EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. (2010)

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.986

Cyclic behavior of precast segmental concrete bridge columns with

high performance or conventional steel reinforcing bars as energy
dissipation bars

Yu-Chen Ou1, ‡ , Mu-Sen Tsai2, § , Kuo-Chun Chang2, ∗, †, ¶ and George C. Lee3, 

1 Department of Construction Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan
2 Department of Civil Engineering, National Taiwan University, Taipei, Taiwan
3 Department of Civil, Structural and Environmental Engineering, University at Buffalo, Buffalo, U.S.A.

SUMMARY
The cyclic behavior of precast segmental concrete bridge columns with high performance (HP) steel
reinforcing bars and that with conventional steel reinforcing bars as energy dissipation (ED) bars were
investigated. The HP steel reinforcing bars are characterized by higher strength, greater ductility, and
superior corrosion resistance compared with the conventional steel reinforcing bars. Three large-scale
columns were tested. One was designed with the HP ED bars and two with the conventional ED bars.
The HP ED bars were fully bonded to the concrete. The conventional ED bars were fully bonded to the
concrete for one column, whereas unbonded for a length to delay fracture of the bars and to increase
energy dissipation for the other column. Test results showed that the column with the HP ED bars had
greater drift capacity, higher lateral strength, and larger energy dissipation than that with fully bonded
conventional ED bars. The column with unbonded conventional ED bars achieved the same drift capacity
and similar energy dissipation capacity as that with the HP ED bars. All the three columns showed good
self-centering capability with residual drifts not greater than 0.4% drift. An analytical model referred to as
joint bar-slip rotation method for pushover analysis of segmental columns with ED bars is proposed. The
model calculates joint rotation from the slip of the ED bars from two sides of the joint. Good agreement
was found between analytical predictions and the envelope responses of the three columns. Copyright q
2010 John Wiley & Sons, Ltd.

Received 12 May 2009; Revised 15 November 2009; Accepted 17 November 2009

KEY WORDS: precast concrete; prestressed concrete; segmental construction; bridge columns; self-
centering; high-performance steel reinforcing bars

∗ Correspondence to: Kuo-Chun Chang, Department of Civil Engineering, National Taiwan University, Taipei, Taiwan.
†
E-mail: ciekuo@ntu.edu.tw
‡
Assistant Professor.
§ Ph.D. Candidate.
¶ Professor and Chairman.
 SUNY Distinguished Professor.

Contract/grant sponsor: National Center for Research on Earthquake Engineering (NCREE) and Federal Highway
Administration (FHWA); contract/grant number: DTFH61-98-C-00094

Copyright q 2010 John Wiley & Sons, Ltd.

 Y.-C. OU ET AL.

1. INTRODUCTION

Cast-in-place construction of concrete bridges typically results in extensive damage to the on-
site environment due to the need for land for construction activities. If the construction is in
busy urban areas, it usually causes prolonged traffic disruption. Traffic disruption not only causes
inconvenience to the traveling public but also increases the consumption of fuel due to detour
and/or traffic jam and hence raises carbon dioxide emissions. Precast segmental construction by
reducing on-site construction activities and time has been proven to be an effective method to
address these issues [1]. However, very few examples of bridges with precast segmental concrete
columns can be found in regions of high seismicity such as the west coast regions of the U.S.,
Japan, [1, 2] and Taiwan due to concerns on their seismic performance.
In a typical design of a precast segmental concrete bridge column, prestressing tendons
throughout the column are stressed to apply compression forces across precast joints, which with
the compression forces from gravity loads provide required flexural and shear strengths at the
joints. Mild steel reinforcement is normally not continuous across the joints and thus contributes
little to the strengths. It is used for positioning transverse reinforcement and for shrinkage and
creep controls. Under lateral loads, the column shows a behavior same as a monolithic column
prior to opening of the joints. Once the joints open, the column exhibits a nonlinear behavior
with little energy dissipation but a small residual drift upon unloading [3–5]. To increase energy
dissipation, Wang et al. [4] and Ou et al. [5] propose to add mild steel reinforcing bars across
the joints. The bars are referred to as energy dissipation (ED) bars to emphasize their function
and to distinguish them from other mild steel bars that are not continuous across the joints. It has
been shown that the use of the ED bars can significantly increase energy dissipation. However,
when the column is subjected to a large lateral displacement, significant joint opening will occur
and likely cause premature fracture of the ED bars. Unbonding the bars for a length can decrease
the strains and delay fracture. Alternatively, fracture can be delayed by using more ductile steel
capable of absorbing greater energy before fracture. In this research, high performance (HP) steel
reinforcing bars commercially known as Enduramet 32, designated as S24100 in ASTM A955
[6], are investigated for use as ED bars. They have superior ductility capacity than conventional
carbon steel reinforcing bars. Additionally, they possess excellent corrosion resistance [7]. This
can address potential corrosion problems resulting from opening of precast joints.
The energy dissipation and residual drift of a segmental column with ED bars increase as the
strength contribution of the ED bars to the lateral strength of the column increases [5]. In Japan,
it has been found difficult to re-center the superstructure of bridge columns with a residual drift
exceeding 1/60 or with a residual displacement more than 15 cm, whichever is smaller [8]. As
a result, the 1996 Japanese seismic design specifications of highway bridges requires that the
residual drift developed at a bridge column after an earthquake not be greater than 1% [9]. Test
results show that the segmental column can maintain a residual drift not greater than 1% provided
the strength contribution of the ED bars is below approximately 35% the lateral strength of the
column [5].
Researchers have previously proposed methods to predict lateral force–displacement relationship
(pushover) of a segmental column [3, 10]. In their method, the concept of plastic hinge lengths is
adopted. The inelastic rotation of a segmental column is calculated from the inelastic curvature of
the section of a precast joint multiplied by a plastic hinge length. In other words, the curvature is
assumed constant from that joint extending a distance equal to the plastic hinge length. This method
is also referred to as monolithic beam analogy, which was proposed in an earlier work on precast

