Documente Academic
Documente Profesional
Documente Cultură
A theory of shear transfer in initially uncracked concrete is pre- in the compression struts formed approximately paral-
sented. The theory is based on the truss model and incorporates a lel to the direction of the cracks. 12. 16 The compression in
softened compression stress-strain relation along the concrete s/ruls.
For reinforced concrete specimens experiencing shear transfer across
the struts and the tension provided by the reinforcing
a plane, acrilical zone in the vicinity of the shear plane is identified. bars across and parallel to the shear plane constitute a
Within this zone, the stress distribution is assumed to be approxi- truss-like action. Although this truss-like action is well
mately uniform after the formation of cracks. The governing equa- recognized, 1•8 direct application of the truss model
tions derived from the theory can then be applied to this crilical zone would result in a much higher prediction of the shear
to obtain strain responses for the given stress conditions. The ulti-
mate shear transfer strength is identified by tracing the complete shear
strength. An attempt was made 8 to bring the predicted
stress-strain history using electronic computer. Comparison of theo- shear strength in line with the measured strength by in-
retical predictions to 32 test results reported in the literature gives troducing shear stresses in the compression strut result-
good agreement. ing in a biaxial failure condition. 17 • 18 However, this ad-
The theory predicts that steel reinforcement parallel to the shear ditional shear stress in the compressional strut consid-
plane also contributes to the shear transfer strength, while the shear-
friction concept in the current design codes recognizes only the con-
erably complicates the truss model theory.
tribution of steel reinforcement crossing the shear plane. Since the The fundamental difficulty in predicting the shear
current design codes are based on test specimens with heavy rein- transfer strength of initially uncracked concrete is in the
forcement parallel to the shear plane, they could be unconservative uncertainty of the compressive strength of the strut. In
for the practical cases where only light reinforcemenl is provided a study of the behavior of reinforced concrete panels
parallel to the shear plane.
Keywords: building codes; failure mechanisms; friction; reinforced concrete;
reinforcing steels; shear properties; shear strength; stress-strain relationships;
structural analysis.
(1)
under predominately shear stresses, it was found that
compressive strength of the diagonal struts formed af-
ter the cracking of the concrete can be much lower than (2)
the standard cylinder strength. 19 •20 This phenomenon
has been called the softening of concrete. This soften- Trrc = (ac - a,) sina COSO! (3)
ing of the concrete struts is related to the tensile strain
in a direction perpendicular to the struts. Using the These relationships can also be presented using Mohr's
softened stress-strain relation proposed in Reference 19, stress circle.
the shear strength and behavior of various reinforced The steel reinforcement is assumed to contribute only
concrete members had been predicted with good accu- normal stresses
racy. 21-2s
a/s = Pt j, (4)
RESEARCH SIGNIFICANCE
In this study, the softened truss model theory was (5)
applied to the shear transfer problem and was found to
be successful in predicting the shear transfer strength as
well as the shear deformations of 32 initially uncracked normal stress in steel in /- and !-direc-
specimens. Contrary to the well-known shear friction tions, respectively
concept, the theory predicts that the ultimate failure is p,, p, reinforcement ratio in /- and !-direc-
caused by the crushing of concrete in the compres- tions, respectively
sional struts formed after cracking of concrete. Fur- j,,j, steel stress in /- and !-directions, re-
thermore, the transverse reinforcement parallel to and spectively
in the vicinity of the shear plane also has an effect on The total stress in a reinforced concrete element is the
the shear strength. Since the ACI shear friction provi- superposition of the concrete stresses, Eq. (1) to (3),
sions are based on test specimens with very high trans- and the reinforcement contribution, Eq. (4) and (5)
150 ACI Structural Journal I March-April 1987
(6) O'"d
(7)
Eq(12a)
Eq(12b)
T 11 = (ad - a,) sina cosa (8) fI
c
-T
Strain transformation conditions (compatibility)
Assuming that the strains are distributed uniformly
in the element, they can be transformed according to
the following equations
(9) E =~
P >-.
(10)
(a) Compression Stress-Strain Relationship
"( 1, = 2(Ed - E,) sina cosa (11)
Material laws
The stress-strain relation in the direction of the
compression strut is represented by the following two
equations suggested by Vecchio and Collins 19 and
(b) Tension Stress-Strain Relationship
shown graphically in Fig. 3(a).
