Professional Documents
Culture Documents
Hubbert and Willis, 1972 Mechanics of Hydr Frac PDF
Hubbert and Willis, 1972 Mechanics of Hydr Frac PDF
W a s h i n g t o n , D.C., and A t h e r t o n , C a l i f o r n i a
Abstract A theoretical e x a m i n a t i o n of the f r a c t u r i n g bedding plane and lifted the overburden, not-
of rocks by means of pressure a p p l i e d in boreholes withstanding the fact that, in the great majority
leads t o the conclusion that, regardless of w h e t h e r the
f r a c t u r i n g f l u i d is of the p e n e t r a t i n g or n o n p e n e t r a t i n g
of cases where pressures were known, they
t y p e , the fractures produced should be a p p r o x i m a t e l y were significantly less than those due to the to-
p e r p e n d i c u l a r to the axis of least stress. The general tal weight of the overburden as determined
state of stress u n d e r g r o u n d is that in which the three from its density.
p r i n c i p a l stresses are u n e q u a l . For tectonically r e l a x e d
areas characterized by normal f a u l t i n g , the least stress Prior to 1948, this prevalent opinion had al-
should b e h o r i z o n t a l ; the fractures p r o d u c e d should be ready been queried by Dickey and Andresen
v e r t i c a l , a n d the injection pressure should be less than (1945) in a study of pressure parting and by
that o f the o v e r b u r d e n . In areas of active tectonic
compression, the least stress should be vertical a n d
Walker (1946, 1949), who, in studies of
equal to the pressure of the o v e r b u r d e n ; the fractures squeeze cementing, pointed out that the pres-
should b e h o r i z o n t a l , a n d injection pressures should be sures required were generally less than those of
equal t o , or g r e a t e r t h a n , the pressure o f the o v e r b u r d e n . the overburden, and inferred that the fractures
Horizontal fractures cannot be p r o d u c e d by hydraulic should be vertical. J. B. Clark (1949), in his
pressures less than the total pressure of the o v e r b u r d e n .
paper introducing hydraulic fracturing, and
These conclusions are c o m p a t i b l e w i t h field experience
in f r a c t u r i n g a n d w i t h the results of l a b o r a t o r y ex- later Howard and Fast (1950) and Scott et al.
perimentation. (1953), postulated that the entire weight of the
overburden need not be lifted in producing hor-
INTRODUCTION
izontal fractures, but that it is only necessary to
lift an "effective overburden," requiring a cor-
The hydraulic-fracturing technique of well respondingly lower pressure. Hubbert (1953a)
stimulation is one of the major developments in pointed out that the normal state of stress un-
petroleum engineering of the last decade. The derground is one of unequal principal stresses
technique was introduced to the petroleum in- and that, in tectonically relaxed areas charac-
dustry in a paper by J. B. Clark of the Stano- terized by normal faults, the least stress should
lind Oil and Gas Company in 1949. Since then, be horizontal. Therefore, in most cases, fractur-
its use has progressively expanded so that, by ing should be possible with pressure less than
the end of 1955, more than 100,000 individual that of the overburden; moreover, such frac-
treatments had been performed. tures should be vertical. Harrison et al. (1954),
The technique itself is mechanically related also on the expectation that the least principal
to three other phenomena concerning which an stress should be horizontal, argued strongly in
extensive literature had previously developed. favor of vertical fracturing.
They are: (1) pressure parting in water-injec- Scott et al. (1953) observed that, by the use
tion wells in secondary-recovery operations, of penetrating fluids, hollow, cylindrical cores
(2) lost circulation during drilling, and (3) the could be ruptured at less than half the pres-
breakdown of formations during squeeze-ce- sures required if nonpenetrating fluids were
menting operations—all of which appear to in- used. They also observed that, with penetrating
volve the formation of open fractures by pres- fluids, the fractures occurred parallel with the
sure applied in a well bore. The most popular bedding, irrespective of the orientation of the
interpretation of this mechanism has been that bedding with respect to the axis of the core;
the pressure had parted the formation along a whereas, with nonpenetrating fluids, the frac-
1
tures tended to be parallel with the axis of the
Manuscript received, July 4, 1972. Published, in core. Reynolds et al. (1954) described addi-
revised form, with permission of Society of Petroleum tional experiments confirming the earlier work
Engineers of AIME; originally published in Transac-
tions of Society of Petroleum Engineers of AIME, 1957, of Scott et al. (1953) and concluded that it
v. 210, p. 153-168. should be possible to produce horizontal frac-
2 tures with penetrating fluids and vertical frac-
U.S. Geological Survey.
