Professional Documents
Culture Documents
High-Strength Concrete Deep Beams With Effective Span and Shear Span Variations
High-Strength Concrete Deep Beams With Effective Span and Shear Span Variations
Nineteen reinforced concrete deep beams with compressive strengths in the of high-strength concrete. Research from References 8
range of 41 MPa ≤ fc′ ≤ 59 MPa (6000 psi ≤ fc′ ≤ 8600 psi) were tested through 10 indicates that ACI Eq. (11-6) overestimates the
under two-point top loading. All the beams were singly reinforced with influence of concrete contribution for high- strength
main steel percentage ρ = 1.23 percent and with nominal percentage of
shear reinforcement ρv = 0.48 percent. The beams were tested for seven
concrete beams with a/d > 4. However, there has been rela-
shear span-depth ratios a/d, ranging from 0.27 to 2.70, and four effective tively little information on the applicability of ACI Eq. (11-
span-depth ratios le /d, ranging from 2.15 to 5.38. Test results indicate that 30) and Eq. (11-31) for higher strength concrete deep beams.
le /d has little influence on the magnitude of the failure load. But for beams Deep beams in this context refer to beams with a/d ≤ 2.7.
with a/d ≥ 1.00, the flexural failure mode becomes dominant with increas- This test program seeks to assess the suitability of ACI
ing le /d. Eq.(11-30) and Eq.(11-31) for deep beams with strengths
The test results are compared with predictions based on the current ACI exceeding 40 MPa (6000 psi). For this purpose, all the test
Building Code. The comparisons reported in this paper will provide an beams had a/d ≤ 2.7 and compressive strength fc′ exceeding
added assurance to designers that the deep-beam provisions in the ACI
code, though essentially based on concrete strengths of less than 41 MPa
40 MPa, with the introduction of two main variables: shear
(6000 psi), will insure safe designs for higher strength deep beams. How- span-depth ratio a/d and effective span-depth ratio le /d. The
ever, the ACI code tends to be rather conservative, as shown by comparison effective span refers to the distance between the two centers
to the Deep-Beam Design Guide issued by the Construction Industry of supports (Fig. 1). The importance of the first parameter a/d
Research and Information Association, London. Nevertheless, the ACI code on the ultimate strength of deep beams has been confirmed by
has the important advantage of being easy to use.
many researchers,8,10,14 but that of le /d has received less atten-
Keywords: building codes; deep beams; deflection; diagonal tension; high-
tion. Earlier work by Manuel, Slight, and Suter15 has indicated
strength concrete; shear strength; spans; strength; tension; web reinforcement. that, although failure load is not significantly affected by le /d,
diagonal and flexural crack widths are influenced.
Reinforced concrete deep beams have useful applications Test results in the present program were compared with
in tall buildings and foundations. With the robust growth of predictions based on the ACI code,1 the CIRIA Guide,16 and
construction work in many Southeast Asian countries, deep Zsutty's proposed equations,17,18 as well as Hsu-Mau's
beam design is now a subject of considerable relevance in explicit equations.19,20
the region. Over there, ACI 318-891 is a major design docu-
ment, particularly as the current British Code BS 81102 does RESEARCH SIGNIFICANCE
not cover deep beam design, and the new Eurocode EC23,4 Concretes of strengths exceeding 40 MPa are increasingly
makes only very brief references to the subject. The investi- being used in the construction industry in many Southeast
gation reported here is one of several deep beam projects Asian countries. The literature provides further empirical
recently initiated at Nanyang Technological University in evidence on the behavior of concrete deep beams with
Singapore.5,6 compressive strengths in the range of 41 MPa ≤ fc′ ≤ 58 MPa
In the ACI code, the shear strength Vc for deep beams is (6000 psi ≤ fc′ ≤ 8600 psi). Tests show that a/d has a
given by ACI Eq. (11-30). This equation is related to the corre- significant influence on the ultimate strength. On the other
sponding equation for ordinary beams, namely ACI Eq. (11-6), hand, le /d has a qualitative influence on the failure mode of
by the multiplying term (3.5 – 2.5(Mu /Vud)). However, ACI’s deep beams. For the same a/d, as le/d increases, the flexural
Eq. (11-6), and hence the code's Eq. (11-30) for deep beams,
are essentially based on the many test results7 on beams with ACI Structural Journal, V. 92, No. 4, July-August 1995.
compressive strengths in the range of 14 to 40 MPa (2000 to Received June 29, 1994, and reviewed under Institute publication policies. Copy-
right © 1995, American Concrete Institute. All rights reserved, including the making
6000 psi). Recent test programs8-13 have been conducted to of copies unless permission is obtained from the copyright proprietors. Pertinent
discussion will be published in the May-June 1996 ACI Structural Journal if received
study the applicability of the two equations for beams made by January 1, 1996.
