READER COMMENTS
Horizontal Shear Strength of Composite
Concrete Beams With a Rough Interface*
by Robert E. Loov and Anil K. Patnaik
Comments by Alan H. Mattock and Authors
ALAN H. MATTOCKt
The authors are to be congratulated on a carefully exe-
cuted experimental investigation that has yielded some
valuable data in a field where data is short. On the basis of
their test results and consideration of prior results, they have
made proposals for changes in the “Shear Friction” provi-
sions (Section 11.7) of the ACI Code," and in the provisions,
for horizontal shear design (Section 17.5).
The proposed Eq, (11-26a), based on Eq, (12) of the
paper, includes the concrete compressive strength f’. How-
ever, nowhere in the paper is it demonstrated that the shear
friction resistance is proportional to /f’, as implied by
Eq, (12). In Fig. 19, the measured shear strength, v,, is plot-
ted against the reinforcement parameter, p,f, for four se-
ries of initially cracked, monolithic push-off specimens,
made of normal weight concrete; the difference between,
the series being the concrete compressive strength, f. The
compressive strengths indicated in Fig. 19 are the average
compressive strengths for each series of tests. (These test
results are taken from Refs. 8, 16, 20 and 36.) Also shown,
in this figure are lines representing Eq. (12) for the four
strengths of concrete.
It can be seen that the lowest line, which corresponds to
ff = 3875 psi (26.7 MPa), is a reasonable lower bound to the
data points for the test series of this strength. However, as,
the concrete strength increases, the calculated values ob-
tained using Eq. (12) do not increase as rapidly as do the ex-
perimental values, indicating that the shear friction resis-
{ance is not proportional to fas assumed in Eq. (12). An
“= PCLJOURNAL, V. 39, No Jamar Febrary 1996, 5p. 8-68.
"ments Rrtesr of Cl Eagan, Univer f Waa, Sate Washing,
108
ples) = O.S((15 +A.t))¢CF pst
$0.25",
Ply ad
Fig. 19. Comparison of Eq. (12) with test data from intially
cracked, monolithic, normal weight concrete push-off
specimens.
additional shortcoming of Eq. (12), as applied to initially
‘racked concrete, is that it predicts there will be significant
shear resistance when there is no reinforcement crossing the
crack; this is not true.
Assuming that the shear friction resistance, vq. is propor-
tional to (9,4) which appears to be reasonable, the data
from the four series of shear transfer tests of initially
cracked, monolithic push-off specimens shown in Fig. 19
was analyzed again, It was determined that the shear fric-
tion resistance is proportional to (f)"". In Fig. 20, (yf) is
PCI JOURNALow fs ~ 3875
x = 5965,
ea f 0 — 6516
p46 279/12.00
sos ph P6079/14.25
bP
vite si)
aN
Waa = ot PMU E428
CE ds * oN PME
Nin = igs = 0.02", sal
et Pte /te
Fig. 20. Comparison of Eq, (14) and Eq, (1) with test data
{rom initially cracked, monolithic, normal weight concrete
push-off specimens.
plotted against (p,f)'"(2)"/f". The mean line through the
data is given by the equation:
(pf) *E°N2.00 € 0.36! (psi)
fe aa)
(Df) MEY 13.820 $ 0.36 (MPa)
This equation yields a test/calculated strength ratio of,
1.00, with a standard deviation of 0.096. It can be seen that
the distribution of the data points about this line is random
with respect to the concrete strength.
Following the philosophy for shear design, where a lower
bound expression is preferred to a mean expression, the sec~
ond equation, shown in Fig. 20, would be suitable for de-
sign in the case of initially cracked, monolithic, normal
weight concrete:
WAP MUEY NAS £0.36" (psi)
My
(as)
(Pf)? ED"VA.536 § 0.36" (MPa)
Assuming @ normal distribution of test data about the
mean, 95 percent of test results should lie above this Tine,
which is 1.645 times the standard deviation below the mean,
line. The distribution of data points in Fig. 20 appears to
confirm this
Eqs. (14) and (15) above relate to shear transfer across a
crack in monolithic normal weight concrete, Tests reported
in Ref. 36 showed that the shear transfer strength across an
initially cracked, rough interface between coneretes cast at
different times was approximately 0.02’A, less than that of
initially cracked monolithic specimens that were otherwise
identical. A reasonable design expression for shear transfer
across an initially cracked, roughened interface between
concretes cast at different times would therefore be:
(uh EY"N4.25 — 002K’ $0.34" (psi)
(oA YEY"13 820 —0.02f! < 0.3¢! (MPa)
‘Soptomber-October 1994
Fig. 21. Comparison of Eq. (16) with test data from initially
cracked, composite, normal weight concrete push-off
specimens.
