.

SPE
SPE 15240

Design Formulae for 2-D and 3-D Vertical Hydraulic Fractures:

Model Comparison and Parametric Studies
by B.R. Meyer, Meyer & Assocs. Inc.
SPE Member

Copyr!ghl 1986, Soc!ely of PeMo!eum Engineers

This paper was prepared for presentation al the Unconvenlcmal Gas Technology Symposium of the Soclely of Petroleum Engmeefs held m Loulswlle,
KY, May 18-21.1986.

This paper was selected for presenlaoon by an SPE Program Commillee follow!ng rewew of mformahon conlained m an abslracl submd~ed by !he
aulhor(e). Conlenls of lhe paper, as presented. have not been rewewed by the Soc!e!y of Petroleum Engineers and are subject 10 Correcllon by the
author(s). The malenal, as presen[ed. does no! necessarily reflecl any position of Ihe Soclely of Pelroleum Engineers, its olhcers. or members Papers
presented al SPE meehngs am sublecl !0 publ!calion rewew by Edmx!al Commmees of Ihe Soc!ely of Petroleum Engineers Perm!ss[on 10 copy IS
reslrtcled to an abs!racl of not more Ihan 300 words. Illusfrauons may no! be copied. The abs!rac! should confam conspicuous acknowledgment of where
and by whom lhepaperispfesenled Wrde Publ,cal[ons Manager. SPE, P O. Box 833836, Richardson, TX 75083-3836. Telex, 730989, SPEDAL.

ABSTRACT and method of

sclution, Some of these simulators
include a comprehensive treatment of fracture
This paper presents asymtotic anaIvtical and containment and proppant transport, while others
numerical solutions (2-D) and
for two-dimensional only calculate fracture propagation characteristics.
three-dinens.ional (3-D) type hydraulic fracture Thus, the user of hydraulic fracturing simulators
geometries. The fracture propagation models shouId have a good understanding of the as=umpi~~ns,
investigated include: a Geertsma-Deklerk (GDK), a limitations and applicability af the code in
F’erl:ins-~ern/Nordgrer IPKN! and a 3-D type model. question, An overview of current hydraulic
fracturing design and treatment technology is given
Comprehensive 2-D and 3-D design formulae for by Veatchl.s. t’eatc3’s paper include; an e::terisiv=
power-law fracturing fluids are given for various list of other authors who have contributed tc 2-D
cases of fluid loss and containment. These formulae and 3-D hydraulic fracttirina simulati,~r, te,:hnnlogv.
can be used to benchmark 2-D and 3-D hydraulic
fracturing simulators. Characteristics of After code faml)iarization, one often specu!ate: as
equilibrium 3-D height growth are also discussed. to the accuracy of the numerics! results. One ~?,
Farametrlc studies based on the design formulae are to help ensure numerical ~imulatlorl ~c~Llracy 1s t(:
performed to show the effect of any one variable, benchmarl, the code against i nown aralytlcal c.
theoretical solutions.
Typical fracture designs are simulated for various
containment and leak-off examples using each of the Simplified design formulae are presented icJ-
fracture propagation models, along with a comparison Irenchmarking GDV, Pt’N and :-r tl’p? h.dr~~i:c
of proppant scheduling and design. The simulaticrr fractures for asymtotic Solutinns of no leal.-of+ :’,6
~ttidies identifv and ‘illustrate the basic large leak-off. Limiting 3-D model +ot-m:.!!a.?s-t
characteristics and design differences of each also presented for well contairied. e~uili hritum ~-,.;
model . penny shape type fractures,

INTRIIDUCTION Cifter code benchmarking, the ne:t phase 15 t~

determine which fracture model is most apglic?.a:?
A number of ?-D and 3-Cf hydraulic fracturing for a given design based on in-s~tu formti!lafi
simulators have been developed that account for many properties and treatment de51qn cofisideratlo;-,=.
~f the complex phenomena affecting the fracturing This also includes determining the effect criti ::!
proces5. Some of the 2-D models are based on the parameters have on fracture characterlstl:=.
GD!I type fracture aeometryl~~ while others are based Numerically simulating
parametric effects i5 u;,uaily
on a PW type fr~cture]~$, A comparison of the very time consuming and computer intense. The m=jor
various 2-D theories and underlying assumptions is parameters affecting fracture characteristics can he
given by Geertsma and Haafkens$. identified bv an analytical parametric stud; t:~ed
on the design formulae given below,
Similarlv, a number of variable height (3-D) models
have been developed~-!t. These models vary A comparison of hydraulic fracturlna Sijlulatiofl
considerably in complexity, underlying assumptions designs for GDK,PKN and 3-D mode15 := presented
illustrating the importance of using the proper
fracture model based on treatment design and
References and illustrations at end of paper. formation properties, Comparisons include fracture

I
 Design Formulae for 2-D and 3-D Vertical Hydraulic Fractures:
2 Hodel Comparison and Parametric Studies SPE 1S240

propagation characteristics and proppant schedulin9

t
design.
a(t) = Vf t)/ J q(7) d~ ,., (2)
s
The simulator used in this study (HFRfJC-11) is a
= Vf t)/qt ; q = cone.tan
t?,ree-dimensi onal hydraulic fracturing simulator
w~th options for both the GDK and PKN type 2-D
The GDK, PKN ano 3-D (penny) fracture lengths
fracture propagation models, This siaulator includes
(radius) for a constant inject on rate and no
3 .:omprshensive treatment of the major coupled
leak-off (a = ) from Eqn. (1) are
+jr~t-order parameters affecting the propagation and
;r~ppant transport processes. The ●ajor fracture
.. i
fluid mechanics phenomena include (1) ~ L(t) = qt/[7v(t Ww(t) Hwl . . . (3)
ur,s{,z,meirlcal confining stress and moduli contrast,
!.-, acljscent layer stre55 gradients, (3) rock Pm L(t) = qticrv(t Ifw(t) H“] . . . (4)
cl=ft:l:lstlon, (4) time dependent flow of power-law
qt 1/2
;:~ids with
fran:por-t
le:k-off
with
c.ro.i,.,:tlor,
and
settling
simulation.
spurt
and
A detailed
lciss, (S) proppant
closure, and (6)
description of
P~ R(t) ,
[ 2 rv(t Ww(t) 1 ,,. (5)

~~,p =ssumptlons, capabilities and solution The fluid loss volume for large leak-off (Vi(t) + qt
lJct!liti.Jlcqv for this simulator are given in or q -I 0) with a constan”t leak-off coefficient and
tie;r~errce [141. no spurt loss is given by

t 1
AJfl~Yll&AL/NUt’lERICAL BENCHHARK FORHULIIE 1
Vi(t)= 2C ~fttft — dt dt
, J~ -~z
Bc:lcfifi!irl.. formulae are presented for GDK, PKN and IJo
7.-0 type fracture
geometries. The governing
e2uatlori5 of mas5 conservation, continuity, . s c A tllz . . . (6)
~ici$~,-opening pressure, momentum and constituitive
relationships for the fracture propagation and where -a s 1/2 (for m = 0).
prnppant transport models are based on the
methodology of Meyer (1’?85)1~. The resultinq width The total leak-off area (one face) for the 2-D and
and !ength equations for no fluid loss and large penny shape type fractures is given by
+luid leak-off are simplified frora the more
comprehensive formulae, 2&lJ: A = Hp L(t) ,.. (7)

S]nce the fracture is assumed to have two symmetric P~ 0 = # R(t)2/2; R < Hp/Z . . . (8)
~ings, only one wing will be analyzed.
A = Hp R(t) \ R )) Hp/2 .,. (9)
ttASS CONSERVATION
where leak-off occurs only in the PaY zone (Hp).
The governing mass conservation equation for an
incompressible slurry in a fracture requires that FRACTURE PROPAGATION
the total volume of slurry injected minus the volume
Df slurry in the fracture (Vf) and the volume of Analytical fracture propagation formulae are
fluid loss to the formation by leak-off (VI) and presented for GDK, PKN and 3-D type geometries.
spurt 10SS (V5P) must be equal to zero. Therefore: These formulae are based on a~ymtotic behavior of
more comprehensive solutions for the limiting cases
of no fluid loss (t = 1) and large leak-off (t + 0).
~q(,)d, - Vf (t) - Vi(t) - Vgp(t) = o . . . (1)
o
Analytic 3-D formulae are given for large and no
where: confining stresses as asymtotit limits for model
validation. The 3-D model is also benchmarked
Vf (t) = 7v(t) Ww(t) Hw L(t) ; GDK
against equilibrium (static) fracture growth
= I’v(t) Uw(t) Hw L(t) ; PKN solutions for large confining stress.

