Documente Academic
Documente Profesional
Documente Cultură
Consequences of recent fracture mechanics studies of concrete for Having realized this fact, ReinhardC,32 recently ana-
analyzing diagonal shear failure of longitudinally reinforced beams or lyzed certain test data for diagonal shear failure on the
one-way slabs without shear reinforcement were studied. The crack-
basis of the classical (linear elastic) fracture mechanics
ing produced by shear was assumed to propagate with a dispersed
zone of microcracks at the fracture front. Dimensional analysis of the and found a relatively good agreement with these test
energy release rate then shows that the nominal shear stress at fail- data. However, when some other available data are
ure should not be a constant but should vary as (1 + did. Aor', in considered, the agreement with the linear elastic frac-
which d = beam depths, d. = maximum aggregate size, and A0 = ture mechanics is not very good and does not seem
constant. For relatively small beams, representing the great majority
much better than for the strength criterion. This is not
of those tested in the laboratories, the nominal stress at failure is
nearly constant; however, for much deeper beams it considerably de- surprising, since the linear elastic fracture mechanics
clines with increasing size. This trend is confirmed by previous exper- has been found to be inapplicable to concrete, as dem-
imental results. In addition to the size effect, a rational formula for onstrated, e.g., by Naus and Kesler.
the effect of steel ratio and shear span is derived. Comparisons with The last few years, however, witnessed an increased
existing test data involving nearly 300 tests indicate that, compared to
interest in basic studies of concrete fracture, and a new
the formulas in the current building codes, the coefficient of varia-
tion of deviations from the formula is reduced to less than one-half. form of fracture mechanics which appears applicable to
concrete has emerged. This new approach does not
Keywords: beams (supports); building codes; cracking (fracturing); dimen-
treat fracture as a point phenomenon, but recognizes
sional analysis; failure; reinforced concrete; shear properties; statistical analy-
sis; structural analysis. that in brittle heterogeneous materials such as concrete
the fracture propagates with a relatively large fracture
STATEMENT OF PROBLEM process zone in which progressive microcracking grad-
Predicting brittle failures of concrete structures due ually reduces the tensile stress to zero. 33-35 The aim of
to tensile cracking of concrete is much more difficult this paper is to explore the consequences of this new,
than predicting ductile failures. The diagonal shear nonlinear fracture mechanics for diagonal shear fail-
failure of longitudinally reinforced beams and one-way ure. The main purpose of fracture mechanics is the
slabs without shear reinforcement is a good example. prediction of the effect of structure size, and it will be
Although great progress has been achieved in various seen that a considerable improvement can be brought
theoretical studies l -8 and extensive experimentation,I,4,8-28 about in this regard.
the scatter of the deviations of test results from the
formulas in current building codes is enormous (see STRUCTURAL SIZE EFFECT IN FRACTURE
Fig. 7.13 in Reference 29) and is much larger than the The structural size effect may be illuminated by con-
scatter of tensile strength or fracture energy. sidering structures of different sizes but geometrically
The current formulas are based on the concept of ten- similar shapes, e.g., beams of the same steel ratio and
sile strength. However, this concept is theoretically jus- the same ratio of depth to shear span. The strength cri-
tified only in the case of ductile failures governed by terion may be stated as UN = Jt' where Jt' = direct ten-
the theory of plasticity. For failures in which the stress sile stength of concrete and UN = nominal stress at fail-
decreases after reaching the strength limit, as is the case ure. For reasons of dimensionality, UN = cNPlbd where
for tensile cracking, the strength concept is inconsistent P = given load, d = characteristic dimension of the
when applied in a continuum analysis. For example, fi- structure,' e.g, the depth of beam, b = thickness, and
nite element analysis of cracking based on the strength CN = constant. Thus, if one considers the plot of log UN
criterion can exhibit a strong spurious dependence on
the choice of mesh size. 3o,31 As is well known from frac- Received May 23, 1983, and reviewed under Instit~te publication. polici.es.
