E. R, SIMONSON
MEMBER SIJE.AIME
A. S. ABOU-SAYED
R. J. CLIFTON
FEBRUARY,197S 27
●
the vertical or overburden stress. s With different stress analysis of the problem and deduce what
stresses between shales and sandstones, it becomes applied loading will produce Klc at the crack tip,
important to consider their effect on fracture This loading will cause further cracking in the
containment, material. A detailed discussion on a crack-length
Hydraulic fracture analysis is inherently a independent method of measuring fracture toughness
three-dimensional problem. Mathematical solutions on geologic material is given in Ref. 2. From this
are extremely complicated and sometimes impossible brief review of fracture mechanics concepts, the
in a great many cases. Three-dimensional solutions following three cases of hydraulic fracturing will
to some problems have been derived using the be considered: (1) the effect of different material
finite-element technique. 6 However, these solutions properties for the pay zone and the barrier
are usually very costly and extremely time consum- formation, (2) the characteristics of fracture
ing, Two-dimensional analyses are on firmvr ground propagation into regions of varying in~situ stress,
and many solutions of two-dimensional crack and (3) the effect of hydrostatic pressure gradients
problems have been worked out. Such simplified on fracture propagation into overlying or underlying
analyses provide considerable insight in under- barrier formations.
standing those parameters and conditions that
influence hydraulic fracture propagation. This CASE 1 — EFFECTS OF DIFFERING
study is limited to treating two-dimensional cracks MATERIAL PROPERTIES
in linear elastic media in which the primary fluid
It is well known that there are differences in
is assumed to act over the entire length of the
mechanical properties between the pay zone forma-
fracture. Furthermore, considerations are given
tion and the barrier formations. The question then is
only to symmetrically loaded cracks (Mode 1),
what role do mechanical properties play in the
A two-dimensional representation of a hydraulic
containment of the hydraulic fracture to the pay
fracture embedded in a sand layer bounded by
zones. The effect is best seen by examining how
barrier formations is shown schematically in Fig.
the stress intensity at the crack tip nearest the
3. The fracture is assumed infinite in extent normal interface ( Kf ) varies as the fracture approaches
to the plane of the page. The section under
the interface,% Fig. 5 illustrates the variation in
investigation is assumed far enough from the
the intensity factors for two cases, These cases
wellbore for its effects to be negligible. These
have been derived for the following set of
assumptions are reasonable when we consider that
mechanical properties,
most hydraulic fractures are assumed to have
lengths many times their heights.
In linear elastic fracture mechanics for Mode 1
cracks, the important parameters are the stress-
intensity factor (Kl) at the crack tip and the critical
value (Klc) of this factor at which crack extension
occurs, The former is a mathematical quantity that
uniquely characterizes the load sensed at the crack
tip. It is given by the limit as r+ O of the expression
for the normal stress component in the vicinity of
the crack tip’ (see Fig. 4); that is, FIG, 4 — STRESS COMPONENTS NEAR THE TIP OF A
-.-.—— ----- .
SHAKP CRACK.
K,=/JpFq&r,o). . . . . . . . (1)
.:-:-:-:-:-: UPPER L -
DIMENSIWESS
.-:-:-:-:-:-: BARR I E R GISFI STRESS
INTENSITY
FACTOR
PAY ‘32+2
ZONE K, (0) 2.0 $-
u
A m
G ‘ SHEAR MODULUS
~’ POISSON’S RA710
(.0 -
I I 1 1 1
0 0,2 04 06 00 0
OIMENSbNLESS DkTANCE “
FROM -u
‘----------- OARRIER +
p-u~ ~<y<e
G2 = 13.38 GPa (1’.94 x 106 psi) -L<y<L
p(y) = p-aa
p~ = 0.14 . . ...(3)
1 p-u~ -L<y <-t,
Case 1 indicates the stiffness of the barrier
Substituting from Eq. 3 in Eq. 2 and performing the
formation as measured by the shear modulus is less
integration gives the following expression for K]:
than the stiffness of the pay zone. For this case,
the stress-intensity factor (Kl)a + m as r/I’+ O. Thus,
the closer the fracture gets to the interface, the
easier it is to extend ar~d the crack eventually will
pass through the interface. In Case 2, however, the . . . . . . . . . . . . . . . . . . , (4)
stiffness of the barrier layer is greater than the Let P = L (1 + c), where c is the fractional distance
stiffness of the pay zone. For this case, the that the crack has propagated into the high stress
stress-intensity factor (Kl)a + O as r/f + 0. This region; then by rearranging terms we obtain
situ ~tion provides a “barrier” effect and tends to
art::st the crack at the interface. Therefore, better
hydraulic fracturing results can be expected from
hydrocarbon-bearing zones that have a lower
stiffness than the adjacent barrier formations.
