You are on page 1of 10

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title No. 112-S13

Inverted-T Beams: Experiments and Strut-and-Tie Modeling


by N. L. Varney, E. Fernández-Gómez, D. B. Garber, W. M. Ghannoum, and O. Bayrak

Contrary to rectangular deep beams, inverted-T beams are loaded In the past two decades, many structural design codes
on a ledge at the bottom chord of the beam. This loading configura- have adopted strut-and-tie modeling (STM) as a more trans-
tion induces a tension field into the web and the resulting complex parent option for the design of deep beams and other struc-
strain distribution renders sectional design provisions inadequate. tures with discontinuities. The current STM provisions were
The applicability of strut-and-tie modeling (STM), developed for
developed for rectangular deep beams and simple structures
rectangular deep beams and simpler, two-dimensional designs, was
with two-dimensional strain distributions, but have not been
evaluated. An experimental study was conducted in which 33 tests
were performed on 22 large-scale reinforced concrete inverted-T experimentally investigated for more complex structural
beams and the effects of the following variables were investigated: elements such as inverted-T beams.
ledge geometry, quantity of web reinforcement, number of point Due to scarcity of experimental research on inverted-T
loads, member depth, and shear span-depth ratio. It was concluded beams, a comprehensive large-scale experimental program
that strut-and-tie modeling, although developed for much simpler was undertaken to examine the behavior of such structural
structural components, offers a simple and accurate design method elements and assess the validity of implementing STM
for the more complex strain distributions in inverted-T beams. design. Thirty-three specimens were tested as part of the
The STM provisions developed for rectangular beams accurately research program. Unlike those found in the literature,
captured both failure mode and ultimate capacity and are recom- the test specimens in this program were considered more
mended for use in inverted-T beam design, as a major conclusion
representative of inverted-T beams designed in practice in
of this research.
terms of their size, geometric and loading properties, and
Keywords: D-region; inverted-T beam; laboratory testing; large-scale; reinforcement details. This paper presents the STM design
nonlinear design; reinforced concrete; shear; shear span; strut-and-tie. provisions as applied to inverted-T beams, the laboratory
test results, and the corresponding design recommendations.
INTRODUCTION
Inverted-T bent caps are often used in construction to RESEARCH SIGNIFICANCE
reduce the overall elevation of bridges and/or to improve Significant diagonal web shear cracking of inverted-T
available clearance beneath the beams, as shown in Fig. 1. bent caps may represent a risk both in terms of strength
The bent caps are beams that support bridge girders on and serviceability. Due to the nonlinear distributions of
ledges near the bottom of the beam, effectively loading the strains in inverted-T beams, STM offers a safe, lower-bound
cap along its tension chord. Within a given cross section design alternative to examine forces and predict the failure
(transverse direction), the loads are transferred from the mode in an element. Current strut-and-tie provisions were
ledges to the bottom of the web and then “hung” vertically developed for rectangular deep beams and have not been
to the compression chord, generating tension fields in the investigated for the three-dimensional state of stress present
web at the loading points. The loads are then transferred in these structures. The research presents an extensive large-
in the longitudinal direction to the supports, as in a typical scale experimental program aimed at assessing the accuracy
compression-chord-loaded beam. This three-dimensional and conservatism of strut-and-tie modeling for the design of
flow of forces, in addition to the deep beam loading condi- inverted-T bent caps. The unique experimental data presented
tions commonly encountered in bent caps, generate regions in this paper and the assessment of STM design provisions
of stress discontinuities that are traditionally designed using is considered to be significant contributions to the literature.
empirical equations and rules of thumb.
Significant web shear cracking of recently built inverted-T BACKGROUND AND STRUT-AND-TIE MODELING
straddle bent caps has been reported in Texas, according to Typically for reinforced concrete beams, a designer makes
the Texas Department of Transportation (TxDOT), trig- the assumption that plane sections remain plane, referred
gering concerns about the current design procedures. Most to as the Bernoulli hypothesis or beam theory. Within this
inverted-T bent caps in Texas are designed using sectional theory, the strains in the beam are presumed to vary linearly
provisions for the web and a series of checks for the ledges through the depth of a section; thus, the beam is said to be
that closely follow the procedures found in the AASHTO dominated by sectional behavior. As shown in Fig. 2(a),
LRFD Bridge Design Specifications.1 Due to the load and
ACI Structural Journal, V. 112, No. 2, March-April 2015.
geometric discontinuities in inverted-T beams, this beam MS No. S-2013-064.R1, doi: 10.14359/51687403, received June 10, 2014, and
theory is not valid; thus, sectional design provisions cannot reviewed under Institute publication policies. Copyright © 2015, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
be used to properly design such structures. obtained from the copyright proprietors. Pertinent discussion including author’s
closure, if any, will be published ten months from this journal’s date if the discussion
is received within four months of the paper’s print publication.