Copyright © 1984, American Concrete Institute. All nghts reserved, mcludm.g
ture mechanics, a theoretically consistent approach the making of copies unless permission is obtained from the copyright propn-
etOTS. Pertinent discussion will be published in the July-August 1985 ACI
must be based on an energy criterion of failure. JOURNAL if received by Apr. I, 1985.
ACI member Jin-Keun Kim is a graduate research assistant and PhD candidate
in civil engineering at Northwestern University. He obtained his BS and MS
degrees in the Department of Architecture from Seoul National University. He
has worked as a teaching assistant at Seoul National University, as an instruc- most existing tests
tor at Ulsan Institute of Technology, Korea, and as a structural designer. His
research interests include inelastic behavior and fracture of concrete and rein-
forced concrete, as well as stress-strain relations for soils.
nonlinear fracture mechanics
"""--
.80+-------~----~~-----+------_+--~
(b) 0.00 .01 .02 .03 .04
~J
p
(6)
We need a simple description of the function j(x), To express the arch action contribution to shear, we
and choose for this purpose may set T = as pbd, and substituting into Eq. (2) we
obtain
(3)
(7)
in which jo is a constant, defining the location of the
compression resultant C at the end of the shear span, x
= a. According to the classical bending theory of rein- Experience shows that the diagonal shear failure hap-
forced concrete beams with only tensile reinforcement pens by inclined cracks whose horizontal projection
and with a negligible tensile capacity of concrete, we roughly equals beam depth d. This suggests taking x =
would have d as the critical cross section for arch-action shear, and
Eq. (7) then reduces to
jo = 1 - 1;3 (c/d) , c/d = (nZpZ + 2np)1h - np (4)
(8)
in which p = steel ratio, n = E/ Ec = ratio of elastic
moduli of steel and concrete, and c = depth to neutral
axis at x = a. Eq. (4) is, however, unnecessarily com- Considering the steel stress as as constant, and substi-
plicated and may be replaced by the following simpler tuting Eq. (5), we further obtain
expression
(9)
(5)
in which k and m are certain constants. These con- in which Cz is some constant.
stants can always be chosen so that the values given by Finally, summing the contributions from composite
Eq. (4) and (5) are almost undistinguishable; see Fig. 3. beam action and arch action, V = VI + Vz, and calcu-
v ::;: kl pP
ro
(fclq + k2 -(a/d)'
p-
) (1 + -d ),V, (11)
>'-.od.
The latter of these equations yields C I-2 as the vertical
in which Ao represents an additional empirical parame- axis intercept, and C I-2/ Ao as the slope of the regres~
ter. sion line. Data that exist for one particular concrete can
be easily and very closely fitted with Eq. (13), as ex-
STATISTICAL ANALYSIS OF EXISTING TEST emplified in Fig. 4 for Walraven's and Kani's test re-
DATA sults. This plot clearly indicates that the size-indepen-
Shear failure of beams is one problem for which ex- dent strength criterion, currently implied in code for-
tensive statistical information has been accumulated mulations, contradicts experimental evidence. So does
over the years. This statistical basis was exploited by the linear fracture mechanics, which corresp~nds in
Zsutty38.39 for the development of a very simple predic- Fig. 4 to the straight line of slope - Yz.
tion formula, the best one proposed up to now. His None of the available data for one particular con-
statistical analysis, however, did not particularly cover crete and fixed p and a/dare, however, sufficiently ex-
the size effect and was made before some of the impor- tensive to allow statistical analysis. For that purpose, all
tant test results on the size effect became available. Eq. the aforementioned available data for the size effect
(11) proposed here has been compared to essentially all must be analyzed collectively. This cannot be done by
important experimental evidence, both that with regard linear regression alone, since coefficients Ao, p, q, and r
to the effect of steel ratio, shear span, and concrete of Eq. (11) are involved nonlinearly. Trial and error
strength, and the more limited one with regard to the approach coupled with nonlinear optimization (Mar-
effect of size (beam depth). The test data used included quardt-Levenberg algorithm)40.41 has been used to de-
those of Moody et aI., 19 Diaz de Cossio et aI., 23 termine the optimal values of these coefficients. The
MatheY,17 Van den Berg,27 Taylor, 26 Rajagopalan,22 optimum fit of the data, achieved with Eq. (11), is
Kani,12.13 Leonhardt and Walther, 16 Walraven,28 Tay- shown in Fig. 5(a) as the linearized regression plot [Eq.
