Performance of High-Strength Concrete Corbels: Aci Structural Journal Technical Paper
Performance of High-Strength Concrete Corbels: Aci Structural Journal Technical Paper
Performance of High-Strength Concrete Corbels: Aci Structural Journal Technical Paper
In this study, 30 high-strength concrete corbels were tested to friction model used in ACI 318-89 has been calibrated for
destruction. Variables considered in the investigation are shear only relatively low-strength concretes. The experimental
span-to-depth ratio, concrete strength (45 to 105 MPa [6500 to results presented in this paper are compared with two design
15,200 psi), and the provision of secondary reinforcement. The models: 1) the ACI 318-1989 model and 2) the plastic truss
investigation examines corbel behavior in the context of the
model of Rogowsky and MacGregor.5 The plastic truss model
previous parameters and compares the experimental results with
the ACI 318-89 design method and the plastic truss model of was chosen from available truss models because its design is
Rogowsky and MacGregor. Particular attention is given to deter- simple, yet it utilizes rigorous plasticity theory.
mining the concrete efficiency factor for members failing in
compression, and the results are compared with the efficiency RESEARCH SIGNIFICANCE
model proposed by Warwick and Foster. The results of the investiga- While a considerable amount of research has been
tion show that good load predictions can be obtained using the performed on corbels in general, very little exists in conjunc-
plastic truss model when combined with the Warwick and Foster tion with high-strength concrete, and, in particular, on
efficiency factor. It is concluded that the design method given in ACI compression failures. This paper examines the behavior of
318-89 is not appropriate for corbels fabricated using high- high-strength concrete (HSC) corbels with emphasis on
strength concrete.
failure through crushing of the compression strut. Previous
research6-8 has shown that the main variables affecting the
Keywords: brackets; corbels; high-strength concretes; shear properties;
efficiency of the compression strut are concrete strength,
structural design; trusses.
shear span-to-effective depth ratio, and the provision of
secondary reinforcement. Each of these variables is studied
INTRODUCTION in the experimental program.
Reinforced concrete corbels and nibs are commonly used
to transfer loads from beams to columns and are similar in DESIGN MODELS
behavior to half-joints used to transfer shear forces in bridges. Plastic truss model
With the growing use of high-strength concrete (HSC), The plastic truss model shown in Fig. 1 consists of
particularly in high-rise structures and long-span bridges, compression struts in uniaxial compression with a uniform
there is a need for experimental data on the performance of distribution of stress across the strut of fc* and a tension tie.
corbels, nibs, and half-joints. This model is chosen due to its simplicity of form and
Yong et al.1,2 conducted a study of 14 reinforced and two rigorous structure. In this model, only two failure modes are
unreinforced HSC corbels. The study concentrated on the possible: failure by crushing of the concrete in the compres-
tension (beam-shear) mode of failure, with only one of the sion strut and failure by yielding of the tension tie. It is also
reinforced specimens reported to have failed prior to notable that in this model bearing failure is simply an exten-
yielding of the primary reinforcement. Where failure sion of the compression failure mode, where crushing occurs
occurred after yielding of the primary reinforcement, they in the concrete strut immediately below the bearing pad.
observed that the behavior of the HSC corbels was similar to Secondary failures, such as anchorage and end splitting
that of normal-strength concrete corbels. Provided that the failure, are accommodated by sensible detailing. Fig. 2
corbels remained under-reinforced, the extent of cracking compares the theoretical compression strut for the plastic
was not affected by concrete strength and the crack patterns truss model with the compression stress contours obtained
were not affected by the amount of primary reinforcement. A for a corbel analyzed using the nonlinear finite element
comparison of the test data with ACI 318-893 and the Hagberg program RECAP developed by Foster.6 It can be seen that
truss model4 was undertaken. Yong et al. concluded that the the plastic truss model compares well with the stress trajec-
Hagberg model provided better accuracy and considered the tories at ultimate given by the FE model.