. . . . . . . . . . . . . . . . . . (5)
CASE 2 — EFFECT OF IN-SITU
Ii p. is the pressure required for crack extension
STRESS VARIATIONS
when c = O, then
There may be differences in in-situ stress between
shales and sandstones. Consider the problem of a K/c
= (60 -Oa)@. .o. .o. o ., .(6)
crack embedded in a homogeneous isotropic medium
subjected to step variations in in-situ stress. This Subsriturion of Eq. 6 in Eq. 5 gives a relation
problem is one in which a hydraulic fracture, by berween P - PO and f when KI = KIC. The result is
some mechanism or other, has extended into
adjacent layers, where possibly different tectonic f
stresses are acting. Fig, 6 is a schematic repre-
sentation of this.
The stress-intensity factor at each end of the
crack is found by superposition of the two problems
shown in Fig, 7.
The stress-intensity factor, Kf, is the sum of the
contribution from Loadings A and B. The contribution Fig. 8 shows a plot of Eq. 7 in terms of excess
from Loading B is zero, and there is a contribution pressure, p - p. vs (!, the distan le crack has
only from Loading A. The value of KI can be advanced into the high stress : *.. The curves
computed directly from the followin~ equation. g~10 in this figure are for a crack heig t 61 m (2oO ft),
a fracture toughness of Klc, = 1.1 MN/m3f2 (1,000
Pe — lb-in.-s/2 ) and for parametric values of the in-situ
stress difference ab - Oa . Thus, if the stress
difference ~b - Oa = 7.58 MPa (1,100 psi), for
example, an overpressure of 3.45 MPa (500 psi)
would be expected if the fracture were to propagate
WSITU”
STRFSS
— 25 m into the region of higher in-situ stress. T~is
analysis,. although simplified, does indicate an
increase in fracture propagation pressure if the
1
fracture extends into a barrier formation with higher
El
in-situ stress, An accurate measure of the fracture
P 2L 2S
p-ffe
REqEII
Ub
2s r
2L p .ve +
‘a
L p-se
%
(h) (s)
FIG. 6 — VERTICAL HYDRAULIC FRACTURE
LOADED UNDER UNIFORM PRESSURE (p) WITH FIG. 7 — EQUIVALENT LOADINGS A AND B FOR THE
DIFFERING HORIZONTAL IN-SITU STRESS. PROBLEM SHOWN IN FIG. 6,
FEBRUARY,197S 2$
.
propagation pressure might cell when the fracture related linearly to the lateral stress 03. It seems
was extending into the barrier zone, provided there plausib~e that the lateral stress increases with
is a significant difference in in-situ stress between depth and that a linear stress gradient is reaIistic.
the barrier layer and the pay zone. The vaIue of the gradient is a topic of debate and
If the in-situ stress in the barrier layer (Ub) is probably will not be resolved, except by direct
less than the in-situ stress in the pay zone (era), a measurement. 13 For this problem, however, assume
situation would exist that required less pressure to there is a linear variation in the applied or tectonic
propagate the fracture in the barrier layer than in stress of go MPa/m as shown in Fig. 9.
the pay zone. Extensive propagation into the barrier The solution for (Kl)l and (Kl )2 again is found by
layer would be highly probable if Ob < aa. Unstable the superposition of the two loadings shown in
propagation into the barrier layers can be expected Fig. 10.
if f’ becomes large enough, even for ~& > Oa. This The contribution to K{ from Loading B is zero.
effect is shown in Fig. 8, where for ab - Ua = 0.34, The only contribution comes from Loading A. The
the pressure, p -p., reaches a maximum at f?’= 160 values for (KI)2 and (Kl)l are calculated as before
m and then decreases with increasing f’. from Eq. 2 with
4
u,
- = b,, . . . . . . . . . . . . . (8)
(?3 P
where
p = Poisson’s ratio.
