Structural behavior of continuous

two-span prestressed concrete T beams

with different tendon profiles

Uksun Kim, Ehab N. Ballu, Seongwon Hong, and Thomas H.-K. Kang

P
restressed concrete members are categorized into two
groups: bonded and unbonded. In a bonded member,
the strand and concrete are assumed to behave as one
body.1,2 Equilibrium and local compatibility equations are
then derived for design purposes. However, in an unbonded
member, the concrete and tendon deform independently, so
one must consider their global compatibility for any analyt-
ical purposes.3,4 In other words, in the unbonded concrete
■ An experimental program was conducted to inves- member, the overall compression of the concrete is equal
tigate the behavior of indeterminate prestressed to the total tendon force along the location of the tendon.
concrete T beams with two different unbonded Therefore, any computational method to calculate the
tendon profiles. Three T beams, at half-scale size, unbonded tendon stresses at various behavioral stages has
were constructed for the test program. The control become a point of interest for researchers and many have
specimen had no eccentricity at the interior support, proposed methods of estimating the value of those stresses,
and the other specimens had eccentricity at the as follows.
interior support.
One of the common methods, load balancing, was proposed
■ The experimental and theoretical results indicate that in 1972 by Lin and Thornton3 to analyze prestressed con-
the specimens with eccentricity at the interior sup- crete members. In 1990, Aalami5 presented detailed concepts
port had a greater number of cracks at the tension and procedures for load balancing with numerical analysis.
zone of the support than the control specimen. This
finding was explained by hyperstatic moment reduc- In 1993, Cohn and Lounis7 presented research on an opti-
ing the negative moment in the control specimen. mum limit design method for prestressed concrete struc-
tures. In 1997, Kodur and Campbell8 evaluated and analyzed
■ The authors conclude that the effect of hyperstatic moment distribution in a two-span, continuous prestressed
moment should be accounted for in the calculation concrete beam, and Lopes et al.9 investigated the degree of
of concrete stress at the service level for continuous distribution in prestressed concrete beams by comparing
prestressed concrete beams with unbonded tendons their experimental data with the corresponding theoretical
and a vertically unconfined support condition. values.

PCI Journal | July–August 2022 79

 In 2010, Nie et al.12 presented an analytical and numerical indeterminate prestressed concrete T beams with two different
model of prestressed, continuous steel-concrete composite unbonded tendon profiles.
beams and Zhou and Zheng13 developed a plastic design
method and formulated the degree of movement distribution T beam specimen 1 (TB1) was designed with a double-harped
for prestressed concrete beams with unbonded tendons by tendon with no eccentricity at the interior support (Fig. 1).
investigating experimental results. T beam specimens 2a and 2b (TB2a and TB2b) had the same
tendon profile, but the eccentricity δ1 was 2.7 in. (68.6 mm)
In 2013, Lou et al.14 analyzed the flexural behavior of continu- at the interior support (Fig. 2). TB1 was a control specimen
ous prestressed concrete beams using a finite-element model. without any eccentricity at the interior support, whereas TB2a
In another paper published the next year, Lou et al.15 analyt- and TB2b were designed to have eccentricity at the interior
ically examined hyperstatic reactions and moment distribu- support.
tion of continuous concrete beams prestressed with external
carbon-fiber-reinforced polymer tendons. Figures 1 and 2 show the dimensions and reinforcement
details for the three specimens. All specimens were 360 in.
In two papers, coauthors Witchukreangkrai, Aravinthan, and (9144 mm) long and 10 in. (254 mm) deep (including 2.5 in.
Mutsuyoshi16,17 proposed a method to enhance the flexural [63.5 mm] of slab depth); the specimens were 13.5 in.
strength of beams that use the highly eccentric tendons. The (343 mm) wide at the top and 6.5 in. (165 mm) wide at
primary moment given by the eccentricity of the prestressing the bottom web. High-strength concrete with compressive
caused the beam to camber upward at the interior supports, strength f c′ of 7.25 ksi (50 MPa) was used to construct all
but restraints at the interior support prevented the upward de- three beams.
flection, which created a hyperstatic moment. This phenome-
non resulted in reactions that were different from the reactions To construct each specimen, four 240 in. (6096 mm) long,
of a similar beam that had no applied prestressing. ASTM A61518 Grade 60 (414 MPa) no. 3 (10M) bars with
a yield strength of bonded reinforcement Fy of 60 ksi
In general, hyperstatic moment occurs when a pin or roller con- (420 MPa) were placed on sawhorses in pairs. Given the
dition of a two-span prestressed concrete beam is used at the in- concrete cover at both ends, the total length for each pair
terior support. There have been few investigations, however, of was 358 in. (9093.2 mm). (These reinforcing bars are
hyperstatic action and mechanism in indeterminate prestressed shown as four no. 3 at interior support in Fig. 1 and 2). To
concrete beams with unconfined interior conditions.4 Further- provide greater flexural strength to the beams, one 120 in.
more, despite the frequent use of a T-shaped section in lieu of a (3048 mm) long, Grade 60 no. 3 bar was located in each
rectangular section, most experimental research on continuous pair. (This reinforcing bar is shown as one no. 3 in Fig. 1
prestressed concrete beams has focused on the behavior of the and 2). Sixty-one 8 × 4.5 in. (203.2 × 114.3 mm) Grade 60
rectangular section. The center of gravity of concrete is much no. 3 stirrups were inserted and spaced 6 in. (152.4 mm)
closer to the top flange of T beams so that tendons are mostly on center. Twelve 11.5 in. (292.1 mm) long, Grade 60 no. 3
located below the center of gravity of concrete. bars were laid and centered perpendicularly at every 30 in.
(762 mm) along the rectangular cage. The unbonded tendons
In the research described in this paper, a series of experiments were designed with double-harped profiles for all spans. The
was conducted using two-span, prestressed high-strength ¼ in. (6.35 mm) diameter, Grade 250 (1725 MPa) sev-
concrete T beams with two different unbonded tendon pro- en-wire strand with an ultimate tensile strength of bonded
files. The experiments were done to observe the structural reinforcement Fu of 250 ksi (1750 MPa) was cut to 408 in.
behavior of the beams that were stressed by the prestressing (10,363.2 mm), cleaned to remove surface rust, and then
strand as well as the resulting crack patterns under the same coated lightly with grease to facilitate insertion of the strand
loading condition. Two seven-wire strands with a draped into the vinyl tube and reduce the friction losses between
tendon profile were provided for each beam. The test program the strand and the vinyl. After placing the strand in the vinyl
analyzed a case where tendon eccentricity was present at the tube, the tendon was inserted inside the cage according to
interior support and a case where it was not. The two verti- the tendon profile.
cal point loads were applied to both spans simultaneously.
Deflection and stress at the prestressing reinforcement were Two tendon profiles were used for the three specimens. For
then measured. Based on the support reaction, the hyperstatic TB1, the control specimen, a tendon end was placed at the
moment at the interior support was analyzed, which was in an centroid (located at the center of gravity of concrete ) of the
unconfined condition. specimen and then the tendon profile was slightly inclined
until it reached a distance of 71 in. (1803 mm) beyond the
Experimental program first (left) end of the beam. From that point, the tendon was
extended horizontally for 38 in. (965 mm) (Fig. 1).
Design of test specimens
The same procedure was repeated for the next span. The
The test program included experiments on three approximate- tendon profile of two specimens (TB2a and TB2b) was the
ly half-scale specimens to understand the overall behavior of same, except that it was located 2.7 in. (68.6 mm) above the

