TECHNICAL PAPER
INTRODUCTION
Reinforced concrete (RC) deep beams are generally used
as load-transferring elements, such as transfer girders, pile
caps, tanks, folded plates, and foundation walls. In buildings,
a deep beam or transfer girder is used when a lower column
on the exterior faade is removed for architectural purposes.
Sometimes the full depth of the floor-to-floor height is
used to transfer the high axial forces of columns above to
the supporting columns below. Large openings through structural members are frequently required for mechanical and
electrical conduits or even for means of passageways, such
as openings for doors and hallways in buildings. ACI 318-08
(ACI Committee 318 2008) defines a deep beam as a structural element in which either the clear span is equal to or less
than four times the overall depth, or the concentrated loads
are applied within a distance equal to or less than two times
the depth from the face of the support. Code-specified empirical formulas used to design these members do not explicitly
address the design of D-regions with openings. Strut-andtie models (STMs) are extensively used for these structures
with D-regions since their implementation in various U.S.
design codes. These models idealize a deep member as a
series of concrete compressive struts and steel tensile ties
connected at joints (called nodes) idealized as frictionless
pins forming a truss. The applied force is transferred
from the loading point to supports only through the STM,
and the remaining concrete between the trusses is neglected
for design and strength calculation purposes. STMs satisfy
any load system based on a statically admissible stress field
that does not exceed the yield criteria and provide safe and
lower-bound designs of discontinuous structures (Schlaich
et al. 1987; Muttoni et al. 1997). Hence, no unique STM
ACI Structural Journal/March-April 2012
193
using STMs. Moreover, the design method is at times ambiguous and cumbersome using these models. Experimental
investigations are required to evaluate the efficacy of these
models for the reliable and consistent prediction of ultimate
strengths and to identify the critical region for controlled
and predictable failure mechanisms. This study evaluates the
behavior of two RC deep beams, with a single large opening,
designed as per STMs. Further, the complex reinforcement
detailing was nearly completely replaced by steel fibers in
two more geometrically similar test specimens. The critical
regions in both types of test specimens are identified and
subsequently strengthened by reinforcing bars. The effect of
strengthening the critical regions of deep beams on their ultimate strengths and failure mechanisms is investigated.
DESCRIPTIONS OF TEST SPECIMENS
Specimen geometry
RC deep beams with large discontinuous regions were
considered as test specimens in this study. The test specimens
were 1/4-scale models of an example deep beam originally
considered by Schlaich et al. (1987) to evaluate the design of
STMs. Geometrically similar test specimens have also been
used previously in other laboratory tests (Maxwell and Breen
2000; Brea and Morrison 2007). All specimens were 74 in.
(1880 mm) long, 46 in. (1170 mm) deep, and 4.5 in. (112 mm)
thick. A 15 in. (380 mm) square opening was located at the
bottom corner near the support of the specimens and at a
distance of 5 in. (125 mm) from the boundaries (Fig. 1). It was
expected that the direct load paths between the loading point
and supports would be interfered with by the position and size
of the selected opening (Brea and Morrison 2007). A total
of four (that is, two RC and two SFRC) deep beam specimens
were tested in this study. The reinforcing bars in the RC
specimens were detailed as per a selected STM discussed in
the following section. Most of the reinforcing bars required
by the STM, however, were eliminated and replaced by steel
fibers in the SFRC specimens, in which steel bars were used
only for longitudinal tensile reinforcement.
Design STM
The principal factor in the design of concrete elements
with discontinuity regions is the selection of a suitable
STM. The position of struts and ties in a model can be based
on the elastic principal stress fields (Schlaich et al. 1987;
MacGregor 1997; Joint ACI-ASCE Committee 445 2002) or
the stress fields at the development plastic hinge mechanisms
(Muttoni et al. 1997). An STM approximately following
the elastic (principal) stress distribution was considered in
this study. It should be noted that the direction of principal
stresses may change after the cracking in concrete; however,
the flow of forces will help decide the position of struts
and ties in the model. The stress distribution shows that the
applied load in the test specimen is transferred directly from
the loading point to the right support through a bottle-shaped
strut; however, the opening near the lower left corner impairs
the direct load transfer from the load point to the left support.
