See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/289846969

Evaluation of High-speed Railway Bridge based on Nondestructive Monitoring

System

Article  in  Applied Sciences · January 2016

DOI: 10.3390/app6010024

CITATIONS READS

10 286

3 authors, including:

Mosbeh Kaloop Emad Elbeltagi

Mansoura University Mansoura University
75 PUBLICATIONS   444 CITATIONS    68 PUBLICATIONS   1,776 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Marine Geophysics View project

Improving the GNSS performance by using single receiver View project

All content following this page was uploaded by Mosbeh Kaloop on 19 January 2016.

The user has requested enhancement of the downloaded file.

 applied
sciences
Article
Evaluation of High-Speed Railway Bridges Based on
a Nondestructive Monitoring System
Mosbeh R. Kaloop 1,2 , Jong Wan Hu 1,3, * and Emad Elbeltagi 4
Received: 14 October 2015; Accepted: 11 January 2016; Published: 18 January 2016
Academic Editor: César M. A. Vasques
1 Department of Civil and Environmental Engineering, Incheon National University, Incheon 406-840,
Korea; mosbeh.kaloop@gmail.com
2 Public Works and Civil Engineering Department, Mansoura University, Mansoura 35516, Egypt
3 Incheon Disaster Prevention Research Center, Incheon National University, Incheon 406-840, Korea
4 Structural Engineering Department, Mansoura University, Mansoura 35516, Egypt; eelbelta@mans.edu.eg
* Correspondence: jongp24@incheon.ac.kr; Tel.: +0082-32-835-8463

Abstract: Recently, trains’ velocities in Korea increased more than the speed used in the design
of some bridges. Accordingly, this paper demonstrates the evaluation of a railway bridge due to
high-speed trains’ movement. A nondestructive monitoring system is used to assess the bridge
performance under train speeds of 290, 360, 400 and 406 km/h. This system is comprised of a wireless
short-term acceleration system and strain monitoring sensors attached to the bridge girder. The results
of the analytical methods in time and frequency domains are presented. The following conclusions
are obtained: the cross-correlation models for accelerations and strain measurements are effective to
predict the performance of the bridge; the static behavior is increased with train speed developments;
and the vibration, torsion, fatigue and frequency contents analyses of the bridge show that the bridge
is safe under applied trains’ speeds.

Keywords: high-speed; bridge; strain; monitoring; prediction

1. Introduction
Nowadays, transportation by high-speed trains is considered as one of the important
transportation facilities in South Korea as well as in the world’s high income countries.Therefore,
increasing of the trains’ speed is one of the main problems facing existing infrastructure facilities.
In Korea, high speed trains startedin 1992 from Seoul to Busan, while the Korea Train eXpress (KTX)
services were launched on 1 April 2004.The bridges and infrastructure of express trains have been
developing continuously from 1970. Most high-speed railway bridges are designed based on 350 km/h
velocity. Therefore,with the velocity increase, existing bridges should be redesigned and evaluated.
The newly-completed train HEMU-430X is currently running at high speed over 400km/h in the
transportation network of Korea. Therefore, this study aims to evaluate the existing composite steel
Kaya Bridge of Seoul-Busan High-Speed Railway under the effect of the high speed train movement.
The acceleration and strain measurements are used to evaluate the composite bridge under velocities
between 290 to 406 km/h.
Lee et al. [1] evaluated steel and pre-stressed concrete (PC) box girder bridgesunder high speed
trains up to 289.3 km/h. From their study, they found that no noticeable differences of dynamic
responses due to the different materials (steel or concrete) could be found. Xia et al. [2] evaluated
the real observation for the multi-span PC of high-speed railway bridges in a time domain. In their
study, they recommended the use of the results as a reference for the design of high-speed railway
bridges. Ding et al. [3] used the long term acceleration measurements to evaluate high-speed railway
steel bridges. More monitoring systems for the effect of high-speed railway trains on bridges can

Appl. Sci. 2016, 6, 24; doi:10.3390/app6010024 www.mdpi.com/journal/applsci

Appl. Sci. 2016, 6, 24 2 of 13

be found in [4–7]. In general, the main objective of the structural health monitoring (SHM) systems
is collecting the observations or information to detect and assess bridge condition, damage, fatigue
Appl. Sci. 2016, 6, 24 2 of 13
and performance for proper and timely maintenance intervention. In order to identify the modes of
bridgein characteristics,
[4–7]. In general, the it ismain
necessary
objective toofexcite the structure
the structural in order to(SHM)
health monitoring producesystemsa response
is collectingat each
relevant mode. The loads and response of structures are parameters
the observations or information to detect and assess bridge condition, damage, fatigue and for the monitoring of the bridge
performance
performance under forcurrent
proper andandtimely
futuremaintenance
loadings conditions.
intervention. Typical
In orderSHM implementations
to identify the modes ofin highway
bridge
characteristics,
and steel bridges are it issummarized
necessary to excite the structure
in [8,9]. In addition,in order
for to produce a response
continuous at each relevant
health monitoring studies,
mode. The loads and response of
the response monitoring technique is more suitable [10]. structures are parameters for the monitoring of the bridge
performance
The evaluation under current and future
methodologies of high loadings
speed conditions.
railway Typical
bridgesSHM are implementations
concluded in [11–13]. in
highway and steel bridges are summarized in [8,9]. In addition, for continuous health monitoring
Sartos et al. [14] assessed the stress/strain levels, load distributions, and fatigue for four different
studies, the response monitoring technique is more suitable [10].
bridges based on strain measurements, and they concluded that the system is effective in the
The evaluation methodologies of high speed railway bridges are concluded in [11–13].
staticSartos
performance
et al. [14]analysis.
assessed the et al. [15] used
Xiastress/strain levels, asimulation model and
load distributions, to evaluate
fatigue for vertical and lateral
four different
bridgebridges based on strain measurements, and they concluded that the system is effective in the static and
behavior under high-speed trains. The results of their study showed that the deflections
accelerations
performance of theanalysis.
bridge girder
Xia et al.are[15]
in accordance
used asimulation with the safety
model and comfort
to evaluate standards
vertical and lateralof bridges
bridge and
running train vehicles.
behavior Ding et al.trains.
under high-speed [3] proposed
The results the parametric
of their study (polynomial
showed that fitting) and nonparametric
the deflections and
accelerations of the bridge girder are in accordance with the safety
(correlation models, mean value control, root mean square (RMS)) statistical methodology for the and comfort standards of bridges
and running
acceleration train vehicles.
measurements Ding the
to study et al. [3] proposed
safety the parametric
and early-warning of the (polynomial
bridge. From fitting)
theirand study,
nonparametric (correlation models, mean value control, root mean
they found that the quadratic polynomial fitting provides a good capability for detecting the abnormal square (RMS)) statistical
methodology for the acceleration measurements to study the safety and early-warning of the bridge.
changes of the transverse acceleration measurements. Furthermore, the correlation models describing
From their study, they found that the quadratic polynomial fitting provides a good capability for
the overall structural behavior of the bridge can be obtained with the support of the health monitoring
detecting the abnormal changes of the transverse acceleration measurements. Furthermore, the
system, which includes
correlation cross-correlation
models describing the overallmodelsstructural forbehavior
accelerations. Liu etcan
of the bridge al.be[6]obtained
concluded with that
the the
numerical simulation gives a good relation between the predicted
support of the health monitoring system, which includes cross-correlation models for accelerations. and the measured responses.
Therefore, the[6]statistical
Liu et al. concludedanalysis can be used
that the numerical to detect
simulation givesfatigue, torsionbetween
a good relation and reliability of structures
the predicted and
basedthe
onmeasured
strain andresponses.
displacement measurements
Therefore, the statistical [16–18].
analysisFurthermore,
can be used toparametric
detect fatigue,models areand
torsion used to
detectreliability of structures
the performance based onbased
of structures strainon and displacement
acceleration andmeasurements
strain measurements[16–18]. [19,20].
Furthermore,
The main
parametric
advantage models
of these are used
methods is to
thedetect
abilitythe to
performance
use them of to structures
evaluate based on acceleration
and detect structural and strain
movements
measurements
and damage. [19,20]. The main advantage of these methods is the ability to use them to evaluate
and detect structural movements and damage.
The proposed study aims to evaluate Kaya bridge performance using a nondestructive monitoring
The proposed study aims to evaluate Kaya bridge performance using a nondestructive
system designed to assess the existing bridge under high-speed train movement as well as investigating
monitoring system designed to assess the existing bridge under high-speed train movement as well
the bridge structural the
as investigating behavior
bridgebased on a behavior
structural simple application
based on a of time series
simple and frequency
application of time series analyses
and for
the acceleration
frequency analyses for the acceleration and strain responses. Finally, the effectiveness of the and
and strain responses. Finally, the effectiveness of the monitoring sensors in time
frequency domains
monitoring is assessed
sensors in time andin order
frequencyto decrease
domainsthe monitoring
is assessed system
in order cost. the monitoring
to decrease
system cost.
2. Kaya Bridge, High-Speed Trainsand Monitoring System Descriptions
2. Kaya Bridge, High-Speed Trainsand Monitoring System Descriptions
Kaya Bridge, shown in Figure 1a, is a composite steel box girder bridge with 50 m span supports
Kaya Bridge,
two line high-speed shown inThe
railways. Figure 1a, is a composite
cross-section of thesteel boxis
bridge girder
shownbridge with 501b.
in Figure m span
Two supports
longitudinal
two line high-speed railways. The cross-section of the bridge is shown in Figure
girders are used with a spacing of 6.5 m and reinforced concrete deck with 14.00 m wide, 1b. Two longitudinal
which is used
girders are used with a spacing of 6.5 m and reinforced concrete deck with 14.00 m wide, which is
to provide the large stiffness for such heavy live loads. The use of a steel box as the deck system is
used to provide the large stiffness for such heavy live loads. The use of a steel box as the deck system
another feature in the design of this bridge, which is adopted to reduce uneven deflection and torsion
is another feature in the design of this bridge, which is adopted to reduce uneven deflection and
of thetorsion
deck. ofThe
thebridge design
deck. The criteria
bridge designatcriteria
the mid span
at the midarespan
presented in Table
are presented 1. 1.
in Table

