You are on page 1of 9

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 107-S72

Behavior of Large Deep Beams Subjected to Monotonic


and Reversed Cyclic Shear
by Boyan I. Mihaylov, Evan C. Bentz, and Michael P. Collins

An experimental study on eight large reinforced concrete deep members without transverse reinforcement can be designed
beams was performed. The variables were the shear span-depth by strut-and-tie models only if a/d is less than approximately 2,
ratio (a/d = 1.55 or 2.29), quantity of stirrups (0.0 or 0.1%), and whereas longer members must be designed using shear equations
type of loading (monotonic or fully reversed cyclic). All members
failed in shear before yielding of the longitudinal reinforcement. for slender beams. This restriction on the applicability of
The provision of only 0.1% of transverse reinforcement significantly strut-and-tie models introduces a discontinuity in the shear
decreased crack widths at service load levels and significantly strength predictions at an a/d of approximately 2. The
increased the shear strength. For members with stirrups, the load- experimental trend shows no such discontinuity. The code-
deformation response measured under monotonic loading prescribed minimum amount of transverse reinforcement for
provided an excellent envelope to the cyclic response and the shear beams also has a discontinuity with the values for deep
strength under reversed cyclic loading was not significantly
reduced. For the members without stirrups—somewhat members being approximately 2.5 times larger than those for
surprisingly—those tested under reversed cyclic loading failed at slender beams. To improve the understanding of deep-beam
significantly higher shear forces than their companion, behavior in the region of these discontinuities, an
monotonically loaded specimens.

Keywords: cyclic loading; deep beams; shear; stirrups; strut action.

INTRODUCTION
Deep beams can be classified as members in which a
significant portion of the shear force is carried by strut
action, where compressive stresses flow directly from the
load to the support. With a few notable exceptions,1-3
research on deep beams has concentrated on simply
supported beams subjected to monotonically increasing
point loads. For point loads, the importance of strut action
can be related to the shear-span-to-depth ratio (a/d). Figure 1
shows the observed influence of a/d on the shear strength of
a series of beams without stirrups tested by Kani et al.4 Also
shown in this figure are the predicted strengths for these beams
given by ACI 318-08.5 It can be seen that for both experiments
and predictions, the shear strength of the longer beams was
not strongly influenced by a/d, whereas that of shorter beams
increased greatly as a/d decreased. Members with large a/d
are dominated by beam action in which the tension in the
longitudinal reinforcement changes along the length of the
beam, requiring shear stresses in the cracked web.6 On the
other hand, for members with low a/d, the beam behaves like
a tied arch with the longitudinal reinforcement having almost
constant force from support to support and the shear being
carried by inclined struts. As evident from Fig. 1, deep
beams can be several times stronger than longer, slender
beams. For the beams in the plot, the transition from strut
action to beam action—and equivalently from deep beams to
slender beams—occurred at an a/d somewhat greater than 2. Fig. 1—Arch action versus beam action. (Note: 1 mm =
In predicting the shear strength of members such as those 0.0394 in.; 1MPa =145 psi = 0.145 ksi.)
shown in Fig. 1, a combination of strut-and-tie models—
which will govern for lower a/d—and shear equations for
slender beams should be used. Because both beam action and ACI Structural Journal, V. 107, No. 6, November-December 2010.
MS No. S-2009-403 received December 14, 2009, and reviewed under Institute
arch action must break down before final failure of the beam, publication policies. Copyright © 2010, American Concrete Institute. All rights reserved,
it is the larger of these two capacities that will be the including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including author’s closure, if any, will be published in the September-
predicted shear failure load. According to the ACI Code, October 2011 ACI Structural Journal if the discussion is received by May 1, 2011.