ACI Structural Journal/March-April 2015 147


Fig. 1—Typical inverted-T bent cap. Fig. 2—(a) B- and D-regions in a rectangular beam; and
(b) corresponding STM.
these regions of linear stress (or strain) are referred to as
B-regions (with the “B” for beam or Bernoulli). Strut-and-tie models consist of three components: struts,
A D-region (with “D” standing for discontinuity or ties, and nodes. These are assembled together to represent
disturbed) can typically be found on either side of a B-region, the flow of forces through a structure, as shown in Fig. 2(b).
as shown in Fig. 2. These regions cannot be designed After calculating the external reactions and defining the
using the sectional procedures as the strain distribution is geometry of the STM, the individual member forces of the
nonlinear and thus the assumptions used to derive the beam truss are determined through statics.
theory are no longer valid. These disturbances are caused Struts are compression elements that vary in shape
either by abrupt changes in geometry or loading. Frame depending on their location. Represented by dashed lines,
corners, dapped ends, openings, and ledges are examples of struts can be bottle-shaped if allowed to spread along their
geometric discontinuities. Point loads such as girder bear- length or prismatic in regions of uniform stresses, such as
ings or support reactions are examples of load discontinu- the compression zone of a beam’s B-region. It is important
ities. According to St. Venant’s Principle, an elastic stress to provide transverse reinforcement to control the tensile
analysis would indicate that the effects of a disturbance stresses caused by the spreading of compressive forces in
dissipate at approximately one member depth away from the bottle-shaped struts.4
discontinuity.2 In other words, a D-region extends approx- Tension elements, or ties, are generally made up of rein-
imately one effective member depth d from the load or forcing steel and denoted by solid lines, as seen in Fig. 2.
geometric change. Enough reinforcement must be provided to carry the tensile
A deep beam is one in which the entire span involves demand of the tie and should be distributed so that the
predominant nonlinear strain distributions through the depth centroid coincides with the tie location. Details such as bar
of the section. For this strain condition to exist, the shear spacing and anchorage are essential for proper STM.
span a must be less than approximately 2 to 2.5 times the Due to the concentration of forces from intersecting truss
effective member depth d. The right shear span in Fig. 2(a) is members, the nodes are the most highly stressed regions of a
entirely composed of D-regions and is thus considered a deep structural member. Three types of nodes can exist within an
beam. A beam with a greater shear span-depth ratio (a/d), STM and are denoted by the intersecting elements. Within
as shown in the left span, is assumed to behave according the nodal designations, “C” refers to a compression element,
to beam theory and can be designed using sectional proce- such as a strut, an externally applied load, or a support reac-
dures. The design of deep beams requires the use of STM or tion, and “T” stands for tension or tie. A CCC node is one
other non-linear procedures outside the scope of this paper. in which only struts intersect, a CCT node has tie(s) inter-
STM offers an approach for obtaining lower-bound solu- secting in only one direction, and a CTT node has ties inter-
tions for the strength design based on simple truss theory. secting in two different directions. The type of node governs
The resultants of complex states of stress are idealized as its behavior and thus its strength.
a system of uniaxial force elements, or a truss, within the The versatility of STM allows it to be used to design any
member as shown in Fig. 2(b). This system will yield a structure and accommodate various load transfer mechanisms.
conservative design if the resulting truss model is in equi- In theory, as long as the principles required to achieve a lower
librium with the external forces and the concrete has enough bound solution are met, any model can be considered a safe
deformation capacity to accommodate the assumed distri- design. A model should, to the furthest extent possible, follow
bution of forces.3 Proper anchorage of the reinforcement is the actual stress field as determined by an elastic stress anal-
crucial. The factored forces also must not exceed the factored ysis. If the model varies substantially from the stress field, the
strengths as shown in the following equation. structure will undergo substantial deformations leading to an
increased chance of cracking. Schlaich et al.3 provides addi-
ϕFn ≥ Fu (1) tional discussion on using STMs to design structural concrete
members. The applicability of STMs has been validated with
experimental testing on rectangular deep beams.5 However,

148 ACI Structural Journal/March-April 2015


experimental testing on more complicated structures such as Table 1—ACI 318-116 and Birrcher et al.4 STM
inverted-T beams is limited. concrete efficiency factors
ACI 318-11
Recent advances in STM provisions
Strut, fce = Node, fce =
ACI 318-116 STM examines the strength of struts and
0.85βsfc′ βs 0.85βnfc′ βn
nodes separately. For the strut-node interface, the smaller
of the nominal compressive strength of the end of the strut, Prismatic 1.00 CCC 1.00
Fns, and of the face of the nodal zone at which the strut acts, Bottle-shaped* 0.75 CCT 0.80
Fnn, is used. These nominal compressive strengths are calcu- Tension flange 0.40 CTT 0.60
lated using their respective cross sectional areas, Acs and Anz,
Birrcher et al.
as shown in following equations. The concrete efficiency
factors, βs and βn, are summarized in Table 1. Node, fcu = Strut-node
mνfc′ Bearing face, ν Back face, ν interface†, ν

Fns = 0.85βs fc′Acs (2) CCC 0.85 f c′


0.45 ≤ 0.85 − ≤ 0.65
CCT 0.70 20
Fnn = 0.85βs fc′Acs (3) *
Without reinforcement satisfying ACI 318 Section A3.3.3, βs = 0.60λ.

Without reinforcement satisfying AASHTO Article 5.6.3.5, ν = 0.45.
Birrcher et al. conducted a thorough investigation of rect-
4

