Impact Response and Capacity of Precast Concrete

Segmental versus Monolithic Bridge Columns

Tin V. Do, S.M.ASCE1; Thong M. Pham2; and Hong Hao, F.ASCE3

Abstract: In this study, the performance of precast concrete segmental bridge columns (PCSBCs) against truck impacts was numerically
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

investigated and compared with a corresponding conventional monolithic column (CMC). The numerical results have shown that
although the impact force time histories of the two columns were quite similar under the same loading conditions, the PCSBC showed a
better performance in terms of the induced bending moment and shear force by high impact force due to shear slippage and joint rocking
between concrete segments. Also, the damage and failure of PCSBCs were localized at the two bottommost segments due to compression
damage and/or combined flexural and shear failure of the concrete segment, whereas failure of the CMC distributed widely with flexural
cracks, shear cracks, and punching shear at multiple sections. Furthermore, the base segment, which was found to absorb about 80% of the
total absorbed energy of the PCSBC, played a crucial role in controlling the failure of the PCSBC. An analytical method to estimate
the bending moment required to open the segment joint and the ultimate bending moment was also developed with consideration of the
dynamic increase factor and the increase in axial force associated with stress wave propagation in the column induced by impact load.
DOI: 10.1061/(ASCE)BE.1943-5592.0001415. © 2019 American Society of Civil Engineers.
Author keywords: Precast; Segmental columns; Bridges; Collisions; Vehicle; Simulation.

Introduction
The demands on accelerated bridge construction (ABC) have
been considerably increased during the last few decades due to
the benefits of ABC compared with the traditional cast-in-place
construction. ABC not only provides a feasible solution to
increase site constructability and construction quality, improve
work zone safety for workers and traveling public, and minimize
traffic disruption during the construction period, but it also adds
practical and economical methods to those of the traditional tech-
nology (Culmo 2011). Because of ABC technology, a precast
concrete segmental bridge column (PCSBC), which is commonly
used in the bridge construction, can meet all the objectives of the
ABC. Many projects using PCSBCs have already been imple-
mented (Ou 2007; ElGawady et al. 2010). A PCSBC has to be
designed to withstand hazardous loads during its service life,
such as seismic load, impact load, blast load, and so forth.
However, due to the lack of understanding of its dynamic behav-
iors under seismic and impact loads, PCSBCs have been limited
1
Ph.D. Candidate, Center for Infrastructural Monitoring and Protection,
School of Civil and Mechanical Engineering, Curtin Univ., Kent St., Bentley,
WA 6102, Australia. Email: tin.v.do@postgrad.curtin.edu.au
2 was observed on a PCSBC that was impacted at a joint between
Research Fellow, Center for Infrastructural Monitoring and Protection,
School of Civil and Mechanical Engineering, Curtin Univ., Kent St.,
Bentley, WA 6102, Australia. Email: thong.pham@curtin.edu.au
3
John Curtin Distinguished Professor, Center for Infrastructural
Monitoring and Protection, School of Civil and Mechanical Engineering,
Curtin Univ., Kent St., Bentley, WA 6102, Australia; Professor, School of
Civil Engineering, Guangzhou Univ., 230 Wai Huan Xi Road, Guangzhou
Higher Education Mega Center, Guangzhou 510006, P.R.China (corre-
sponding author). Email: hong.hao@curtin.edu.au
Note. This manuscript was submitted on January 22, 2018; approved
on December 18, 2018; published online on April 10, 2019. Discussion pe-
riod open until September 10, 2019; separate discussions must be submit-
ted for individual papers. This paper is part of the Journal of Bridge
Engineering, © ASCE, ISSN 1084-0702.
mainly in areas of low seismicity and low traffic flow. To over-
come these difficulties, the performance and capacities of
PCSBCs under cyclic loading and seismic loading have been
experimentally (Hewes and Priestley 2002; Billington and Yoon
2004; Ou 2007; Li et al. 2017b), numerically (Dawood et al.
2014; Hung et al. 2017; Li et al. 2017a), and analytically (Ou
2007; Chou et al. 2013) investigated. The investigations found
that PCSBCs have many structural advantages, e.g., higher duc-
tility, greater self-centering capacity, and less column damage
compared with the reference conventional monolithic column
(CMC).
However, studies of the behaviors of PCSBCs under other
extreme loading conditions, such as truck impacts or blast loads,
are very limited with very few reports found in open literature
(Chung et al. 2014; Zhang et al. 2016a, b; Hao et al. 2017; Li
et al. 2017c; Do et al. 2018a, b). An experimental study on
PCSBCs under pendulum impact by Zhang et al. (2016a) indi-
cated that the PCSBCs experienced a flexural response when an
impactor hit the center of the column. Compressive damage at
the segment corner was observed on the impacted segment and
the base segment due to the rocking and rotation of the concrete
segment leading to the column failure. In the latest experiment
study by Hao et al. (2017), a combined flexural and shear failure
the base segment and the second-base segment. The flexural
compressive damage of concrete together with the diagonal shear
failure of the base segment led to collapse of the PCSBC. Also,
when the impactor smashed into the center of the base segment,
the column slipped away without flexural deformation and then
collapsed because of excessive damage to the concrete segment.
These experimental studies showed that under different loading
conditions, the PCSBC showed various types of failure modes,
i.e., flexural failure, combined shear and flexural damage, and
shear failure. To control the impact response of PCSBCs sub-
jected to vehicle collisions, effects of critical parameters on their
impact performance were numerically examined by Do et al.

© ASCE 04019050-1 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 (2018b). By increasing the initial prestress level and reducing the
number of segments, smaller lateral/residual displacement
occurred in the PCSBCs. The height-to-depth ratio of the con-
crete segment was also suggested to be smaller than 2 to reduce
flexural cracks and failure of the column.
Research Significance
Although the previous studies provide an overall impact perform-
ance and response of the PCSBC, the characteristics of axial force,
bending moment, shear force, and failure modes induced by vehicle
impact have not been well investigated. For safe and economic
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.
Numerical Validation
Numerical Model Development
In this study, numerical models were built and verified against ex-
perimental pendulum impact tests on PCSBCs and a CMC by
Zhang et al. (2016a). The schematic view of the pendulum impact
test setup, dimensions, and the design of the two columns are pre-
sented in Fig. 1(a). Two finite-element (FE) models were developed
in ANSYS-APDL/LS-DYNA to simulate the pendulum impact
tests of the PCSBCs and CMC [Fig. 1(b and c)]. In this study, the
numerical model was first built in the ANSYS APDL environment
before exporting a keyword file to LS-DYNA for implicit and

designs of PCSBCs to resist vehicle impact, it is important to under-
stand these response characteristics. This study attempts to investi-
gate the induced bending moment, shear force, and failure modes of
PCSBCs under truck impact. An analytical approach to estimate the
bending moment capacities including the moment that results in the
opening of the segment joint and the ultimate bending moment is
also proposed.
explicit simulation. In LS-DYNA, hexahedral elements with one
integration point (constant stress solid elements) were used to
model the footing, concrete column, impactor, added mass, anchors,
and the prestress tendons, whereas three nodes-beam elements
(Hughes-Liu with cross-sectional integration) were used to repre-
sent reinforcement bars and energy dissipation (ED) bars. The LS-
DYNA contact keyword *Contact_Automatic_Surface_to_Surface

4 Bolts 400
160 Steel plates
M16 240x90x10

400

300
100
Added mass
(288 kg)
The hole Section A-A
left for 100 I6a50
tendon I20 4I6

50
4I6

Steel impactor
I4a40
(300 kg)
800

800
I4
800
400

Steel frame ED bars

A A I6a50
160

4 Bolts Footing
I6a50
140

140

PCSBC CMC
(a)

Anchor

Steel plate

Concrete block Reinforcing

cage

Concrete segments
Impactor

Tendon

Longitudinal
bars

Stirrups
Footing
ED bars
(b) (c)

Fig. 1. Experimental test and numerical model of the scaled columns under impact load: (a) impact test setup and the design of the scaled columns
(data from Zhang et al. 2016a); (b) simulation of PCSBC; and (c) simulation of CMC. ED = energy dissipation.

