You are on page 1of 5

Considerations for Optimum

Fracture Geometry Design


....L. Elbel, SPE, Dowell Schlumberger

Summary. This paper gives production forecasts for (1) various equal-proppant-volume fracture geometries with various
formation permeabilities, (2) equal fracture lengths with different proppant volumes, and (3) equal fracture lengths and proppant
volumes with various proppant distributions. It shows conditions when Prats' relationship for maximum productivity is true and
when higher values of dimensionless fracture conductivity are beneficial. These should give the design engineer useful tools when
optimizing fracture design.

Introduction
Hydraulic fracturing has been successful in increasing the productive simulator can model the performance of a well containing a fracture
reserves in many oil and gas reservoirs. As wells are completed of constant width with different conductivities in various fracture-
in formations of lower permeability, larger fracturing treatments length segments. This would be the same as constant proppant per-
are required for economical recovery. Therefore, the fracturing de- meability with different fracture widths when non-Darcy effects are
sign engineer wants to design the optimum fracturing geometry. ignored.
Prats 1 addressed this problem by showing that for a given fracture Three situations were simulated. The first situation was for a con-
volume (proppant volume), there is a fracture-width-to-fracture- stant or predetermined fracture volume with five different fracture
length relationship for achieving maximum productivity. He showed lengths in four reservoirs of different permeabilities. The second
that this relationship was 2Lf !b= 1. 59k/k. Prats' findings are valid was for a constant or predetermined fracture length but with different
after long producing times or for cases of high formation permea- proppant volumes. The third was for a situation having a constant
bility. In formations with low permeability, Morse and Von Gon- fracture length and propped volume but with the propped width vary-
ten 2 showed that early transient production rates before a ing over the length. The first two situations were to test the validi-
pseudosteady-statecondition is reached can be important in deter- ty of the guidelines mentioned previously; the last was to examine
mining the economics of hydraulic fracturing treatments. the differences between the constant-propped-width constraint of
More recently, Cinco-Ley et al. 3 and Agarwal et ai. 4 published the type curves and the varying widths, which may be more common
dimensionless type curves that show the relationship of production in practice.
rates for various fracture conductivities with time. These curves The term "optimization" in this paper is based solely on maxi-
show that for a constant fracture length, higher .fracture conduc- mum production and is not meant to take the place of the proper
tivities are beneficial and a rule of thumb of CfD = 10 to 30 has economic optimization that would consider well spacing, formation
been proposed as the desired dimensionless fracture conductivi- thickness, porosity, production, and treatment cost.
ties. 5 ,6 The CfD is defined as
Equal-Fracture-Volume, Varying-Length Cases. Table 1 gives
kfb the reservoir properties used for the various simulations. It can be
CfD = - (1) noted that the properties may seem unrealistic in their combination
kLf of dimensions. The thickness was varied to maintain a constant for-
mation kh product, which allowed all the combinations to have the
Prats' optimum fracture-width-to-fracture-Iength relationship for same proppant volume and the same combination of CfD and frac-
maximum productivity corresponds to a dimensionless fracture con- ture penetration. This resulted in the use of some unusually low
ductivity, CfD , of 1.26. This shows that for a given volume of fracture conductivities for the simulations; however, because the
proppant, the optimum fracture conductivity, kfb, should equal study was performed to evaluate CfD , a dimensionless term, the
1.26 times the product of fracture length, Lf , and formation per- conclusions should not be affected. A 9OO-acre [364-ha] spacing
meability, k, or kfb= 1.26kLf . This relationship is true for cumula- was used to accommodate the 2,898-ft [883-m] fracture penetra-
tive production after long periods of time (10 to 20 years) and for tion. Note that the formation porosity is decreased as the permea-
the cases studied by Prats. In low-permeability formations, how- bility is decreased, and although this will affect the transient time,
ever, higher values of CfD will result in more production during it was done to give more realistic conditions. Table 2 shows the
the early-time periods. combination of CfD and Lf that will be obtained for equal proppant
In the fracturing of deeper formations with potential high closure volumes for the cases in Table 1. These combinations will always
stress on the proppants, more expensive proppants are often re- be the same in this constant-volume study because the kh is the same
quired to maintain fracture conductivity. In these cases, the prop- for all cases.
pant cost becomes a higher proportion of the treatment cost, and Fig. 1 shows the cumulative production for Reservoir A having
proppant volume has some limiting effect on treatment sizing. When a permeability of 1 md with the five different combinations of frac-
fluid costs are relatively high, fluid volume or fracture length is ture length and CfD. The same conclusion as given by Prats-that
the limiting parameter. for a constant-volume fracture, CfD = 1.26 is the optimum-can
This paper evaluates two conditions: equal-length and equal- be made. Of interest is the order of optimization of the combinations
volume fractures. The design engineer may design for a certain frac- of CfD and Lf in Table 3. .
ture length and try to obtain CfD = 10 +. This would require a cer- Fig. 2 shows the effect of early-time transient behavior for Reser-
tain fracture volume that, according to Prats' study, would be better voir B having a formation permeability of 0.1 md. It shows that
when distributed over a greater length, resulting in CfD = 1.26. for the first 180 days, the order of CfD for maximum productivity
To evaluate the guidelines using CfD in fracture-destgn optimi- is 5, 3, 10, 1.26, and 0.6. After 500 days, the order is 3, 5, 1.26,
zation, production was simulated for various conditions with a finite- 10, and 0.6. It is interesting that the 2,898-ft [883-m] fracture having
difference, two-dimensional, single-phase, reservoir simulator. This a CfD of 0.6 is better than the 71O-ft [216-m] fracture having a
CfD of 10. After about 15 years, the order is the same as with For-
Copyright 1988 Society of Petroleum Engineers mation A and agrees with Prats' conclusion.

