You are on page 1of 10

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 107-S32

Shear Strengths of Prestressed Concrete Beams Part 1:


Experiments and Shear Design Equations
by Arghadeep Laskar, Thomas T. C. Hsu, and Y. L. Mo

In this paper, Part 1, five full-scale prestressed concrete I-beams The test results and the shear analysis of the five beams, as
were tested to study the effect of three variables: the shear span- well as those reported in the literature, were used to develop
depth ratio (a/d), the transverse steel ratio (ρ t), and the presence of a new equation for the shear strength of prestressed concrete
harped strands on the web shear and the flexural shear capacity. beams. In this new equation, the shear capacity of a
The results from these tests, together with those found in literature, prestressed beam is simply a function of the a/d ratio, the
were used to develop an accurate, yet simple, equation for the compressive strength of concrete f c′ , the web cross section
shear strengths of prestressed concrete beams. This new equation bwd, and the ρt ratio of the beams. Notably, this study shows
is a function of the shear span-depth ratio (a/d), the strength of that the present ACI Code and AASHTO Specifications
concrete f c ′ , the web area bwd, and the ρ t ratio. unnecessarily include two complicated variables, the prestress
Although the ACI and AASHTO shear provisions include two
force and the angle of failure crack. These two variables are
other variables, namely the prestress force and the angle of failure shown experimentally to have no significant effect on the
crack, this study showed that these two variables had no significant shear capacity (Laskar et al. 2006; Lyngberg 1976).
effect on the shear capacity. In addition, a new formula was The maximum shear strength, which ensures the yielding
derived for the maximum shear strengths to preclude the web of transverse steel before the web crushing of concrete, is
crushing of concrete before the yielding of transverse steel. The very different in ACI and AASHTO. In this study, a new
ACI minimum stirrup requirement was also evaluated. formula is proposed that reduces the unwarranted conservatism
In the Part 2 paper, given in this same journal issue, the proposed of the ACI Code, and guards against the unsafe nature of the
shear design method will be compared to the shear provisions in AASHTO Specifications for high-strength concrete.
the ACI 318-08 and the 2007 AASHTO LRFD Specifications. The ACI minimum stirrup requirement is checked by the
prestressed I-beams available in literature. It was found that
Keywords: beams; full-scale tests; prestressed concrete; shear design; the ACI requirement, shown later as Eq. (18), is not conservative
shear strength. when the a/d ratios are in the range from 2 to 4.
In Part 2, the proposed shear design procedures for
INTRODUCTION prestressed concrete beams are analyzed and compared to
Prestressed concrete I-beams are used extensively as the the shear design provisions of ACI and AASHTO.
primary superstructure components of highway bridges.
This research aims to solve one of the most troublesome RESEARCH SIGNIFICANCE
problems in prestressed concrete beams, namely the shear Through the testing of five girders and the analysis of
problem. In fact, there is, at present, no rational model to 143 girders available in literature, the shear strengths of
predict the shear behavior of prestressed concrete structures prestressed concrete bridge girders were found to be strongly
and the various modes of shear failures. Because of this affected by the a/d, and weakly influenced by the prestress
deficiency, all of the shear design provisions, such as force and the angle of failure crack. These findings led to the
those in the ACI Building Codes and the AASHTO development of an accurate, yet simple, equation for the
LRFD Specifications, are empirical, complicated, and shear analysis and design of prestressed I-girders. This new
have severe limitations. shear design method can replace the shear provisions in the
ACI Building Code and the AASHTO LRFD Specifications.
This paper describes the laboratory tests on five full-scale
prestressed concrete I-beams and the development of a
EXPERIMENTAL WORK
rational and simple method to calculate the shear strength of Test program and specimens
prestressed concrete beams. Three of the prestressed Five beams having cross sections known as Texas Department
concrete beams were designed and tested to fail in web shear of Transportation (TxDOT) Type A and shown in Fig. 1 were
mode and the other two in flexural shear. It was observed designed to study their behavior in web shear and flexural
that the shear span-depth ratio (a/d) was a primary variable shear. The total length of each of the beams was 7.62 m (25 ft)
affecting the shear strength of prestressed concrete beams. whereas the span length was 7.32 m (24 ft). The position of
The concrete contribution term Vc in web shear failure was the vertical loads on the beams, together with the support
higher than that recognized by ACI 318-08 (ACI Committee positions, is shown in Fig. 2. Three of the five beams
318 2008) and much higher than that allowed by the AASHTO
LRFD Specifications (AASHTO 2007). The proposed Vc
term was derived from the shear resistance of concrete along ACI Structural Journal, V. 107, No. 3, May-June 2010.
an inclined failure crack, rather than the tensile resistance of MS No. S-2009-079.R1 received March 23, 2009, and reviewed under Institute
publication policies. Copyright © 2010, American Concrete Institute. All rights reserved,
concrete across the failure crack, as assumed in the shear including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including author’s closure, if any, will be published in the March-
provisions of the previously referenced codes. April 2011 ACI Structural Journal if the discussion is received by November 1, 2010.