© ASCE 04019050-2 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 was used to simulate the contact between concrete segments and the
tendons, the steel impactor and the column, and between two con-
crete segments. In addition, a perfect bond between reinforcement
bars and surrounding concrete was assumed in the simulation.
The LS-DYNA material model *Mat_Concrete_Damage_Rel3
(MAT_072R3) was used for modeling the concrete material. The
recommendation by Hao and Hao (2014) for the dynamic increase
factor (DIF) of the concrete compressive and tensile strengths was
implemented in this study because it improved some limitations
from previous models. The lateral inertia confinement and the end
friction confinement from the experimental test were eliminated
when deriving the proposed equations so that it better reflects the
dynamic properties of concrete. The compressive DIF (CDIF) and
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.
corresponding number of the CMC was 104,311 elements, i.e.,
102,547 solid elements and 1,764 beam elements. The detailed in-
formation of the PCSBC model was comprehensively presented in
Do et al. (2018b) and that of the CMC model was given in Do et al.
(2018a).
Results and Validation
Numerical analysis results and experimental testing results of the
PCSBC are compared in Fig. 2. Fig. 2(a) demonstrates that five
peaks of the impact force time histories were observed in both
the numerical simulation and laboratory test. The first peak of
the impact force (PIF) and impact duration in the FE model were

20.70 kN and 93 ms, which compared well to 20.91 kN and 93 ms in

tensile DIF (TDIF) of concrete at the strain rate ɛ_ d are presented in
the experimental test, respectively. The impulse predicted in the nu-
the following equations:
merical model was 537.4 N·s, which was just 3.7% less than that of
fcd the experimental test (557.8 N·s). Moreover, as given in Fig. 2(b),
CDIF ¼ the displacement time histories at the column midheight from the FE
fcs
8 model were in good agreement with the testing results in which the
< 0:0419ðlog ɛ_ d Þ þ 1:2165 ðɛ_ d  30s1 Þ maximum value was 32.75 mm in the simulation and 32.80 mm in
¼ the test. The global trends of the displacement response histories
: 0:8988ðlog ɛ_ d Þ2  2:8255ðlog ɛ_ d Þ þ 3:4907 ðɛ_ d > 30s1 Þ were also well predicted. Figs. 3(a and b) demonstrate that the defor-
(1) mation of the PCSBC including the joint opening between the
impacted segment and its adjacent segments, the damage at the seg-
ment, and the joint opening at the base were all well simulated in the
ftd numerical model compared with the experimental test.
TDIF ¼
fts The impact responses of the CMCs, including the impact force,
8 column midheight displacement, and plastic strain from the simula-
>
> 0:26ðlog ɛ_ d Þ þ 2:06 ðɛ_  1s1 Þ
< tion, were also verified against the experimental test, as given in
¼ 2ðlog ɛ_ d Þ þ 2:06 ð1s1 < ɛ_  2s1 Þ
>
>
:
1:44331ðlog ɛ_ d Þ þ 2:2276 ð2s1 < ɛ_ d  150s1 Þ 25 Experiment (Zhang et al. 2016a)
First peak Simulation
(2)
20
Impact force (kN)

where fcd and ftd = dynamic compressive strength and dynamic ten- Second peak
sile strength, respectively; and fcs fts = static compressive and tensile 15
Third peak
strength, respectively. Fourth peak
Also, the LS-DYNA erosion feature *MAT_ADD_EROSION 10
Fifth peak
was used to simulate the concrete material damage to avoid compu-
tation overflow. This study used the maximum principle strain at 5
failure as a criterion to delete failed concrete elements. The value of
0.7 was used for the erosion criterion of the concrete of the two col- 0
umns after trials, which yielded a good prediction of the column 0 20 40 60 80 100
damage. The steel reinforcements including longitudinal bars, stir- Time (ms)
rups, and ED bars were simulated by using the elastic-plastic mate- (a)
rial model *Mat_Piecewise_Linear_Plasticity (MAT_24). The DIF
35 Experiment (Zhang et al. 2016a)
of these steel reinforcements proposed by Malvar and Crawford
Simulation
(1998) was used
28
Displacement (mm)

 0:0740:04fy
ɛ_ 414
DIF ¼ (3) 21
104

where fy = yield strength of steel in megapascals. 14

For the anchor plate and steel impactor, the LS-DYNA elastic
material model *MAT_ELASTIC (MAT_001) was selected. To 7
represent a prestress tendon in the PCSBC, *Mat_Elastic_
Plastic_Thermal (MAT_004) was used. 0
The minimum mesh size of the concrete elements and steel rein- 0 100 200 300 400 500
Time (ms)
forcements was 5 mm, which was obtained after conducting a con-
(b)
vergence test on these columns, and the maximum mesh size used
in the simulation to model the added mass and impactor was
Fig. 2. Model verification of PCBSC: (a) impact force time history;
50 mm. In summary, the PCSBC consisted of 126,407 elements, and (b) displacement at the center of the column.
i.e., 124,247 solid elements and 2,160 beam elements, and the

© ASCE 04019050-3 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 30 Experiment (Zhang et al. 2016a)
Simulation
25

Impact force (kN)

20

15

10

0
0 10 20 30 40 50
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

Time (ms)
(a)

12 Experiment (Zhang et al. 2016a)

Simulation
(a) (b)

Displacement (mm)
9

Concrete damage 0

-3
0 50 100 150 200 250 300 350
Time (ms)
(b)

Fig. 4. Model verification of CMC: (a) impact force time history; and
Flexural crack (b) displacement at the center of the column.

Shear crack
Numerical Models of Bridge Columns under
Truck Impacts

(c) (d)
Fig. 3. Plastic strain of the columns under impact: (a) PCSBC simula-
tion; (b) PCSBC experiment; (c) CMC simulation; and (d) CMC experi-
ment. [Reprinted (b) and (d) from International Journal of Impact
Engineering, Vol. 95, X. Zhang, H. Hao, and C. Li, “Experimental
investigation of the response of precast segmental columns subjected to
impact loading,” pp. 105–124, © 2016, with permission from Elsevier.]
Figs. 3(b) and 4. The PIF and impact duration in the FE model were
23.7 kN and 3 ms, respectively, and the corresponding results in the
experiment were 22 kN and 30 ms [Fig. 4(a)]. Also, the FE model
gave a good prediction in terms of the lateral displacement time his-
tory at the column midheight [Fig. 4(b)]. The maximum and resid-
ual lateral displacement of the beam in the simulation were 7.6 and
1.8 mm, respectively, compared with 7.5 and 1.5 mm in the test.
Furthermore, the column damage, i.e., flexural cracks at the column
midheight, shear cracks at the column base, and concrete damage at
the column top, observed in the test was also predicted by the nu-
merical model, as presented in Figs. 3(c and d).
The previously mentioned comparisons indicated that the cur-
rent numerical models reliably predicted the impact responses of
the PCSBC and the CMC, and they were able to simulate the lateral
shear slip, local damage, plastic deformation, and the failure modes
of the column under impact load.
Bridge and Truck Model
Based on the validated models, FE models of two bridge models
with PCSBCs and CMCs, respectively, were developed in this sec-
tion. The previous study by Consolazio and Davidson (2008) indi-
cated that the dynamic behaviors of multispan bridge structures
could be accurately predicted by an analysis model that consists of
one bridge column and two superstructure spans. This simulation
approach was also used in previous studies (El-Tawil et al. 2005;
Abdelkarim and ElGawady 2017). As such, each bridge model con-
sisting of one single column, two superstructure beams, footing,
and two concrete abutments was considered in this study, as pre-
sented in Fig. 5(a). The overall dimensions and properties of super-
structures were obtained from Megally et al. (2001), and the span
length was assumed to be 40 m. The superstructure’s mass was
transmitted to the column through a trapezoidal cap beam that
was placed on top of the column. Note that no rubber or bearing pad
was used to connect the superstructures and the cap beam in this
study because of its insignificant effect on the impact behaviors of a
bridge column (El-Tawil et al. 2005). In these simulations, the
superstructures were assumed to rest on top of the cap beam with a
coefficient of friction between the concrete and concrete surface of
0.6 (ACI 2008). The other end of the superstructure was designed
to rest on a simplified solid block, which represents the abutments
[Fig. 5(a)]. The total dead load consisting of the superstructures
and the substructures was about 4,600 kN, which is equal to 10%

© ASCE 04019050-4 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 Superstructures

Abutment Cap beam

Concrete segments
Abutment

Footing
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

(a)

150 Steel duct: D100 mm

Thickness: 10 mm
3200 1200 3200 7600
Tendon
D50 mm
1750

1750
24D20
1920 1920 1920 1920 1920

1200
D14a200
9600

9600
1200
A A A-A B B
24D30

D14a200
1200

1500
1500

5200 1200 5200

PCSBC B-B CMC

(b)

Anchors
1200

Cap beam Cap beam

Tendon

Concrete segments Column

1200
PCSBC- cross section
Tendons Reinforcements
Steel duct
Steel duct
1200

Reinforcements Footing

1200
Footing
CMC - cross section

PCSBC CMC
(c)

Fig. 5. Design and simulation of the PCSBC and CMC: (a) three-dimensional (3D) view of the PCSBC with superstructures; (b) column design and
dimensions; and (c) numerical model of the PCSBC and CMC.