726 ACI Structural Journal/November-December 2010


EXPERIMENTAL PROGRAM
Boyan I. Mihaylov is a Postdoctoral Fellow at the University of Toronto, Toronto,
ON, Canada. He received his PhD at the ROSE School, Pavia, Italy, in 2009. Specimens
The experimental program involved eight three-point
Evan C. Bentz, FACI, is an Associate Professor of civil engineering at the University
of Toronto and is a member of ACI Committee 365, Service Life Prediction, and Joint bending tests of large deep beams intended to represent shear
ACI-ASCE Committee 445, Shear and Torsion. critical members from engineering practice. The experimental
variables were the a/d ratio, transverse reinforcement ratio
Michael P. Collins, FACI, is a University Professor and the Bahen-Tanenbaum
Professor of Civil Engineering at the University of Toronto. He is a member of ρv, and type of loading as summarized in Table 1. In the
Joint ACI-ASCE Committee 445, Shear and Torsion. specimen names, the first symbol S or L stands for short or
long, the second symbol 0 or 1 means no transverse rein-
forcement or 0.1% transverse reinforcement, and the third
experimental study involving large-scale shear tests of symbol M or C stands for monotonic or cyclic loading.
beams with or without small amounts of transverse
The chosen amount of transverse reinforcement
reinforcement was conducted and is described in this paper.
approximately corresponds to the ACI-specified minimum
Pairs of nominally identical specimens were subjected either amount for the longer members; and for consistency, the same
to quasi-static monotonic loading or quasi-static fully amount was used for the shorter members. All specimens were
reversed cyclic loading. 1200 mm (47.2 in.) deep and 400 mm (15.7 in.) wide with
The influence of cyclic loading on the shear capacity of similar longitudinal reinforcement and material properties (refer
reinforced concrete members has been extensively studied for to Fig. 2). The top and bottom reinforcement ratio ρl was 0.70%,
regions where flexural hinging is expected to occur.7,8 For these and the bars were anchored beyond the supports by friction-
regions, it is generally accepted that cyclic loading causes a welded circular anchor heads. Table 1 lists the compressive
noticeable reduction in shear capacity, particularly at higher
ductility levels. The situation is less studied for those regions
Table 1—Tests
where flexural hinging is not expected to occur. For buildings
and bridges proportioned using capacity design principles,8 most fc′, MPa fr , MPa Load
Test a/d ρv (psi) (psi) type
of the structure will be protected from flexural hinging but will
S0M Monotonic
still be subjected to reversed cyclic shear forces. In such regions, 0 34.2 4.32
S0C (4960) (627) Cyclic
will reversed cyclic loading reduce shear capacity?
1.55
S1M 33.0 3.90 Monotonic
0.10%
RESEARCH SIGNIFICANCE S1C (4790) (566) Cyclic
Many reinforced concrete structures contain members in L0M Monotonic
29.1 3.74
which a significant portion of the applied shear force is 0 (4220) (542)
L0C Cyclic
carried by strut action. This research is aimed at improving 2.29
understanding of the behavior of such members under both L1M 37.8 3.65 Monotonic
0.10%
L1C (5480) (529)
monotonic and reversed cyclic loading. Cyclic

Fig. 2—Test specimens; dimensions in mm. (Note: 1 mm = 0.0394 in.; 1 MPa = 145 psi = 0.145 ksi.)

