I. M., and Prediction of Vertical Hydraulic Fracture
Mach, J. M.: Production Systems Analysis of Migration using Ccinpressional and Shear Wave
Vertically Fractured Wells, paper SPE/DOE Slowness, paper SPE/DOE 13895 presented at the
10842 presented at the 1982 SPE/DOE Unconven- 1985 SPE/DOE Joint Symposium on Low-Permeabil-
tional Gas Recovery Symposium, Pittsburgh, May ity Reservoirs, Denver, May 19-22.
6-18.
12. Dobkins, T. A.: Improved Methods to Determine
3. Brown? K. B., and Beggs, H. D.: The Technology Hydraulic Fracture Height, J. Pet. Tech.
of Artificial Lift Methods, Volumes 1-4, (April 1981) 719-726.
PennWell Books, Tulsa, Oklahoma (1977).
13. Perkins, T. K. and Kern, L. R.: Widths of
4. McLeod, H.: The Effect of Perforating Condi- Hydraulic Fractures, J. Pet. Tech. (Sept.
tions on Well Performance, J. Pet. Teci. (Jan. 1961) 937-949; Trans., AIME, 222.
1983) 31-41.
140 Khristianovich. S. A. and Zheltov. Y. P.: For-
5* Montgomery, C. T. and Steanson. L E.: ProD- mation of Vertical Fractures by M~ans of-Highly
pant-Selection -- The Key to ~uccessful Fra~- Viscous Liquid, Proceedings of the 4th World
ture Stimulation, paper SPE 12616 presented at Pet. Congr., Rome (1955).
cp x O.001* = Pa-s
(Oc s &)~~.~ ~ o~
\,
ft x 0.3048* = m
gal x 3.785 412 E-03 = m3
in. x 2.54* = cm
lbm x 0.453 5924 = kg
10-3PM2
md X 0.9869 =
2
urn-m
md-ft/cp x 0.308 E-4 =
Pans
*Conversion is exact.
TABLE 1
*
CASE NO. 1
hn = 13 ft (3.9 m)
TABLE 2
CASE ~. 2
Kohn
. 64.7 md-ft/cp (0.0195 ~m2-m/Pa.s)
o
Pi = 6,100 psia (42,058 kPa)
hn = 26 ft (7.9 m)
B. = 1.55 RVB/STB
I
hn = 10 ft
J. Jo=0.41 Cp
k =5md
~o= 1.7 RVB/STB
kfw = 1,000 mdft
#
~~. ------------------------------
Me
I
..
.._..--. ..
------------------------ : 1k =Imd
t
L 1 1 I I I I 1 I I I
00 200 400 600 800 1,000
FRACTURE HALFLENGTH (Xf,ft)
Figure 1. Example of the effect matrix permeability and fracture conductivity have on
cumulative production (Pi = 3,464 psig, P = 0.4 Cp, B= 1.7 RVB/STB, hn = 10
ft, pumped-off operating conditions).
37,000 -
P. = 3,464 psia
hn = 10 ft
po= 0.41 Cp
A = 1.7 RVB/STB
34,000 -
kfw = 1,000 md-ft
31,000 -
28,000 -
25,0001
22,000
40 nnn
I 2,UUU
t
[
I 1 1 I 1 1 1 1 I
16,000~
200 400 600 800 1
FRACTURE HALF-LENGTH, (Xf, ft)
Figure 2. Example of the effect operating conditions and the plumbing system configu-
ration have on cumulative production.
20/40 Ottowa Sand
O. Z-in. Average Fracture Width
I -. ,. . . . ... . .. . . 1
t mcum
gr/ 7 7 7 Y Y l-.
2,000
1,000 [
/////A;( 0
0
4
%,7s0
v
/;i/>>>>Y
1/////////
J /.,,.4,
17,500
/
PROPPANT
./
VOLUME
J, ,44
175,000
(lb)
1.750.000
d
I
Figure 6.
TREATMENT VOLUME _
F
-- Bauxite
--- Intermediate Strength
Precured ResinCoated Sand
-- ResinCoated Sand
Otbwo Sand
1 ,ooo~ ,
t
.-
IL -J
.=---e.--s=.-=. -
-_-_
I I I
D
\___ -- -- 8 yr
- 4 yr
..-. -----
L
lyr
100 - i
-
I
!
v 1
,o~
FRACTURE HALFLENGTH (f
*
I Figure 7. The effect time has on the detertn{na-
tion of the optimum economic stimula-
6,000 8,000 10.000 [psi) tion design.
120 160 200 240 280 i%+
CLOSURE (d Temperature)
al
I Revenue
or MSCF
per BBL
Ecanamic Optimum I
<
>: =q
+C
z:
I
Wa
ul~
w
/
I
E= ,=-r----
rL
+ 0 I k.,
+C /
W3 I
ZCJ I \ 1
:+<, , ~ / / Ecanernic
Optimum _
I
I Lawer Net Operating
n I
Revenue per SBL
/ I
I or MSCF
500 1,000 1,500 2,000 2,500 3,000 3
FRACTURE HALF-LENGTH
FRACTURE LENGTH (X,,ft)
Figure 8. The effect net discounted revenue has
Figure 5. The relationship between fluid loss, on the determination of the optimum
fracture half-length and fracturing economic stimulation design.
fluid volume.
----------------------- Scheduled Prappont
- Scheduled Fluid
Actual Pracmant Pummsd
------------- -------Actual fluidPumped
100,000) P
vl
Q
/
~
75,000 -
Lb-
~w
Oz
f ~ ~ 50,000 -
20
<>
t n
n Screenout ------- ---=
25,000 -
o
w
, I
a
I
FRACTURE HALFLENGTH (ft) 15 20 25 :
WW
250~
I
7501
:!
-e-c.
:[ /
v / ~z...-
_=--- -----$-9;0:000
Revenue Potential
I I i I I 1
!30 100 150 200 250 300 350 4 )
TIME (days)
)
Figure 13. Case No. 1: Canwarison of discounted
TIME (days)
I revenue potential.
300,000
Gross Frocture Height
/80
s
- 200,000 - z/6C
1-
~ - Scheduled Treotment
v
g 100,000 . s 130,000 /
7
WE 1437,9
1,8C
o _ Design
0 000 Actual
:- 1,20
x
w
z
3
-1
0
>
+
z
<
L 60
R
c)
x
1
/ 50 ft
0,/~
o 600 1,200 1,800 2,4C
FRACTURE HALFLENGTH (ft)
Figure 15. Case No. 2: The relationship between PUMPING TIME (rein)
1,500[
I Forecast
Actual
I ! ---
rlgure ii% Case No. 2: Actuai production vs fore-
I
~ .5 I 1 I cast: with and MCthQUt the ra<trirtinn.
. -- -. . - , ,. .
o 500 1,20J3 1,800 2,4I
FRACTURE HALFLENGTH (ft)
SPE 14379
Production Wellhead
Pressure = 1,000 psig
a
Production
I
Curtoiied
1 1
1 f)n 9Arl -ten A
, z .-r. d.. T
TIME (days)
WE ]4379