© ASCE 04019050-5 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 of the axial capacity of each column. The detailed dimensions of
the two bridge specimens are presented in Fig. 5(b) and Table 1.
The bridge column, cap beam, tendons, superstructures, concrete
abutments, and footing of these models were modeled by using
solid elements (constant stress solid elements), and steel rein-
forcements were simulated by beam elements (Hughes-Liu with
cross-sectional integration) [Fig. 5(c)]. Note that the longitudinal
bars were discontinuous at segment joints and no ED bars were
used in the PCSBC.
A medium-duty, Ford truck (35,400 elements) model was used
to simulate the collision on the bridge columns. This model was
shared by Sharma et al. (2012) and Abdelkarim and ElGawady
(2016); it was used to examine the dynamic behavior of structures
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.
so that the influence of the engine mass can be clearly observed.
Moreover, the vehicle velocity also was varied from 60 to 140 km/h
in the simulations. In this study, the top of the footing was assumed
to be placed under the ground level of 0.5 m.
Modeling Procedure
To apply an initial prestressing force in the tendon, a temperature-
induced shrinkage option was employed that was used in previous
studies (Jiang and Chorzepa 2015; Do et al. 2018b). In this method,
the *DYNAMIC RELAXATION (DR) function was used to calcu-
late the initial stress on concrete structures and tendons before trans-
ferring those results as an input data to an explicit analysis

subjected to vehicle collisions in previous studies (El-Tawil et al.
2005; Sharma et al. 2012; Agrawal et al. 2013; Chen et al. 2015;
Abdelkarim and ElGawady 2016). The accuracy of the model was
verified by the Federal Highway Administration (FHWA)/National
Highway Traffic Safety Administration (NHTSA) National Crash
Analysis Center at George Washington University. The previous
study by Do et al. (2018a) showed that the PIF noticeably depended
on the kinetic energy of the truck’s engine. Thus, to investigate the
performances of the columns under different PIFs, the truck veloc-
ity and truck engine’s mass were varied in this study. The engine’s
mass was varied from 0.64 t to 2 and 3 t by changing the mass den-
sity of the material model. By increasing the mass of the engine, the
cargo mass was reduced from 3 to 0.64 t to keep the total vehicle
mass of 8 t unchanged. The purpose of these analyses is to investi-
gate the influence of the engine mass on the response of the col-
umns, which was usually neglected in previous studies and design
guides. The total vehicle mass of 8 t was kept the same in this study
Table 1. Detailed dimensions of the two bridge specimens
(Livermore Software Technology Corporation 2007). An example
of this method was reported in Do et al. (2018b). In this study, the
prestress force in each tendon of the PCSBC was 1,500 kN, which
was equal to 40% of the yielding capacity of the tendons. As a
result, the four tendons yielded a total resultant force of 6,000 kN
on the PCSBC, which was equal to 13% of the axial compressive
strength of the PCSBC. In the implicit simulation, the gravity load
of the structures was applied by gradually increasing the gravity
acceleration in both the PCSBC and the CMC. Note that the abrupt
application of the gravity acceleration (9.81 m/s2) will lead to unde-
sirable dynamic responses of the structures, i.e., the vertical vibra-
tions of the superstructures (Consolazio et al. 2009). Therefore, the
gravity acceleration should be applied for a relatively long duration;
for example, this study used the duration of 150 ms before the vehi-
cle model collided with the bridge column to mitigate the unex-
pected variation of the gravity load. In the following sections, the
time was set to zero when the vehicle starts to collide on the bridge
column. The simulation process of the study is presented in Fig. 6.

Parameters PCSBC CMC Numerical Results

Column height (mm) 9,600 9,600 To examine the performance and capacity of the PCSBC against
Number of segments 5 — truck impacts, a series of numerical simulations was performed and
Segment height (mm) 1,920 —
the numerical results are presented in Table 2. The numerical results
Section width (mm) 1,200 1,200
of the PCSBC are also compared with the corresponding results of
Section depth (mm) 1,200 1,200
CMC in terms of impact force time histories, bending moment,
Longitudinal steel 24D20 (discontinuous) 24D30 (continuous)
shear force, and failure modes.
Lateral steel D14a200 D14a200
Tendon diameter (mm) 50 —
Number of tendons 4 — Impact Force Time History

Note: — = not applicable. For concrete bridge columns, the impact force time history under
truck impact normally included two different peaks caused by the
truck’s engine and cargo (Abdelkarim and ElGawady 2016; Do
et al. 2018a) in which the PIF could be associated with the engine,
Prestress load Gravity load Truck impact simulation the cargo impact depending on the vehicle, and column properties
and interaction between vehicle and cargo. Cargo impact occurs af-
(Dynamic relaxation) 150 (ms)
ter engine impact, therefore, cargo impact could generate a larger
impact force than engine impact only if the column survives the
engine impact without experiencing significant damage. If engine
Axial force

Transient stage impact causes significant damage to the column, which softens the
column, then the subsequent cargo impact is likely to generate a
long duration impact but not necessarily a large impact force. For
Time is set to zero example, the numerical studies by Chen et al. (2016, 2017) showed
that the second PIF from the cargo collision was larger than the first
PIF caused by the engine collision because the column was assumed
Time
to be rigid; therefore, concrete damage and column failure were not
modeled. When the cargo started to collide on the column, the con-
Initial process Explicit analysis
tact stiffness between the vehicle model and the column was similar
Fig. 6. Modeling procedure.
to the initial contact stiffness of the column. With very high kinetic
energy from the cargo due to the cargo’s mass, the second peak was

© ASCE 04019050-6 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 Table 2. PCSBC and CMC under different initial loading conditions

Initial condition PCSBC CMC

Total truck’s Engine’s Velocity Momentum Kinetic energy Impulse
Case mass (t) mass (t) (km/h) (t·m/s) (kN·m) PIF (kN) Impulse (kN·s) PIF (kN) (kN·s)
C1 8.0 0.64 60 133.3 1,111.1 1,981 143.4 1,868 130.2
C2 8.0 0.64 80 177.8 1,975.3 3,182 179.0 3,460 176.4
C3 8.0 0.64 90 200.0 2,500.0 4,848 203.2 4,596 199.8
C4 8.0 0.64 100 222.2 3,086.4 7,891 225.3 8,260 220.5
C5 8.0 0.64 110 244.4 3,734.6 9,680 245.4 9,660 249.6
C6 8.0 0.64 120 266.7 4,444.4 12,149 274.1 12,000 266.1
C7 8.0 0.64 140 311.1 6,049.4 16,086 —a 16,400 —a
C8 8.0 2.00 100 222.2 3,086.4 19,326 223.0 18,500 223.5
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

C9 8.0 3.00 100 222.2 3,086.4 24,476 —b 23,333 —b

a
Simulation was terminated due to severe damage of the vehicle model.
b
Simulation was terminated due to severe damage of the column.

