Professional Documents
Culture Documents
SPE-686 - (Mechanics of Hydraulic Fracturing by Hubbert, 1957) PDF
SPE-686 - (Mechanics of Hydraulic Fracturing by Hubbert, 1957) PDF
E
Mechanics of Hydraulic Fracturing
M. KING HUBBERT
MEMBER AIME SHELL DEVELOPMENT CO.
DAVID G. WILLIS HOUSTON, TEX.
JUNIOR MEMBER AIME
T. P. 4597
LEAST PRINCIPAL
STRESS
FI G. l-STRESS EL EMEt"T AI'OD PRE FE RRED PLAI'OE OF FIG. 4 - S EC TION SHOWIN G ApPROXIMAT E S TRESS CONDITIONS IN
FRACTURE. SAND Box.
ment, however, a, will be horizontal, increasing as the approaches 90 and the plane becomes normal to the
partition is moved, and a, will be vertical and equal to least principal stress a" the normal stress becomes equal
the pressure of the overlying material. A third type of to aa and the shear stress again disappears. This figure
failure, known as transcurrent faulting, is not demon- completely describes all possible combinations of nor-
strated in the sand-box experiment. This occurs when mal and shear stress acting on planes perpendicular to
the greatest and least principal stresses are both hori- the plane of a, and aa.
zontal and failure occurs by horizontal motion along a It is next necessary to determine the combination of
vertical plane. In all three kinds of faults, failure occurs shear and normal stresses which will induce failure.
at some critical relationship between a, and a,. This information may be obtained from a standard soil-
To determine this critical relationship it is first neces- mechanics test which is illustrated in Fig. 7. A horizon-
sary to obtain an expression for the values of the normal tally divided box is filled with sand which is then placed
stress a and shear stress T acting across a plane perpen- under a vertical load. The shearing force which is nec-
dicular to the a"a,-plane and making an arbitrary angle essary to displace the upper box is then measured for
a with the direction of least principal stress aa. As various values of vertical stress. In this way it is found
shown in Fig. 5, this may be done by balancing the that the shearing stress for failure is directly propor-
equilibrium forces which act upon a small triangular tional to the normal stress, or that
prism of the sand. The resulting expressions for a and r
Tare - - = tan cf> , (2)
a
at + a,
a=
2 + 2
U'1, - (13
COS 2a, 1 where cf> is known as the angle of internal friction and
I
(1) is a characteristic of the material. For loose sand cf> is
(11 - (J'3 approximately 30. These critical stress values may be
r= sin 2a. plotted on a Mohr diagram, as shown in Fig. 8. The two
2
diagonal lines comprise the Mohr envelopes of the mate-
A very convenient method of graphically represent- rial and the area between them represents stable combi-
ing these expressions, known as the Mohr stress repre- nations of shear stress and normal stress, whereas the
sentation, consists in plotting values of normal and
area exterior to the envelopes represents unstable condi-
shear stress from Eq. 1 with respect to a,T-coordinate
tions. Fig. 8 thus indicates the stability region within
axes for all possible values of the angle a, as shown in which the permissible values of a and r are clearly
Fig. 6. The locus of all a,r-values is a circle and it can
defined. The stress circles may then be plotted in con-
junction with the Mohr envelopes to determine the con-
ditions of faulting. This is illustrated in Fig. 9 for both
normal and thrust faulting. In both cases one of the
principal stresses will be equal to the overburden pres-
sure, or a z In the case of normal faulting the horizontal
principal stress is progressively reduced thereby increas-
M~---------------------------------"
FIG. 5-STRESSES a AND r ON PLANE OF ARBITRARY
ANGLE a.