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 CYCLIC BEHAVIOR OF SEGMENTAL COLUMNS WITH HP

concrete beam-column joints [11]. However, this analogy is against experimental observations
where the inelastic rotation concentrates at the joint instead of spreading over a certain length. To
resolve such disagreement, this research proposes a method of pushover analysis referred to as
joint bar-slip rotation method in which inelastic rotation is directly computed from the slip of the
ED bars across the joint.
Three large-scale precast segmental concrete columns were tested in this research under lateral
cyclic loading to demonstrate the cyclic performance of segmental columns with HP reinforcing
bars as ED bars in comparison with conventional reinforcing bars. The joint bar-slip rotation method
was verified by comparing with the envelope responses of the three columns under cyclic loading
in terms of drift versus lateral force, ED bar strain, tendon force, and joint opening relationships.

2. SPECIMEN DESIGN AND TEST SETUP

The design and test setup of the segmental columns tested are shown in Figures 1(a) and (b),
respectively. Table I lists major design parameters. Each column consisted of four precast segments
with a hollow cross section and one precast cap beam. Prestressing tendons were anchored at the
underside of the foundation at one end and anchored on the top of the cap beam at the other
end. The tendons were unbonded to the concrete and passed through ducts in the foundation,
through hollow core of the segments and through ducts in the cap beam. Note that ‘unbonded’
does not mean the tendons are ‘ungrouted’ against corrosion. The tendons can be placed in smooth
polyethylene pipes that are not bonded to concrete and fully grouted for corrosion protection.
The tendons can also be epoxy coated for enhanced corrosion resistance. The total prestressing
force was 1042 kN, which was carried by four tendons, each consisting of two D15 seven-wire
strands. The prestressing force was determined to ensure no opening of the precast joints under
a lateral force corresponding to an assumed moderate earthquake. The prestress corresponded
to 55% tendon yield stress. This is lower than typically used for prestressed concrete. Lower
initial prestress and unbonding the tendons were intended to minimize yielding of the tendons
and hence preserve an axial force necessary for self-centering capability. The specified gravity
load was 1456 kN or 0.1 f co  A , typical for a bridge column, where f  is the specified concrete
g co
strength, 27.6 MPa, and A g is the gross cross-sectional area of the column. The gravity load was
applied to the cap beam by two hydraulic actuators and remained constant throughout the testing.
The displacement-controlled lateral cyclic loading was applied by a hydraulic actuator at one end
anchored to the reaction wall and at the other end to the cap beam. The drift levels included 0.25,
0.375, 0.5, 0.75, 1.0, 1.5, 2.0, 3.0, 4.0, 5, and 6% with each drift level repeated twice.
Conventional reinforcing bars used in this research were low-alloy steel deformed bars
conforming to ASTM A706 [12], typical for seismic design. Figure 2 shows the monotonic tension
responses of a conventional reinforcing bar and a HP reinforcing bar. The uniform elongation
before necking of the HP reinforcing bar was 48%, more than three times that of the conventional
reinforcing bar, 13%. In addition, the HP reinforcing bar had higher yield and ultimate strengths
than the conventional reinforcing bar as listed in Table II. At the time of this writing, the price of
the HP reinforcing bar is approximately six times that of the conventional reinforcing bar. The
three columns were designed with 0.5% ED bar ratio. The ED bar contribution to the column
lateral strength, denoted as ED and defined by Equation (1), was calculated to be lower than 35%
for the three columns. The 35% limitation was to ensure that the residual drift would not exceed

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 Y.-C. OU ET AL.

(a) (b)

Figure 1. (a) Specimen design and (b) test setup.

Table I. Design parameters.