= 2 Pv(t) Ww(t) R(t)z ; ‘enny

GDK Analytical Formulae

tA C(t),t)
The Geertsisa-Deklerk type geometry ●odel is a
Vi(t) = 2
JJ 8 0 lt - 7(A) ‘Adt
constant
plane strain.
perpendicular to
height
The
the
fracture which assumes
cross
horizontal

direction
sectional
of
area
fracture
Vsp(t) = 2 sp A(t) propagation is rectangular in sh.=.o ““I a constant
nidth (vertically unbo, ~eometrv) .
llma
7(A) - t [A/rJ(t)l Characteristics include a pressure
profile with time. This model I ~able for
The fracture voluae Vf is reported only for the fractures with length to half cios less
analytical propagation solutions sought. The time than one (L/H < 1).
dependent leak-off area and propagation parameter
are given by A(t) and @a, respectively. The shape of the fracture, except very near the tip,
is aleiost elliptical and represented by
Fracture efficiency defined as the ratio of fracture
volume to the total volume of slurry injected i W(t,t) = klw(t) (1-1~)11: . . . (10)
SPE 15240 Bruce R. Ileyer 3

values reported by Geertsma et al.$ were 0,48 and

where W([,t) is the width at any position t (/ =
0,68, respectively,
x/L(t)) and I$w(t) is the maximum wellbore width
(wW(t) = w(o,t)).
Substituting an apparent viscosity

The GDK width-opening pressure relationship is;$ bq n’-l

Z7w@(l - v)
Iiapp =
‘a [ -w 1 ... (19)
Ww(t) = L(t)hP(O,t) . . . (11)
G
into the Newtonian version of Eqn. (12) and
The maximum GDK wellbore width for power-law type rearranging also yieIds Eqns, (i3) and (14),
fluids ba5ed on the methodology of lleyeri~ is Therefore, use of an apparent viscosity is
justifiable for the GDK model since the gamma
coefficients are rheology independent.
4 bn’yw,
Ww(t) =
7+ 7P Large Fluid Leak-off
[ “a +[~1”’L(t)21 ::+:,,)

where q is the injection rate for a single ning, For large leak-off (Vl + qt or q + O), the fracture
length in terms of Q from Eqns, (l), (6) and (7) is

No Fluid Loss 1/2

Qt
L(t) = — . . . (20)
The resulting width and length equations obtained by 2mCHp
substituting Eqn. (3) into (121 and rearranging for
the case of no leak-off (rI = 1) are for a two-wing fracture, This is the same result a5
given in References [11 and [51.

(1-u ,o.n’+~2(n!+2) Substituting Eqn. (20) into ( 2) and rearranging

1
i/(n’+ 2)
Ww(’ t yields the 6DK wellbore width foraula for large
leak-off
..! (13)
and :tn’+:
Ww(t) = a Ka(Q/HwJn ‘(Q/CHp t:n”+ 2
[% 1’1
L(t . b t n- . . . (21)
(1 G
- U)ka { % “+ ‘l’(;’+ ‘) “+ ‘
[ J
. . . (14) i/(zn’+ 2)
where a = [ 3n’7w,/(7p7+ *2)1 and the
where gamma coefficients are time and rheology independent
1

a=
[
3n’7wo/(7f 7 p 7;)
1
2(27’+ 2)
as given by Eqn. (16).

PKN (Mzalvtical Formulae

1

7p/(~6. ‘ 7W0
b=[yf
12n 72n’+
“ *)]2(n’+ 2) The Perkins-Kern/Nordgren type geometry model is a
constant height fracture with an elliptical shaped
2n’+ 1 “
Ka= K’
[-—1 3n , . . . [15)
width vertically
Characteristics
(vertically
include an
bounded
increasing
geometry).
pressure
G = EI[2 I+U)] profile with time. This model is most applicable
for fractures with length to half height ratios
Q=2q; (injection rate two wings) greater than one (L/H > 1).
Analytically and numerically the values for the
gamma coefficients The fracture shape is elliptical in the (ertlc.{1
were found to be approximately
plane for constant formation properties
7 > 0.795 , 7P 2 0.7046
wo 2 1(2
14(x,c,t) = If(x,o,t) (1 - G ) ,.. (22}
. . . ( 6)
7V = U14 7+
3 = 1.0 where W(x,L,t) is the fracture width at any pfislti~n
‘Z (L = z/H) and W(x,O,t) is the maximum Jr
The width equation for a Newtonian fluid (n’= 1) s centerline width at any position x,

I The PKN width-ooenina ., ore5sure relationship is’t

‘w(t) = a[ %“ {%1’1’” ““ c s o ’17)
rw(l - v)
where the proportionality constant a is 1.325 for a W():,o,t) = Hw OP(x.t) . . . (23)
two-wing and 1.874 for a one-wing fracture. The G
values reported by Geertsma et a~.$ were 1.32 and The major difference between the GDK and PKN madels
1.87, respectively, is the width-opening pressure relationship, for the
GDK model Ww * L(t)flP whereas, in the PKN model Ww “
Similarly the length is HWOP, The difference between the GDK vertical
ilk constant width and PKN elliptical width profile does
G
L(t) =b (I
2/3 not control model behavior. If this were the case,
t . . . !18)
- “), { +}” both modeIs would have the same characteristics with
[
] only slight differences in the width and length
where the proportionality constant b is 0.4804 for a proportionality con5tants.
two-sided and 0,6793 for a one-5ided fracture. The
—
 Design Formulae for 2-D and 3-D Vertical Hydraulic Fractures:
‘.,
4 ., /’4 ti!Jdel Comparisprr and Parametric Studies SPE 15240

The ,naiinum FKN wellbore width for power-law type 1/5

l-v Q2

1
fluids isi~ 1/5
UW(t) z a yII t. . .. (301
i ~
: bn’pwo
[
Ww(t) =
rf rp where the proportionality constant a is 1,918 for a
[ ‘a +l~ln”w’’t’l ~n+~(24)
tuo-wing and 2.530 for a one-wing fracture, The
#here q IS the single wing injection rate, nuaerical values of Nordgren~ were 1.89 and 2,50,
respectively.

No Fluid Loss
Similarly the length is
:te rcstiltinq width and length equations obtained by 115
G Q)

1
~;~~=:.ittiting Eqn. (4) into (24) and rearranging for VI J
L(t) = b t .,. (31)
-f~ 1~~l.-off (a = 1) are (1 - V)ll T;
[

where the proportionality constant b is 0.4426 for a

two-sided and 0,6709 for a one-sided fracture, The
values reported by Nordgren~ were 0.448 and 0,6S0,
. . . (25)
respectively,
, r,.2
Ilaking an apparent viscosity substitution into the
Newtonian version of Eqn. (24) does not yield the
same non-Newtoniarl proportionality constants.
Therefore, unlike t~~ Gil’ model, an apparent
where viscosity substitution for the PKN ❑odel is not
1
tn’+ 3 justifiable because of the flow behavior index
?.
[‘n ‘rwoi(rf I’p rv)
1 .,. (27) dependent gamma coefficients.

Large Fluid Leak-off

For large leak-off (Vl + qt or a + 0), the fracture

length in terms of Q from Eqns. (1), (6) and (7) is
The gamma coefficients for no leak-off and constant
1/2
formation properties are found to be approximately Qt
L(t) = — . . . (32)
2sCHP
H-FM Numerical SolutlO~

for a two-wing fracture. This is the same result as

rwo = I.O given in References [41 and [51.
rp : l/(n’t2)
Substituting Eqrr. (32) into (24) and rearranging
9. J07S . . . (28)
rf : 1 - (1 - 3fi/16)n” yields the PKN inaximum nellbore width formula

rv & 1/4 (n’+ Z)/(n’+ 3) i

i-u
‘(QHw/~Hp)]2n’+ 2 t,n?+ \
Ww(t) = a ~ Ka(Q/HH)n
for no viscous thinningl~ the friction coefficient [
(flow channel factor) is equal to the Newtonian . . . (33)
value (rf = 3x!16).
The above gamma coefficients
are based on nuraerical simulation results i/(ln’+ 2)
which where a = [ 3n rwO/(rprf x)]
include simultaneously solving the fracture
continuity and momentum equations for no leak-off.
These results are similar to the assumption of a Numerically the gamma coefficients for large
constant slurry velocity in the fracture. These leak-off (Q + O) were found to be approximately
coefficients are exact as n’ + O.
H-PKN Numerical Solution
The equivalent gamma coefficients as formulated from
the Perkins and KernJ analytical solution for a rW$ = 1.0
constant fracture flux (q(x) = q(0)) are rp 2 f(n’)/(n’+2); f(n’) 2 1.0 + 0.65n’/(n’+3)
o.407\
PK Analytical Solution3 rf 2 1 - (1 - 3#/16)n’ . . . (34)

rwo = 1.0 rv s (~/4)/(1 + rp)

rp = l/(2n’+2) for no viscous thinning I’f is 3s/16. The above

. . . 129)
I’f = ?rI16 gamaa coefficients are exact for n’ = O.