3 tures with nonpenetrating fluids.
Consultant.
239
240 M. King Hubbert and David G. Willis
w^*
c which is necessary to displace the upper box is
measured for various values of vertical stress.
In this way, it is found that the shearing stress
for failure is directly proportional to the nor-
\ o-
mal stress, or that
tan , (2)
loose sand, <f> is approximately 30°. These criti- reaches a value which is about three times the
cal stress values may be plotted on a Mohr dia- least principal stress, and that the failure will
gram, as shown in Figure 8. The two diagonal occur along a plane making an angle of about
lines comprise the Mohr envelopes of the mate- 60° to the least principal stress. Also, for a
rial, and the area between them represents sta- fixed vertical stress trz, the horizontal stress may
ble combinations of shear stress and normal have any value between the extreme limits of
stress, whereas the area exterior to the enve- one third and three times uz.
lopes represents unstable conditions. Figure 8
thus indicates the stability region within which M O H R D I A G R A M FOR ROCKS
the permissible values of <r and T are clearly de- The foregoing theoretical analysis is directly
fined. The stress circles may then be plotted in applicable to solid rocks provided the Mohr en-
conjunction with the Mohr envelopes to deter- velopes have been experimentally determined.
mine the conditions of faulting. Such plots are In order to make such determinations, it is nec-
illustrated in Figure 9 for both normal and essary to subject rock specimens to a series of
thrust faulting. In both cases, one of the princi- triaxial compression tests under wide ranges of
pal stresses will be equal to the overburden values of greatest and least principal stresses a1
pressure, or o> In the case of normal faulting, and CT3. It has been found that, at sufficiently
the horizontal principal stress is progressively high pressures, nearly all rocks deform plasti-
reduced, thereby increasing the radius of the
stress circle until it becomes tangent to the
Mohr envelopes. At this point, unstable condi-
tions of shear and normal stress are reached,
and faulting occurs on a plane making an angle
of 45° + <f>/2 with the least stress. For sand
having an angle of internal friction of 30°, the
normal fault would have a dip of 60°, which
agrees with the previous experiments. For the
case of thrust faulting, the least principal stress
would be vertical and would remain equal to
the overburden pressure, whereas the horizon-
tal stress would increase progressively until un-
stable conditions occurred and faulting took
place on a plane making an angle of 45° -f <p/2
with the least principal stress, or 45° — <£/2
with the horizontal. For sand, this would be a
dip of about 30°, which also agrees with the
experiment.
It can be seen that, for sand having an angle
of internal friction of 30°, failure will occur in FIG. 8—Mohr envelopes for sand showing curves
either case when the greatest principal stress of values of a and T at which slippage occurs.
Mechanics of Hydraulic Fracturing 243
FIG 9—Mohr diagram showing possible range of horizontal stress for a given vertical stress as. Horizontal
stress can have any value ranging from approximately one third vertical stress, corresponding to normal fault-
ing, to approximately three times vertical stress, corresponding to reverse faulting.
cally and the Mohr envelopes become approxi- ues usually between 20° and 50°, and most
mately parallel with the <r-axis. However, at commonly not far from 30°, and T0 is the
lower pressures, most rocks fail by brittle frac- shearing strength of the material for zero nor-
ture, and within this domain the envelopes are mal stress (Fig. 10).
approximated by the equation Fortunately, equation 3 is applicable to most
r — + (j„ + a tan <j>), (3)
rocks within drillable depths. Exceptions would
occur in the cases of the plastic behavior of
where the angle of internal friction, <j>, has val- rock salt and unconsolidated clays.