Gravity load Prestressing force ED bar ratio L au
Column (kN) (kN) (%) (mm)
C5C-FB 1456 1042 0.5 0
C5C-E32 1456 1042 0.5 0
C5C-UB 1456 1042 0.5 400

1% before failure of the column as previously mentioned. The value of ED for each column will
be presented later using measured column strengths.
V − V0
ED = (1)
V
where V = lateral strength of a column with ED bars; and V0 = lateral strength of that column
without considering the ED bars.
The ED bars were inserted through corrugated steel ducts during assembling of the columns. The
ducts were grouted using a cement-based grout with an actual compression strength of 49 MPa.
The ED bars were terminated in segment S3. This was because the peak moment demand at
the joint between segments S3 and S4, denoted as joint S3-S4, was found to be lower than the
moment that would result in a compression depth lower than half the sectional diameter. The
embedded lengths of the bars into segment S3 were determined by multiplying the development
length calculated from AASHTO Section 5.11.2.1 [13] by a ratio of the predicted bar stress

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 CYCLIC BEHAVIOR OF SEGMENTAL COLUMNS WITH HP

1000

800

Stress (MPa)
600

400
Conventional
200 HP

0
0 10 20 30 40 50 60
Strain (%)

Figure 2. Monotonic tension behavior of conventional and HP ED bars.

Table II. Measured material properties.

Conventional Prestressing
Concrete Grout reinforcing bars HP reinforcing bars tendons
Compressive Compressive Yield Peak Yield Peak Yield Peak
strength strength strength strength strength strength strength strength
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
45 49 454 665 580 939 1682 1852

at that joint to the ultimate stress of the bar. For the HP ED bars, the computed development
length was further multiplied by 1.1 to take into account the higher ratio of the ultimate to yield
strengths of the bars (Table II). Columns C5C-FB and C5C-UB used conventional reinforcing
bars, whereas column C5C-E32 used the HP reinforcing bars as ED bars. The ED bars of columns
C5C-FB and C5C-E32 were fully bonded, i.e. bonded to the concrete along their entire length,
whereas those of column C5C-UB were unbonded starting from joint foundation-S1 into the
foundation for a length of 400 mm. The length is denoted as additionally unbonded length or
L au . The unbonding was done by wrapping the bars with duct tape. The 400-mm unbonded
length was to ensure no fracture of the ED bars up to 5% drift. This will be further discussed
later.

3. ANALYTICAL MODEL FOR PUSHOVER

A method referred to as joint bar-slip rotation method is proposed for pushover analysis. This
method is based on the experimental observation that most of the column rotations resulted from
joint rotations as conceptually illustrated in Figure 3. Test results of the three columns, which
will be further discussed later, showed that only a few cracks could be found on the surface of

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 Y.-C. OU ET AL.

Pg
F
sE 2 sp 2

Crack

Tendon
ED bar

sE1 sp1
h
Segment
Unbonded joints
region p
s

dp
dt

s c hi

Strain distribution at joint

Figure 3. A segmental column under lateral force.

the segments and were smaller than 0.15 mm at the peak lateral forces of the columns, indicating
small flexural rotations of the segment bodies. To be consistent with this observation, joint rotation
in the proposed method is calculated directly from the slip of the ED bars instead of assuming
a uniform distribution of the curvature calculated from the joint over a length extending to the
segment body.
In the conventional method to calculate the pushover curve of a cantilever concrete column,
the lateral displacement results from three actions, flexural rotations, slip of reinforcing bars
out of the foundation, and shear deformations. The first part, flexural rotations, consists of
elastic and inelastic components. Plastic hinge lengths typically are used in the calculation of
the inelastic components of flexural rotations and slip of reinforcing bars out of the founda-
tion. The lateral displacement in the proposed method arises from joint bar-slip rotations, shear
deformations, and elastic flexural rotations. The method is outlined as follows and illustrated in
Figure 3.
The cross sections of the precast joints of a segmental column are first analyzed under various
axial forces to obtain relationships of curvature  to other cross-sectional parameters including the
moment about centroid M, neutral axis depth c, strain of the ED bar closest to the tension face εs ,
and concrete strain at the extreme compression fiber εc . Unbonded prestressing tendons are not
explicitly modeled in these analyses. Their effects are represented by changing of the axial force
acting on the cross sections.