Ascending branch
Descending branch
a = _
d A
l
J: 1 _ (EdiE 0 1/A) 2
-
2 - 1/A
l (12b)
(15a)
taken as - 0.002. A is a coefficient to take care of the derivative of ad with respect to Ed in Eq. (l2a), and then
softening phenomenom and is expressed by taking Ed equal to zero.
For the descending branch after cracking the rela-
tionship is assumed to be
if E, > E0 (15b)
Substituting E1 and E, from Eq. (9) and (10) into Eq.
(13), the expression for A can be simplified to
A=
R d
(14) where !a concrete cracking stress, taken as
4Jf: if 1: is expressed in psi. E,, = concrete cracking
strain = f j E,. This expression is not the same as the
where E, = E1 + E, - Ed. This relationship for E, can equation given in Reference 19. In Reference 19, the
easily be observed from Mohr's strain circle . 26 term under the square root in the denominator is E,
The stress-strain relation in a direction perpendicular rather thanE, - E This correction makes the resulting
0 •
to the compression strut is shown in Fig. 3(b). Before Eq. (15b) consistent with the condition that a, = ;:,
the concrete is cracked, the ascending linear relation- when E, = E". The effect of this modification is minor,
ship is since Ecr is usually much smaller than E,.
ACI Structural Journal I March-April 1987 151
Eliminating P 1 from Eq. (18) and (19), but keeping Ka
and Kr as variables
(20)
TR. Denoting
1 (21)
where Ka coefficient describing the nonuniform Substituting Eq. (17) into Eq. (24) and using Eq. (10)
distribution of stress a1 for E1
coefficient describing the nonuniform
(ad - a,)K sina cosa
distribution of stress Tc
b thickness of test specimen (Fig. 4) = ad sin 2 a + a, cos 2a + P1/ry (25a)
h width of test specimen in the longitudi-
nal direction (Fig. 4) and (ad - a,)K sina cosa
length of shear plane in the transverse = ad sin 2a + a, cos 2a
direction (Fig. 4) + p1Es(Ed sin2a + E, cos 2a) (25b)
152 ACI Structural Journal I March-April 1987
Note that Eq. (23) and (25) are expressed in terms of 9. Select another value of Ed and repeat Steps 1
the six unknowns listed in the preceding paragraphs. By through 8. In this way, a set of solution for various Ed
selecting a value for Ed, the other five unknowns, ad, a, values can be obtained.
E, a, and A, can be solved by Eq. (12), (14), (15), (23), 10. The value of Tit, E1, E, "f 1,, j,, and j, can be calcu-
and (25). Ed is selected because it is expected to vary lated from Eq. (8), (9), (10), (11), (16) and (17), respec-
monotonically as the load is increased. Once these six tively. The relationship of any two variables, such as T1,
unknowns are obtained, the stress and strain in the 1-t versus "( 1,, can be plotted.
axis (T,, E,, E,, "f1, j,, j,) can be easily calculated.
The iterative procedure to solve the five nonlinear si- COMPARISON WITH TESTS
multaneous algebraic equations is as follows: The theory described in the previous steps will now
1. Select a value for Ed. be applied to the shear transfer problem. A typical test
2. Assume a value of a,. specimen for shear transfer across a vertical shear plane
3. Solve forE, from the stress-strain curves of Eq. is shown in Fig. 4. To apply the theory, it is necessary
(15) to know the ratios among the three in-plane stresses a,,
a, and Tit. These stresses should also be uniformly dis-
E, I[_0.005 (I,a,. tributed over the region of interest. However, exami-
nation of the test specimen in Fig. 4 shows that the
stresses cannot be expected to distribute uniformly over
a, the entire specimen, nor can they be expected to be
E =-
, E,. uniform in the central test region. Before cracking, the
shear stress along the shear plane T 1, should be consid-
4. Find A from Eq. (14) erably larger near the two ends of the shear plane where
the open slot disrupts the smooth geometry and intro-
duces local stress concentration. For the same reason,
the transverse normal stress in the direction of the load
a, is larger near the two ends of the shear plane. The
normal stress in the longintudinal direction a, is small
5. Find a" from Eq. (12) and can be neglected.
After diagonal cracking, a cracked region is ob-
served in the vicinity of the shear plane and eventually
leads to failure. This cracked region will be called the
critical zone and is the shaded area shown in Fig. 4.