ACI Structural Journal/November-December 2010 727


strength of the concrete at the time of testing of the beam and is 10% of the failure load of the previously tested companion
the average strength of three 150 x 300 mm (6 x 12 in.) cylinders. monotonically loaded specimen. Each monotonic test took
The modulus of rupture fr is also listed. The age of the specimens approximately 7 hours, whereas each cyclic test required 3 to
at the time of testing was approximately 150 days. 7 days to complete. In both cases, approximately 10% of the
total testing time was spent applying the loads, whereas the
Loading remaining time was used to measure and record the state of
Figure 3 shows the planned loading protocol for the the specimen.
reversed cyclic tests. The beams were tested using direct
loading and direct supports. In the downward cycle of Instrumentation
loading (positive load), the load was applied by the head of In each experiment, the applied force and support reactions
the testing machine. For the upward direction of loading, the were measured with five load cells, the relative displacements
load was applied by four hydraulic jacks pushing up on the between 72 grid points were measured using demountable
bottom face of the beam with the reaction provided by hold- displacement transducers, the diagonal and vertical web
down rods at the ends of the beams (refer to Fig. 4). The deformations and vertical member displacements were
reversal of the load, including the self-weight of the beam, measured with 16 linearly variable differential transformers
was assumed to take place under a bottom load of one half (LVDTs), reinforcement strains were measured with up to
the self weight, G/2, which corresponds to zero midspan 38 strain gauges, and crack widths were measured with crack
moment and zero shear at the middle of the shear spans. To comparators at approximately 70 locations. The load cells,
place similar demands under both positive and negative LVDTs, and strain gauges provided continuous readings,
loading, the loading protocol was kept symmetric with whereas the demountable displacement transducer measurements
respect to this G/2 load level. The first full unloading was and crack widths were recorded at a number of load stages,
done after major shear cracks approached the top loading as indicated in Fig. 3. The vertical displacements at midspan
plate. As shown in Fig. 3, the planned loading protocol used were measured with respect to a rigid rod hanging from
groups of three cycles with the same load amplitudes. The mounts at the ends of the beam. More information about the
load increment between groups of cycles was approximately experimental program can be found elsewhere.9

TEST RESULTS AND DISCUSSION


Load-displacement response
Figure 5 shows the measured load-displacement response
of all eight tested beams. The curves shown for reversed
cyclic tests S0C, S1C, L0C, and L1C are the positive
envelopes of the full hysteretic responses. As can be seen,
the shorter specimens are stronger and stiffer than the longer
specimens; and prior to first diagonal cracking, there is no
significant difference in the measured responses of the four
specimens within each length group. After first diagonal
cracking, the two members in each length group with shear
reinforcement display a stiffer response than the two without
shear reinforcement. For the members without shear
reinforcement, there is a further change in response once the
diagonal cracking reaches the supports. At this stage, the
observed response of the specimens subjected to cyclic
loading diverges from those observed for the monotonic test.

Fig. 3—Cyclic loading protocol.

Fig. 4—Specimen S1C after failure. Fig. 5—Load-displacement response.

728 ACI Structural Journal/November-December 2010


Fig. 6—Behavior of Specimen L0M. (Note: 1 kN = 0.225 kip; 1 MPa = 145 psi = 0.145 ksi; 1 mm = 0.0394 in.)

All beams failed in shear with crushing of the concrete near The incremental midspan displacement is indicated at the
the applied load prior to yielding of the longitudinal bottom right corner of the plots as a fraction of the
reinforcement and with no evidence of anchorage failure. displacement at peak load. Thus, only 18% of the total
Surprisingly, the two beams without stirrups subjected to displacement was caused by the 500 kN (112 kips) applied
cyclic load were significantly stronger than the between LS0 and LS2, whereas 16% was caused by the 40 kN
corresponding beams tested monotonically. This and other (9 kips) applied between LS6 and LS7. The scale diagrams of
aspects of the behavior of the specimens are discussed in the incremental displaced shapes have been magnified such
more detail as follows. that the maximum incremental vertical displacement equals
200 mm (7.9 in.). The dashed lines on the crack diagrams
Test L0M depict the variation of the longitudinal stress measured along
Specimen L0M, which had no stirrups and had an a/d of the bottom reinforcing bars. Finally, at the bottom of the
2.29, is in the zone of transition between deep and slender figure is a photograph of Specimen L0M taken after failure.
beams. Figure 6 shows crack diagrams at seven load stages Initially (LS0 to LS2), it can be seen that the
(LS0 to LS7) and the measured change of deformed shape deformations were predominantly flexural and the stresses
between each two consecutive load stages (“incremental” along the bottom reinforcement varied in a manner similar to
deformed shapes). These six incremental deformed shapes the bending moment. The first diagonal crack developed in
illustrate where damage in the specimen is happening during the right-hand-side shear span between LS2 and LS3,
each load increment. The final deformed shape gives the causing an asymmetrical incremental deformed shape with a
total deformation at LS7, which was 98% of the failure load. concentration of shear deformations along the diagonal