rv = %/4 (h’+ 2)/(2n’+ 3) The width proportionality constant a in Eqn. (33)

These coefficients are easily obtained fron Eqns, for a Newtonian fluid is 1.430 for a two-wing and
(C-7) through (C-1O) of Reference [31. 2.023 for a one-wing fracture. The values reported
by Nordgren* were 1.425 and 2.016, respectively.
The width equation for a Newtonian fluid (n’= 1)
from Eqn, (25) is
 . SF’E 15240 Bruce R. Meyer 5

rhe analytical fracture width profile for large Equilibrium models cannot accurately predict
fluid loss (n * 0) and n’ = 1 is’ vertical propagation unless the fracture is well
contained. The fracture shape predicted by
Wlt,t) 2 ti2 tl% equilibrium models for smal 1 confining stress
— = [i arcsint + (1-t ) - fi/2 .!1 . (35)
W(o,t) contrasts tends to grow into a star shape fracture
whose height is extremely large around the injection
where t (( = x/L(t)) is the dimensionless fracture welll). In many cases the fracture half height is
length. calculated to be greater than the fracture length
for formations with weak to moderate confinin9
Numerically the width profile for n + O and n’ = 1 barriers. This is not realistic and a drastic
is approximately misuse of these models.

H(t,t)/N(o,t) ~ (1 - t)@’~*’$ . . . (36) A distinct characteristic of equilibrium models is

the prediction of fracture pressures 13P less than
with only slight deviations near the fracture tip. the confining stress contrast llu.
Table 1 compares Eqns. (35) and (36) for various
values of t. As shown, the exact and numerical Equilibrium fracture height growth can be calculated
solutions are in excellent agreement. from the equations of Simonson et aJ. (1978)it for
2-D line cracks. Iforita et= al (19B5)1J provides a
3-D Analytical Formulae more general formulation for ellipsoidal type
fractures.
The three-dimensional fracture propagation modeli$
(M-3D) used is a variable height model with The stress intensity factor K1 for an ellipsoidal
characteristics of both the GDK and PKN type fracture with a very elongated shape (line crack)
fractures. This madel includes lateral and vertical and equal moduli in the adjacent strata is~~ilfl)~l$
momentum driven fracture propagation. +1 ilt
Slt
Ki = (Hlx) hPz(L) {~} dc . , (37)
This model approaches the PI(N solution for large J -i
ffacture length to height ratios and large confining
stresses. For no confining stress or moduli where H is the fracture half height and c = z/H.
contrast, the 3-D model reduces to the familiar
penny (circular) shape fracture. The fracture differential pressure for symmetrical
confining stress contrast is
The pressure initially decreases with time (after
breakdown) until it reaches the uPPer and lower 13PZ(C) = Pf - 51 ; ICI 4 HR/H
strata. Once re~ching the barriers, the Pressure . . . (30)
differential will increase for large confining = Pf - 52 : ICI J HR/H
stresses or continue decreasing for weak barriers.
The pressure will behave similar to the familiar where Pf is the fracture pressure, HR is the half
Nolte and Smith1t\t7 type pressure profiles for pay zone height, S1 is the minimum horizontal stress
various confining stress contrast gradients. in the pay zone and S2 is the confining barrier
stress (S2 > S1),
This section contains 3-D asymtotic fracture
characteristic formulae for formations with Integrating Eqn, (37) with the conditions given in
infinite, large (au/llP ) 1) and zero confining Eqn. (38) results in
stress contrasts.
2 HR
ir2
JLP = Kc/(xH) + Au ~arccos ~ . . . (Y~;
Infinite Confining Stress Cantrast

Theoretically, for an infinite confining stress where aP = Pf - 51 , 00 = s: - S1 and Kc is the

contrast (Off + * or Lu/ilP .>> 1 and L/H >> 1), the fracture toughness (Kc = Ki).
3-D solution approaches the analytical PKN formulae.
Benchmarking could also include 3-D simulation where Eqn. (39) shows that fractLlre toughness KC is ~n~t

the upper and lower fracture heights reach confining important when the fracture half height H I% =xali.
barriers. Once reaching these barriers the fracture Rearranging Eqn. (39) for a negligible first i~r:.
—
will propagate laterally with a constant wellbore results in the equilibrium condition
height and eventually obtain a uniform constant
height throughout the fracture.
rLP
Heq/HR = l/[eos f/2 ~ ] . . . (4(,;
Large Confining Stress Contrast

where Heq is the equilibrium (static) fracture haif

The equilibrium fracture model discussed here 15
height for a 2-D line crack.
represented by an equilibrium (static) height growth
condition rather than by momentum (dynamic). For
From Eqn. (40) it is apparent that the fracture
this condition to exist the fracture must be well
pressure at the wellbore IIP must be less than the
contained13. Therefore, a well contained fracture
confining stress contrast hu for an equilibrium
must have a confining stress contrast clr greater or
fracture to exist. In fact, if the net pressure
much greater than the net fracture pressure hP
differential is equal to the confining stress
(Jiu/hP ) 1 or >> 1). The fractura length must also
contrast (AP/8u = 1), the equilibrium fracture
be greater or much greater than the fracture half (Heq/HR + 9),
height would approach infinity
height (L(t)/H(t) ) 1).
Consequently, an equilibrium model forces the net
-4aK
 Design Formulae for 2-D and 3-D Vertical Hydraulic Fractures:
b Hodel Comparison and Parametric Studies 5PE 15240

ressure r3P to be less than Au (Kc/(sH)i/z (< hP). 2

heretically, as the confining stress contrast S+n’ ~ t-n,
l-v —
pproaches zero, the pressure differential will not Uw(t) = a ~ Ka Qr t]n ‘d , . . (46)
[ 1
pproach zero and the fracture approaches a penny
hape with equal fracture half height and length and
H(t) + L(t)). Therefore, the condition H(t) <
G “,+2 m Zn’+t
{t) must e:{is.t for positive confining stress R(t) = t~n +4 . . s (47)
antrast in a homogeneous modulus formation. b[ (1 - u)KaQ 1’
quillbriurn models violate this condition for 1 Ow where 2
.>rlfininq stresses.
a = 3“ rwll(l’f Pp PV‘-n’’*)]’’”+’
[’
1
No Confining Stress Contrast n
b = [rf Pp/(4]’ 24’’”I’w r?’’+*~] . . .
(40)
“he theoretical 3-D solution for no confining stress
or,t-zst (OJ = 0) is that of a penny or circular
,rl<pe fracture. The analytical solutions to follow Numerically the a and b proportionality constants
,.EI ~.$ this type. are given in Table 2 as a function of n’ for o + 1.

“?rr general 3-0 width-opening pressure relationship The width equation for a Newtonian fluid from Eqn.
,~.~~ ,~1~ (45) is
rw(l - v) llb
illx,i,,t) =
G
Hx(x,t)AP(x,O,t) . (41)
Uw(t) = c
[
l-v
y u Q R(t)
1 ... (49)

ihere H:: is the total fracture height at any

)c=#ltlon :<, For a penny shape fracture Iiw * R(t) where the proportionality constant c (IHP3D = 3) is
\F((J.t). 2.189 for a two-wing and 2,602 for a one-wing
fracture. The values reported by Geertsma et al.l
rhree geometry options are available for simulating were 2.149 and 2.556, respectively. These values
ttle fracture shape as the confining stress were calculated based on a reported c value of 2.00
~pprcaches zero. These 3-D width profile options for v = 0.25.
(IWP3D) are:
Substituting an apparent viscosity into Eqn. (49)
1) The fracture is elliptical in shape (IUP3D=1). results in a different Eqn. (45) proportionality
constant because the gamma coefficients are n’
W(r,t) = W(O,t) [1 - (r/R)zli/z . . . (42)
dependent.

ihere W(r,t) is the fracture width ●t any position Largt Fluid L@ak-off
, W(O,t) is the maximum wellbore width and R is the
‘racture radius at any tine t. The governing radius formulae for large leak-off (a
+ 0) from Eqns. (l), (6), (S) and (9) are
!) The shape of the fracture is deter~ined fros Eqn.
[41) based on the H-PKN type fracture pressure R < Hp/2:
)rofile (IWP3D=2) Q ti’z 112
R(t) = [ “1 ~
1 . . . (50)
W(r,t) = W(O,t) dP(r,t)/flP(O,t) . . . (43)
R >) HP12:
~here W(r,t) is the width at any position r and
Q t’f’
i(O,t) is the maximum wellbore width. R(t) = — . . . (51)
2RCHP

S) The circular fracture is approximately parabolic

(IWP3D=3) in shape$ These are exact analytical solutions for rI + O.