As shown in Fig. 2, the STM considered in this study is basically a modified model proposed by Schlaich et al. (1987).
This model was also previously used by Brea and Morrison
(2007) for comparison with laboratory test data. The right
part of the STM beyond the loading point consisted of a truss
system to resist the transverse tension and the compression
in the bottle-shaped concrete strut formed due to the direct
flow of forces from the loading point to the support, whereas
ACI Structural Journal/March-April 2012
Fig. 3Layout of reinforcement (two layers for each bar location) and locations of strain
gauges: (a) Specimen RC1; (b) Specimen RC2; (c) Specimen SFRC1; and (d) Specimen SFRC2.
the left part consisted of two truss systems that transfer the
applied load to the left support around the opening.
Specimen reinforcement
The test RC specimens were designed for an ultimate
strength of 31.3 kips (139kN), the same as that used by
a previous study (Brea and Morrison 2007). The nominal
values of compressive and tensile strengths of the concrete
and reinforcement bars were assumed as 5000 and 60,000 psi
(35 and 414 MPa), respectively. A strength-reduction factor
equal to 0.75 was used for all struts, nodes, and ties in
the STM (Brea and Morrison 2007). Detailed information about the strut geometry and forces, and the efficiency
factor bs can be found elsewhere (Brea and Morrison
2008). No. 3 (10mm) bars with a nominal area of 0.11 in.2
(71 mm2) were used as steel reinforcing bars, provided in
two layers in the RC specimens with a clear concrete cover
of 1 in. (25 mm) to the edges of the reinforcing bars. All
reinforcing bars ending near the edges of the specimens
were provided with standard hooks with required development lengths to provide sufficient anchorage and avoid their
pullout. Secondary reinforcing bars for the temperature
and shrinkage cracking for walls were not provided in this
study because these bars can significantly enhance the loadcarrying capacities of test specimens (Brea and Morrison
2007). Thus, the minimum vertical and horizontal reinforcement requirements as per the ACI 318-08, Section 14.3.1,
provisions for walls were not satisfied for these specimens.
Although the secondary reinforcement was expected to
increase the ductility of the concrete, thus allowing a trusslike plastic mechanism to form, it was hardly realized, as
evidenced by prior experimental results (Maxwell and Breen
2000; Brea and Morrison 2007; Kuchma et al. 2008). It
should also be noted that steel fibers not only enhance the
ACI Structural Journal/March-April 2012
Nominal (28-day),
psi (MPa)
RC1
RC2
SFRC1
SFRC2
Measured (day of
testing), psi (MPa)
6185 (43)
5000 (35)
6645 (46)
5867 (41)
60,000 (414)
81,240 (560)
126,700 (874)
6225 (43)
Fig. 4Flexural behavior of SFRC: (a) load-displacement behavior of ASTM beams; and
(b) multiple cracks in ASTM beams. (Note: 1 kip = 4.45 kN; 1 in. = 25.4 mm.)
of cracks emanating from the corner of the opening, two
No. 6 (20mm) bars were placed diagonally in two layers
near the opening of Specimen SFRC2, as shown in Fig.3(d).
These bars were purposely overdesigned to prevent the
failure initiating from the corner so as to achieve better
stress distribution in the specimen. These diagonal bars were
oriented as normal to the line connecting the loading point to
the corner of the opening. Unlike the RC specimens, the SFRC
specimens did not require any reinforcement detailing, which
resulted in a very fast and simple construction process.
