Vous êtes sur la page 1sur 10

A MICROCRACK

MODELAND ITS SIGNIFICANCETO HYDRAULICFRACTURING


AND FRACTURE
TOUGHNESS
TESTING

Richard A. Schmidt, Sandia National Laboratories, Albuquerque,NM 87185

ABSTRACT Most investigations to date that deal with


fracture mechanics of rock have emphasized direct
A better understanding of the fracture process measurements of fracture parameters (e.g., Schmidt,
in rock is needed to explain certain observed 1976; Barker, 1977) and direct applications of
bhavior from fracture testing and to provide fracture mechanics principles (e.g., Simonson et al.,
assistance in selecting appropriate methods for 1978; Hanson et al., 1978; Warpinski et al., 1979).
applying fracture mechanics to actual rock mechanics Some emphasis has focused on descriptions of the
problems. First, the widely accepted formulation fracture process (e.g., Kobayashi and Fourney, 1978),
for the crack tip plastic zone in metals is reviewed. but much more work is needed if one is to fully
Observations on how rock fracture differs from understand the results of these fracture measure-

metal fracture motivates a change from the Von ments and to have confidence in the appropriateness
Mises yield condition to a maximum normal stress of these applications.
criterion. This results in a crack tip microcrack
zone description that is consistent with observed Fracture mechanics is based on the stess inten-

fracture toughness behavior for rock. sity factor, K, which essentially describes the entire
stress field at a crack tip in a linear elastic mat-
The model is extended to include the application erial. When a sufficient level of stress intensity

of confining stresses. This includes anisotropic is applied to a material the crack will extend. This
in situ stress states as well as hydrostatic com- critical value of K is referred to as fracture tough-
pression. Verification of this model rests on ness, Kic, whenlinear elastic conditions prevail.
its ability to explain observed fracture behavior; Since stresses are known to be very large at the
no direct verification is available as yet. Finally, crack tip, a process zone develops within which the

a suggestion is made for improving hydraulic fracture stresses are inelastic (plastic in metals). Linear
containment models based on average crack-tip stresses elastic conditions prevail, then, as long as this
rather than the stress intensity factor. zone of inelastic behavior remains small relative

to the other dimensions of the problem.


INTRODUCTION

The apparent fracture toughness of rock has been


In the past five years or so there has been a shown to depend on crack length (Ingraffea and Schmidt,
growing interest in applications of linear elastic 1978; Schmidt and Lutz, 1979) which is similar to the
fracture mechanics to problems involving rock. Much behavior of many metallic materials (Nelson et al.,
of this interest in rock fracture stems from energy 1972; Jones and Brown, 1970). This dependence in
research. Engineers who are concerned with optimum metals is attributed to the size of the crack tip
resource recovery find they must deal with rock as plastic zone relative to the dimensions of the
a structural material. Resource recovery techniques specimen. For metals the effect of this plastic
such as stimulation of oil and gas wells by hydraulic zone is also reflected in a thickness effect, but
fracturing, fragmentation of oil shale beds for in the fracture toughness of rock (and concrete) does
situ retorting, and thermal-stress induced fracturing not appear to be affected by thickness at all. The
of geothermal sources, for example, would benefit implica.tion here is that a crack tip process zone
from a better understanding of the fracture process occurs in rock, but that it is somewhat different
and requirements for crack propagation. from the plastic zone in metals. A description of

581
this crack tip process zone for rock may provide
(o1 o2)2+ (o2 - o3)2+ (o3 - Ol)2
an explanation of the fracture toughness effects (2)