SPE Production Engineering, August 1988 323


TABLE 1-PROPERTIES USED IN SIMULATION TABLE 3-0RDER OF OPTIMIZATION OF
COMBINATIONS OF C,o AND L, FOR RESERVOIR A
Reservoir pressure, psi 5,000
Producing pressure, psi 2,000 L,
Reservoir temperature, of 200
0.6
..J!!L
Gas specific gravity (air = 1) 1.26 2,000
Spacing square, acres 900 3.0 1,296
Formations 5 1,000
0.6 2,898
A B C D 10 710
Permeability, md 1.0 0.1 0.01 0.001
Porosity 0.2 0.15 0.05 0.03
Thickness, ft 1 10 100 1,000
2,400
2,200
2,000
1,800
TABLE 2-CONSTANT PROPPANTVOLUME
iL 1,600
~ 1,400
Fracture Length
~ 1,200
(ft)
Cl~ 1,000
2,898 0.6 800
2,000 1.26 600
1,296 3 400
1,000 5.0 200
710 10.0 o 0~---:~--:-:-:-::-~=----=-~---::-=:--~::---:-:=-~4,ooo

1,ooo,...-------------------, Fig. 2-Cumulatlve production for constant-proppant-volume


900 Formation B, k=O.1 md.
800
700
t5 600 3,000 ...--------------""'7'~-------.,
2,800
::;
~ 500 2,600
~ 400 C to Lt 2,400
Cl 2.200
300 --- 10 710
- 1.26 2,000 2,000
200 --
_.-
0.6
5
2,898
1,000
t5 1.800
::; 1,600
100 .....- 3 1,296
~ 1,400
~ 1.200 C to L,
500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 Cl 1.000
TIME (days) 800
600
400
Fig. 1-Cumulatlve productIon for constant-proppant·volume 200
Formation A, k=1.0 md. o 0~---,~-~:::--.,..,...J:=----=-~-=-'===--~:-::---;:-':::::-----:-!4,000