One additional modification in the theoreti- lated "effective overburden pressure" invoked
cal analysis is needed before it is directly appli- by the Stanolind group and others to justify
cable to geologic phenomena. Sedimentary their assumption of horizontal fracturing in re-
rocks are both porous and permeable, and the : r sponse to pressure less than that of the overbur-
pore spaces are almost invariably occupied by den.
fluids, usually water, at some pressure, p. It is Therefore, the entire Mohr stress analysis is
necessary to know the effect which is produced directly applicable to porous rocks containing
by the fluid pressure upon the mechanical prop- fluid under pressure, provided that only effec-
erties of the rock. tive stresses are used.
This question was specifically investigated by It is interesting to consider the behavior of
McHenry (1948), who ran a large series of the effective vertical stress under various fluid-
tests on duplicate specimens with and without pressure conditions. Under any condition, the
enclosure by impermeable jackets, using nitro- total vertical stress, S,, is nearly equal to the
gen gas to produce the pressure (p). He found weight of the overlying material per unit area.
for the unjacketed specimens that, when the ax- The effective vertical stress, <rz, however, is
ial compressive stress, S, was corrected for the given by
opposing fluid pressure, p, the value of the re-
sidual effective stress, o- = S — p, at which fail- o,=&,- p. (5)
ure occurred was to a close approximation con- Under normal hydrostatic conditions, p is
stant and independent of the pressure (p) of somewhat less than half the total pressure of
the permeating fluid. the overburden, and the effective vertical stress
This result is directly applicable to the be- is therefore slightly more than half the overbur-
havior of rocks underground. Porous sedimen- den pressure. However, with abnormally high
tary rocks are normally saturated with fluid un- fluid pressures, such as occur in some parts of
der pressure and constitute a mixed solid-fluid the Gulf Coast, the effective vertical stress is
stress system. The stress field existing in this correspondingly reduced, and, in the extreme
system may be divided into two partial stresses: case of fluid pressure equal to the total over-
(1) the hydrostatic pressure, p, which pervades burden pressure, the effective vertical stress be-
both the fluid and solid constituents of the sys- comes zero.
tem, and (2) an additional stress in the solid In regard to the mechanical properties of
constituent only. The total stress is the sum of rocks, for loosely consolidated sediments such
these two. as those of the Gulf Coast area, the limiting en-
If, across a plane of arbitrary orientation, S velopes on the Mohr diagram will approximate
and T are the normal and tangential compo- those for loose sand shown in Figure 8. In
nents, respectively, of the total stress, and o- older and stronger rocks, the Mohr envelopes
and T the corresponding components of the are given approximately by equation 3. These
solid stress, then, by superposition, envelopes are similar to those for loose sand ex-
cept, as shown in Figure 10, they project to an
S = <r + V, and
(4)
intersection at some distance left of the origin,
T
1 = T indicating that the rocks have some degree of
are the equations relating the stress fields. tensile strength and also have a shear strength
The pressure (p) produces no shearing stress of finite magnitude T„ when the normal stress is
and hence has no tendency to cause deforma- equal to zero. The Mohr envelopes for tests on
tion. Moreover, as demonstrated by the work a sandstone and an anhydrite made by J. W.
of McHenry, it has no significant effect on the Handin of the Shell Development Exploration
properties of the rock. Therefore, with respect and Production Research Laboratory are
to the stress of components <x and r, the rock shown in Figures 11 and 12.
has the same properties underground as those In either case, however, it will be observed
exhibited in the triaxial testing machine using that, at other than shallow depths, the value of
jacketed specimens. cr3, the least stress, at the time of faulting will
This fact long has been recognized in soil be of the order of one third the value of o^, the
mechanics, where the partial stress of compo- greatest stress.
nents a and T is known as the effective stress Because these are the extreme states of stress
and the pressure (p) as the neutral stress (Ter- at which failure occurs, it follows that the stress
zaghi, 1943). The effective stress defined in this differences which prevail when actual faulting
manner is not to be confused with the postu- is not taking place are somewhat less than these
Mechanics of Hydraulic Fracturing 245
limits. However, in most regions a certain type However, in regions which are being short-
of deformation is usually repetitive over long ened, either by folding or thrust faulting, the
geologic periods of time, indicating that the least stress should be vertical and equal to the
stresses of a given type are psrsistent and not effective pressure of the overburden, whereas
far from the breaking point most of the time. the greatest stress should be horizontal and
The orientation of the trajectories of the probably between two and three times the
principal stresses in space is largely determined effective overburden pressure.