ACI member Susanto Teng is a lecturer in the Division of Structures and Construc- Materials
tion, NTU. He received his PhD from the University of Iowa in 1991 prior to joining
NTU. His research interests include structural concrete deep beams, time-dependent
In the concrete mix design, ordinary portland cement was
effects and cracking in structural concrete members, and behavior of concrete struc- used in conjunction with 10-mm chippings to achieve an
tures under dynamic loads. aggregate-cement ratio of 3.52. The water-cement ratio was
Lingwei Guan is a lecturer in the Division of Structures and Construction, NTU. He
kept at 0.31 with the addition of admixtures. The beam
obtained his BE and ME from Northern Jiaotong University, Beijing, China, and his specimens and the 100-mm (4-in.) British standard cubes
MS and PhD from Rutgers, the State University of New Jersey. His research is on rein- were cured in the same condition for more than 28 days. In
forced concrete deep beams and the performance assessment of nonload- bearing
building facades.
Table 1, the cube compressive strengths fcu were obtained on
the same day as the beam test. The cylinder compressive
strengths fc′ were calculated from the fcu values using the
failure mode becomes predominant. This investigation has formula proposed by Neville21
given added assurance that ACI Eq. (11-30) and Eq. (11-31)
are safe and convenient design tools, while rather fc′ = [0.76 + 0.2log10(fcu /19.59)]fcu (1)
conservative. The CIRIA Guide of the U.K. consistently
yields very good agreement with the experimental results,
though it is not as easy to use as the ACI code. Two sizes of hot-rolled steel bars were used for reinforcement:
20-mm (0.8-in.)-diameter deformed bars for the main longi-
EXPERIMENTAL PROGRAM tudinal reinforcement and 10-mm (0.4-in.) plain bars for the
Specimen details stirrups. Their respective yield strengths were 504.8 and
The test specimens (Fig. 1) consisted of 19 simply 375.2 MPa (73 and 54 ksi).
supported beams 500 mm (19.7 in.) deep and 110 mm (4.3 in.)
wide. The effective span le varied between 1000 and 2500 mm Test procedure
(39.4 and 98.4 in.). The beams had a tension steel area As = Vertical deflections were monitored by the LVDTs. At
628 mm2, providing ρ = 1.23 percent. All the beams had 10- each load increment, the test data were captured by a data
TEST RESULTS AND DISCUSSIONS the beams are marked boldly. For all the beams, fine vertical
Table 2 lists the measured diagonal cracking strength Vcr , flexural cracks formed first, usually near to midspan. The
serviceability strength Vser, and ultimate strength VnTEST. The widths of the initiating flexural cracks were very small, in the
second term Vser was defined by ACI 318-89, Commentary order of 0.04 mm (0.0015 in.). On further loading, diagonal
Clause R10.6.4, as the limiting crack width of 0.013 in. In cracks of widths generally greater than 0.10 mm (0.004 in.)
Table 3, comparisons of test results with the ACI code and occurred independently near the middle of shear span,
design equations by CIRIA, Zsutty, and Hsu-Mau are presented. crossing the mid-depth of the beam instantly. The flexural
cracks that had formed earlier then stabilized and stopped
General behavior propagating. New inclined cracks were formed in the shear
Fig. 2 shows the structural response of specimens with le /d span in the load-support direction until failure took place.
= 5.38. The trend is also typical of other le /d series. The For Beams A-0.27-5.38 and B-0.54-5.38, failure was in the
midspan deflection increases with higher a/d, as expected. shear mode, which is characteristically brittle and sudden,
For the beams with a/d ≥ 1.62, the increase in central deflec- accompanied by a loud noise. Concrete crushing of the
tion is not as significant as for those with lower a/d. In compression zone near the load point was observed for
general, all beams with the same a/d had a similar order of Beam A-0.27-5.38, while Beam B-0.54-5.38 failed in the
deflection, though, with increasing le /d, the deflection tends diagonal tension mode. In both instances, horizontal cracks
to increase. The midspan deflection at failure was less than initiating from two vertical end faces of the beams were
le /200, showing that, as expected, deflection is not a problem found on failure. With increasing a/d, the failure modes were
in deep beams. mixed, with flexure becoming more dominant. Beams D-
Fig. 2 also indicates the total load (2Vcr) associated with 1.08-5.38 and E-1.62-5.38 had almost fully developed diag-
the first occurrence of the diagonal crack, which started at onal cracks in the shear span, but failure was delayed until
about d/3 above the bottom face of the beam, and propagated vertical flexural cracks in the constant moment region
simultaneously toward the loading and support points. The branched upward and outward, causing a ductile collapse.