ACI 318 limit of 5.5 MPa (800 psi) on the shear stress at the
interface to be overly conservative. In this paper, 30 HSC
corbels are tested, predominantly failing in compression. ACI Structural Journal, V. 93, No. 5, September-October 1996.
Received Feb. 27,1995, and reviewed under Institute publication policies. Copy-
Design methods currently available include empirical right © 1996, American Concrete Institute. All rights reserved, including the making of
design, design based on stress analysis, design based on the copies unless permission is obtained from the copyright proprietors. Pertinent discussion
will be published in the July-August 1997 ACI Structural Journal if received by
shear friction model, and strut-and-tie modeling. The shear March 1, 1997.
Rex E. Powell is a structural engineer with Morrison Whitten and Nicey. He is a grad-
uate of the Victoria University of Technology, Victoria, Australia, and received a Mas-
ter of Engineering Science degree from the University of New South Wales.
In the simplest form of the plastic truss model (shown in Eq. (4) was calibrated for shallow beams failing in shear
Fig. 1), the ultimate capacity of the section in tension is for a limited range of conventional-strength concretes. In the
same publication, however, Nielsen et al.7 used the relation-
w ship for the design of deep beams and corbels without veri-
V u = A st f sy ---- (1) fication. This relationship has been picked up in the
Ω
literature10 and has been adopted by AS 360011 for use
and in compression is (without restriction) in the design of nonflexural members.
Rogowsky and MacGregor5 suggest that the strength of
V u = f c∗ bw (2) the compression strut is reduced by an uneven stress distri-
bution acting within the strut. They proposed that an effi-
where Ast is the area of main tension reinforcement, fsy is the ciency factor of 0.85 be used for corbels with 0.15 ≤ a/d ≤ 1.
yield strength of the tension reinforcement, w is the effective Schlaich and Schäfer12 observed that the shape of the
width of the bearing plate (which may be less than the phys- concrete strut is bowed, and that this produces transverse
ical width), Ω is the effective anchorage depth (Ω = d – tensile forces acting across the compression strut. These
2 2
d – 2aw – w ), b is the width of the corbel, and fc* is the forces can lead to a premature diagonal splitting failure if
effective strength of the concrete compression strut given by sufficient secondary reinforcement is not provided. Foster6
added that the shape of the strut is not constant but varies
f c∗ = νf c ′ (3) from an exaggerated bowed shape at low loads to a shape
approximating that used in Rogowsky and MacGregor’s5
fc′ is the characteristic cylinder strength of the concrete in model near ultimate.
uniaxial compression, and ν is an efficiency factor. The Warwick and Foster8 investigated the influence of shear
corbel’s capacity is taken as the lower of Eq. (1) and (2). span-to-effective depth ratio (a/d), concrete strength, and the
Secondary reinforcement is required in corbels to guard presence of secondary reinforcement on concrete efficiency.
against diagonal splitting and interface shear. Mattock9 noted They found that the a/d ratio and concrete strength were the
that diagonal splitting failure was not observed, provided that most important parameters affecting concrete efficiency, and
an area of secondary reinforcement is provided that has a proposed the following relationship on the condition that a
force capacity of one-half that of the main reinforcement and minimum amount of secondary reinforcement is provided
fc ′ with 0.2 fc′ > 5.5 MPa (800 psi), where b is the width of the
a 2
ν = 1.25 – --------- – 0.72 --- + 0.18 ⎛ ---⎞ ≤ 0.85… for a ⁄ d < 2
a
⎝ d⎠ corbel and fc′ is the uniaxial compression strength of
500 d
concrete. It is not clear to the authors why the limit on
concrete strength was imposed on the design—presumably
fc ′ because of the relatively low strength of the corbels tested in
- … for a ⁄ d ≥ 2
ν = 0.53 – -------- (5)
500 1983. However, this limit requires review if the full advan-
tage of HSC is to be available to the designer. In the analysis
Other efficiency models have been proposed and detailed of the test data that follows, comparisons are made with and
by Rogowsky and MacGregor10 and Foster and Gilbert.13 without this limit.