330 ACI Structural Journal/May-June 2010


Twelve 12.7 mm (0.5 in.) diameter, seven-wire, low-relaxation
Arghadeep Laskar is an Engineer at WorleyParsons Inc., Houston, TX, working on
the design and analysis of offshore structures. He was previously a Graduate strands were used as the prestressing steel. The prestressing
Research Assistant in the Department of Civil and Environmental Engineering, strands had an ultimate strength of 1862 MPa (270 ksi). The
University of Houston, Houston, TX. He received his PhD in May 2009.
strands were prestressed by hydraulic jacks in a long-line
Thomas T. C. Hsu, FACI, is Moores Professor in the Department of Civil and prestressing bed with a steel form. Hold-downs were installed
Environmental Engineering, University of Houston. He is the recipient of many on the bed to harp the strands at the desired inclination.
national awards, including the ACI Arthur J. Boase Award in 2007, the ACI Arthur R.
Anderson Award in 1990, and the ACI Wason Medal for Materials Research in 1965. One day after casting of the beams, the prestressing
strands were slowly released to minimize the bond slip
Y. L. Mo, FACI, is a Professor in the Department of Civil and Environmental
Engineering, and the Director of Thomas T. C. Hsu Structural Research Laboratory at the between the strands and the concrete. The compressive
University of Houston. He is a member of Joint ASCE-ACI Committees 445, Shear strength of concrete at the release of prestress was approximately
and Torsion, and 447, Finite Element Analysis of Reinforced Concrete Structures.
27.6 MPa (4000 psi). For Beams B3 and B5, the anchors of
the hold-down rods were removed after the release of the
prestress, and the rod holes were grouted by mortar.
(Beams B1, B2, and B3) were designed to fail in web shear
mode with an a/d ratio of 1.61, whereas the remaining two
Loading equipment
(Beams B4 and B5) were designed to fail in flexural shear
The beams were subjected to vertical loading up to their
mode with an a/d ratio of 4.29. One web shear specimen
maximum shear capacity in a steel loading frame, as shown
(Beam B3) and one flexural shear specimen (Beam B5) had in Fig. 3. Two of the four actuators (B and C), each attached
harped prestressing strands. Figure 1 shows the locations of to a vertical steel frame, were used to apply the vertical loads
the strands at the end cross section. Another variable in the on the beams. Each of the two actuators had a capacity of
test program was the amount of shear reinforcement (ρt), 1423 kN (320 kips). Actuator frame B was installed on the
which was either 0.17% (minimum) provided by No. 2 north end of the beam. Actuator frame C was installed on the
reinforcing bars at 250 mm (10 in.) spacing, or 0.95% provided south end of the beam. The loads from Actuators B and C
by No. 4 reinforcing bars at 178 mm (7 in.) spacing. Table 1 gives were simultaneously applied at 0.914 m (3 ft) from the
the three test variables (a/d, ρt, strand configuration) for Beams supports (both north and south supports) to create an a/d
B1 to B5. ratio of 1.61 for Beams B1, B2, and B3; and at 2.44 m (8 ft)

Fig. 1—Cross section of Beams B1 to B5. (Note: dimensions are in mm; 1 mm = 0.0394 in.)

Table 1—Test variables and failure loads of beams


Concrete Maximum shear Maximum moment
compressive Ultimate corresponding to corresponding to
Tendon profile Transverse cylinder strength at Failure Ultimate shear moment ultimate moment, ultimate shear,
Beam a/d (straight/harped) steel ratio, % test, MPa mode capacity, kN capacity, kN-m kN kN-m
B1-North 1.61 Straight 0.17 72.4 Web shear 840.2 — — 768.3
B1-South 1.61 Straight 0.17 72.4 Web shear 771.7 — — 705.7
B2-North 1.61 Straight 0.95 74.5 Web shear 894.5 — — 817.9
B2-South 1.61 Straight 0.95 74.5 Web shear 1040.8 — — 951.7
Flexure/
B3-North 1.61 Harped 0.95 64.6 — 927.3 1014.1 —
web shear
B4-South 4.29 Straight 0.17 71.0 Flexural shear 430.6 — — 1049.9
Flexure/
B5-North 4.29 Harped 0.17 64.5 — 1062.9 435.9 —
flexural shear
Note: 1 MPa = 145 psi; 1 kN = 0.225 kips; 1 kN-m = 0.74 kip-ft.

ACI Structural Journal/May-June 2010 331


from the supports to create an a/d ratio of 4.29 for Beams B4 hold. The cracks marked on one of the specimen grids is
and B5. shown in Fig. 4. The crack widths were measured using
Actuator loads were applied through a roller assembly hand-held microscopes having a measuring precision of
consisting of two 152.4 x 304.8 x 50.8 mm (6 x 12 x 2 in.) 0.0254 mm (0.001 in.).
bearing plates and two rollers of 50.8 mm (2 in.) diameter Further details about the test specimens and testing procedure
and 304.8 mm (12 in.) length. Lead sheets were also used can be obtained from the research report by Laskar et al. (2006).
between the load bearing plates and the beam surface.
The loads and displacements of the actuators were TEST RESULTS
controlled by the MTS system. Actuators B and C were first Shear strengths
programmed with a load control mode of 22.2 kN/minute Table 1 shows the ultimate strengths at failure for the five
(5 kips/minute). When the slope of the load-displacement test beams. It can be seen that both ends of Beams B1 and B2
curve started decreasing, the control mode was switched to a failed in web shear mode. It was possible to induce two web
displacement control of 5.08 mm/hour (0.2 in./hour). This shear failures in each of these two beams because after the
step continued until shear failure occurred near either end of beams failed at one end, the testing could be continued in
the beam. A full test lasted 8 to10 hours. The displacement displacement control mode until a second failure occurred at
control feature was essential in capturing the ductility/ the other end. The tests could be continued on the existing
brittleness behavior of beams failing in shear. supports of the beams because the first failure damaged only
a short length of the beams adjacent to the support.
The north half of Beam B3 failed in flexure at the hold-
Instrumentations
down section due to poor grouting of the holes for the hold-
During testing, linear voltage differential transformers down rods. Because the hole extends from the hold-down
(LVDTs) were used to measure the displacements within the point up to the top flange of the beams, the flexural capacity
failure regions of the beam adjacent to the points of load of the hold-down section was weakened. Fortunately, the
application, as shown in Fig. 2. A set of six LVDTs forming shear span at the north end almost reached its web shear
a rosette was installed on both faces of the beams to get the capacity because the concrete in the web region was close to
average deformations within the failure zone (Fig. 4). crushing at failure. The load-deformation curve of B3-North
Several LVDTs were also placed vertically under the in Fig. 5 shows that the north end had reached its flat top. The
beam, both at the supports and at the points of loading, to south end of the beam could not be tested after the failure of
measure the total and net deflections of the beam. The net the north half of the beam.
deflection of the beam was obtained by subtracting the
deflections measured under the loading points from the
support deflections, if any.
Electrical-resistance strain gauges were installed on both
legs of the vertical stirrups inside the beams to monitor the
steel strains during the load test. The locations of strain
gauges on the stirrups were selected to intersect the predicted
shear failure planes of the beams (Laskar et al. 2006). On
average, each beam was instrumented with 30 LVDTs and
16 strain gauges to record the structural behavior of the
beam. Data from these sensors were continuously monitored
during the testing and stored by an acquisition system.
Shear cracks that formed on the beam web during the load
test were marked at regular intervals by putting the test on

Fig. 3—Test setup.

Fig. 2—Loading positions and LVDT rosettes in Beams B1 to B5.


(Note: 1 mm = 0.0394 in.) Fig. 4—LVDT setup on Beam B4.

