You are on page 1of 12

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 106-S63

Evaluation of Load Transfer and Strut Strength of Deep


Beams with Short Longitudinal Bar Anchorages
by Sergio F. Breña and Nathan C. Roy

This paper presents laboratory results of 12 deep beams in which


the longitudinal reinforcement was anchored into the support
using short straight bar anchorages. The shortest anchorage
lengths provided were below 50% of those required by ACI 318-08
Chapter 12 provisions. Four different specimen groups were
constructed using three different shear span-depth ratios (a/d) and
two longitudinal bar sizes. Most of the beams failed by strut
crushing after yielding of the main longitudinal reinforcement at
midspan. Only those specimens with the shortest anchorage length
in each group developed concrete splitting failures along the
anchorage region. The effect of a/d and anchorage length on strut
strength and load transfer mechanism observed in the tests is
presented and discussed. Test results indicate that a significant Fig. 1—Strut-and-tie models consistent with load transfer
portion of the applied shear force may be transferred through truss mechanisms in deep beams.
action even in beams with low a/d. In addition to well-known variables
such as a/d and transverse reinforcement content, short anchorage
length affected the load transfer characteristics of the deep beams of each node over each support. Anchoring these horizontal
tested in this research. bars, which often have large diameter and are placed in
multiple layers to satisfy the tie force requirement, becomes
Keywords: anchorage length; deep beams; strength; strut-and-tie models. very difficult and results in reinforcing bar congestion near
the ends of the beam. To anchor the bottom tie according to
INTRODUCTION code procedures (for example, ACI 318), hooked bars or tee-
Design of deep beams using strut-and-tie models involves headed reinforcement are almost always used.
the selection of an adequate model that captures the anticipated The anchorage condition over the support of deep beams
behavior with reasonable accuracy. Model selection is typically might benefit from the presence of transverse confining
based on knowledge about the load transfer mechanisms stresses that generate from the reaction and diagonal strut
present in deep beams. It is now commonly accepted that that meet at the end node. The beneficial effect of transverse
deep beams transfer shear through one of two basic load normal stresses on the anchorage condition of deep beams
transfer mechanisms identified as either a tied-arch mechanism was examined and discussed in detail by Roy and Breña
or truss mechanism (Fig. 1). The two most important factors (2008). The test results indicated that beams with straight bar
affecting the type of load transfer mechanism are the shear anchorages of only approximately 50% of the length
span-depth ratio (a/d) of the beams and the amount of trans- required by provisions in Chapter 12 of ACI 318-08 (ACI
verse reinforcement. There seems to be general agreement that Committee 318 2008) were able to develop peak loads
as a/d increases, a higher fraction of load is transferred comparable with beams where longer anchorages were
through truss action (Fig. 1(b)). Conversely, beams with provided. In some of these beams with a/d as low as 1.0 or
lower a/d transfer a larger portion of the load through tied- 1.5, limited or no yielding was measured in longitudinal
arch action (Fig. 1(a)). Beams with intermediate a/d transfer reinforcement at the face of the support node indicating that
load through a combination of these two mechanisms, but load transfer by tied-arch action was not predominant even
there is disagreement on how to estimate the fraction transferred in beams with this low a/d. Furthermore, reinforcing bar
by each assumed model. Similarly, larger amounts of transverse yielding at midspan was observed in many of these beams and
reinforcement lead to a higher fraction of load transferred by maintained to load levels that caused crushing of the main struts
truss action. A minimum amount of web reinforcement (vertical forming after cracking of the beams. These results indicate
and horizontal) is required to prevent splitting of diagonal struts that an alternate load path was developed in the beams with
forming between load and supports of deep beams. short anchorages (truss action), which allowed development
Load transfer through tied-arch action has important of higher load without placing the high anchorage demand at
implications in design. As can be observed by equilibrium of the support node that is generated from tied-arch action.
the simplified model shown in Fig. 1(a) depicting a tied arch The purpose of this paper is to examine and discuss the
model, the horizontal tensile force in the tie at the bottom of behavior of struts and nodes in these deep beams fabricated
the model is required to maintain its maximum value
throughout the span of the beam. This assumed behavior ACI Structural Journal, V. 106, No. 5, September-October 2009.
MS No. S-2008-205 received June 24, 2008, and reviewed under Institute publication
places a high anchorage demand at nodes located above the policies. Copyright © 2009, American Concrete Institute. All rights reserved, including the
supports, because if one designs the horizontal tie to yield at making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including author’s closure, if any, will be published in the July-
ultimate load, then yielding has to be maintained at the face August 2010 ACI Structural Journal if the discussion is received by March 1, 2010.

678 ACI Structural Journal/September-October 2009


sectional area throughout their length. For bottle-shaped
Sergio F. Breña, FACI, is an Associate Professor in the Department of Civil and
Environmental Engineering at the University of Massachusetts Amherst, Amherst, struts, βs equals 0.75 if the strut is crossed by the minimum
MA. He is Secretary of ACI Committee 369, Seismic Repair and Rehabilitation, and a amount of reinforcement specified in ACI 318-08 section
member of the ACI Publications Committee; ACI Committees 374, Performance- A.3.3, or 0.60λ (λ is a lightweight aggregate concrete factor)
Based Seismic Design of Concrete Buildings, and 440, Fiber Reinforced Polymer
Reinforcement; Joint ACI-ASCE Committee 445, Shear and Torsion; and E803, if the reinforcement does not comply with ACI 318-08,
Faculty Network Coordinating Committee. Section A.3.3, or no reinforcement is present.
Nathan C. Roy is a Structural Engineer with LeMessurier Consultants, Cambridge,
Similarly, the nominal strength of nodal zones (Fnn) in
MA. He received his BS and MS in civil engineering from the University of Massachusetts ACI 318-08 is calculated using
Amherst in 2004 and 2006, respectively.