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 CYCLIC BEHAVIOR OF SEGMENTAL COLUMNS WITH HP

For a given curvature  at the lowest joint, the following iteration scheme is carried out to
determine the lateral force and corresponding lateral displacement of the column.
(a) At step k, the values of M, c, εs , and εc for the cross section of the lowest joint are
interpolated from the relationships previously obtained with an axial force (gravity load plus
prestressing force) from step k −1. The lateral force F is calculated by Equation (2). With
the lateral force known, moments at the other joints can be calculated from equilibrium and
other cross-sectional parameters determined by interpolation.
F = (M − Pg + M p )/ h (2)
where Pg = gravity load; M p = moment by the tendons about the centroid of the cross section
at the lowest joint;  = lateral displacement at the loading point of the column from step
k −1; and h = distance from the loading point to the lowest joint, 4050 mm for the columns
investigated herein. Note in this research the lowest joint is located at the base of the column.
(b) The amounts of slip of the ED bars closest to the tension face at the lowest joint as well
as those for the other joints are calculated. The slip results from strain penetrations of the
bar into two sides of the joint as illustrated in Figure 4. To calculate the slip, a simplified
representation of bond stress distribution is adopted and modified [14]. The bond stress is
assumed to be equal to u b1 when the maximum bar strain is smaller than or equal to the yield
strain and equal to u b2 when the maximum  bar strain is largerthan the yield strain. Values of
u b1 and u b2 herein are assumed to be 2.7 f c (MPa) and 0.5 f c (MPa), respectively. These
were obtained from testing of individual ED bar embedded in a grouted steel corrugated
duct. Figure 5 shows one of the test results with two dotted lines representing u b1 and u b2 .
With the assumed bond stresses, the bar strain distribution can be determined. The amount
of the slip S E from two sides of the joint is then calculated by the area enclosed by the bar
strain distribution as given in Equations (3)–(6).

S E = εsm (L d1 + L au ) for εsm ε y (3)

S E = ε y L d1 +(εsm +ε y )L d2 +εsm L au for εsm >ε y (4)

ub1 Ld1
y
ub 2 Ld 2
Joint

Lau sm Lau
sm

ub 2 Ld 2
y

ub1 Ld1

ED bar Bond stress Bar strain

Figure 4. Bond stress and bar strain.

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 Y.-C. OU ET AL.

30

20
ub1=2.7

Bond stress (MPa)

fc'
10

-10 ub2=0.5 fc'

-20

-30
0 2 4 6 8 10 12 14
Actuator displacement (mm)

Figure 5. Cyclic behavior of an ED bar grouted in a duct.

where εsm = maximum bar strain; ε y = yield strain; and L d1 and L d2 = lengths along which
constant bond stresses u b1 and u b2 are assumed, respectively, and are defined as follows.
f sm db f y db
L d1 =  (5)
4u b1 4u b1
( f sm − f y )db
L d2 = (6)
4u b2
where f sm = bar stress corresponding to εsm and db = bar diameter. In previous studies [4, 10],
the bond-slip behavior of an ED bar at the segment joint was modeled using the concept
in which the maximum strain of the bar was assumed uniformly distributed over a length
referred to as equivalent unbonded length. The length was assumed to be constant during
analysis. The modeling approach in this research provides a more realistic representation of
the bond-slip behavior. A non-uniform distribution of ED bar strain that is closer to the actual
strain distribution is assumed. In addition, the length over which the strain is distributed
is allowed to change during analysis. However, the finite element method developed in
previous studies [4, 10] cannot use this approach.
(c) Rotation of a joint  under consideration is calculated by
SE
= (7)
dt −c
where dt = distance from the ED bar to the compression face.
(d) The lateral displacement at the loading point of the column  is calculated by
  F F
= i (h −h i )+ + (8)
i j k vj k 

where h i = height of joint i, kvj = effective shear stiffness for segment j of the column and
defined by (9), and k = effective flexural stiffness and defined by Equation (10). The first,

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 CYCLIC BEHAVIOR OF SEGMENTAL COLUMNS WITH HP

second, and third terms of Equation (8) represent displacements resulting from joint bar-slip
rotations, shear deformations, and elastic flexural rotations, respectively.

Gj Aj
kvj = (9)
h j
3(E I )eff
k = (10)
h3

where G j = effective modulus of elasticity in shear for segment j and defined by E csj /2
(1+), where E csj = secant modulus of concrete corresponding to the extreme fiber compres-
sive strain of the core concrete at the mid height of segment j and  = poisson s ratio and
equal to 0.3; A j = area in compression at the mid height of segment j ; h j = height of
segment j ; and (E I )eff = effective flexural stiffness and assumed equal to E c Ig and 0.7E c Ig
before and after decompression of thelowest joint, respectively, where E c = modulus of
elasticity of concrete and equal to 4730 f c (MPa) [15], and Ig = moment of inertia of gross
concrete section;
(e) The increase of the strain in a tendon ε p is assumed to result only from openings of joints
and is computed by

  S pi  i (d p −ci )
ε p = ε pi = = (11)
i i hp i hp

where Spi = amount of opening of joint i at the location of the tendon under consideration,
h p = unbonded length of the tendon, i = rotation of joint i, d p = the distance from the
compression face to the tendon, and ci = neutral axial depth from the compression face
of joint i. With the change of the strain known, the new tendon force and hence the
new axial force are calculated. The new axial force is then used in the next step k +1.
Procedures from (a)–(e) are repeated until the axial force from step k −1 is close to that from
step k.
The curvature  is increased until any of the three following limit states is reached.
(1) Failure of the confined concrete: defined by εc equal to εcu , where εcu is when transverse
confining steel fractures and is defined by