Eq. 8 shows that the overburden stress al is
2.05
1,3s
0.69
0.34
oR=- 1
that down ward migration is probable. Conversely, or downward migration, the source of pressurization
if pa > gp, then (Kl)l > (K1)2 and upward migration (perforation holes) would be “covered up, ” and
is most probable. Thus, it is conceivable that crack pressurization would not be possible. It is
vertical motion of the crack could be controlled by conceivable that as the crack extends away from
—
using hydraulic fracture fluids with various densi- the wellbore, it would propagate upward and
ties, depending on whether upward (g. > gp), downward at fixed height because near the wellbore
downward (g ~ > ga), or both upward and downward the propping agent might hold the crack open,
(equal probability, gp = ~) migration is desired. allowing fluid to flow out inro the fracture. The
Another interesting observation can be made by important thing to note is that preferred upward or
further examination of Eq. 10. If the difference in downward crack migration is entirely possible. By
stress-intensity factors (KI)2 - (Kl)l is equal to adjustment of the hydraulic fracture fluid density,
Klc, then crack extension would be certain for the probability of producing horizontality propagating
(KI)2 > (Kl)l > 0. The relationship between this fractures can be maximized. To dc this, the
difference and the difference in gp and ga is given variation of in-situ stress with depth must be
by determined.
t?p -
Klc
‘u=iJz ’””’””’”””’
, (11) CONCLUSIONS
Three cases of hydraulic fra ;ture containment
Fig. 11 shows a plot of this equation for various were discussed in relationship to linear elastic
values of the fracrure roughness Kl=. Consider two fracture mechanics. Analyses of fracture contain-
cases, both of which give the same value for ]gp - ment as a two-dimensional problem yielded several
~ I = 5.7 kPa/m. Case 1 assumes the pressure fundamental results that can be applied in general
gradient is caused by water pressure on the crack to the design of massive hydraulic fractures. The
face, gp ==9.73 kN/m (0.43 psi/ft) and Case 2 following conclusions resulted from these analyses.
assumes a 21.2 -kN/m3 (18 lb/gal) mud is pressur- 1. Hydraulic fractures in a pay zone located
izing the crack, gp = 21.2 kPa/m (0.93 psi/ft). For between two adj scent barrier layers tend to be
both cases, the tectonic str~ss variation is 15.5 contained, provided the stiffness of the pay zone
k Pa/m (0.68 psi/ft), Case 1 corresponds to upward is less than the stiffness of the barrier layers, If
migration of the crack and Case 2 corresponds to the opposltt condition exists, barrier penetration
downward migration. The crack height for which is likely.
Igp - %1 = 5.7 kPa/m and K}= = 1.65 MN/m3/2 2. Migration of a hydraulic fracture, either upward
(1,500 lb-in. ‘3/2) is approximately 54,8 m (180 ft), or downward in an isotropic, homogeneous medium
For crack heights less than this value, AKI = I(KJ)2 may be controlled by the derisity of hydraulic
- (Kl) 1 I is less than Klc. For crack heights fracturing fluid. If the fluid density gradient is
exceeding 54.8 m (180 ft), the stress-intensity factor greater (less) than the minimum horizontal in-situ
is equal to Klc at one end of the crack and is stress gradient, then downward (upward) migration
negative at the other end. Negative values for K1 is probable.
imply the crack is closed and the crack height must 3. If there is a difference in in-situ stress
be reduced to the point where KI = 0, Therefore, the between the bounding layers and the pay zone, with
crack reaches a critical height, at which point a greater in-situ stress in the bounding layers, then
crack of constant height propagates either upward or those layers serve as a barrier to vertical extension
downward, depending on the value of gp -~. In the of the fracture. It may be possible to detect fracture
field, this cannot occur to any great extent near the propagation into the barrier formation by an increase
wellbore because as the crack closes during upward in pumping pressure, occurring as the fracture
crosses the interface and extends into the barrier
40 [ 1
layer. The pressure increase for a given extension
I
K,c. 1.65 MN/ms’a into the barrier is a‘ funcrion of the difference in
n’ in-situ stress between the barrier and the pay zone
30 I*I
● layers and the height of the pay zone.
1
09 -tlr +J
The mechanical properties of the pay zone and
!lp- % the barrier formation, as well as the minimum
(kPo/rn) 20 [\ horizontal in-situ stresses for these layers, play
important roles in the prediction of hydraulic
fracture containment.
NOMENCLATURE
g = acceleration of gravity
“CRACK LENGTH (m)
gp = pressure gradient caused by fluid pressure
FIG. 11 — PLOT OF THE DIFFERENCE BETWEEN go = stress gradient caused by variation in in-situ
STRESS AND PRESSURE GRADIENTS VS CRACK
LENGTH Wf?H FRACTURE TOUGHNESS (&c) AS A
stress with depth
PARAMETER, G = shear modulus
FSBRUARY,191B
M
w