angular deep beam behavior to improve upon the current well these provisions captured the strength and behavior of
STM provisions and recommend modifications to both the inverted-T beams with their additional tension field.
the ACI 318-116 and AASHTO LRFD1 codes. Potentially, The effectiveness of the Birrcher et al.4 STM design provi-
the most significant modifications that affect the design sions were demonstrated by using an extensive database of
of inverted T-beams focused on the node strength. This rectangular deep beams. Improvements were made in the
procedure simplifies the design of struts by considering overall conservatism and accuracy as well as simplicity of
the strut-node interfaces, which implicitly accounts for the STM for deep beams as compared to current ACI 318-116 and
strut capacity and eliminates redundant stress checks at the AASHTO1 procedures. However, the application of STMs
same location. was not investigated experimentally for more complicated
In Birrcher et al.,4 the design strength of the node, Fn, is deter- three-dimensional structures such as inverted-T beams. Due
mined by the limiting compressive stress at the faces of the node to their unique geometry, certain assumptions not addressed
by the concrete efficiency factor ν in the following equation. in the current STM procedures had to be made in the design
of inverted-T beams.
Fs = 0.85νfc′Acn (4)
Strut-and-tie modeling of inverted-T beam specimens
The appropriate concrete efficiency factor, ν, is used to Inverted-T beams transfer load in three dimensions: from
reduce the compressive strength of the concrete in the node the ledges to the web, from the tension to the compression
depending on the type of node (CCC, CCT, or CTT) and chord, and from the loading points to the supports. To capture
face (bearing face, back face, strut-node interface) under this behavior, it is necessary to consider a three-dimensional
consideration. The factors developed by Birrcher et al.4 are strut-and-tie model. To simplify the analysis, the model is
summarized in Table 1 along with the existing factors in divided into two compatible two-dimensional models.
Appendix A of ACI 318-11.6 It should be noted that Birrcher The STM design of an inverted-T beam is often iterative,
et al.4 recommended removing reference to CTT nodal as many factors are interdependent. First, the overall geom-
regions as they are typically smeared nodes and emphasis etry was determined based on the experimental variables
for deep beam design should be placed on the more critical under consideration, and the preliminary loads and corre-
CCC and CCT nodal regions. The cross-sectional area of the sponding reactions were calculated. The STMs were then
node, Acn, such as the strut-node interface shown in Fig. 3, is detailed to carry the required loads. Diagonal shear cracking
limited in the perpendicular direction by either the width of was a primary concern in this study, thus the beams were
the bearing plate or web width bw. designed to ensure a shear failure.
For bearing areas smaller than the width of the struc- Defining geometry of longitudinal strut-and-tie model—
tural member, the concrete strength for all the faces in that An example of a simple longitudinal STM for an inverted-T
node was increased to account for triaxial confinement. The beam with two shear spans is shown in Fig. 4. Each tie was
triaxial confinement factor, m, is in Article 5.7.5 of AASHTO aligned with the centroid of the reinforcing bars. Vertical
LRFD1 and Section 10.14.1 of ACI 318-11.6 hanger bars were placed at each load point with the tie corre-
It can be noted from Table 1 that the efficiency factor for sponding to the center of the bearing pad. A 45-degree spread
a strut-node interface is given as the same for both CCC and on the ledge under the loading plates defined the width of
CCT nodes according to the recommendations put forth by hanger ties.7 For cut-off ledges, load spread was limited on
Birrcher et al.4 These provisions do not reduce the nodal one side, as shown in the STM.
strength due to the presence of a tension field in an inverted-T The horizontal tie along the bottom of the beam was
beam as the node below the applied load is a CCT node, aligned with the centroid of the flexural reinforcement. The
rather than a CCC node. It was of interest to observe how width of the tie was assumed to be twice the distance from

ACI Structural Journal/March-April 2015 149


Fig. 3—Geometry of CCT Support Node A.

Fig. 4—Longitudinal strut-and-tie model.


the extreme tension fiber to the centroid of the steel as shown ensure they were greater than or equal to 25 degrees but
in the detail of Node A in Fig. 3. less than 65 degrees. This limit was enforced to prevent
All ties must be properly anchored to achieve the assumed an incompatibility of strains.4 A smaller angle would result
stress distribution. ACI requires that the yield strength of the in the tension tie overlapping more of the diagonal strut,
tie be developed at the critical point where the centroid of decreasing its effectiveness. The forces in each element were
the tie meets the end of the extended nodal zone or edge of calculated using statics and the element size and location
the diagonal strut as shown in Fig. 3. adjusted as needed.
The location of the intermediate (stirrup) Tie BC for the Defining geometry of cross-sectional strut-and-tie
two-panel mode in Fig. 4 was determined using the technique model—A cross-section STM was required to design the
proposed by Wight and Parra-Montesinos8 that any stirrup that ledge of the inverted-T beam. The external loads were
intersects an adjacent strut at an angle greater than 25 degrees applied equally to each ledge as shown in Fig. 5. The hanger
can be engaged as part of the vertical tie in the CCT node. ties in the longitudinal model corresponded to the vertical
A line projected at a 25 degree angle from the edge of the reinforcement. The top of the ledge reinforcement corre-
support plate at Node A to the top of the beam defined the sponded to the horizontal ledge tie. The centroid of the hori-
left limit of the tie. The right limit was defined by the edge of zontal strut shown was located at the depth of the flexural
the 45-degree load spread for hanger Tie DE and all stirrups reinforcement from the longitudinal model. A diagonal strut
that fell within the rectangular shaded region were considered transferred the load from the bearing plate to the bottom of
part of the vertical Tie BC, as shown in Fig. 4. The 45-degree the hanger bars.
load spread was an initial assumption validated with data from Shear spans under investigation—Once the forces in the
strain gauges applied to the hanger reinforcement. truss members were calculated and the nodes checked using
The horizontal strut along the top of the beam was the Birrcher et al.4 recommendations, the required steel
assumed to be prismatic with a depth equal to the depth area was determined to satisfy the tie tensile forces. Proper
of the equivalent rectangular compression stress block as anchorage of the ties was provided within the extended
defined from a typical flexural analysis. Although not tech- nodal regions.
nically valid in a D-region due to the nonlinear distribution Shear spans a equal to 2.50d and 1.85d were examined, as
of strains, defining the depth of the strut using a flexural shown in Fig. 4, with the two-panel model on the left end of
analysis is considered conservative and the assumption is well- the beam and the single diagonal strut on the right. The a/d
established in practice.9 is defined within the context of this paper as the ratio of the
Diagonal bottle-shaped struts, represented by dashed lines, distance from the center of the support to the center of the
complete the flow of forces in the longitudinal strut-and-tie nearest loading point a, with respect to the effective depth d.
model. The angles between struts and ties were checked to The specimens were designed to be shear critical with two