A typical width of this zone was observed to be about
2 to 3 in. for a 10 in. wide specimenY Within this zone,
the extensive cracking of the concrete had an effect of
redistributing the shear stress and the transverse nor-
6. Solve for a from Eq. (23) mal stress more evenly along the shear plane. The
cracking also reduced the stiffness in the zone as com-
pared to that outside of the zone. This would cause a
redistribution of the compression stress in the trans-
verse direction to become more evenly distributed
across sections perpendicular to the shear plane. Thus,
within this critical zone the stresses might be assumed
to be uniform and the theory developed in the previous
then calculate sin 2a, sina, and cosa. sections could be used. More specifically, the shear
7. Solve for a, from Eq. (25) stress T" was estimated as the average stress over the
entire shear plane (i.e., Kr = 1); and the compressive
stress in the transverse direction a, was estimated as the
aAK sina cosa
a, average stress over a cross-sectional plane perpendicu-
lar to the shear plane (i.e., Ka = 1). The K ratio, there-
fore, becomes 1/h as shown in Fig. 4. The normal stress
aiK sina cosa - sin 2a) in the longitudinal direction a, is assumed to be zero
- p,£s(Edsin 2a + E,COS 2a) [Fig. 4(b)].
K sina cosa + cos 2a In determining the reinforcement ratio, the cross-
sectional area of the longitudinal steel across the shear
8. If the calculated a, is close enough to the assumed plane is divided by the area of the shear plane to obtain
a, value, a set of solution ad, a, E, a, and A has been p 1, and the area of the transverse steel is divided by the
obtained for the selected Ed value. Otherwise, a new a, cross-sectional area of a plane perpendicular to the
is calculated by a bisection method and Steps 2 to 7 are shear plane to obtain p,. As the cracks in the critical
repeated. zone are constrained by the two rows of transverse
ACI Structural Journal I March-April 1987 153
steel, and the transverse steel is almost uniformly de-
1500 ployed over the whole section, the use of the average
steel ratio over the whole section for the transverse re-
inforcement ratio in the critical zone is considered ap-
propriate.
~
The initially uncracked shear transfer tests reported
Ill
.e in the literature were studied by tracing the shear stress-
1-- 1000 shear strain history by the method as previously de-
ui scribed. Fig. 5 and 6 show the shear stress versus shear
VI
w strain curves for Specimens M2 and M6, respectively,
cc:
1-
VI
obtained from References 13 and 16. For convenience,
cc: the starting point is taken at the zero stress state and
c SPECIMEN M2
w
:z:
successive tracing is done from uncracked state to
NON-SOFTENED CONCRETE
VI
500 cracked state of the concrete, even though the imposed
SOFTENED CONCRETE ratio between the normal stress and shear stress is only
-x- TEST applicable to the cracked state as explained previously.
6. YIELDING OF LONGIT. STEEL
Fig. 5 and 6 each provide three curves: one experi-
mental and two theoretical. One theoretical curve is
0 PEAK STRESS IN CONCRETE
based on the softened compression stress-strain rela-
tionship given by Eq. (12) and (14) and shown in Fig.
0 0.005 0.01 0.015 0.02 0.025 3(a), while the other one utilizes the nonsoftened
SHEAR STRAIN, )" compression stress-strain curve specified by the CEB-
FIP Model CodeY The CEB-FIP curve has a para-
bolic-rectangular shape. The ascending parabolic curve
up to a strain of 0.002 is described by Eq. (12a), if the
Fig. 5-Shear stress-shear strain curves for Specimen
coefficient 'A is taken as unity, and the continuing hori-
M2 (1 psi = 6.895 kPa)
zontal branch terminates at a strain of 0.0035. Fig. 5
and 6 show that the theoretical curves using the soft-
ened compression stress-strain curve agree very well
with the experimental curve. In contrast, the theoreti-
cal curves based on the nonsoftened compression stress-
2000
strain curve overestimates considerably the maximum
stress as well as the strain at maximum stress.