Fnn = fceAnz (3)


using short longitudinal bar anchorages. The strength of
struts is investigated based on the results of the tests. A critical where Anz represents the area of the node face perpendicular
evaluation of the apparent load transfer mechanisms in this type to the applied force on the node that results in the highest
of beams was conducted to provide insight into the effect of stress on the nodal zone. In the case of nodes, fce is calculated
short bar anchorages on how these beams transfer shear. using Eq. (2) with βn instead of βs. The factor βn depends on
the type of node, and it equals 1.0 for nodal zones with only
RESEARCH SIGNIFICANCE compression forces acting on them, 0.80 in nodal zones
Design of deep beams is commonly done using strut-and- where one tie is anchored, or 0.60 for nodal zones anchoring
tie models following Appendix A in ACI 318-08. Engineers two or more ties. These nominal strengths computed by Eq. (1)
are given freedom to select the strut-and-tie model for design and (3) must be multiplied by a strength reduction factor of
with little guidance as to the merits of each model. The 0.75 to calculate the design strength according to ACI 318.
choice of model is largely based on knowledge of load
transfer mechanisms for beams with different a/d. Deep COMMONLY ACCEPTED FORCE TRANSFER
beams with a/d < 1.0 are commonly assumed to transfer the MECHANISMS FOR DEEP BEAMS
total load through tied-arch action, whereas the most widely Rather than relying on empirical procedures as was done
accepted load transfer mechanism in slender beams with a/d ≥ in the past, strut-and-tie models are now used extensively in
4.0 is based on truss action. Loads in deep beams with inter- the design of deep beams. These models are commonly
mediate a/d (between 1.0 and 4.0) are transferred through a constructed to approximately follow the force transfer path
combination of these two mechanisms. This paper will anticipated in deep beams. The two most commonly
provide data about load transfer mechanisms in deep beams accepted strut-and-tie models used in deep beam design are
with short longitudinal bar anchorages. The information tied arch or truss models depending on the a/d of the beam.
provided in this paper will assist in implementation of strut- A tied arch model is predominantly used in design of short
and-tie models and an improved understanding of the merits beams (a/d equal to 1.0 or less), with the assumption that
of different models when used in design. load is transferred directly from load point into support
through the formation of concrete struts (direct strut mechanism).
The horizontal component of each strut at the support is set in
ACI 318-08 STRUT AND NODE
STRENGTH EQUATIONS equilibrium by a horizontal tie extending the full length of
Design provisions for strut-and-tie models are contained the beam (Fig. 1(a)). The tie force in this model is constant
in Appendix A of ACI 318-08. The strength of elements throughout the span, so the longitudinal reinforcement
designed using strut-and-tie models, according to these forming the tie has to be anchored at the face of the node over
design provisions, is governed by the strength of the weakest each support to develop the yield stress fy. The critical
element of the strut-and-tie model (strut, tie, or node). Forces section for development of reinforcement in ACI 318-08 is
acting in struts, ties, or nodes must be equated to the resistance defined as the section where the centroid of the reinforcement
of each element to solve for the load carrying capacity of the leaves the extended nodal zone.
system. The focus in this paper will be those models For longer deep beams (a/d of 4.0 or above), load transfer
governed by either strut or node strength. It will be assumed occurs indirectly from load point into support through two or
that tie strength does not govern the capacity of the chosen more struts that form between diagonal cracks within the
strut-and-tie model. The nominal strength of struts in a strut- shear span of the beam (Fig. 1(b)). The vertical component
and-tie model built according to ACI 318-08 is determined from of struts that reach the bottom longitudinal reinforcement
within the shear span is set in equilibrium by vertical tie
Fns = fce Acs (1) forces generated in stirrups that enclose the longitudinal
reinforcement and are well anchored in the compression
zone of the beams. The bottom longitudinal reinforcement is
fce = 0.85βs fc′ (2) again required to equilibrate the horizontal component of
strut forces at the bottom nodes, and compression stresses
where Acs is the minimum cross sectional area (strut width w near the top of the beam equilibrate horizontal strut force
times element thickness b) at one of the two ends of the strut; components at top nodes. The force transfer mechanism in
fce is equal to the effective concrete strut strength; and βs is this type of beams closely resembles forces in a truss, with
an effective concrete strength factor dependent on strut geometry top and bottom chords carrying compressive and tensile
and transverse strain conditions. The reduction in uniaxial forces, respectively, diagonal web members carrying
concrete compressive strength, when subjected to off-axis compressive forces, and vertical web members carrying
tensile strains, is well known. In ACI 318-08, this decrease tensile forces.
in strut strength is captured through the use of βs. The factor In beams of intermediate length (a/d between 1.0 and 4.0),
βs equals 1.0 for prismatic struts that have a constant cross- a combination of force transfer mechanisms is commonly

ACI Structural Journal/September-October 2009 679


Table 1—As-built properties of specimens and observed failure parameters
Group a/d Specimen fc′ , MPa (psi) Beam width b, mm (in.) Peak load Pu, kN (kip) Failure mode Pu/Py* Pu/Pcr-d†
DB1.0-1.00 33.3 (4830) 165 (6.5) 677 (152) Strut – S 1.43 2.17
DB1.0-0.75 31.7 (4600) 173 (6.8) 743 (167) Strut – H 1.55 2.10
1.0
DB1.0-0.50 30.6 (4440) 157 (6.2) 729 (164) Strut – H 1.53 2.34
DB1.0-0.32 27.0 (3915) 152 (6.0) 667 (150) Strut – S 1.43 2.14
DB1.0-0.75L 29.9 (4340) 155 (6.1) 741 (167) Strut – H 1.15 2.39
1.0L
DB1.0-0.28L 29.4 (4265) 155 (6.1) 642 (144) Anchorage ‡ 2.06
DB1.5-0.75 32.7 (4745) 152 (6.0) 459 (103) Strut – S 1.49 1.72
1.5 DB1.5-0.50 34.1 (4945) 152 (6.0) 423 (95) Strut – H 1.44 1.58
DB1.5-0.38 33.8 (4900) 152 (6.0) 427 (96) Anchorage 1.45 2.40
DB2.0-0.75 34.7 (5035) 155 (6.1) 313 (70) Strut – H 1.56 2.33
2.0 DB2.0-0.50 33.0 (4790) 155 (6.1) 297 (67) Strut – H 1.40 1.68
DB2.0-0.43 35.6 (5165) 155 (6.1) 266 (60) Anchorage 1.28 1.50
*
Ratio of maximum load to yield load.
†Ratio of maximum load to diagonal cracking load.