by the condition which they must satisfy at the In regions of transcurrent faulting, both the
surface of the earth. This is a surface along greatest and least stresses should be horizontal,
which no shear stresses can exist. For unequal and the intermediate stress, <r2, should equal the
stresses, the only planes on which the shear effective vertical stress.
stresses are zero are those perpendicular to the As an example, the Tertiary sedimentary
principal stresses; therefore, one of the three strata of the Texas and Louisiana Gulf Coast
trajectories of principal stress must terminate have undergone recurrent normal faulting
perpendicular to the surface of the ground, and throughout Tertiary time and to the present.
the other two must be parallel with this surface. Thus, a normal-fault stress system must have
Thus, in regions of gentle topography and sim- been continuously present, which intermittently
ple geologic structures, the principal stresses reached the breaking points for the rocks, caus-
should be, respectively, nearly horizontal and ing the stresses temporarily to relax and then
vertical, with the vertical stress equal to the gradually to build up again. Hence, during
pressure of the overlying material. most of this time, including the present, a stress
Therefore, in geologic regions where normal state must have existed in this region for which
faulting is taking place, the greatest stress, alt the least stress has been horizontal and proba-
should be approximately vertical and equal to bly between one half and one third of the effec-
the effective pressure of the overburden, tive pressure of the overburden. Because the
whereas the least stress, cr3, should be horizon- faults in this area, except around salt domes,
tal and most probably between one half and are mostly parallel with the strike of the rocks,
one third the effective pressure of the overbur- the axis of least stress must be parallel with the
den. dip.
/ ^
4000
2000
x
/ / \
A| / \\
v*
\
1 1 1 1 i
C 2000 4000 6000 8000 IOP0O 12,000 14,000
KG/CM2
FIG. 11—Mohr envelopes for Oil Creek Sandstone (measurements by John Handin).
246 M. King Hubbert and David G. Willis
BLAINE ANHYDRITE
(DRY,24°C, 0 - 2 0 0 0 ATM PRESSURE)
22°-43°
4000 5000
KG/CM!
A large part of the region of West Texas and analyze the stress conditions around the bore-
the Mid-Continent is a region of tectonic relax- hole and to determine the actual conditions un-
ation characterized by older normal faults. The der which hydraulic tension fractures will be
situation here is somewhat more ambiguous formed.
than that of the Gulf Coast because faulting in
these regions is not now active. However, be- STRESS DISTORTIONS CAUSED BY BOREHOLE
cause evidence of horizontal compression is The presence of a well bore distorts the
lacking, it is still reasonable to assume that a preexisting stress field in the rock. An approxi-
relaxed stress state in these areas is the more mate calculation of this distortion may be made
probable one at present. by assuming that the rock is elastic, the bore-
In contrast, California is in a region where hole smooth and cylindrical, and the borehole
active tectonic deformation is occurring at axis vertical and parallel with one of the preex-
present, as indicated by the recurrence of earth- isting regional principal stresses. In general,
quakes, by extensive folding and faulting of the none of these assumptions is precisely correct,
rocks during Holocene time, and by slippages but they will provide a close approximation to
along faults and measurable movements of ele- the actual stresses. The stresses to be calculated
vation bench marks during the last few de- should all be viewed as the effective stresses
cades. All three of the types of stress pattern carried by the rock in addition to a hydrostatic
described probably occur in different parts of fluid pressure, p, which exists within the well
this region; but, in areas still undergoing active bore as well as in the rock. The calculation is
compression, the greatest stress must be essen- made from the solution in elastic theory for the
tially horizontal, whereas the least stress would stresses in an infinite plate containing a circular
be the effective weight of the overburden. hole with its axis perpendicular to the plate.