ACI Structural Journal/November-December 2010 729


crack. The shaded area of concrete in the LS3 crack diagram width of the tooth. It is of interest that this load increment,
can be viewed as a cantilever fixed at the top part of the which caused the large shear deformations on the left side of
beam, which Kani et al.4 called a “tooth.” The bond forces the beam, was not associated with large shear deformations
between the bottom reinforcement and the concrete tend to on the right side of the beam, even though there was a 1.05 mm
bend the cantilever toward midspan, while the aggregate (0.041 in.) wide diagonal crack on this side. Between LS5
and LS6, the existing diagonal cracks increased in width but
interlock (AI) and dowel action (DA) resist this bending.
did not significantly propagate and no new significant cracks
Both of these resisting mechanisms degrade as cracks widen; formed. Despite this, the load increment caused substantial
and by LS4, this cantilever had failed as shown by the now shear deformations, indicating that failure of the aggregate
nearly uniform stress in longitudinal reinforcement over the interlock of the crack surfaces had occurred. The load
increment between LS6 and LS7 caused failure of more
concrete teeth near the left support, resulting in a diagonal
crack that reached the edge of the supporting plate on this
side. The loss of the teeth resulted in almost constant tension
in the bottom reinforcement at LS7; therefore, the load-bearing
mechanism had switched from beam action to almost pure
arch action. The final failure involved crushing of the
concrete near the left edge of the loading plate above the
diagonal crack, and the beam failed under a load of 801 kN
(180.0 kips) and a midspan displacement of 10.0 mm (0.394 in.).
Figure 7 shows photographs of the zone of crushing at LS7
and at the end of the test. For convenience, this zone is called
the critical loading zone (CLZ).
It is interesting to note that the demountable displacement
transducer readings between targets on the same concrete
tooth showed the presence of small diagonal compressive
stresses even after the tooth had failed. This indicated that a
small part of the shear was carried under the main diagonal
crack up to the failure of the beam. This shear-carrying
mechanism can be referred to as “residual beam action.”

Effect of transverse reinforcement and a/d


The influence of minimum transverse reinforcement can
be seen by comparing the behavior of Specimen L0M to
Specimen L1M. The top two crack diagrams in Fig. 8 show
the two beams under an applied load of approximately
800 kN (180 kips). It can be seen that minimum stirrups
Fig. 7—Critical loading zone of Specimen L0M: (a) at LS7; restrained both the propagation of the diagonal cracks and
and (b) after failure. their width. The maximum crack width in Specimen L0M
was six times larger than that in Specimen L1M. The failure
of a concrete tooth in Specimen L1M (refer to the shaded
area) was followed by the formation of a stiff truss
mechanism consisting of compression struts supported by
stirrups. The third crack diagram in Fig. 8 shows Specimen
L1M just prior to failure, which occurred at a load of 1295 kN
(291 kips) and a midspan displacement of 14.2 mm (0.56 in.)
with crushing of the left critical loading zone. It is of interest
that although the failure load of member Specimen L1M was
62% higher than that of Specimen L0M, the crack diagrams
that correspond to loads close to failure are very similar in
both the pattern of cracking and the crack widths.
The top two diagrams in Fig. 9 illustrate the fully developed
arch action in Specimen L0M and the fully developed truss
action in Specimen L1M at failure. Strain measurements
showed that the stirrups in Specimen L1M yielded upon
diagonal cracking; therefore, the force in each pair of stirrup
legs was 69.5 kN (15.6 kips). The other internal forces shown
in Fig. 9 were obtained from equilibrium. It can be seen that
the critical loading zones of the two beams crushed under
diagonal forces of similar magnitudes (1129/987 = 1.14); but
as a result of the truss action in Specimen L1M, its ultimate
load was significantly larger than that of Specimen L0M
Fig. 8—Crack diagrams of Specimens L0M and L1M. (Note: (1295/801 = 1.62). The bottom two plots in Fig. 9 show
1 kN = 0.225 kip; 1 mm = 0.0394 in.) excellent agreement between the calculated and measured