332 ACI Structural Journal/May-June 2010


Beams B4 and B5 were designed to fail in flexural shear failure crack, T is the tensile force in the prestressing tendons
mode in a region adjacent to the load point at one-third of the at the ultimate load of the beams, and θ is the angle between
span. The failure destroyed one-third of the beam length, so the failure crack and the longitudinal axis.
it was not possible to devise a loading scheme to create The crack angle θ of web shear failure in Beams B1 to B3
another flexural shear failure on the other end. Hence, each was observed to be less than 45 degrees by a few degrees. To
of these two beams could provide only one set of data. check the validity of Eq. (1), θ was also determined using an
Specimen B5 also failed in flexure at a distance of 3.35 m eight-step iteration procedure described in Laskar et al.
(11 ft) from the end due to a weak section created by the poor (2006). It was found that the calculated θ angles for
grouting of the hole for the hold-down rod. Similar to Beams B1 to B3 were approximately the same as those
Beam B3, Fig. 5 showed that the flexural shear failure load observed in tests.
of Beam B5 was very close to its failure load in flexure. The term (S – Tcos(θ))/sinθ in Eq. (1) is the “contribution
of concrete in shear,” Vc. To avoid the excessive complexity
Load-deflection curves involved in the calculation of S, T, and θ, it was decided to
From the shape of the load-deflection curves shown in Fig. 5, it
can be seen that the specimens designed for web shear failure

derive the Vc term directly from tests. In Eq. (1), F V is the
contribution of steel in shear, denoted as Vs. Thus
(Beams B1, B2, and B3) had much higher shear capacities
compared to the specimens designed to fail in flexural shear
Vn = Vc + Vs (2)
(Beams B4 and B5); however, the specimens that failed in
flexural shear had higher ductility. Both the strengths and
deflections were predicted accurately by the flexural analysis Contribution of steel (Vs)
(Laskar et al. 2006). The Vs term in Eq. (2) must be based on the observed
failure crack. In beams that failed in flexural shear, such as
Strains at failure B4 and B5, the angle θ of the cracks would theoretically be
The ultimate strains at the failure regions of Beams B1 to much less than 45 degrees according to the principal
B5 were measured by the LVDT rosettes shown in Fig. 4. compressive direction of concrete in the prestressed webs.
These strains were used in the theoretical shear analysis of Such a crack with a small θ angle is shown in Fig. 7 by the
the beams to rigorously calculate their ultimate capacities large crack that passed through Grid 4 at the bottom edge of
(Laskar et al. 2006). the web of Beam B4; however, this crack with a small θ
The ultimate strains of transverse stirrups in all five beams
were measured by gluing an electrical strain gauge to each of
the two stirrup legs in the failure regions. Three stirrups were
instrumented with strain gauges in Beams B1, B4, and B5,
and four stirrups were instrumented in Beams B2 and B3.
The locations of the strain gauges on the stirrups were
selected to intersect the failure plane. The ultimate strains of
the stirrups were beyond the yield strain at failure, with many
stirrups far into the strain hardening range. The stirrup forces
were also used to derive the ultimate shear capacities of the
beams in the report of Laskar et al. (2006).

SHEAR ANALYSIS OF PRESTRESSED BEAMS


Shear model
The concept of shear resistance developed by Loov
(2002), Fig. 6, was used to derive the ultimate web shear
capacity of Beams B1 to B3. According to this model, the
contribution of concrete to the shear capacity of the beam
stems from the shear stress in the concrete along a failure Fig. 5—Load-deflection curves of Beams B1 to B5. (Note:
crack, represented by the force S. Loov’s concept is very 1 kN = 0.225 kips; 1 mm = 0.0394 in.)
different from the concept of the existing design methods
(ACI and AASHTO), which assume that the concrete
contribution to the shear capacity of beams is derived from
the tensile stress across the cracks.
Assuming the failure surface to be an inclined plane, and
taking the force equilibrium of the free body along the crack
direction, Fig. 6, the shear capacity of the beam, Vn, can be
calculated as

S – Tcosθ
V n = ------------------------ +
sinθ ∑ FV (1)

where ∑ F V is the summation of vertical forces of the


stirrups intersected by the failure crack at the ultimate load, Fig. 6—Analytical model used for calculating web shear
S is the ultimate shear force of concrete in the direction of capacities of beams.

ACI Structural Journal/May-June 2010 333


angle did not develop to form a failure surface at ultimate This concept by Loov (2002), which was also supported by
load. In fact, Fig. 7 shows another large crack with a θ angle Kim and Mander (2005, 2007), gives the number of stirrups
of approximately 45 degrees, developed and passed through crossing the crack as (d/s – 1), and the Vs term as
Grid 8 at the bottom edge of the web. This approximately
45-degree crack was observed to cause failure.
V s = A v f y ⎛ --- – 1⎞
d
For design, the failure crack can simply be assumed to (3)
⎝s ⎠
incline at an angle of 45 degrees, similar to the ACI Code.
The assumption of a 45-degree failure crack can also be
supported by a study of shear energy dissipations in the Besides being more rational and conservative, Eq. (3) has
failure zone. According to Laskar (2009), a crack with an additional practical advantage: all of the tedious stirrup
inclination close to 45 degrees would result in minimum spacing limitations to guard against the unsafe nature of the
shear energy dissipation. smeared stirrups assumption can be eliminated.
In the ACI Code, the 45-degree crack is located in a
manner as shown in Fig. 8(a). This ACI concept of smearing Contribution of concrete (Vc)
the stirrups results in an average number of stirrups, d/s, Two trends were observed from the test results of
crossing the crack. In this paper, however, we used a more Beams B1 to B5. First, the Vc term in the shear strength of
realistic concept of seeking a minimum shear resistance prestressed beams was observed to be a strong function of
among a series of individual stirrups, as shown in Fig. 8(b). the a/d ratio because the web shear strengths of Beams B1 to
B3 with a/d = 1.61 were much higher than the flexural shear
strengths of Beams B4 and B5 with a/d = 4.29. This parameter,
a/d, must be implemented in the new shear equation.
Second, Laskar et al. (2006) showed that the shear strength
was not significantly affected by the angle of failure plane,
both in the case of web shear failure (Beams B1, B2, and B3)
and in the case of flexural shear failure (Beams B4 and B5).
This angle variable need not be included in the new equation.
The tests conducted by Lyngberg (1976) convincingly
showed that the prestress force could also be neglected in the
new shear equation. Nine 600 mm (23.6 in.) deep beams
were tested, in which the major variable was the intensity of
prestress. The cross section, web reinforcement, ultimate
moment, and shear span were held constant. The results
showed that the shear strength was not influenced by the
amount of prestress. For practical application and to be
Fig. 7—Flexural shear failure region in Beam B4. consistent with ACI 318-08, however, the effective prestress
force in the beams should not be less than 40% of the tensile
strength of the flexural reinforcement.
Lyngberg’s tests also provided another important
observation when compared to the UH test specimens. The
two groups of test specimens were similar in size and shape,
except that Lyngberg’s specimens had wide top flanges
(700 mm [27.6 in.]) and the UH specimens had narrow top
flanges (305 mm [12 in.]). Because good agreement in shear
strengths was observed between these two groups of tests, it
was concluded that the top flange width was not a significant
variable affecting the shear strength, and that the web region,
defined as bwd, was the primary shear-resisting component.
From the study of RC shear elements with concrete
strengths up to 100 MPa (14,500 psi) (Zhang and Hsu 1998),
it was observed that the softened compressive strength of
concrete struts is proportional to the square root of the
compressive strength, f c′ (for more details, refer to a
following section on maximum shear strength). Therefore,
the shear force that causes the crushing of concrete in the
web of a beam must be proportional to f c′ bwd. This
parameter could be used to normalize a shear force into a
nondimensional quantity.
The test results of Beams B1 to B5 reported in this
research, as well as those of Hernandez (1958), MacGregor
(1960), Mattock and Kaar (1961), Bruce (1962), Hanson and
Hulsbos (1965), Lyngberg (1976), Elzanaty et al. (1986),
Robertson and Durrani (1987), Kaufman and Ramirez
Fig. 8—Determination of number of stirrups for contribution of (1988), and Shahawy and Batchelor (1996) were used to
steel Vs. implement the parameter a/d into the new shear equation.