It should be understood that the foregoing This solution was first obtained by Kirsch
analysis of faulting is used only as a means of (1898) and later was given by Timoshenko
estimating the state of stress underground, and (1934) and by Miles and Topping (1949).
that the shearing mechanism of faulting is quite Expressed in polar coordinates with the cen-
distinct from the mechanism of producing hy- ter of the hole as the origin, the plane-stress
draulic fractures, which are essentially tension components at a point 6,r, exterior to the hole
phenomena. However, an understanding of the in a plate with an otherwise uniform uniaxial
regional subsurface stresses makes it possible to stress, dA, are given by
Mechanics of Hydraulic Fracturing 247
+
5V 7
*Y^\~
*r,^
Fin. 13—Superposition of stress states about a well bore as a result of two horizontal
principal stresses of equal magnitude.
248 M. King Hubbert and David G. Willis
trj»—
<i*0- B «4C
(0)
(t>)
FIG. 18—Idealized diagram of two possible types of pressure behavior during fracture treatment
depending upon various underground conditions.
ing or to a vertical fracture in a situation where upon both the rock and the contained fluid in
the ratio <rB/aA of the horizontal principal each of the walls of the fracture. Let the nor-
stresses was greater than 2.0. mal component of this force acting upon the
rock content of unit bulk volume be Hnl and
Effect of Penetrating Fluids that upon the fluid content, Hn2. Since the vol-
When a penetrating fluid is used in a fractur- ume of the fluid per unit bulk volume is the po-
ing operation, a more complicated mechanical rosity, /, and that of the rock (1 — / ) , it
situation exists. As noted previously, the total follows that
normal stress S across any plane may be re- 3 (Am)
solved into the sum of a residual solid stress o- Hnl = - ( l - . / J - U l i , and (8)
and the fluid pressure p, or S — a + p. dn
Furthermore, with a nonpenetrating fluid, an d(Ap)
H„ -f (9)
increase in pressure (Ap) equal to <r, or a total dn
pressure P — p + Ap equal to S, is required to However, because of viscous coupling, the
hold open and extend a fracture along this force Hn2 acting upon the fluid is transmitted
plane. entirely to the rock, so that the total outward
For the case of a penetrating fluid, an incre- force exerted upon the rock per unit of bulk
ment of pressure in the fracture, which now volume will be:
may be designated by Apn, will produce an out-
a(Ap)
ward flow of fluid into the rock with a resulting H„ = H„i + Hni (10)
variable increment Ap to the pressure within dn
the formation. The gradient of this incremental Similarly, the total o u t w a r d force p e r unit
pressure will exert an outward-directed force a r e a of t h e fracture wall will be the integral of
Mechanics of Hydraulic Fracturing 251
all the forces exerted upon the rock contained whether the fluid is penetrating or nonpenetrat-
within a column of unit area of cross section ing, is also equal to the overburden pressure. It
normal to the fracture, or thus appears to be mechanically impossible for
F r" d(Av) r° horizontal fractures to be produced with total
T = - -^-dn = - \cl(Ap) = Apo. (11) fluid pressures less than the total overburden
.4 Jo dn J Apo
pressure.
In order for the fracture to be held open and Because the great majority of fracturing op-
extended, this outward-directed force per unit erations in the Gulf Coast, Mid-Continent, and
area must be equal to <r. Therefore, for this West Texas-New Mexico regions require injec-
case tion pressures less than the overburden pres-
P = p + Ap = p + a = S, (12) sure, it is difficult to escape the conclusion that
which is exactly the same as the pressure re- most of these fractures are vertical. Further-
quired to hold the fracture open when a non- more, because the minimum pressures should
penetrating fluid is used. be independent of the fluids used, there appears
In the case of radial flow away from a well to be no valid basis for the claims that vertical
bore, the situation differs somewhat from that versus horizontal fracturing can be controlled
of flow away from a plane fracture. In the ra- by variations in the penetrability of the fractur-
dial-flow case, a force acts outward whose mag- ing fluids. In either case, it appears, the orienta-
nitude per unit bulk volume is tion of the fractures should be controlled by the
preexisting stress field of the rocks into which
H = - grad (Ap), (13)
the fluid is injected.