It should be mentioned that the truss model theory is
not intended for the prediction of behavior before
cracking. Tests in Fig. 5 and 6 show quite reasonably
that the specimens before cracking are considerably
1500 stiffer than those predicted. Only when the ultimate
·;;
strength is approached can the predicted shear stresses
.!: //'-
and shear strains become valid .
l-
f -----i(
ui I
Push-off tests
Ill I
w
1!: 1000 I
I
The two specimens M2 and M6, discussed previously
Ill
I and shown in Fig. 4, are subjected to the so-called
a:: I
c push-off loading. A total of 20 push-off tests is re-
w
:1:
Ill ported in References 13, 16, and 28 for initially un-
- SPECIMEN M8 -
cracked specimens. The test results are compiled in Ta-
-0- NON-SOFTENED CONCRETE ble 1, including the predicted shear stresses 7"'"'·'' the
500 -e- SOFTENED CONCRETE shear strains at peak stress 'Ymax,n and the longitudinal
-X- TEST steel strains at peak stress E1• The shear stresses were
-6- YIELDING OF LONGIT. STEEL computed from Tmax,c = p,lbl, assuming K = l. Also,
-0- PEAK STRESS IN CONCRETE
assuming Ka = 1, then K = 1/h. For specimens No. 1
through 14, K = 10/10 = 1, and for specimens No. 15
through 20, K = 10/12 = 0.83. This means that the
0.002 0.004 0.006 0.008 0.01
shear stresses are assumed to be uniformly distributed
SHEAR STRAIN, 'Y across the shear plane and the transverse stresses are
uniformly distributed on the plane perpendicular to
that.
Fig. 6-Shear stress-shear strain curves for Specimen A comparison of the calculated and experimental
M6 (1 psi = 6.895 kPa) maximum shear stresses is given in Fig. 7. The agree-
154 ACI Structural Journal I March-April1987
ment between the calculated and the measured shear
strengths is indeed very good. The mean value of the ·;
2- 1500
ratio of the measured shear strength to calculated shear "'"'w
strength is 1.054 and the standard deviation is 0.069. ...a:
<f)
y
I I. lA 0.0044 0.0568 50,700 3920 0.0008 0.3965 62.79 3.415 0.314 750 684 1.0965 4.333
2 l.IB 0.0044 0.0568 48,000 4340 0.0008 0.3749 63.81 3.977 0.356 844 696 1.2126 4.700 y
3 1.2A 0.0088 0.0568 50,700 3840 0.0010 0.4966 57.72 2.113 0.242 1000 923 1.0834 3.933 y
4 1.2B 0.0088 0.0568 48,000 4180 0.0010 0.4701 58.91 2.538 0.287 980 930 1.0538 4.267 y
5 1.3A 0.0132 0.0568 50,700 3840 0.0016 0.5859 54.05 1.769 0.172 1100 1109 0.9919 4.887 y
6 1.3B 0.0132 0.0568 48,000 3920 0.0013 0.5617 54.85 1.716 0.195 1070 1098 0.9745 4.246 y
7 1.4A 0.0176 0.0568 50,700 4510 0.0017 0.5996 53.18 1.658 0.181 1360 1326 1.0256 5.028 N
8 1.4B 0.0176 0.0568 48,000 3855 0.0017 0.6139 52.74 1.479 0.141 1280 1173 1.0912 4.836 N
9 1.5A 0.0220 0.0568 50,700 4510 0.0017 0.6161 52.39 1.429 0.159 1400 1377 1.0167 4.822 N
10 1.5B 0.0220 0.0568 48,000 4065 0.0017 0.6253 52.09 1.325 0.132 1384 1268 1.0915 4.711 N
11 1.6A 0.0264 0.0568 50,700 4310 0.0017 0.6332 51.67 1.218 0.130 1432 1366 1.0483 4.614 N
12 1.6B 0.0264 0.0568 48,000 4050 0.0016 0.6333 51.66 1.153 0.120 1420 1300 1.0923 4.355 N
13 6.1 0.0044 0.0568 48,000 3960 0.0008 0.3893 63.21 3.597 0.321 800 672 1.1905 4.441 y
14 6.2 0.0220 0.0568 48,000 3930 0.0017 0.6282 52.01 1.291 0.125 1240 1235 1.0041 4.672 N