W(r,t) = W(O,t) (1 - r/R)rwth . . . (44) The penny shape fracture width formulae obtained by
substituting Eqns, (50) and (51) into (45) are
inhere the gamma KOefflClent Pwth VarieS frOm 1/2 (n’
❑ 1) to 2/3 (n’ = O) for no fluid loss (a + 1). R ( Hp/2:
1
The maximum wellbore width for po!!er-la: fluids
i-v——
based on the methodology of Ileyert$ is Ww(t) = a ~ KaQ t C i (52)

Zn’+
2
[
1
4 3n’rw, 1-V where

‘1
4 (3/2)n’rwt ‘/(2n’+t)
Ww(t) =
rf rp
Ka ~ qn’R(t) *-n
. . . (45) a =
[ rf rp fI~-n “1
. . . (s3)
[
and
NO Fluid LOSS R >> Hp/2:
—
he penny shape width and radius formulae obtained
IIy substituting
for no leak-off
Eqn.
are
(5) into (45) ●nd rearranging Wwlt)
1-s
= ● ~
[ ‘a “n’[$l’-n’r”’+’
t* o’54)
Qae
“.”
there 3n’ rwo i/(2n’+2) of power-law fluids results in
~. narrower and longer fractures for non-constant width
[ rf rp ITt-n ‘1 , s , ’55) ‘iscous ‘binning
fractures (i.e. PKN and 3-D), A parametric study
shows that the PKN fracture width and length ratios
‘numerically, the above nidth proportionality
for viscous thinning versus no thinning (l’{ = 3x/16)
:onstants are tabulated in Table 3 as a function of
are
1’ for large leak-off (n + O)t Since t5e
Proportionality constant a is not a function of the 1-II
*W = ti’ = [(3s/16)/rfll’(*’’’+3);
volume coefficient rv! it is the same for all three
]enny shape models (1WP3D = i, 2 k 3). and . . . [56)

PARRHETRIC STUDIES iw = t(3fi/16)/rfll’(*n’+2) ;a+o

Numerical simulation studies are usually performed nhere *W and PI are the width and length ratios (t +
to determine the importance and effect a parameter 1; 0.838 f tw ~ 1 & 1 4 ?1 f 1.193 and a + O; 0.767
has on fracture design. These studies are typically f !W ~ 1 k?l = 1).
based on changing one parameter while holding the
remaining constant. This method is time consuming COMPARISON OF 2-D AND 3-D DES113NS
and cannot a priori identify the major variables to
base the study. Usually, hydraulic fracture designs are based on
utilizing an inhouse or service company GDk or PKN
The analytical forfriulae given for no leak-off (n + type model. Frequently, only crne model is available
1) and large fluid loss (u, + 0) can bound the effect and is used for all simulation designs irrespective
parameters have on fracture characteristics. of the utility. If more than one model is
available, a choice is made as to which is most
For example, it is evident that the GDK pressure applicable. Next the 2-D height or confining stress
decreases with time (ilP ‘“ ta, where c = -n’/(n’+2), contrast must be detemined, Even with a realistic
a+ 1 and o = -n’/(2n’+2), rI + O) and the PKN 2-D height, 2-D versus 3-D simulation for weak
pressure profile increases with time (LP * t’, where barriers is sometimes questionable.
a = l/[~n’+3), Q + t and a = l/(4n’+4J, n + 0). The
penny shape pressure profile decreases with time. A comparison of 2-LI and 3-D fracture propagation
and proppant scheduling characteristics is given for
The effect of injection rate on fracture GDK , PKN and 3-D type models, Examples are
characteristics for a given slurry voluae is another presented for a wide range of confining stresses and
useful parametric study. Holding Qt constant, the leak-off. The 3-D simulations are for ItiP3D=2,
GDK width is shown to increase with flow rate Q (Ww Viscous thinning is not included in this study for
* Q“, where a = n’/(2n’+4), n + 1 and II = the PKN or 3-D models (1’f = 3x/16),
(n’+f)/(2n”+2), rI + 0) as does the PKN neiibore
#idth (Ww * Qa, where E = n’/(2n’+3), n + 1 and t = The simulation studies were based on the input data
(2n’+1)/(4n’+4), n + O). in Table 4, The formation and treatment data were
similar for each case. The equivalent 2-D height
Thest ~c,m,l;ae demonstrate the non-linear behavior was calculated from the 3-D fracture profile. The
of most parameters. For example, doubling total pay zone height was 150 feet. 011 cases were
Poisson’s ratio from 0,20 to 0.40 has a for a constant total leak-off coefficient C, with
negligible effect on the wellbore width (i.e. GDK: leak-off only to the pay zone,
iIlw(ti=0.2(1)/Ww(v=0.40) = 1,049; PKN: WW(V=0,20)
/Ww(v=0.40) = 1.059 for t = 1 and n’ = 1). This The effect of proppant concentration on slurrv
nonlinearity greatly minimizes the effect of data rheology properties (n’ k K.’) and sett!inq rats was
uncertainty. not included in this study.

The rule of thumb based on a given injection rate FRACTURE DESIGN CHARACTERISTICS
per unit fracture height (Q/Hw) for optimized
fracture characteristics also seem5 to be well A symmetrical confining stress contrast range of -50
founded from the 2-D formulae. to IOJ psi (-50 4 du + IOJ) was chosen to illustrate
the GDK, PI!N and 3-D model characteristics over s
Parameters with the largest exponents in the width, wide range of fracture length to height ratios..
length and radius formulae have the largest effect
on fracture design. Therefore, the greatest The 2-D fracture height was calculated from the
attention should be given to these parameters for simulated 3-D height profile based on
fracture design and optimization,

H2d = tHw + 2(1-1) ~iH(t) dt . . . 0 0 . (57:

The effect of slurry efficiency on fracture geometry o
can be approximated by replacing time t with tt for
large values of a. The effect on fracture length is where H2d is the equivalent 2-D height, Hw is the
much greater than on the width or pressure because 3-D wellbore height and H(t) is the 3-D height
of the larger time exponent power (PKN; L * te where profile as a function of dimensionless length /.
a = 4/5 for n’=1 and n + 1). The approximate PKN The weight factor t (O 4 } 4 1) was set equal to one
fracture lengtti and width ratios for s = 1 and half for this study. This value of t was chosen
11 . 1/2 are L(rl=l)/L(l=l/2) 2 1.74 and rather than zero to weigh the 2-D height toward the
Ww(rI=l)/Ww(rI=l/2) g 1.15 for n’=1. Fluid loss also weIlbore. This seems realistic since the greatest
has a major impact on sand scheduling design. fracture volume per unit length is near the
wellbore. Using the maxinuim 3-D wellbore height
‘aa7
“./r
 Design Formulae for 2-D and 3-D Vertical Hydraulic Fractures
8 Hodel Comparison and Parametric Studies SPE 15240