334 ACI Structural Journal/May-June 2010


The concrete shear contribution Vc of all of the specimens 25% less than the web crushing failure load of the ends with
were calculated by subtracting the steel contribution Vs as perfectly anchored strands.
per Eq. (3), from the total shear capacities of the beams. The Because shear failures at beam ends (a/d ratios less than
normalized concrete shear stress Vc/ f c′ bwd of the specimens 1.6) are significantly affected by the bond slip of prestressing
was obtained thereafter and its variation versus a/d was strands, the limitation in Eq. (4) at low a/d ratios must be a
plotted in Fig. 9. A conservative Vc/ f c′ bwd versus a/d curve function of the anchorage length of the strands. A satisfactory
has an expression of anchorage length for shear should be one that ensures the
yielding of the transverse stirrups at shear failure in the web.
1.17 The upper limit in Eq. (4) for a/d less than 1.6 appears to be
- f c ′ ( MPa )b w d ≤ 0.83 f c ′ ( MPa )b w d
V c = ------------------- (4)
(a ⁄ d)
0.7 valid for the TxDOT Type A beams tested at UH. The UH
beams have 12 seven-wire strands of 12.7 mm (0.5 in.)
diameter, with an extension of 152 mm (6 in.) beyond the
where f c′ ( MPa ) ≈ 12 f c′ ( psi ) ⋅ b w equals the width of the
support resultant and a supporting plate of 152 mm (6 in.) wide.
web of the prestressed beam; and d is the effective depth
More research is required to determine a satisfactory anchorage
from the centroid of the tendons to the top compression fiber
length of prestressed strands to prevent shear/bond failure.
of the prestressed beam. To be consistent with ACI 318-08,
the value of d is not taken to be less than 80% of the total
depth of the beam. Shear strengths
The expression Assuming that the transverse steel yields at failure, the
final equation for the shear strength of prestressed I-beams is
– 0.7 obtained by substituting Eq. (3) and (4) into Eq. (2)
V c = 1.17 ( a ⁄ d ) f c′ ( MPa )b w d
– 0.7
( V c = 14 [ a ⁄ d ] f c′ [ psi ]b w d )
V n = -------------------- f c′ ( MPa )b w d + A v f y⎛ d
--- – 1⎞
1.17
(5a)
0.7 ⎝s ⎠
(a ⁄ d)
shown in Eq. (4) was substantiated by the large-sized test
specimens of Mattock and Kaar (1961) with a beam height
of 648 mm (25.5 in.). The shear strengths of their beams where 1.17(a/d)–0.7 ≤ 0.83.
agreed very well with the proposed expression in Fig. 9 for The ratio a/d in Eq. (5a) represents the effect of moment
a/d ratios of 4.5, 3.25, 2.0, 1.5, and all the way down to 1.0. M (relative to shear V) on the shear strength Vn. This moment
The upper limit of effect was quantified by Joint ACI-ASCE Committee 326,
(1962) using a parameter M/Vd along the length of a beam
with a uniformly distributed load as well as other types of
V c ≤ 0.83 f c′ ( MPa )b w d ( V c ≤ 10 f c′ [ psi ]b w d ) loading. The parameter M/Vd degenerated into a/d for beams
with concentrated loads, which are the easiest to apply and
in Eq. (4) is imposed for simply supported beams to ensure are employed by most researchers. Replacing a/d by M/Vd in
that the end anchorage will be sufficient to produce yielding Eq. (5a) results in the following expression for all types of
of the stirrups, as observed in the web shear failure of loading
Beams B1 to B3. Mattock and Kaar (1961) tested continuous
beams, which did not have the anchorage problems at beam
V u d⎞ 0.7
V n = 1.17 ⎛ -------- f c′ ( MPa )b w d + A v f y⎛ --- – 1⎞
ends associated with simply supported beams. d
- (5b)
⎝ Mu ⎠ ⎝s ⎠
Shear/bond failure at beam ends
Shahawy and Batchelor (1996) showed that the prestressed where 1.17(Vud/Mu)0.7 ≤ 0.83.
strands at the ends of the AASHTO Type II pretensioned
Using the U.S. Customary system, Eq. (5b) becomes
concrete beams would slip before the web crushing of shear
failures. They referred to this type of failure as shear/bond
failure. When the a/d ratio is greater than approximately 1.6,
the bond slip did not appear to have a noticeable effect on the
shear strengths; however, in their B1 series of 10 beams
where a/d ratios varying from 1.37 to 1.52, the bond slip
appeared to have reduced the shear strengths. It was not
clear, however, whether the transverse stirrups in the shear
span had yielded.
Ma et al. (2000) tested two NU1100 pretensioned beams
with a/d ratios that varied from 1.16 to 1.28. At both ends of
their Beam A, the strands were extended beyond the end
face, bent upward, and securely anchored into a large
concrete diaphragm attached to the end face. Because bond
slip was not allowed to occur, Beam A failed in web
crushing, a typical shear failure, at both ends. One failed end
of Beam A was then sawed off, leaving the ends of strands
flush with the end face. The application of shear loads then
produced a shear/bond failure, rather than a web crushing Fig. 9—Variation of normalized concrete shear with a/d.
failure. The shear/bond failure load of the saw-cut end was (Note: 1 MPa = 145 psi.)