150 ACI Structural Journal/March-April 2015


web shear failure modes, stirrup tie yielding for a/d of 2.50
and diagonal strut crushing for a/d of 1.85. The a/d of 2.50
was chosen to evaluate the limit of deep beam behavior and
compare with the Birrcher et al.4 studies. For this longer a/d,
the intermediate (stirrup) tie was designed to govern, thus
the capacity was determined by the quantity of stirrups in
the tie region. For specimens tested at an a/d of 1.85, the
strut-node interfaces of the diagonal strut were designed
to govern. Thus the capacity of the specimens tested at the
shorter a/d were determined by the size of the node (Fig. 3)
and the compressive strength of the concrete.
Resistance and load factors are required for STM design
but were taken as 1.0 for the purpose of this investigation as
nominal strengths were computed and compared with exper-
imental strengths. In general, no serviceability checks were
made before testing the specimens. Rather, the cracking Fig. 5—Cross-sectional strut-and-tie model.
data obtained from loading the beams were used to validate Fig. 7. Each beam was tested at an a/d of 1.85 or 2.5 to
current serviceability provisions and/or make recommen- observe the two web shear failure modes: diagonal strut
dations for application to inverted-T beams.7 For step-by- crushing or stirrup tie yielding.
step STM design examples for inverted-T bent caps, refer to Ledge length—The varying ledge lengths of inverted-T
Williams et al.9 bent caps were simplified to three types. A “cut-off” ledge
is one in which the ledge was interrupted just past the edge
DATABASE OF EXPERIMENTS CONDUCTED ON of the bearing pad of the outermost girder. If the ledge ran
INVERTED-T BEAMS continuously to the support, it was considered a “long”
A thorough literature review was conducted prior to ledge. In a bent cap with a “short” ledge, the ledge continued
establishing the experimental investigation. A total of 97 a distance approximately equal to the depth of the ledge past
tension chord-loaded specimens reported within 13 unique the outermost girder, as shown in Fig. 7.
sources10-22 were ultimately compiled in a collection data- As previously mentioned, inverted-T bent caps are
base.7 Two sets of filters were used to develop the final tension chord-loaded structures in which the bridge girders
inverted-T database to meet the purposes of this project. supported on the ledges induce a tension field in the web.
The first filter focused on data required to develop STMs. The size of this tension field is determined by the ledge
Specimens not loaded to failure10,11; with complicated length and, as in the case of short and long ledges, the
support conditions, geometry, or reinforcement details12-21; tension field can engage all the “hanger” bars within the
and with lack of information essential for the construction of 45-degree load spread. For the cut-off ledges, the force can
STMs12-18 were eliminated. only spread on one side of the bearing plate, concentrating
The majority of the specimens found in the literature the load in a smaller area and increasing the tensile stresses.
were unrepresentative of the bent caps in service in Texas, Furthermore, by extending the ledge to the entire length of
10-12,16,19,21,22
requiring additional filters. A scaled comparison a beam, the capacity of the support node can be increased.
of the cross sections of the specimens from the literature, the The additional cross-sectional area in a longer ledge length
inverted-T beams tested in the current project, and distressed can provide confinement in the nodal region and increase
in-service bent caps within Texas is presented in Fig. 6. the bearing width at the support as compared to beams with
The specimens are identified by their reference number. short and cut-off ledges.
A notable difference in size exists between the in-service Ledge depth—To fully capture the effect of ledge geom-
bent caps (hatched) and the majority of specimens found in etry, two ledge depths were investigated as shown in Fig. 7.
the literature (solid). A complete discussion of the filtering Shallow ledge specimens had depth equal to one-third and
process is provided in Larson et al.7 deep ledge specimens were one-half the total height of the
In summary, all of the 97 specimens from the 13 sources beam. The ledge depths were chosen to give an adequate
were filtered out due to the reasons stated above, reinforcing range of those seen in practice. As with the ledge length, the
the need for a large number of specimens to evaluate the ledge depth also has an effect on the width over which this
behavior of inverted-T beams and investigate the applica- tension field spreads, to a lesser extent. Deeper ledges allow
bility of STM design provisions. the forces to spread over a wider area, decreasing the tensile
stress in the web.
EXPERIMENTAL INVESTIGATION Reinforcement ratio—Two amounts of orthogonal web
Experimental variables reinforcement were chosen with areas of steel equal to 0.3%
The five variables investigated are as follows: the length and 0.6% of the effective web area, as shown in Fig. 7. In most
of the ledge beyond the bearing of the exterior stringer, the tests, the amount of vertical and horizontal web reinforce-
depth of the ledge, the amount of web reinforcement (trans- ment, ρv and ρh, was equal. Two specimens were designed
verse and longitudinal), the number of point loads (girders) with 0.3% in the horizontal direction (skin reinforcement) and
on the ledge, and the height of the member, as shown in 0.6% in the vertical direction (shear stirrups). The reinforce-

ACI Structural Journal/March-April 2015 151


ment ratio of 0.003 (0.3%) corresponds to No. 4 (No. 13) bars total force to multiple locations and preventing local ledge
at 6.5 in. (165 mm) on center at each face of the beam. Like- failure. The spacing of the three point loads was representa-
wise, a 0.006 (0.6%) ratio corresponds to No. 5 (No. 16) bars tive of in-service girders. They were also used to help quan-
at 5 in. (127 mm) on center. The lower limit of 0.3% is the tify the effect of multiple girders on bridge bent caps. Due
AASHTO LRFD1 minimum required skin reinforcement for to limitations, laboratory testing is typically performed with
deep beams. The upper limit of 0.006 (0.6%) was selected to one loading point, but bent caps in service support multiple
encompass the maximum reinforcement ratio (0.57%) found girders on each side. By comparing beams tested at one and
in the in-service distressed bents. The size and spacing of the three points, the validity of beams tested at a single load
bars provides typical crack control. point could be assessed.
Number of point loads—The beams in this investiga- Web depth—A review of the literature revealed a signifi-
tion were loaded at either one or three points along their cant difference in the size of the in-service bent caps when
length, as shown in Fig. 7. The load at each point was compared to the specimens used to calibrate the shear provi-
equally divided and applied to the ledge on both sides of sions in the current design code.7 Full-scale specimens with
the web using a U-shaped frame. Specimens with multiple web depths of 42 and 75 in. (1067 and 1905 mm) were
point loads allowed for shallower ledges by distributing the constructed and tested for the experimental program to fill
in this gap and validate the STM design provisions for use in
larger inverted-T beams.