1.4s f yt εsu
εcu = 0.004+ 
(12)
f cc

where s = volumetric ratio of confining steel; f yt = yield strength of confining steel; εsu =
steel strain at maximum tensile stress, 0.12; and f cc  = strength of confined concrete in

compression [16].
(2) Fracture of the ED bar: defined by when the damage index D reaches one. The maximum
strain history of the bar under consideration is first decomposed to cycles of various ampli-
tudes. The number of cycles to failure Nfi corresponding to amplitude i is calculated by
Equations (13) and (14) for conventional and HP reinforcing bars, respectively. The damage
index D is then calculated by Equation (15). Equations (13) and (14) were obtained from

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 Y.-C. OU ET AL.

low-cycle fatigue tests of the two types of reinforcing steel bars. The tests were carried out
under strain control with constant amplitude according to ASTM E606-92 [17].
εap = 0.096(2N f )−0.348 (13)
εap = 0.213(2N f )−0.443 (14)
 ni
D= (15)
i Nfi
where εap = plastic strain amplitude, N f = number of cycles to failure, n i = number of cycles
corresponding to strain amplitude i, and Nfi = number of cycles to failure corresponding to
strain amplitude i.
(3) The lateral force of the column drops below 80% of the peak value (lateral strength). This
is mainly caused by the P- effect.

4. EXPERIMENTAL RESULTS

4.1. General observations

The lateral force–displacement relationships of the three columns before fracture of the ED bars
are shown in Figures 6(a), (c), and (e) for columns C5C-FB, C5C-E32, and C5C-UB, respectively.
The lateral force was calculated from the force of the lateral actuator subtracted by the horizontal
components of the forces of the two vertical actuators. All the three columns failed due to fracture
of the ED bars that occurred at 3, 6, and 6% drifts, respectively, as shown in Figures 6(b),(d),
and (f). Note that the sudden drops of lateral forces in these plots indicate occurrences of ED bar
fracture. Test results in terms of peak values are summarized in Table III. The ultimate drifts of
the three columns were defined as the drift level at which two cycles of loading were completed
without fracture of the ED bars. Thus, the drift capacities were set as 2, 5, and 5%, respectively.
For all the three columns, flexural cracks first appeared on the east and west surfaces at 1.5% drift
and then propagated toward the centroid as appeared on the north and south surfaces (Figure 7)
until 3% drift at which the lateral forces of the columns began to decrease due to the P- effect.
At the peak drifts during cyclic loading between 1.5–3% drifts, the widths of these cracks were
measured to be between 0.06 and 0.15 mm, much smaller than the amount of corresponding joint
opening that was on the order of 10–20 mm (Figure 8). This forms the basis of the joint bar-slip
rotation method previously stated. Most of the cracks occurred on segment S1. Column C5C-E32
had a few cracks on segment S2 due to a higher lateral strength resulting from the use of the
HP ED bars (Figure 7(b)). Spalling of the cover concrete only occurred on the compression faces
around joint foundation-S1 as shown by the regions with black color and was found to be minor.
Opening of joint foundation-S1 was quite significant as shown in the left three plots of Figure 8
and increased approximately linearly with the increase of drift levels. Opening of joint S1-S2 was
much smaller than that of joint foundation-S1 as shown in the right three plots of Figure 8 and
saturated after the peak lateral forces had been reached. Opening of the other joints were found
to be negligible and hence not presented. The left three plots of Figure 9 show the strains of
the ED bars at the west walls of the columns at joint foundation-S1. Owing to large amounts of
slip between the bars and surrounding grout, the strain gauges failed between 1 and 2% drifts.
Further discussion on ED bar strains will be presented later. The right three plots of Figure 9 show

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 CYCLIC BEHAVIOR OF SEGMENTAL COLUMNS WITH HP

400 400
Experimental
Analytical
Lateral force (kN)

Lateral force (kN)

200 200

0 0

-200 -200

-400 -400
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
(a) Drift (%) (b) Drift (%)
400 400

200 200
Lateral force (kN)

Lateral force (kN)

0 0

-200 -200

-400 -400
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
(c) Drift (%) (d) Drift (%)
400 400

200 200
Lateral force (kN)

Lateral force (kN)

0 0

-200 -200

-400 -400
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
(e) Drift (%) (f) Drift (%)

Figure 6. Lateral force versus drift relationships: (a) C5C-FB: up to 2% drift; (b) C5C-FB: up
to 3% drift; (c) C5C-E32: up to 5% drift; (d) C5C-E32: up to 6% drift; (e) C5C-UB: up to 5%
drift; and (f) C5C-UB: up to 6% drift.

the forces of the tendons at the west side with the measured initial tendon forces Pi indicated.
Prestress loss were observed and mainly attributed to localized yielding around anchorages and at
the locations where the tendons were bent against the top surface of the foundation during cyclic
loading. The prestress loss decreased the total axial forces by approximately 6% that had little
influence on the performance of the columns. Figure 10(a) shows column C5C-E32 at 6% drift.
Large amount of opening of joint foundation-S1 and fracture of one of the ED bars can be clearly
seen in Figure 10(b).