15 Ml 0.0044 0.0587 50,900 4180 0.0008 0.3806 63.03 3.778 0.386 760 695 1.0935 4.660 4.000 y
16 M2 0.0088 0.0587 52,700 3900 0.0011 0.4924 57.33 2.348 0.318 980 932 1.0515 4.423 3.333 y
17 M3 0.0132 0.0587 52,300 3995 0.0015 0.5689 53.81 1.812 0.273 1110 1131 0.9814 4.486 4.667 y
18 M4 0.0176 0.0587 50,900 4150 0.0017 0.5995 52.35 1.588 0.253 1140 1233 0.9246 5.072 4.667 y
19 M5 0.0220 0.0587 52,700 3935 0.0017 0.6202 51.39 1.314 0.220 1280 1225 1.0449 4.814 6.667 N
20 M6 0.0264 0.0587 52,700 4120 0.0017 0.6281 50.88 1.197 0.219 1320 !304 1.0123 4.713 3.333 N
K = 1.0 for No. 1 through 14; K = 0.83 for No. 15 through 20. No. I through 14 are taken from Reference 28 and No. 15 through 20 are taken from References
13 and 16. 'Ym~., -y,,,,, = experimental and calculated shear strains at maximum stress, respectively. Y = steel yielded; N = steel not yielded.'" 1/A a ,, and, are
values at maximum stress. S = longitudinal steel. I psi = 6.895 kPa. ' ' ' ' '
rr:
y
a:
C(
a:
C( 1.0
1-
Ul I
w w a: oT
::r:: ::r:: C(
max,c
!/) !/) w I
::r::
x x Ul 500 practical
I
I *Tmax,t
C(
:::E
C(
:::E 0.5 x
C(
regk»n - 1
I TEST
:-sPECIMEN
a:w <..i
-1
:; I
ll. C(
><
w
(,)
0.02 0.04 0.06 0.08 0.1
0 0.5 1.0 1.5 2.0 2.5
TRANSVERSE STEEL RATIO, p1
K-RATIO
Fig. 8-Ejject ojK-ratio on calculated maximum shear Fig. 9-Ejject of transverse steel ratio on calculated
stresses maximum shear stresses (1 psi = 6.895 kPa)
Number Specimen p, psi psi psi 0.0025* 0.0055* 0.011''' 0.022* 0.0293* 0.044* 0.0587' 0.0733* 0.088*
16 M2 0.0088 52,700 3900 980 700 777 829 877 895 918 932 942 950
20 M6 0.0264 52,700 4120 1320 1174 1209 1243 1273 1284 1297 1304 1310 1313
*Assumed p,.
tp, in actual test specimen.
I psi = 6.895 kPa.
insensitivity of the shear strength to the assumed K-ra- specimens, M2 and M6 (p, = 0.0088 and 0.0264). The
tio simply means that the shear strength is not sensitive transverse reinforcement ratio is assumed to vary from
to the unevenness of the compressive stress distribution. p, = 0.0025 to 0.088 for cases shown in Table 3. The
The choice of the simple expression K = 1/h would effect of the transverse steel ratio on the maximum
have the advantage of simplicity and would provide shear stress is plotted in Fig. 9. Fig. 9 shows that
sufficient accuracy. changing the reinforcement ratio from the actual 0.0587
The sensitivity of shear strength to the assumption to 0.0293 or from 0.0587 to 0.088 results in a change of
made in determining the amount of transverse steel is shear strength by Jess than 4 percent. If, however, the
studied by comparing the results for two of the test transverse steel ratio is reduced from the test specimen
156 ACI Structural Journal I March-April 1987
<'>
0...
8.0 .--------r------.---.------::l
~
.....
Ill
E
)>...
rri 6.0
Ill
w
a:
I-
I/)
><
4.0
""
::IE
I- ......
"':ccz
a:
I-
I/)
2.0 p = 12"
a:
1:
SPECIMEN 15-20
""
w
:1:
Ill
a:w 0 2.0 4.0 6.0 8.0
a..
><
w
CALC. SHEAR STRAIN AT MAX. STRESS, Y max,c (X10" 3 )
I
the shear plane is generally not available in practical
~ ~
structures. Thus, design guidelines based solely on these
tests may not be conservative.