Oi - 0"3
To -2-Stn2a
UNSTABLE
STRESSE.S
REGION
OF
/ I I
/ / I
/ / I
--/ / I
./
BLAINE ANHYDRITE
6000 OIL CREEK SANDSTONE
(DRY, 24C. 0-2000 ATM PRESSURE)
(DRY, 24C., 0-2000 ATM PRESSURE) 3000
~'22'-43'
~--- ~'45'
4000
2000
FIG. ll~MOHR ENVELOPES FOR OIL CREEK SANDSTONE FIG. 12-MOHR ENVELOPES FOR BLAINE ANHYDRITE
(MEASUREMENTS BY JOHN HANDlN). (MEASUREMENTS BY JOHN HANDIN).
"9";
I:,
,
,, ,
, :1
, i1
, :\
, 1
/'
/ : t-- -
+ '"(~~
"A
H-I-+-H-I-+-I.-i [lpeO
- - - - - -- -- -
2"A~ vI---
1/
"9
"8
('a;;~ 10)
FIG. I3-SUPERPOSITION OF STRESS STATES ABOUT A WELLBORE DUE TO Two HORIZONTAL PRINCIPAL STRESSES OF EQUAL MAGNITUDE.
l~
THE EFFECf OF PRESSURE APPLIED IN
THE BOREHOLE 0", = + IIp ;: ,
("'~
32
"A stresses which were calculated previously. This is illus-
LI P"O l trated in Fig_ 16 in which a pressure equal to 1.60"A is
applied to the wellbore for the case in which (fBI0"1
~t--
16"A v~
---- -- -- = 1.4, and is just sufficient to reduce the circumferen-
tial stress to zero across one vertical plane at the walls
of the hole. In all cases when the O"BIO"A-ratio is greater
than 1, the vertical plane across which 0", first becomes
tal zero as the wellbore pressure is increased is that perpen-
dicular to 0".1, the least horizontal stress.
('j
A the strongest rocks. However, as observation of any
l outcrop will demonstrate, flawless specimens of linear
dimensions greater than a few feet rarely occur. In ad-
/ -- - - -
~c_
- - - - -- dition to the bedding laminations across which the
"A '--
tensile strength ordinarily is a minimum, the rocks
usually are interesected by one or more systems of
joints comprising partings with only slight normal dis-
(b)
--~=\
/'
/ '
I , ,
,
/
'"
;
r ,, ,,
~
t , r- r-r--
8,,/ J
C:) -\ -- - -- - -- - --
i.
A
"B
(cr"3.0)
A
(e)
FIG. I4--STRESS STATES ABOUT A BOREHOLE FOR REGIONAL-STRESS FIG. I5-STRESSES CAUSED BY A PRESSURE l:J.p
RATIOS (fBI (fA OF 1.4, 2.0, AND 3.0. WITHIN THE WELLBORE.
/ I 1 \
/ r---I
"
+
1
~/ 1
-""" -
T
(0 S
32(/ A
-160-
I~ ------ -
____ /"l----.
- - -
0,;+08 Llp
(~'14)
A
FIG. 16--SUPERPOSITION OF THE STRESSES DUE TO A PRESSURE !'::.p OF 1.6 ITA UPON THE STRESSES AROUND
A WELLBORE WHEN ITB/ITA Is 1.1.
placements. Across these joint surfaces the tensile equal to the component of the undistorted stress field
strength is reduced essentially to zero. normal to the plane of the fracture. A pressure only
In any section of a well bore a few tens of feet in slightly greafer than this will extend the fracture in-
length, it is probable that many such joints have been definitely provided it can be transmitted to the leading
intersected. It appears likely, therefore, that the tensile edge. This can be seen from an analysis of an ideally
strengths of most rocks that are to be subjected to hy- elastic solid, as shown in Fig. 17. The normal stresses
draulic fracturing by pressure applied in wellbores is across the plane of a fracture near its leading edge are
effectively zero, and that the pressure required to pro- shown for the case in which the applied pressure !'::.p
duce a parting in the rocks is only that required to re- is slightly greater than the original undistorted stress
duce the compressive stresses across some plane in the field ITA. This solution is derived directly from the solu-
walls of the hole to zero. tion for ilie stresses in a semi-infinite solid produced
As the pressure is increased, the plane along which by a distributed load, which is presented by Timo-
a fracture will commence will be that across which shenko."