80 PCI Journal | July–August 2022

 13.5 13.5

4 no.3 8 no.3

Two ¼ diameter
3 no.3 4 no.3
Two ¼ diameter

7 wire strands
7 wire strands
7.5 2.5

7.5 2.5
c.g.c. c.g.c.
5.8

5.8
6.5 6.5
2.7

1.5

1.5
Section at interior support Section at midspan
c.g.c. 71 P P P P
38 71 71 38 71

no.3 Two ¼ diameter

1 no.3 8 4 7 wires strands
2 no.3
120 no.3 no.3 120 6
180 180
360

Figure 1. Dimensions and reinforcement details for the control specimen (TB1). Note: Units are in inches. c.g.c. = center of gravity
of concrete; P = vertical load. No. 3 = 10M; 1 in. = 25.4 mm.

13.5 13.5
Two ¼ diameter

Two ¼ diameter
4 no.3 8 no.3

3 no.3 4 no.3
7 wire strands

7 wire strands
7.5 2.5

7.5 2.5

c.g.c. c.g.c.
5.8

5.8

6.5 6.5
2.7

1.5

1.5

Section at interior support Section at midspan

c.g.c. 71 P P P P
38 71 71 38 71

no.3 Two ¼ diameter

1 no.3 8 4 7 wires strands
2 no.3
120 no.3 no.3 120 6
180 180
360

Figure 2. Dimensions and details of reinforcement of T beam specimens 2a and 2b (TB2a and TB2b). Note: Units are in inches.
c.g.c. = center of gravity of concrete; P = vertical load. No. 3 = 10M; 1 in. = 25.4 mm.

PCI Journal | July–August 2022 81

 centroid of the T beam at the interior support, allowing 1.5 in. under the beam where the maximum theoretical deflection
(38.1 mm) of concrete cover from the top (Fig. 2). would occur. All LVDT wires were attached to the bottom of
the specimen at the north and south spans. All load cells and
Twelve strain gauges were placed on the reinforcing bars LVDTs were connected to the dual display multimeters, load
of each beam (Fig. 3). Six strain gauges (SG1, SG2, SG3, cell indicators, and data acquisition system.
SG9, SG10, and SG11) were installed on the reinforcing
bars in each span, where maximum positive moment at each An electric hydraulic jack was used to stress the strands up to
span was expected, with a gauge length of 8 in. (203.2 mm). 70% of the specified tensile strength of prestressing steel 0.7fpu
Three strain gauges (SG5, SG6, and SG7) were placed on (189 or 6.8 kip [1.3 or 30.2 kN]) after seating loss. To achieve
the reinforcing bars at the interior support, which yielded the initial stress of 0.7fpu, the prestressing load was incremen-
maximum negative bending moment. SG5 and SG7 were tally applied at 900, 3500, and 7000 lb (4, 15.6, and 31.1 kN)
positioned 8 in. (203 mm) from SG6. SG4 and SG8 were (jacking load per each strand). At each increment, the prestress-
placed on reinforcing bars at the bottom of the beam, where ing process was stopped while readings from the load cells
the second point load would be applied, to measure the ten- were recorded manually. The readings included prestressing
sion stress, and SG12 was located on a reinforcing bar at the force in both strands, deflection in the north and south spans,
bottom of the interior support to measure the compression reactions, and strand elongation. In addition, the jack pressure
stress at that point. data from the hydraulic jack dial were documented. Steel plates
measuring 6 × 6 × 0.5 in. (152.4 × 152.4 × 12.7 mm) were
Loading and measurement placed between the specimen and load cell. Because of the nar-
row web width, there was an interference problem between the
After the beam was placed in the loading frame (Fig. 4), east and west load cells, so a circular hollow bar was inserted
the 12 strain gauges were connected to the data acquisition between the load cells and the steel plate (Fig. 4).
system. A special load cell frame was designed and fabricated
to house the load cells at the supports. All three supports— After the strands were tensioned, the jack was released and
one each at the two ends and one in the interior—had the the tendon force was maintained by the chucks; then the
same support condition, which was a steel roller (Fig. 4). beam was ready to be loaded vertically. Two actuators (55 kip
To observe the actual deflection under a given loading, two [244.7 kN] capacity) were used to apply the vertical load to
linear variable displacement transducers (LVDTs) were placed both spans simultaneously as a two-point load. The data were