Yielding not observed.
Note: S is strut crushing observed on test end (straight anchorage) of specimen; and H is strut crushing observed on far end (hooked anchorage) of specimen.

assumed. Guidance on the fraction of force transferred by Specimens are designated according to a/d and anchorage
either truss or tied arch action is sparse. Federation International length of straight bars on the test side of the beams. The first
de La Precontrainte Recommendations (FIP 1999) indicate that two digits in the beam designation correspond to a/d (1.0,
the force transfer in beams with a/d between 0.5 and 2.0 1.0L, 1.5, or 2.0), and the last three digits correspond to the
occurs through a combination of truss and tied arch action. ratio of provided-to-design anchorage length on the test side
Force transfer in beams transitions from entirely tied arch of the specimens. The design anchorage length is defined as
action to entirely truss action as a/d increases from 0.5 to 2.0 the length calculated using Chapter 12 of ACI 318-08. This
using the following relationship ratio varied in each group of beams as listed in the specimen
designation in Table 1, but all groups had beams with 75 and 50%
of the anchorage length required by ACI 318-08, Chapter 12.
F 1 = --- ⎛ ------ – 1⎞ F
1 2a
(4) Specimen nominal dimensions and reinforcing patterns
3⎝ z ⎠
are shown in Fig. 2. All beams had a nominal width of 152 mm
(6 in.) and total depths of 635, 457, or 356 mm (25, 18, or 14 in.)
where F represents the total force being transferred, F1 is the for beams with a/d of 1.0, 1.5, or 2.0, respectively. Longitudinal
force transferred through truss action, and z is the internal reinforcement consisted of No. 5 or No. 6 main bottom bars and
lever arm (distance between top and bottom chords of the No. 3 top bars needed for reinforcing cage fabrication. In
model). Using strain gauge data obtained from instrumented specimens in Group 1.0L, the longitudinal bar size used was
stirrups in tests of deep beams with a/z = 1.49, Uribe and No. 6, whereas No. 5 bars were used in the other three
Alcocer (2001) found that the direct strut mechanism groups. Vertical stirrups were fabricated using deformed
contributed slightly more than implied by the FIP (1999) wire (D4) with an actual yield stress of 605 MPa (88 ksi).
recommendations although the implied failure mode was Horizontal bars made from D4 wire were placed at 152 mm
consistent with test observations. (6 in.) spaces in the beam web, close to each of the lateral
The models mentioned above only account for the contribution beam faces. Web reinforcement complied with section A.3.3
of horizontal reinforcement near the bottom of the beam. in ACI 318-08.
Vertical web reinforcement is only explicitly included in Beams were designed using a nominal compressive
truss models but not in tied arch models. Horizontal web strength of concrete equal to 28 MPa (4 ksi) and a reinforcing
reinforcement is not included in either model and its effect is steel nominal yield stress fy equal to 414 MPa (60 ksi). Just
only considered when determining the strength of struts. before testing each beam, specimen dimensions were verified
and companion concrete cylinders were tested to determine the
DESCRIPTION OF LABORATORY TESTS as-built geometry of the beams and the actual strength of
Tests reported in this paper were conducted for simply concrete. Due to formwork flexibility, the actual width of the
supported deep beams subjected to a single concentrated beams varied slightly from the nominal value of 152 mm (6 in.).
load at midspan. Laboratory specimens consisted of 12 deep The measured concrete compressive strength along with the as-
beams divided into four different groups according to three built width of each specimen at the time of testing is listed in
different a/d (1.0, 1.5, or 2.0) and two sizes of the main Table 1. Further details of the beam reinforcement configuration
longitudinal bars (No. 5 or No. 6). The main variable in each and provided anchorage length on the test side of the beams are
group of beams was the anchorage condition of longitudinal presented in a previous paper (Roy and Breña 2008).
bars over one of the supports. On the test side, the longitudinal
reinforcement was continued on different distances past the Test setup and instrumentation
support node in the various specimens (straight bar The span in all beams was equal to 1.22 m (48 in.).
anchorage). Longitudinal reinforcement on the far side of the Specimens were subjected to a single concentrated force at
beams was anchored past the support using a standard midspan. Thick steel plates (25 mm [1 in.]) were placed below
90-degree hook to preclude anchorage failures there. the loading point and above reaction points to avoid localized

680 ACI Structural Journal/September-October 2009


Fig. 2—Specimen geometry, reinforcement, and experimental test setup.

crushing at the nodal zone. The beam supports consisted of a and support on the test side of the beams. In Groups 1.5 and
pin (far end of beam) and a roller (test end of beam) in all tests. 2.0, two possible strut orientations were monitored by placing
A 445 kN (100 kip) load cell was placed underneath each instruments that followed a direct path (tied-arch model) or an
support to measure reactions throughout the tests. indirect path (truss model) from loading point to supports.
External and internal instrumentation was placed at Figure 3 illustrates the potentiometer placement for the four
selected locations in the specimens to relate the measured specimen groups tested in this research.
response with parameters from common strut-and-tie models The stress condition in the extended nodal zone on the test
used for design of deep beams. To investigate the effect of side of the beams was of interest because of the short
the short reinforcing bar anchorage on tie stresses, reinforcing longitudinal bar anchorage used in the design of the specimens.
bars were instrumented internally using strain gauges. The Properly anchored straight bars extending past the back face of
results of these measurements were presented and discussed the nodal zone are assumed to generate a uniform stress
in detail in a separate paper (Roy and Breña 2008). The distribution on the vertical node face. These stresses are
instrumentation that was used to determine the behavior of essential to preserve node equilibrium in strut-and-tie
struts in the four specimen groups is described in detail in this models and avoid node failure. Strains in the nodal zone on
section. the test side of the specimens were measured using surface
During testing, linear potentiometers were attached on both strain gauges bonded to the surface of the concrete in a 0-45-90
beam faces to threaded rods embedded into the concrete degree rosette pattern. The instrument placement within the
before beam casting and defined the control points to calculate extended nodal zone is illustrated in Fig. 3(d).
axial strains developed along struts. The rod placement was Strain gauges were also bonded to reinforcing bars at
selected to measure concrete deformation along the direction midspan and within the extended nodal zone over the
of main struts in selected strut-and-tie models of the different supports. These gauges were used to evaluate the development
specimen groups depending on beam a/d. For Specimen of yield stresses, primarily within the extended nodal zone
Groups 1.0 and 1.0L, the potentiometers measured axial on the test side of the specimens as an indicator of the load
deformation of a direct strut forming between loading point transfer mechanism being developed in the beams. Details of