(a)

Figure 1. Cont.
Appl. Sci. 2016, 6, 24 3 of 13
Appl. Sci. 2016, 6, 24 3 of 13

Appl. Sci. 2016, 6, 24 3 of 13

(b)
Figure 1. Kaya bridge (a) view and (b) cross section (dimensions in mm).
Figure 1. Kaya bridge (a) view and(b) (b) cross section (dimensions in mm).
Figure 1. KayaTable
bridge1.Railway
(a) view and (b) design
Bridge cross section
criteria(dimensions
[5]. in mm).
Table 1. Railway Bridge design criteria [5].
Review Table
Factor Criteria
1.Railway Bridge design criteria [5].Note
VerticalFactor
acceleration 4.9 m/s2 For concrete
Note track
Review Criteria Note
Review Factor Criteria
Displacement
Vertical as safety
acceleration 82
4.94.9 mm
m/s 2 Under
For 350 km/h
concrete track
Vertical acceleration m/s2 For concrete track
Displacement
Displacement as comfort
as safety 8222mmmm Under
Under350350km/h
km/h
Displacement as safety 82 mm Under 350 km/h
Displacement as twist
Track comfort 22
0.4mm
mm/m Under 350
By dynamic km/h
analysis
Displacement
Track twist as comfort 22
0.4 mm/mmm Under 350 km/h
By dynamic analysis
Track twist 0.4 mm/m By dynamic analysis
The next-generation high-speed developed train (HEMU-430X) is intended to travel with a
maximum
The speed of 430 km/h
next-generation (Figure 2).The high-speed
high-speed train is controlled tois pass over the
tobridge in
The next-generation high-speed developed train(HEMU-430X)
developed train (HEMU-430X) intended
is intended to travel travel
with a with
the passage track with four different speeds, i.e., 290, 360, 400, and 406 km/h.
a maximum
maximum speed
speedof of
430 km/h
430 km/h(Figure 2).Thehigh-speed
(Figure 2).The high-speedtraintrain
is is controlled
controlled to pass
to pass overover the bridge
the bridge in in
the passage tracktrack
the passage with four
with different
four different speeds,i.e.,
speeds, i.e.,290,
290, 360, 400,and
360, 400, and406406km/h.
km/h.

Figure 2. Axial spacing and loading of the high-speed train.

Figure of
To monitor the Figure
behavior 2. 2. Axial spacing
the bridge
Axial spacing and
asand loading
perloading
the passingof theof
of thehigh-speed train. trains, a wireless SHM
the high speed
high-speed train.
system was designed and installed on the Kaya Bridge shortly after it was opened to use the
To monitor the behavior of the bridge as per the passing of the high speed trains, a wireless SHM
HEMU-430X train, as shown in Figure 3. This on-line concise SHM system was designed with a
To monitor
system the behavior
was designed of the bridge
and installed on theasKayaper the Bridgepassing
shortlyofafter
the high
it wasspeedopened trains,
to usea wireless
the
minimum of 20 sensors to monitor the key parameters. Moreover, a total of 20 accelerometer and
SHMHEMU-430X
system wastrain,designed
as shown andininstalled
Figure 3. on Thisthe Kayaconcise
on-line BridgeSHM shortlysystemafter wasit designed
was opened with toa use
strain gauge sensors (five sensors on each railway track)with sampling frequency of 100Hz were
minimum of 20
the HEMU-430X sensors
train, as to monitor
shown in the key3.parameters.
Figure This on-line Moreover,
concise aSHM
total system
of 20 accelerometer
was designed and with
installed on the main girders under each railway track in each direction with equal-spacing of 8.3 m.
strain gauge
a minimum sensors
of 20were
sensors (five sensors on
to monitor theeach
key railway
parameters. track)with sampling frequency of 100Hz were and
The sensors installed at the bottom flange of the mainMoreover,
girders in the a total of 20
vertical accelerometer
direction to detect
installed
strainand
gauge on the
sensors main girders
(five sensors under each
onand railway
eachfatigue
railway track in
track) each direction
withassampling with equal-spacing of 8.3Hzm.
monitor the vertical vibration of the bridge, illustratedfrequency
in Figure 3.ofThe 100entirewere
The sensors were installed at the bottom flange of the main girders in the vertical direction to detect
installed on the
system mainofgirders
consists a set ofunder
sensors,each railway
data track data
acquisition, in each direction with
transmission, data equal-spacing
management and of a8.3 m.
and monitor the vertical vibration and fatigue of the bridge, as illustrated in Figure 3. The entire
structural evaluation mechanism. The primary purpose of the system
The sensors were installed at the bottom flange of the main girders in the vertical direction to detect is to monitor in-service
system consists of a set of sensors, data acquisition, data transmission, data management and a
performance
and monitor of the vibration
theevaluation
vertical bridge structure
and The underofhigh-speed
fatigue bridge,trains,
thepurpose as and to provide early warning of
structural mechanism. primary of illustrated
the systeminisFigure 3. The
to monitor entire system
in-service
abnormal
consists changes in
of a set ofofsensors, in-service
data performance
acquisition, of the bridge. Herein, the structural parameters to be
performance the bridge structure underdata transmission,
high-speed trains, anddatatomanagement
provide earlyand a structural
warning of
monitored in the Kaya Bridge were determined using the structural sensitivity analysis with the finite
evaluation
abnormalmechanism.
changes in The primary
in-service purpose of
performance of the
thebridge.
systemHerein,
is to monitor in-service
the structural performance
parameters to be of
element model under the designed train speeds action, as shown in Kim et al. [5].Other important
the bridge structure
monitored under
in the Kaya high-speed
Bridge trains, and
were determined using to the
provide early
structural warninganalysis
sensitivity of abnormal
with thechanges
finite in
parameters including accelerations, strains, deformations and fatigue of the steel box girder bridge
element
in-service model
performanceunder the designed train speeds action, as shown in Kim et al. [5].Other important
were monitored and of the bridge.
studied Herein, speeds
under different the structural parameters
of the passing high-speed to betrains.
monitored in the Kaya
Bridgeparameters including using
were determined accelerations, strains, sensitivity
the structural deformations and fatigue
analysis withofthe thefinite
steel element
box girder bridge
model under
were monitored and studied under different speeds of the passing high-speed
the designed train speeds action, as shown in Kim et al. [5]. Other important parameters including trains.
accelerations, strains, deformations and fatigue of the steel box girder bridge were monitored and
studied under different speeds of the passing high-speed trains.
Appl. Sci. 2016, 6, 24 4 of 13
Appl. Sci. 2016, 6, 24 4 of 13

Appl. Sci. 2016, 6, 24 4 of 13

Figure 3. Structural health monitoring (SHM) system locations and components.

