Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
00
Printed in Great Britain. Pergamon Press pk.
Abstract-For quasi-brittle materials like concrete, rock and ceramic it is not possible to determine
fracture toughness using laboratory-sized specimens and linear elastic fracture mechanics. This
inapplicability of the conventional fracture mechanics is due to fracture process zone, R-curve type
of behavior and precritical stable crack growth. In this keynote paper, a two-parameter fracture
model developed at Northwestern University is described. The model is based on observations of
fracture processes using laser holography as well as acoustic emission measurements.
Since G,, is closely related to K,,, the size-dependency of G,, is essentially the same as that of K,,.
s. P. SHAH
1000, I
0.8 . Experimental results
Corpinteri (1981 I
- 0.8
800 --
l\
/ l
A. :
-0.6 :
_;
6 600 - J'
/’ E
P !il
,a
"
I
ti G
b-5.9‘f‘ n- -%@ - 0.4
400- -- ,
I 2ooO_10-/o.2 0.6
0.2 0.4 06 0.8 Notch-depth rcitio (a,/bf
(al $L (b)
Fig. 1. Theoretical prediction and experimental results of size effect on conventional K,, of TPB
specimens.
Fracture energy, G,
The fracture energy has similar physical meaning as the critical strain energy release rate.
However, unlike the determination of G,, which is directly related to the peak load, the fracture
energy is determined from the work to break a specimen into two halves. Ideal values of Gr should
be calculated using direct uniaxial tensile tests. Due to the difficulties incurred in performing the
direct uniaxial tensile tests (i.e. prevention of eccentricity and prediction of the failure location etc.),
a three-point bend notched beam test was proposed[ l] to calculate the fracture energy. By recording
the complete (from zero load to zero load) load-load line deflection curve in a stable 3-PB notched
beam test, the fracture energy can be calculated.
Based on the proposal[l], Gr values have been experimentally determined by several
laboratories[2-4]. The Gr values were reported to be dependent on specimen sizes. For the same
notch-depth ratio (which is proposed to be 0.5), larger specimen sizes always yields higher values
of GF. For the specimens of the same size with different notch-depth ratios, lower notch-depth
ratios give higher Gr values, The size-dependency of GF has been attributed to the settlement of
support and energy contributed from tne noncritical sections.
Transducers
/ \
\A/ .-scurc*
Fig. 2. Explanatory diagram to show the difference in arrival times at four transducers.
is completely different from ductile materials due to lack of large plastic deformation. Since
different fracture mechanisms are involved, the conventional methods proposed to measure the
CTOD, values for metal are not applicable to concrete. The CTOD of concrete is mainly due to
the formation of stable crack growth. Thus, it is expected that the CTOD, can be related with the
critical crack extension.
Loading grips
Extensometer
Disk
storoge
,I
I
I
Fig. 3. Acoustic emission monitoring set-up.
S. P. SHAH
800
(a)
Loading stage 2
Loading stage 3
Fig. 5. Microscopically observed crack lengths, acoustic emission source locations and effective
crack lengths (Ref. 5).
determined by an ultrasonic system. To find the rate of AE events, the oscilloscope is connected
to an IBM-AT by a GPIB (IEEE 488) interface bus[5].
The Acoustic emission set-up is shown in Figs 3 and 4. Four transducers (PZTSA, Dunnegan
Corp Type MtOOD] were attached to the specimen with methyl cyanacrylate adhesive. The AE
signals from the four transducers were amplified by a 70 dB bandpass filter (120 kHz-1.2 MHz].
The signals were digitized and stored on diskettes which were used for future analysis. The first
arrival of the P waves at each transducer was found and the difference in arrival times of the AE
signal at the 4 transducers was used to locate the source of the AE signals. Nicolet digital
oscilloscope with sampling rate of 2 MHz was used. A least squares iteration program{81 was used
for source location. To count the number of AE events, the oscilloscope trigger used to send a
message to the IBM-AT by a GPIB bus. Each time an AE event occurred, the oscilloscope triggered
and the GPIB message updated the count on the IBM-AT.