North span SG7 SG6 SG5 South span

SG10 SG8 SG12 SG4 SG2

SG11 SG9 SG3 SG1
PT strands with vinyl tube
TB1
North span SG7 SG6 SG5 South span

SG10 SG8 SG12 SG4 SG2

SG11 SG9 SG3 SG1
PT strands with vinyl tube
TB2a and TB2b

Figure 3. Location of strain gauges placed on longitudinal reinforcing bars. Note: PT = post-tensioned; SG1 = strain gauge 1; SG2
= strain gauge 2; SG3 = strain gauge 3; SG4 = strain gauge 4; SG5 = strain gauge 5; SG6 = strain gauge 6; SG7 = strain gauge 7;
SG8 = strain gauge 8; SG9 = strain gauge 9; SG10 = strain gauge 10; SG11 = strain gauge 11; SG12 = strain gauge 12; TB1 = T beam
specimen 1; TB2a = T beam specimen 2a; TB2b = T beam specimen 2b.

82 PCI Journal | July–August 2022

 East-3144-10K Concrete T beam Loading frame
Load cell
North span South span

Actuator Steel jaw

with 55 kip wedged
capacity

N LVDT
Chuck body
S LVDT GL
71 109 109 71
Wast-3144-10k
Load cell 1221HL-50k-B
1221HL-50K-B Load cell 1221HL-50k-B
Load cell Load cell

East 3144 10K Load cell

South span West 3144 10K Load cell
Strand

Steel roller
c.g.c.

Chuck Jack
Steel
plate T beam

1221HL-50K-B Load cell

1221HL 50K B Load cell

Figure 4. Loading frame, load cell details, linear variable displacement transducers (LVDTs), steel jaw, and chuck. Note: Units are
in inches. c.g.c. = center of gravity of concrete; NLVDT = north-span linear variable displacement transducer; SLVDT = south-
span linear variable displacement transducer. 1 in. = 25.4 mm; 1 kip = 4.448 kN.

recorded in increments of 1 second. The loading process was cracking loads were recorded at their corresponding locations.
stopped at every 1 kip (4.448 kN) increment to record the The same procedure was conducted for all three specimens.
tendon force and reaction data. The loading was stopped when
the stress applied to the strand reached 0.7fpu. Test results

Crack propagation patterns were also recorded manually Responses of deflection versus
throughout the test. The first cracking load was recorded. After load curve
each additional loading, new cracks developed, and these were
traced and labeled with a number at the top of the crack. All Figure 5 presents for each specimen the total deflection

PCI Journal | July–August 2022 83

 Figure 5. Deflection versus applied load curve. Note: NLVDT = north-span linear variable displacement transducer; SLVDT =
south-span linear variable displacement transducer; STROKE1 = linear actuator stroke at midspan (south); STROKE2 = linear
actuator stroke at midspan (north); TB1 = T beam specimen 1; TB2a = T beam specimen 2a; TB2b = T beam specimen 2b. 1” = 1
in. = 25.4 mm; 1 kip = 4.448 kN.

versus applied load curves at two locations: midspan and asymmetric behavior indicates that the tendon behavior might
71 in. (1803 mm) from the exterior supports. The deflections be asymmetrical. The reason why deflection of the north span
were measured at the locations of the two LVDTs and the in TB2b was lower than that of the south span might be that
strokes (at the actuators), and the tests were stopped before the south span had a fixed end of tendons, even though the
the beam failed. In the cases of TB1 and TB2a, the north span tendon profiles of TB2b were installed symmetrically. The
exhibited greater deflection than the south span when sub- differences between the deflections recorded from the LVDTs
jected to the same applied load, whereas the deflection at the and the strokes can be attributed to the fact that continuity in
south span was greater than at the north span for TB2b. This the two-span beams shifted the maximum deflection position

84 PCI Journal | July–August 2022

from the midspan toward the exterior supports where the load that the larger load capacity is anticipated due to the larger
at the first harp was applied. negative moment capacity at the beam section near the interi-
or support.
The prestressing tendon profile did not have a significant ef-
fect on the initial stiffness of the specimens for the following Behavior of bonded reinforcing bars
reasons:
Figure 6 shows relationships for the three specimens between
• The concrete stress produced by the prestressing balanced the applied load and strain on bonded longitudinal reinforcing
some of the stress of the applied load and correspond- bars. The detailed locations of strain gauges were illustrat-
ingly reduced the deflection, resulting in essentially no ed in Fig. 3. The reinforcing bar strains of TB1 and TB2b
cracking for the specimens. at SG1, located at the starting point of the first harp at the
south span, were 2832 and 2492 με, respectively, under the
• The initial stiffness, which is mostly based on the sec- applied load of 14.7 kip (65.4 kN). For TB2a, the reinforcing
tional properties of the structural member, was the same bar yielded at a load of 14.8 kip (65.8 kN) and a maximum
for all specimens. strain of 2291 με, which yielded perfectly plastic deformation.
At the SG3 position, the yielding point of the reinforcing
Even though testing of the beams was stopped before their bar in TB1 and TB2b was detected at approximately 14 kip
ultimate limit state was reached, the larger measured load for (62.3 kN) with a strain of 2820 με, whereas the reinforc-
TB1 than for TB2a and TB2b at the same deflection signals ing bar in TB2a yielded at the higher load (approximately

Figure 6. Reinforcement strain results. Note: SG1 = strain gauge 1; SG2 = strain gauge 2; SG3 = strain gauge 3; SG4 = strain gauge
4; SG6 = strain gauge 6; SG7 = strain gauge 7; SG8 = strain gauge 8; SG10 = strain gauge 10; SG12 = strain gauge 12; TB1 =
T beam specimen 1; TB2a = T beam specimen 2a; TB2b = T beam specimen 2b. 1 kip = 4.448 kN.