beams exhibited considerable strength reserve after diagonal Beam G-2.70-5.38 had flexural cracks that became inclined
cracking. In Table 2, the diagonal cracking strength Vcr is and joined the later-formed diagonal cracks, a mode charac-
between 20 and 35 percent of the ultimate shear strength teristic of ductile flexural-shear failure.
VnTEST. Likewise, the serviceability load Vser is in the range Table 2 shows the failure modes of the beams. Those
of 30 percent ≤ Vser/VnTEST ≤ 50 percent. beams that had significant diagonal cracking but eventually
Fig. 3 shows the cracks at failure of the beams with le /d = failed in flexure are labeled as “sh-flex” (for shear-flexure).
5.38, together with the loads at which each crack was first Those with significant flexural cracking but that eventually
observed, and the extent of the crack at each load. Those failed in shear are labeled as “flex-sh” (for flexure-shear). In
cracks believed to be the cause of the immediate failure of both Series D and E beams, the failure mode changed from
(a) (b)
Fig. 4(a)—Ultimate stresses vnTEST versus effective span-depth ratio le /d; and (b) diagonal cracking stresses vcr versus effective
span depth ratio le /d.
account of the shear-strength reserve of deep beams after the combined nominal shear strength Vn (= Vc + Vs) shall not
diagonal cracking has occurred; this factor shall not exceed exceed 8 f c′ bd.
2.5. The ACI code restricts Vc from Eq. (2) in exceeding In the CIRIA Guide, two sets of design recommendations
3.5 f c′ bd, and Vc from Eq. (3) in exceeding 6 f c′ bd. Also, are given for the design of deep beams, viz., the “Simple
motion about the loading point. Thus, the lower a reinforce- Vs = λ2 ∑ --------------------------------------- bh a
bh a
2
(6)
ment bar intersects the diagonal crack, the more effective it i= 1
A v f yv d It is clear that the ACI code [Eq. (2) through (4)] has the
V s = ---------------
- (9)
sV lowest standard deviation of 0.181 among the four methods
(Column 2 of Table 3), and the predictions are conservative.
Hsu-Mau’s explicit method19,20 The results demonstrate that the ACI code predictions are
safe for deep beams with 41 MPa ≤ fc′ ≤ 58 MPa (6000 psi ≤
fc′ ≤ 8600 psi), even though Eq. (2) through (4) were obtained
2 2
V n = 0.5f c ′ K ( ω 1 + 0.03 ) + K ( ω 1 + 0.03 ) + 4 ( w 1 + 0.03 ) ( ω t + 0.03 ) from tests on beams with lower concrete strength.7 The study
× bdv ≤ 0.3bdv (10) also shows that the predictions are generally very conserva-
tive, with a mean of 0.548; the conservatism reduces with
where increasing a/d and with increasing le/d.
⎛ ρf ⎞ ⎛ ρv ft ⎞
Fig. 7 shows the reserve capacity (1 – Vn/VnTEST) as well
ω 1 = ⎜ -------l ≤ 0.26⎟ , ω t = ⎜ --------- ≤ 0.12⎟ and as the relative Vc /VnTEST and Vs /VnTEST components in
⎝ fc ′ ⎠ ⎝ fc ′ ⎠
Vn/VnTEST using the ACI equations. The reserve capacity
2d varies widely, from a maximum of 70 percent at the low end
--------v of a/d = 0.27 to practically zero at the high end of a/d = 2.70
h 0 < a ⁄ h ≤ 0.5
K= d 0.5 < a ⁄ h ≤ 2 (Beam G-2.70-5.38). Within each of the beam series A
----v- --- ⎛ --- – ---⎞
4 h 1
h 3 ⎝ a 2⎠ a⁄h>2 through G (i.e., for each constant value of a/d), the relative
0 shear reinforcement contribution Vs /VnTEST increases with
le/d. This increase of Vs /VnTEST can in fact be predicted from
Comparison of shear design equations Eq. (4); since the test beams had no horizontal web reinforce-
Table 3 and Fig. 6(a) through (d) compare the four design ment (i.e., Avh = 0), then Vs must increase with the clear span-
methods for the test beams. The material safety factors for depth ratio ln/d. Referring again to Fig. 7, it can be seen that
concrete and steel reinforcement have been set to unity for the increase in Vs /VnTEST is more marked for those beam
comparison purpose. The ratio Vn/VnTEST, where Vn is the series with higher a/d. In other words, the strength reserve (1
calculated strength and VnTEST the measured strength, is an – Vn /VnTEST) decreases as a/d increases. It is worth noting
indication of the agreement between the two. that the ACI code recommends that Eq. (4) be used only for
beams with ln/d ≤ 5, thus restricting the value of Vs. In Fig. 7, fact that the CIRIA equations are intended for a/d ≤ 0.90.