Mixture compositions and material properties
A concrete mixture of nominal 28-day compressive
strength equal to 5000 psi (35 MPa) was used in all test
specimens. A mixture design was carried out to achieve the
target compressive strength of concrete and use the optimum
quantity and similar proportions of materials in both the
RC and SFRC specimens. The design mixture proportion
(by weight) used for all specimens was 1.0 (cement):0.5
(fly ash):1.7 (sand):1.0 (coarse aggregate). Type I portland
cement, Class C fly ash, and coarse aggregates of maximum
size limited to 0.5 in. (13 mm) were used in the concrete
mixture. A constant water-cementitious materials ratio (w/cm)
of 0.4 was used in all specimens. No chemical admixtures
or high-range water-reducing admixtures were added to the
concrete mixture. Both SFRC specimens consisted of endhooked steel fibers (diameter = 0.03 in. [0.75 mm]; length =
2.4 in. [60 mm]; aspect ratio = 80; tensile strength = 152.3 ksi
[1050 MPa]) of volume equal to 1.5% of the total volume of
the specimen. Based on an earlier study by Liao et al. (2010)
for a highly flowable mixture, the weight of cement and
steel fibers used in the SFRC specimen was 312 and 77 lb
(142 and 35 kg), respectively.
Standard tests (ASTM C31/C3M-09, ASTM C39/C39M-09,
and ACI 318-08) were carried out to evaluate the compres196
condition, whereas the roller at the right support represented a pin support. However, the presence of horizontal
restraints at the supports had a negligible effect on the strutand-tie forces (Brea and Morrison 2007). Several sensors
were used to measure the applied load and the response of
the specimens at different load levels. A load cell with a
capacity of 600kips (2670kN) was used at the loading point
to measure the magnitude of monotonic load applied to the
specimens. Uniaxial 120 Ohm electrical strain gauges with a
gauge length of 0.2 in. (5 mm) were attached to the surface
of the steel reinforcing bars at specified locations to measure
the magnitude of strain and, hence, to compute the tie forces
at various load levels (Fig. 3). These locations were finalized
where maximum strain would be expected so that the calculated tie forces could be compared with the predicted forces
from the STM. Four LVDTs were also used on the surface of
the specimens to measure the deformation of concrete struts
formed in the specimens during testing (Fig.1). A linear
potentiometer was used below the load point to measure the
deflection of the specimens and two additional linear potentiometers were used to measure the displacement or slippage
of both supports, if any.
TEST RESULTS
The performance of the test specimens was evaluated in
terms of the following parameters: overall cracking, loaddeflection response, failure mechanism, ultimate strength,
and variation of tie forces. A detailed discussion on these
parameters is presented in the following sections.
Overall cracking
The propagation of cracks in the test specimens was
mapped after each load increment of 5 kip (22.5kN) intervals. The first crack in Specimen RC1 was noticed near the
supports at a load level of 25 kips (112.5kN). As shown
in Fig. 6(a), diagonal cracks initiated from the opening
at a 30 kip (135 kN) load level and propagated with
Fig. 6Crack propagation in test specimens: (a) Specimen RC1; (b) Specimen RC2; (c) Specimen SFRC1; and (d) Specimen SFRC2. (Note: Dimensions in kips; 1 kip = 4.45 kN.)
ACI Structural Journal/March-April 2012
197
Fig. 7Crushing of concrete strut near supports: (a) Specimen RC1; and (b) Specimen SFRC1.
the further increase of load up to 60 kips (270kN). At
a 50 kip (225 kN) load level, several instances of cracking
and crushing of concrete were observed near the right support
of Specimen RC1 because of the lack of confinement action
to the concrete. The width of the crack originating from the
opening to the loading point was 0.016 in. (0.40mm) at
a 65 kip (293kN) load level. A major flexural crack running
almost the full depth of Specimen RC1 was observed just
below the load point at a load level of 70 kips (315kN). The
loading in Specimen RC1 was stopped at this point due to
severe local damage and instability at the support regions.