PCI Journal | July–August 2022 85

 16.5 kip [63.4 kN]). In TB1 and TB2b, the reinforcing bar at the interior support—upward, which caused an increase
strain deformation curves at SG4, placed at the second harp in the compression stress at the bottom fiber of the beam at
point of the south span, were almost identical to each other that location, where no cracks developed on either the top
and did not reach steel yielding. Strain gauges SG6 and SG7 or bottom surfaces.
were installed at the interior support at the tension zone. At
the location of SG6, the reinforcing bar in TB1 yielded at Stress at prestressing reinforcement
a load of 15.1 kip (67.2 kN) and the maximum strain was
2883 με. The strain slopes of TB1 and TB2a recorded by SG7 The stress in the unbonded prestressing reinforcement fps at
were almost identical to one another: reinforcement in TB1 nominal flexural strength can be calculated by the method set
and TB2a yielded at loads of 13.5 and 13.2 kip (60.1 and forth in section 20.3.2.4.1 of the American Concrete Institute’s
58.7 kN), respectively, and the maximum strains were 2318 Building Code Requirements for Structural Concrete (ACI 318-
and 2354 με, respectively. 19) and Commentary (ACI 318R-19),19 as shown in Table 1.

SG12 was placed exactly at the interior support in the com- Determining the stress in the prestressing strand was com-
pression zone. Despite the fluctuation in some of the data, the plicated because of the many variables and parameters. In
slopes of the lines were close to each other. The applied load addition, because the test was stopped before failure, the final
and reinforcing bar strain for TB1 were 14.5 kip (64.5 kN) point was not the stress in the unbonded prestressing steel at
and -674 με, respectively. For TB2b, they were 13.7 kip nominal flexural strength. The decision to stop before failure
(60.9 kN) and -667 με, respectively. The steel bars at the was due to safety issues and because the allowable deflec-
interior support yielded first. Strain gauge data for TB2a were tion limit had been exceeded. Because the distance from the
not available. extreme compression fiber to the centroid of reinforcement
steel at the interior support was different for TB1 and TB2a,
For all specimens, all of the nonprestressed steel bars at the slopes of the applied load–strand tension relationship were
the point where the first load was applied yielded at a also different for the two specimens (Fig. 7). Specifically, the
comparable loading state. This behavior can be due to the slope of the applied load–strand tension relationship for TB2a,
maximum positive moment occurring at the first point load which had a larger tendon profile depth, was greater than the
from external support for the north and south spans and slope for TB1. The first cracking in TB2a and TB2b occurred
the maximum negative moment occurring at the interior at a lower applied load than the first cracking in TB1; howev-
support. In both spans, the steel bars at the second point er, the unbonded tendons of TB2a and TB2b seemed to take
load did not yield. The steel bars at the bottom of the inte- over the larger external load after cracking. The data from a
rior support did not reach the compressive yielding point, load cell in TB2b were almost the same as those from TB1
and as a result, the prestressing had no noticeable effect at up to 14 kip (62.3 MPa), after which monitoring was stopped.
the compression zone of the interior support. The primary The other load cell of TB2b malfunctioned and could not be
moment due to prestressing deflected the beam—including read from the beginning.

Table 1. Specified stress in unbonded prestressing reinforcement fps at nominal flexural strength for unbonded
tendons
ln
fps, ksi
h

fc′
fse + 60,000 +
(100ρ ) ρ

≤35 The least of

fse + 60,000

fpy

fc′
(
fse + 10,000 + 300ρ
ρ )
>35 The least of
fse + 30,000

fpy

Source: Equation from American Concrete Institute (ACI) 318-19 section 20.3.2.4.1.
Note: Aps = area of prestressing longitudinal tension reinforcement; b = width of compression face of member; dp = distance from extreme compression
fiber to centroid of prestressing reinforcement; fc = specified compressive strength of concrete; fpy = specified yield strength of prestressing reinforce-
ment; fse = effective prestress; h = height of member; ln = length of clear span measured from face to face of supports; ρp = ratio of Aps to bdp. 1 ksi =
6.985 MPa.

86 PCI Journal | July–August 2022

Figure 7. Applied load–strand tension relationships for specimens TB1 and TB2a Note: TB1 = T beam specimen 1; TB2a = T beam
specimen 2a. 1 kip = 4.448 kN.

North span, east view

North span, west view

Center of specimen, top view

Lesser top cracks at interior support

South span, east view

South span, west view

Figure 8. Crack patterns for T beam specimen 1 (TB1).

PCI Journal | July–August 2022 87

 North span, east view

North span, west view

Center of specimen, top view

More top cracks at interior support

South span, east view

South span, west view

Figure 9. Crack patterns for T beam specimen 2a (TB2a).