would o<er estimate the height near the tip. wellbore pressure differential (integrated
vertically) is approxi~ately 0.714 (IPW for hu + O
Table 5 compares fracture characteristics at the end (n’=O,S). For Ex, 2, the average 3:D fJPw is about
of pumping for GDK, PKN and 3-D designs as a !10 versus 104 aad 132 psi for the 6DK and PKN
f~lr,
ct ion of confining stress and leak-off. Two models, respectively, The 3-D pressure is
e::ample sets were run with different leak-off approximately constant at any position t for large
coefficients. Examples 1-9 were siaulated with a confining stresses as is the PKN pressure.
total Ieal.-off coefficient of 1.0 x 10-} and Exs.
](1-i8 with a coefficient of 0.5 x 1(1-3 ft/mini/~, The SDK wellbore width is shown to increase with
increasing confining stress, This is expected for
Th:+ GDI: model compares favorably to the 3-D results cases of no fluid 10s5 and most other cases as
/or i~ngth to height ratio~ of order one and the PKN evident from Eqn. (13) (Ww*(ta/Hwl/t), u =
n.;c:! comndres very favorably for ratios greater l/(n’+2)), The PKN width is shown to initially
!h?ln Ln$?. This IS noticeable in all characteristics increase and then decrease with increasing confining
:.c:ot the 6Uk maximum wellbore width and pressure, stress (decreasing height) for large fluid loss
I*> ciscussed below. For near penny shape fractures (Exs. 1-9), For low leak-off (Ex5. 10-18), the PKN
.4;1 icndels gl’ie comparable results. Figures 1 midth increases far increasing values of confining
-t,,u:.lgh 5 show GDK, PEN and 3-D fracture profile stress. These trends are verified by substituting
,;,;+er=r,c~s for no, moderate and large confining nt for t and fracture h’eight into Eqn. !2s) . The
stress contrasts. general behavior of the 3-D model for large
confining stl”esses can be obtained from the PKN
ltlE ira.:ture efficiency at the end of pumping for equations.
)arqe (E;.s. 1-7’) and moderate leak-off (ExE.. 10-18)
iS :Ilol+il t,j decrease with increasing confining The GDK wellbore width from Table 5 does not appear
>~rzz5, The major reason is that with leak-off only to match the PKN or 3-D widths. For low confining
ta the Fay zone, the leak-off area increases with stresses a better match would be expected since the
increasing confining stress because of the increased GDK model is generally more applicable at small
iracture length. The efficiencies for all three length to height ratios. The reason for this
models are reasonably close with the 3-D efficiency apparent anomaly is that the maximum rather than
being t~le lowest because of the slightly greater average wellbore widths are reported. Since the GDK
ler,qt~,~ In the pay zone. The GDK efficiency is width is vertically constant, the average wellbore
hlqher than both the F’KN and 3-D models for large width is equal to the maximum width. The PKN and 3-D
confining stresses because of the simulated shorter (IWP3D=2) average width is 1/4 times the maximum
fracture lengths. For low confining stresses the width for constant formation properties. Thus, the
F’KN efficiency is the greatest due to the shorter average widths in Ex. 2 (flv = O) are 0.230, 0.270
ler,gths, and 0,251 inches for the GDK, PKN and 3-D models.
The average widths in Ex. 9 (Aw = 104) are 0.425,
The fracture height is shown to decrease with 0.271 and 0,271 for GDK, PKN and 3-D, respectively.
Increasing confining stress, The fracture Comparison of average widths shows that the GDK
approaches a penny shape for low confining stresses widths are very close to the 3-D model at low
and a constant height for large confining stresses confining stresses and much larger than the 3-D or
with a wellbore height equal to the pay zone height. PKN widths at large length to height ratios.
The .2-D fracture length propagation behavior for
varlng efficiency and confining stress can be Table 5 shows that decreasing the leak-off
illustrated by substituting at for t in the GDK and coefficient by a factor of trio, the fracture length,
PKN length equations. For the GDK model,
the if height and efficiency increase for a given confining
fracture height (Hw+l/2) decreases faster than stress contrast (Exs. 5 & 14), as expected. The PKN
efficiency (na) for increasing confining stress, the and 3-D pressures, however, are lower for small
fracture length will increase (L * rIa/Hw+l/Z, a = confining stresses and higher for large confining
(n’+1)/(n’+2)). A similar analogy can be applied to stresses. The lower pressures are a result of a
the PI(N and 3-D models, greater increase in fracture height than width at
low confining stresses (dPw *Hw/Hw).
The GDK pressure was found to decrease and the PKN
pressure to increase with time as theoretically Table 6 shows a comparison between the equilibrium
discussed above. For the 3-D model the pressure fracture height Eqn. (40) and the simulated 3-D
initially decreases until the fracture reaches the wellbore half height for Exs. 4-9 and 13-18. An
confining barriers; and then either increases or equilibrium height for stresses less than 250 psi is
continues to decrease depending on the confining not possible since fIPw > /lu which implies the
stress contrast, The 3-D wellbore pressure increased fracture height goes to infinity (Kc/(tH)l’2 << hP).
for cases of large confining stress (Exs. 6-9 and
15-18) and continued to decrease for cases of low As the fracture pressure to confining stress
confining stress (Exs, 1-3 and 10-12). contrast ratio (fIPw/hr) decreases, the equilibrium
heights are shown to approach the 3-D sisulated
The wellbore pressure
at the end of pumping for results. For a confining stress of 250 psi, the
increasing confirming stress is shonn to increase error is 52-54%, while at a confining stress of 1000
greatly for the PKN and 3-D models and slightly for psi, thE error is 2-3%. This illustrates that
the GDK model. The 6DK pressure at sma;l confining equilibrium models are valid for large confining
stresses is shown to be much lower than the 3-D stresses. For 1Ow confining stresses the
model. The reason is that the 3-D pressure (6P) is equilibria height greatly over estimates the 3-D
proportimal to the maxinu~ wellbore width and height. The reason is that at low confining
decreases vertically (Eqn. (41)). The average 3-D stresses fracture propagation is momentum driven and

-on
. Bruce R. Meyer 9
SPE 15240

lot in “static” equilibrium. To prevent calculating dimensionless fracture conductivity is defined as

in infinite fracture height, an equilibrium model
$ill force the fracturing pressure to
be less than FCD = ~1 Ueff(/)Heff(t)Kf(t)d//(KRLPHP) . . . (58)
:he confining stres5 contrast. However, the o
:alrulated equilibrium height may be greater than where Weff is the effective propped width, Heff is
the length for positive confining stress, which Is the effective propped height in the pay zone and Kf
impossible. is the fracture permeability at any position /. The
reservoir permeability is given by KR.
rhe 2-D models do a fairly good job of modelling 3-D
fracture characteristics for a height weighting Table 7 shows that the amount of pad volume required
factor \ of one half. Refinement of $ to a slightly increases with increasing confining stress. This is
smaller value may have produced better results fnr a result of decreased fracture efficiency as the
some of the examples. The other observation is that leak-off area in the pay zone increases. Noticeable
the GDK model is more applicable for predictin9 is that the 3-D designs require a greater pad volume
fracture characteristics at low confining stresses than the GDK or PKN models. This is due to the
while the PKN model is more applicable at larger loner 3-D efficiency and fracture shape near the tip
length to height ratios. (H(&,t)/HR ratio). Decreasing the 2-D height
weighting factor t to slightly less than 0.5 would
Table 5 suggests that if the fracture lflltlallY also bring the 2-D results in better agreement with
grows as a penny shape and then is contained with a the 3-D model.
bounded width, the PKN model would be the better
model to use. This is more evident when sand The pad volume required is shown to be greater for
scheduling is included in the design. However, if the larger leak-off coefficient cases at a given
the fracture width is vertically unbounded (constant confining stress. Although changing the total
width) the GDK model may be preferable to both the leak-off coefficient by a factor of two did affect
PKN and 3-D models for calculating all fracture the fracture characteristics it has a much greater
characteristics except pressure. impact on the proppant scheduling, especially on the
amount of pad and final batch volumes.
PROPPANT SCHEDULIN6
Figure 6 shows a typical concentration profile in
The proppant transport model calculates ProPPant the 3-D fracture at the end of pumping for Ex. 5.
placement, suspended proppant and bank heights, The solid horizontal lines represent the inlet
along with slurry concentrations during pumping and concentration schedule for each batch. The
at closure. The model tracks the slurry and concentration profiles are shown to increase as a
proppant for variable width and height fractures function of the position in the fracture with the
from initial injection until closure. The design is first schedules injected having the greatest change
checked for bridging- and screening-out. Details of in concentration levels. Figure 6 clearly
this model are given in Reference t141. identii.es the need to accurately track every
element of slurry and proppant rather than using
Table 4 lists the formation, treatment and sand spatially or batch averaged values. Noticeable is
scheduling input data. All designs are based on a the exponential behavior of fluid 10SS near the
total injected volume of 84,000 gallons slurry at a fracture tip.
constant rate of 20 bpm.
The GDK proppant scheduling is reasonable [lose to
Proppant scheduling was calculated using the the 3-D scheduling for low confining stress (Exs. 3
automatic sand scheduling option. The design was & 12) but under estimates the pad volume fof large
scheduled in one ibm/gallon liquid increments to confining stresses. This is due to the Iaryer ED!.
achieve a final maximum concentration of 5.5 efficiency and relatively wider width and ahol t?r
ibm/gallon in each batch at the end of pu~ping. length. All three models show the same gene:al
‘designing for a uniform concentration per unit area characteristics of increased pad with decrca$lr(q
is not practical for variable width fractures. It efficiency and increasing confining :tress. ThE :-2
is more important to have a higher concentration and PKN models require more pad because O+ iowcf
near the wellbore than near the fracture tip. The efficiencies and narrower fracture wldttis near the
maximum proppant concentration of 5.5 lbmlgallon was tip.
chosen for illustrative purposes only. )luch higher
concentrations would typically be pumped in many The propped fracture length
is shown to lncrca~.e and
formations to increase the fracture conductivity. A the dimensionless to decrease
conductivity with
constant settling velocity in each batch of i.25 increasing confining stress. This is due to u!:mpl,,~~
ft/min ~as introduced to show the effect settling less proppant at lower efficiencies for @ ai~en
has on the propped fracture length. volume of slurry injected and an increased proPFed
length. The GDK model is closer to the 3-D solutloo
A comparison of proppant transport and scheduling at low stresses and the PKN at higher stresses. The
characteristics is given in Table 7 for 6DK, PKN and GDK propped length is also shown to be shorter and
Z-D type fracture designs. The designs are for a the conductivity greater. This is due to the
wide range of confining stress contrasts (OO = O, relative wider GDK widths and shorter lengths as the
35(I k 750 psi) and moderate (ExE.. 12, 14 & 16) to fracture length to height ratio increases. Thu~,
high (Exs. 3, 5 $ 7) fluid loss cases. for massive fractures a GDK model would require
significantly more slurry and proppant than the PKN
The fracture efficiency rI (end of pumping), propped or 3-D models fcr a given design fracture length or
fracture length Lp and dimensionless fracture propped length. The 6DK .sodel would also predict
conductivity FCD are given for each design. The greater conductivities. All aodels, however, show