the compressive stress is first reduced to zero. In the The tensile stress near the edge of the fracture ap-
case of a smooth cylindrical wellbore, this plane must proaches an infinite magnitude for a perfectly elastic
be vertical and perpendicular to the least principal re- material. For actual materials this stress will still be
gional stress. For the cases illustrated in Fig. 14, the so large that a pressure !'::.p only slightly greater than
least compressive stress across a vertical plane at the ITA will extend the fracture indefinitely. The minimum
walls of the hole varies from twice ITA to zero. There- down-the-hole injection pressure required to hold open
fore, the down-the-hole pressure required to start a and extend a fracture is therefore slightly in excess of
vertical fracture with a non-penetrating fluid may vary the original undistorted regional stress normal to the
from a value of twice the least horizontal regional stress plane of the fracture. The actual injection pressure will
to zero, depending upon the ITB/IT..-ratio.
It can be seen from Eq. 7 that pressure inside a
cylindrical hole in an infinite solid can produce no axial
tension, suggesting that it is impossible to initiate hori-
zontal fractures. However, under actual conditions in
wellbores, end effects should occur at well bottoms or
in packed-off intervals in which axial forces equal to
the pressure times the area of the cross section of the
hole would be exerted upon the ends of the interval.
Furthermore, irregularities exist in the walls of the bore- ~--'--'---rT'T
"A
hole which should permit internal pressures to produce ~=-rr~~~t+~~~+4~
tension.
In particular, as has been suggested by Bugbee,' the
initial fractures may often be joints which have separated
sufficiently to allow the entrance of the fluid, in which
case it is only necessary to apply sufficient pressure to
hold open and extend the fracture.
INJECTION PRESSURES
, I
Once a fracture has been started the fluid penetrates
the parting of the rocks and pressure is applied to the II
walls of the fracture. This reduces the stress concen- FIG. 17-STRESSES IN THE VICINITY OF A CRACK
tration that previously existed in the vicinity of the IN A STRESSED ELASTIC MATERIAL WHEN THE PRES
SURE ACTING ON THE WALLS OF THE CRACK Is
well bore and the pressure !'::.p required to hold the frac- SLIGHTLY GREATER THAN THE STRESS WITHIN THE
ture open in the case of a non-penetrating fluid is then MATERIAL.
behavior during a fracturing treatment. These are illus- H" = - to(t::,p) (9)
trated in Fig. 18. The pressures t::,p are increases meas- n_ on '
ured with respect to the original fluid pressure in the However, due to viscous coupling the force H n ,
rocks. In one case the breakdown pressure may be acting upon the fluid is entirely transmitted to the rock
substantially higher than the injection pressure. This so that the total outward force exerted upon the rock
would probably correspond to a horizontal fracture per unit of bulk volume will be
from a relatively smooth well bore or to a vertical frac- o(t::,p)
ture under conditions in which the two horizontal prin- Hn = HnI + H n, = - - - (10)
cipal stresses ITA and <TB were nearly equal. In the second on
case there is no distinct pressure breakdown during Similarly, the total outward force per unit area of
the treatment indicating that the pressure required to the fracture wall will be the integral of all the forces
start the fracture is less than or equal to the injection exerted upon the rock contained within a column of
pressure. This would correspond to a horizontal or ver- unit area of cross section normal to the fracture, or
tial fracture starting from a pre-existing opening or to
a vertical fracture in a situation where the ratio <TB/<TA
of the horizontal principal stresses was greater than 2.0.