Specimen description
A large testing program was required to fully evaluate all
of the experimental variables. The width of the beams was
proportioned to maximize the cross-sectional area of the
specimen, while keeping it narrow enough to ease instal-
lation and removal from the test setup. Typical dimensions
and reinforcement layouts are shown in Fig. 8 and bearing
plate sizes are given in Table 2. Flexural reinforcement
was comprised of 12 No. 11 (No. 36) bars for the 42 in.
(1067 mm) specimens and 22 for the 75 in. (1905 mm)
specimens. Hanger reinforcement was comprised of No. 6
(No. 19) bars and was detailed based on the estimated
ultimate load. Ledge reinforcement was either No. 5 or 6
(No. 16 or 19) bars, depending on the depth of the ledge and
the resulting demand. The web width was kept constant at
21 in. (533 mm) for each beam in the experimental program,
including the 75 in. (1905 mm) beams. The width of the
ledge was also the same, 10.5 in. (267 mm), on each side. All
other dimensions varied. To distinguish between the speci-
mens in Table 2 and their respective variables, the following
Fig. 6—Scaled cross sections of literature specimens with corre- nomenclature was developed
sponding reference number, current specimens, and in-service
bent caps. (Note: Dimensions in inches; 1 in. = 25.4 mm.) Sample specimen designation: DC1-42-1.85-03

Fig. 7—Experimental variables.

152 ACI Structural Journal/March-April 2015


Fig. 8—Typical reinforcement layout and dimensions. (Note: Dimensions in inches; 1 in. = 25.4 mm.)
where the first term refers to the ledge depth, either deep (D) crack, then in 100 kip (445 kN) increments to failure. After
or shallow (S). The second term refers to the ledge length— each load increment, cracks were marked and diagonal crack
cut-off (C), short (S), or long (L). The third term refers to widths were measured and recorded as part of the service-
the number of point loads, either one (1) or three (3). The ability considerations of the experimental program.7
next number is the web depth in terms of inches, 42 or 75 Each beam was designed with two test regions. Speci-
(1067 or 1905 mm, respectively). Next is the a/d, either 1.85 mens with a single point load were loaded a distance from
or 2.5. The final term is the web reinforcement ratio, either one support corresponding to the desired a/d. After a shear
0.3% (03) or 0.6% (06). failure was achieved, the load was removed and post-
The specimens were constructed using conventional mate- tensioning clamps were installed to strengthen the first test
rials and methods. Steel formwork was used to expedite the region. Then the hydraulic ram and U-frame were moved
fabrication process and to ensure dimensional accuracy. and the test procedure was repeated. Specimens with three
Beams were tested approximately 28 days after concrete loading points were designed such that both ends were
placement. Domestic Grade 60 deformed bars satisfying the tested simultaneously and monitored until a shear failure
requirements of ASTM A61523 were used for all steel rein- was achieved at one end of the beam. The load was then
forcement. Cross-sectional dimensions of the bars complied removed, post-tensioned clamps were installed in the failed
with the nominal sizes given in ASTM A615.23 The tensile test region, and the load was reapplied at the same location
strength of the coupons was measured in accordance with until the opposite end of the beam failed in shear, as shown
ASTM A370.24 Material properties, including reinforcement in Fig. 9. Vtest, the maximum shear carried in the critical
and concrete strength, are provided in Table 2. section of the test region, including the self-weight of the
beam and test setup, is provided in Table 2.
Testing procedure
The specimens were tested at the University of Texas at COMPARISON OF STRUT-AND-TIE MODELS AND
Austin’s Phil M. Ferguson Structural Engineering Labora- EXPERIMENTAL RESULTS
tory. The upside-down simply supported beam test setup A summary of the experimental versus calculated shear
used for testing is shown in Fig. 9. The load was applied strengths (Vtest/Vcalc) is provided in Table 2, where Vcalc is the
via a 5 million pound (22,200 kN) capacity, double-acting shear capacity calculated using the measured material proper-
hydraulic ram for single point load tests, and three 2 million ties and the Birrcher et al.4 STMs as implemented for inverted-
pound (9000 kN) capacity rams for multiple point load spec- T beams. As shown in the table, all values of Vtest/Vcalc
imens. U-shaped frames applied load evenly to the ledges of are greater than 1.0, indicating that the STM provisions
the test specimens. At each support, six 3 in. (76 mm) diam- as implemented for inverted-T beams are conservative for
eter threaded rods reacted against a 7000 lb (31 kN) transfer all specimens tested. With the large number of specimens,
girder to resist the applied load. 500 kip (2200 kN) capacity direct comparisons investigated each variable independently
load cells were placed between the transfer beam and the while keeping all others constant as discussed extensively by
reaction nut at each of the 12 rods to measure the applied Larson et al.7 In these direct comparisons, the STMs showed
shear at each throughout the loading history. no bias to ledge depth, number of point loads, beam depth,
Test specimens were monotonically loaded in 50 kip or chord loading; that is, the effects of these variables were
(222 kN) increments to the appearance of the first diagonal adequately captured as no trends were observed. The STMs