larger compared with the first peak. In the present study, damage to
the concrete column and column deformation by the bumper and
engine impact was simulated, leading to a significant reduction of
contact stiffness of the column. As a result, the second impact force
from the cargo collision, which depends on the interaction between
vehicle and column, was smaller than the first impact force,
although the kinetic energy was larger. The results indicated that
the assumption of the rigid column does not reflect the actual inter-
action between the vehicle and column. Hence, the term PIF stands
for the first peak of the impact force in this study. It should be noted
that the PIF caused by the vehicle bumper is usually small compared
with the engine and cargo impact; therefore, it is not explicitly dis-
cussed. Also, note that if stronger columns were considered and
they survived the engine impact without suffering a large amount of
damage, then the PIF could correspond to the cargo impact.
However, because the primary objective of the present study is to
compare the performances of CMCs and PCSBCs subjected to ve-
hicle impact, modeling stronger columns to get the larger impact
force from cargo impact is not performed. Instead, the impact force
due to engine impact is discussed in detail because it has been
neglected in many previous studies and not covered in the current
design guides.
The impact force time histories of the PCSBC and the CMC
against truck impacts are presented in Fig. 7. Interestingly, although
these two columns had different lateral stiffnesses due to the dissim-
ilar initial axial force and the discontinuous concrete segments of
the PCSBC, the impact force time histories were almost identical
for PIF, duration, and impulse when they were subjected to the
same loading condition (Table 2 and Fig. 7). These results indicated
that the influence of the global stiffness of the bridge column on the
PIF was insignificant. Similar observations were also reported in
previous studies on concrete beams: the impact force was found to
be dependent primarily on the local stiffness only (Pham and Hao
2017a). The marginal effect of the global column stiffness on the
PIF when concrete structures were subjected to vehicle or ship
impacts also has been previously reported (Sha and Hao 2013; Do
et al. 2018b).
Meanwhile, an increase in the vehicle velocity corresponded to a
substantial increase in the PIF on the PCSBC, as illustrated in
Figs. 7(a–g). The PIF increased significantly from around 1,981 to
16,400 kN when the velocity rose from 60 to 140 km/h.
Interestingly, with the same vehicle mass and vehicle velocity, the
contribution of the engine’s mass on the PIF was also noteworthy.
Consider the velocity of 100 km/h and the total mass of 8 t, increasing
the engine’s mass from 0.64 to 3 t, the PIF increased proportionally
from 7,891 to 24,476 kN, as illustrated in Figs. 7(d, h, and i). In gen-
eral, in all the cases the PIFs of the two columns were almost identi-
cal. However, if the impact force was not intensive and fast enough
to cause slippage or local damage, the impact force time history,
affected by the interaction between the global/local stiffness of the
column and the impact energy, was only slightly different.
Otherwise, under intensive impact in which only the local stiff-
ness governs the impact force, the impact force time histories of
the two columns were almost identical. The comparisons of the
PIF and impulse of the two columns are also presented in Fig. 8.
The numerical results in this study also show good agreement
with the empirical equations, which have been proposed by Do
et al. (2018a).
Column Shear Force
Fig. 9 illustrates the typical shear force time histories of the PCSBC
and the CMC under truck impact (C6). When the impact force
reached a peak at 12,149 kN, the shear force at the column base
of the two columns also increased to the highest value of about
7,500 kN [Fig. 9(a)]. Note that the shear force at the column base
was substantially smaller than the PIF because of the contribution
of the inertia force to resist the impact, as discussed in Do et al.
(2018a), implying that directly applying PIF in the equivalent static
analysis without considering the distribution of the inertia force
would significantly overestimate the shear force in the column. The
stress wave propagated from the impact area to the column top caus-
ing the maximum value of the shear force at about 3,800 kN
[Fig. 9(b)]. The time lag between the PIF and the maximum shear
force at the column top was about 10 ms. When the impact force
dropped to about 2,500 kN after 30 ms, with the contribution of the
inertia force, the shear force time histories at the column base fluctu-
ated around the impact force values, and those at the column top
oscillated around the zero level. Fig. 9 demonstrates that the shear
force time histories of the PCSBC at the column base fluctuated
with large amplitude and high frequency, whereas those of the
CMC were almost equal to the impact force because the PCSBC
experienced the high-frequency vibration of the concrete segment
during impact loading (Do et al. 2018b). This caused the variation
of the inertia force distributed along the segment. Because the
PCSBC was discontinuous at the segment joints, the stiffness of
the column was smaller than the CMC. The high-frequency vibra-
tion of the shear force with larger amplitude compared with the
corresponding CMC was therefore mainly associated with the
vibrations of the individual segment instead of the segmental
column.

© ASCE 04019050-7 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 6,000 6,000 6,000
PCSBC PCSBC

Impact force (kN)

Impact force (kN)

PCSBC
Impact force (kN)
CMC CMC CMC
4,500 4,500 4,500

3,000 3,000 3,000

1,500 1,500 1,500

0 0 0
0 40 80 120 160 200 0 40 80 120 160 200 0 40 80 120 160 200
Time (ms) Time (ms) Time (ms)
(a) (b) (c)
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

10,000 10,000 14,000

PCSBC PCSBC PCSBC
Impact force (kN)

Impact force (kN)

Impact force (kN)

CMC CMC CMC
7,500 7,500 10,500

5,000 5,000 7,000

2,500 2,500 3,500

0 0 0
0 40 80 120 160 200 0 40 80 120 160 200 0 40 80 120 160 200
Time (ms) Time (ms) Time (ms)
(d) (e) (f)

18,000 20,000 28,000

PCSBC PCSBC PCSBC
Impact force (kN)

Impact force (kN)

Impact force (kN)

CMC CMC CMC
13,500 15,000 21,000

9,000 Simulation was terminated 10,000 14,000

due to vehicle model failure Simulation was terminated
due to the column colapse
4,500 5,000 7,000

0 0 0
0 20 40 60 80 0 40 80 120 160 200 0 25 50 75 100 125
Time (ms) Time (ms) Time (ms)
(g) (h) (i)

Fig. 7. Impact force time histories with different initial conditions: (a) C1; (b) C2; (c) C3; (d) C4; (e) C5; (f) C6; (g) C7; (h) C8; and (i) C9.

The maximum value of the shear force at the column ends of the Moreover, an excessive local failure occurred at the base segment,
PCSBC and CMC are compared in Fig. 10. This figure demon- which is evident by diagonal shear failure in the PCSBC. Meanwhile,
strates that minor differences at the two ends could be found a diagonal shear failure and punching shear failure were observed in
between the PCSBC and the CMC when the PIF was smaller than the CMC when the shear force reached the column’s shear capacity
12,149 kN (C1–C6) because no shear damage or slippage at the at about 10,500 kN. Furthermore, due to the slippages, segment
impact area (between Segment 1 and Segment 2) was observed dur- vibrations, and the large deformation of the base segment, which dis-
ing the force phase in these columns [Fig. 11(a)]. This led to the sipated a large amount of the impact energy, the shear force at the
similar impact energy from the collision transferred to the PCSBC column top of the PCSBC was smaller than that of the CMC.
and the CMC, and the similar responses of the two columns. As a Envelopes of the shear force diagram of the PCSBC and the CMC in
result, the induced shear forces in the two columns had a trivial the last three loading conditions are also compared and presented in
difference (Fig. 10). Also, no shear crack or shear failure was Fig. 10(c). These results showed that both columns react similarly to
observed in these columns when the shear force at the base was truck impact when no slippage between the segments occurs in the
less than 7,500 kN (PIF = 12,149 kN). However, considerable dif- PCSBC. However, under high impact force, the slippage between
ferences in the maximum shear force at the column ends were the segments reduced the shear force at the column ends of the
observed when these columns were subjected to more intensive PCSBC, resulting in less shear damage of the column, as will be pre-
impact loading (C7–C9), as presented in Fig. 10. When the PIF sented in the subsequent section.
increased from 12,149 kN (C7) to 24,476 kN (C9), the shear force
at the column base of the PCSBC was nearly steady at 7,700 kN,
Column Bending Moment
whereas that of the CMC continued increasing to about 10,500 kN
before leveling off [Fig. 10(a)] because the shear force (7,700 kN) Because of the variation of the inertia force, which distributed along
reached the antislip capacity of the PCSBC, which caused the slip- the column, the bending moment shape of the PCSBC and the CMC
page between the base segment and the footing [Fig. 11(a)]. varied considerably, as presented and compared in Fig. 12(a) (C6).

© ASCE 04019050-8 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 40,000 When the impact force increased to the highest value of 12,149 kN
Do et al. (2018a)
PCSBC
at t = 20.5 ms, the bending moment at the second joint (between
CMC Segment 1 and Segment 2) reached the maximum positive value at
30,000
4,171 kN·m. After about 0.5 ms, the bending moment of the
PIF (kN)
PCSBC at the base also increased to the maximum negative value at
20,000 4,533 kN·m. The compression stress also spread to the column top
leading to the vibration of the entire column. When the bending
10,000 moment at the column top appeared (t  25.5 ms), the bending
moment at the intermediate section reached its maximum negative
value. It is noteworthy that the negative bending moment, which
0 occurred on the top part of the column, was caused by the inertia
0 7 14 21 28 35 42
1/2 force that was distributed along the column after the PIF and the
(0.5meV2)1/2 (kN.m)
inertia resistance of the superstructures (Do et al. 2018a).
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