Figure 3. Structural health monitoring (SHM) system locations and components.
3. Evaluation of the Bridge Condition Using SHM Monitoring Data
3. Evaluation
Kimof the
et al. [5]Bridge Condition
designed a 3-D finite Using
elementSHM
modelMonitoring DataBridge using ANSYS V10.0
(FEM) of the Kaya
Figure 3. Structural health monitoring (SHM) system locations and components.
Kim software (ANSYS,
et al. [5] designed Canonsburg,
a 3-D finite PA, element
USA, 2005) to support
model (FEM)the proposed
of the Kaya movement
Bridge using analysis
ANSYS of V10.0
different train’s speeds. Train speeds from 200 to 450 km/h were studied. The FEM model results are
software (ANSYS, of
3. Evaluation Canonsburg, PA, USA, Using
the Bridge Condition 2005) toSHM support the proposed
Monitoring Data movement analysis of different
concluded as follows: (i) the natural frequency modes of the bridge are 3.186 (1st bending), 3.689 (2nd
train’s speeds.
bending) Train
Kim et and speeds
al. [5]5.913
designed from 200 to
a 3-D finite
(1st torsion) 450
Hz for km/h
element were
model
the first, studied.
(FEM)
second and The
of the
third FEM
Kaya model
Bridge
modes, results
using ANSYS
respectively; are the
(ii) concluded
V10.0
as follows:
software (i) (ANSYS,
maximum thevertical
natural frequency
acceleration
Canonsburg, and modes
PA, USA,of2005)
displacementthe bridge
are
to 3.5 are 3.186
m/s
support 2 and (1st occurred
the 6proposed
mm bending), 3.689
with
movement train (2nd
speedbending)
analysis of
and 5.913250(1st
different torsion)
andtrain’s
280 km/h, Hzrespectively;
speeds. for thespeeds
Train first,(iii)
second
the 200
from andto third
maximum modes,
wererespectively;
acceleration
450 km/h and displacement
studied. The FEM(ii) the maximum
for
model the vertical
train are
results
400 km/h
acceleration
concluded and speed
as are 1.6
displacement
follows: (i) m/s and
are 3.8
the2natural mm,
3.5frequency2 and
m/srespectively.6 mm
modes Therefore,
ofoccurred
the bridge comparing
with the
train
are 3.186 FEM
speed
(1st results
250 and
bending), and
3.689the
280
(2ndkm/h,
bridge design
bending) and criteria
5.913 in Table
(1st 1 shows
torsion) Hz thatthe
for the first,
bridgesecond
is safe under
and applied
third loads.respectively; (ii) the
modes,
respectively; (iii) the maximum acceleration and displacement for the train 400 km/h speed are
maximum
2 and The vertical
current study utilizesand
acceleration the real monitoringare
displacement data that
3.5the
m/s is collected
2 and 6 mmfrom the SHM
occurred withsystem
train of
speed
1.6 m/s Kaya
3.8 mm,
Bridge to
respectively.
assess and
Therefore,
evaluate the real
comparing
bridge condition
FEM
in
results
terms of
and
its
the bridge
vibration, static
design
strain,
criteria
250 and 280 km/h, respectively; (iii) the maximum acceleration and displacement for the train
in Table 1 shows that the bridge is safe under applied loads.
torsional and fatigue behavior of the steel deck as well as evaluating and comparing the frequency
400 km/h speed are 1.6 m/s2 and 3.8 mm, respectively. Therefore, comparing the FEM results and the
The current
contents forstudy utilizes the of
the measurements real
themonitoring data that is collected from the SHM system of Kaya
monitoring sensors.
bridge design criteria in Table 1 shows that the bridge is safe under applied loads.
Bridge to Theassess and evaluate
currentofstudy
the
utilizes
real
the
bridge condition
real monitoring
in terms of its vibration, static strain, torsional
3.1. Evaluation the Bridge Girder Vibration Behavior data that is collected from the SHM system of
and fatigue behavior
Kaya Bridge of the
to assess andsteel deck as
evaluate thewell
real as evaluating
bridge conditionand incomparing
terms of its the frequency
vibration, staticcontents
strain, for
torsionalFigure
the measurements and 4a,b illustrate
of the
fatigue the typical
monitoring
behavior of the vertical
sensors.
steel deckacceleration
as well astime historiesand
evaluating of the passage monitoring
comparing the frequency
points of the girder measured from accelerometers 1 to 5 for the 290 km/h and 406 km/h speed,
contents for the measurements of the monitoring sensors.
respectively.
3.1. Evaluation It can
of the be seen
Bridge thatVibration
Girder the dt, (dt Behavior
= t2 (leave time) − t1 (entrance time)) values are reported as
3.1.5.9, 2.92, 1.9 and 1.85 s with the speeds 290, 360, 400 and 406 km/h, respectively. This means that the
Evaluation of the Bridge Girder Vibration Behavior
Figure 4a,btime
vibration illustrate
effect the typical
on the vertical
bridge decreasedacceleration
by 68.65%time as thehistories of theincreased
train speed passage from
monitoring
points of290Figure
the 4a,b
km/h.illustrate
girder
to 406 measured thefrom
In addition, typical
it vertical
thatacceleration
accelerometers
is observed 1 to 5time
the vibrations for histories
290 of
themonitored
of the thepoints
km/h passagea monitoring
andat406 km/h
speed of speed,
points
respectively. ofItthe
290 km/h are
cangirder measured
approximately
be seen thatthe
thefrom
dt, accelerometers
same, while
(dt = tas 1time)
the speed
(leave to 5´ fort the
increased, the290 km/h
entrance
(entrance andexit
and
time)) 406 km/hare
speed,
monitoring
values reported
2 1
points are experiencing
respectively. It can be high
seen vibration
that the dt, response
(dt = t compared
2 (leave time) totthe
− mid span
1 (entrance point.values are reported as
time))
as 5.9, 2.92, 1.9 and 1.85 s with the speeds 290, 360, 400 and 406 km/h, respectively. This means that
5.9, 2.92, 1.9 and 1.85 s with the speeds 290, 360, 400 and 406 km/h, respectively. This means that the
the vibration time effect on the bridge decreased by 68.65% as the train speed increased from 290 to
vibration time effect on the bridge decreased by 68.65% as the train speed increased from
406 km/h. In addition, it is observed that the vibrations of the monitored points at a speed of 290 km/h
290 to 406 km/h. In addition, it is observed that the vibrations of the monitored points at a speed of
are approximately the same, while
290 km/h are approximately as thewhile
the same, speed as increased, the entrance
the speed increased, and exitand
the entrance monitoring points are
exit monitoring
experiencing high vibration response compared to the mid span
points are experiencing high vibration response compared to the mid span point. point.

(a) (b)

Figure 4. Vertical acceleration time histories of the girder caused by a high-speed train. (a)passage
response points for 290 km/h; (b) passage response points for 406 km/h.

(a) (b)

Figure 4. Vertical acceleration time histories of the girder caused by a high-speed train. (a)passage
Figure 4. Vertical acceleration time histories of the girder caused by a high-speed train. (a) passage
response points for 290 km/h; (b) passage response points for 406 km/h.
response points for 290 km/h; (b) passage response points for 406 km/h.
Appl. Sci. 2016, 6, 24 5 of 13