ACI Structural Journal/September-October 2009 681


Observed failure mode
Failure of all specimens, except those with the shortest
anchorage length, was characterized by strut crushing either
on the test side or on the far side of the beams independent
of provided anchorage length. Strut crushing was typically
observed at the boundary between the top end of the strut and
the nodal zone under the loading point (refer to Fig. 4). Only
specimens DB1.0-0.28L, DB1.5-0.38, and DB2.0-0.43
exhibited an anchorage failure on the test side of the beams
prior to strut crushing. Strains and stresses in struts on the
test side of the beams were able to be examined in detail for
all other specimens that failed by strut crushing. Detailed
cracking maps for all specimens are presented and discussed
in detail by Roy and Breña (2008).
Figure 4 illustrates typical cracking and strut crushing in
three different specimens from groups representing different
a/d ratios. Other specimens failing by strut crushing in each
group exhibited similar cracking and crushing characteris-
tics. As shown in Fig. 4(a) and (b), the inclination of diagonal
cracks and strut crushing in beams from Specimen Groups 1.0
and 1.5 approximately followed the direction of a direct strut
between the load point and one of the supports. A compar-
ison of Fig. 4(c) and (d), which show the front and back face
of a specimen in Group 2.0, illustrates how the inclination of
diagonal cracks and strut crushing roughly followed the
direction of an indirect (truss) load-transfer mechanism that
formed between loading point and support. This behavior is
consistent with the recognized influence of a/d on behavior
of deep or slender beams (MacGregor and Wight 2005).

Comparison of strains in nodal zone with strains


measured in direct strut
Compressive strains were measured in all specimen
groups along lines that approximately coincided with the
expected direction that struts would form if loads were transferred
through tied-arch action. In Specimen Groups 1.5 and 2.0,
compressive strains were also measured along the direction
of struts in a strut-and-tie model representing transfer of
shear through truss action. Additionally, strains were
measured within the extended nodal zone on the test side of
Fig. 3—Specimen instrumentation used in struts and nodes.
the specimens using surface strain gauges bonded at an angle
of 45 degrees with the horizontal in all specimens. Strains
reinforcing bar gauge placement and a description of the measured in the struts and nodes are compared for the three
measured strains at these locations is provided by Roy and specimens with longest anchorage lengths in Group 1.0 in
Breña (2008). Fig. 5. In this group, the direct strut forming between load
point and support was inclined at approximately 45 degrees
TEST RESULTS so the readings on the two elements of the strut-and-tie tied-
All specimens failed at higher loads than anticipated by the arch model (strut and node) could be compared directly.
strut-and-tie models used for design. Even specimens with the Several distinct features can be observed from the load-
shortest longitudinal bar anchorages in each group (DB1.0-0.32, strain plots in Fig. 5. The strain measured in different
DB1.0L-0.28, DB1.5-0.38, and DB2.0-0.43) were capable of instruments at lower loads is approximately equal in
reaching a peak load approaching the maximum load Specimens DB1.0-1.00 and DB1.0-0.75, but quite different
measured in specimens within their corresponding group and in Specimen DB1.0-0.50. Strains at lower loads, below a
exhibited yielding of the main reinforcement and moderate load approximately corresponding to first diagonal cracking,
ductility before failure. Ultimate load was between 1.15 and seemed to be increasing linearly. This load level is indicated
1.56 times higher than the load corresponding to yielding of by an oval or circle in the figures. After this point, the readings
longitudinal reinforcement at midspan, indicating the in different instruments departed significantly with strains
development of an alternate load transfer mechanism. In measured in Instruments L2 or L3, positioned within the strut
specimens where anchorage failure was observed on the test width registering the largest value and the strain measured
end of the beams, the ratio between ultimate and yield load within the extended nodal zone being one of the lowest. The
was similar to or lower than other specimens in the same difference in maximum compressive strain at peak load
group (Table 1). The loads measured at flexural and diagonal depended largely on whether crushing and diagonal cracks
cracking, yielding, and failure of the specimens are listed in formed within the strut width on the test side of the beams or
the companion paper (Roy and Breña 2008). not. In Specimen DB1.0-1.00, crushing was observed at the

682 ACI Structural Journal/September-October 2009


Fig. 5—Measured strain in direct strut and nodal zone
strain at 45 degrees.

top of the diagonal strut on the test side and explains the
large compressive strains registered in this specimen. Strut
crushing occurred on the far side of Specimens DB1.0-0.75
and DB1.0-0.50, explaining the lower compressive strains
recorded at failure compared with Specimen DB1.0-1.00. In
Specimen DB1.0-0.75, diagonal cracks did not form within
the estimated width of the direct strut, whereas cracks
formed at approximately the same inclination as the strut in
Specimen DB1.0-0.50. The presence of diagonal cracks
within the strut width may have attributed to a stiffness
reduction of the strut, leading to higher compressive strains.
Fig. 4—Selected pictures illustrating strut crushing in Instruments L2 and L3 in Specimen Groups 1.0 and 1.0L
specimens from different groups. were thought to be the most representative of observed strut

ACI Structural Journal/September-October 2009 683


of anchorage length was observed in strain development for
specimens in these groups. Also, due to the larger a/d in
these specimen groups, it is conceivable that load transfer
would occur predominantly by truss action so anchorage
differences would not affect development of direct strut
strains as much. It is interesting to note that strut strains in
Specimen DB2.0-0.43, the specimen with the shortest
anchorage length of this group, became tensile as load
increased, perhaps due to the critical diagonal crack
widening, excessively creating a vertical offset between the
monitoring points for instrument L1.