Some typical results from the AE source location studies are shown in Fig. 5. Single-edged-
notched specimens of mortar were tested in a closed-loop testing machine. The crack mouth
opening displacement (CMOD) was used as a feedback control. In Fig. 5, load vs CMOD curves
Fracture process zone and fracture parameters
-60
-70 I I -16
0 I 2 3
of a specimen is given. The location of AE events observed during the different stages of
load-CMOD response are also given along with microscopically observed surface crack length. The
following conclusions can be summarized from the study:
1. Microcracking localizes before peak load and the microcrack cluster progresses forward
with increasing CMOD.
2. Few events occur ahead of the microscopically observed crack tip but events continue to
occur far behind the crack tip indicating closing pressure due to ligament connection,
3. Rate of AE events reach a maximum immediately after peak load.
4. Effective crack length estimated from the modified LEFM approach (to be described later)
is consistent with optical and AE measurements.
Laser holography
The optical arrangement for obtaining split-beam transmission (Lieth-Upatniks) hologram is
shown in Fig. 6. A hologram is produced as a result of the interference of the object beam and
the reference beam. When two holograms of the object at two levels of loading are superimposed
(sandwich holograms), fringes are observed. The number of fringes give the displacement along
the sensitivity vector. With this arrangement we can observe crack profile as the crack propagates
with an increase in load with an accuracy of about l/2 micron.
Sample crack profiles for a (center cracked plate) mortar specimen are shown in Fig. 7. Note
that the radius of the loading hole was l/2 in. and the length of the notch was 7/8 in. It can be
observed that the crack tip opening displacements will sometimes increase over several load
intervals before the crack propagates. A plot of CTOD of the former crack tip just after
propagation is shown in Fig. 8. On the same plot, critical crack opening displacement calculated
from the notched beam test (this will be defined later) by Jenq and Shah is also shown. Although
the holographic measurements show a considerable scatter, the idea of critical crack tip opening
displacement appears correct from these data.
75
60:
50 -
40 -
-------------_-___ *-----
----__----
Jenq ond Shoh [S] + ;-*-----
0
Notched beam test
210
30 - 0
0 a
20 -
+ 0 O 06 0
IO -
d
0 I 2 3’
Fig. 8. Crack opening displacement (COD) of former crack tip just after propagation.
specimens. Several nonlinear fracture models have been proposed to account for the nonlinear
fracture process zone. A model proposed by Jenq and Shah is briefly described in this paper.
where
K, = net stress intensity factor for mode I
K,, = net stress intensity factor for mode II
a(w) = tensile traction acting on the crack surface
CTOD = crack tip opening displacement
T(S) = shear traction acting on the crack surface
CTSD = crack tip sliding displacement.
In the present derivation, the energy is assumed to dissipate along the crack. Equation (1)
includes most of the basic fracture parameter (i.e. Ki, K,, , CTOD, CTSD) that might be involved
in a general failure condition. In eq. (I), the shear traction (r) is a function of the sliding
displacement, friction coefficient (p) and possibly, crack opening displacement. The effect of the
shear traction to the structural system is similar to the slip type of plastic deformation in metal.
Therefore, the contribution of energy by shear traction will not be conservative and the global
mechanical behavior may not be predicted just based on energy criteria if the shear traction is not
negligible.
It should be noted that if no tractions exist along the crack surface, the eq. (2) will reduce
to the Griffith crack, i.e.
K2 K2
- -
G-J-_-!+” (3)
E’ E”
Fracture process zone and fracture parameters 11
i CMOD. A CMOD
Fig. 9. Explanation of the terms used in the two parameter fracture model of concrete (Ref. 13).
However, if one assumed that there will be no stress concentration (K, and K,, are equal to zero),
then, Dugdale-Barenblatt type of model will results, i.e.
CTOD CTSD
G=J= CT(W) dw t(S) ds.
+ (4)
s0 s 0
The original idea of the tractions is attributed to the atomic type of force in
Dugdale-Barenblatt model which is not obtained from the global uniaxial stress-strain or
stress-separation relationships.
For fiber reinforced concrete, eq. (2) can be used as an approximate solution of the energy
consumption and the tractions due to the fiber pullout and sliding can be determined from fiber
pullout test. Detailed application of fracture mechanics to steel fiber reinforced concrete is given
in [9].