and is needed to help provide guidelines for fracture =2Oy2s


testing.
whereOys is the uniaxial yield stressobtainedfrom
a standard tensile test.
In addition, inconsistencies have arisen in
applications of fracture mechanics principles to rock.
Through a transformation of stresses, Equation 1
For example, various analyses (e.g., Hanson et al.,
can be rewritten in terms of principal stresses as
1978) have indicated that a fracture approaching a
material interface should be arrested if the two
K o o
materials have different elastic moduli. However, O1 = cos (1 + sin3)
direct observations have demonstrated results to the (3)
contrary (Warpinski et al., 1980). A description 02= .K cos0 (1 - sin7)
0
of the crack tip process zone may help explain this
discrepancy and also suggest improved fracture
The "out of plane" stress, o3, must nowbe defined
criteria. In order to develop a model for this
order to use the yield condition of Equation 2. Two
crack tip process zone it may be helpful to review
extreme cases are usually described:
the steps that led to the description of the plastic
zone that occurs at a crack tip in metals.
Plane stress: o3 = 0
(4)
CRACK TIP PLASTIC ZONE IN METALS
Plane strain: o3 = v(o1 + o2)
The linear elastic stress field that occurs at
where v is Poisson's ratio. Substituting Equations
a crack tip can be described by the following set of
3 and 4 into 2 and solving for r, the result is
equations* (Irwin, 1957; Paris and Sih, 1965):

Plane stress:
K o o 3o
cos (1 - sin sin--) 2

o - K 0 0 30 (1) (1+ 73 sin20+ cos0)


Y cos (1 + sin sin-)
(5)
K 0 0 30 Plane strain:
sin cos cos-

where r and 0 are crack


in Figure 1. These equations,
tip coordinates
however,
as given
indicate
sin
20
+(1-2v)2(1
+cos
0
that stresses become infinite at the crack tip. (Broek, 1974).
In reality, though, the material yields, stresses
remain finite, and a plastic zone forms. Equation 5 is plotted (in non-dimensional form)
mFigure 2 assuming a Poisson's ratio of 1/3 for
A rigorous solution for the size and shape of the plane strain case. Recall that these boundaries
this plastic zone is not available, but an approximate merely define the region where the elastic stress
description is obtained by determining the boundary solution predicts that yielding will occur. If
from the elastic solution within which the stresses
the stresses inside this zone are limited to the
are greater than the yield stress. This is typically yield stress, there must be an increase in stress
done (McClintock and Irwin, 1965) by using a Von outside the zone to carry the load. This results
Mises yield condition which can best be described
in an increase in the actual plastic zone size. It
in termsof the principal stresses (Ol, o2, and o3) is estimated that the actual plastic zone has linear
as follows:
dimensions that are approximately double the lengths
predicted from Equation 5 (Irwin, 1960).
*Only the opening mode, Mode I, will be considered

582
The variation of apparent fracture toughness in a manner similar to metals. However, limited
of metals with crack lenth (Figure 3) can now be data indicate that thickness has little or no

understood. The toughness values for small crack effect on toughness (Schmidt and Lutz, 1979)(Fig-

lengths differ from Kic becausethe relative size ure 6). This observation appears to be consistent
of the plastic zone is not small enough for linear with data for concrete (Mindess and Nadeau, 1976).
elastic conditions to prevail. Years of testing These findings indicate that a crack tip process
many metallic alloys have resulted in an empirical zone exists for rock, but its description must

determination, that measured toughness will be differ from the plastic zone model.

sufficiently close to KIc whenall in-plane specimen


dimensions are at least 25 times larger than the The stress state at a crack tip is not con-
plastic zone. The ASt criterion (ASTM, 1972) for ducive to plastic flow in most rock, unless con-

a valid test for KIc is ditions of high hydrostatic compression or high


temperatures are present. The inelastic behavior

KIc
/2 (6) at the crack tip is usually described as micro-
crack
length
2.5X--
s/ ' cracking (e.g., Hardy, 1973; Hoagland et al., 1973).
The variation of apparent fracture toughness A crack tip microcrack zone in rock can be con-
with specimen thickness (Figure 4) is also explained sidered analogous to the crack tip plastic zone
from Equation 5 and Figure 2. The surface of a in metals.

specimen is in plane stress, while the interior of


a thick specimen is under conditions of plane strain, Microcracking occurs as a result of tensile
which produces a much smaller plastic zone. Varying stress while plasticity results from shear stresses.
the specimen thickness changes the relative amounts It seems logical, then, to develop a model of the
of plasticity thus changing the toughness. Testing microcrack zone by simply following the plastic
conditions that can be adequately described as pre- zone formulation, while replacing the usual Von Mises
dominantly plane strain occur only for sufficiently yield condition with the criterion of maximum
thick specimens. Empirical results of many tests normal stress that is more appropriate for the
have determined that the thickness requirement for tensile failure in rock.