ACI Structural Journal/May-June 2010 335


V u d⎞ 0.7 V u d⎞ 0.7
V u = 14 ( 1.17 ) ⎛ --------
- f c′ (psi) b w d + Av f y ⎛ d
--- – 1⎞ (5c) V n = 1.17 ⎛ --------
- f c′ ( MPa )b w d + A v f y⎛ --- – 1⎞ + V p (6a)
d
⎝ Mu ⎠ ⎝s ⎠ ⎝ Mu ⎠ ⎝s ⎠

where 14(Vud/Mu)0.7 ≤ 10. where 1.17(Vud/Mu)0.7 ≤ 0.83.


Using the U.S. Customary system, Eq. (6a) becomes
In Eq. (5b) or (5c), the three quantities Mu, Vu, and d in
Vud/Mu are calculated at the same given section under
V u d⎞ 0.7
consideration. V n = 14 ⎛ --------
- f c′ ( psi )b w d + A v f y⎛ d
--- – 1⎞ + V p (6b)
⎝ Mu ⎠ ⎝s ⎠
Contribution of harped strands to shear strengths
The additional shear strengths contributed by harped where 14(Vu d/Mu)0.7 ≤ 10.
strands were studied by comparing the shear strengths of It should be emphasized that the V p term in Eq. (6a) is
Beam B3 (with harped strands) to that of Beam B2 (without applicable to prestressed beams failing in web shear as well
harped strands) in the case of web shear, and by comparing as in flexural shear. This is not the same as the provisions of
the shear strength of Beam B5 (with harped strands) to that the ACI Code, which uses the vertical component of the
of Beam B4 (without harped strands) in the case of flexural prestressing force in harped strands in the case of web shear
shear. It was clear that shear capacity of prestressed concrete failure only, and not in the case of flexural shear failure.
beams is enhanced due to the harping of the prestressing
strands in both cases of web shear and flexural shear. The MAXIMUM SHEAR STRENGTHS
vertical component of the effective prestressing force in the Maximum shear strengths in code provisions
harped strands at the section, Vp, used in the ACI Code and Because Eq. (6a) is based on the yielding of transverse
the AASHTO 2007 specifications, appeared to be a steel reinforcement, a maximum shear strength Vn,max must
conservative estimate of the observed additional strength. be defined to ensure that transverse steel will yield before the
web crushing of concrete. ACI Code (1963) specifies
Hence, for a beam with harped strands, the ultimate shear
capacity can be conservatively taken as
V n, max = 0.83 f c′ ( MPa )b w d ( 10 f c′ [ psi ]b w d ) (7)

and AASHTO Specifications (2007) specifies

V n, max = 0.25f c′ b w d v or 0.225f c ′b w d (8)


if d v is assumed to be 0.9d

The fact that Vn,max is proportional to f c′ in the ACI Code,


Eq. (7), and is proportional to fc′ in the ASSHTO specifications,
Eq. (8), testifies to the confusion surrounding the formulas
for Vn,max. These two equations are plotted in Fig. 10.
The ACI equation

V n, max = 0.83 f c′ ( MPa )b w d ( 10 f c′ [ psi ]b w d )


Fig. 10—Variation of ultimate strength with concrete
strength. (Note: 1 MPa = 145 psi.) was first proposed by Joint ACI-ASCE Committee 326
(1962) and was incorporated into the ACI Code (1963) for
nonprestressed beams. It was slightly liberalized in 1971 for
application to both nonprestressed and prestressed concrete
beams (ACI 1971), in the form of limiting the Vs term to a
maximum contribution of steel

V s, max = 0.66 f c′ ( MPa )b w d ( 8 f c′ [ psi ]b w d ) (9)

If Vc is taken conservatively as

0.17 f c′ ( MPa )b w d ( 2 f c′ [ psi ]b w d )

then the Vu,max in the 1971 ACI Code is identical to that in


the 1963 ACI Code. Because the Vc term could be greater than

0.17 f c′ ( MPa )b w d
Fig. 11—Variation of normalized ultimate shear capacities
of beams with a/d. (Note: 1 MPa = 145 psi.) ( 2 f c′ [ psi ]b w d )

336 ACI Structural Journal/May-June 2010


the Vn,max for prestressed beams in the 1971 ACI Code is ζ = f ( f c′ )f ( ε 1 ) (12)
allowed to be somewhat greater than
where ε1 is the principal tensile strain. The function f(fc′ ) was
0.83 f c′ ( MPa )b w d ( 10 f c ′ [ psi ]b w d ) determined by panel element tests to be