Appl. Sci. 2016, 6, 24 5 of 13

Figure 5 shows that the vertical acceleration of the mid-span point of the girder at 406 km/h
is much
Appl.smaller
Sci. 2016, 6,than
24 that of 290 km/h. In addition, it is noticed that the vibration
Figure 5 shows that the vertical acceleration of the mid-span point of the girder at 406 km/h is
response 5 of for
13 the
passage
muchway is higher
smaller thanthan
that that
of 290 ofkm/h.
the opposite side itatis400
In addition, and 406
noticed thatkm/h, while at
the vibration 290 andfor
response 360the km/h,
Figure 5 shows that the vertical acceleration of the mid-span point of the girder at 406 km/h is
the acceleration responses for the passage and opposite sides are equal. These
passage way is higher than that of the opposite side at 400 and 406 km/h, while at 290 and 360 km/h, indicate that, although
much smaller than that of 290 km/h. In addition, it is noticed that the vibration response for the
the structural layouts
the acceleration of the five
responses for monitoring
the passage andpoints of thesides
opposite mainaregirder
equal.are the same,
These indicate there
that,isalthough
a significant
passage way is higher than that of the opposite side at 400 and 406 km/h, while at 290 and 360 km/h,
the structural
difference betweenlayouts of the five
the vertical monitoring
vibration points of the
characteristics mainsides
of two girderofare thethe same, due
bridge theretoisthe a rail
the acceleration responses for the passage and opposite sides are equal. These indicate that, although
significant
irregularity in difference
the vertical between
directionsthe with
vertical vibration
different characteristics
speeds. Thus, of two
there is a sides
need oftothe bridge due
monitor the to
vertical
the structural layouts of the five monitoring points of the main girder are the same, there is a
the rail irregularity
accelerations in the
in vertical directions sowith different
realizespeeds. Thus, there isfora need to monitor
significantofdifference
the span the the
between long term vibration
vertical as to anomaly
characteristics of twoalarms
sides of the vibration
bridge due behavior
to
the vertical accelerations of the span in the long term so as to realize anomaly alarms for vibration
of thethe
main girder.In addition,
rail irregularity the measured
in the vertical vertical
directions with acceleration
different is smaller
speeds. Thus, there is than
a need the
to FEM
monitor results
behavior of the main girder.In addition, the measured vertical acceleration is smaller than the FEM
by 62%the according
vertical accelerations of thecriteria
to the design span inof the
thelong term (as
bridge so as to realize
shown Tableanomaly
1),which alarms for vibration
means that the real
results by 62% according to the design criteria of the bridge (as shown Table 1),which means that the
behavior
response of the main girder.In addition, the measured vertical acceleration
is safe. Furthermore, the girder torsional behavior should be studied with new development is smaller than the FEM
real response is safe. Furthermore, the girder torsional behavior should be studied with new
results
trains’ by 62% according
speeds. trains’ speeds. to the design criteria of the bridge (as shown Table 1),which means that the
development
real response is safe. Furthermore, the girder torsional behavior should be studied with new
development
0.8 trains’ speeds. 0.8
290 290
0.6 360 0.6 360
0.8 400 0.8 400
406
290 406
290
0.4 0.4
0.6 360 0.6 360
400 400
0.2 406 0.2 406
0.4 0.4
(ms -2)

(ms-2)
0 0
0.2 0.2
Acc (ms -2Acc

Acc (ms-2Acc
)

)
-0.2 -0.2
0 0

-0.4 -0.4
-0.2 -0.2

-0.6 -0.6
-0.4 -0.4

-0.8 -0.8
-0.60 2 4 6 8 10 12 14 -0.60 2 4 6 8 10 12 14
Time (Sec.) Time (Sec.)

-0.8
0 2 4 (a)
6 8 10 12 14
-0.8
0 2 4
(b)
6 8 10 12 14
Time (Sec.) Time (Sec.)

Figure 5. Vertical acceleration

(a) time histories of the passage and opposite mid
(b)span girder point. (a)
Figure 5. Vertical acceleration time histories of the passage and opposite mid span girder point.
passage response of mid span point; (b) opposite response of mid span point.
(a) passage
Figure response
5. Verticalof mid span time
acceleration point; (b) opposite
histories response
of the passage ofopposite
and mid span point.
mid span girder point. (a)
passage response of mid span point; (b) opposite response of mid span point.
Figure 6 illustrates the dynamic torsional behavior of the bridge girder based on vertical
Figure 6 illustrates
acceleration the dynamic
measurements [21].The torsional behaviorofofthethegirder
torsional behavior bridge
duegirder based
to train on vertical
passage at
Figure 6 illustrates the dynamic torsional behavior of the bridge girder based on vertical
406 km/hmeasurements
acceleration is shown in Figure 6a. While the torsional behavior ofgirder
the middue
span point atpassage
differentatspeeds
acceleration measurements [21].The torsional behavior of the girder due to train passage atkm/h
[21].The torsional behavior of the to train 406
is
is shown shown in Figure
in Figure 6a. 6b.
While the
406 km/h is shown in Figure 6a.torsional
While thebehavior
torsional of the mid
behavior span
of the point
mid spanatpoint
different speeds
at different is shown
speeds
in Figure 6b.
is shown in Figure 6b. a − a
T= (1)
pass a ´a
a − opps
a T“ (1)
where T, a ,a T = l
and l are the torsion, acceleration for the passage and opposite points and (1)

T,, aaopps, aand l and

wheredistances
T, apass between sensors’ positions.
are the torsion, acceleration for the
where l are the torsion, acceleration forpassage and opposite
the passage points
and opposite and distances
points and
between sensors’ positions.
distances0.5 between sensors’ positions. 0.05
0 P1 290
0
-0.5
0.50 2 4 6 8 10 12 14 0.05
-0.05
0.1 P1 0 2 4 6 8 10 12 29014
) (rad/s )

0
2

0 P2 0.10
-0.5 360
-0.1 0 2 4 6 8 10 12 14 -0.05
0
Torsional
) (rad/s )

0.10 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
2

0.2 0.1
Torsional (rad/s

0 P2 -0.1
0 P3 0 2 4 6 8 10 12 36014
Torsional

-0.1 0.20
-0.2 0 2 4 6 8 10 12 14
2
Torsional (rad/s

0.20 2 4 6 8 10 12 14 400
0.2 -0.1
0
0 P3 0 2 4 6 8 10 12 14
0 P4 0.2
-0.2 -0.2
0 2 4 6 8 10 12 40014
-0.2 0 2 4 6 8 10 12 14
0.20 2 4 6 8 10 12 14 0.20
1 P4 406
0 -0.2
0
0 P5 0 2 4 6 8 10 12 14
-0.2
0.2
-1 0 2 4 6 8 10 12 14 -0.2
10 2 4 6 8 10 12 14 0 2 4 6 8 10 12 40614
Time (sec) P5 0 Time (sec)
0
-1
0 2 4 (a)
6 8 10 12 14
-0.2
0 2 4 (b)
6 8 10 12 14
Time (sec) Time (sec)

Figure 6. Torsional measurements

(a) for the bridge girder. (a)the torsional of the
(b)monitoring points at
406 km/h passage; (b)mid span point torsional values.
Figure 6. Torsional measurements for the bridge girder. (a)the torsional of the monitoring points at
Figure 6. Torsional measurements for the bridge girder. (a) the torsional of the monitoring points at
406 km/h passage; (b)mid span point torsional values.
It is noticed
406 km/h that
passage; (b) the
midtorsional values
span point increased
torsional as the train velocity increased with maximum
values.
values at points 1 and 5. As the speed increased from 290 to 406 km/h, the torsional values increased
It is noticed that the torsional values increased as the train velocity increased with maximum
values
It at points
is noticed 1 and
that the5.torsional
As the speed increased
values from as
increased 290the
to 406 km/h,
train the torsional
velocity values
increased increased
with maximum
values at points 1 and 5. As the speed increased from 290 to 406 km/h, the torsional values increased by
Appl. Sci. 2016, 6, 24 6 of 13

61.20%, while they increased by 22.33% as the train speed increased from 360 to 406 km/h. This means
Appl. Sci. 2016, 6, 24 6 of 13
that the bridge deck torsional response is higher than the vibration response. The girder torsional
value by
at the midwhile
span they
point, 0.103 rad/s 2 , is within the design values, as shown in Table 1.
61.20%, increased by 22.33% as the train speed increased from 360 to 406 km/h. This
The
means statistical parameters,
that the bridge maximum
deck torsional and isroot
response mean
higher than square (RMS),
the vibration for theThe
response. acceleration
girder
measurements and torsional
torsional value at the midcalculations at 360
span point, 0.103 and
rad/s 2 , 406 km/h
is within are
the presented
design values,inasTable
shown 2. in
The 3601.km/h
Table
The statistical parameters, maximum and root mean square
is assumed as the base-speed for the study of the bridge safety because the bridge design speed is(RMS), for the acceleration
measurements
350 km/h. Furthermore, and torsional calculations
the Bessel at 360 and
filter cut-off 406 km/hofare
frequency 30presented in Table
Hz is applied to2.remove
The 360 the
km/hnoise
is assumed as the base-speed for the study of the bridge safety because the
measurements of the accelerometer [21]. It is noticed that the maximum acceleration occurred at points bridge design speed is
350 km/h. Furthermore, the Bessel filter cut-off frequency of 30 Hz is applied to remove the noise
1 and 5 for the passage and opposite directions at 406 km/h speed, respectively, with values within the
measurements of the accelerometer [21]. It is noticed that the maximum acceleration occurred at
design limits as shown in Table 1. While there is no high relative change at points 2 to 4, the RMS of
points 1 and 5 for the passage and opposite directions at 406 km/h speed, respectively, with values
the 360 km/h
within theisdesign
smaller than
limits as that
shown of in
the 406 1.km/h
Table Whileon passage
there is no highway,relative vice versa
while change in the2opposite
at points to 4,
way except for point 5. The maximum torsion occurred at point 5 at a speed
the RMS of the 360 km/h is smaller than that of the 406 km/h on passage way, while vice versa of 406 km/h with a value
in the
very close to the design value (Table 1). Thus, it is recommended to limit
opposite way except for point 5. The maximum torsion occurred at point 5 at a speed of 406 km/h the train speed to 400 km/h
only. In addition,
with the effective
a value very torsion
close to the designtest should
value (Tablebe 1).assessed
Thus, it isatrecommended
the end points of the
to limit thebridge (point of
train speed
to 400torsion).
maximum km/h only. In addition, the effective torsion test should be assessed at the end points of the
bridge (point of maximum torsion).
Table 2. Maximum acceleration, root mean square (RMS) and torsional acceleration for the filtration of
Table
vibration 2. Maximum acceleration, root mean square (RMS) and torsional acceleration for the filtration
measurements.
of vibration measurements.