~ao
.,-.
 Design Forsulae for 2-D and 3-D Vertical Hydraulic Fractures:
1(1 Hodel Comparison and Parametric Studies

he saline general trends. 4. All three aodels show the saise general proppant
scheduling and transport characteristics.
‘lgures 7 and S show the GDK suspended proppant and However, the PKN ●nd 3-D odels,● generally,
,ani..ed profiles at the end of pumping and closure require ● ore pad because of the loner
or Ex. 12. Figure S shons. that most of the efficiencies and narrower fracture widths near
~,-oppant settled beloa the pay, which explains why the tip.
i,~ propped fracture length is less than in Ex. 3.
‘F,ls behavior na~ typical for all three models. The 5. Elemental tracking of the slurry and proppant is
:ettllrlg of sand below the pay zone in Ex. 12 is due essential for realistic transport design. Fluid
o the larger fracture efficiency which results in a loss near the tip is exponential-like in
r’nq&.- closure time. This allows the sand to settle behavior.
.:\ b~l~w the pay zone before the fracture CIOSeS.

“:1+ ;.ropped fracture length is defined at the point b. Fracture efficiency has a large impact on
.r.~re the sand bank falls below the pay zone. proppant scheduling design and fracture
propagation, The total leak-aff coefficient is
“,fiicil fracture profiles for each model at the end ane of the most critical parameters in fracture
.+ pumping and closure are shawn in Figures 9 design.
:,,
-,~:.:gfii4 for Ex. 5. These Figures show the
:,si~fld.?d and settled bank profiles at the end af 7. Most parameter exponents in the width and length
..(tl]plr,g
and final propped bank height at closure. equatians are less or much less than unity which
:he proppant transport characteristics are shown to minimize their effect on fracture design. The
)P ~lmllar for each model. greatest attention should be given to parameters
with the largest exponents for fracture design
rhe SDI and F’KN models were shown to provide optimization.
~sn~rally acceptable results as cainpared to the 3-D
ncd$?l. This, of course, assumes that a good NOMENCLATURE
~=tillate of the 2-D height is used. Again, a little
jetter reiults may have been obtained if a slightly ~ = Leak-off area (one face)
sm~ller 4 factor was used, which would bring the 2-D c = Tatal leak-off coefficient
efficiencies and pad volumes into better agreement E = Yaung’s modulus
~,th ihe T-D r@5Ult5. The amount of pad volume is FCD = Dimensionless fracture conductivity
~lso qo~erned by the height prafile near the tip. G = Shear aodulus
~ot- a leading edge model (H(t=l)=HR) the amaunt af H = Fracture half height
~ad val ume required would be smaller, Therefore, H2d = Equivalent 2-D fracture height
lumping more pad with higher concentrations near the Heq = Equilibrium half height
end of the job would make for a better design and Hp = Pay zone height
oreclude the risk of screening- or bridging-out near HR = Pay zone half height (HR=HP/2)
the tip. Alsa, it is more impartant to have a high Hw = Total wellbore height
conductivity fracture near the wellbore. K’ = Slurry consistency index
Ka = Apparent consistency index
CDNCLUS1ONS Kc = Fracture toughness
K1 = Stress intensity factor
Comprehensive 2-D and 3-D design formulae have been L ❑ Fracture half length
presented for power-laa fracturing fluids. These Lp = Propped fracture half length
farmulae can be used far parametric studies and n’ = Slurry flow behaviar index
benchmarking GDK, PKN and 3-D type fracture p = Pressure
prapagatian models. The major difference betneen pf = Fracture pressure
the GDK and PKN models is the width-apening pressure q = Fiaw rate lone wing)
relationship and not the vertical width profile. Q = Total flow rate
r = Radial coordinate
Fracture and proppant scheduling design comparisons R ❑ Fracture radius
for 2-D and 3-D models were performed. The major S1 = Minimum in-situ horizontal stress
conclusions are: S2 = Canfining stress (adjacent layer)
sp = Spurt loss coefficient
1. Two-dimensional GDK and PKN type models can t = Time
provide acceptable fracture characteristics for Vf = Fracture volume (one wing)
various degrees of containment provided an v, = Volume loss by leak-off
equivalent 2-D height is established. = Volume loss by spurt
Vsp
M = Width
2. The GDK madel is generally mare applicable at Ww = Haximum wellbore width
low confining stresses (L/H 4 1) and the PKN ~ ❑ Lateral coordinate along frac length
model at high confining stresses (L/H * 1). Far ~ = Vertical coordinate
large length to height ratios the 6DK ●adel
predicts wider wellbore widths and shorter = Leak-aff area propagation parameter
‘a
lengths than the PKN or 3-D type models. 7f,rf = Friction coefficients
= Pressure profile coefficients
7
7P$rP
,. Equilibrium 3-D models should only be used for 7V,I’V = Fracture volume coefficients
well contained fractures where “static” 7W*I’W = Width-opening pressure coefs.
equilibrium occurs. For small confining stress 7W,,I’W,= Haximun width-opening pressure coefs.
contrasts, an equilibrium model ●ay greatly over fjp = Pressure differential, (Pf-Sl)
estimate the wellbare height. aPz = Vertical pressure function, Eqn. (38)

AM
SPE 15240 Bruce R. Meyer 11

)0 = Confining stress contrast, (SZ - S1) of Stress Variations, ” Sot. Pet. Engr. J., Dec.
= Dimensionless vertical coordinate, c ❑ z/H 1983.
I = Fracture efficiency
I = Viscosity 12. Abou-Sayed, A. S. and Sinha, K. P.: “Evaluation
lapp = Apparent viscosity, Eqn. (19) of the Influence of In-Situ Reservoir
w = Poisson’s ratio renditions on the 6eometry of Hydraulic
= Dimensionless lateral coordinate, t = x/L Fractures Using a 3-D Simulator,” Part 1-SPE
r = Fluid loss delay time 12877, Part 2-SF’E 12878, 1984.
1 = 2-D height weighting factor
r = Viscous thinning rdtio 13. florita, N., Iihitfill, D. L. and Wahl, H. A.:
“Stress-Intensity Factor and Fracture
Subscripts Cross-Sectional Shape Predictions From a 3-D
Ilodel for Hydraulically Induced Fractures,” SPE
D = Dimensionless
142b2, Sept. 1985,
Pff = Effective
f = Fracture, friction
14. Heyer, B, R.: “Frac Model in 3D - 4 Parts,” Oil
1 = Lower strata, length
and Gas Journal, June 17, July 1, July 22 and
R = Reservoir
July 29, 1985.
u = Upper strata
w = Width, wellbore
. 15. Veatch, R. W., Jr.: ‘Overview of Current
x-direction
Hydraulic Fracturing Design and Treatment
2 = z-direction
Technology - Part i and 2,” Sot. Pet. Engr.
AIME, 1983.
REFERENCES
lb. Nolte, K. G. and Smith, Ii. B.: “Interpretation
1. Geertsma, J. and Deklerk F.: “II Rapid Method of Fracturing Pressures,” J. Pet. Tech., P.
of Predicting Width and Extent of Hydraulically 1776, Sept. 19S1
Induced Fractures, ” J. Pet. Tech., P. 1571,
Dec. 19&9. 17. Nolte, K. G.: “Fracture Design Considerations
Based on Pressure Analysis,” SPE 10911, May
9
L. Daneshy, A. A.: “On the Design of Vertical 19S2.
Hydraulic Fractures,” J. Pet. Tech., P. S3,
Jan. 1973 1s. Simonson, E. R., Abou-Sayed, A. S. and Jones,
!-), H.: “Containment of tlassive Hvdraulic
7
., . Perkins, T. K. and Kern, L. R.: “Widths of Fractures,” Sot. Pet. Engr. J., Vol. 14, No. 1,
Hydraulic Fractures,” J. Pet. Tech., P. 937, p. 27, Feb. 1978.
Sept. 1961.