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 Y.-C. OU ET AL.

Table III. Test results.

Lateral strength Ultimate drift ED Max eq Max residual
Column (kN) (%) (%) (%) (%)
C5C-FB 370 2.0 25 10 0.1
C5C-E32 386 5.0 28 15 0.4
C5C-UB 363 5.0 23 16 0.4

S3 S E N W S E N W S E N W

S2

S1

(a) (b) (c)

Figure 7. Cracks and spalling of concrete at the end of tests: (a) C5C-FB; (b) C5C-E32; and (c) C5C-UB.

4.2. Performance of HP ED bars and effects of unbonding

It can be seen from Figure 9 that for a given drift, the conventional ED bar and HP ED bar of columns
C5C-FB and C5C-E32, respectively, were subjected to a similar strain. For example, both were
subjected to a strain of approximately 2.5% at 1% drift. For column C5C-FB, six of the critical ED
bars, which were located at the east and west walls at joint foundation-S1 of the column fractured
at 3% drift. The fracture caused a significant drop in the lateral force as shown in Figure 6(b).
For column C5C-E32, the HP ED bars were able to sustain cyclic loading up to 5% drift of the
column without any fracture. Compared to column C5C-FB, the drift capacity of column C5C-E32
was greatly increased from 2–5% and hence the maximum energy dissipation in terms of equivalent
viscous damping ratio eq was improved from 10–15% (Table III). Additionally, column C5C-E32
showed a higher lateral strength. These demonstrated the capability of the HP ED bar to increase
the drift capacity, energy dissipation and lateral strength of a segmental column. Instead of using
the HP ED bars, fracture of the conventional ED bars can be delayed by unbonding the bars for a
length. By comparing the ED bar strains of columns C5C-FB and C5C-UB as illustrated in Figures
9(a) and (e), respectively, it can be seen that unbonding for 400 mm effectively decreased the ED
bar strains. For example, at 0.5% drift, the maximum ED bar strains were 1.2 and 0.26% for
columns C5C-FB and C5C-UB, respectively. Unbonding decreased the rate of increase of lateral
force as expected but had little effects on the lateral strength (Table III). It was because with such
an unbonded length, the maximum strains of the ED bars at the peak force of the column still well
exceeded the yield strain. The corresponding stresses between the two columns did not vary to an
extent that could cause a significant difference in the column lateral strengths. Unbonding slightly
decreased energy dissipation for a given drift (Figure 11) but increased maximum energy dissipation
by delaying fracture of the bars. The column with unbonding achieved similar performance to that
with the HP ED bars in terms of drift capacity and energy dissipation (Table III and Figure 11).

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 CYCLIC BEHAVIOR OF SEGMENTAL COLUMNS WITH HP

40 4
Experimental
30 Analytical 3
Opening (mm)

Opening (mm)
20 2

10 1

0 0

-10 -1
-5 -4 -3 -2 -1 0 1 2 3 4 5 -5 -4 -3 -2 -1 0 1 2 3 4 5
(a) Drift (%) (b) Drift (%)
40 4

30 3
Opening (mm)

Opening (mm)
20 2

10 1

0 0

-10 -1
-5 -4 -3 -2 -1 0 1 2 3 4 5 -5 -4 -3 -2 -1 0 1 2 3 4 5
(c) Drift (%) (d) Drift (%)
40 4

30 3
Opening (mm)

Opening (mm)

20 2

10 1

0 0

-10 -1
-5 -4 -3 -2 -1 0 1 2 3 4 5 -5 -4 -3 -2 -1 0 1 2 3 4 5
(e) Drift (%) (f) Drift (%)

Figure 8. Joint opening versus drift relationships: (a) C5C-FB, joint found-S1; (b) C5C-FB,
joint S1-S2; (c) C5C-E32, joint found-S1; (d) C5C-E32, joint S1-S2; (e) C5C-UB, joint
found-S1; and (f) C5C-UB, joint S1-S2.

However, unbonding requires additional labor work. In addition, unlike the prestressing tendons,
which are fully unbonded and hence can be replaced if corrosion occurs, it is difficult to replace
the ED bars. Better corrosion resistance is expected for the column with the HP ED bars than
with unbonded conventional ED bars, because not only the HP ED bars have superior corrosion
resistance but they are also fully grouted, which further enhances corrosion protection.

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 Y.-C. OU ET AL.

8 500
Experimental
6 Analytical 400

Force (kN)
Strain (%)

4 300

2 Pi=253 kN
200

0
100
-3 -2 -1 0 1 2 3 -5 -4 -3 -2 -1 0 1 2 3 4 5
(a) Drift (%) (b) Drift (%)
2.5 500

2
400
1.5

Force (kN)
Strain (%)

1 300

0.5
Pi=249 kN
200
0

-0.5 100
-3 -2 -1 0 1 2 3 -5 -4 -3 -2 -1 0 1 2 3 4 5
(c) Drift (%) (d) Drift (%)

2.5 500

2
400
1.5
Force (kN)
Strain (%)

1 300

0.5 Pi=265 kN
200
0

-0.5 100
-3 -2 -1 0 1 2 3 -5 -4 -3 -2 -1 0 1 2 3 4 5
(e) Drift (%) (f) Drift (%)

Figure 9. ED bar strain (at the west wall at joint found-S1) and tendon force (at the west side) versus drift
relationships: (a) C5C-FB, ED strain; (b) C5C-FB, tendon force; (c) C5C-E32, ED strain; (d) C5C-E32,
tendon force; (e) C5C-UB, ED strain; and (f) C5C-UB, tendon force.