Acceleration (m/s2(m/s
Acceleration ) 2) RMS
RMS 2)2 )
(m/s
(m/s Max Torsion
Max
(rad/s2 )
Point Passage Opposite Passage Opposite Torsion
Point Passage Opposite Passage Opposite
406 (rad/s
360)
2
360 406 360 406 360 406 360 406
1 0.104 360 1.251 406 0.196360 406
0.079 360
0.006 406
0.035 360
0.008 406 0.004360 406
0.059 0.191
2 10.169 0.1040.1831.251 0.231
0.196 0.088
0.079 0.006
0.008 0.035
0.009 0.008
0.011 0.0040.0060.059 0.191
0.072 0.191
3 20.244 0.1690.3190.183 0.279
0.231 0.146
0.088 0.010
0.008 0.012
0.009 0.013 0.0060.0070.072
0.011 0.079
0.191 0.037
4 30.569 0.2440.6220.319 0.252
0.279 0.239
0.146 0.014
0.010 0.016
0.012 0.012 0.0070.0080.079
0.013 0.244
0.037 0.055
5 - 0.401 0.169 2.497 - 0.013 0.008 0.035 - 0.343
4 0.569 0.622 0.252 0.239 0.014 0.016 0.012 0.008 0.244 0.055
5 - 0.401 0.169 2.497 - 0.013 0.008 0.035 - 0.343
Ding et al. [3,17] concluded that the quadratic linear fitting for the maximum and RMS are
calculatedDing et al.
for the [3,17] concluded
acceleration that the quadratic
measurements linear to
can be used fitting forthe
detect theperformance
maximum andofRMS are
high-speed
calculated
railway bridgesfor the acceleration
based on one spanmeasurements
monitoring.can be used
Herein, to method
this detect theisperformance of high-speed
applied to detect and check
railway bridges
the performance based
of the on one
bridge span
due monitoring.
to different Herein,
speeds onthis
themethod
two waysis applied
of thetotrack.
detectThe
andmaximum
check
the performance of the bridge due to different speeds on the two ways of the track. The maximum
and RMS of 150 acceleration of the original measurements are shown in Figure 7. The quadratic
and RMS of 150 acceleration of the original measurements are shown in Figure 7. The quadratic linear
linear fitting is suitable for the maximum acceleration (Figure 7a), while the RMS shows no correlation
fitting is suitable for the maximum acceleration (Figure 7a), while the RMS shows no correlation
between the passage
between andand
the passage opposite ways.
opposite ways.Therefore,
Therefore, the
the maximum acceleration
maximum acceleration fitting
fitting cancan be used
be used to to
checkcheck
the safety of the bridge.
the safety of the bridge.
0.6 -3
x 10
2 6
y = - 0.34*x + 1.2*x - 0.001
0.5
5

0.4
Acc Max. of Opp (ms-2)

RMS of Acc Opp (ms-2)

4
0.3

3
0.2

2
0.1

0 1

-0.1 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0 1 2 3 4 5 6 7 8
Acc Max. of Pass (ms -2) RMS of Acc Pass (ms -2) -3
x 10

(a) (b)

Figure 7. Cross-correlation between the passage and opposite way responses for 360 km/h. (a)
Figure 7. Cross-correlation between the passage and opposite way responses for 360 km/h.
cross-correlation maximum acceleration; (b) cross-correlation roote mean square (RMS) acceleration.
(a) cross-correlation maximum acceleration; (b) cross-correlation roote mean square (RMS) acceleration.
Appl. Sci. 2016, 6, 24 7 of 13

In this case, the fitting equation is used to predict the opposite way measurements for the 400 and
406 km/h, as2016,
Appl. Sci. shown6, 24 in Figure 8. The correlation coefficient of the measured maximum values
7 of 13 and
predicted values with 400 and 406 km/h is 0.999 and 0.998 for opposite accelerations, respectively.
The resultsInindicate
this case,that
the good
fitting cross-correlation
equation is used to predict the opposite
exists between theway measurements
maximum valuesforof
thethe
400opposite
and 406 km/h, as shown in Figure 8. The correlation coefficient of the measured maximum values
way at the two train speeds. Herein, the control chart can be used to monitor the changes in the
and predicted values with 400 and 406 km/h is 0.999 and 0.998 for opposite accelerations, respectively.
verticalThe
accelerations caused by deterioration of the vibration behavior. Firstly, the condition index
results indicate that good cross-correlation exists between the maximum values of the opposite
e (e = Max (measurements)
way at the two train speeds. ´ Max (prediction))
Herein, forchart
the control early warning
can be used of abnormal
to monitor the vibration
changes in behavior
the is
definedvertical
as theaccelerations
difference between
caused bythe measuredofand
deterioration predicted
the vibration maximum
behavior. values
Firstly, of the accelerations
the condition index e in
(e = Max
the opposite (measurements)
way. Then, a mean − Max (prediction))
value for early
control chart warning of to
is employed abnormal
monitor vibration
the timebehavior
seriesisof e with
regard defined as the difference
to the opposite between the
accelerations. measured
For onlineand predicted maximum
monitoring, values ofparameter
the controlling the accelerations
is chosen so
in the opposite way. Then, a mean value control chart is employed to monitor the time series of e with
that, when the structural vibration is in good condition, all observation samples fall between the
regard to the opposite accelerations. For online monitoring, the controlling parameter is chosen so
controlthat,
limits.
when When the new vibration
the structural measurement is made,
is in good the structural
condition, all observationabnormal
samplesvibration condition
fall between the can
be detected if an unusual number of samples fall beyond the control limits.
control limits. When the new measurement is made, the structural abnormal vibration condition canIn this study, the control
condition is assumed
be detected if an with
unusual290number
km/h of found within
samples 0.025 to
fall beyond the´0.03
controlm/s 2 . The
limits. calculated
In this study, thecondition
control index
for the condition
400 and is 406assumed
km/hwith 290 km/h0.0043
is between found within 0.025 tom/s
and ´0.028 2
−0.03 m/s 2 . The calculated
. Hence, a long-termcondition index of the
monitoring
for the 400 and 406 km/h is between 0.0043 and −0.028 m/s2. Hence, a long-term monitoring of the
maximum values of the opposite accelerations can help in the early-warning of the vibration behavior
maximum values of the opposite accelerations can help in the early-warning of the vibration behavior
deterioration. These results are identical with the Ding et al. [3,17] conclusions.
deterioration. These results are identical with the Ding et al. [3,17] conclusions.