4. Nordgren, R. P.: “Propagation of a Vertical

Hydraulic Fracture,” Sot. Pet. Engr. J.* P.
306, Aug. 1972.

5. Geertsma, J. and Haafkens, R.: “A Comparison of

the Theories for Predicting Width and Extent of Tablt 1 - Exact and Approximate PKN Width Profiles
for Large Leak-off (1 + o)
Vertical Hydraulically Induced Fractures,” ——
T
Trans. ASME, p. S, Ilarch 1979. r pKN Hidth Profile
~ Error 7
lPosition w(t,t)/w(o,t)
h. Settari, A,, and Price, H. S.: “Simulation of
Hydraulic Fracturing in Low Permeability Exact Approximate ,, -“–;
Reservoirs,” SPE/DOE S939, 19S0. t = x/L Eqn (35) Eqn (3L?
——.-. I
(
7. Advani, S, H., et al.: “wFluid Flow, Structural 1.000000 i).;! ~ sO I
0.00 1.OOOOOO
and Fracture Mechanics Modeling Associated With 0.98(1320 (:,<,;:
0,05 0.980091
Hydraulic Stimulation Operations,” SPE/DOE 0,959995 (l,(!;; !
O.10 0.939598
10846, 1982. ,.,
.()k O I
0.20 0.916615 0.917165
~m:,~:, \
0.30 0.870459 0.s70914
8. Settari, A. and Cleary, It. P.: “Three- (I.4(I 0,820309 ().820415 (1.(ii?
Dimensional Simulation of Hydraulic 0.764453 -0,06;
0.50 0.7b4966
Fracturing,” SPE 10504, 19S2. 0,701130 -0.2(1:
0.60 0.702554
0.70 ().629834 0.6271b9 -0,4:7
9. Settari, A. and Cleary, M. P.: “Development and 0.5359s1 -0.793
0,80 0.540267
Testing of a Pseudo-Three-Dimensional Hodel of -1.521
0.90 0.416060 0.409732
Hydraulic Fracture 6eometry (P3DH),” SPE 10505, 0.~20621 0.313221 -~. ~(!s
0.95
1982. ().000000 (1,(!0[1
1.00 ().000000
—--J
10, Cleary, H. P.: ‘Comprehensive Design Formulae
for Hydraulic Fracturing,” SPE 9259, Sept.
19ao.

11. Palmer, 1. D. and Carroll, H. E., Jr.: ‘3D

Hydraulic Fracture Propagation in the Presence

.-.
*UI
 15/40

Table 2 - Penny Shape Width ●nd Radiue Proportionality

cOIIStJIItS - NO Leak-off (U + I)

Flow a coef., width Eqn (46) b coef., radiue Eqn (47)

Ind*x
I14P3D IWP3D
n’ 1 ~ 3 1 2 3

(),O 1.6290 1.7225 1.8082 0.5414 0.5724 0.6009

O.i 1.6652 1.7442 1.8232 0.5355 0.5650 0.5948
0.2 j.bqbq 1.7636 1.83b2 0.5304 0.5584 0.5893
l),: 1.7248 1.71309 1.8475 0,5261 0.5525 0.5S44
0.4 1.7494 1.7’764 1,8572 0.5224 0.5472 0.5800
0.5 1.7712 1.8103 1.8657 0.5192 0.5424 0.5761
O.b 1,7904 1.8228 1.8731 0.5164 0.5381 0.5726
il,7 1.S073 1.8341 1.8795 0.5140 0.5341 0.5694
().8 1.8224 1.S442 1.8850 0.5119 0.5305 0.5665
i].? 1.8357 1.8534 1.8898 0.5100 0.5272 0.5638
1.0 1.8475 1.S617 1.S939 0.5084 0.5241 0.5b14

Table 3
Penny Shape 14idth Proportionality
Constant5 - Large Leak-off (n + O) Table 4 - Sum*ary of Input Data

I Formation data
Flow 1 Hidth Proportionality
Index Constant, a Young’s modulus,* psi. . . . . . . 6.0 X 104
-r Poisson’s ratio* . . . . . . . . . 0.250
n Eqn (53)* Eqn (55)+ Ilinieum horizontal stress, psi . . 10,000
(,,() 0.957a 0.4709 Stress contra5t,* psi. . . . . . -50 to lot
0,1 1.0494 0.5767 Total pay zon~ height, ft. . . . . 150
{),2 1.131’3 0,6731 Reservoir pressure, psi. . . . , . 8,000
0.3 1.20A4 0.766s Reservoir compressibility, psi-l . 9.18 x 10-f
0.4 1.2737 0.8571 Permeability, md . . . . . . . . . 0.50
().5 1.3347 0,9437 Porosity. . . . . . . . . . . . . 0.10
O.b 1.3900 1,0264 Fluid viscosity, cp. . . . . . . . 1.00
0,7 1.4404 1,1050
Treatnent data
0.8 1. 48b4 1.1797
0.’? 1.5285 1.2506 Injection rate (Z-wings), bbllmin. 20
1.0 1.5671 1.3177 Total volume injected, gal . . . . 84,000
Flow behavior index. . . . . . . 0.50
● R < Hp/2; + R >> Hp/2 0.05
Consistency index, lbf-sn:/ftz . .
Total leak-off coef., ft/nin112. . 1 x 10-3 (Exs. 1-9)
5 X 10-* (Exs. 10-18)
Spurt 10s5 coefficient, ga Iftt. , o
Fracture pereeabilit; , ●d. ,.. 1 x !Of

Sand scheduling data

Concentration, Ibm/gal liq . . 1 to 5 (one lb incr)

Maximum concentration, Ibm/gal iq 5.5 (end of pumping)
Proppant diaseter, in. . . . . 0.025 to 0.05
Settled porosity . . . . . . . 0.40
..
Pronuant density. lba/ftJ. . . . 162
Settling velocity, ft/ain. . . . . 1.25

●Barriers and reservoir properties are assumed to be equal in

these examples. Formation properties, however, can ba
srzecified for each layer.
 WE 15240

Table 5 - Fracture Propagation Characteristics for 2-D ●nd 3-D Models

tress Pressure Length Width’ Height*” Efficiency

ff,psi DPW, psi L,ft WM, in Hu,], ft a

6DK PKN 3-D ;~k. PKN = 6DK FKN 3-D 6DK PKN 3-D 6DK PKN 3-D

104 115 137 355 333 353 0.217 0.341 0.315 395 395 434 0.711 0.742 0.690
104 132 155 374 3b4 308 0.230 0.344 0.319 347 347 38B 0.b96 0.717 0.bb5
105 lb7 181 41)a 421 441 0.253 0.348 0.331 278 278 311 O.bbS 0.b72 0,b29
107 231 2i8 65? 511 544 0.288 0.351 0.355 203 203 244 oBb~6 o.boo 0.571
108 304 272 507 595 63! 0.322 0,352 0,3bb 154 154 lBb 0.587 0.532 0.520
109 3B1 341 54B b66 701 0.352 0.351 0.366 123 123 144 0.553 0.475 o,4h9
110 456 422 503 7:5 74A 0.379 0.350 0.359 102 102 114 0.524 0.424 0.423
111 504 475 bOl 75b 7b9 0,393 0.349 0.355 92 92 100 0,509 0.399 0,397
113 1114 b13 641 S17 816 0,425 0.34S 0.345 75 7s 75 0.476 0.34b 0,34b

-5!) 100 109 131 379 34b 376 0.224 0.352 0.328 429 429 470 0.B47 0.8b7 0.034
() 101 127 1!9 402 384 420 0.238 0.356 0.334 375 Z75 420 0,837 0.852 O.BIS
100 101 163 17& 446 460 49i 0.2b7 0.3b4 0.S31 297 297 332 (),s19 0.s22 0.793
250 102 237 219 519 594 b43 0.311 0.373 0,384 210 210 255 0.790 0.770 0,750
350 102 313 27b 5%0 714 7b7 0.350 0.379 0.39B lbl lbl 19b 0.7k5 0.723 0.712
500 103 397 351 b38 830 879 0.387 0.3B3 0.402 129 129 153 0.741 0.677 0.b71
750 104 4@4 442 s90 937 9b9 0.421 0.386 0.398 106 10b 120 0.720 0.b34 0.631
1000 104 540 505 720 999 1020 0.441 0.3S7 0.39b 9b 96 104 0.707 0.b09 0.606
10J 105 S91 b$’1 792 l14b 1145 0.439 0.369 0.368 7!3 75 75 0.678 0,550 0.549

* See Table 4 for ingbi data. C is l.OXIO-] fw Exs. 1-9 and 0.5x10-1 ft/ininL/2 for Exs. 1O-1B.
+ welloorc?
ttaximurn width. The av~. PKN and 3-D (I!4P3D=2) wellbore width is 2 x/4 WM.
E* ~axlfium wellb~re half hei9ht-
Tlre equivalent 2-D height was calculated from the 3-D model based on
the arithmetic inean of the wellbore heiqht and average 3-D fracture height profile (Eqn 57, t=l/2).