,1,--------:-----------,
(a) Walraven a/d=3 (b) Kanl a/d=3
pO.8 " p2.75 "
1:=4000 (pII) 1,'=4000 (pII)
'~
- - -- Itrongth
- . -ar-ylold
- -crltorion
--
'~.~
~ -.1
-=-to>
.3
-.2
nonlinear fractur.
'""
m.ChanIC.~
A,,-13.'~.
>,,_25 2
-.3 +----+-----+----*---_+_~
.75 1.00 1.25 1.50 1.75 .75
"". . 1.00 1.25 1.50 1.75
..
K Bhal (1968) .trength or yield criterion
• Swamy· (1971) .0 +-----'~---L.'--'---"....,--'---------1
.. e
3
. ·. , oj
..
•
• K «
• • K
Fig. 5 - Comparison with existing test data for beams of different sizes
(13)], and in Fig. 5(b) as the plot of the logarithm of line plot represent the errors. A plot of this type is
the nominal shear strength versus the logarithm of rel- shown, for the proposed Eq.(1l), in Fig. 6(d). The
ative size dl da. Due to combining test data for differ- standard deviation of the vertical errors with regard to
ent concretes, different beams, and from different lab- the regression line, and the correlation coefficient r, are
oratories, the scatter is now much larger; however, the listed in the figure. It should be mentioned that the val-
size effect is clearly confirmed. The strength criterion, ues of f: were considered in psi (1 psi = 6895 Pa), and
implied in the current design approach, would corre- that the cubic strength of concrete fcc (fcc is in psi), when
spond to a horizontal line in both Fig. 5(a) and 5(b), indicated, was converted to the cylindrical strength ac-
which would obviously contradict the test data. On the cording to the formula f: = [0.76 + 0.2 log (fccl
other hand, a straight line of slope - Y2 in Fig. 5(b) 2840)] f«.42
would give too strong a size effect, in clear disagree- As a result of all these statistical comparisons, the
ment with test results. following formula is proposed for the mean ultimate
Note also that if the range of beam sizes were re- nominal shear strength
duced to about one-half, no clear size effect would be
apparent and the strength criterion would represent the 104;>
data for the smaller beams as well as the present the- v = [.JT[ + 3000 "';pl(ald)S] (14)
u "';1 + d/25d.
ory. This accentuates the need of testing beams of
widely different sizes if any evidence on the size effect
should be obtained. Fig. 5(b) also brings to light the For the sake of comparison, the formulas used in
gradual transition from a strength criterion that applies ACI Standard 318-77 43 and in the CEB-FIP Model
for small beams to a linear fracture mechanics criterion Code 197844 have also been used to fit these data. How-
that would no doubt be applicable to very large beams. ever, the coefficients of these formulas had to be dis-
To obtain more extensive evidence, and also to ver- regarded because they are not intended to give the mean
ify the dependence on p, aid, and f:, numerous further values of the ultimate nominal shear strength but the
data were included, although each of them, taken values for the initiation of cracking. These formulas
alone, provides no information on the size effect as may be written as
such. In their majority, these further data correspond
to the smallest beams admissible for a given aggregate,
and thus they cannot be expected to strengthen the evi-
v = Min (kl .JT: + k zp ~~, 3.5 .JT:) (ACI) (15)
/'
/'
--,r:..
., 2.8
+i
.;
+ /,
/
~+
+/'
/'
]
~
2.2
Fig, 6 - Comparison of various formulas with the bulk of the existing test data on
the ultimate shear strength
TRd 0.01 fck + 0.06 for fCk ~ 20 Table 1 - Coefficients obtained by nonlinear
TRd 0.008 fck + 0.1 for fck > 20 (17) regression for Vu
K = Max( 1. 6 - d, 1) Number of
p = Min(A/bd, 0.02) for CEB-FIP available
Model test data k, k, A. p q r
ACI318-77 377 1.64 7423
In Eq. (17), d is in meters, but fck in the present form CEB-FIP 377 1.31 54.7
must be given in MPa. Coefficients kl and k2 in these Z tt {a/d~2.5 296 58.4 0.38 0.35 0.28
su Y a/d<2.5 81 7829 0.554 -0.0057 1.50
formulas have been optimized to obtain the best fit of Proposed model 296 7.23 3284 25 0.29 0.52 2.51
all the 377 data points. The resulting optimum fits are
shown for the ACI and the CEB-FIP formulas in Fig.