Fig. 3 shows that for Reservoir C with a permeability of 0.01


md, the case with CfD = 10 is optimum for about 100 days even Fig. 3-Cumulatlve production for constant·proppant-volume
though the differences with CfD's of 3 and 5 are negligible. After FormatIon C, k=O.01 md.
about 700 days, the combination with CfD =3 shows the best pro-
duction. The combination for the CfD = 1.26 overtakes the one with
CfD = 10 after about 850 days. Also, the case with CfD = 1.26 and cumulative for that combination, however, will not be the best until
fracture length of 2,000 ft [610 m] is better than the l,ooo-ft [305-m] about 18 years.
fracture with CfD =5 after about 10 years. The 2,898-ft [883-m] Examination of Figs. 1 through 4 shows that the case with
fracture with CfD =0.6 overtakes the 71o-ft [215-m] fracture after CfD=3 is optimum or near optimum for the permeabilities exam-
about 4,000 days. ined over a 1O-year period. This differs from Prats' optimum of
It should be mentioned that for all cases the first few days of pro- 1.26. Prats stated that his optimum for the constant-volume case
duction were always higher with CfD = 10 and decreased in de- may not be accurate for cases where the fracture penetrates more
scending order of CfD' This follows previous observations 3•7 that than 50% of the drainage radius. A CfD=3 should be considered
the fracture conductivity near the wellbore determines the initial as near optimum when the proppant volume has been predetermined
production rates and pressure response and that the fracture length as a result of high proppant cost, and other parameters are known
does not affect the initial production. to some degree of accuracy.
Fig. 4 shows the cumulative production for a formation with a
permeability of 0.001 md. In this case, the combination with C/D = Constant-Fracture-Length Case. For a constant fracture length,
10 has the highest cumulative production for the first 450 days. the CfD can be optimized by determining the dimensionless-time
After that time, the fracture with CfD =5 has the highest cumulative range for the period of time of interest and then comparing with
for 20 years. the Agarwal et al. or Cinco-Ley et at. type curves to estimate the
Figs. 5 through 8 show the production rates of the various for- relative effect of various values of CfD' This method has led to
mations resulting from the various combinations of fracture length the rule of thumb of designing for CfD = 10+. Fig. 9 shows the
and CfD. Note that the optimum rates of the combinations change difference in cumulative production of Formation C for a fracture
at earlier times than for cumulative production because of the long length of 2,000 ft [610 m] with CfD values of 0.6, 1.26, 5, and
transient times. Fig. 6 shows the production rates for the forma- 10. Fig. 10 shows the production rates for the four cases. The pre-
tion with a permeability of 0.1 md and the various fracture-length ferred choice, CfD = 10, seems obvious, but the extra cost in ob-
combinations. After about 2,200 days, the combination with taining CfD = 10 (eight times the amount of proppant necessary to
CfD = 1.26 and Lr2,000 ft [610 m] will have the highest rate. The obtain CfD = 1.26) should be considered.

324 SPE Production Engineering, August 1988


5,000 .----------------,;--~~----, 4,000 ..----------------C-,o-----,
4,500
4,000
3,500

5'::;: 3,000 C
u..
::
2,500
~:~ -- 3
......... 5

()
~ 2,500 ~ 2,000
c5 2,000 CT 1,500
1,500
1,000
1,000
500 500
Ol.-_-'--_ _'---_-'--_ _'---_-'-_ _.L.-_---'-_---'
500 1,000 1,500 2,000 2,500 3,000 4,000 o 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000
TIME (days) TIME (days)

Fig. 4-Cumulatlve production for constant-proppant-volume Fig. 8-Productlon rate for constant-proppant-volume Forma-
Formation D, k=0.001 md. tion D, k=0.001 md.

800
Cm
700 - - 10
- 1.26
600 - - 0.6
-5
.._-- 3
C 500
u..
()
(J) 400
~
CT 300

200

100

0
0 4,000
o0""'"~~~~1.J,O~OO~~~~2,..J.OO~0~~~~3,...L00~0~~~~4....J,OOO
TIME (days)
Fig. 5-Productlon rate for constant-proppant-vOlume Forma-
tion A, k= 1.0 md. fig. 9-Cumulatlve production for constant fracture length
with different C'D'S, Formation C, k=0.01 md.