f ( f c′ ) = -------------------------- ⎛ ----------------------⎞
The ACI formula for Vs,max has continued to be used up to 5.8 0.48
(13)
the present time (ACI 2008). f c′ ( MPa ) ⎝ f c′ [ psi ]⎠
On the other hand, the AASHTO formula Vn,max =
0.225fc′ bwd was introduced into the first edition of the
AASHTO LRFD Specifications (1994), based on the truss Substituting Eq. (13) into Eq. (12), and then into Eq. (11) and
model concept first introduced by the Canadian Code (1977) (10) results in
and the CEB-FIP Code (1978). A second edition of
AASHTO (1998) followed, and interim revisions were made V n, max = constant f c′ f ( ε 1 )b w ( 0.9d ) ( 0.5 ) (14)
from 1999 to 2003. The analysis in this paper is based on the
fourth edition of AASHTO (2007).
Both the ACI and AASHTO formulas can be checked by where the constant is 5.8 for fc′ in MPa (or 0.48 for fc′ in psi).
the prestressed beams of Bennett and Balasooriya (1971), Equation (14) shows that Vn,max should be a function of f c′
Rangan (1991), and Ma et al. (2000), which are over- for fc′ up to 100 MPa (14,500 psi).
reinforced in shear (refer to Fig. 11). These experimental
data are also shown in Fig. 10. It can be seen that the ACI Proposed maximum shear strength
formula is way too conservative. The AASHTO formula is Equation (14) can be simplified as follows
more reasonable when compared to the test data; however,
the AASHTO formula is expected to be unsafe for beams
with a concrete strength higher than 60 MPa (8700 psi), in V n, max = C 1 f c′ b w d (15)
view of the panel element tests in the following section.
where C1 is a constant to be determined by the shear tests of
Maximum shear strength in panel elements prestressed beams.
The shear resistance of a prestressed I-beam is contributed Before deciding on the constant C1 for the maximum shear
mainly by its web. The state of stress and strain in the web can be strength, we must first calibrate the balanced condition defined as
simulated by a two-dimensional (2D) panel element subjected to
pure shear stress. In the 45-degree direction, such an element is
also subjected to a biaxial stress condition with equal magnitude of V n, b = C b f c′ b w d (16)
principal compressive stress and principal tensile stress. Such a
biaxial stress condition can be applied to an element by placing it where Vn,b is the balanced shear force and Cb is the constant
in a universal panel tester (Hsu et al. 1995). Using the softened corresponding to the balanced condition. The balanced
truss model (Zhang and Hsu 1998), the maximum shear strength condition occurs when the beam is reinforced in a condition
of the element can be expressed as where the yielding of the transverse steel occurs simultaneously
with the web crushing of the concrete. When Vn < Vn,b, the
c beam is defined as under-reinforced in shear, where the
V n, max = σ 2 b w ( 0.9d )sinθcosθ (10) transverse steel yields before the crushing of concrete. When
Vn > Vn,b, the beam is defined as over-reinforced in shear,
c
where σ 2 is the compression strength of the concrete struts, where the concrete crushes without the yielding of steel.
0.9d is the height of the truss measured from the centroid of The balanced constant Cb can be calibrated by comparing
the steel to the centroid of the compression zone, and θ is the the over-reinforced beams versus the under-reinforced
angle of the failure surface with respect to the longitudinal beams, as shown in Fig. 11. Over-reinforced beams have
axis of the beam (refer to Fig. 6). The value θ equals 45 degrees been tested by three groups of researchers (Bennett and
when an element is subjected to pure shear. Balasooriya 1971; Rangan 1991; Ma et al. 2000). Under-
To develop an expression for Vn,max that is applicable to reinforced beams have been tested by all the other
the whole range of concrete strengths from 20 MPa researchers. Figure 11 plots Vn/ f c′ bwd against a/d. It can
(3000 psi) to 100 MPa (14,500 psi), Zhang and Hsu (1998) be seen that all of the over-reinforced beams had a Vn/
tested full-sized reinforced concrete (RC) panel elements f c′ bwd value above 1.5 for fc′ in MPa (or 18 for fc′ in psi)
(1.4 x1.4 x 0.178 m [55 x 55 x 7 in.]) with concrete strengths and almost all of the under-reinforced beams have a Vn/
up to 100 MPa (14,500 psi). Because the strength of the f c′ bwd value below 1.5 for fc′ in MPa (or 18 for fc′ in psi).
concrete struts in the principal compressive direction is Therefore, the balanced constant can be taken as Cb = 1.5 for
“softened” by the perpendicular principal tensile strain, the fc′ in MPa (or 18 for fc′ in psi).
c
“effective compressive strength” of concrete struts σ 2 is To provide some ductility in shear failure, the constant C1 for
maximum shear strength must be taken as less than Cb ⋅ C1 < Cb
c is also desirable because Rangan’s over-reinforced beams have
σ 2 = ζ f c′ (11) large web stiffeners under the loads to prevent local compression
failures. In view of the fact that the prestressed beams used
where ζ is the softening coefficient. Zhang and Hsu (1998) in practice do not contain web stiffeners, it was decided to
summarized the extensive panel tests at UH and showed that choose a conservative C1 value of 1.33 for fc′ in MPa (or 16 for
ζ is a product of two functions, that is fc′ in psi) and Vn,max can thus be expressed as