such a high Vs value occurs in Beam G-2.70-5.38; reference to However, more results in the higher range of a/d are needed
Table 3 shows that Vn/VnTEST, i.e., (Vs + Vc)/VnTEST, is, in before any firm recommendations can be given.
fact, 1.038, so that the “strength reserve” is –3.8% in this case. Zsutty’s predictions [Eq. (7) through (9)] give a mean of
A comparison of Vc /Vcr for the ACI code is included in Vn/VnTEST close to 0.787, but the standard deviation is the
Column 6 of Table 3. Basically, Vcr is the diagonal cracking highest among the four methods considered (Table 3,
strength of the test beams with nominal web reinforcement. Column 4). For beams with a/d ≥ 1.62, the predictions are
However, it has been shown9,10 that for beams with nominal unconservative, although Eq. (8) is intended to be valid up to
web reinforcement, the Vcr values obtained are similar to the a/d < 2.5.18 This suggests that Zsutty’s equations may not be
diagonal cracking strengths of singly reinforced beams with suitable for beams with fc′ > 40 MPa (6000 psi) and a/d ≥ 1.6.
no web reinforcement. For an ordinary beam without web Hsu and Mau’s19,20 explicit equations [Eq. (10)] with a
reinforcement, the failure load corresponds to Vcr , as the beam standard deviation of 0.254 tend to overestimate the shear
cannot mobilize any shear reserve after diagonal cracking. Vc strength, even for beams at a/d = 0.27 (Table 3, Column 5).
in Eq.(2) was developed by ACI on this basis. But for a deep With increasing a/d, the overestimation becomes more
beam, there is a significant shear strength reserve after diagonal pronounced. However, Hsu and Mau suggest that the
cracking has occurred. Using the ACI code for Vc /Vcr yields an implicit iterative version would yield better accuracy, but
average value of 1.675, which is close to the ratio of the this is not considered here.
imposed limitations of 6 f c′ bd [Eq. (3) for deep beams] to
3.5 f c′ bd [Eq. (2) for ordinary beams]. This experimental CONCLUSIONS
comparison gives some added assurance of the validity of the Based on the test results and their comparison with the
imposed restrictions on Vc in the ACI code. design equations, the following conclusions can be made:
The CIRIA Guide [Eq. (5) and (6)] gives an average mean 1. ACI’s code Eq. (11-30) and (11-31) can be used for
of 0.757 for Vn/VnTEST and achieves the second lowest stan- designing deep beams with concrete compressive strengths
dard deviation of 0.191 (Table 3, Column 3). Unlike the ACI in the range of 41 MPa ≤ fc′ ≤ 58 MPa (6000 psi ≤ fc′ ≤ 8600
code, the CIRIA Guide does not use the parameter le /d; psi). The equations can be very conservative at the low end
hence, using the CIRIA equations, Beams A-0.27-2.15 and of a/d and this conservatism reduces with both increasing a/d
A-0.27-5.38 would have exactly the same shear strength if and le /d. Overall, ACI's code strength predictions gave the
the concrete strengths had been the same. This also helps to lowest standard deviation among the four methods. This
explain why the Vn/VnTEST ratios did not fluctuate widely means that, used with suitable safety factors, the ACI code
with the le /d ratio when Vn was calculated from the CIRIA will consistently give very good results.
Guide. Note also that CIRIA’s predictions are conservative; 2. CIRIA's predictions are also conservative, but with a
the only exception is Beam E-1.62-5.38,. This is despite the higher mean of predicted to tested shear strength of 0.757.