In Fig. 21, this equation is compared to strength data re-
ported in Ref. 36 for two series of initially cracked, normal
weight concrete, composite specimens that had roughened
interfaces as required by Section 11.7.9 of ACI 318-92.’ In
Series B, the concrete compressive strength of both con-
cretes was approximately 6000 psi (41.4 MPa) at the time of
test. In Series D at the time of test, the initially cast concrete
had a compressive strength of approximately 6000 psi
(41.4 MPa), while the second cast concrete had an average
strength of 3000 psi (20.7 MPa). The best correlation was,
obtained when the average compressive strength of the two
concretes was used for f’ in Eq. (16). The line correspond-
ing to Eq, (16) is seen to be a reasonable lower bound to the
test data for these two series.
‘The authors’ comments on the importance of ensuring ad-
equate roughness of the surface of the precast unit in com-
posite construction cannot be over-emphasized. If the inter-
face is not adequately roughened, the shear strength of the
interface can be drastically reduced. Because of this, itis
believed that in order to ensure that adequate roughness is
provided in practice, some numerical measure of that rough-
ness should continue to be specified in the ACI Code,’ even
if itis @ Title less than the '/ in, (6.35 mm) amplitude eui-
rently required.
The authors’ comments regarding the misuse of the equi-
librium method when designing the interface reinforcement
of a composite beam subject to a uniformly distributed load
are correct. It was in an effort to overcome this problem
that the new Section 17.5.3.1 was introduced in ACI 318-
92." This requires that when the equilibrium method is
used, the tie reinforcement shall be distributed over the half
span so that “the tie area to tie spacing ratio along the
member shall approximately reflect the distribution of
shear forces in the member.”
The use of the expression (V,Q.)/(Jab,) to calculate the
horizontal shear stress, v,4, at the interface in a composite
beam was dropped from the ACI Building Code in the inter-
est of simplification of the design process. Also, its applica-
107bility to composite beams in which the precast beam is pre-
stressed is questionable, particularly when the tendons are
draped so that the actual effective depth of 4, atthe support
may be quite small. This was an additional reason why it
was decided to express v,, as a simple nominal stress,
Vaylbd, where dis taken as the greater of the actual effec-
tive depth and 0.8 times the total depth of the section (Sec-
tion 17.5.2.5 of ACI 318-92)
AUTHORS’ CLOSURE by
ROBERT E. LOOV* and ANIL K. PATNAIKt
‘The comments made by Dr. Mattock are greatly appreci-
ated, The authors wish to respond to several of them.
The effect of variations in { is a most interesting issue.
As indicated in Eq. (5), Walraven et al." found that the
shear strength is related to (f')'** with additional effects of
(f'P applied to the exponent of the clamping stress, p,f,.
‘These factors were determined based on data from 88 push-
off tests of precracked monolithically cast specimens with
concrete cylinder strengths ranging from 16.4 to 68.3 MPa
(2385 to 9900 psi). In a discussion of Walraven’s paper,
Mattock" proposed that the influence of concrete strength
be considered additive to that of the tie steel. The suggested
exponent was (£/)'* {see Eq, (6).
Based on these previous results from push-off test an in-
termediate rounded value of ((/)"* was tentatively chosen by
the authors for the proposed design equation.
In the series of tests reported in this paper, four beams
were made with different flange strengths in order to ascer-
tain whether the variation in the horizontal shear strength of
composite beams was affected by concrete strength in a
manner similar to that found for push-off tests. As shown in
Table 3, the flanges of Beams 13 and 14 were made with
lower strengths while Beams 15 and 16 were made with
higher strengths than that nced fo the other 1? heams
‘As was discussed in the paper, Beam 14 highlights the
need for a rough interface, but the test result is not compara-
ble to the remaining test results. Fig, 14 shows that the use,
of (f°) brings the results from the three remaining tests di-
rectly into line with the results from the other 12 beams.
(Note, however, that Beams 15 and 16 failed in flexure and
hhad not reached their limiting interface shear strength.)
In his comments, Dr. Mattock has focused his attention
con precracked push-off tests. He has combined six test re-
sults from Ref. 20 (NI to N6) and eight from Ref. 36 (A1 to
AT and AGA) along with 26 of the tests analyzed by Wal-
raven et al. Based on the 40 test results chosen, he found,
+ Profenar of Gv aginering, University of Calay, Calgary Alerts, Canad,
7 Scr Engines, Whol Gland Paes, Pe, Wesem Awa, Ausab
108
Additional Reference
36. Mattock, A. H., “Shear Transfer Under Monotonic
Loading, Across an Interface Between Concretes Cast
at Different Times,” Report SM 76-3, Department of
Civil Engineering, University of Washington, Seattle,
WA, 1976, 67 pp. (Part I of Final Report for National
Science Foundation Grant No. ENG74-2113.)
ou
03
‘02
R + nterett
‘toe
a See 8 ae
on,
ug an a2 a Osos
as
ye
Fig, 22, Test data for 102 monolithically cast, intially
cracked, normal weight concrete push-off specimens.
Fig. 23. Mean and fith percentile equations based on 82
specimens with (p, f/f’ < 0.36.
PCI JOURNAL