ACI Structural Journal/May-June 2010 337


represented by solid points whereas those that failed in a
ductile manner were represented by hollow points.
Figure 12 shows that some beams tested by Hernandez (1958),
MacGregor (1960), Hanson and Hulsbos (1965), and Robertson
and Durrani (1987) failed in a brittle manner although they had
shear stirrups greater than that required by Eq. (18) and (20). These
beams have a/d ratios in the range from 2.0 to 4.0, the region
known as “Kani’s Valley” (Kani 1964).
Two recommendations can be drawn from Fig. 12: (a) ACI
Eq. (18) is applicable to all a/d, except in the range from 2 to
4. In other words, when 0.25 < Vu d/Mu < 0.5, the ACI
minimum stirrup required by Eq. (18) should be doubled;
and (b) ACI 318-08 Eq. (19) should be removed as in the
AASHTO Specifications, because this formula could be
unconservative for prestressed I-beams.
Fig. 12—Variation of minimum shear reinforcement with a/d Notice that the six beams tested by Teoh et al. (2002) with
ratio. (Note: 1 MPa = 145 psi.) very small amounts of stirrups (ρfy/ f c′ varying from
0.045 to 0.079 for fc′ in MPa or 0.54 to 0.95 for fc′ in psi) are
not included in Fig. 12. These 700 mm (27.5 in.) deep, high-
V n, max = 1.33 f c′ ( MPa )b w d ( 16 f c′ [ psi ]b w d ) (17) strength concrete I-beams were simply supported and
subjected to a single point load at midspan. The beams were
Equation (17) is also plotted in Fig. 10. It can be seen that provided with strong stiffeners at the supports and under the
the UH formula is the most reasonable for concrete strengths loading point at midspan. The large stiffeners, together with
up to 100 MPa (14,500 psi). The ACI formula is obviously the large top and bottom flanges, formed a Vierendeel truss.
too conservative, and the AASHTO formula may be A Vierendeel truss, which has rectangular openings and
seriously unsafe when fc′ is greater than 60 MPa (8700 psi). requires no diagonal bracings, can resist the midspan load
even in the absence of the webs. The large deformations
MINIMUM SHEAR REINFORCEMENT undergone by the beams were caused by the flexural
Minimum shear reinforcement in prestressed I-girders is deformations of the Vierendeel truss and were not due to
required to prevent the brittle failure of the girders due to the any shear deformations of the web. Furthermore, the beams
fracture of the shear reinforcement shortly after the formation of were symmetrically loaded and were tested using a servo-
the inclined shear cracks. The ACI Code (2008) states that control system, which helped to promote ductile failure.
the minimum area of shear reinforcement, Av,min, provided at Because of these reasons, the ductile behavior of beams
a spacing s in a member having width bw and effective depth tested by Teoh et al. (2002) should not be used to justify a
d, should be greater than the smaller of two quantities low requirement for minimum stirrups.
expressed in the following
CONCLUSIONS
1. Five full-scale prestressed concrete I-beams were
bw s bw s
A v, min = 0.0625 f c′ ( MPa ) -------
- ≥ 0.35 -------
- (18) designed, cast, and tested to study their behavior in web
fy fy shear and flexural shear failure modes. The test program
bw s b w s⎞ included three variables: the a/d, the transverse steel ratio,
⎛A = 0.75 f c′ [ psi ] -------
- ≥ 50 -------
- and the effect of harped strands.
⎝ v, min fy fy ⎠ 2. A simple and accurate shear equation, Eq. (6a), based on
Loov’s rational concept, was developed using the test results
Aps f pu s d of the five prestressed beams tested at UH, as well as the
A v, min = ----------------
- ------ (19) 143 test beams available in literature. In this equation, the
80f yt d b w
concrete contribution (Vc) term is proportional to f c′ and
inversely proportional to (a/d)0.7. For the steel contribution (Vs)
The minimum shear reinforcement in the AASHTO term, the number of transverse steel bars intersecting a
specifications (2007) is 45-degree crack is taken as (d/s – 1). In addition, the
contribution of the harped strands, Vp, is taken into account
bw s for both web shear and flexural shear failure modes.
A v, min = 0.083 f c′ ( MPa ) -------
- (20) 3. In contrast to the ACI Code and the AASHTO specifications,
fy
an accurate equation for shear strengths of prestressed concrete
⎛A b w s⎞ beams can be obtained without taking into account the amount of
= f c′ [ psi ] -------
-
⎝ v, min fy ⎠ prestressing force and the angle of failure crack.
4. A new formula, Eq. (17), was derived for the maximum shear
Equations (18) and (20) are evaluated by the test beams strength (Vn,max) to ensure the yielding of transverse steel before
the crushing of concrete. Vn,max is a function of f c′ for concrete
found in literature as follows: first, all of the I-beams having
strengths from 20 MPa (3000 psi) to 100 MPa (14,500 psi).
ρt fy less than 1.38 MPa (200 psi) were selected. Second, the
nondimensional reinforcement index ρt fy/ f c′ of these
beams were plotted against their corresponding a/d ratios, as NOTATION
1 = direction of applied principal tensile stress
shown in Fig. 12. Third, the beams that failed in a brittle 2 = direction of applied principal compressive stress
manner due to the low amount of shear reinforcement were Aps = cross-sectional area of prestressing steel