F
A
'"
o(t::,p)
- --:an-dn
J = - f0
d (t::,p) = t::,p"
/!"Po
I -LEGEND-
~~
TE R T I ARY INTRUSI V ES
- I
CRE TAC EOUS AND EOC EN E
ti2m PALEOZO IC
exactly perpendicular to the trajectories of the least the polyethylene container laterally, thereby forcing it
principal stress. into an elliptical cross section, and producing a com-
pression in one horizontal direction and an extension
EXPERIMENTAL FRACTURING at right angles in the other. The least principal stress
DEMONSTRATION was therefore horizontal, and vertical fractures should
In order to verify the inferences obtained theoretically, be expected in a vertical plane, as shown in Fig. 21.
a series of simple laboratory experiments has been per- In other experiments the container was wrapped with
formed. The general procedure was to produce fractures rubber tubing stretched in tension, Fig. 22, thus pro-
on a small scale by injecting a "fracturing fluid" into a ducing radial compression and a vertical extension. In
weak elastic solid which had previously been stressed. this case, the least principal stress was vertical, and
Ordinary gelatin (12 per cent solution) was used for horizontal fractures could be expected, as shown in
the solid, as it was sufficiently weak to fracture easily, Fig. 22.
was readily molded with a simulated wellbore, and was, The plaster slurry was injected from an aspirator
almost perfectly elastic under short-time application of blJttle to which air pressure was applied by means of
stresses. A plaster-of-paris slurry was used as a fractur .. Ii squeeze bulb.
ing fluid since this could be made thin enough to flow
easily and could also be allowed to set and thus provid,~;
a permanent record of the fractures produced.
It is interesting to note that, in a model experimen t
conducted in this way, the stress distributions are en ..
tirely independent of scale. Provided the material is
elastic, similitude will exist no matter on what length
scale the experiment is conducted.
The experimental arrangement consisted of a 2-gal
polyethylene bottle, with its top cut off, used as a con-
tainer in which was placed a glass tubing assembly con-
sisting of an inner mold and concentric outer casings.
The container was sufficiently flexible to transmit ex-
ternally applied stresses to the gelatin. The procedure
was to place the glass tubing assembly in the liquid
gelatin, and after solidification to withdraw the inner
mold leaving a "wellbore" cased above and below an
FIG . 20-PHOTOGRAPH OF W ES T SPAN ISH PEAK FRO M THE
open-hole section. Stresses were then applied to the NORTHWEST, SHOWING DIKES CUTTING FLATLYING EOCENE
gela6n in two ways. The first, Fig. 21, was to squeeze STRATA (G. W. STOSE, U. S. G EOL OGICAL SURVEY).
FAVORED F RACTURE
DIREC TI ON
(
//
LE AS T PRINCIPA L
STRESS
LEAST PRINCIPAL
ST RESS
FAVORED
FRACTURE
DIRECTION -"2.._ _,-
[<'IG. 2S-VERTICAL FRACTURE IN STRATIFIED GELATI". FIG. 26-HoRIZONTAL FRACTURE IN STRATIFIED GELATIN.
We are not in agreement with the authors' conclu- The authors have stated that low breakdown pres-
sions and feel that they have made a number of sim- sures in the bulk of a large number of fracturing oper-
plified theoretical assumptions which they have failed ations imply the creation of vertical fractures. They
to support. Attempts to extrapolate their type of ex- have not supported this with experimental results. Our
perimentation to reservoir conditions and to imply far- laboratory and field experimentations have led us to
reaching conclusions is, to us, a serious mistake. the opposite conclusion. In our opinion, their basic as-
We are convinced that horizontal fractures can be sumption of 1 Ib of overburden pressure per foot of
created at any depth because we have seen the re- depth has not been supported experimentally.
sults of field experiments which indicated them. Spin-
We believe that fractures may be either horizontal
ner' surveys and radioactive surveys run for the pur-
pose of locating fractures are easier to believe than or vertical. If incipient fractures (horizontal or vertical)
laboratory experiments in non-porous and non-per- exist, any type of fluid will extend them, rather than
meable gelatin. Careful analysis of 18 surveys which change their direction. If no fractures exist, we believe
gave positive results indicated 14 horizontal and four it has been proven that horizontal fractures can be
vertical fractures. started at any depth by penetrating breakdown liquids.