730 ACI Structural Journal/November-December 2010


pattern of strains in the bottom reinforcement. It can be seen and it can be seen that the beam with stirrups had smaller
that the truss action in Specimen L1M is characterized by a crack widths. Also, note the difference in the longitudinal
decrease in tension in the longitudinal steel as the moment reinforcement stress profiles. The member without stirrups
decreases toward the supports, whereas in Specimen L0M, shows high steel stresses near the supports, indicating tied-
the tied-arch action results in an almost constant tension arch action. On the other hand, the member with stirrups
force in the longitudinal reinforcement. Thus, the provision showed a significant reduction in longitudinal steel stress
of just a minimum amount of transverse reinforcement not near the supports, indicating the effectiveness of truss action
only greatly reduces crack widths at similar load levels but even for this short span. For a deep beam with an a/d of 1.55,
also enables the diagonal compressive stresses in the concrete the ACI Code specifies a minimum quantity of 0.25%
to follow a more efficient path in carrying the shear. transverse and 0.15% longitudinal shear reinforcement.
For the shorter beams, the influence of a small amount of Specimen S1M contained just 0.10% of stirrups and no
transverse reinforcement can be seen by comparing the distributed longitudinal shear reinforcement. Nevertheless, these
behaviors of Specimens S0M to S1M, shown in Fig. 10. The stirrups provided significant crack control and increased the
top two crack diagrams compare the beams at the same load; shear strength by 31%.

Fig. 9—Arch action and truss action of Specimens L0M and L1M. (Note: 1 kN =
0.225 kip; 1 mm = 0.0394 in.)

Fig. 10—Crack diagrams of Specimens S0M and S1M. Fig. 11—Influence of a/d on shear strength. (Note: 1 MPa =
(Note: 1 kN = 0.225 kip; 1 mm = 0.0394 in.; 1 MPa = 145 psi.) 145 psi; 1 mm = 0.0394 in.)

ACI Structural Journal/November-December 2010 731


It was seen in Fig. 1 that the provisions of ACI 318-08 six of these specimens had similar concrete strengths, an
predicted the influence of the a/d on the shear strength of the effective depth of approximately 1100 mm (43.3 in.), and
relatively heavily reinforced (ρl = 2.7%) 540 mm (21.3 in.) five of the six specimens had 0.7% of longitudinal
deep specimens tested by Kani et al.4 Figure 11 shows a reinforcement. It can be seen that for the three specimens
similar comparison for the four monotonically loaded without stirrups, the ACI predictions are no longer consistently
members of this study, along with two geometrically similar conservative as they are in Fig. 1. Specimen 10T0 failed at
specimens (10T0 and 10T24) tested by Higgins et al.10 All just 62% of the ACI-predicted shear strength due to the size
effect in shear, and this unconservative ACI prediction is
consistent with many other tests of large, lightly reinforced
slender members.6 As the a/d ratio is reduced, the observed
shear strengths of the members without stirrups can be seen
to increase almost linearly. The ACI-predicted strength for
Specimen L0M is excellent, although the mode of failure is
predicted to be by beam action breakdown; and the observed
failure mode was by diagonal strut crushing. Specimen S0M is
unconservatively predicted by the ACI Code, though the
prediction of failure by strut crushing is correct.
Comparing Specimens 10T0 and 10T24 in Fig. 11, it can
be seen that for these large, slender beams the provision of
approximately minimum stirrups greatly enhanced the shear
strength largely due to the stirrups mitigating the size effect.
For the deeper members, the provision of just 0.1% of
transverse reinforcement also significantly increased the
shear capacity with the increase in capacity at a/d of 2.29 and
a/d of 1.55 being approximately equal. The ACI predictions
for these members are more accurate than for the members
without stirrups. However, the ACI strut-and-tie predictions
underestimate the beneficial effect of adding stirrups. Details
on the ACI strut-and-tie calculations are given elsewhere.9