and the effect of this distributed field of force is
to diminish the stress concentration at the face PREDICTED I N J E C T I O N PRESSURES
of the hole, in turn reducing the excess pressure It is interesting to estimate the actual values
that otherwise would be required to produce of the minimum injection pressures under con-
breakdown. Once the fracture is started, how- ditions of incipient normal faulting such as
ever, the flow field and the stress field become may exist in many parts of the Gulf Coast area.
those associated with a plane fracture given in As has been pointed out (equation 5 ) , the
equation 12. undisturbed effective vertical stress az is equal
Therefore, the only effect of using a pene- to the total pressure of the overburden Sz less
trating fluid is in the reduction of the break- the original fluid pressure p existing within
down pressure. The minimum injection pres- the rocks prior to disturbances such as fluid
sure, for both penetrating and nonpenetrating withdrawals.
fluids, must be greater than the preexisting nor- Under conditions of incipient normal fault-
mal stress across the plane of the fracture. ing, the least principal stress, <rA, will be hori-
zontal and will have a value of approximately
Orientation of Fractures Produced one third the effective overburden pressure, <rz.
Considering the earlier postulate that the Therefore,
fractures should occur along planes normal to «A. 9* (& - p)/3. (H)
the least principal stress, the minimum injection Because the additional fluid pressure, Ap, re-
pressure should thus be equal to the least prin- quired to hold open and extend a fracture
cipal stress. Considering the injection pressures should be equal to the least principal stress,
and fracture orientations for various tectonic then
conditions, it follows that, in regions character-
ized by active normal faulting, vertical frac- Ap ^ (& - p)/3. (15)
tures should be formed with injection pressures However, the total injection pressure, P, is
less than the overburden pressure; whereas, in given by
regions characterized by active thrust faulting, P = Ap + p. (16)
horizontal fractures should be formed with in-
jection pressures equal to, or greater than, the Therefore,
overburden pressure. P si (S. + 2p)/3. (17)
In the particular case of horizontal fractur- Dividing by depth z gives
ing, the total normal stress across the plane of
P/z S* (S./* + 2p/z)/3, (18)
the fracture is equal to the pressure due to the
total weight of the overburden; therefore, the which is the approximate expression for the
minimum injection pressure, regardless of minimum injection pressure required per unit
252 M. King Hubbert and David G. Willis
of depth in an area of incipient normal fault- tions in the Gulf Coast, Mid-Continent, and
ing. West Texas regions. With but few exceptions,
The value of Sz/z is approximately equal to the injection pressures have been substantially
1.00 psi per foot of depth for normal sedimen- less than the total overburden pressure, thus
tary rocks in most areas. Under normal hydro- implying that vertical fractures are actually be-
static fluid-pressure conditions, p/z is about ing formed.
0.46 psi per foot of depth. Substituting these In addition to the preceding data, the occur-
values into equation 18 gives rence of lost circulation throughout the Gulf
Coast area at pressure substantially less than
P/z ^ 0.64 psi/ft
that due to the weight of the overburden sup-
as the approximate minimum value that should ports the conclusion that the least stress should
be expected in the Gulf Coast. be horizontal in this area.
Let us consider the values of P which would In much of California, however, tectonic
occur under conditions in which the original compression is taking place, and in these areas
fluid pressure was other than hydrostatic. In horizontal fractures should occur with injection
those cases of an original fluid pressure less pressures greater than the total overburden
than hydrostatic, it can be seen from equation pressure. Although comparatively few fractur-
17 that P would be correspondingly reduced. ing operations have been performed in Califor-
However, where abnormally high original fluid nia, extremely high pressures are required with
pressures prevail, P would become higher until, injection pressures commonly greater than the
in the limit, when the original pressure p ap- overburden pressure (W. E. Hassebroek, per-
proaches the total overburden pressure Sz, P sonal commun.).