EVALUATION OF STRUT STRENGTH AND


EFFECTIVE STRENGTH FACTOR
A key parameter in the analysis and design of concrete
elements using strut-and-tie models according to ACI 318-08 is
the definition of a strut effective strength factor used to calculate
strut strength. The ACI strut effective strength factor (βs) is
meant to capture the lower compressive strength of concrete
when subjected to off-axis tensile strains as commonly
occurs in webs of beams. The instrumentation used to
monitor strains in struts in the specimens was used with an
assumed strut-and-tie model to determine experimental values
of strut-effective strength factors for the different specimens,
considering that the majority of the specimens tested in this
research failed by strut crushing.
To facilitate calculations to determine the strut strength
from experimental results, the total load in the specimens
was assumed to be transferred through tied-arch action from
the loading point to support in all specimens. This assumption
is not entirely correct because part of the load is also transferred
through truss action, particularly in the more slender specimen
groups. The number and inclination of struts entering the top
node adjacent to the loading point of the specimens is illustrated
in Fig. 7. Only this node was considered critical, because strut
crushing was always observed at or near the top node. A
single strut entering the top node, as opposed to various struts
representing the different load transfer mechanisms, may be
used to represent the resultant of all struts. This resultant would
have a different angle than a strut following a direct path from
load to support because of the presence of other load transfer
mechanisms. The slight difference in angle between the resultant
strut and a direct strut, however, is likely to affect strut strength
calculations minimally (Fig. 7).
Fig. 6—Direct strut strain development in specimens within A tied-arch strut-and-tie model consistent with the peak
different groups. measured load in the tests was then developed for strut
strength evaluation. The geometry of the tied-arch model
behavior, so only these are used to determine the average was established so that node strength was not reached at the
strut strain in subsequent calculations. peak load because no crushing was observed in the nodal
region of the beams. The procedure used followed the one
Effect of anchorage length on strain proposed by Wight and Parra-Montesinos (2003), where the
developed in direct strut strut inclination angle and width are determined iteratively.
The influence of anchorage length on strain developed in From symmetry of the test setup, the total applied force on
the direct strut on the test side of specimens with different a/d the beams resulted in equal shear force transferred to each
was examined by comparing strains developed in struts support. The force in the direct strut forming between load
within specimens of the same group (Fig. 6). The largest and support in the tied-arch model is FS = V/sinα, where α is
differences in strut strains were observed in Group 1.0 specimens the angle of inclination of the strut measured from horizontal
(Fig. 6(a)), with lower compressive strains developing (in (Fig. 8). The angle α depends on the horizontal distance
general) as anchorage length decreased. The only specimen between the resultant of one-half of the applied force and the
in Group 1.0 that did not fit this trend was Specimen DB1.0-0.75, support (560 mm [22 in.]), and the vertical distance between
in which no diagonal cracks formed within the direct strut width the top and bottom chords in the model (z). The top-node
as previously discussed. Differences in strut strains were less height hnode-top was initially assumed equal to 50 mm (2 in.)
pronounced in specimens within Groups 1.5 and 2.0, as to initiate iterations. After determining FS for the initial α,
shown in Fig. 6(b) and (c), so no clear detrimental influence the top horizontal compressive force was calculated as

684 ACI Structural Journal/September-October 2009


Fig. 7—Replacement of forces at top node with single
diagonal resultant.

FC = F S cosα; this force was then equated to the nominal


strength of the top node given by Eq. (2) and (3) with βn = 1.0
to solve for the required top-node height to avoid node crushing

FC F S cos α V
h node-top = --------
- = ------------------- = -------------------------------- (5)
f ce b 0.85f c′ b 0.85f c′ b tan α

The depth between top and bottom chords in the model, z,


was then revised and a new strut angle α was determined to
avoid nodal failure

h node-top Fig. 8—Model and node geometry used for strut strength
z = d – -------------------
- (6) evaluation.
2

Iterations were conducted until hnode-top was approximately Effective concrete strength factor
the same in two subsequent calculation cycles. In all iterations, Strut stresses fstrut determined using the peak measured
the height of the bottom node was assumed equal to 102 mm shear force in the tests was equated to the ACI strut strength
(4 in.), twice the distance between the bottom face of the equation (Eq. (2)) to estimate the strut effective strength
beams and the centroid of the bottom longitudinal reinforcement. factor βs for all specimens. The resulting values for βs ranged
Once convergence was achieved and an appropriate strut between the value corresponding to prismatic and bottle-
inclination was found, the top and bottom widths of the strut shaped struts as defined in ACI 318-08, with higher average
were determined from values obtained in Specimen Groups 1.0 and 1.0L. A plot
showing the decrease of βs with increasing a/d is shown in
wstrut = Lplatesinα + hnodecosα (7) Fig. 9. For comparison, results from deep beams tests
reported by Quintero-Febres et al. (2006) are also shown in
where the plate widths and node heights used for the top and the figure (hollow symbols). The decreasing trend in βs, with
bottom nodes in the previous equation corresponded to those an increase in a/d, is less pronounced in their specimens but
shown in Fig. 8(a). Because the bottom plate is wider than the plot clearly demonstrates an influence of a/d on the
half the top plate, the top end of the strut ended up governing concrete strut strength factor (βs). It is also of interest to note
strut strength in all the specimens. This is consistent with the that the value recommended for prismatic struts in ACI 318
location of observed concrete spalling near the top of the (βs = 1.00) was not reached even for deep beams with the
struts in all specimens that failed by strut crushing (refer to lowest a/d. On the other hand, the value assumed for bottle-
Fig. 4). The stress at the top end of the strut was then shaped struts satisfying web reinforcement requirements of
calculated using ACI A.3.3 (βs = 0.75) was conservative for all ranges of a/d;
that is, higher βs values were obtained in all specimens. Short
FS bar anchorages did not seem to influence βs values significantly
f strut = ----------------
- (8) as observed in values reported in Table 2 for specimens with
w s-top b even the shortest anchorage length within each group.