Effect of load reversals


Figure 12 shows the full load-displacement responses of
the eight beams tested. Each of the four plots compares the
response of companion specimens subjected to monotonic or
reversed cyclic loading. For each monotonic/cyclic pair of
specimens, the initial load-deformation responses, up to first
load reversal of the cyclic specimen, were essentially
identical and, hence, plotted on top of each other. The four
cyclically tested beams had very similar hysteretic behavior
characterized by small energy dissipation (narrow hysteretic
loops), small cyclic degradation (a small increase in
displacements under consecutive cycles with constant load
amplitude), small residual displacements, and low ductility
(steep drop of resistance after peak load). The beams with
stirrups had slightly wider hysteretic loops as a result of the
yielding of the transverse reinforcement, which prevented full
closure of the shear cracks upon unloading. In general, residual
deformations were observed mainly along the major diagonal
cracks. All of the beams failed in shear by crushing in the CLZ.
The crack diagrams for the four cyclically loaded specimens
at the last recorded positive load stage prior to failure are
shown in Fig. 13. In each case, the pattern of the cracks
formed under positive loading was essentially the same as
that for the companion monotonically loaded test. For
example, compare Specimens L0C and L1C in Fig. 13 with
Specimens L0M and L1M in Fig. 8. In both cases, the crack
patterns and maximum crack widths are very similar.
Further, it can be seen that the crack pattern formed under
negative loading caused the total crack pattern to be symmetrical
about the midheight of the beam.
While it was only possible to show a few measured crack
widths in the previous crack diagrams, the maximum crack
width was determined from the crack widths measured at
Fig. 12—Full load-displacement response. approximately 70 locations at each load stage. Figure 14

732 ACI Structural Journal/November-December 2010


compares the maximum crack width observed for the positive load Δu, and displacement-to-shear-span ratios Δu/a. The
load stages with the applied positive midspan load for all ratio Δu/a is comparable to story drift in a column and for
eight beams. It can be seen from this plot that there is very these members, in which yielding of longitudinal reinforcement
little difference between the maximum crack widths did not occur, the values can be seen to be approximately 0.5%.
observed at any load for the cyclically loaded tests and that Table 2 also shows maximum applied point loads at shear
measured for the companion monotonic test. Note that the failure Pu and maximum shears (including self-weight) Vu at
members with transverse reinforcement (for example, the middle of the shear span. The final column in the table
Specimen L1M) displayed much better crack control than shows the ratios of the strength of the cyclically loaded
members with no stirrups (Specimen L0M). This was true beams to the strength of the companion monotonically
even for the shorter members with stirrups that contained
only 40% of the ACI minimum shear reinforcement for deep
beams, indicating that the provision of even very small
amounts of transverse reinforcement can have a significant
beneficial effect on crack widths. It is of interest that both
members with and without transverse reinforcement developed
crack widths greater than 2.5 mm (0.079 in.) prior to failure.
A major objective of the experimental work summarized
in this paper was to determine whether reversed cyclic
loading caused a significant reduction in the shear capacity
of deep beams. It is evident from Fig. 12 that no large reduction
in shear capacity was observed. The members with shear
reinforcement showed remarkable agreement between the
envelope of the cyclic response and the monotonic load-
deformation response including post-peak behavior.
Surprisingly, the two beams without stirrups subjected to
cyclic load were significantly stronger than the corresponding
beams tested monotonically. Particularly striking was the case of
Specimen S0C, which was 62% stronger than Specimen S0M.
An explanation for why Specimen S0C was stronger than
Specimen S0M can be found by looking at the CLZs of the
two beams. The concrete around the top loading plate of
Specimen S0C was expected to be weaker than that of
Specimen S0M because flexural tension cracks would occur
in this region under load reversals. However, it was found
that because the longitudinal reinforcement did not yield,
these flexural tension cracks closed completely when the
load reverted to the positive direction; thus, there was no
apparent reduction in the ability of this region to carry
diagonal compressive stresses. The small displacement
shifts between the hysteretic loops with the same load Fig. 13—Crack diagrams: Specimens S0C, S1C, L0C, and
amplitude was consistent with this observation. Another L1C. (Note: 1 kN = 0.225 kip; 1 mm = 0.0394 in.)
important factor controlling the strength of the CLZs was
their geometry. As shown in Fig. 10 and 13, the depth of the
CLZ at the face of the loading plate was 130 mm (5.1 in.) for
Specimen S0C as compared to only 70 mm (2.8 in.) for
Specimen S0M. This deeper CLZ likely enabled the strut in
Specimen S0C to resist the much higher loads. It is important
to note that the geometry of the critical zone in a cyclic test
was not influenced by the load reversals because the main
diagonal crack was formed by the initial monotonic part of
the loading protocol. The exact path of the critical diagonal
crack is a function of random factors such as local variations
of material properties over the shear span. It is suspected that
if Specimen S0C had been monotonically loaded to failure,
it would have been at least 62% stronger than the
corresponding monotonically loaded Specimen S0M.
Equally large variations in the shear strengths of nominally
identical monotonically loaded deep beams without stirrups
have been reported by other investigators.1