338 ACI Structural Journal/May-June 2010


As = cross-sectional area of longitudinal steel Bennett, E. W., and Balasooriya, B. M. A., 1971, “Shear Strength of
Av = cross-sectional area of stirrup (including two legs for closed stirrup) Prestressed Beams with Thin Webs Failing in Inclined Compression,” ACI
Av,min = minimum required cross-sectional area of stirrup JOURNAL, Proceedings V. 69, No. 3, Mar., pp 204-212.
a = shear span of prestressed concrete beams Bruce, R. N., 1962, “An Experimental Study of the Action of Web
bw = width of web of prestressed I-beams Reinforcement in Prestressed Concrete Beams,” PhD dissertation, University of
C1 = constant term used to express maximum design shear capacity Illinois at Urbana-Champaign, Urbana, IL, 249 pp.
of prestressed beams CSA Code, 1977, “Code for the Design of Concrete Structures and
d = effective depth of prestressed beams measured from centroid of Buildings, CAN3-A23.3-M77),” Canadian Standards Association,
tendons to top fiber of concrete Rexdale, Toronto, ON, Canada, 131 pp.
dv = lever arm between resultant compressive and tensile forces in CEB-FIP, 1978, “Model Code for Concrete Structures,” CEB-FIP
AASHTO code International Recommendation, third edition, Comite Euro-International
fc ′ = cylinder compressive strength of concrete du Beton (CEB), Paris, 348 pp.
fc ′ = square root of cylinder compressive strength of concrete (same Elzanaty, A. H.; Nilson, A. H.; and Slate, F. O., 1986, “Shear Capacity of
units as fc′) Prestressed Concrete Beams Using High Strength Concrete,” ACI
fpu = ultimate strength of prestressing bars JOURNAL, Proceedings V. 83, No. 3, May-June, pp. 359-368.
fy = yield strength of bare mild steel bars
Hanson, J. M., and Hulsbos, C. L., 1965, “Overload Behavior of
fyt = yield strength of transverse steel
Pretensioned Prestressed Concrete I-Beams with Web Reinforcement,”
ΣFV = summation of stirrup forces crossing failure crack in prestressed beam
Highway Research Record 76, Highway Research Board, pp. 1-31.
h = depth of prestressed beams
l = direction of longitudinal reinforcements Hernandez, G., 1958, “Strength of Prestressed Concrete Beams with
ln = clear span of PC beams Web Reinforcement,” Report, The Engineering Experiment Station,
M = bending moment in prestressed beams University of Illinois at Urbana-Champaign, Urbana, IL, 135 pp.
Mn = nominal bending moment at given section of prestressed beams Hsu, T. T. C.; Belarbi, A.; and Pang, X. B., 1995, “A Universal Panel
Mu = bending moment at given section of prestressed beams Tester,” Journal of Testing and Evaluations, ASTM, V. 23, No. 1, pp. 41-49.
Pe = effective prestressing force Kani, G. N. J., 1964, “The Riddle of Shear Failure and its Solution,” ACI
R = normal force at failure surface in prestressed beams JOURNAL, Proceedings V. 61, No. 4, Apr., pp. 441-468.
S = concrete shear force along inclined shear crack of prestressed beam Kaufman, M. K., and Ramirez, J. A., 1988, “Re-evaluation of Ultimate
s = stirrup spacing Shear Behavior of High-Strength Concrete Prestressed I-Beams,” ACI
T = tensile force of tendons in prestressed beams Structural Journal, V. 85, No. 3, May-June, pp. 295-303.
t = direction of transverse reinforcements Kim, J. H., and Mander, J. B., 2005, “Theoretical Shear Strength of
V = shear resistance composed of Vc and Vs Concrete Columns Due to Transverse Steel,” Journal of Structural
Vcal = shear capacity of prestressed beams calculated by various methods Engineering, ASCE, V. 131, No. 1, Jan., pp. 197-199.
Vexp = experimental shear capacity of prestressed beams Kim, J. H., and Mander, J. B., 2007, “Influence of Transverse
Vc = concrete contribution to shear resistance of prestressed beams Reinforcement on Elastic Shear Stiffness of Cracked Concrete Elements,”
Vn = nominal shear strength at given section of prestressed beams Journal of Structural Engineering, ASCE, V. 131, No. 1, Jan., pp.197-199.
Vp = vertical component of prestressing force in beams with draped Laskar, A.; Wang, J.; Hsu, T. T. C.; and Mo, Y. L., 2006, “Rational Shear
tendons Provisions for AASHTO LRFD Specifications,” Technical Report 0-4759-1 to
Vs = steel contribution to shear resistance of prestressed beams Texas Department of Transportation, Department of Civil and Environmental
Vs,max = maximum shear capacity of stirrups in prestressed beams Engineering, University of Houston, Houston, TX, 216 pp.
Vu = ultimate shear force of prestressed beams Laskar, A., 2009, “Shear Behavior and Design of Prestressed Concrete
Vu,max = maximum shear strength of prestressed beams Members,” PhD dissertation, Department of Civil and Environmental
x = distance from support of prestressed beams Engineering, University of Houston, Houston, TX, 322 pp.
β = factor indicating ability of diagonally cracked concrete to transmit Loov, R. E., 2002, “Shear Design of Uniformly Loaded Beams,”
tension as per AASHTO specifications Presented at the Sixth International Conference on Short and Medium Span
ε1 = principal tensile strain Bridges, Vancouver, BC, Canada, pp. 515-522.
θ = angle of inclination of diagonal compressive stresses as per
Lyngberg, B. S., 1976, “Ultimate Shear Resistance of Partially
AASHTO specifications (same as crack angle)
Prestressed Reinforced Concrete I-Beams,” ACI JOURNAL, Proceedings
ρt = transverse steel ratio
V. 73, No. 4, Apr., pp. 214-222.
σc2 = compressive strength of concrete struts
ζ = softening coefficient of concrete in compression Ma, Z. J.; Tadros, M. K.; and Baishya, M., 2000, “Shear Behavior of
Pretensioned High-Strength Concrete Bridge I-Girders,” ACI Structural
Journal, V. 97, No. 1, Jan.-Feb., pp. 185-193.
REFERENCES MacGregor, J. G., 1960, “Strength and Behavior of Prestressed Concrete
AASHTO, 1994, “AASHTO LRFD Bridge Design Specifications,” first Beams with Web Reinforcement,” PhD thesis, Department of Civil
edition, American Association of State Highway and Transportation Engineering, University of Illinois, Urbana, IL, 295 pp.
Officials (AASHTO), Washington, DC, 1091 pp. Mattock, A. H., and Kaar, P. H., 1961, “Precast-Prestressed Concrete
AASHTO, 1998, “AASHTO LRFD Bridge Design Specifications,” Bridges—4: Shear Tests of Continuous Girders,” Journal of the PCA
second edition, American Association of State Highway and Transportation Research Development Laboratories, pp. 19-47.
Officials (AASHTO), Washington, DC, 1116 pp. (Including interim
Rangan, B. V., 1991, “Web Crushing Strength of Reinforced and
revisions for 1999 through 2003.)
Prestressed Concrete Beams,” ACI Structural Journal, V. 88, No. 1, Jan.-
AASHTO, 2007, “AASHTO LRFD Bridge Design Specifications,”
Feb., pp. 12-16.
fourth edition, American Association of State Highway and Transportation
Officials (AASHTO), Washington, DC, 1518 pp. Robertson, I. N., and Durrani, A. J., 1987, “Shear Strength of Prestressed
ACI Committee 318, 1963, “Building Code Requirements for Reinforced Concrete T Beams with Welded Wire Fabric as Shear Reinforcement,” PCI
Concrete (ACI 318-63) and Commentary (ACI 318R-63),” American Concrete Journal, Precast/Prestressed Concrete Institute, V. 32, No. 2, pp. 46-61.
Institute, Farmington Hills, MI, 91 pp. Shahawy, M. A., and Batchelor, B., 1996, “Shear Behavior of Full-Scale
ACI Committee 318, 1971, “Building Code Requirements for Reinforced Prestressed Concrete Girders: Comparison between AASTHO Specifications and
Concrete (ACI 318-71) and Commentary (ACI 318R-71),” American Concrete LRFD Code,” PCI Journal, Precast/Prestressed Concrete Institute, V. 41, No. 3,
Institute, Farmington Hills, MI, 78 pp. pp. 48-62.
ACI Committee 318, 2008, “Building Code Requirements for Structural Teoh, B. K.; Mansur, M. A.; and Wee, T. H., 2002, “Behavior of High-
Concrete (ACI 318-08) and Commentary,” American Concrete Institute, Strength Concrete I-Beams with Low Shear Reinforcement,” ACI Structural
Farmington Hills, MI, 473 pp. Journal, V. 93, No. 3, May-June, pp. 299-307.
Joint ACI-ASCE Committee 326, 1962, “Shear and Diagonal Tension,” Zhang, L. X., and Hsu, T. T. C., 1998, “Behavior and Analysis of
ACI JOURNAL, Proceedings V. 59, No. 1, Jan., pp. 1-30 (Chapters 1-4); No. 2, 100 MPa Concrete Membrane Elements,” Journal of Structural Engineering,
Feb., pp. 277-334 (Chapters 5-7); No. 3, Mar., pp. 352-396 (Chapter 8). ASCE, V. 124, No. 1, Jan., pp. 24-34.

ACI Structural Journal/May-June 2010 339