4.3. Self-centering capability

All the three columns showed good self-centering capability with residual drifts not greater than
0.4% (Table III and Figure 12). The columns exhibited a similar pattern of residual drift histories.
The residual drift of column C5C-FB prior to failure was only 0.1%, much lower than the other
two columns as it failed at a lower drift. The values of ED of the three columns calculated from

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 CYCLIC BEHAVIOR OF SEGMENTAL COLUMNS WITH HP

Fracture of ED bar

(b)

(a)

Figure 10. Column C5C-E32 at 6% drift: (a) north-east view; and (b) close view of joint foundation-S1.

16
Equivalent viscous damping ratio (%)

12

8
C5C-FB

C5C-E32
4
C5C-UB

0
0 1 2 3 4 5
Drift (%)

Figure 11. Equivalent viscous damping ratio.

measured lateral strengths ranged from 23–28% (Table III). The value of V0 in Equation (1) was
equal to 278 kN and was obtained by testing of a corresponding column without ED bars [5].
As previously mentioned, ED was limited to 35% to ensure a residual drift smaller than 1%.
In other words, the lateral strength contribution from axial forces including gravity loads and
prestressing forces needs to be greater than 65%. In contrast, lateral strength contribution from
axial forces, typically only from gravity loads, in a conventional concrete bridge column normally
ranges from 20–40%. This means a large residual drift may result for a conventional column after
large earthquakes. However, for a segmental column with ED limited to 35% to achieve the same

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 Y.-C. OU ET AL.

0.5

0.4 C5C-FB

C5C-E32

Residual drift (%)

0.3 C5C-UB

0.2

0.1

0
0 1 2 3 4 5
Drift (%)

Figure 12. Residual drift.

lateral strength as a conventional column, the size and/or compressive strength of concrete cross
section may need to be increased to accommodate additional prestressing forces.

5. COMPARISONS BETWEEN ANALYTICAL AND EXPERIMENTAL RESULTS

The envelope responses of the three columns were compared with results of analytical pushover
analyses using the joint bar-slip rotation method. Good agreement can be observed for lateral
forces (Figure 6), amounts of joint opening (Figure 8), and maximum strains and forces from the
ED bars and tendons at the west sides of the columns (Figure 9). Only the rotations of joints
foundation-S1 and S1-S2 were considered in the analytical model. Fracture of the ED bars was
predicted by the damage index defined by Equations (13)–(15). The damage indices for the ED
bars nearest to the tension faces of columns C5C-FB, C5C-E32, and C5C-UB were calculated to
be 0.5, 0.6, and 0.4, respectively, for when the columns were cyclically loaded up to 3, 6, and 6%
drifts, when fracture of the bars occurred. The bars fractured at damage indices smaller one. This
was likely due to additional strains arising from bending action by joint rotation and from buckling
action of the bars. These two actions were not considered in the proposed pushover model or in the
fatigue tests used to derive Equations (13) and (14). They were less pronounced at smaller drifts
as was evident in the comparisons of ED bar strains between experimental and analytical results
in Figure 9. Note that the ED bars were confined by grouted steel corrugated ducts that were
located in the middle of the walls of the segments and hence buckling of the ED bars at fracture
(e.g. Figure 10(b)) appeared to be less significant than typically seen in conventional columns.
The maximum strains of the ED bars at fracture were predicted by the pushover analysis to be
7, 13, and 6% for columns C5C-FB, C5C-E32, and C5C-UB, respectively. As the two actions
previously mentioned were not included in the analysis, actual maximum strains were expected
to be larger. The maximum confined concrete strains were calculated to be 0.01, 0.02, and 0.01
for columns C5C-FB, C5C-E32, and C5C-UB, respectively. These values were smaller than the
ultimate value 0.03 calculated by Equation (12). This is consistent with the test results where

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 CYCLIC BEHAVIOR OF SEGMENTAL COLUMNS WITH HP

damage of core concrete was negligible and no fracture of transverse reinforcement was found
(e.g. Figures 10(b)).