(a)
Moreover, the intermediate section, which happened when the col-
umn top started vibrating and was located between the column top
280
o
45 benchmark and the impact point, varied considerably under different loading
PCSBC conditions as defined and explained in Do et al. (2018a). Then, the
240 CMC negative bending moment at the column top together with the posi-
Impulse (kN.s)

tive bending moment at the intermediate section increased to its

200 highest value at about t = 32.5 ms because of the effect of the cap
beam and the superstructures. Fig. 12(a) also compares the bending
moment diagrams of the bridge column by using the equivalent
160 static force (ESF) (AASHTO 2012) with the bending moment enve-
lope from the dynamic simulation. It is clear that the ESF resulted in
120 the highest bending moment at the base, whereas the actual truck
120 160 200 240 280 impact caused the large bending moment at multiple sections of the
Momentum (kN.s) column, i.e., the column base, the segment joint close to the impact
(b) point, the intermediate section, and the column top. The ESF did
not yield a negative moment along the column, whereas the numeri-
Fig. 8. (a) The PIF-initial kinetic energy of the engine relation; and cal simulation showed that the magnitude of the negative moment
(b) vehicle momentum–impulse conversion. at the intermediate section was even greater than the negative
moment at the base. It is, therefore, essential to note that the use of
12,500 the ESF model could lead to an underestimation of the impact
Shear force_PCSBC responses of the structures.
10,000 Shear force_CMC The comparisons of the bending moment diagrams between the
Impact force
PCSBC and CMC against different truck impact conditions are also
7,500
Force (kN)

presented in Fig. 12(b). Similar to the shear force, when the PIF was
5,000 smaller than 12,149 N (C1–C6), only a minor difference could be
found between the two columns because there were no significant
2,500 shear slippages between the concrete segments; thus, the PCSBC
under these loading conditions behaved like a CMC. However, very
0
large differences could be observed between the two columns under
-2,500 higher impact energy (C7–C9) when the slippage between Segment
0 30 60 90 120 150 180 1 and Segment 2 occurred, as given in Fig. 12(b), because a large
Time (ms) amount of the impact energy was absorbed due to shear slippages
(a) and rocking of the segments in PCSBC. Thus, the bending moment
diagrams of the PCSBC at the four critical sections were considerably
12,500
smaller than those of the CMC. Additionally, in the CMC, the bend-
Shear force_PCSBC
10,000 Shear force_CMC ing moment at the intermediate section proportionally increased with
Impact force the PIF while its location moved downward, which causes an uncer-
Force (kN)

7,500
tain parameter in the design stage. For the PCSBC, even the PIF kept
5,000 increasing, the bending moment at the intermediate section was
2,500 nearly steady, and its location was close to the segment joint due to
the rocking of the segment. These bending moment diagrams again
0 demonstrate that the PCSBC outperforms the CMC when the bridge
-2,500 column is subjected to high impact energy due to joint sliding and
joint opening, which absorb a significant amount of impact energy.
-5,000
0 30 60 90 120 150 180
Time (ms) Failures Modes
(b)
The comparison of the PCSBC and the CMC under truck impact in
Fig. 9. Shear force time histories of the two columns under truck terms of cracks and failure modes is presented in Fig. 13. Although
impact (C6) at the (a) column base; and (b) column top. the impact force time histories were almost identical, these columns
responded differently with distinguished types of column damage

© ASCE 04019050-9 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 25,000
Shear force_PCSBC
Shear force_CMC
20,000
PIF_PCSBC
Force (kN)
PIF_CMC
15,000
10,000
5,000
(C1-C6) (C7-C9)
0 0
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.
0 5,000 10,000 15,000 20,000 25,000
PIF (kN)
(a)
PCSBC
5,751 5,392
3,203
5,640
10,267 10,758
7,655 C7 8,424
(c)
Fig. 10. Comparisons of the shear force between the PCSBC and CMC: (a) maximum shear force at the column base; (b) maximum shear force at the
column top; and (c) envelopes of the shear force.
and failures. As presented in Fig. 13(a), the failure mode of the
CMC varied significantly from the flexural crack at the impact area
to the local punching shear failure with the column damage spread-
ing from the column base to its top. When the impact force was
small (C1–C6), minor concrete damage at the impact area and a
flexural crack at the impact point and the column top were observed.
Increasing the PIF to about 16,400 kN (C7), flexural cracks
appeared at the column midheight due to the positive bending
moment at the intermediate section, and a large diagonal shear crack
occurred in the negative side of the column top. Additionally, when
the truck impacted the column with the velocity of 100 km/h and
the engine’s mass of 2 t (C8), yielding the PIF of 18,400 kN, a diag-
onal shear failure was observed at the column base. That large PIF
also yielded other shear cracks at two-thirds of the column. By fur-
ther increasing the velocity to 140 km/h with the engine’s mass of
2 t (PIF = 30,000 kN), the column exhibited a severe punching shear
damage at the impacted area leading to the collapse of the CMC
(Do et al. 2018a). These simulated cracks and failure modes pro-
vided an explanation for the different failures of the bridge column
under vehicle impacts in reality documented by Buth et al. (2010);
these were underestimated by the equivalent static analysis.
For the PCSBC under vehicle impact, when slippage occurred at
the segment joint due to the PIF, the relative displacement between
the segments remained stable during the entire impact process
because the contact forces from the tendon and the segments were
inadequate to pull the segments back to their original position. This
observation was reported in Do et al. (2018b). Therefore, the slip-
page at the segment joints under the PIF, which was presented in
Fig. 11(a), was also the residual slippage between the segments.
© ASCE
J. Bridge Eng., 2019, 24(6): 04019050
25,000
Shear force_PCSBC
Shear force_CMC
20,000
PIF_PCSBC
Force (kN)
PIF_CMC
15,000
10,000
(C1-C6) (C7-C9)
5,000
0 5,000 10,000 15,000 20,000 25,000
PIF (kN)
(b)
CMC
5,527
3,443
5,637 6,198
9,198
6,840 8,510
10,459
C8 7,417 C9
Moreover, when the impact force reached the peak, the opening
also occurred at the first two segments of the PCSBC, as presented
in Fig. 11(b). For Cases 1–6, after the impact force phase, the open-
ing between the segments closed due to the effects of the initial pre-
stress tendon. For Cases 8–9, the combined flexural-shear failure
occurred at the first segment, which caused the collapse of the col-
umn. Hence, in this study, no residual opening was obtained.
Because of the advantages of the segment-to-segment sliding
and rocking, the behaviors of the PCSBC thus differed from the
CMC. As presented in Fig. 13(b), when the impact force was small
(C1-C6), only local concrete damage at the impacted area was
observed, whereas no flexural crack developed along the column.
Although the PIF increased considerably from 16,086 kN (C7) to
about 24,476 kN (C9), the failures of the concrete segment were
similar. When the truck engine collided on the column, a large slip
and opening between the first and the second segment were gener-
ated. Because of the column rocking, the concrete compressive
damage was, therefore, produced in the compression area of the two
segments. Meanwhile, the large shear force from the impact area
also transferred to the footing through the base segment leading to a
predominant diagonal crack of the base segment. That diagonal
shear crack together with the flexural bending damage at the base
joint caused severe damage of the base segment leading to the col-
lapse of the column. Importantly, no sliding, cracks, and concrete
damage were observed in the other sections of the PCSBC. The fail-
ure mode of the PCSBC in this study was consistent with that in the
experimental pendulum impact tests (Hao et al. 2017), confirming
again the reliability of these numerical results. Whether the integrity
of the upper part of the columns above the midheight was
04019050-10 J. Bridge Eng.
 12 all the column elements. The internal energy of each element is
Joint 1 (Footing and Segment 1)
Joint 2 (Segment 1 and segment 2)
computed from the six directions of element stress and strain
10 (Livermore Software Technology Corporation 2007). In each direc-
Joint sliding (mm)
8 tion, the energy is defined by multiplying the stress, incremental
strain, and the element volume (Livermore Software Technology
6 Corporation 2007). Fig. 14(a) indicates that the two columns
(C1-C6) (C7-C9)

4 showed a similar amount of the ED when no shear slippages or

minor joint opening occurred in the PCSBC (C1–C4). When the
2 segmental joint started to open at the first two joints, which
absorbed an amount of the impacted energy in the C5–C6, the
0
0 5,000 10,000 15,000 20,000 25,000 absorbed energy by the PCSBC was thus slightly higher than that
PIF (kN) of the CMC. When the shear slippages between the segments due to
(a) the PIF occurred in the PCSBC (C8–C9), the absorbed energy by
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

the PCSBC (1,041 kN·m, C9) was about 2.5 times higher than that
12 Joint 1 (Footing and Segment 1) of the CMC (405.8 kN·m, C9). This observation indicated that due
Joint opening (JO) (mm)