Crack distribution and propagation were not pin or roller connected, hyperstatic moment and
reaction at the initial stages were calculated and compared
Figures 8 and 9 show that cracks in TB1 and TB2a occurred with the experimental data. The direct method procedure
first at the tension face (top) of the interior support and then was initiated by prestressing strands that were converted into
cracking started at the tension face (bottom) at the midspan corresponding equivalent loading and applied to the concrete.
in both the north and south spans. The cracks in the interior Figure 10 shows the equivalent loading due to the prestress-
supports distributed uniformly from both sides, but there were ing tendon profile in the concrete of TB1 with its correspond-
fewer cracks in TB1 than in TB2a and TB2b. The reason for the ing primary, hyperstatic, and combined moments. Note that
differences in the two TB2 specimens’ behavior was the posi- unlike hyperstatic moment (actual beam internal moment due
tive hyperstatic moment induced in the supports. This moment to external reaction), primary moment is not an actual internal
was greater for TB1 than for TB2a and TB2b and, as a result, moment of the beam but is an assumed effect on concrete only
there was a greater reduction of negative moment in TB1. On when tendons are replaced by equivalent load. In real-life sce-
the other hand, the distribution of the cracks at the north and narios, beams would be seated on top of the interior support
south spans ranged from the first point load applied, which was and vertical upward restraint would not exist. For this reason,
71 in. (1803.4 mm) from the external support in both spans, to the procedure was also applied with no interior support reac-
the midspan of each span. This behavior was due to the maxi- tion to account for zero hyperstatic moment (Tables 2 and 3).
mum positive moment occurring at that location. Figure 11 displays the equivalent loading due to prestressing
tendon profile in the concrete of TB2a and TB2b with its cor-
Investigation of hyperstatic actions responding primary, hyperstatic, and combined moments.

Direct method Comparison of theoretical

and experimental results
In general, hyperstatic moments of continuous indeterminate
structure are determined through the direct method (also Theoretical results were first compared based on the roller
called load-balancing method).5 Although the interior supports support condition with vertical restraint with the experimental

88 PCI Journal | July–August 2022

 c.g.c.
L1 L2 L1 L1 L2 L1

F F
δ 2θ1
1

Fsinθ1 Fsinθ1
H2
Fcosθ1 Fcosθ1
H1 H1
 
1  tan 1   (radian)
 L1 
H 1  F1 H 2  2 F 1

A D E B C

Primary
M1

M 1D ( 5.1) M 1E M 2B ( 4.6)

Hyperstatic
M2

M bB ( 4.6)

Prestressed
Mb
in concrete
M bE ( 2.3)
M bD ( 3.3)

Figure 10. Equivalent load to concrete beam and corresponding moment diagrams primary, hyperstatic, and combined mo-
ments in concrete of T beam specimen 1 (TB1). Note: Units are in kip-ft. F = prestressing force; H1 = equivalent concentrated load
calculated by θ1; H2 = equivalent concentrated load calculated by 2θ1; L1 = distance between the end support and the concentrat-
ed loading point; L2 = distance between two concentrated loading points; Mb = combined moment; MbB = combined moment at
point B; MbD = combined moment at point D; MbE = combined moment at point E; M1 = primary moment; M1D = primary moment at
point D; M2 = hyperstatic moment; M2B = hyperstatic moment at point B; δ = distance between the line of prestressing force and
the lowest sag point of tendon profile; θ1 = angle of the profile at H1; θ2 = angle of the profile at H2. 1 kip-ft = 1.356 kN-m.

Table 2. Comparison of theoretical and experimental changes at the interior support immediately after transfer

Experimental reaction
Theoretical reaction
reduction/theoretical
Theoretical reduction,* lb
Experimental reaction reduction, %
Specimen hyperstatic
With Without reaction reduction, lb With Without
moment, kip-ft
vertical vertical vertical vertical
restraint restraint restraint restraint

TB1 4.63 617 0 546 88.4 Infinite

TB2a 2.91 388 0 282 72.7 Infinite

TB2b 2.91 388 0 365 94.1 Infinite

Note: TB1 = T beam specimen 1; TB2a = T beam specimen 2a; TB2b = T beam specimen 2b. 1 lb = 4.448 N; 1 kip-ft = 1.356 kN-m.

PCI Journal | July–August 2022 89

 Table 3. Comparison of theoretical and experimental changes at the interior support after 24 hours

Experimental reaction
Theoretical reaction
reduction/theoretical
Theoretical reduction, lb
Experimental reaction reduction, %
Specimen hyperstatic
With Without reaction reduction, lb With Without
moment, kip-ft
vertical vertical vertical vertical
restraint restraint restraint restraint

TB1 4.63 617 0 549 88.9 Infinite

TB2a 2.91 388 0 311 80.2 Infinite

TB2b 2.91 388 0 241 62.1 Infinite

Note: TB1 = T beam specimen 1; TB2a = T beam specimen 2a; TB2b = T beam specimen 2b. 1 lb = 4.448 N; 1 kip-ft = 1.356 kN-m.

c.g.c.
L1 L2 L1 L1 L2 L1

F  3  2 2 F
δ θ2 δ1 θ2

Fsinθ1 Fsinθ1
H4
Fcosθ1 Fcosθ1
H1 H3
     1 
1  tan 1   (radian)  2  tan 1  
L
 1  L1 
H 1 =Fθ1 H 3 =Fθ2 H 4 =Fθ3 = 2Fθ2

A D E B C

M 1B ( 3.1)
Primary
M1

M 1D ( 5.0) M 1D M 2B ( 2.9)

Hyperstatic
M2
M bB ( 6.0)

Resulting
Mb
in concrete
M bE ( 3.2)
M bD ( 3.8)