6(a) and (b), and the values of the optimum coeffi- in which kl' r, p, and q are four empirical constants.
cients are listed for these formulas as well as the pres- The values of these constants have been optimized
ent model in Table 1. The scatter apparent from these again to obtain the best possible fit of the 377 data
figures and quantified by the values of standard devia- points used in Fig. 6. The resulting plot is shown in Fig.
tion and the correlation coefficient in these figures is 6(c). It may be noted that this formula agrees with the
obviously much larger than the scatter for the pro- data nearly as well as the proposed formula. However,
posed formula, especially in the case of CEB-FIP for- the size effect evidenced in Fig. 5 is not modeled by this
mula. It must be kept in mind, however, that these for- formula, although it could be introduced by mUltiply-
mulas are not intended to give the ultimate strength but ing the formula with the function ct>(A). Note also that,
the initiation of diagonal shear cracking (as loosely as in contrast to the present formula, Zsutty's formula is
it may be defined). purely empirical, not based on some mechanics analy-
The best previous formula is doubtless that of sis.
Zsutty,l8 which reads As another useful statistic, one may consider the
population of the values of Y = (Vtes/veal) - 1, in which
kl Veal is the calculated value and Vtest is the measured value
v = q
(18)
(a/ d)' pi' .t: of nominal shear strength. These values are plotted as
ACI JOURNAL I September-October 1984 461
a function of the logarithm of relative size in Fig. 7. The statistics for the nominal cracking shear stress Vc
The proposed formula [Fig. 7(d)] appears best. Zsut- may also be worked out for the variable Y = (Vtes/Vcal)
ty's is nearly as good, although it does not reflect the - 1. These values are plotted for all the four formulas
declining trend of the data as a function of dl da, ap- against the logarithm of the relative size dlda in Fig. 9.
parent from Fig. 7(c) as well as Fig. 7(d). The scatter It is apparent that, for crack initiation, the size effect is
for the ACI formula [Fig. 7(a)] is much larger, and much less pronounced, nearly undetectable. In fact, for
even more so for the CEB-FIP formula [Fig. 7(b)]. De- the true crack initation, the size effect should be non-
spite this larger scatter, a declining trend with regard to existent, since the beams do not contain any initial
the size is noticeable in Fig. 7(a) and (c). stress concentrator (a notch). The fact that any size ef-
fect seems to be apparent indicates that the observed
DESIGN PHILOSOPHY: CRACK INITIATION OR values of Vc did not in fact correspond to the true crack
FAILURE? initiation, which, of course, is very difficult to define as
The philosophy of the present design codes is to the cracking begins by a gradual formation of invisible
achieve a certain safety not against the ultimate load in microcracks.
diagonal shear failure but against the load for which The fact that no significant size effect on the nomi-
the initiation of diagonal shear cracks is observed. nal shear stress at cracking is observed while at the
Therefore, comparisons have also been made with the same time the size effect is clearly confirmed by tests of
(much less numerous) available test data on the nomi- the ultimate nominal shear strength raises a question
nal cracking shear stress VC. Fig. 8 shows such a com- with regard to the present design philosophy of design-
parison in terms of the measured Vc. against the calcu- ing against crack initiation rather than ultimate failure.