metallic materials (ASTM, 1972) is


The maximum normal stress criterion is simply
(7) stated as

thickness
>2.5
(KI---
c)2 - Oys
(8)
For this reasonKic, which is defined as the limiting
value of toughness, is often referred to as the plane where ou is the ultimate tensile stress. The most
strain fracture toughness. appropriate choice for a value of ou for a given
rock is unclear, since tensile tests are usually per-
It is interesting to note that there has been formed on samples that are considerably larger than
little direct verification of the size or shape of the expected microcrack zone size. But, for lack of
plastic zones in actual materials. Nevertheless, the a better choice, ou will, for now, be considered
foregoing description is widely accepted for the value measured in an unconfined, direct-pull,

engineering purposes. tension test. (Note that tensile stress is taken

as positive to be consistent with the plastic zone


CRACK TIP PROCESS ZONE FOR ROCKS formulation.)

Basic Formulation Combining Equations 8 and 3 and solving for r


gives

The apparent fracture toughness of rock is


known to depend on crack length (Ingraffea and (9)
Schmidt, 1978; Schmidt and Lutz, 1979)(Figure 5) r(O)= u cos
(1+sin )

583
This describes the contour within which the theoretical Microcracks would be expected to occur perpendicular

(linear elastic) maximum normal stress is greater to this orientation and this array of microcracks

than somevalue, ou. As before, the actual stress is displayed by short line segments in Figure 8.

can not becomeinfinite, and if limited by ou a


stress redistribution must take place which will Effect of Hydrostatic Compression
increase the size of this zone.

Another difference between the plastic zone


Equation 9 can be non-dimensionalized and is model and the microcrack zone model can be seen

plotted in Figure 7. The zone size is very similar if hydrostatic compression is applied. (Many
to that of the plane stress plastic zone plotted in applications involving rock fracture take place
Figure 2. (In fact, if a Tresca yield condition where significant confining stresses are present.)
had been used instead of Von Mises, (Broek, 1974) Hydrostatic compression of magnitude can be
the two zones would be identical in shape and superposed in Equations 3 and 4 as follows:
size.) The similar size probably accounts for the
K
fact that the present fracture toughness data for coso (1 + sin)
o
-
rock, although limited, appear to follow a limiting (13)
criterion similar to that for metals (Schmidt, 1976). K
That is coso (1 - sin)
o
-

(10)
crack
length
>2.5(KI--
c)2 -- OU
plane stress: o3 = -

plane strain: o3 = v(a1 + o2)


The most obvious difference between the micro-

crack zone and the plastic zone models is that the


If these equations are substituted into the Von
microcrack zone, Equation 9, remains unchanged
Mises yield criterion (Equation 2), the same plastic
whether the out-of-plane stress, o3, is for plane zone contours are achieved as before (Equation 5).
stress, plane strain, or any state in between. In
That is, hydrostatic compression has no effect
other words, the zone shape and size as displayed
on the plastic zone size. However, upon substitution
in Figure 7 holds whether the crack tip location is
of Equation 13 into the maximum normal stress
on a free surface (plane stress) or deep within the
criterion (Equation 8) the resulting microcrack
material (plane strain). This is apparently the
zone description becomes
reason that rock and concrete, unlike metals, exhibit
no effect of thickness on fracture toughness. (14)
r(8) Ou
+ cos (1+sin )
Knowledge of the orientation of the maximum
normal stress throughout the microcrack zone may The zone size and shape are the same as before
give some indication of the orientation of the micro- except that ou is replaced by 0u +.
cracks within the zone. Referring to the original
transformation of stress, the orientation, e, of the This decrease in microcrack zone size with

maximum normal stress is hydrostatic compression accounts for the increase


in fracture toughness with confining pressure

1 tan-1 xy (11) observed by Schmidt and Huddle (1976) for Indiana


= 1 ' --
limestone (Figure 9). The reduced process zone
(ox - Oy)
must somehow elevate the fracture toughness. This
is similar to the increase in apparent fracture
Substituting Equation 1 into Equation 11 yields
toughness with crack length observed for this
material (Figure 5), since an increase in crack
(12)
length is comparable to a decrease in the size
of the microcrack zone.