Joint 2 (Segment 1 and segment 2)
10 Opening moment (OP) (Eq.8)
8 absorption under truck collision.
Mop=2,120 (kNm)
6 kinetic energy (IKE) of the vehicle model is also presented
4
M=1,961 (kNm)
2 JO=0.00 (mm)
0
0 1,400 2,800 4,200 5,600
Bending moment - M (kN.m)
(b)
Fig. 11. (a) Joint sliding; and (b) joint opening of the PCSBC under
truck impacts.
continuous or discontinuous did not affect the response of the col-
umn at the PIF. As presented in Fig. 12(a), when the impact force
reached the PIF, only a part of the column in the vicinity of the
impact point reacted to the impact, whereas other parts of the col-
umn remained in the stationary condition. This observation was
also obtained in previous studies (Pham and Hao 2017a, b; Zhao
et al. 2017, 2018). If the PIF is intensive enough to cause failure in
the PCSBC and CMC, then the shear cracks immediately occur at
the impact area with no involvement of the other parts of the col-
umn. In Cases 7–9, the flexural-shear failure of the PCSBC and the
diagonal shear or punching shear of the CMC happened when the
impact force time histories reached the peak. Thus, the top parts of
the two columns were not involved in resisting the PIF. After that
period, the stress wave propagated from the impact point to the col-
umn top, which caused another flexural-shear crack in the CMC,
whereas the PCSBC observed the joint opening at the other segmen-
tal joints. These results proved the merits of the PCSBC in control-
ling the damage and failure of the bridge column under truck
impacts compared with the CMC, in which the PCSBC failed at the
base segment due to the combined shear and bending damage;
whereas damage occurred at multiple sections of the CMC. In gen-
eral, very localized damage was observed in the PCSBC, whereas
distributed damage occurred in the CMC.
Discussion and Analytical Investigation
Energy Absorption
The energy absorption of the PCSBC and the CMC under truck
impacts is presented and compared in Fig. 14(a). The absorbed
energy of the column is defined by adding up the internal energy of
to the opening and sliding at the joints between the concrete seg-
ments, the PCSBC outperforms the CMC in terms of the energy
The ratio of the absorbed energy by the PCSBC to the initial
M=6,200 (kNm)
JO=10.36 (mm)
in Fig. 14(b). The figure demonstrates that when quadrupling the
IKE of the truck model from 1,111 kN·m (C1, PIF = 1,981 kN) to
M=2,165 (kNm)
JO=0.10 (mm) 4,444 kN·m (C6, PIF = 12,148 kN), the absorbed energy of the
PCSBC showed a moderate increase from 0.95% to about 4.07%.
7,000 This happens because in these cases the response of the column
was almost in the elastic range with no cracks or shear slippages
as previously mentioned, whereas the truck model, which was
less stiff, experienced a large deformation and absorbed most of
the impact energy. The PIF showed a significant influence on the
energy absorption; for example, C9 and C5 had a similar IKE, but
the PIF of C9 was greater than that of C5. The energy absorption
of the PCSBC in C9 was about 34.07% (1051.6 kN·m) of the
total kinetic energy (3086.4 kN·m), whereas the corresponding
energy absorption in C5 was 3.80% (57.18 kN·m) because the
column damage, i.e., large diagonal shear cracks, flexural dam-
age, and concrete damage of the column in C9, absorbed the
impact energy. This proves again that the PIF corresponding to
the engine impact plays a crucial role in the performances of the
structures against truck impact.
The ratio between the energy absorption by Segment 1 and the
total energy absorption is also presented in Fig. 14(b). The ratio pro-
portionally increased with the PIF from 50% (C1) to 79% (C6)
when the PIF increased from 1,981 to 12,148 kN, respectively. That
ratio remained nearly unchanged (80%) after the diagonal shear
cracks appeared on Segment 1 (C7–C9), although the PIF and the
total absorbed energy still increased. In Fig. 4(b) the impacted seg-
ment (base segment) is the key segment of the PCSBC when the
column is collided by a truck. The base segment absorbs a large
amount of the impact energy and governs the capacity of the column
because the failure of the base segment leads to the collapse of all
the bridge structures.
Bending Moment Capacity
Opening Bending Moment
Under lateral impact forces, the behavior of the segmental joint can
be characterized by three main stages, i.e., initial stage, opening
stage, and ultimate stage. The section equilibrium analysis at the
interface between the base segment and the footing is presented in
Fig. 15. At the initial stage [Fig. 15(b)], the column is under com-
pression by the prestressing force and the gravity load. When the
lateral force increases, the compressive stress develops on one side
while stress on another side decreases. During this stage, the col-
umn is still under compression and all the segment joints remained

© ASCE 04019050-11 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 PCSBC CMC ESF according to AASHTO (2012)
4,297
Negative Positive 3,711 ESF
3,646 3,892 4,583
3,122 4,710
4,120
4,481
4,171 t = 29.5 ms t = 32.5 ms
Loading point 4,154
4,360
4,147 2,141
5,582 4,533 2,982
t = 20.5 ms t = 21 ms t = 25.5 ms When the bending Envelope of bending
moment at the top reaches moment
the highest value
(a)

PCSBC CMC
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

2,997 2,516 3,346 3,112

4,168 4,297

4,559 4,710
3,402

2,904 3,542 4,282

3,402 4,360
3,341 3,539 3,163 5,582
4,613 4,526
C3 C4 C5 C6

4,287 4,794 3,388

4,868
4,240
5,970 6,867
3,675 4,201

3,986 5,227 5,618

5,358 6,239 6,607
7,420 4,811 7,883 6,150 8,815 6,166
C7 C8 C9
(b)

Fig. 12. Column bending moment diagrams under truck impacts: (a) bending moment variation along the columns after the impact force reaches the
peak (C6); and (b) envelopes of the bending moment. ESF = equivalent static force.

in contact [Fig. 15(b)]. From the prestressing force of one tendon,
Po = number of tendons; n = total gravity load; and W = initial strain,
ɛo , on a concrete section can be estimated as follows:
nPo þ W T segment joint must be less than 2  ð0:2  0:3Þ fc0 = 0.4–0.6 fc0 .
ɛo ¼ ¼ (4)
S1 S2 Eo S1 S2 Eo
where S1, S2 = section depth and the section width, respectively;
Eo = Young’s modulus of concrete; and T = total vertical force on
the PCSBC at the initial stage.
The opening moment at the base joint [Fig. 15(c)] can be esti-
mated by the following equation:
S1ð=2
n n
Mop ¼ S2 x f ðɛx Þdx þ ðPo þ DPÞd  ðPo  DPÞd
2 2
S1 =2
where d = distance from the tendon to the origin [Fig. 15(a)];
DP = deformation of the prestressed tendon; x = distance from the
infinitesimal, dx, to the origin; and f ðɛx Þ = stress of concrete corre-
sponding to the strain, ɛx . In this analytical method, the stress–strain
curve of concrete proposed by Thorenfeldt et al. (1987) is used.
Normally, the suggested initial stress in PCSBC caused by the
prestressing force and the self-weight was from 0.2 to 0.3 fc0 (Ou
2007), where fc0 = compressive strength of the concrete. In this
study, the initial stress on the column was approximately 0.2 fc0 .
Hence, in the opening stage [Fig. 15(c)], the maximum stress on the
According to Thorenfeldt et al. (1987), the change of the concrete
modulus is minor when the stress is smaller than 0:6 fc0 . Thus, it is
assumed that the modulus of elasticity of concrete is unchanged up
to this stage. As a result, the stress of concrete can be derived as
S1 þ 2x
f ðɛx Þ ¼ Eo ɛx ¼ Eo ɛo (6)
S1
Thus, the opening bending moment at the segment joint can be
expressed by
(5)
TS1
Mop ¼ þ nDP1 d (7)
6
In general, the opening moment is dependent on the section geom-
etry, vertical force, and the stress increase in tendons. However,
unlike the column under static or cyclic load, in which the elongation
in the tendons was normally observed due to the deformation of the
column during the loading process (Hewes and Priestley 2002; Ou
2007; Sideris et al. 2014), under truck impacts, no additional defor-
mation of the prestressed tendons was recorded during the loading

© ASCE 04019050-12 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Failure modes of the bridge columns under truck impacts: (a) CMCs; and (b) PCSBCs.

phase. There was no additional deformation because when the
impact reached its peak, just a part of the column was activated
and responded to the impact force with no movement at the col-
umn top [Fig. 12(a)]. Thus, the influence of the column deforma-
tion on the tendon during this period is neglected. Moreover, no
contact between the concrete segments and the tendons during the
PIF was observed because the joint sliding generally occurred af-
ter the joints opened, as presented in Fig. 11. Therefore, the vibra-
tion of the prestressed tendon due to the interaction between the
segments and the tendons did not happen. As a result, the change
of the prestress tendon force before the joint opens is minimum.
The opening moment thus can be approximated by the following
equation:
TS1
Mop ¼ (8)
6 section is, therefore, equated as
In this study, the total vertical load from the prestress tendons
and the gravity load was 10,600 kN. From Eq. (8), the opening
moment is 2,120 kN·m. This result is compared with that from the
numerical simulation. Very good agreement is achieved, as given in
the Fig. 11(b).
Ultimate Bending Moment
The entire column failed if the base segment was severely damaged
due to the combined flexural bending and the diagonal shear cracks
at the PIF. This was usually associated with damages in the second
segment due to the flexural compression while the other segments
were still intact. These failures of the concrete segments (Segments
1 and 2) occurred immediately when the impact force reached the
peak (C7–C9). The column response at the ultimate stage is pre-
sented in Fig. 15(d). The equilibrium of the axial force acting on the

© ASCE 04019050-13 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 1,250 S1ð=2
n

Energy absorption (kN.m)

PCSBC
C9 S2 DIF f ðɛx Þ dx ¼ W þ a PIF þ ðP0  DPL2 þ DPL1 Þ
1,000 CMC 2
S1 =2cult

750 n
C8 þ ðP0 þ DPR2  DPR1 Þ (9)
2
500
where cult = compressive depth at the ultimate stage; DIF = dynamic
C6
250 increase factor; a  PIF = increase of the axial force due to the
C1 stress propagation caused by the PIF ða ¼ 0:2289Þ as given in
0 Fig. 16; DPL1 ; DPL2 = elongation and shrinkage of the tendons on
0 7,000 14,000 21,000 28,000 the left side (impacted side), respectively; and DPR1 ; DPR2 = shrink-
PIF (kN) age and elongation of the tendons on the right side, respectively.
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.