(a) (b)

Figure 8. Predicting the effect of cross-correlation between maximum values of accelerations by using
8. Predicting
Figure quadratic the effect of cross-correlation between maximum values of accelerations by using
polynomials. (a) 400 km/h; (b) 406 km/h.
quadratic polynomials. (a) 400 km/h; (b) 406 km/h.
3.2. Evaluation of the Bridge Girders’ Train Responses
3.2. Evaluation
The of the Bridge
measured Girders’
vertical strainTrain Responses
histories of the monitoring points of the passage way for the train
speeds 290 and 406 km/h are presented in Figure 9a,b. The vertical strain of the mid-span point of
The measured vertical strain histories of the monitoring points of the passage way for the train
passage and opposite ways for different train speeds are compared in Figure 9c,d. As the train speeds
speedsincrease
290 and to 406
290, km/h
360, 400areandpresented
406 km/h, the in strain
Figure 9a,b. The
responses (dt) vertical
decrease strain of the
to 2.1, 1.93, 1.64mid-span
and 1.61 s, point of
passagerespectively.
and opposite ways for different train speeds are compared in Figure
This shows that the time of static strain responses decreases by 23.33% when the 9c,d. As thespeed
train speeds
increase to 290,from
changes 360,290 400and
and406 406 km/h,
km/h. the strain
In addition, theresponses
maximum (dt) decrease
strain responsetoin2.1,the1.93, 1.64and
passage and 1.61 s,
opposite
respectively. ways
This occurred
shows thatat the
a speed
time of of
290static
km/h.strain
The results as suchdecreases
responses show that the bystatic and when
23.33% dynamic the speed
behavior of the bridge is higher with low train speeds. In addition, the
changes from 290 and 406 km/h. In addition, the maximum strain response in the passage and opposite strain measurements of the
bridge points are highly correlated (0.95) with each speed change. This means that the strain
ways occurred at a speed of 290 km/h. The results as such show that the static and dynamic behavior
measurement of the mid span point can be used to detect the performance of the whole bridge. This
of the bridge is will
situation higher withthe
decrease low train
cost speeds.
of the In addition,
monitoring system due thetostrain
the usemeasurements
of one monitoring ofpoint
the bridge
only. points
are highly
The correlated (0.95)
high correlation with
(0.99) ofeach
strainspeed
response change.
for theThis means
passage and that the ways
opposite strainoccurred
measurement of the mid
at 400 and
406 km/h.
span point can be It used
meansto that the strain
detect response of the
the performance oftwo
the speeds
wholeisbridge.
approximately equal in will
This situation the static
decrease the
cost of response. However,
the monitoring to show
system due clearly
to thetheuserelationship between dynamic
of one monitoring and static
point only. The response of the
high correlation (0.99)
bridge, the dynamic increment factor should be calculated and analyzed. The Savitzky-Golay finite
of strain response for the passage and opposite ways occurred at 400 and 406 km/h. It means that
impulse response(FIR) smoothing filter is applied to detect the static strain of the bridge. The first
the strain response of the two speeds is approximately equal in the static response. However, to show
polynomial order with 101 frame size is utilized in this study. Figure 10a shows the strain
clearlymeasurements
the relationship between
and filter data of dynamic and static
the mid span response
of passage directionof with
the bridge, the dynamic
a train speed of 406 km/h.increment
factor should be the
Therefore, calculated
dynamic and analyzed.
increment factor The
(DF) Savitzky-Golay
can be calculated as finite impulse
follows [22,23]:response(FIR) smoothing
filter is applied to detect the static strain of the bridge. The first polynomial order with 101 frame size
is utilized in this study. Figure 10a shows the strain measurements and filter data of the mid span of
Appl. Sci. 2016, 6, 24 8 of 13

passage direction with a train speed of 406 km/h. Therefore, the dynamic increment factor (DF) can be
Appl. Sci.
calculated as2016, 6, 24 [22,23]:
follows 8 of 13
Adyn
ˆ ˙
DF “ 1 ` A (2)
DF = 1 + Astc (2)
A
where Appl.
AdynSci.and
2016,A 24 are the maximum absolute of dynamic and static amplitude of the strain,
6,stc 8 of 13as shown
where A and A are the maximum absolute of dynamic and static amplitude of the strain, as
in Figure 10a. The dynamic factors of passage (P)
shown in Figure 10a. The dynamic factors of passage
and opposite (O) directions for the monitored
(P) and opposite (O) directions for the
points
A
are illustrated
monitoredinpoints
Figure illustrated in Figure DF
are10b. 10b.= 1 + (2)
A
where A15 and A are the maximum absolute of dynamic and5 static amplitude of the strain, as
shown in10
Figure 10a. The dynamic factors of passage (P) and
0
opposite (O) directions for the
monitored5
points are illustrated in Figure 10b.
0 -5

15 5
strain (μs)

-5

strain (μs)
-10
10
-10
0
5
-15 -15 P1
P1
0 -5 P2
-20 P2
P3
P3 -20 P4
strain (μs)

-5

strain (μs)
-25 P4
-10 P5
P5
-10
-30 -25
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
-15 -15 P1
Time (Sec.) Time (Sec.)
P1
P2
P2
-20
(a) P3 -20
(b) P3
P4
-25 P4
P5
10 P5 15
-30 -25
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
5 Time (Sec.) 10 Time (Sec.)

(a) 5 (b)
0

10 15 0
-5
Strain (μs)

Strain (μs)

5 10
-5
-10
5
0 -10

-15 0
-5 290 -15 290
Strain (μs)

Strain (μs)

360 360
-5
-20 400 -20 400
-10 406
406
-10
-25 -25
-150 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
Time (Sec.) 290 -15 Time (Sec.) 290
360 360
-20 400 400
(c) 406
-20
(d) 406

-25 -25
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
Figure 9. Strain measurements of the main girder induced by high-speed train. (a) passing response
Time (Sec.) Time (Sec.)

Figure 9. Strain measurements of the main girder induced by high-speed train. (a) passing response
(c) passing response points for 406 km/h; (c) passage(d)
points for 290 km/h; (b) response mid span points;
points(d)
foropposite
290 km/h; (b) passing
response
response points for 406 km/h; (c) passage response mid span points;
mid spanofpoints.
Figure 9. Strain measurements the main girder induced by high-speed train. (a) passing response
(d) opposite response mid span points.
points for 290 km/h; (b) passing response points for 406 km/h; (c) passage response mid span points;
The dynamic
(d) opposite factormid
response calculation shows that the DF of 290 km/h train is higher than other train
span points.
speeds at the passage and opposite
The dynamic factor calculation shows that directions. Furthermore, the DF
the DF of 290 of the
km/h opposite
train direction
is higher thanis other
highertrain
than Thepassage
the dynamicdirection
factor calculation
with all shows
train that the
speeds DF of
except 290 (P3)
point km/hwith
trainspeeds
is higher
of thanand
400 other
406train
km/h.
speeds at the passage and opposite directions. Furthermore, the DF of the opposite direction is higher
speeds at the passage and opposite directions. Furthermore, the DF of the opposite direction is higher
than the passage direction with all train speeds except point (P3) with speeds of 400 and 406 km/h.
than the passage direction with all train speeds except point (P3) with speeds of 400 and 406 km/h.

(a) (b)
(a) strain and dynamic factor of the strain. (a) measured
Figure 10. Measured static (b) and filtered static
strain; (b) dynamic factor of the monitoring points.
Figure 10. Measured static strain and dynamic factor of the strain. (a) measured and filtered static
Figure 10. Measured static strain and dynamic factor of the strain. (a) measured and filtered static
strain; (b) dynamic factor of the monitoring points.
strain;From
(b) dynamic factor
Figure 10b, of the
it can bemonitoring points.
seen that the DFs for the development speeds are less than two. It
meansFrom
that Figure 10b, it can
the dominant be seen that
performance ofthe
theDFs for is
bridge thestatic
development speeds
with speeds are 400
of 360, less and
than406
two.km/h
It at
means
all that thepoints,
monitoring dominant performance
while the dynamicof the bridge is static
performance with speeds
occurred at the of 360, 400monitoring
opposite and 406 km/h at
direction
From Figure 10b, it can be seen that the DFs for the development speeds are less than two. It means
all monitoring
with train speedpoints, whileand
290 km/h the dynamic
point (O1)performance
with train occurred
speed 400 at km/h.
the opposite
Thesemonitoring direction
results indicate that the
that the dominant performance of the bridge is static with speeds of 360, 400 and 406 km/h at all
with train speed 290 km/h and point (O1) with train speed 400 km/h. These results indicate that the
Appl. Sci. 2016, 6, 24 9 of 13