1,600 . . . - - - - - - - - - - - - - - - - : - - - - _ _ _ _ ,
C,o
1,400 - - 10 6,000 r-;-
.. - - - - - - - - - - - - - - - - - - - - ,
-5
1,200 . t.
,1\, -- 3 - . - C ID =10 L, =2,000
- - 0.6
......... 1.26 ------- C ID =5 L, =2,000
c 1,000 '\\ .......... C ID =1.26 L, -2,000
t5
(J) 800 f-
~~
~.~
- - C ID =0.6 L,=2,000

; 600 f- "'~.:::::.. ~
400 f--
----.:..._------ ~---

200 f-
o L-_...LI_~IL-_ _'__I_.....lI__-.L.-I-...LI__---JI'_:_::_--'
o 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000
TIME (days)

O .........~~~ .........~~~~~~~~~"-':-~~~~
Fig. 6-Productlon rate for constant-proppant-volume Forma- o 1,000 2,000 3,000 4,000
tion B, k=0.1 md. TIME(days)

Fig. 10-Productlon rate for constant fracture length with


different C'D'S, Formation C, k=0.01 md.
Cm
- - 0.6
-10 Constant Fracture Length and Volume With Varying Proppant
-- 5
-- 3 Distribution. When some CfD is chosen as optimum, we can con-
.......- 126
sider one more criterion, varying fracture conductivity. 7,8 Bennett
~ 1,200
() et ai. 7 showed that the fracture conductivity near the wellbore will
~ 1,000 determine the initial production rate and that at a later time the pro-
-; 800 duction will be determined by the arithmetic average of the varying
600 fracture conductivities. The proppant volume for a constant fracture
400 length is directly proportional to the value of CfD' and it can also
200
be shown that it is directly proportional to the average CfD .
Ol.-_-'--_ _l.-_-'--_---JL-_--'-_..........l_ _-'-_-' Fig. 11 shows the cumulative production for five cases of varying
o 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 fracture conductivities for Formation C that result in an average
TIME (days) CfD of 5. Therefore, these are cases with equal proppant volume
and equal length. The fracture conductivity for each quarter seg-
Fig. 7-Production rate for constant-proppant-volume Forma-
tion C, k=0.01 md. ment of fracture penetration is shown. The conductivities corre-
spond to the proppant volume in that segment. This figure shows

SPE Production Engineering, August 1988 325


o
tr
::::- 10.---------------------.,

W
l-
e{
II::
2,000 ...J
e{
U
o
II::
a..
---- 83 65 39 13 u
C!)Q. 1,000 w
............. 100 50 37 13 II::
-- 50505050 VI 10. 1
-.- 87 87 13 13 VI
W
---- 162 13 13 13 ...J
Z
o
o 0......,~~~~1,J..00~0~~~~2w,0~00~~~~3~,OLO~0~~~4.w,000 Vi
z
TIME (days) w
:::::l:
o
Fig. 11-Cumulative production, constant fracture length,
variable fracture conductivity.
Zone 1 Zone 2 Zone 3 Zone 4