ACI Structural Journal/March-April 2015 153


Table 2—Summary of beam details
Reinforcement fy, ksi
Support plate, Load
Specimen in. plate, in. No. 11 No. 6 No. 5 No. 4 fc′, ksi Vcalc, kip Vtest, kip Vtest/Vcalc
DS1-42-1.85-03 16 x 20 26 x 9 69.24 63.38 64.69 63.14 5.27 463 712 1.54
DS1-42-2.50-03 16 x 20 26 x 9 69.24 63.38 64.69 63.14 5.39 202 406 2.01
DS1-42-1.85-06 16 x 20 26 x 9 64.13 63.38 60.68 N/A 5.02 479 621 1.30
DS1-42-2.50-06 16 x 20 26 x 9 64.13 63.38 60.68 N/A 5.09 339 503 1.49
DL1-42-1.85-06 16 x 20 26 x 9 67.90 63.38 64.69 N/A 4.83 464 741 1.60
DL1-42-2.50-06 16 x 20 26 x 9 67.90 63.38 64.69 N/A 4.99 353 622 1.76
SS3-42-1.85-03 16 x 20 18 x 9 68.60 64.68 62.75 67.25 5.89 456 523 1.15
SS3-42-2.50-03 16 x 20 18 x 9 68.60 64.68 62.75 67.25 5.89 215 447 2.08
SS3-42-2.50-06 16 x 20 18 x 9 69.50 61.83 60.90 N/A 6.26 415 516 1.24
SC3-42-2.50-03 16 x 20 18 x 9 66.20 63.50 60.25 64.27 5.87 257 329 1.28
SC3-42-1.85-03 16 x 20 18 x 9 66.20 63.50 60.25 64.27 5.87 427 483 1.13
DS3-42-2.50-03 16 x 20 18 x 9 63.60 62.63 60.22 64.58 5.69 236 430 1.82
DL1-42-1.85-03 16 x 20 26 x 9 71.01 61.90 64.29 64.43 4.93 468 626 1.34
DL1-42-2.50-03 16 x 20 26 x 9 71.01 61.90 64.29 64.43 4.93 235 510 2.17
SL3-42-1.85-03 16 x 20 18 x 9 75.18 60.62 63.58 65.57 5.04 409 571 1.39
SL3-42-1.85-06 16 x 20 18 x 9 70.38 63.26 64.80 62.62 5.25 424 744 1.76
DC1-42-1.85-06 30 x 20 30 x 10 73.30 63.98 60.81 N/A 3.73 428 519 1.21
SS1-75-1.85-03 16 x 20 30 x 10 66.10 61.97 64.69 65.08 3.13 389 745 1.92
DC3-42-1.85-03 16 x 20 18 x 9 63.63 66.00 63.09 63.16 4.57 370 395 1.07
DS3-42-1.85-03 16 x 20 18 x 9 63.63 66.00 63.09 63.16 4.57 370 454 1.23
SS1-42-2.50-03 16 x 20 26 x 9 65.44 69.57 77.76 66.58 5.70 205 398 1.94
SS1-42-1.85-03 16 x 20 26 x 9 65.44 69.57 77.76 66.58 5.72 501 583 1.16
DC1-42-2.50-03 16 x 20 18 x 9 70.06 64.13 69.77 62.44 4.04 259 365 1.46
DL3-42-1.85-03 16 x 20 18 x 9 70.06 64.13 69.77 62.44 4.20 359 629 1.75
SL1-42-2.50-03 16 x 20 26 x 9 68.70 71.41 N/A 64.47 4.28 261 498 1.91
SC1-42-2.50-03 16 x 20 26 x 9 68.70 71.41 N/A 64.47 4.28 259 319 1.24
DS1-42-1.85-06/03 16 x 20 26 x 9 65.80 70.92 64.94 65.18 4.17 416 739 1.78
DS1-42-2.50-06/03 16 x 20 26 x 9 65.80 70.92 64.94 65.18 4.17 362 539 1.49
SC1-42-1.85-03 30 x 20 26 x 9 66.36 64.04 N/A 67.28 4.33 443 451 1.05
DC1-42-1.85-03 30 x 20 26 x 9 66.36 64.04 N/A 67.28 4.30 474 517 1.09
SC1-42-1.85-03 *
30 x 20 30 x 10 70.47 63.12 N/A 68.56 3.01 362 456 1.26
DC1-42-1.85-03* 30 x 20 30 x 10 70.47 63.12 N/A 68.56 3.00 362 424 1.17
SS1-75-2.50-03 16 x 20 26 x 9 65.22 63.85 63.62 63.76 5.16 357 649 1.82
*
Ledge length set equal to load plate length.
Notes: Shaded values indicate failure modes other than web shear; 1 in. = 25.4 mm; 1 ksi = 6.89 MPa; N/A is not available.

did show limited bias to ledge length and reinforcement friction, ledge tie failure, and punching shear occurred. In
ratio, but produced conservative results in all cases with these few cases, the failure mode was related to the second
reasonable safety margins. weakest element in the STM, which changed depending
Overall, the STM procedures offer a more transparent of the experimental variables. Nonetheless, each of the 33
approach to designing inverted-T deep beams than sectional specimens carried loads well above the calculated web shear
design, as they inherently consider all failure modes for the capacity and thus, the strength estimates were conservative.
ledges, web, and bearing points. The web shear failure mode The statistical results for the strength ratios of the 33 test
predicted by the STMs, either crushing of the strut-node specimens in the experimental program are summarized in
interface or yielding of the intermediate tie, was observed in Table 3. As shown in the table, the design method yielded
all specimens except for the five shaded in Table 2, in which conservative and accurate estimates of strength with a
flexure (crushing of the compression stress block), shear

154 ACI Structural Journal/March-April 2015


Table 3—Summary of experimental/calculated
shear capacity
Rectangular deep beams,4
Inverted T-beams, 33 tests 179 tests
Birrcher Birrcher
Vtest/Vcalc et al.4 ACI 3186 et al.4 ACI 3186
Minimum 1.05 1.04 0.73 0.87
Maximum 2.17 2.17 4.14 9.80
Mean 1.50 1.57 1.54 1.80
Unconservative 0.0% 0.0% 0.6% 1.7%
Coefficient of
0.22 0.20 0.28 0.58
variation*
Fig. 9—Test setup: specimen at failure of second test region.
*
Coefficient of variation is standard deviation divided by mean.