lated Vc> as obtained by optimizing the coefficients in Since the ultimate value of p decreases with size, there
the formula in Eq. (11), (15), (16), (17), and (18). (Note obviously exists a certain sufficiently large size for
that all the existing test data for the initiation of diag- which the ultimate p ceases to be larger than the value
onal shear cracking are confined to the values of dl d a of p for crack initiation. This is, of course, natural to
between 10 and 24.) In Fig. 8 the proposed type of for- expect. It is known from nonlinear fracture mechanics
mula again gives the best agreement with test data, al- that the strength reserve due to stable crack growth be-
though only slightly better than Zsutty's formula. The comes smaller as the size increases and vanishes when a
improvement compared to the ACI formula is, in terms certain size is exceeded.
of Vc> not very significant, but it is more significant Thus, designing against the crack initiation rather
compared to the CEB-FIP formula. The values of the than ultimate failure does not assure a uniform safety
optimized coefficients used in plotting Fig. 8 are listed margin. The safety margin decreases with increasing
in Table 2. dlda, and for a sufficiently large dlda the safety margin
.9~~------------------------,
(0) ACI (b) CEB-FIP
~ +7400 P Vu d ~ 3.5v'1;;
.7 .
c= 1.64
• Mu
C= 1.3
. TRdX:( l-t.55p)
.5
:
'1 :
..
.3 ·~ I
•
:~I i1 , .•
J •
~.
.1
.',
f·' :-I·,r: ..:;
f. •
"" . ·. ".. . :.
~ -.1
.4 r--c(;--'c)C"Z:C:S""'U=TTC-- - - - - - - - - - - - - - - - - - - - - ,
Y (d) PROPOSED FORMULA
.2
t
.
~
-.2 C=60 Vf~ pd/a for aid;;; 2.5 •
C=150 Vf~p (d/a)' for a/d<2.5
-.4 +-----+-----+-----+----+-----+----....;
.6 .8 f.O f.2 f.4 f.6 f.8.6 .8 1.0 1.2 1.4 1.6 1.8
log (d/d.)
.....
'Ii~ 2.3
.
-=-- 2.1
2.3
.5
~
l 2.1
1.9
No. of dolo 118 No. of dolo 118
S.D. = 16.54 (pil ) S.D. = 16.39 (pil )
=0.124 =0.925
..·•
.15
•
!.
., i
.05
·.i . i ..•;• .
I .1 l
-.05
·• •
'""'
~ -.15
-.:..
S -.25 +---+---!------;e----4----+--~
i!
• S
. r *•
.1-----. -..r t·---1::~25
f
. 05
i . I'
.
-.051-_ ••
•·L • I
i: -.15 : ;
-.05
•i
••
t. '---
-.1 5 +---+---!----e-----;----4--~ -.25 + - - - + - - - + - - - > - - - - e - - - - - ; - - - - - 4
.9 1.0 1.1 1.2 1.3 1.4 1.5 .9 1.0 1.1 1.2 1.4 1.5
log (d/d.)
Fig. 9 - Same comparison with test data as in Fig. 8, but in different variable
2.8
.~
, '"
'~" ".." ~
$ 2.4
~
•
;.
"
'"
-'l 2.0
1:i
~
~
;l
'"t:l
~
~ 1.6
3.2
'"'"
~ (e) ZSUTTY (d) PROPOSED FORMULA
.t
V]
~
~
"'V]" 2.8
, .......
(
~
"'il
-: *+ +
++:t.... . , t;
~ .... + :",,++
S 2.4
:1.+ • .:
"il
.~
,:..i1'-:;:
~
0
:;,; ;$r
2.0
:-i":}:
/ 1Ju~CF'60 'If; pd/a for a/d~ 2.5 v. ~ (Vf;+3000Yp/(a/d)")
/ 1Ju~CF'150 'l./f,p (d/a)' for a/d<2.5 l+d 25d.
CF~O.75
1.61;C.6----;~-2~.0:---+--~2.-4-~----:2+:.8----;~----::<3.2 1.6 2.0 2.4 2.8 3.2
Nominal Ultimate Shear Stress Calculated by Models
Fig. 10 - Comparison of the design formulas with the bulk of existing data
464 ACI JOURNAL I September-October 1984
mula, which is proposed here as a replacement for the bers of points that lie above the inclined straight line
present code formulation of ACI or CEB-FIP are n = 249, 289, 286, and 286, respectively.