where the as-built width of the specimens, b, was used and EXPERIMENTAL EVALUATION OF SHEAR
ws-top is the width of the strut determined from Eq. (7) at the TRANSFER MECHANISM
top node. Table 2 summarizes the final strut angle, top-node As mentioned previously, two major force transfer paths
height, and strut stress at the top end for the assumed tied from load point application to support (direct or indirect
arch model. path) commonly occur in deep beams with different a/d. The

ACI Structural Journal/September-October 2009 685


Table 2—Model parameters for strut strength evaluation
Applied shear V, Beam width b, Top node height Strut width (top end)
Specimen kN (kip) mm (in.) Strut angle α, deg hnode-top , mm (in.) ws-top , mm (in.) fstrut, MPa (ksi) βs
DB1.0-1.00 338 (76) 165 (6.5) 44.3 74.3 (2.9) 124 (4.89) 23.6 (3.43) 0.84
DB1.0-0.75 371 (83.5) 173 (6.8) 44.0 82.6 (3.3) 130 (5.12) 23.8 (3.45) 0.88
DB1.0-0.50 365 (82) 157 (6.2) 43.8 92.9 (3.7) 137 (5.41) 24.4 (3.53) 0.94
DB1.0-0.32 334 (75) 152 (6.0) 43.6 100.3 (3.9) 143 (5.62) 22.3 (3.23) 0.97
DB1.0-0.75L 371 (83.5) 155 (6.1) 43.6 99.1 (3.9) 142 (5.59) 24.5 (3.55) 0.97
DB1.0-0.28L 320 (72) 155 (6.1) 44.0 85.8 (3.4) 132 (5.21) 22.5 (3.27) 0.90
DB1.5-0.75 229 (51.5) 152 (6.0) 32.9 83.5 (3.3) 125 (4.93) 22.1 (3.2) 0.79
DB1.5-0.50 211 (47.5) 152 (6.0) 33.3 72.9 (2.9) 117 (4.59) 21.6 (3.14) 0.75
DB1.5-0.38 214 (48) 152 (6.0) 33.3 74.4 (2.9) 118 (4.64) 21.7 (3.14) 0.75
DB2.0-0.75 156 (35) 155 (6.1) 25.5 71.6 (2.8) 108 (4.26) 21.6 (3.13) 0.73
DB2.0-0.50 149 (33.5) 155 (6.1) 25.5 72.1 (2.8) 109 (4.28) 20.6 (2.98) 0.73
DB2.0-0.43 133 (30) 155 (6.1) 26.0 58.3 (2.3) 97 (3.82) 20.2 (2.94) 0.67

direct path is often represented using a tied-arch strut-and-tie


model and the indirect path is often related to a truss model.
An experimental evaluation of the fraction of the total shear
transferred through both load mechanisms was only
conducted for specimens in Groups 1.5 and 2.0 using the
average strain data measured during the tests. Control points
in these specimen groups were set following the approximate
direction of tied-arch (direct) and truss-model struts (indirect)
forming between loading point and support, as shown in
Fig. 3(b) and (c). Instrumentation in Specimen Groups 1.0 and
1.0L was only placed following the direction of a direct strut as
assumed in a tied-arch model (Fig. 3(a)) so no information
could be obtained regarding the shear transferred by truss
action, in these groups, if any. This instrumentation, in
combination with assumed strut-and-tie models, permitted an Fig. 9—Variation of strut factor as a function of shear span-
depth ratio.
estimation of the fraction of shear transferred from load point to
support through each mechanism.
Figure 10 shows average strains in struts measured during where εco = 2fc′ /Ec represents the strain at peak stress (fc′ )
the tests for specimens in Groups 1.5 and 2.0. Average and the modulus of concrete Ec was equal to 4730 f c′ MPa
strains were measured in struts that followed a direction (57,000 f c′ psi). Because Eq. (9) was used to estimate stresses
approximately parallel to that predicted from a truss load from strains measured experimentally, k was equal to 1.0
transfer model (refer to sketches in Fig. 10). In a truss model instead of the typically-assumed value of 0.85. Only the
with the two struts inclined at the same angle as shown, the stress at maximum measured strain was of interest for
forces in diagonal struts are theoretically equal. The test estimation of the fraction of load transferred through
results showed, however, that strains measured along the two truss action.
struts developing in this hypothetical load-path model were Strut stresses determined from strain measurements taken
different. This phenomenon may be due to the presence of on Instruments L2 and L3 (Fig. 10) were used to calculate
diagonal cracks crossing the interior strut where Instruments the fraction of shear transferred through truss action (Vtruss).
L3 were placed, resulting in lower measured compressive Figure 11 illustrates the assumed truss model used, where the
strains compared with those measured in Instrument L2. top node geometry under the point load was assumed, as
Because these observed differences were larger in specimens illustrated in Fig. 11(b). Forces acting at the top node were
determined using Eq. (10) through (12) for an assumed value
in Group 2.0, only the measurements taken in the exterior
of shear transferred through truss action (Vtruss)
strut along L2 were used to estimate stresses at the peak load
measured in the tests in Specimen Group 2.0. The average
strain values measured in Instruments L2 and L3 at peak load V truss
F S-truss = ------------
- (10)
were used to estimate strut stresses in Specimen Group 1.5. sin γ
The uniaxial stress-strain model proposed by Hognestad
(1951) was used to compute stresses from measured average V truss a V truss a
strains. To capture the stress-strain response of the concrete F C ( R ) = ---------------
- = ----------------------------------
- (11)
z d – h node-top ( R )
struts to the point of peak stress, the ascending portion in the
model as described in Eq. (9) was used
FC(L) = FC(R) – Fscosγ (12)
2ε ε 2
f c = kf c′ --------c – ⎛ ------c-⎞ (9) where the shear span a equals 610 mm (24 in.), and the angle
ε co ⎝ ε co⎠ that the strut makes with the horizontal, γ, was determined

686 ACI Structural Journal/September-October 2009


Fig. 11—Truss model for load transfer: (a) geometry of
model; and (b) top node details.