Figure 11. Equivalent load to concrete and corresponding moment diagrams for primary, hyperstatic, and combined moments in
concrete of T beam specimens 2a and 2b (TB2a and TB2b). Note: Units are in kip-ft. F = prestressing force; H1 = equivalent con-
centrated load calculated by θ1; H3 = equivalent concentrated load calculated by θ2; H4 = equivalent concentrated load calculated
by θ3; L1 = distance between the end support and the concentrated loading point; L2 = distance between two concentrated load-
ing points; Mb = combined moment; MbB = combined moment at point B; MbD = combined moment at point D; MbE = combined
moment at point E; M1 = primary moment; M1B = primary moment at point B; M1D = primary moment at point D; M2 = hyperstatic
moment; M2B = hyperstatic moment at point B; δ = distance between the line of prestressing force and the lowest sag point of
tendon profile; δ1 = distance between the line of prestressing force and the highest point of tendon profile at center; θ1 = angle of
the profile at H1; θ2 = angle of the profile at H3; θ3 = angle of the profile at H4. 1 kip-ft = 1.356 kN-m.

90 PCI Journal | July–August 2022

results. Tables 2 and 3 compare the theoretical5 and exper- Conclusion
imental reaction values at the interior support before the exter-
nal loading was applied. The theoretical reaction is simply de- Approximately half-scale tests were carried out to analyze
termined from hyperstatic moment, with and without vertical and evaluate the structural performance of indeterminate
restraint. The experimental data recorded for TB1 show the prestressed concrete T beams with distinctive tendon profiles.
closest correspondence to the theoretical calculations (88.4% The following conclusions were reached based on the investi-
and 88.9% at the time of stressing and after 24 hours, respec- gation and comparison of experimental and theoretical results:
tively). Unlike TB1, TB2a and TB2b had eccentricity at the
interior support, which might have caused the strand to take • The tendon profile did not have a noticeable effect on the
more time to redistribute the internal forces. Cold welding deflection of the unbonded prestressed beam. The eccen-
(or contact welding) was considered the primary reason why tricity at the interior support reduced the number of cracks
hyperstatic moment at the interior support occurred. Even at the external load-induced tension zone (top) at the interi-
though the interior support was not restrained, the self-weight or support, affected by the effect of the hyperstatic action,
of the beam worked as a restraint so that the interior support which increased the positive moment at that location.
of the beam became a partial roller condition. The primary
moment caused the beam at the interior support to deflect • Because the maximum positive moment developed at the
upward, but the hyperstatic moment tried to reduce that first point load from the external support for the north and
upward movement. Once the hyperstatic moment occurred at south spans and the maximum negative moment occurred
the interior support, the hyperstatic moment value could be at the interior support, the reinforcing bars at the interior
interpolated at any other point along the span of the T beam. support yielded first and the steel bars at the point where
the first load was applied yielded at a comparable loading
Figure 12 shows reaction-deflection curves for all three state. In both spans, the steel bars at the second point
specimens. As mentioned earlier, the deflection was mea- load from the external support did not yield. Moreover,
sured at a horizontal distance of 71 in. (1803 mm) from the the steel bars at the bottom of the interior support did not
exterior supports and the average value of two deflections reach the compressive yielding point, which indicated
was used for plotting the reaction-deflection curve at the that the prestressing did not have a noticeable effect at the
interior support. In Fig. 12, the black solid lines represent compression zone (bottom) of the interior support. The
the reaction at each support—interior support, south support, moment in concrete from the equivalent load replacing
and north support—and the black dashed lines were calcu- tendons resulted in an increase in the compression stress
lated from the total applied load multiplied by the proportion at the top of TB2a and TB2b at the interior support,
4.23/6.23, which is obtained from elastic analysis of an where no cracks developed on either the top or bottom
indeterminate beam and expected to be the same until plastic concrete surface at that location prior to loading.
moment is reached. The gray dashed lines show the differ-
ences between the experimental reactions and the portions of • Because of the difference between TB1 and TB2a in
measured total actuator. As the external loading increased, terms of the distance from the extreme compression fiber
the differences decreased and approached zero, signifying to the centroid of the steel strands at the interior support,
that hyperstatic reaction decreased at the interior support. the slopes of the applied load–strand tension relation-
In addition, the differences also decreased at the south and ship for the two specimens also differed. Specifically,
north support for all three specimens. There was hyperstatic the slope of the strain–applied load tension relationship
reaction due to cold welding effect immediately before the for TB2a, which had a larger tendon profile depth, was
external loading; hyperstatic reaction then decreased as the greater than that for TB1.
loading increased and eventually approached zero, which
indicates that the cold welding effect disappeared. The • During loading, cracks occurred at the tension face of the
reason may be that cracking occurred at the interior support interior support and then cracking began in the midspan
and the hyperstatic moment tended to be almost constant in both the north and south spans. Uniformly distributed
because the restrained relative rotation no longer existed at cracks were observed at the interior support from both
the interior support. sides. Because the positive hyperstatic moment at the
interior support was greater for TB1 than for TB2a and
Kim and Kang4 reported that the hyperstatic moment in- TB2b, the negative moment at the same support was less
creased dramatically as the external load was increased for in TB1 than in the other two specimens. As a result, TB1
unbonded post-tensioned beams with three to five draped had fewer cracks than TB2a and TB2b. In contrast, the
tendons per continuous beam; however, additional experimen- cracks at the midspan of the north and south spans for all
tal research is warranted to confirm this finding. Unconfined specimens were distributed from the first point load to
interior support condition is common for continuous pre- the midspan because of the maximum positive moment
stressed concrete beams in bridge structures, and ACI 318- occurring at the zone.
1919 requires inclusion of the values of concrete stress (at the
service level) and moment demand (at the ultimate level) due • The interior support was a roller condition with no
to hyperstatic actions in the structural design. vertical restraint, and common statics suggest that