Fig. 11 shows the same comparisons as Fig. 10, but
in different variables. The ordinates are, similarly to
v 8* (fJ'1 + 3000 ~~)
~
d VIc a 5 ' Fig. 9, the values of Y = (Vtes/Vcal) - 1. The compari-
1 + -- sons lead to similar conclusions as Fig. 10. (Note that
25da only 25 data points of the 77 data points of Kani 13 are
Vud plotted in Fig. 7 and 11, although all data points agree
with a = (19)
Mu with the formula well; this is because they all refer to
the same beam size and are all crowded in such a small
Here a = a/ d for the case of concentrated load (Fig. spot that they could not be graphically distinguished.)
2), and a = fl4d for that of uniform load (heref:must From Fig. l1(a) we see that the points laying signifi-
be in psi). This formula is shown by the straight line in cantly below the ACI formula correspond indeed to
Fig. lO(d). Unlike Fig. 10(a), no data points fall signif- large beam sizes, which again confirms our previous
,icantly below this formula. Note that the band of data argument about the size effect and how it affects the
points based on this formula becomes much narrower safety margin.
than for the present code formulations [Fig. lO(a),(b)]
and is also somewhat narrower than for Zsutty's for- REMARK ON THE EFFECT OF SHEAR
mula shown in Fig. lO(c). This formula was also scaled REINFORCEMENT
(replacing kl in Eq. (18) by ko = 0.75k l) so that only a It is certainly a reasonable design approach to as-
few data points would lie below the formula. sume, as is done in current codes, that the ultimate load
The fact that no data points in Fig. lO(d) for the in presence of shear reinforcement (stirrups) is a sum of
present formulation lie high above the straight inclined the ultimate load in absence of the shear reinforcement
line means that the proposed formula is overall eco- plus the additional capacity due to the shear reinforce-
nomic. The economy may be quantitatively character- ment alone, obtained by plastic analysis. This ap-
ized by factor cf>. = E; tJ./ny where y = ordinate of data proach, however, is not as easy as it might seem, since
centroid, tJ.; = vertical deviation of data points from the presence of shear reinforcement mitigates the size
the straight line (i = 1,2, . . . n), and n = number of effect, as shown in Reference 34 by a similar dimen-
all points. Only the points lying above the straight line sional analysis. No meaningful experimental evidence
are counted. The smaller is cf>.. the better the economy. seems to be available for the size effect in presence of
Calculations yield cf>. = 0.595, 0.844, 0.448, and 0.290 shear reinforcement. It is, nevertheless, theoretically
for Fig. lO(a), (b), (c), and (d), respectively. The num- evident that the reduction in the loss of safety margin
(b) CEB-FIP
,6
..
,..
:
~
t
~8.
:t..
.2 • t
:'1.
0';0 • J . .
L
'"""
~;>
0,0
o ..
...-!. ,.. .
0
$ ~
.
'- 'I
0)
.s -.2
.5
(e) ZSUTTY
. r ..
(d) PROPOSED FORMULA
.3
...". : iI 0
I..
.:
• .~. I ..1~ !~
f.· I
.
.1
00 0 •
C,=O.75 , ~
-.1
C=C,150 Vf~p (d/a)' for a/d<2.5
C=C"60 ~Pd/a for a/d~2.5
-.3 + - - - - + - - - + - - - - - i - - + - - - + - - - - - - "
.6 .8 1.0 1.2 1.4 1.6 1.8.6 .8 1.0 1.2 1.< 1.6 1.8
log (d/d.)
W = -1 (P)'
- bd'f(cx" a" ~;) (21)
mda (1 _E') f:'
E, 2£,
(24)
2£, bd
oW
= Gfb (22)
U
N
= BI"
'J, ( + - d)
1
}...Pa
-tl,
(25)
oa
in which Gf is the fracture energy, a material property characterizing in which B = [(1 - E,IE,)/f,)'12 and -,...., = mI,/f,. Band -,.... are con-
the energy consumed per unit extension of the fracture, per unit stants when geometrically similar structures of different sizes are
thickness. considered. Thus, Eq. (25) proves our starting equation, Eq. (1).