584
Many applications at even moderate depths and Equation 15 is seen to reduce to Equation 14.
beneath the earth's surface involve confining The predicted zone configurations for various
stress that are comparable to and often greater values of $ are plotted in Figure 11.

than ou. This causesa large decreasein the


predicted microcrack zone size from the unconfined APPLICATIONS

size. For example, the calculated process zone

size for Nevada


ash-fall tuff (Kic = 500kPa This microcrack model has been shown useful

andou = 700 kPa) is 13 cmwhenunconfined. But, in describing behavior and general trends in fracture
the zone size drops to 0.10 cm for a confining toughness testing of rock and concrete. However,
stress of only 7 MPa (Warpinski et al., 1980). no direct confirmation of this model has been

(This confining stress is typical for hydraulic achieved as yet. But then, little direct con-
fracture experiments being conducted in this material firmation of the crack tip plastic zone model for
at the Nevada Test Site.) metals is available either, and that model has
been applied successfully for twenty years.

Effect of Anisotropic In Situ Stress State


One problem area that may benefit from this
In general, the in situ state of stress in microcrack model involves hydraulic fracture
most applications is not hydrostatic. The addition containment. The approach to the containment
of a completely general in situ stress state is, problem has focused on the behavior of a crack

however, rather laborious. A simplication is as it approaches a material interface, from sand-


possible considering the crack to be a hydraulic stone to shale for example. Various analyses
fracture which is, therefore, oriented perpendicular (e.g., Hanson et al., 1978) have indicated that
to the least compressive principal stress. A the interface should arrest the crack (providing

notation is adopted in Figure 10 that defines o containment) whenever the two materials have

to be the principal stress (compressive) that is different elastic moduli. This concept is based

perpendicular to the crack, and oil to be the on the observation that the stress intensity factor,

principal stress parallel to the crack. Since K, drops sharply to zero as the crack approaches
these stresses are oriented with respect to the a high modulus material from a low modulus material.
crack, they must first be added to the stress (When the materials are reversed, the stress
in Equation 1. Transforming these stresses to intensity still goes to zero, but not until the
principal stresses, substituting into Equation 8, interface is just penetrated.) However, direct
and solving for r yields observations of hydraulic fractures have demon-
strated results to the contrary (Warpinski et al.,

r(e,$)- i KZc 2 f(e,S) 19 80 ).


2 ou +
05) One principal difficulty with these analyses
is that the fracture criterion relies solely on
f(,B)
=[ 1 - 82 1+sin
sin
--B the value of the stress intensity factor. K is
defined as the strength of the square-root
+ (1 + sin sin-- 6) - (1 - ) cos
2
singularity in stress at the crack tip, but the

oll - o1 nature of this singularity changes as an interface


$ -
2(ou +) is approached. When the crack reaches the inter-
face, the singularity is no longer square-root
- 11
+
2
l
and K goes to zero; but the stresses still remain
singular (infinite). Other singularities now
The term, $, indicates the degree of in dominate the fracture growth.
situ stress anisotropy and is seen to vary from
0 to 1. However, values above 0.5 are uncommon. It has been suggested (Erdogan, 1979) that a
The case of $ = 0 is hydrostatic compression, fracture criterion based on some average crack-