PCI Journal | July–August 2022 91

 Portion of measured total actuator load
Measured reaction
Difference

Figure 12. Deflection versus support reaction curves for all three specimens. Note: TB1-IS = T beam specimen 1 at interior sup-
port; TB1-NS = T beam specimen 1 at north support; TB1-SS = T beam specimen 1 at south support; TB2a-IS = T beam specimen
2a at interior support; TB2a-NS = T beam specimen 2a at north support; TB2a-SS = T beam specimen 2a at south support; TB2b-
IS = T beam specimen 2b at interior support; TB2b-NS = T beam specimen 2b at north support; TB2b-SS = T beam specimen 2b
at south support. 1 in. = 25.4 mm; 1 kip = 4.448 kN.

92 PCI Journal | July–August 2022

 hyperstatic moment would therefore not occur at the 6. Alkhairi, F. M., and A. E. Naaman. 1993. “Analysis of
interior support. However, differences between the Beams Prestressed with Unbonded Internal or External
theoretical and experimental reaction values were Tendons.” Journal of Structural Engineering 119 (9):
observed at the interior support after prestressing. The 2680–2700.
observed behavior indicates that hyperstatic moment
occurred at the interior support due to the cold welding 7. Cohn, M. Z., and Z. Lounis. 1993. “Optimum Limit De-
effect and partial restraint existed. For all three spec- sign of Continuous Prestressed Concrete Beams.” Journal
imens, hyperstatic reaction then decreased as loading of Structural Engineering 119 (12): 3551–3570.
increased and this reaction eventually approached
zero, which may imply that the contact welding almost 8. Kodur, V. K. P., and T. I. Campbell. 1997. “Evaluation of
disappeared. Based on the current study only, the Moment Redistribution in a Two-Span Continuous Pre-
effect of hyperstatic moment may need to be included stressed Concrete Beam.” ACI Structural Journal 93 (6):
in the calculation of concrete stresses at the service 721–728.
level, but it may not need to be included in the process
of ultimate strength design for beams with vertically 9. Lopes, S. M. R., J. Harrop, and A. E. Gamble. 1997.
unconfined supports. That is, the balancing moment “Study of Moment Redistribution in Prestressed Con-
Mb consisting of primary and (secondary) hyperstatic crete Beams.” Journal of Structural Engineering 123 (5):
moment (M1 + M2) should be accounted for during the 561–566.
allowable stress design. Additional research for contin-
uous prestressed concrete beams would be necessary to 10. Barr, P. J., B. M. Kukay, and M. W. Halling. 2008.
test this hypothesis. “Comparison of Prestress Losses for a Prestress Concrete
Bridge Made with High-Performance Concrete.” Journal
Acknowledgments of Bridge Engineering 13 (5): 468–475. https://doi.org/10
.1061/(ASCE)1084-0702(2008)13:5(468).
The work presented in this paper was funded by California
State University, Fullerton, and the Korea Foundation of 11. Ayoub, A., and F. C. Filippou. 2010. “Finite-Element
Nuclear Safety (grant 2003007-0120-CG100). The views ex- Model for Pretensioned Prestressed Concrete Girders.”
pressed are those of authors and do not necessarily represent Journal of Structural Engineering 136 (4): 401–409.
those of the sponsors.
12. Nie, J., M. Tao, C. S. Cai, and S. Li. 2011. “Analytical
References and Numerical Modeling of Prestressed Continuous
Steel-Concrete Composite Beams.” Journal of Structural
1. Tadros, M. K., N. Al-Omaishi, S. J. Seguirant, and J. Engineering 137 (12): 1405–1418.
G. Gall. 2003. Prestress Losses in Pretensioned High-
Strength Concrete Bridge Girders. NCHRP (National 13. Zhou, W., and W. Z. Zheng. 2010. “Experimental Re-
Cooperative Highway Research Program) report 496. search on Plastic Design Method and Moment Redistri-
Washington, DC: Transportation Research Board. https:// bution in Continuous Concrete Beams Prestressed with
onlinepubs.trb.org/onlinepubs/nchrp/nchrp_rpt_496.pdf. Unbonded Tendons.” Magazine of Concrete Research
62 (1): 51–64.
2. Kang, T. H.-K., and M. Ibrahim Ary. 2012, “Shear-
Strengthening of Reinforced and Prestressed Concrete 14. Lou, T., S. M. R. Lopes, and A. V. Lopes. 2013. “Flexural
Beams Using FRP: Part II—Experimental Investigation.” Response of Continuous Concrete Beams Prestressed
International Journal of Concrete Structures and Materials with External Tendons.” Journal of Bridge Engineering
6: 49–57. https://doi.org/10.1007/s40069-012-0005-0. 18 (6): 525–537.

3. Lin, T. Y., and K. Thornton. 1972. “Secondary Moment 15. Lou, T., S. M. R. Lopes, and A. V. Lopes. 2014. “External
and Moment Redistribution in Continuous Prestressed CFRP Tendon Members: Secondary Reactions and Mo-
Concrete Beams.” PCI Journal 17 (1): 8–20. https://doi ment Redistribution.” Composites: Part B 57: 250–261.
.org/10.15554/pcij.01011972.8.20.
16. Witchukreangkrai, E., H. Mutsuyoshi, and T. Aravinthan.
4. Kim, K., and T. H.-K. Kang. 2019. “Experiments on 2003. “Secondary Moment and Moment Redistribution in
Continuous Unbonded Post-Tensioned Beams with 2,400 a Two-Span Continuous PC beam with Large Eccentrici-
MPa (350 ksi) Strands.” ACI Structural Journal 116 (5): ty.” Japanese Concrete Institution 25 (2): 775–780.
125–136.
17. Aravinthan, T., E. Witchukreangkrai, and H. Mutsuyoshi.
5. Aalami, B. O. 1990. “Load Balancing: A Comprehen- 2005. “Flexural Behavior of Two-Span Continuous Pre-
sive Solution to Post-Tensioning. ACI Structural Journal stressed Concrete Girders with Highly Eccentric External
87 (6): 662–670. Tendons.” ACI Structural Journal 102 (3): 402–411.