In contrast to the single diagonal and flexural cracks noticed
in Specimen RC1 during the testing, as shown in Fig. 6(a),
Specimen RC2 showed several distributed cracks up to
failure. The first crack in Specimen RC2 was noticed at a
load level of 20 kips (90kN). As shown in Fig.6(b), several
diagonal cracks initiated at a load level of 40kips (180kN)
around the opening, and the maximum width of the crack
was measured as 0.012in. (0.3mm). These diagonal cracks
propagated toward the loading point as the applied load
levels increased. Flexural cracks started from the bottom of
the specimen below the load point at a load level of 65 kips
(293kN) and the maximum width of the crack measured was
0.03in. (0.75mm). This was consistent with Specimen RC1,
in which a major flexural crack was observed at 70 kips
(315 kN). Beyond the load level of 85 kips (382.5kN), the
diagonal cracks started to propagate horizontally toward
the load point. The width of the major flexural crack was
0.1in. (2.54 mm) at the load level of 95 kips (428 kN), and
this crack propagated toward the load point up to 130kips
(585kN) before complete failure. Unlike Specimen RC1,
Specimen RC2 did not exhibit any local damages near the
supports because of the sufficient concrete confinement
provided through steel caging near the supports. However,
both RC specimens showed that initial cracks propagating
from the opening did not cause the failure of the specimens;
instead, flexural cracks developing at the higher load levels
controlled the failure mechanism and the ultimate strength.
Hence, the RC specimens designed based on an STM effectively transferred the applied forces to the supports without
any local damage around the opening regions.
The first (minor) crack was observed in Specimen SFRC1
near the support at a load level of 15 kips (67.5kN). A small
crack 0.004 in. (0.10 mm) wide formed near the bottom
middle span of the specimen at a 20 kip (90kN) load level,
198
RC1
RC2
SFRC1
70.3 (313)
14.3 (64)
31.3 (139)
SFRC2
Plain concrete*
68.2 (303)
132.1 (588)
65.0 (289)
96.8 (431)
Modeled by CAST; only bottom reinforcing bars were used as that of SFRC specimens. Nominal strengths are computed by using nominal material properties of material with
strength-reduction factor of 0.75. Expected strengths are calculated by using actual material properties.
greater than 1.0%, which was depicted by the strain-hardening behavior in the load-deflection response. The ratios of
measured ultimate forces to the computed values in various
ties were nearly 1.5 for Specimen RC2 with the strengthened
boundaries. Although the STM adequately identified the
locations of critical ties (with maximum strain levels) in the
specimens, the model failed to capture the important role of
anchorage bars in the load-sharing mechanism, particularly
if the vertical segments of the openings were strengthened
against premature local failures. The STM underestimated
the forces in the ties located at the bottom of the specimen
and in the bottle-shaped struts. In addition to these ties,
both horizontal and vertical steel reinforcing bars around
the opening were found to be critical for these specimens.
Table3 also summarizes the force carried by the steel reinforcing bars used as ties in the SFRC specimens. Both the
bottom longitudinal tensile bars and diagonal bars carried an
equal proportion of forces at the failure stage, indicating a
significant role of the diagonal bars in enhancing the performance of Specimen SFRC2. The steel reinforcing bars in
the SFRC specimens, however, carried a smaller amount
of force as compared to the RC specimens, indicating the
reduction in strain demand on the reinforcing bars in the
SFRC specimens due to additional tensile strength provided
by the steel fibers.
SUMMARY AND CONCLUSIONS
The sufficient plastic redistribution of internal forces is
essential for a structure to sustain expected and unexpected
loads and to fail in a ductile manner if overloaded. In RC
members, due to the brittle nature of concrete, this redistribution primarily relies on the steel reinforcing bars and
their layouts, in which bars are placed at locations where the
concrete is overly stressed beyond its cracking strength. For
typical concrete members with simple and regular geometries, those locations can be easily predicted by classical
elastic theory. It is well known, however, that the stress
pattern is highly nonlinear and deviates considerably from
the classical elastic theory for RC members with significant
geometric discontinuities. These members with significant
geometric discontinuities and complex stress distributions
under loading require considerable analyses and usually
complicated reinforcement detailing. The reinforcement
detailing of these concrete members based on STMs can
be quite complicated and, very often, these models cannot
predict the failure mechanism due to localized damages.