LID, =0.25 L lD2 =0.25 L I03 =O.25 L lD4 =0.25


that the constant-width fracture shown here as having a fracture
conductivity of 50 md-ft [15 md'm] does not result in the maximum e fO , C'02 C ID3 C ID4 (;,0
.................... CASE 2 10.00 5.00 3.75 1.25 5.00
production. It also shows that for this case, varying the conductivity
. _ . - CASE 5 15.00 2.50 1.50 1.00 5.00
would have only minimum effect, except for the case having a con-
ductivity of 162 md-ft [49 md'm] in the segment adjacent to the
wellbore. Fig. 11 also shows that for a given proppant volume, Fig. 12-Effect of varying fracture conductivity (after Bennett
varying the conductivity monotonically away from the wellbore can et al. 7 ).
be beneficial.
Fig. 12 (Fig. 2 in Ref. 7) shows the effect of varying conductivity
for two cases having an average CfD of 5. This figure is a log-log permeability, rectangular drainage would be more appropriate. 9
plot in dimensionless terms of l/qD and tD' It shows that Case 5 If the fracture azimuth 13 is known, the wells can be spaced for rec-
has a higher production over the first log cycle of time and that tangular drainage for more efficient recovery. 14
Case 2 has higher production over the third and fourth cycles. Be- Because the observations in this study depend on correct estimates
cause the time period for the third cycle is 100 times longer than of the formation permeability, the conclusions may not be useful
the time period for the first cycle, it is obvious that the cumulative if this value is not known. As engineers, we are charged to use
production would be greater for Case 2. The significance of the the best methods with the best available information. It is hoped
difference can be estimated by relating the time to the corresponding th~t this study will help the engineer determine when more time
value of tD' If 1 day represents a tD of 10 - 5 , the first log cycle should be spent for better optimization of fracture treatments:
will be 9 days, the second 90 days, etc. If 1 day represents a tD
of 10 -2, the difference between Cases 2 and 5 will probably be Conclusions
insignificant, and varying the proppant concentration will be of little The study was made to evaluate guidelines used in the industry,
consequence. The importance of the accuracy of the parameters basing a fracture design on the dimensionless term CfD for maxi-
affecting the calculation of tD should be noted. Also, as Bennett mum productivity. The following observations have been made for
et al. stated, the fracture conductivity should decrease monotonically cases where the cost of the treatment is not considered.
from the wellbore to the fracture tip. 1. Prats' optimum fracture-length-to-width ratio corresponding
to CfD = 1.26 for specified proppant-volume treatment is valid in
Discussion formations having a high permeability (k> 1.0 md) or when con-
In designing a hydraulic fracture treatment, the engineer attempts sidering cumulative production long after.a pseudosteady period
to determine the fracture length and conductivity for optimum has been reached.
results. These decisions are best determined by simulation and ec- 2. In low-permeability formations (k<O.Ol md), the long period
onomic studies that consider the early transient behavior of pro- of time in transient production makes higher values of CfD more
duction, as 'well as spacing. 9,10 However, this study can give the economical.
following guidelines. If a predetermined proppant volume has been 3. The simulations in this study over the permeability range of
chosen, using a minimum fluid volume and therefore m.inimum 0.1 to 0.001 md show that CfD =3 should be considered when the
length and maximum CfD may not result in maximum production. proppant volume has been predetermined.
If the formation permeability is in the range of 1 md or greater, 4. Varying the proppant distribution for a given proppant volume
CfD = 1.26 would be optimum; if the permeability is less than 0.1 and fracture length can affect the production rates, depending on
md, CfD =3 or greater would probably be warranted. If a fracture the tD range and CfD value.
length has been predetermined, higher values of CfD are warrant-
ed, depending on the dimensionless time. Nomenclature
The optimization process must consider over which time the max- b = propped fracture width, ft [m)
imum productivity is desired, on the basis of present worth or cash CfD = dimensionless fracture conductivity
flow. It should also consider the expected life of the fracture effec- Gp = cumulative gas produced
tive_ness. There is evidence that fracture conductivity decreases with h = formation height, ft [m]
time, but tests for longer than 1 year have not been made. Other k = formation permeability, md
considerations include being able to obtain the correct CfD because
kf = proppant permeability, md
of errors in proppant-permeability measurement, formation-
permeability measurement, proppant placement, and other factors Lf = fracture length, wellbore to tip, ft [m)
that affect the fracture conductivity. 11,12 Consider the consequence q = production rate, surface volume rate, STB
of a design for CfD = 1.26 but obtaining CfD =0.6 as a result of [stock-tank m 3 ]
an error in proppant permeability or width by 50% (Fig. 6). qD = dimensionless production rate
The simulations in this study were performed with a square t = producing time, days
drainage, whereas for the case of long fractures and low formation tD = dimensionless time