PCI Journal | July–August 2022 93

 18. ASTM International. 2016. Standard Specification for Mb = balancing moment
Deformed and Plain Carbon-Steel Bars for Concrete
Reinforcement. ASTM A615/A615M-16. West Consho- MbB = balancing moment at point B
hocken, PA: ASTM International. https://doi.org/10.1520
/A0615_A0615M-16. MbD = balancing moment at point D

19. ACI (American Concrete Institute) Committee 318. 2019. MbE = balancing moment at point E
Building Code Requirement for Structural Concrete (ACI
318-19) and Commentary (ACI 318R-19). Farmington M1 = primary moment
Hills, MI: ACI.
M1B = primary moment at point B
Notation
M1d = primary moment at point D
Aps = area of prestressing longitudinal tension reinforce-
ment M2 = hyperstatic moment

b = width of compression face of member M2B = hyperstatic moment at point B

dp = distance from extreme compression fiber to centroid P = vertical load

of prestressing reinforcement
δ = distance between the center of gravity of concrete
f c′ = specified compressive strength of concrete and the lowest sag point of tendon profile

fps = stress in unbonded prestressing reinforcement δ1 = distance between the center of gravity of concrete
and the highest point of tendon profile at center
fpu = specified tensile strength of prestressing reinforce-
ment θ1 = angle of the profile at H1

fpy = specified yield strength of prestressing reinforce- θ2 = angle of the profile at H3

ment
θ3 = angle of the profile at H4
fse = effective prestress
ρp = ratio of Aps to bdp
F = prestressing force

Fu = ultimate tensile strength of bonded reinforcement

Fy = yield strength of bonded reinforcement

h = height of member

H1 = equivalent concentrated load calculated by θ1

H2 = equivalent concentrated load calculated by 2θ1

H3 = equivalent concentrated load calculated by θ2

H4 = equivalent concentrated load calculated by θ3

ln = length of clear span measured from face to face of

supports

L1 = distance between the end support and the concen-

trated loading point

L2 = distance between two concentrated loading points

in a span

94 PCI Journal | July–August 2022

About the authors

Uksun Kim, PhD, PE, is a  eongwon Hong, PhD, is an

S
professor in the Department of associate professor in the Depart-
Civil and Environmental Engi- ment of Safety Engineering at
neering at California State Korea National University of
University, Fullerton. Transportation and was a postdoc-
toral researcher in the Engineering
Research Institute at Seoul
National University.
Ehab N. Ballu is a former master’s
student in the Department of Civil Thomas H.-K. Kang, PhD, PE, is a
and Environmental Engineering at professor in the Department of
California State University, Architecture and Architectural
Fullerton. Engineering and the Engineering
Research Institute at Seoul
National University.

PCI Journal | July–August 2022 95

 Abstract Keywords

Approximately half-scale tests were conducted to inves- Continuous beam, intermediate prestressed concrete
tigate the behavior of indeterminate prestressed concrete beam, hyperstatic action, support condition, T beam,
T beams with two different unbonded tendon profiles. unbonded tendon.
Three T beams were employed for the tests: T beam
specimen 1 (TB1) was the control specimen and had no Review policy
eccentricity at the interior support and T beam speci-
mens 2a and 2b (TB2a and TB2b) had eccentricity at the This paper was reviewed in accordance with the Pre-
interior support. The results indicate that the specimens cast/Prestressed Concrete Institute’s peer-review pro-
with eccentricity at the interior support had a greater cess. The Precast/Prestressed Concrete Institute is not
number of cracks than the control specimen at the ten- responsible for statements made by authors of papers
sion zone of the support. This finding was explained by in PCI Journal. No payment is offered.
hyperstatic moment reducing the negative moment in the
control. The trend in the slope of the strain-applied load Publishing details
tension relationship was different in TB1 than in TB2a
due to a difference between the specimens in terms of This paper appears in PCI Journal (ISSN 0887-9672)
the distance from the extreme compression fiber to the V. 67, No. 4, July–August 2022, and can be found
centroid of reinforcement at the support. Although the at https://doi.org/10.15554/pcij67.4-01. PCI Journal
interior support was a roller with no vertical restraint, is published bimonthly by the Precast/Prestressed
hyperstatic moment at the support occurred due to a Concrete Institute, 8770 W. Bryn Mawr Ave., Suite
cold welding effect, creating a partial roller condition. 1150, Chicago, IL 60631. Copyright © 2022, Precast/
The hyperstatic action then decreased as the loading was Prestressed Concrete Institute.
increased. In light of this, the authors conclude that the
effect of hyperstatic moment should be accounted for in Reader comments
the calculation of concrete stress at the service level for
continuous prestressed concrete beams with unbonded Please address any reader comments to PCI Journal
tendons and a vertically unconfined support condition. editor-in-chief Tom Klemens at [email protected] or
Precast/Prestressed Concrete Institute, c/o PCI Journal,
8770 W. Bryn Mawr Ave., Suite 1150, Chicago, IL
60631. J

96 PCI Journal | July–August 2022