monitoring points, while the dynamic performance occurred at the opposite monitoring direction with
train speed 290 km/h and point (O1) with train speed 400 km/h. These results indicate that the static
behavior increases with increased train speeds. Therefore, the fatigue and frequency behavior should
be studiedAppl.to Sci.
investigate
2016, 6, 24 the safety of the bridge under high train speed effect. 9 of 13
Herein, the cross-correlation evaluation is used to predict the dynamic behavior of strain
contents. static
The behavior increases with increased train speeds. Therefore, the fatigue and frequency behavior
same conditions used in the acceleration analysis are used in this part. Figure 11
should be studied to investigate the safety of the bridge under high train speed effect.
presentsAppl.
theSci.
cross-correlation
Herein, and the cubic fitting of the maximum dynamic of strain strain
2016, 6, 24the cross-correlation evaluation is used to predict the dynamic behavior of9 of
measurements.
13
The relationship
contents. between the maximum
The same conditions used in dynamic
the accelerationof strain contents
analysis are used in in the passage
this part. Figureand 11 opposite
directionsstatic
forbehavior
presents
thethetrainincreases
speed with
of increased
cross-correlation and km/h
360 trainisfitting
the cubic speeds. ofTherefore,
the maximum
illustrated the fatigue
in Figure dynamic
11a.and frequency
ofWhile behavior
strain measurements.
Figure 11b shows the
shouldrelationship
be studied to investigate the the safety ofdynamic
the bridge ofunder
strainhigh train in
speed
the effect.
predictionThe of the oppositebetween direction maximum
contents of the dynamic contents
strain for 406 passage and opposite
km/h, the statistical analysis
Herein, for
directions thethecross-correlation
train speed of 360 evaluation
km/h is isillustrated
used to in predict
Figurethe
11a.dynamic behavior
While Figure 11b of strainthe
shows
of cubic contents.
and quadratic
predictionTheofsamethefitting
oppositeshows
conditions usedthat
direction theacceleration
in contents
the correlation coefficient
analysis
of the dynamic are used
strain ofin
for quadratic
406 this isstatistical
part.the
km/h, 0.90,11so the cubic
Figure
fitting inpresents
this case
analysisthe is better
cubic and to
ofcross-correlation predict
and the
quadratic the
cubic
fitting dynamic
fitting
shows behavior
of the
that the maximum ofdynamic
correlation the strain of contents.
of strain
coefficient is The
measurements.
quadratic 0.90, comparison
so
betweenThe therelationship
the cubic fitting
results between
of acceleration the maximum
in this case isand
better todynamic
strain predict
dynamicof strain
the dynamiccontents
contents in the
behavior
shows passage
ofthat the and
the strain opposite
contents.
dynamic The
evaluation of
directions
comparison for the train the speed of 360 km/h is illustrated in Figure 11a. While Figure 11b the
shows the
acceleration is betterbetween
and effective results
to of acceleration
assess and strain
the vibration dynamic
of contents
the bridge, shows that
but the strain dynamic
dynamic contents
prediction
evaluation ofofthe opposite direction
acceleration is better and contents of the
effective dynamic
to assess strain for of406
the vibration thekm/h,
bridge,thebut
statistical
the strain
can be used to
analysis
dynamic decrease
of contents
cubic andcanthe monitoring
quadratic
be used fitting cost.
shows
to decrease thethat the correlation
monitoring cost. coefficient of quadratic is 0.90, so
the cubic fitting in this case is better to predict the dynamic behavior of the strain contents. The
7 18
comparison ybetween the results of acceleration and strain dynamic
= 0.1*x - 0.87*x + 2.9*x - 0.44 R=0.94 contents shows that the dynamic
3 2
16
6
evaluation of acceleration is better and effective to assess the 14
vibration of the bridge, but the strain
5
dynamic contents can be used to decrease the monitoring cost. 12
Max Str. predicted (μs)
Max. of Str. Opp (μs)

4
10
7 18
3 8R=0.94
y = 0.1*x 3 - 0.87*x 2 + 2.9*x - 0.44 16
6
6
2
14
5 4
1 12
Max Str. predicted (μs)
Max. of Str. Opp (μs)

2
4
10
0
0
3 8
-1 -2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 6 0 1 2 3 4 5 6 7 8
2 Max. of Str. Pass (μs) Max Str. measured (μs)
4
1 (a) (b)
2

0
Figure 11. Cross-correlation and prediction of maximum strain dynamic. (a) cross-correlation 0

-1 Cross-correlation and prediction of maximum

Figure 11.maximum -2 strain dynamic. (a) 7 cross-correlation
0 0.5 1 dynamic
1.5 2 of
2.5 strain;
3 3.5(b) 4predicting
4.5 5 effect of cross-correlation.
0 1 2 3 4 5 6 8
of Str. Pass (μs) Max Str. measured (μs)
maximum dynamic of Max. strain; (b) predicting effect of cross-correlation.
To perform fatigue (a) evaluation, a simplified rain-flow cycle counting(b) algorithm was used first to
process
Figurestrain history data andand
11. Cross-correlation the prediction
spectrum of of stress
maximum matrix obtained
strain dynamic.by statistical analysis [17,24].
(a) cross-correlation
To perform
Themaximum
fatigue
spectra dynamic
evaluation,
of stressofmatrix
a simplified
calculated using
rain-flow cycle counting algorithm was used first to
strain; (b) predicting effectthe strain history data with train speeds of 360 and
of cross-correlation.
process strain history
406 km/h for thedata and(left)
passage theand spectrum
opposite of stress
(right) matrix
as shown in obtained by statistical
Figure 12, respectively.In analysis [17,24].
addition,
The spectra theTo
of perform
stress fatigue
maximum matrix evaluation,
calculated
stress cycles a simplified
usinginthe
are presented rain-flow
strain
Figure cycle
12. iscounting
Ithistory data
observed algorithmthewas
with
that used
train
maximum firststress
speeds to of 360 and
process strainobtained
amplitude, history data
from and thehistory
strain spectrum of stress
curves under matrix
two obtained
trains’ by for
effects statistical
the analysis
passage and [17,24].
opposite
406 km/h for the passage (left) and opposite (right) as shown in Figure 13, respectively. In addition,
The spectra of
directions, stress matrix
is smaller than 2.5calculated using the
MPa. Therefore, onlystrain history
a small number dataofwith
stresstrain speeds
cycles occurof
at 360 and
the higher
the maximum
406 km/h
stress
stress
for the
range.
cycles
passage
Most
are presented
cycles(left)
occurand in Figure
in opposite
the region(right)
13. Itmean
is observed
as shown
of stress in Figure
and
that
amplitude
the maximum
12, respectively.In
from −0.5 to 0.5
stress amplitude,
addition,
MPa and
obtainedthe
from
0 to strain
maximum
0.5 history
MPa, respectively. curves
stress cycles Thus, under
are presented
the meantwoin
andtrains’
Figure effects
12.
amplitude is for
Itvalues ofthe
observedthe passage
that for
stress and
the the opposite
maximum
two stress
trains directions, is
in the
amplitude,
smaller thantwo2.5 obtained
MPa.
directions from strainonly
Therefore,
are equal. history curves number
a small under twoof trains’
stress effects for the
cycles passage
occur and higher
at the oppositestress range.
directions, is smaller than 2.5 MPa. Therefore, only a small number of stress cycles occur at the higher
Most cycles occur in the region of stress mean and amplitude from ´0.5 to 0.5 MPa and 0 to 0.5 MPa,
stress range. Most cycles occur in the region of stress mean and amplitude from −0.5 to 0.5 MPa and
respectively. Thus,
0 to 0.5 the mean and
MPa, respectively. Thus,amplitude
the mean and values of the
amplitude stress
values forstress
of the the twofor thetrains in the
two trains two directions
in the
are equal.
two directions
150 are equal.
Number of cycles

100
X: 1.24e-005
Y: -0.06603
Z: 164
50

150 0
Number of cycles

-2.5
100 -2 2
X: 1.24e-005
-1.5 Y: -0.06603 1.5
-1 Z: 164 1
50
-0.5 0.5
Mean (MPa) 0
Amplitude (MPa)
0

-2.5 (a)
-2 2
-1.5 1.5
-1 1
-0.5 0.5
Mean (MPa) 0
Amplitude (MPa)

(a)

Figure 12. Cont.