also approaches the total overburden pressure A phenomenon very similar to artificial for-
and fracturing will occur at pressures only mation fracturing, but on a much larger scale,
slightly greater than the original fluid pressure. is that of dike emplacement. It has been
Walker (1949) has described an interesting pointed out by Anderson (1951) that igneous
example of lost circulation which might be ex- dikes should be injected along planes perpen-
plained on the basis of the foregoing analysis. dicular to the axis of least principal stress. This
In a Gulf Coast well drilling below 10,000 ft situation is entirely analogous to that for artifi-
(3,050 m ) , the specific weight of the drilling cial formation fracturing. A remarkable field
mud, which was a little over 18 lb/gal, had to example of the effect of a regional stress pat-
be kept constant to within 0.3 lb/gal, or about tern upon the orientation of igneous dikes is the
2 percent, to prevent either lost circulation Spanish Peaks igneous complex in Colorado.
when the density was too high or "kicking" by A map of this area is shown in Figure 19,
the formation fluids when the density was too and a photograph of West Spanish Peak from
low. the northwest, showing dikes cutting flat-lying
Eocene strata, is given in Figure 20. Ode
F I E L D EVIDENCE
(1957) has made a mathematical solution of
Present field data derived from experience the regional stress field which would most likely
with hydraulic fracturing, squeeze cementing, result from the presence of the structural fea-
and lost circulation are fully consistent with the tures in the area. A comparison of the radial-
foregoing conclusions. In the Gulf Coast area, dike system with the mathematical solution
recent normal faulting indicates that vertical shows the dikes to be almost exactly perpendic-
fractures should be formed with injection pres- ular to the trajectories of the least principal
sures less than the total overburden pressure. In stress.
the Mid-Continent and West Texas regions, old
normal faulting, although representing more EXPERIMENTAL FRACTURING DEMONSTRATION
ambiguous evidence, also favors vertical In order to verify the inferences obtained
fracturing. theoretically, a series of simple laboratory ex-
Howard and Fast (1950) have summarized periments has been performed. The general
the pressure data from 161 squeeze-cementing procedure was to produce fractures on a small
and acidizing jobs performed in the Gulf Coast scale by injecting a "fracturing fluid" into a
and West Texas-New Mexico areas. Also, pub- weak elastic solid which previously had been
lished data by Harrison et al. (1954) and Scott stressed. Ordinary gelatin (12-percent solution)
et al. (1953) describe injection pressures for was used for the solid, because it was suffi-
large samples from hydraulic-fracturing opera- ciently weak to fracture easily, was readily
Mechanics of Hydraulic Fracturing 253
Warn.
FIG. 20—Photograph of West Spanish Peak from northwest, showing dikes cutting flat-lying Eocene strata
(G. W. Stose, U.S. Geological Survey).
254 M. King Hubbert and David G. Willis
values of the flow velocity, q, for this case will single field be vertical, but ihey also should
be very much less than for the radial-flow case. have roughly the same direction of strike.
Hence, for a given rate of oil production, a ver- 7. Vertical fractures intersecting horizontal
tical fracture across the oil-water interface in a permeability barriers will facilitate the vertical
uniform sandstone, instead of causing an in- flow of fluids. However, in the absence of such
crease of water production, actually should barriers, vertical fractures across the oil-water
serve to reduce markedly the water coning and, or gas-oil interface will tend to reduce the con-
consequently, to decrease the production of wa- ing of water or gas into the oil section for a
ter—a result in accord with reports of field ex- given rate of oil production.
perience wherein fracturing near the water ta-
ble has not resulted in increased water produc- REFERENCES
tion. Anderson, E. M., 1951, The dynamics of faulting and
dyke formation with applications to Britain: Edin-
burgh and London, Oliver and Boyd, 2d ed.
CONCLUSIONS Bugbee, J. M., 1943, Reservoir analysis and geologic
From the foregoing analysis of the problem structure: AIME Trans., v. 151, p. 99.
Clark, J. B., 1949, A hydraulic process for increasing
of hydraulic fracturing of wells, the following the productivity of wells: AIME Trans., v. 186, p.
general conclusions appear to be warranted. 1-8.