Also compared in Table 1 are the shear strains at
peak stress for the six specimens for which the test val- I[ I JJ I 7
•
ues were reported from measured slip across the shear
plane. A test shear $train is calculated by dividing the
measured slip by the gage distance across the shear
r--- h = 14" ---1
plane. A comparison of the calculated and experimen-
tal shear strains at maximum stress is given in Fig. 10. Fig. 11-Push-off test specimen with longitudinal ten-
The predicted shear strains at peak stress are in gener- sion stress (Reference 14) (1 in. = 25.4 mm)
ally good agreement with the test values. The scatter
can be explained by the fact that the slip across the tensile stress is kept constant during the test, the shear
shear plane was measured at one level across the shear strength is expected to be weakened. Test results are
plane, while the predicted shear strain is an average available to show this weakening effect. 14 The present
value. theory can be applied easily to such cases by assigning
a constant normal stress in the longitudinal direction.
Push-off tests with imposed longitudinal tensile In these test specimens K = 12/14 = 0.86. Shown in
stress Table 4 are the measured and calculated shear strengths
A tensile stress is applied in the longitudinal direc- for six tests, including four with imposed longitudinal
tion to a test specimen, 14•15 as shown in Fig. 11. If the tension. The agreement is acceptable but not as good as
>< ><
<1: <1: 0.5
::1! ::1!
SPECIMEN 21-26
Fig. 13-Crack pattern of push-off specimens with
ri cj longitudinal tension stress (Reference 14)
w ...J
Q. <1:
>< ()
w
0 200 400
that of the previous cases. One possible reason for the
LONGITUDINAL TENSION STRESS, "£. (psi)
generally higher test values is the existence of the addi-
tional steel bars to apply the longitudinal stresses and
Fig. 12-Effect of longitudinal tension stress on maxi- the additional ties for the transverse steel bars. This
mum shear strength (1 psi = 6.895 kPa) additional steel does not pass through the shear plane
and is therefore not taken into account in the calcula-
tion. However, it may be close enough to the shear
plane to be partially effective in the truss model action.
The effect of the longitudinal tensile stress on the
shear strength is plotted in Fig. 12, showing the ratio of
the calculated to experimental maximum shear stresses
for these six specimens as a function of the imposed
longitudinal tension stress. The effect of the longitudi-
nal tension stress is correctly predicted.
Table 4 shows that the test value of shear strain at
T I
peak stress is reasonably close to the predicted value for
I
I I the case with no applied tension. However, in the two
cases with imposed tension the calculated values are
three to four times greater than the test values. A close
look at the reported cracking pattern, 14 shown in Fig.
Side View 13, reveals the cause. For the specimens with imposed
longitudinal tension, E4U and E6U, the critical crack
zone is much narrower than that for specimen E1 U,
where no tension is imposed. Since the gage length
across the shear plane in these tests is kept constant and
is much larger than the width of the critical crack zone
in Specimens E4U and E6U, the resulting shear strains
based on this large gage length would show much re-
duced values. Should the observed narrower width of
crack zone be used as the gage length, the shear strains
of Specimens E4U and E6U would be closer to the val-
ues predicted by the theory.
14-r-- 1 3., - - - - . . ..
!
1
Pull-off tests
T
4.75"
(.
I
I
I
'-
9.5"
shear stress at the shear plane and a tensile stress in the
transverse direction (Fig. 14). 29 In the theoretical calcu-
lation, the tensile stress is modelled by a negative stress
I
~3"-+
I
- il-----
l nized that the tensile stress in the transverse direction in
the critical zone may be larger than assumed, since the
tension is transmitted almost directly through the an-
choring bars within the critical zone. The local distri-
bution of tensile stress is difficult to estimate and the
Fig. 14-Pull-off test specimen (Reference 29) (1 in. calculated shear is expected to be somewhat greater
25.4 mm) than the actual shear strength because the tensile stress
158 ACI Structural Journal I March-April 1987
would be underestimated using an average value. This 1500 /
/
/
/ /
.
~
values. )( / /~o,..
,/ ,.
/
...E / / '
/ /
vi 1000
DISCUSSION /
.
til "/
w
The proposed theory for shear transfer across anini- a:
1- /
/
/
.-
tially uncracked plane in reinforced concrete is shown til
/
/
/
a:
to be sound. Excellent agreement between predicted <
w /
/
/
/