To examine the validity of eq. (2) and the associated fracture parameters, we first wanted to
simplify the studied problems so that the influence of normal and shear tractions are minimized.
The simplest fracture problem of this type is encountered in mode I failure of plain concrete. Only
two fracture parameters are involved if the tractions along the crack surface are assumed to be
negligible. These two fracture parameters are the critical stress intensity factor (Ki,) and the critical
crack tip opening displacement (CTOD) (Fig. 9) which is used to quantify the stable crack growth
prior to peak load.
(Fig. 9). One can determine these parameters from pertinent linear elastic fracture mechanics
(LEFM) equations and from the measured values of initial compliance (Ci in Fig. 9) and unloading
compliance (C, in Fig. 9) at or near the maximum load as described next.
In general, LEFM yields the following equations:
K, = oJna F,(a) (5)
CMOD = 4a a/E. V,(a) (6)
CTOD = CMOD * Z(a, /?) (7)
where 0 = applied stress, E = Young’s modulus, D = depth of the specimen, CMOD = crack
mouth opening displacement, CTOD = crack tip opening displacement, a = a/D = notch-depth
ratio, a = a, + Aup, a, = initial notch length, Au, = pre-peak crack extension, /I = Q/U and El, V,
are geometry correction factors. Knowing the initial compliance (Ci in Fig. 9) and using eq. (5)
and a = a,, the value of Young’s modulus E can be determined. The substituting this value of E
in eq. (6), an effective crack length a is calculated such that the compliance obtained using eq. (6)
is equal to the measured elastic compliance at the peak load (C, in Fig. 9). Kk And CTOD, are
then calculated at peak load using eqs (5)-(7) and the effective crack length a. Thus, the two fracture
parameters, K;, and CTOD, (as well as Young’s modulus E) can be determined from a single
notched beam test.
If the fracture parameters K3, and CTOD, and Young’s modulus E are known, the peak stress
and hence peak load of any structure can be calculated using the eqs (5)-(7). Combining eqs (5)-(7)
leads to:
K =CTOD.E+ F,(a)
I
4Ja ’ v,(a).
(8)
z@, 6)’
The value of effective a at the critical load is determined from eq. (8) for a given set of values of
KC, and CTOD,. Then using eq. (5) the corresponding stress value at the critical load can be
determined.
Based on tests of three different sizes of single edge notched beam specimens, Jenq and
Shah[l 1,121 showed that the two fracture parameters Ki, and CTOD, thus determined are
size-independent (Fig. 10). The superscript in Ks, denotes inclusion of prepeak stable crack growth.
TPFM requires the use of a closed loop testing system with COD as the control feedback signal
so as to obtain stable load-COD response. But both the fracture parameters Ki, CTOD, (and E)
can be obtained from a single test. These results were validated recently by Alvarado, Shah and
I .4
0 I 2 3 4 5
AU dimensions in inches
I inch = 25.4mm
I p.s.i. = 0.0069 MPo
IO I I I I I
0 2 4 6 6 10
John[l4] and Miller et a1.[7] based on a mode I fracture study using a plate specimen with a central
notch and a point load applied at the center of the notch.
Knowing the fracture parameters Kf, and CTOD,, and Young’s modulus, E the peak load
for any structure can be predicted by a simple iterative procedure using LEFM equations similar
to eqs (S)--(S). Note that this model is valid for both notched (a,, # 0) and unnotched (a~ = 0)
specimens. This model will reduce to the single parameter LEFM approach when CTOD, -P 0 (i.e.
a, = q,). Combining Kfc, CTOD, and E, Jenq and Shah[2] obtain a material parameter, Q, which
is similar to the characteristic length, I,, proposed by Hillerborg et a1.[15]. Q Can be expressed as:
Q =[“‘~coDc] (8)
Jenq and Shah[OO]and John and Shah[l6] have used this model to predict: (1) tensile strength; (2)
split-cylinder strength; (3) size effect of R,, (Fig. 1); (4) size effect on modulus of rupture; (5) strain
rate effect; and (6) size effect in mixed-mode fracture and shear.