Summary of experimental results


Table 2 summarizes some important results from the tests.
The table gives maximum crack widths wmax measured at the
last load stage prior to failure, midspan deflections at peak Fig. 14—Crack widths.

ACI Structural Journal/November-December 2010 733


Table 2—Summary of experimental results
Cyclic/
Test a/d ρv fy, MPa fc′, MPa wmax, mm Δu, mm Δu/a, % Pu, kN Vu, kN monotonic
S0M 2.50 6.4 0.38 1420 721
0 34.2 1.62
S0C 3.70 10.9 0.64 2301 1162
1.55
S1M 2.30 7.7 0.45 1860 941
0.497 33.0 1.00
S1C 4.20 8.4 0.49 1864 943
L0M 2.00 10.0 0.40 801 416
0 29.1 1.19
L0C 4.00 11.1 0.44 953 492
2.29
L1M 3.50 14.2 0.57 1295 663
0.497 37.8 0.97
L1C 3.50 13.7 0.55 1253 642
Note: d = 1095 mm; b = 400 mm; and ρl = 0.70%; 1 kN = 0.225 kip; 1 MPa = 145 psi = 0.145 ksi; and 1 mm = 0.0394 in.

loaded beams. As can be seen, no significant degradation in 6. In assessing the shear strength of members with less than
shear capacity was observed. As discussed previously, it is the minimum vertical and horizontal shear reinforcement for
believed that the 1.62 and 1.19 ratios for members without deep beams specified by ACI 318-08,5 it is useful to
stirrups primarily result from the inherent variability of the recognize that members that contain only 0.001bws of
shear strength of such members. The provision of just 0.10% vertical shear reinforcement and no horizontal shear
transverse reinforcement essentially eliminated this variability. reinforcement, as was the case in this study, still show
greatly reduced crack widths at a given load and an increase
CONCLUSIONS in shear capacity compared to members without stirrups.
In this paper, load-displacement curves, crack diagrams, 7. Based on the experiments described in this paper, it can
deformed shapes, steel strain measurements, and photographs be concluded that deep members subjected to reversed cyclic
were used to establish a comprehensive picture of the behavior loading where direct struts can form between the applied
of eight deep beams subjected to shear. The variables were the load and the reaction location can be designed for shear
a/d (1.55 or 2.29), the amount of transverse reinforcement using the same provisions as those for monotonically loaded
(zero or 0.1%), and the type of loading (monotonic or members, provided that the longitudinal reinforcement
reversed cyclic). The following conclusions were reached: remains in the elastic range.
1. For all beams tested, there was a considerable increase
in the load that could be resisted by the beam after the ACKNOWLEDGMENTS
formation of the first significant diagonal crack. Prior to The research summarized in this paper was funded by the Natural
shear failure, extensive diagonal cracking had occurred Sciences and Engineering Research Council of Canada; its long-term
over the full length of the beam with the maximum width support is greatly appreciated. The headed reinforcement was donated by
of these cracks being approximately 4 mm (0.16 in.). The Headed Reinforcement Corporation; its support is also appreciated.
overall transition from beam action to arch action was
evident from the changing pattern of the strains in the REFERENCES