FC (R ) F S-truss cos γ V truss


h node-top ( R ) = ------------ = ----------------------------
- = ------------------------------ (13)
f ce b 0.85f c′ b 0.85f c′ b tan γ

FC (L ) F C ( R ) – F S-truss cos γ
h node-top ( L ) = -----------
- = -----------------------------------------------
- (14)
f ce b 0.85f c′ b

Δh node-top = h node-top ( R ) – h node-top ( L ) (15)


Fig. 10—Average strains measured in struts of strut-and-tie
truss model: (a) Group 1.5 exterior strut; (b) Group 1.5 L plate-top
interior strut; (c) Group 2.0 exterior strut; and (d) Group 2.0 - sin γ + Δh node-top cos γ
w S-truss = -------------------- (16)
interior strut. 2

Notice that the node heights calculated in Eq. (13) and (14)
represent the minimum heights required to avoid nodal zone
assuming an initial depth of the top node (hnode-top(R)) of crushing failure at a stress equal to 0.85fc′ . With the node
approximately 75 mm (3 in.). With known values of these geometry defined, a revised value for the diagonal strut force
forces acting at the top node, the node dimensions were FS-truss was estimated using the strut stresses determined
adjusted so that nodal failure would not occur to be consistent during the tests (fS-truss(test)) applying Eq. (17), which then
with the observed behavior during the tests. The revised permitted evaluation of the shear force transferred into the
height of the right face of the node (hnode-top(R)), the height support by truss action (Vtruss) through the use of
of the left face of the node (hnode-top(L)), and strut width
(wS-truss) were determined using FS-truss = fS-truss(test)wS-truss b (17)

ACI Structural Journal/September-October 2009 687


Table 3—Geometry of strut-and-tie models for load transfer evaluation
Truss model Tied arch model
fS-truss(test), fS-TA(test), f S – TA ( test )
Truss strut Tied arch strut ---------------------
Specimen angle γ, deg MPa ws-truss, mm FS-truss, kN angle α, deg FS-TA, kN ws-TA, mm fS-TA, MPa MPa f S – TA
DB1.5-0.75 52.7 9.8 93 139 32.9 219 92 15.7 14.5 1.08
DB1.5-0.50 52.7 9.0 91 125 33.3 203 88 15.2 16.8 0.90
DB1.5-0.38 52.7 17.8 105 284 33.3 * * * 15.3 *
DB2.0-0.75 43.7 12.9 91 181 25.5 69 56 8.0 9.1 0.88
DB2.0-0.50 43.7 11.6 90 162 25.5 85 60 9.2 10.4 0.88
DB2.0-0.43 43.7 16.8 99 258 26.0 * * * 5.9 *
*Specimen where all load was transferred through truss action.
Note: 1 MPa = 145 psi; 1 mm = 0.0394 in.; 1 kN = 0.2248 kip.

Vtruss = FSsinγ (18) Table 4—Fraction of shear transferred by


individual load transfer mechanisms
With these revised values of FS-truss and Vtruss, Eq. (11) Vtest, Vtruss, Vtruss/Vtot
Specimen kN kN Vtruss/Vtot VTA, kN VTA/Vtot (FIP)
through (18) were applied again to adjust the top node geometry
and forces until convergence was achieved in two subsequent DB1.5-0.75 229 110 0.48 119 0.52 0.79
cycles within reasonable tolerance. The effective depth z was DB1.5-0.50 211 100 0.47 112 0.53 0.77
calculated in each iteration using the most current value for DB1.5-0.38 214 226 1.06 * * 0.78
hnode-top(R). Table 3 summarizes the diagonal strut force, DB2.0-0.75 156 125 0.81 30 0.19 1.00
FS-truss, corresponding to the measured strut stress and the DB2.0-0.50 149 112 0.75 36 0.25 1.00
corresponding strut width determined from the procedure DB2.0-0.43 133 178 1.34 * * 1.00
described above. Note: 1 kN = 0.2248 kip.
The fraction of shear force transferred by truss action is
summarized in Table 4 for each of the specimens in Groups shear force was assumed to be transferred by truss action in
1.5 and 2.0. It can be observed that this fraction depends on specimens with the shortest anchorage lengths in each group.
a/d, as would be expected. For specimens where anchorage
Table 3 lists relevant parameters of the diagonal strut in
failures were not observed, the fraction of total shear transferred
the tied-arch model for each specimen in Groups 1.5 and 2.0.
through truss action was approximately 0.50 and 0.80 for
The value of these parameters was obtained, assuming that
Groups 1.5 and 2.0, respectively. Interestingly, specimens with
only a fraction of the total shear corresponding to VTA was
the shortest anchorage length in these two groups, in which
applied to the model. The assumed angle of inclination of the
anchorage failure was observed during the tests, seemed to
diagonal strut in the model (α) was calculated assuming an
transfer the entire applied shear by truss action. The difference
approximate node height of 75 mm (3 in.). The diagonal strut
in observed load transfer mechanism of specimens with
force FS-TA required for equilibrium was calculated using
inadequate anchorage length can be attributed to the
Eq. (19), from which the minimum node height to avoid
inability to develop high stresses at the bottom node over the
nodal crushing was estimated using Eq. (20). The diagonal
support, as required by a tied-arch model. For comparison,
strut width at the face of the top node was estimated using
the value considered transferred by truss action included in
Eq. (21).
the “FIP Recommendations 1999” (Federation International
de La Precontrainte 1999) (Eq. (4)) is listed in the last
column of Table 4. The values determined in these tests are V TA V test – V truss
F S – TA = ----------
- = -----------------------------
- (19)
consistently lower than those computed using the FIP sin α sin α
recommendations by approximately 40% for specimens in
Group 1.5 and 20% for specimens in Group 2.0. ACI 318-08
does not give guidance as to specific strut-and-tie models F S – TA
h node-TA = ------------------- (20)
to use for design of deep beams as indicated earlier in 0.85f c′ b
this paper.
According to the assumed load transfer models, the difference L plate-top
in total shear not transferred by truss action must be transferred - sin α + h node-TA cos α
w S – TA = -------------------- (21)
by tied-arch action (VTA = Vtest – Vtruss). A tied arch model 2
similar to that illustrated in Fig. 8 was used in combination
with strut stresses experimentally determined by using readings With these properties and forces determined, a diagonal
from Instrument L1 (Fig. 3(b)) to assess whether the procedure strut stress could be determined analytically (fS-TA) and
to estimate Vtruss previously described gave reasonable compared with the strut stress experimentally estimated
results. The assessment was based on comparing the strut ( fS-TA(test)). These values and their ratio are listed in the last
stress determined from the assumed tied-arch model with three columns of Table 3. As shown in the table, strut stress
strut stresses determined experimentally. For this comparison, values determined analytically with an assumed tied-arch
the fraction of shear force transferred by tied-arch action was model compare favorably with values experimentally
calculated as the difference between the total applied shear determined. The maximum difference observed was
force and the fraction of shear transferred by truss action as approximately 12% in specimens in Group 2.0, giving
previously calculated. As previously discussed, the total confidence in the procedure and models adopted for the