of anchorage failure. In all cases, the horizontal reinforcement (for Groups SA, SC, SD, PB, PD-PF, and PG2) contained a
in the corbels passed inside the vertical column bars. Series SD mix of silica fume and powdered superplasticizer. Where
corbels had the same dimensions and reinforcement layout required, additional superplasticizer was added at the site to
as Series SC but with mechanical anchorages similar to achieve the desired workability. Characteristic concrete
Series P corbels. strengths were obtained from 300 x 150-mm-diameter
For Specimens SA2 and SB2, additional confinement cylinders for Groups SB, PA, PC, and PG1, and 200 x
reinforcement was provided in each corbel along the lines of 100-mm-diameter cylinders for the HSC specimens. Details
the predicted compression struts. The aim of this experiment of the mix proportions for Specimens SC are given in Selim.15,16
was to indicate if there was any strength advantage in more
tightly confining the compression strut. No significant gain Instrumentation
in strength was achieved and this reinforcement layout was Each corbel was instrumented to record strains in the
not used in any further tests. middle layer of the main tension reinforcement and the
One of the main aims of the experimental program was to deflection at the tip of the corbel. For all but Series PD and
look at the behavior of HSC corbels failing in compression. PE corbels, strains were measured using a demec strain gage
To determine the efficiency factor required by the plastic over a gage length of 250 mm. Three or four gage lengths
truss model, it is necessary to know the width of the were used for each specimen, depending on the shear span-
compression strut. This was obtained by using stiff bearing to-effective depth ratio (a/d). Steel pins 40-mm long and 6 mm
plates and insuring that any bearing reinforcement used did in diameter were welded onto the tension tie reinforcement.
not lie outside the line of the bearing plate. Thus, provided that Demec targets were then glued to the pins and strains were
failure is in a compression mode (discussed below), the effec- measured. A sample section of reinforcement was fabricated
tive bearing width w, given in Eq. (2), equals the plate width. with two welded lugs and tested in tension to evaluate the
effects of welding on the strength properties of the bar. No
Materials detrimental effects were recorded either on the yield point or
All concrete was supplied by a local ready-mix supplier ductility of the bar. Details of strain gage, locations, and
and had a maximum aggregate size of 10 mm. All HSC mixes target connection are given in Fig. 4. For Series PD and PE
Fig. 6—Load-versus-deflection results for: (a) Corbel SC1-4; (b) Corbel PG2.
Figure 7 illustrates typical crack patterns for the three specimen gave a high efficiency factor when analyzed using
observed modes of failure. Detailed results for individual the plastic truss model. Cracks observed from a compression
corbels are given in Selim et al.15 and Powell and Foster.17 failure were finer, more numerous, and more evenly distrib-
uted across the specimen than for specimens that failed by
ANALYSIS OF RESULTS diagonal splitting. As expected, the most ductile failures
All corbels without any secondary reinforcement failed by observed were tension failures.
diagonal splitting. This failure mode is brittle and occurs The load-versus-strain relationship for specimens that
with little warning. Specimens failing in this manner typically failed by either diagonal splitting or compression comprise
displayed a well-defined diagonal crack that formed between two phases. Phase I is the precracking state where the tension
the inside edge of the loading plate and the column-corbel forces are carried by the concrete section. In this stage,
junction. Corbels containing secondary reinforcement strains measured in the main tension reinforcement were
generally failed in the more ductile compression mode. The small and irregular. In Phase II, the cracking phase, strains
exception was Specimen PA2, which failed by diagonal increased in a linear fashion until failure. Specimens that
splitting; note, however, that later analysis shows that this failed in tension exhibited a third phase that occurs on yielding
of the primary tension reinforcement. Phase III is shown by Table 3 and Fig. 8 indicate that the plastic truss model is a
a plateau in the load-versus-strain graph [see Fig. 6(a)]. good tool for designing corbels of high-strength concrete
Phases II and III are consistent with the plastic truss theory. when combined with the efficiency factor relationship
The ultimate load predicted by the plastic truss model, proposed by Warwick and Foster. The mean predicted to
with efficiency factors given by Warwick and Foster,8 Nielsen experimental failure load for specimens containing
et al.,7 and ACI 318-893 (with and without the limitation on secondary reinforcement and failing in compression using
fc′) are given in Table 3. The results for specimens this model is 0.86, with a standard deviation of 0.10.