585
REF ERENC E S
tip stress may provide a realistic alternative.
This criterion would say that fracture growth
would occur when the crack tip stress, calculated ASTM, 1972, "Tentative Method of Test for Plane
assuming linear elastic behavior and averaged over Strain Fracture Toughness of Metallic Materials

a set distance, reaches a critical value. Such an (ASTMDesignation: E399-72T)," 1972 Annual Book
average stress is directly proportional to stress of Standards, Part 31, Am. Soc. fom Testing and
intensity. The advantage is that this new criterion Mat'is, Philadelphia, PA.
would involve contributions from all stress

singularities when a material interface is approached. Barker, L. M., 1977, "A Simplified Method for
In other words, this criterion continues to be Measuring Plane Strain Fracture Toughness,"
functional when the stress intensity approach Ensineerin Fracture Mechanics, , p. 361.
collapses. The microcrack model is utilized here
to ake a realistic choice for the length over Broek, D., 1974, Elementary Engineerin S Fracture
which the stresses should be averaged. Mechanics, Noordhoff International Publishing,
Leyden, Netherlands, p. 97.
Quantitative results using this approach are
not available at this time. However, preliminary Clark, J. A., 1980, private communication.
calculations (Clark, 1980) do show that the average
stress, unlike K, does not go to zero as the inter- Erdogan, F., 1979, private communication.
face is approached. This, at least, allows that
conditions can exist for which a crack will cross Hanson, M. E., Anderson, G. D., Shaffer, R. J.,
a material interface. Emerson, D. 0., Heard, H. C., and Haimson, B. C.,
1978, "Theoretical and Experimental Research
SUMMARY on Hydraulic Fracturing," Proceedings, Fourth
Annual DOESymposiumon EnhancedOil and Gas
A microcrack model has been formulated to Recovery and Improved Drillin Methods, Tulsa,
describe the crack tip process zone in rock. The OK, p. 29.
model development parallels the plastic zone
formulation with the substitution of a maximum Ingraffea, A. R., and Schmidt, R. A., 1978,
normal stress criterion. The model is consistent "Experimental Verification of a Fracture Mechanics
with observed behavior from fracture toughness Model for Tensile Strength Prediction of Indiana
testing of rock. This fracture behavior includes: Limestone," Proceedings of 19th U.S. Symposium
1) apparent fracture toughness of rock increases on Rock Mechanics, Stateline, NV, p. 247.
with crack length as it does for many metals,
2) minimum crack length requirements for valid Irwin, G. R., 1957, "Analysis of Stresses and Strains

Kic tests coincide well with the ASTMcriterion Near the End of a Crack Traversing a Plate,"
for metals, 3) fracture toughness is not affected Journal of Applied Mechanics, 24, p. 361.
by specimen thickness (unlike metals), and 4)
fracture toughness increases with confining pressure Irwin, G. R., 1960, "Plastic Zone Near a Crack
(hydrostatic compression). and Fracture Toughness," Proceedinss 7th Sagamore
Conference, p. IV-63.
Previous analyses for hydraulic fracture con-
tainment are inconsistent with observed behavior. Jones, M. H., and Brown, Jr., W. F., 1970, "The
Suggestions are made on how one could modify Influence of Crack Length and Thickness in Plane-
these analyses using the microcrack model and Strain Fracture Toughness Tests," Review of
preliminary calculations show this to be a more Developments in Plane Strain Fracture Toughness
realistic approach. Testins, ASTMSTP 463, Am. Soc. for Testing and
Mat'is, p. 63.

586
Kobayashi, T., and Fourney, W. L., 1978, "Experi- Warpinski, N. R., Schmidt, R. A., Northrop, D. A.,
mental Characterization of the Development of 1980, "In Situ Stresses: The Predominant Influence
the Micro-Crack Process Zone at a Crack Tip in on Hydraulic Fracture Containment," Proceedings
Rock Under Load," Proceedings of 19th U.S. of 1980 SPE/DOESymposium
on Unconventional Gas
Symposium
on Rock Mechanics, Stateline, NV, p. 243. Recovery, SPE 8932, Pittsburgh, PA.

McClintock, F. A., and Irwin, G. R., 1965, "Plasticity


Aspects of Fracture Mechanics," Fracture Toughness
Testin and Its Applications, ASTMSTP 381, Am.
Soc. for Testing and Mat'is, p. 84.

Mindess, S., and Nadeau, J. S., 1976, Effect of

NotchWidth on Kic for Mortar and Concrete,"


Cementand Concrete Research, , p. 529.