688 ACI Structural Journal/September-October 2009


experimental evaluation of shear force transfer. The fraction conservative for the beams tested in this research. βs seems to
of shear force transferred by tied-arch action is listed in decrease with increasing a/d, however, which would be
Table 4 for the two groups of specimens. consistent with the presence of higher longitudinal strains
developing in the web. Further research seems warranted in
CONCLUSIONS this area.
Tests of deep beams having three different a/d, where the 5. Although short bar anchorages should not be encouraged in
longitudinal reinforcement was anchored using shorter design to ensure ductility of beams prior to failure, the tests
lengths than required by the ACI 318-08 Code, were reported reported in this paper show that even with short anchorages
in this paper. Two parameters that are relevant to the development yielding of the longitudinal reinforcement occurred at
of strut-and-tie models for this class of deep beams were critically midspan, and, depending on the load transfer mechanism
examined in the paper (effective concrete strength factor βs and that is mobilized, could spread all the way into the support
the force transfer mechanism). The transverse reinforcement node. The high fraction of applied shear carried through truss
content in all beams was kept constant in order to isolate the action in beams with a/d greater than 1.0 and short
effects of a/d and anchorage conditions. The effects of lateral anchorages allowed bottom bars to develop stresses lower
confining stresses along the straight anchored bars were than yield at the support node and still be able to carry the
reported in a previous paper (Roy and Breña 2008). The applied load safely. The tensile force demand at the support
following conclusions can be drawn from the test results node decreased when truss action developed in the beams.
presented in this paper, which are strictly applicable only for
the range of variables included in the present study: ACKNOWLEDGMENTS
1. Longitudinal bars in the beams reported in this paper Graduate studies for the second author were supported through a Brack
were anchored at nodes over a support generating transverse Structural Engineering Graduate Student Fellowship at the University of
stresses along the bar within the nodal region. These transverse Massachusetts Amherst. The authors would like to express their most
sincere gratitude to R. Brack for establishing the graduate fellowship. In
stresses prevented bars from pulling out from the nodal zone kind material donations received for specimen fabrication from Barker Steel
and promoted stress redistribution within the beams after bar and Nucor Steel Connecticut are deeply appreciated.
yielding. Even beams with short longitudinal bar anchorages
were capable of redistributing load into a load path other REFERENCES
than directly from load to support, which allowed development ACI Committee 318, 2008, “Building Code Requirements for Structural
of higher loads leading to strut crushing at failure of most Concrete (ACI 318-08) and Commentary,” American Concrete Institute,
of the specimens. Farmington Hills, MI, 465 pp.
2. Because of the transverse stresses developed within the Federation Internationale de la Precontrainte, Commision 3 Recommenda-
tions, 1999, “FIP Practical Design of Structural Concrete,” FIP Symposium,
support node, deep beams with short longitudinal bar London, UK, 113 pp.
anchorages were able to transfer a fraction of the total load Hognestad, E., 1951, “A Study of Combined Bending and Axial Load in
from loading point to support through truss action. This Reinforced Concrete Members,” Bulletin Series No. 399, University of Illinois
phenomenon is not expected to occur in beams supported Engineering Experiment Station, Urbana, IL, V. 49, No. 22, 128 pp.
through indirect supports where no clamping stresses are MacGregor, J. G., and Wight, J. K., 2005, Reinforced Concrete:
developed on longitudinal bars anchored in the nodal region. Mechanics and Design, fourth edition, Prentice Hall, NJ, 1132 pp.
Quintero-Febres, C. G.; Parra-Montesinos, G.; and Wight, J. K., 2006,
3. The fraction of load transfer by truss action in deep “Strength of Struts in Deep Concrete Members Designed Using Strut-and-Tie
beams is significant, even for relatively low a/d. In these Method,” ACI Structural Journal, V. 103, No. 4, July-Aug., pp. 577-586.
tests, the fraction of load transferred by truss action in beams Roy, N. C., and Breña, S. F., 2008, “Behavior of Deep Beams with Short
with a/d = 1.5 and a/d = 2.0 was approximately 50% and Longitudinal Bar Anchorages,” ACI Structural Journal, V. 105, No. 4,
80% of the total shear, respectively. These percentages July-Aug., pp. 460-470.
would likely be different in beams with other amounts of Uribe, C. M., and Alcocer, S. M., 2001, “Behavior of Deep Beams
Designed using Strut-and-Tie Models,” Centro Nacional de Prevención de
transverse reinforcement. Desastres, México, D.F., 247 pp. (in Spanish)
4. The value of the effective concrete strength factor (βs) Wight, J. K., and Parra-Montesinos, G., 2003, “Strut and Tie Model for
used in ACI 318-08 for bottle-shaped struts (βs = 0.75) was Deep Beam Design,” Concrete International, V. 25, No. 5, May, pp. 63-70.

ACI Structural Journal/September-October 2009 689

You might also like