containing secondary reinforcement and failing in compres- The plastic truss model for corbels failing in tension gives
sion are compared in Fig. 8 for each of the design models.
good results for the specimens with no secondary reinforce-
A comparison between the ACI 318-89 provisions and the
ment, but generally underestimates the capacity of the
experimental results shows that when the limits on concrete
strength are included, the ACI method is grossly conserva- section when horizontal stirrups are added. The plastic truss
tive. When the limit on concrete strength is removed, the model used does not consider the effect of the secondary
results are grossly nonconservative. reinforcement in increasing the tension capacity of the spec-
Results predicted by plastic truss theory in combination imen, although a more complex model could. The difficulty
with the Nielsen et al.7 efficiency relationship underesti- in adopting a more advanced model is in guaranteeing that
mated the experimental results to an extent similar to the all the secondary reinforcement is at yield at failure. In the
ACI procedure. Nielsen’s effective strength relationship authors’ opinion, it is the role of the secondary reinforcement
(fc*) is a second order polynomial in fc′ and has a maximum to guard against interface shear and diagonal splitting fail-
value of 32 MPa when fc′ = 80 MPa. Actual values of effec- ures and to improve the performance of the compression
tive concrete strength of up to 79 MPa (Specimen PB1) were strut by reducing transverse strains. Strains in the secondary
recorded in this study. reinforcement should not be assumed to be at yield at failure.
For corbels failing in compression, the efficiency factor 3. Providing secondary reinforcement reduces crack
is a measure of the performance of the compression strut. widths, improves ductility, and for beams failing in compres-
The detrimental effect of transverse strains on the ultimate sion may change the failure mode from diagonal splitting to
capacity of concrete in compression is well documented.18 compression strut crushing. A minimum quantity of hori-
The effect of providing horizontal reinforcement is to zontal stirrups similar to that for normal-strength concrete
increase the effectiveness of the compressive strut. The should be used in corbels fabricated with HSC.
provision of secondary reinforcement generally improved 4. The ACI 318-89 design method is not recommended for
the compressive strength of the corbels. For specimens with use with corbels designed with high and very high-strength
low a/d (such as Series SA), the transverse strains are small concretes.
and the variation in measured ultimate strengths is within the 5. The plastic truss model provides a good tool for
normal range of experimental scatter for the testing of designing HSC corbels and is best used in conjunction with
concrete structures. the efficiency factor proposed by Warwick and Foster.
CONCLUSIONS ACKNOWLEDGMENTS
In this investigation, 30 corbels were prepared and tested This research was funded through Faculty Research Grants from the
under vertical loading. The main test variables were concrete Faculty of Engineering at the University of New South Wales. The authors
strength (45 to 105 MPa [6500-15,200 psi]), shear span-to- are grateful for this assistance. The contributions of Warwick Faulkner and
Derek Graham to the experimental program are gratefully acknowledged.
depth ratio, and the provision of secondary reinforcement.
The following conclusions can be drawn based on the test results:
1. The first cracks are flexural cracks propagating from the REFERENCES
1. Yong, Y. K.; Douglas, H.; McCloskey, H.; and Nawy, E. G., “Rein-
corbel-column intersection, and the flexural cracking load forced Corbels of High-Strength Concrete,” High-Strength Concrete, SP
decreases with an increase in the shear span-to-depth ratio. 87, American Concrete Institute, Detroit, 1985, pp. 197-212.
2. Corbels fabricated from HSC behaved similarly to those 2. Yong, Y. K., and Balagura, P., “Behavior of Reinforced High-Strength
made of normal-strength concrete. Concrete Corbels,” Journal of Structural Engineering, ASCE, V. 120, No.