Tabls

r
—- >
6 - Comparison
.——
U! Equilibrium

W
and 3-D Hellbore Heights

&Ilp=l=’‘eirtH@ ‘t ‘r:or
+0 i :: ‘Y-:riun
5
‘-:: ‘
:50 272 219 1B6
54
18 ~
‘

b 500 341 157 144 9

7 , 750 422 11s 114 4
a 1000 475 102 100 2
9 104 613 75 75 0

13 250 219 , 3B7 255 52

14 350 27b 230 196 17
{5 500 351 166 153 9
lb 750 442 125 120 4
ii
1000 505 107 104 3
If! 10J 691 75 75 ()
~ 1

*See Table 4 for input fata. C is l,9x10-i tor Exs. 4-9 and
+0,5x10-1 ft/#in1~2 for ExE.. 13-113.
Equilibrium height Eqn (40), Heq = HRICOS $% HR =75 ft.
 s 15240

able 7 - Proppant Scheduling Characteristics for 2-D and 3-D Hodels

haracteristics Proppant Scheduling (gallons slurry/batch) ~

k
“ess Leak-off Coef. Leak-off Coef.
Proppant
Iu C = 0.001 ft/mini’2 c= 0.0005 ft/mini’2
Concentration
Exam~le 3 Example 12

1s1 lhnt/gal liq. GDK PKN 3-D GtIK PKN 3-D

=100 Fad O 12400 11900 1968(J 4020 3430 8140

1 531O* 5420* 2B70* 2160*
4440* ;;::; :::::
2 S240 e3eo 5510
3 13640 13800 10470 9770 9760 7600
4 24450 Z459CI 22590 27420 27810 23570
5 19960 19’710 22880 3b190 35940 39730

n O.hbfi 0.672 0.b29 0.819 0,822 0.793

Lp, +t S94 408 397 184 181 197
FED 3.07 2,82 3.10 2.94 2.95 4.11

Example 5 Example 14

psi Ibm/gal liq. GDK PKN 3-!) GDK PKN 3-D ~

=550 Pad O 18620 23170 2B71O 6430 8360 13200

1 6500* 7260* 3470* 3460* 4560* 2420*
~ 9240 9b70 6200 6080 7500 47s0
3 13640 132:0 10610 11s80 13370 9700
4 21050 18560 19170 27350 26500 :4260
5 14950 12130 15840 28609 23710 29670

n 0.587 0.532 0.520 0.765 0,723 0.712

Lp, ft 4!34 5b5 596 399 580 695
FCD 2.57 1.93 1.81 2,66 1.37 1.11
.—
Example 7 Exaaple lb

)si ibm/gal liq. GDK PKN 3-D “– 6KIK PKN 3-D

=750 Pad II 24070 33540 36120 S990 14280 17670

1 7iO0 7640* 5040* 4370* 6330* 4oso*
2 9500 92S0 7376 7210 9250 6050
3 13020 11330 10460 12970 14200 11910
4 18310 14000 15020 26270 22970 22990
5 12000 8240 9990 2~190 lb970 20500

8 0,524 0,424 0,.423 0.720 0.634 0.631

Lp, ft 552 678 69S 632 02b ?19
FCD 2.b2 1.51 1.4b 2,23 i.21 1.01

Reduction of proppant diameter from 0.05 to 0.025 inches to prevent

bridging-out in park or all of the schedule.
.
.
sPE 15240

N -.-”__

. . . . . . .:-. ~- ----- .-—

?-—-___
,,.
M -——---::= . -- ——.- . .. .
H -.
,, A
L ----
J/u ...
..
.. ---- ---- . . . .. -
.
;: ,, ... ., i’
.q)k w! “ It

~M1--.--.. . . . . . . ---- ~,.-.

r
&

*
---
,~., : .’tJyl{ ?~~-. _.-. -.—- m---
..””
-. .- -. - .-.
.. —— -
.-
,. —....--
_.. ---
----
+$ . .. .. - .
---— -.. —— ---- ---- -. —.- -—. . .. .—----- ----

,{J ‘?,{ :{tiif+1Ii

“’t{.r
.- -:t

F,g. 1–F,,cl”,e
wWe,-nO 8tress C.”!,, St. E, 2 Ftg. 2—Fmcwre P,OI,I,S—1OWstress .ontr.st.
E,,4

y~i -
,., --- ——.
+ _
—- -_. ———. — —.. -.
.;? - -— —__
-—___
------ ------- --- ------- ---- ---- =---—_- ----
----.---

--- ---
: . . .. . . . ..-------- ------ ------ ---- =------- -----
——
- it;,. - _ ——.-—
. — —-— —- _—— ———
.—-— _— —

Fig. 3-F,8cIu,, )wdiks—medium stress

cc.nl,a.t. E..
E

.- i
N@~— ——_
------ ------- ------ ---- ----- - ------- ---
H -
GIIK
--:.””
----- ------- ---- ----- ---- - -=----= - ----
r
~oP -- ————
, _— .—
“

- Jhl-
—— —. .-— ..-. ——

F,g4—Fm.X.r.
p,ofil..—~g4
stress
contmsl.
E. 7
 .
.
xii- ?,#E 15 .24:()
r!
. .
i
j m
G --.-.:..-...---.--.-=— .- —-—----------
---
t, ..’-- .- ,

..
. ---- ---- ---- -,. - —- .— -. :: -
f .lgQ ..----’””
.—— “’ ‘- —— .—. . ---
t—
.

.—-—— -—. -. - -. .. ---- .. - ... . _

---- ..-,
. ---
.-..... ,>’
\ ‘~i ‘“”- ..
(! ; --- ““ /.
.- ----- ...
tj ----
, 4b - _._—.”

k.
— ——
l--

; ——
ii’”
1-

FractureprofI!e (JD!.;
389 .-

.
lee; ------ --.--- .-.. m.& -

Pay‘
zone
-UN : ------- -------- -----
1
-200+
Bank height : * I
‘~..
-300 *
——
I
“w+ f

Mle &H31Mt4B8
- w Me feoeoH
3wt
.
w 502

Fig. 7- tmnport ~-ad al pum$h& m. !2. !% O-Qm mnIQOll ~. 2X. 12.

5

SE i 524,0
.

.
Fractureprohle (CDK)
H ~ ~-— ___ Suspendedheight
; lee:
------ ------ ------‘—-------
------ -- -----y- i .----
i -:
/:
~ Pay
il- ; ~o~
.
t’ 4------- ------- ------- ------- -------- I . . . . .
t -lee J,
Bankheight
~’

109 al 4a5ee bee

m%!
- ft

t
Fractureprofile(GDK)
+
H’ ,
; laa*
------- ------- ------ ------ -------- - 1------
!
!:
1“ . Ba* heisht (closure) ~ Pay
e? -—~ Zow
, I
k I
L
f’ -.---- ------ .------ ------ ------ ------- ----

I
t
.

lee Zee
-ft 4WW6M
Im!M

I
fracture
profile (PM)

H SUSIWnded
heieht
/ lee+
------- -------- ------- ------- ---
G I

H
I
I
+
I
f ------- ------- ------- ------- -------- ----
.l~ ~
t
I Bankheight
J

100 a 4aiae Me
IJ?#-ft
 a

sit i 5 /4,() “-
Fractureprofile (?kN) -_ .
.
I

Mb +-
------- -------- -------- ----..”- -------- ----

~.
.- .. . . .
.; — -—.Bankheight:: lmre’
.. I
+—.-.
.
: u------ ------- .------ ------- ------- ---- _-,
- ~~
.,
+
! 1

* —.—
—— .—
M 280 w w 5aQ ~w
LI?WH
- ft

t -“--.. ..
A

“J& -
“——-
------------/’---”---
!mk hegn~_
_—-—
—------
-~ .-–
-.—
tee Zw 390 w! 6BB w
IJMIH-4!

~&
r-—___

—---wle
+

ME4 -------
------
------
-------
‘r_’’-’’’”
“--- ------

f --------------------------..-
/--’ --------
t -ml 4
~
k

lee ml 3U0 !iMiee7ell

WGIH-%