326 SPE Production Engineering, August 1988


Reference. mum Hydraulic Fracture Design," paper SPE 16435 presented at the
1987 SPEIDOE Symposium on Low Permeability Reservoirs, Denver,
1. Prats, M.: "Effect of Vertical Fractures on Reservoir Behavior-Incom-
May 18-19.
pressible Fluid Case," SPEJ (June 1%1) 105-18; Trans., AIME, 222.
11. Montgomery, C.T. and Steanson, R.E.: "Proppant Seleetion-The Key
2. Morse, R.A. and Von Gonten, W.D.: "Productivity ofVertica\ly Frac-
to Successful Fracture Stimulation," lPT (Dec. 1985) 2163-72.
tured Wells Prior to Stabilized Flow," paper SPE 3631 presented at
12. Cooke, C.E. Jr.: "Effect of Fracturing Fluids on Fracture Conduc-
the 1971 SPE Annual Meeting, New Orleans, Oct. 3-6.
tivity," lPT (Oct. 1975) 1273-82.
3. Cinco-Ley, H., Samaniego-V., F., and Dominguez, N.: "Transient
13. Smith, M.B. et at.: "The Azimuth of Deep Penetrating Fractures in
Pressure Behavior for a Well with a Finite-Conductivity Vertical Frac-
the Wattenberg Field," lPT (Feb. 1978) 185-93.
ture," SPEJ (Aug. 1978) 253-64; Trans., AIME, 265.
14. Elhel, J.L.: "Designing Hydraulic Fractures for Efficient Reserve
4. Agarwal, R.G., Carter, R.D., and Pollock, C.B.. : "Evaluation and
Performance of Low-Permeability Gas Wells Stimulated by Massive Recovery," paper SPE 15231 presented at the 1986 SPE Unconven-
Hydraulic Fracturing," lPT(March 1979) 362-72; Trans., AIME, 267. tional Gas Technology Symposium, Louisville, KY, May 18-21.
5. Holditch, S.A.: "Criteria of Propping Agent Selection," Norton Co.,
Dallas (Jan. 1979). 81 Metric Conver.lon Factors
6. Veatch, R.W. Jr. : "Overview of Current Hydraulic Fracturing Treat- acres x 4.046 873 E+03
ment and Design Technology," lPT (April 1983) 677-87. ft x 3.048* E-Ol
7. Bennett, C.O. et al.: "Influence of Fracture Heterogeneity and Wing ft3 x 2.831 685 E-02
Length on the Response of Vertically Fractured Wells," SPEJ (April OF (OF-32)/1.8
1983) 219-30.
psi x 6.894 757 E+OO
8. Rosato, N.D.: "The Transient Behavior of Finite Conductivity and In-
fInite Conductivity Vertical Fractures," MS thesis, U. of Tulsa, Tulsa,
OK (1980). .
9. Holditch, S.A. et al.: "The Optimization of Well Spacing and Fracture -Conversion Iactor Is exact. SPEPE
Length in Low Permeability Gas Reservoirs," paper SPE 7496 presented
at the 1978 SPE Annual Technical Conference and Exhibition, Houston,
Oct. 1-3. Original SPE manuscript r-wed lor review May 19, 1985. Paper accepted lor publicallon
March 27, 1987. Revised manuscript received 8ept. 10, 1987. Paper (SPE 13866) first
10. Meng, H.Z. and Brown, K.E.: "Coupling of Production Forecasting, ~ at the 1985 SPE SymposIum on Low PermeabIlity Gas ~ held In Denver,
Fracture Geometry Requirements, and Treatment Scheduling in Opti- May 19-22.

SPE Production Engineering, August 1988 327

You might also like