minimum Vtest/Vcalc value of 1.05, a maximum of 2.17, and


an average of 1.51 for the inverted-T beams.
A comparison of STM procedures by Birrcher et al.4 and
ACI 318-116 revealed similar levels of accuracy for the
inverted-T beams as shown in Fig. 10. The average Vtest/Vcalc
values in Table 3 were equal to 1.50 and 1.57, respectively,
suggesting that ACI 318-116 is slightly more conservative.
The significant difference between the two STM procedures
is the treatment of the struts and nodes, thus most of the spec-
imens designed to fail due to yielding of the stirrup tie had the
same calculated capacity. However, ACI 318-116 predicted a
diagonal strut failure occurring in the cross-sectional model
before web shear in seven of the 33 specimens. This was due
to the low efficiency factor βs for struts in tension flanges of Fig. 10—Comparison of Vtest/Vcalc for Birrcher et al.4 and
beams, as shown in Table 1, which resulted in an increased ACI 318-116 STMs.
conservatism for several of the specimens shown in Fig. 10. The following variables were investigated to encompass the
The Birrcher et al.4 STMs do not account for this out-of- full behavior of inverted-T beams: the length and depth of
plane tension. If only web shear failure modes were consid- the ledge, the quantity of web reinforcement, the number of
ered using ACI 318-11,6 the mean Vtest/Vcalc would decrease point loads, member depth, and the shear span-depth ratio.
to 1.50. With these results, previously proposed strut-and-tie provi-
In comparison with the rectangular deep beams, it can be sions were assessed for their applicability to inverted-T
concluded that the Birrcher et al.4 STM provisions provided beams. Both ACI 318-116 STM and Birrcher et al.4 design
equal, if not slightly better, predictions of shear strength. provisions yield accurate and reasonably conservative results
The overall mean was similar with 1.50 for the inverted for tension chord-loaded beams. The following conclusions
T-beams and 1.54 for the rectangular bent caps evaluated summarize the views of the authors:
with the Birrcher et al.4 STMs. The scatter in the results was • Use of STM is recommended for the design of
also decreased for the inverted-T beams when the minimum inverted-T beams. A comparison between the ulti-
and maximum Vtest/Vcalc values were compared. The stan- mate shear capacity obtained from the test results and
dard deviation and coefficient of variation also decreased. the nominal shear capacity from the STM calculations4,6
Furthermore, no unconservative prediction of strength was revealed conservative strength estimates for every spec-
noted for the inverted-T beams, while a small number of imen. Furthermore, the Birrcher et al.4 STMs accurately
rectangular deep beams had Vcalc values greater than Vtest. predicted the web shear failure mode for 28 of the 33 spec-
The comparison of the rectangular deep beams as eval- imens. For the five that did not fail in shear, the calculated
uated using ACI 318-116 STMs is also provided in Table 3 shear capacity was exceeded and the actual failure mode
to demonstrate its effectiveness. A significant improvement was the second weakest element in the model. Within
was observed for minimum, maximum, and average Vtest/Vcalc these provisions, a minimum web reinforcement ratio is
values. It was for this reason that Birrcher et al.4 STMs were given as 0.3% in each orthogonal direction and is also
investigated for inverted-T beams. recommended for inverted-T beam design.7
• Valid assumptions were made in implementing the
SUMMARY AND CONCLUSIONS STM provisions for inverted-T beams. The geom-
In this investigation, the behavior of inverted-T beams etry of inverted-T beams requires the use of a three-
was studied through a comprehensive experimental program dimensional STM model or two equivalent and compat-
composed of 33 tests on 22 large-scale beams. Most of the ible two-dimensional models. Recommendations were
beams were designed with two test regions, one on each end.