Also, the actual stress fields in such members are typically
very different from those predicted by STMs, as indicated by
many experimental investigations.
This study investigates the behavior of deep beams with
large openings that were designed using STMs. Two RC and
two SFRC test specimens were tested under monotonically
increased loads. A nearly self-consolidating SFRC mixture
was used without any workability issues even with 1.5% of
steel fibers by volume. Reinforcing bars in the SFRC specimens were required at only a few critical locations. Those
bars served as ductile links to prevent the breakdown of
the highly stressed regions before the fully plastic redistribution of internal forces through steel fibers. The main
objectives of this study were: 1) to investigate the effect of
local strengthening on the load-transferring mechanism and
failure modes of test specimens; 2) to study the behavior of
SFRC specimens and compare that behavior with that of the
RC specimens designed using STMs; and 3) to identify the
201
E24
RC specimen
E19
E18
E86
VT
E90
E10
E2
E14
0.11
0.11
0.11
0.11
0.11
0.11
0.11
0.11
10
0.11
11
0.11
12
0.11
13
0.11
14
0.11
15
0.11
16
0.11
17
0.11
20
0.11
18
0.11
19
0.11
22
0.11
770.0
22.33
2.5
2000.0
58.00
6.4
1380.0
40.02
4.4
1640.0
47.56
5.2
960.0
27.84
3.1
740.0
21.46
2.4
2400.0
69.60
7.7
2540.0
73.66
8.1
50.0
1.45
0.1
10.0
0.29
0.0
30.0
0.87
0.1
2010.0
58.29
6.4
2230.0
64.67
7.1
1800.0
52.20
5.7
1570.0
45.53
5.0
320.0
9.28
1.0
25.6
22.0
12.8
12.0
11.5
25.6
24.7
9.7
30.0
0.87
0.1
1830.0
53.07
5.8
1210.0
35.09
3.9
4.9
20.4
0.80
5.4
0.25
15.8
1.23
0.1
0.01
6.5
14.2
1.23
12.8
0.50
11.9
0.48
7.7
0.80
SFRC specimen
E18
DG
0.11
462.1
13.40
1.5
0.11
490.4
14.22
1.6
0.11
491.9
14.27
1.6
0.11
389.6
11.30
1.2
0.44
0.44
3.0
2.8
Tu /Tcalc
79.97
8.8
2802.3
81.27
8.9
4375.4
81.85
9.0
2447.8
70.99
7.8
3102.2
81.40
9.0
15,994.3
92.00
10.1
13,520.2
90.00
9.9
11,966.7
88.00
9.7
15,984.6
94.00
10.3
871.8
5.19
0.6
178.9
6.29
0.7
217.0
3.35
0.4
115.6
88.06
9.7
3036.5
81.30
8.9
5538.4
82.00
9.0
7723.8
84.00
9.2
11,107.1
87.00
9.6
10,409.9
86.50
9.5
1197.3
34.72
3.8
Microstrain
Stress, ksi
Microstrain
Stress, ksi
Strain gauge
Tie No.
Specimen type
17.9
33.6
1.31
20.0
0.91
20.0
1.57
1.1
0.10
1.1
19.4
1.68
36.4
1.42
38.2
1.55
7.6
0.79
27.01
3.0
1086.1
31.50
3.5
675.3
19.58
2.2
897.6
26.03
2.9
382.2
11.08
4.9
42.5
1.23
0.5
6.4
5.0
5.4
REFERENCES
203
NOTES:
204