Nelson, F. G., Schilling, P. E., and Kaufman, S. G.,


1972, "The Effect of Specimen Size on the Results
of Plane-Strain Fracture-Toughness Tests,"
Ensineerin8 Fracture Mechanics, , p. 33.

CRACK__. x
Paris, P. C., and Sih, G. C., 1965, "Stress Analysis
of Cracks," Fracture Toughness Testin and Its
Applications, ASTMSTP 381, Am. Soc. for Testing
Figure 1. Crack Tip Coordinates.
and Mat'is, p. 30.

Simonson, E. R., Abou-Sayed, A. S., and Clifton, R. J.,


1978, "Containment of Massive Hydraulic Fractures,"
PLANE
Society of Petroleum Engineers Journal, 18, p. 27. STRESS

Schmidt, R. A., 1976, "Fracture Toughness Testing


PLANE
of Limestone," Experimental Mechanics, 16, p. 161. STRAIN

Schmidt, R. A., and Huddle, C. W., 1977, "Effect


of Confining Pressure on Fracture Toughness
r
of Indiana Limestone," International Journal
of Rock Mechanics and Minin8 Science, 14, p. 289. CRACK
.1 .2

Schmidt, R. A., and Lutz, T. J., 1979, "Kic and


JIc of WesterlyGranite - Effects of Thickness
and In-Plane Dimensions," Fracture Mechanics
Applied to Brittle Materials, ASTMSTP 678, Am.
Soc. for Testing and Mat'is, p. 166.

Warpinski, N. R., Schmidt, R. A., Cooper, P. W.,


Walling, H. C., and Northrop, D. A., 1979,
"High-Energy Gas Frac: Multiple Fracturing in Figure 2. Plastic Zone Contours Using Von Mises Yield
Condition.
a Wellbore," Proceedings of 20th U.S. Symposium
on Rock Mechanics, Austin, TX, p. 143.

587
PLANE
SMALL-SCALE STRESS
YIELDING
KIc

PLANE
K
STRA IN

2.5
(Klcl
\O'y
S/
CRACK LENGTH TH1CKNESS

Figure 3. Effect of Crack Length on Apparent Fracture Figure 4. Effect of Thickness on Apparent Fracture
Toughness, Generalized for Metals. Toughness, Generalized for Metals.

lOOO

K
Ic

8oo

6OO
o,.ooit tl
Schmidt,
R.A.,
Exp__.
_M_ech_.,
1_6,
161 (1976)
I-L.I

o/
o
Three
point
bend
oCenter
specimens
notched
panels
from
t t[ t t
400

z
I-L.I / 111
/

200 I 1
0 10 20 30 40 50 6O

0.5 1.0 1.5 2.0 inches


CRACK LENGTH

Figure 5. Effect of Crack Length on Apparent Fracture Toughness of Indiana


Limestone (From Ingraffea and Schmidt, 1978).

588
3.0

leastsquaresfit

20 40 60 80 100 120

THICKNESS ( mm )

Figure 6. Effect of Thickness


on FractureToughness
of WesterlyGranite(FromSchmidtand
Lutz, 1979).

_ r r

Figure 7. Microcrack Zone Contour Using Maximum


Figure 8. ExpectedOrientation of Microcracks
Within Process Zone.
Normal Stress Criterion.

589
,o

nconfined three-point-bend

nconfinedsingle-edge-notch
__ I I
2 4 6 8 10
ksi
I
20 40 60
MPa

CONFINING PRESSURE

Figure 9. Effect of Confining Pressure (Hydrostatic Compression)


on Fracture Toughness of Indiana Limestone (From
Schmidt and Huddle, 1976).

1,o[ ,5
o

,5 '

,1,3

-1.0 i i
-l,0 -,5 0 ,5 1.0

Figure 10. Orientation of In Situ Stresses Relative


to Crack.
Figure 11. Microcrack Zone Contours for Various
Degrees of In Situ Stress Anisotropy,

59O

Vous aimerez peut-être aussi