(a) As presented in Fig. 11(b), the joint opening at these two sections
at the ultimate stage were almost similar (C8, C9), which led to the
100 magnitude of the elongation and shrinkage of the tendons at the
Segment 1 / Total absorbed energy
impact side to be equal to the corresponding elongation and
80
Ratio (%)

60
10,000

Increase of axial force(kN)

Increase of axial force
40 Total absorbed energy / IKE 8,000 Fit curve

20 0.95% 3.80% 6,000 y=0.2289x

4.07% 34.07% 2
(R =0.961)
0 4,000
0 7,000 14,000 21,000 28,000
PIF (kN) 2,000
(b)
0
0 7,000 14,000 21,000 28,000
Fig. 14. Energy absorption of the PCSBCs and CMCs under truck PIF (kN)
impacts: (a) comparison between the PCSBC and CMC; and (b) PCSBC.
IKE = initial kinetic energy. Fig. 16. Axial force versus PIF.

S1
dx
nP W nP
2 o 2 o
S2
d

x
(a) εo
(b)
Mult
No joint opening
n (Po-ΔP +ΔP ) W+ΔPIF n (Po+ΔP -ΔP )
L2 L1 R2 R1
2 2
Mop
ΔPL2 cult
n(P +ΔP) W n(P -ΔP) ΔPR2
2 o 2 o PIF εult
Strain distribution
Impact force ΔPL1 ΔPR1 cult

f(εx)
2εo
εx Stress distribution

(c) (d)

Fig. 15. Segmental joint behavior under impact force: (a) section properties; (b) initial stage; (c) opening stage; and (d) ultimate stage.

© ASCE 04019050-14 J. Bridge Eng.

J. Bridge Eng., 2019, 24(6): 04019050

 shrinkage of the tendons at the other side (DPL1 ¼ DPR2 ; DPL2 ¼
DPR1 ). Further investigation, by examining the prestress load in the
tendons at the ultimate stage, showed only a minor difference in the
tendon force at the two sides of the columns (less than 3%), which is
the difference between the impact response and quasi-static response
of the PCSBC. In the static analysis, when the load is applied at the
column top, the entire tendon is elongated in one side while the ten-
don of the other side shrinks (Ou 2007; Bu et al. 2016). Based on the
previously mentioned observation, Eq. (9) can be rewritten as
S1ð=2
S2 DIF f ðɛx Þdx ¼ T þ a PIF þ nðDPL1  DPL2 Þ
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.
S1 =2cult
The ultimate bending moment can be expressed as
S1ð=2
Mult ¼ S2 DIF x f ðɛx Þdx
S1 =2cult
Also, the tendon elongation and shrinkage can be achieved from
the rotation at the segment, u p , which was defined by Hewes and
Priestley (2002)
 