CONCLUDING REMARKS
It is interesting to note that the critical crack extension (Au,) calculated using the two fracture
parameters (KS, and CTOD,) gives a good indication of relative fracture toughness of materials.
For example, critical crack extension decreases with an increase in compressive strength (Fig. 11).
This means that for very high strength concrete, the precritical crack growth tends to be negligible
and as a result, the material behaves more classically brittle. The same observation can be made
with respect to the effect of increasing strain rate. It was observed that the higher the strain rate,
the lower the valve of Aa, (Fig. 12). Thus, the extent of critical crack extension (which can be
ascertained from the two fracture parameters) appear to be a useful index for toughness for
quasi-brittle material. For a classically brittle material, Au, = 0 and LEFM is applicable. For
concrete, Aa, can be significant and as a result, one fracture parameter is not adequate. The value
of Aa, will depend on material composition, strain rate, size and geometry of the structure.
Acknowledgement-This research was supported by a grant from the Air Force Office of Scientific Research (AFOSR-85
0261) under a program managed by Dr Spencer T. Wu.
REFERENCES
[I] Proposed RILEM recommendation, Determination of the fracture energy of mortar and concrete by means of three
point bend tests on notched beams. Division of Building Materials, Lund Institute of Technology, Sweden (1982).
12) P. Nallathambi, B. L. Karihaloo and B. S. Heaton, Various size effect in fracture of concrete. Cement Concr. Res.
15, ooo (1985).
[3] A. Hillerborg, Concrete fracture energy tests performed by nine laboratories according to a draft RILEM
recommendation. Report to RILEM TC SO.FMC, Division of Building Materials, Lund Institute of Technology,
Sweden (1983).
[4] A. Hillerborg, Additional concrete fracture energy tests performed by six laboratories according to a draft RILEM
recommendations. Report to RILEM TC 50-FMC, Division of a Building Materials, Lund Institute of Technology,
Sweden (1984).
[5] A. K. Maji and S. P. Shah, Process zone and acoustic emission measurements in concrete. Exp. Me&t., 27-33 (March
1988).
[6] A. K. Maji and S. P. Shah, Appli~tion of acoustic emission and laser holography to study ~crofracture in concrete.
ACI SP-112 (1989).
[7] R. A. Miller, S. P. Shah and H. I. Bjelkhagen, Crack profiles in mortar measurement by laser holographic
interferometry. To be published in Exper. Mech. (April 1989).
[S] J. F. Labuz, S. P. Shah and C. H. Dowding, The fracture process zone in granite: evidence and effect. Int. J. Rock
Mech. Mineral Sci. (August 1987).
191 Y. S. Jenq, Fracture mechanics of cementitious composites. Ph.D. Thesis, Northwestern University, Chapter 5 (June
1987).
[IO] Y. S. Jenq and S. P. Shah, Nonlinear fracture parameters for cement based composites: theory and experiments. in
Application of Fracture Mechanics lo Cementitious Composites (Edited by S. P. Shah), pp. 319-360. Martinus Nijhoff
Company, Dordrecht (1986).
[I I] Y. S. Jenq and S. P. Shah, A fracture toughness criterion for concrete. Engng Fracture Mech. 21, 1055-1069 (1985).
[12] Y. S. Jenq and S. P. Shah, Two parameter fracture model for concrete. J. Engng Mech. 111, 1227-1241 (October 1985).
[13] Y. S. Jenq, Fracture mechanics studies of cementitious composites. Ph.D. Dissertation, Northwestern University,
Evanston, Illinois (1987).
[14] M. Alvarado, S. P. Shah and R. John, Mode I fracture in concrete using center cracked specimens. Accepted for
publication in >. Erzgng Mech. Die. (March 1988).
[I 51 A. Hillerborg, M. Modeer and P. E. Petersson, Analysis of crack formation and crack growth in concrete by means
of fracture mechanics and finite elements. J. Cemenr Concr. Res. 6, 773-782 (1976).
[ 161 R. John and S. P. Shah, Fracture of concrete subjected to impact loading. Cement Concr. Aggregates 8,24-32 (Summer
1986).