ACI Structural Journal/March-April 2015 155


given to aid in developing these models. A 45-degree load 4. Birrcher, D.; Tuchscherer, R.; Huizinga, M.; Bayrak, O.; Wood, S.;
and Jirsa, J., “Strength and Serviceability Design of Reinforced Concrete
spread at each load point satisfactorily models the hanger Deep Beams,” Report No. 0-5253-1, Center for Transportation Research,
and ledge reinforcement that engaged during loading. the University of Texas at Austin, Austin, TX, 2009, 400 pp.
The Birrcher et al.4 STM provisions provided accurate predic- 5. Tuchscherer, R.; Birrcher, D.; and Bayrak, O., “Experimental Exam-
ination of ACI 318 Strut and Tie Modeling Provisions,” Symposium
tions of failure mode and capacity for the inverted-T beams Honoring James O. Jirsa’s Contributions in Structural Concrete: A Time to
but further investigation is recommended for struts in tension Reflect, SP-296, J. A. Pincheira and S. M. Alcocer, eds., American Concrete
members and tension flanges in other structural members. Institute, Farmington Hills, MI, 2014, 20 pp.
6. ACI Committee 318, “Building Code Requirements for Structural
Concrete (ACI 318-11) and Commentary,” American Concrete Institute,
AUTHOR BIOS Farmington Hills, MI, 2011, 503 pp.
ACI member Nancy Larson Varney is a Staff II Structural Engineer with 7. Larson, N.; Fernández-Gómez, E.; Garber, D.; Bayrak, O.; and Ghan-
Simpson Gumpertz & Heger, Inc. She received her BS from Lehigh Univer- noum, W., “Strength and Serviceability Design of Reinforced Concrete
sity, Bethlehem, PA, in 2008, and her MS and PhD from the University of Inverted T-Beams,” Report No. 0-6416, Center for Transportation Research,
Texas at Austin, Austin, TX, in 2010 and 2013, respectively. Her research University of Texas at Austin, Austin, TX, 2012, 234 pp.
interests include strut-and-tie modeling of reinforced concrete. 8. Wight, J. K., and Parra-Montesinos, G., “Use of Strut-and-Tie Model
for Deep Beam Design as per ACI 318 Code,” Concrete International,
ACI member Eulalio Fernández-Gómez is a structural engineer at V. 25, No. 5, May 2003, pp. 63-70.
Osseous Structural
Engineering, Ciudad Juárez, Mexico. He received his 9. Williams, C.; Deschenes, D.; and Bayrak, O., “Strut-and-Tie Model
BS from Universidad Autónoma de Chihuahua, Chihuahua, Mexico, in Design Examples for Bridges,” Report No. 5-5253-01-1, Center for Trans-
2004, and his MS and PhD from the University of Texas at Austin in 2009 portation Research, University of Texas at Austin, Austin, TX, 2012, 276
and 2012, respectively. pp.
10. Furlong, R. W.; Ferguson, P. M.; and Ma, J. S., “Shear and Anchorage
ACI member David B. Garber is an Assistant Professor at Florida Inter- Study of Reinforcement in Inverted T-Beam Bent Cap Girders,” Report No.
national University, Miami, FL. He received his BS from Johns Hopkins 113-4, Center for Highway Research, University of Texas at Austin, Austin,
University, Baltimore, MD, in 2009, and his MS and PhD from the Univer- TX, 1971, 81 pp.
sity of Texas at Austin in 2011 and 2014, respectively. His research inter- 11. Cussens, A. R., and Besser, I. I., “Shear Strength of Reinforced
ests include plasticity in structural concrete and behavior of prestressed Concrete Wall-Beams under Combined Top and Bottom Loads,” The Struc-
concrete members. tural Engineer, V. 63, No. 15, Sept. 1985, pp. 50-56.
12. Graf, O.; Brenner, E.; and Bay, H., “Versuche mit einem wandartigen
Wassim M. Ghannoum is an Assistant Professor in the Department of Trager aus Stahlbeton,” Deutscher Ausschuss fur Stahlbeton, V. 99, 1943,
Civil, Environmental, and Architectural Engineering at the University of pp. 41-54.
Texas at Austin. He is Chair of ACI Committee 369, Seismic Repair and 13. Ferguson, P. M., “Some Implications of Recent Diagonal Tension
Rehabilitation, and a member of ACI Subcommittee 318-R, High-Strength Tests,” ACI Journal Proceedings, V. 53, No. 8, Aug. 1956, pp. 157-172.
Reinforcement (Structural Concrete Building Code), and Joint ACI-ASCE 14. Schütt, H., “Über das Tragvermögen wandartiger Stahlbetonträger,”
Committees 441, Reinforced Concrete Columns, and 447, Finite Element Beton und Stahlbetonbau, V. 10, Oct. 1956, pp. 220-224.
Analysis of Reinforced Concrete Structures. 15. Taylor, R., “Some Shear Tests on Reinforced Concrete Beams
without Shear Reinforcement,” Magazine of Concrete Research, V. 12, No.
Oguzhan Bayrak, FACI, is a Professor in the Department of Civil, Envi- 36, 1960, pp. 145-154. doi: 10.1680/macr.1960.12.36.145
ronmental, and Architectural Engineering and holds the Charles Elmer 16. Furlong, R. W., and Mirza, S. A., “Strength and Serviceability of
Rowe Fellowship in Engineering at the University of Texas at Austin, where Inverted T-Beam Bent Caps Subject to Combined Flexure, Shear, and
he serves as Director of the Phil M. Ferguson Structural Engineering Torsion,” Report No. 153-1F, Center for Highway Research, University of
Laboratory. He is a member of ACI Committees 341, Earthquake-Resistant Texas at Austin, Austin, TX, 1974, 156 pp.
Concrete Bridges, and S803, Faculty Network Coordinating Committee; 17. Smith, K. N., and Fereig, S. M., “Effect of Loading and Supporting
and Joint ACI-ASCE Committees 441, Reinforced Concrete Columns, and Condidtions on the Shear Strength of Deep Beams,” Shear in Reinforced
445, Shear and Torsion. Concrete, SP-42, American Concrete Institute, Farmington Hills, MI, 1974,
pp. 441-460.
18. Fereig, S. M., and Smith, K. N., “Indirect Loading on Beams with
ACKNOWLEDGMENTS Short Shear Spans,” ACI Journal Proceedings, V. 74, No. 5, May 1977,
The authors wish to thank the Texas Department of Transportation for
pp. 220-222.
providing the financial support for this investigation, and the contribu-
19. Leonhardt, F., and Walther, R., “Wandartige Träger,” Deutscher
tions of Project Director J. Farris and TxDOT Project Advisors including
Ausschuss für Stahlbeton, V. 178, 1966.
C. Holle, D. Van Landuyt, G. Yowell, M. Stroope, N. Nemec, and R. Lopez.
20. Galal, K., and Sekar, M., “Rehabilitation of RC Inverted-T Girders
The contribution of the students and the staff at the Ferguson Structural
Using Anchored CFRP Sheets,” Composites. Part B, Engineering, V. 39,
Engineering Laboratory is also greatly appreciated. Opinions, findings,
No. 4, 2008, pp. 604-617. doi: 10.1016/j.compositesb.2007.09.001
conclusions, and recommendations in this paper are those of the authors.
21. Zhu, R. R.-H.; Dhonde, H.; and Hsu, T. T., “Crack Control for Ledges
in Inverted ‘T’ Bent Caps,” TxDOT Project 0-1854, University of Houston,
REFERENCES Houston, TX, 2003, 4 pp.
1. AASHTO LRFD, Bridge Design Specifications, American Associa- 22. Tan, K. H.; Kong, F. K.; and Weng, L. W., “High Strength Concrete
tion of State Highway and Transportation Officials, Washington, DC, 2012, Deep Beams Subjected to Combined Top- and Bottom-Loading,” The
1960 pp. Structural Engineer, V. 75, No. 11, 1997, pp. 191-197.
2. Joint ACI-ASCE Committee 445, “Recent Approaches to Shear 23. ASTM A615/A615M-08, “Standard Specification for Deformed and
Design of Structural Concrete (ACI 445R-99),” American Concrete Insti- Plain Carbon-Steel Bars for Concrete Reinforcement,” ASTM Interna-
tute, Farmington Hills, MI, 1999, 56 pp. tional, West Conshohocken, PA, 2008, 5 pp.
3. Schlaich, J.; Schäfer, K.; and Jennewein, M., “Toward a Consis- 24. ASTM A370-08a, “Standard Test Methods for Mechanical Testing of
tent Design of Structural Concrete,” PCI Journal, V. 32, No. 3, 1987, Steel Products,” ASTM International, West Conshohocken, PA, 2008, 47 pp.
pp. 74-150. doi: 10.15554/pcij.05011987.74.150

156 ACI Structural Journal/March-April 2015

You might also like