ɛcu 2ɛ0
up ¼  Lp
cult S1
where cult = depth of section under compression; u p = plastic rota-
tion angle; and Lp ¼ 1; 920 ðmmÞ is the plastic hinge length, which
is recommended for equalizing the segment height of the PCSBC
under truck impact.
In this study, the DIF of the concrete material at the ultimate
strength was 1.325, corresponding to the strain rate of 65 s−1 (Hao
and Hao 2014). However, that DIF in each concrete element in the
compressive area varied. Therefore, in the analytical approach, the
average DIF of 1.16 is used for concrete material in the compressive
area. As a result, the ultimate bending moment and the joint opening
are about 6,863 kN·m and 9.14 mm. The corresponding results from
the numerical simulation were about 6,200 kN·m and 10.36 mm,
respectively [Fig. 11(b)].
Conclusions
In this study, the performance and capacity of the PCSBCs sub-
jected to truck impacts have been investigated and compared with
the reference CMC. The findings of the present study can be sum-
marized as follows:
1. Under different truck impact conditions, similar impact force
time histories for PCSBCs and CMCs were observed because
the vehicle–column interaction was mainly governed by the
column local stiffness instead of the global stiffness.
2. The influences of the PIF on the induced bending moment
and shear force diagrams of the PCSBC and CMC have been
presented. The bending moment of the PCSBC was signifi-
cantly smaller than that of the CMC under high impact force
due to the shear slippage and opening of segmental joints in
PCSBC.
3. The failure of the PCSBCs mainly occurred at the two bottom-
most segments with the combined compression, shear, and flex-
ural failure. The results also proved the advantages of the
© ASCE
J. Bridge Eng., 2019, 24(6): 04019050
PCSBC in localizing the damage of bridge structures under
truck impact compared with the CMC.
4. Under truck impact, the base segment was a crucial element of
the PCSBC, which could absorb up to 80% of the total energy
by the entire column. The failure of this segment may lead to
the total collapse of the bridge structure. Thus, the capability of
the base segment needs to be considered carefully in the design
stages.
5. The analytical method to estimate the opening bending
moment and the ultimate bending moment of the segmental
joint has been proposed. In this method, the DIF and the
(10) increase of the axial force due to the PIF have been taken into
consideration.
In general, PCSBCs exhibit better impact-resistant performance
than CMCs. The damage of the PCSBCs is localized at the two bot-
tommost segments; thus, it is recommended to strengthen these two
segments instead of the entire column in the case of CMCs.
(11)
Acknowledgments
The authors are grateful for the financial support by the Australian
Research Council (ARC). The first author expresses his gratitude
to the Curtin University of Technology for the full Ph.D.
scholarship.
(12)
References
AASHTO. 2012. AASHTO LRFD bridge design specifications: Customary
U.S. units. 6th ed. Washington, DC: AASHTO.
Abdelkarim, O. I., and M. A. ElGawady. 2016. “Performance of hollow-
core FRP–concrete–steel bridge columns subjected to vehicle collision.”
Eng. Struct. 123 (Sep): 517–531.
Abdelkarim, O. I., and M. A. ElGawady. 2017. “Performance of bridge
piers under vehicle collision.” Eng. Struct. 140 (Jun): 337–352. https://
doi.org/10.1016/j.engstruct.2017.02.054.
ACI (American Concrete Institute). 2008. Building code requirements for
structural concrete (ACI 318-08) and commentary (ACI318R-08). ACI
318. Farmington Hills, MI: ACI.
Agrawal, A. K., G. Y. Liu, and S. Alampalli. 2013. “Effects of truck impacts
on bridge piers.” Adv. Mater. Res. 639–649: 13–25. https://doi.org/10
.4028/www.scientific.net/AMR.639-640.13.
Billington, S., and J. Yoon. 2004. “Cyclic response of unbonded postten-
sioned precast columns with ductile fiber-reinforced concrete.” J. Bridge
Eng. 9 (4): 353–363. https://doi.org/10.1061/(ASCE)1084-0702(2004)9:
4(353).
Bu, Z., Y. Ou, J. Song, and G. Lee. 2016. “Hysteretic modeling of unbonded
posttensioned precast segmental bridge columns with circular section
based on cyclic loading test.” J. Bridge Eng. 21 (6): 04016016. https://
doi.org/10.1061/(ASCE)BE.1943-5592.0000902.
Buth, C. E., W. F. Williams, M. S. Brackin, D. Lord, S. R. Geedipally,
and A. Y. Abu-Odeh. 2010. Analysis of large truck collisions with
bridge piers: Phase 1. Rep. of guidelines for designing bridge piers
and abutments for vehicle collisions. Rep. FHWA/TX-10/9-4973-1.
Washington, DC: FHWA.
Chen, L., S. El-Tawil, and Y. Xiao. 2016. “Reduced models for simulating
collisions between trucks and bridge piers.” J. Bridge Eng. 21 (6):
04016020. https://doi.org/10.1061/(ASCE)BE.1943-5592.0000810.
Chen, L., S. El-Tawil, and Y. Xiao. 2017. “Response spectrum-based
method for calculating the reaction force of piers subjected to truck col-
lisions.” Eng. Struct. 150 (Nov): 852–863. https://doi.org/10.1016/j
.engstruct.2017.07.092.
Chen, L., Y. Xiao, G. Xiao, C. Liu, and A. K. Agrawal. 2015. “Test and nu-
merical simulation of truck collision with anti-ram bollards.” Int. J.
Impact Eng. 75 (Jan): 30–39. https://doi.org/10.1016/j.ijimpeng.2014
.07.011.
04019050-15 J. Bridge Eng.
 Chou, C.-C., H.-J. Chang, and J. T. Hewes. 2013. “Two-plastic-hinge and
two dimensional finite element models for post-tensioned precast con-
crete segmental bridge columns.” Eng. Struct. 46 (Jan): 205–217.
https://doi.org/10.1016/j.engstruct.2012.07.009.
Chung, C. H., J. Lee, and J. H. Gil. 2014. “Structural performance evalua-
tion of a precast prefabricated bridge column under vehicle impact load-
ing.” Struct. Infrastruct. Eng. 10 (6): 777–791. https://doi.org/10.1080
/15732479.2013.767841.
Consolazio, G., and M. Davidson. 2008. “Simplified dynamic analysis of
barge collision for bridge design.” Transp. Res. Rec. 2050 (1): 13–25.
https://doi.org/10.3141/2050-02.
Consolazio, G. R., D. J. Getter, and M. T. Davidson. 2009. A static analysis
method for barge-impact design of bridges with consideration of
dynamic amplification. Rep. No. 2009/68091. Gainesville, FL: Univ. of
Downloaded from ascelibrary.org by Louisiana Dept Trans & Dev on 06/19/19. Copyright ASCE. For personal use only; all rights reserved.
Florida, Dept. of Civil and Coastal Engineering.
Culmo, M. P. 2011. Accelerated bridge construction: Experience in design,
fabrication and erection of prefabricated bridge elements and systems.
Washington, DC: FHWA.
Dawood, H., M. Elgawady, and J. Hewes. 2014. “Factors affecting the seis-
mic behavior of segmental precast bridge columns.” Front. Struct. Civ.
Eng. 8 (4): 388–398. https://doi.org/10.1007/s11709-014-0264-8.
Do, T. V., T. M. Pham, and H. Hao. 2018a. “Dynamic responses and failure
modes of bridge columns under vehicle collision.” Eng. Struct. 156
(Feb): 243–259. https://doi.org/10.1016/j.engstruct.2017.11.053.
Do, T. V., T. M. Pham, and H. Hao. 2018b. “Numerical investigation of the
behavior of precast concrete segmental columns subjected to vehicle
collision.” Eng. Struct. 156 (Feb): 375–393. https://doi.org/10.1016/j
.engstruct.2017.11.033.
ElGawady, M., A. Booker, and H. Dawood. 2010. “Seismic behavior of
posttensioned concrete-filled fiber tubes.” J. Compos. Constr. 14 (5):
616–628. https://doi.org/10.1061/(ASCE)CC.1943-5614.0000107.
El-Tawil, S., E. Severino, and P. Fonseca. 2005. “Vehicle collision with
bridge piers.” J. Bridge Eng. 10 (3): 345–353. https://doi.org/10.1061
/(ASCE)1084-0702(2005)10:3(345).
Hao, H., X. Zhang, C. Li, and T. V. Do. 2017. “Impact response and mitiga-
tion of precast concrete segmental columns.” In Proc., 12th Int. Conf. on
Shock and Impact Loads on Structures. Singapore: Ci-Premier Pte.
Hao, Y., and H. Hao. 2014. “Influence of the concrete DIF model on the nu-
merical predictions of RC wall responses to blast loadings.” Eng. Struct.
73 (Aug): 24–38. https://doi.org/10.1016/j.engstruct.2014.04.042.
Hewes, J. T., and M. N. Priestley. 2002. Seismic design and performance of
precast concrete segmental bridge columns. Rep. No. SSRP-2001/25.
La Jolla, San Diego: Dept. of Structural Engineering Univ. of
California, San Diego.
Hung, H.-H., Y.-C. Sung, K.-C. Lin, C.-R. Jiang, and K.-C. Chang. 2017.
“Experimental study and numerical simulation of precast segmental
bridge columns with semi-rigid connections.” Eng. Struct. 136 (Apr):
12–25. https://doi.org/10.1016/j.engstruct.2017.01.012.
Jiang, H., and M. G. Chorzepa. 2015. “An effective numerical simulation
methodology to predict the impact response of pre-stressed concrete
members.” Eng. Fail. Anal. 55 (Sep): 63–78. https://doi.org/10.1016/j
.engfailanal.2015.05.006.
Li, C., H. Hao, and K. Bi. 2017a. “Numerical study on the seismic perform-
ance of precast segmental concrete columns under cyclic loading.” Eng.
Struct. 148 (Oct): 373–386. https://doi.org/10.1016/j.engstruct.2017.06
.062.
© ASCE 04019050-16
J. Bridge Eng., 2019, 24(6): 04019050
Li, C., H. Hao, X. Zhang, and K. Bi. 2017b. “Experimental study of precast
segmental columns with unbonded tendons under cyclic loading.” Adv.
Struct. Eng. 21 (3): 319–334.
Li, J., H. Hao, and C. Wu. 2017c. “Numerical study of precast segmental
column under blast loads.” Eng. Struct. 134 (Mar): 125–137. https://doi
.org/10.1016/j.engstruct.2016.12.028.
Livermore Software Technology Corporation. 2007. LS-Dyna keyword
user’s manual 971. Livermore, CA: Livermore Software Technology
Corporation.
Malvar, L. J., and J. E. Crawford. 1998. “Dynamic increase factors for steel
reinforcing bars [C].” In Proc., 28th DDESB Seminar. Orlando, FL: ANSI.
Megally, S. H., M. Garg, F. Seible, and R. K. Dowell. 2001. Seismic per-
formance of precast segmental bridge superstructures. Rep. No. SSRP-
2001/24, 24. La Jolla, San Diego: Dept. of Structural Engineering Univ.
of California, San Diego.
Ou, Y.-C. 2007. “Precast segmental post-tensioned concrete bridge columns
for seismic regions.” Ph.D. thesis, State Univ. of New York at Buffalo.
Pham, T. M., and H. Hao. 2017a. “Effect of the plastic hinge and boundary
conditions on the impact behavior of reinforced concrete beams.” Int. J.
Impact Eng. 102 (Apr): 74–85. https://doi.org/10.1016/j.ijimpeng.2016
.12.005.
Pham, T. M., and H. Hao. 2017b. “Plastic hinges and inertia forces in RC
beams under impact loads.” Int. J. Impact Eng. 103 (May): 1–11. https://
doi.org/10.1016/j.ijimpeng.2016.12.016.
Sha, Y., and H. Hao. 2013. “Laboratory tests and numerical simulations of
barge impact on circular reinforced concrete piers.” Eng. Struct. 46
(Jan): 593–605. https://doi.org/10.1016/j.engstruct.2012.09.002.
Sharma, H., S. Hurlebaus, and P. Gardoni. 2012. “Performance-based
response evaluation of reinforced concrete columns subject to vehicle
impact.” Int. J. Impact Eng. 43 (May): 52–62. https://doi.org/10.1016/j
.ijimpeng.2011.11.007.
Sideris, P., A. Aref, and A. Filiatrault. 2014. “Quasi-static cyclic testing of a
large-scale hybrid sliding-rocking segmental column with slip-dominant
joints.” J. Bridge Eng. 19 (10): 04014036. https://doi.org/10.1061
/(ASCE)BE.1943-5592.0000605.
Thorenfeldt, E., A. Tomaszewicz, and J. Jensen. 1987. “Mechanical proper-
ties of high-strength concrete and application in design.” In Proc., Symp.
on Utilization of High Strength Concrete, 149–159. Throndheim,
Norway: Tapir.
Zhang, X., H. Hao, and C. Li. 2016a. “Experimental investigation of the
response of precast segmental columns subjected to impact loading.”
Int. J. Impact Eng. 95 (Sep): 105–124. https://doi.org/10.1016/j
.ijimpeng.2016.05.005.
Zhang, X., H. Hao, and C. Li. 2016b. “The effect of concrete shear key
on the performance of segmental columns subjected to impact load-
ing.” Adv. Struct. Eng. 20 (3): 352–373. https://doi.org/10.1177
/1369433216650210.
Zhao, D., W. Yi, and S. Kunnath. 2017. “Shear mechanisms in rein-
forced concrete beams under impact loading.” J. Struct. Eng.
143 (9): 04017089. https://doi.org/10.1061/(ASCE)ST.1943-541X
.0001818.
Zhao, D.-B., W.-J. Yi, and S. K. Kunnath. 2018. “Numerical simulation and
shear resistance of reinforced concrete beams under impact.” Eng.
Struct. 166 (Jul): 387–401. https://doi.org/10.1016/j.engstruct.2018.03
.072.
J. Bridge Eng.