Crawford, P. B„ and R. E. Collins, 1954, Estimated
1. The state of stress underground is not, in effect of vertical fractures on secondary recovery:
general, hydrostatic, but depends upon tectonic AIME Trans., v. 201, p. 192.
conditions. In tectonically relaxed areas charac- Dickey, P. A., and K. H. Andresen, 1945, The behav-
terized by normal faulting, the least stress will ior of water-input wells: Drilling and Prod. Prac-
tices, v. 34.
be approximately horizontal; whereas, in areas Harrison, Eugene, W. F. Kieschnick, Jr., and W. J.
of tectonic compression characterized by fold- McGuire, 1954, The mechanics of fracture induction
ing and thrust faulting, the least stress will be and extension: AIME Trans., v. 201, p. 252.
approximately vertical and, provided the defor- Howard, G. C , and C. R. Fast, 1950, Squeeze cement-
mation is not too great, approximately equal to ing operations: AIME Trans., v. 189, p. 53.
Hubbert, M. K., 1940, The theory of ground-water mo-
the overburden pressure. tion: Jour. Geology, v. 48, p. 785-944.
2. Hydraulically induced fractures should be 1951, Mechanical basis for certain familiar geo-
formed approximately perpendicular to the logic structures: Geol. Soc. America Bull., v. 62, no.
4, p. 355-372.
least principal stress. Therefore, in tectonically 1953a, Discussion of paper by Scott, Bearden,
relaxed areas, they should be vertical, whereas, and Howard, "Rock rupture as affected by fluid
in tectonically compressed areas, they should properties": AIME Trans., v. 198, p. 122.
be horizontal. 1953b, Entrapment of petroleum under hydrody-
namic conditions: Am. Assoc. Petroleum Geologists
3. Rupture or breakdown pressures are af- Bull., v. 37, no. 8, p. 1954-2026.
fected by the values of the preexisting regional Hunter, Z. Z., 1956, 8V2 Million extra barrels in 6
stresses, by the hole geometry including any years: Oil and Gas Jour., August 27, p. 86.
preexisting fissures, and by the penetrating Kirsch, G., 1898, Die Theorie der Elastizitat und die
quality of the fluid. Bediirfnisse der Festigkeitslehre: Zeitschr. Ver.
Deutsch. Ingenieure, v. 42, p. 797.
4. Minimum injection pressures depend McHenry, Douglas, 1948, The effect of uplift pressure
solely upon the magnitude of the least principal on the shearing strength of concrete: Troisieme Con-
regional stress and are not affected by the hole gres des Grands Barrages, Stockholm, Sweden.
geometry or the penetrating quality of the fluid. Miles, A. J., and A. D. Topping, 1949, Stresses around
a deep well: AIME Trans., v. 179, p. 186
In tectonically relaxed areas, the fractures Ode, H., 1957, Mechanical analysis of the dike pattern
should be vertical and should be formed with of the Spanish Peaks area, Colorado: Geol. Soc.
injection pressures less than the total overbur- America Bull., v. 68, no. 5, p. 567-575.
den pressure. In tectonically compressed areas, Reynolds, J. J., P. E. Bocquet, and R. C. Clark, Jr.,
provided the deformation is not too great, the 1954, A method of creating vertical hydraulic frac-
tures: Drilling and Prod. Practice, p. 206.
fractures should be horizontal and should re- Scott, P. P., Jr., W. G. Bearden, and G. C. Howard,
quire injection pressures equal to, or greater 1953, Rock rupture as affected by fluid properties:
than, the total overburden pressures. AIME Trans., v. 198, p. 111.
Terzaghi, Karl, 1943, Theoretical soil mechanics: New
5. It does not appear to be mechanically York, John Wiley and Sons.
possible for horizontal fractures to be produced Timoshenko, S., 1934, Theory of elasticity: New York
in relatively undeformed rocks by means of to- and London, McGraw-Hill.
tal injection pressures which are less than the Walker, A. W., 1946, Discussion of paper by A. J. Tep-
total pressure of the overburden. litz and W. E. Hassebroek, "An investigation of oil-
well cementing": Drilling and Prod. Practice, p. 102.
6. In geologically simple and tectonically re- 1949, Squeeze cementing: World Oil, Septem-
laxed areas, not only should the fractures in a ber, p. 87.