6. CONCLUSIONS

The cyclic behaviors were investigated for three large-scale precast segmental concrete bridge
columns designed with fully bonded conventional ED bars, with fully bonded HP ED bars, and
with conventional ED bars that were unbonded for a length. Important conclusions are summarized
as follows.
(1) All the three columns failed due to fracture of the ED bars. The column with the HP ED
bars showed higher drift capacity, greater energy dissipation, and higher lateral strength than
that with fully bonded conventional ED bars. Unbonding the conventional ED bars for a
length of 400 mm effectively delayed fracture of the bars while slightly decreased the lateral
strength. The column with such unbonded bars achieved the same drift capacity and similar
energy dissipation capability as that with the HP ED bars. However, unbonding requires
considerable labor work and weakens corrosion protection of the bars.
(2) At the ultimate state, the extent of cracking and crushing of concrete was minor. Most of
the column flexural rotations resulted from rotations of joints foundation-S1 and S1-S2.
Rotations of the other joints and the segment bodies were found to be negligible.
(3) All the three columns showed good self-centering capability with residual drifts not greater
than 0.4%. The measured strength contributions of the ED bars to the lateral strengths of
the three columns ranged from 23–28%.
(4) The proposed joint bar-slip rotation method provides satisfactory predictions of the envelope
responses of the columns in terms of lateral forces, amounts of joint opening, ED bar strains,
and tendon forces.

ACKNOWLEDGEMENTS
This cooperative research is funded by the National Center for Research on Earthquake Engineering
(NCREE) on the Taiwan side and by the Federal Highway Administration (FHWA) (Grant DTFH61-98-
C-00094) on the U.S. side. Their support is gratefully acknowledged.

REFERENCES
1. FHWA. Prefabricated bridge elements and systems—get in, get out, stay out, Federal Highway Administration,
Washington, DC, U.S.A., 2009. Available from: http://www.fhwa.dot.gov/bridge/prefab/index.htm.
2. P.S. Mitsubishi Construction Co., Ltd. Precast column method, 2009. Available from:
http://www.psmic.co.jp/gijyutu/d kabu01.html.
3. Hewes JT, Priestley MJN. Seismic design and performance of precast concrete segmental bridge columns. Report.
No. SSRP-2001/25, University of California, San Diego, CA, U.S.A., 2002.
4. Wang J-C, Ou Y-C, Chang KC, Lee GC. Large-scale seismic tests of tall concrete bridge columns with precast
segmental construction. Earthquake Engineering and Structural Dynamics 2008; 37(12):1449–1465.
5. Ou Y-C, Wang P-H, Tsai M-S, Chang K-C, Lee GC. Large-scale experimental study of precast segmental
unbonded post-tensioned concrete bridge columns for seismic regions. Journal of Structural Engineering 2009;
10.1061/(ASCE)ST.1943-541X.0000110.
6. ASTM A955-04a. Standard Specification for Deformed and Plain Stainless Steel Bars for Concrete Reinforcement.
ASTM International: West Conshohocken, PA, U.S.A., 2004.

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe
 Y.-C. OU ET AL.

7. Schnell RE, Bergmann MP. Improving tomorrow’s infrastructure: extending the life of concrete structures with
solid stainless steel reinforcing bar. Proceedings of the 3rd NYC Bridge Conference, New York, NY, U.S.A.,
27–28 August 2007.
8. Zatar WA, Mutsuyoshi H. Residual displacements of concrete bridge piers subjected to near field earthquakes.
ACI Structural Journal 2002; 99(4):740–749.
9. Kawashima K. Seismic performance of RC bridge piers in Japan: an evaluation after the 1995 Hyogo-ken nanbu
earthquake. Progress in Structural Engineering and Materials 2000; 2(1):82–91.
10. Ou Y-C, Chiewanichakorn M, Aref AJ, Lee GC. Seismic performance of segmental precast unbonded post-
tensioned concrete bridge columns. Journal of Structural Engineering 2007; 133(9):1636–1647.
11. Pampanin S, Priestley MJN, Sritharan S. Analytical modeling of the seismic behavior of precast concrete frames
designed with ductile connections. Journal of Earthquake Engineering 2001; 5(3):329–367.
12. ASTM A706-04a. Standard Specification for Low-alloy Steel Deformed and Plain Bars for Concrete Reinforcement.
ASTM International: West Conshohocken, PA, U.S.A., 2004.
13. AASHTO. AASHTO LRFD Bridge Design Specifications (4th edn). American Association of State Highway and
Transportation Officials: Washington, DC, U.S.A., 2007.
14. Sezen H, Setzler E. Reinforcement slip in reinforced concrete columns. ACI Structural Journal 2008; 105(3):
280–289.
15. ACI Committee 318. Building Code Requirements for Structural Concrete (ACI 318-08) and Commentary
(ACI 318R-08). American Concrete Institute: Farmington Hills, MI, U.S.A., 2008.
16. Mander JB, Priestley MJN, Park R. Observed stress-strain behaviour of confined concrete. Journal of Structural
Engineering 1988; 114(6):1827–1849.
17. ASTM E606-92. Standard Practice for Strain-controlled Fatigue Testing. ASTM International: West
Conshohocken, PA, U.S.A., 1992.

Copyright q 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. (2010)
DOI: 10.1002/eqe