longitudinal reinforcement, which became nearly uniform 1. Rogowsky, D. M.; MacGregor, J. G.; and Ong, S. Y., “Tests of
Reinforced Concrete Deep Beams,” ACI JOURNAL, Proceedings V. 83,
from support to support prior to failure. No. 4, July-Aug., 1986, pp. 614-623.
2. The extent of cracking and the width of cracks were 2. Asin, M., and Walraven, J. C., “Numerical Analysis of Reinforced
similar for members with or without stirrups at shear failure. Concrete Continuous Deep Beams,” Heron, V. 40, No. 2, 1995, pp. 163-178.
Members with stirrups, however, resisted a larger shear force 3. Alcocer, S. M., and Uribe, C. M., “Monotonic and Cyclic Behaviour
and showed significantly narrower cracks at a given shear of Deep Beams Designed Using Strut-and-Tie Models,” ACI Structural
force compared to members without stirrups. Journal, V. 105, No. 3, May-June 2008, pp. 327-337.
3. Transverse reinforcement enhances the shear strength of 4. Kani, M.; Huggins, M.; and Wittkopp, R., “Kani on Shear in Reinforced
Concrete,” University of Toronto Press, Toronto, ON, Canada, 1979, 225 pp.
deep beams partly by decreasing the diagonal compression 5. ACI Committee 318, “Building Code Requirements for Structural
demands on the CLZs through the development of truss action. Concrete (ACI 318-08) and Commentary,” American Concrete Institute,
4. For members without transverse reinforcement, the Farmington Hills, MI, 2008, 473 pp.
monotonic response was an envelope to the cyclic response 6. Collins, M. P.; Bentz, E. C.; and Sherwood, E. G., “Where is Shear
up until the point when full arch action developed. After this Reinforcement Required? Review of Research Results and Design Procedures,”
ACI Structural Journal, V. 105, No. 5, Sept.-Oct. 2008, pp. 590-600.
stage, the cyclic experiments continued to a significantly
7. Wight, J. K., and Sozen, M. A., “Shear Strength Decay of RC
higher level of shear than the monotonic specimens. It is Columns under Shear Reversals,” Proceedings, ASCE, V. 101, No. ST5,
believed that this surprising result was due to random variations in May 1975, pp. 1053-1065.
the path of the diagonal crack near the CLZs. 8. Priestley, M. J. N.; Calvi, G. M.; and Kowalsky, M. J., Displacement-
5. For members with transverse reinforcement, the load- Based Seismic Design of Structures, IUSS Press, Pavia, Italy, 2007, 721 pp.
deformation response measured under monotonic loading 9. Mihaylov, B. I., “Behavior of Deep Reinforced Concrete Beams under
provided an excellent envelope to the cyclic response. Monotonic and Reversed Cyclic Load,” doctoral thesis, European School for
Advanced Studies in Reduction of Seismic Risk, Pavia, Italy, 2008, 379 pp.
Stirrups reduced the scatter in ultimate response by 10. Higgins, C.; Potisuk, T.; Farrow, W. C.; Robelo, M. J.; McAuliffe, T. K.;
constraining the random nature of the critical crack location and Nicholas, B. S., “Tests of RC Deck Girders with 1950s Vintage Details,”
and reducing the importance of the geometry of the CLZs. Journal of Bridge Engineering, ASCE, V. 12, No. 5, 2007, pp. 621-631.

734 ACI Structural Journal/November-December 2010

You might also like