Appl. Sci. 2016, 6, 24 10 of 13
Appl. Sci. 2016, 6, 24 10 of 13

100
100

Number of cycles
Number of cycles
X: 3.828e-006
X: 4.496e-005
Y: -0.02062
Y: 0.4223
50 Z: 126.5
50 Z: 132

0 0

-2
-3
2.5 -1.5 1.2
-2
2 -1 1
-1 1.5 0.8
1 -0.5 0.6
0.4
0 0.5 0 0.2
Mean (MPa) Mean (MPa)
Amplitude (MPa) Amplitude (MPa)

(b)
Figure 12. Rain-flow matrix of mid-span stress (a) 360 km/h; (b) 406 km/h.
Figure 13. Rain-flow matrix of mid-span stress (a) 360 km/h; (b) 406 km/h.
The maximum number of stress cycles at 406 km/h is smaller than that occurring at 360 km/h in
the two directions. The results show that the fatigue stress and number of cycles limit are 29 MPa and
The maximum number of stress cycles at 406 km/h is smaller than that occurring at 360 km/h
2 × 106, respectively [25], as recommended by Eurocode 3. The value of the equivalent stress
in the two amplitude
directions. andThe resultsofshow
the number cycles that
whenthe fatigue stress
the high-speed and number
train passes of cycles
through bridge limit
is far less are 29 MPa
than
6 , respectively
and 2 ˆ 10this [25],However,
value for two trains. as recommended by Eurocode
the fatigue behavior 3. The
of the bridge deckvalue
satisfiesof the
the equivalent stress
requirement
amplitudeofand the the
infinite-fatigue-life
number of cycles design when
method.the high-speed train passes through bridge is far less than

this value for

3.3. two trains. However,
Acceleration-Strain the
Frequency fatigue
Domain behavior of the bridge deck satisfies the requirement of
Evaluation
the infinite-fatigue-life design method.
The frequency contents of strain and acceleration measurements for the mid-span monitoring
points in the passage and opposite directions are illustrated in Figure 13. The cross spectrum density
3.3. Acceleration-Strain
function in Matlab Frequency
(VersionDomain Evaluation
7.6, MathWorks, Natics, MA, USA, 2008) is used to calculate the
frequency contents.Based on the FEM [5] analysis, the band-pass filters in between 1 to 45 Hz with
The frequency contents of strain and acceleration measurements for the mid-span monitoring
101 hamming window are used to filter the measured data to include the static and dynamic
points in the passage
frequency and opposite
contents directions
of the bridge. From Figure are13,illustrated
the frequencyin Figureare14.
contents The3.906
3.223, cross
Hz spectrum
and 3.223, density
function in4.199
Matlab (Version
Hz for 290 and 7.6, MathWorks,
360 km/h Natics,
at the opposite and MA, USA,
passage 2008) respectively.
directions, is used to calculate
In addition,thethefrequency
frequency
contents.Based on thecontents
FEM equal (4.297 Hz)
[5] analysis, theforband-pass
the 400 andfilters
406 km/h at the two
in between 1 todirections.
45 Hz with From101the hamming
comparison of the FEM frequency and real data, it is observed that the first dynamic mode changes
window are used to filter the measured data to include the static and dynamic frequency contents of
increased with increasing the trains’ speeds. The changes of passage frequency from the first bending
the bridge.FEMFrom Figuremode
frequency 14, the frequency
are 18.5%, 24.2%,contents
25.8% for the arespeeds
3.223,290,
3.906 Hz406
360 and and 3.223,
km/h, 4.199 HzThe
respectively. for 290 and
360 km/h strain
at thefrequencies
oppositecontents
and passage
at the twodirections,
directions with respectively.
the effect of Inall addition,
trains’ speeds thearefrequency
similar. In contents
equal (4.297addition,
Hz) for thethestatic frequency
400 and 406 contents
km/hareatclearly
the two shown with strain From
directions. measurements only. The low
the comparison of the FEM
frequencies are 0.781, 0.977, 1.172, 1.172 Hz and 0.879, 1.074, 1.172, 1.172 Hz of the opposite and
frequency and real data, it is observed that the first dynamic mode changes increased with increasing
passage directions for the 290, 360, 400, 406 km/h train speeds, respectively. It means that the strain
the trains’ measurements
speeds. The changesare enough of to
passage
estimatefrequency
the static and from the first
dynamic bending
behavior FEM frequency
in frequency domain. mode are
18.5%, 24.2%, 25.8%from
Moreover, for the speeds 290,
the comparison 360 and
between the 406
first km/h, respectively.
mode contents The strain and
of the measurements frequencies
the FEM contents
at the two calculations,
directions itwithcan be theconcluded
effect ofthat
allthe bridgespeeds
trains’ is safe under its currentIn
are similar. dynamic behavior
addition, the with thefrequency
static
development speeds of trains.
contents are clearly shown with strain measurements only. The low frequencies are 0.781, 0.977, 1.172,
The Matlab Spectrogram toolbox is used to extract the three dimensional time-frequency maps
1.172 Hz and 0.879,
for the 1.074,
passage 1.172,
train at the1.172
mid-spanHz point
of the of opposite
the bridge and passage
at speeds 290 and directions
406 km/h, for the 290,
as shown in 360, 400,
406 km/h train
Figurespeeds, respectively.
14. The results show thatIt means
the power that the strain
spectrum densitymeasurements
(PSD) at 290 km/hare enough
speed is lowertothan
estimate the
the PSD amplitude
static and dynamic behavior at 406
in km/h. The PSD
frequency frequencyMoreover,
domain. responses’ amplitude
from thedifferences
comparison between trains the first
between
passage and departure (load and unload) show small values at 406 km/h. Therefore, it is concluded
mode contents of the measurements and the FEM calculations, it can be concluded that the bridge is
that the dynamic behavior of the bridge at train speeds of 406 km/h is greater than the 290 km/h.
safe underHowever,
its currentit is dynamic
concluded thatbehavior with
the bridge theatdevelopment
is safe speeds
a speed of 406 km/h, but itofshould
trains.
be continuously
The Matlab
monitored Spectrogram
if trains speeds toolbox is used
are increased aboveto this
extract
value.the three dimensional
Moreover, the increase of PSD time-frequency
with train maps
speeds train
for the passage indicates that mid-span
at the the simple beam
pointgirders
of theof bridge
steel bridges are very290
at speeds sensitive
and 406to train
km/h,induced
as shown in
vibrations, and, therefore, may be not suitable for an increase in the speed of train traffic.
Figure 15. The results show that the power spectrum density (PSD) at 290 km/h speed is lower than
the PSD amplitude at 406 km/h. The PSD frequency responses’ amplitude differences between trains
passage and departure (load and unload) show small values at 406 km/h. Therefore, it is concluded
that the dynamic behavior of the bridge at train speeds of 406 km/h is greater than the 290 km/h.
However, it is concluded that the bridge is safe at a speed of 406 km/h, but it should be continuously
monitored if trains speeds are increased above this value. Moreover, the increase of PSD with train
speeds indicates that the simple beam girders of steel bridges are very sensitive to train induced
vibrations, and, therefore, may be not suitable for an increase in the speed of train traffic.
Appl. Sci. 2016, 6, 24 11 of 13
Appl. Sci. 2016, 6, 24 11 of 13
Appl. Sci. 2016, 6, 24 11 of 13
2 2
10 10
290 290
12 360 2 360
10
10 10
400
290 0
400
290
10 406
01 406
360 360
10
10
400 0
400
10 406
0 406
10-1
10 -2
10

Magnitude
Magnitude

-1
10-2
10 -2
10

Magnitude
Magnitude

-4
-2
-3 10
10
10
-4
-3 10
10-4
10
-6
10
-4
10-5
10 -6
10
-5
-6 -8
10
10 10
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
-6 Frequency (HZ) -8
Frequency (HZ)
10 10
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
(a)
Frequency (HZ)
(b)
Frequency (HZ)
4 4
10 10
(a) 290 (b) 290
4
360 34 360
10 10
10
2 400
290 400
290
10 406
406
360 3 360
102
10
2 400 400
10 406
406 2
0 101
10
10
Magnitude

Magnitude
1
0 100
10
10
Magnitude

Magnitude
-2
10 0
-1
10
10
-2
10 -1
10-2
10
-4
10
-2
-3
10
10
-4
10
-3
10
-6
10-4
10
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
-6
Frequency (HZ) -4
Frequency (HZ)
10 10
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
(c)
Frequency (HZ)
(d)
Frequency (HZ)

(c) (d)
Figure 13. Acceleration and strain frequency contents. (a) passage-acceleration; (b) opposite-
Figure 14. Acceleration and strain frequency contents. (a) passage-acceleration; (b) opposite-acceleration;
acceleration; (c) passage-strain;
Figure 13. Acceleration (d) opposite-strain.
and strain frequency contents. (a) passage-acceleration; (b) opposite-
(c) passage-strain; (d) opposite-strain.
acceleration; (c) passage-strain; (d) opposite-strain.

(a) (b)
(a) (b)
Figure 14. Time-frequency acceleration measurements for the trains speeds (a) 290 km/h and (b) 406 km/h.
Figure 14. Time-frequency acceleration measurements for the trains speeds (a) 290 km/h and (b) 406 km/h.
Figure
4. Summary15. Time-frequency
and Conclusionsacceleration measurements for the trains speeds (a) 290 km/h and
(b) 406 km/h.and Conclusions
4. Summary
This paper aims to evaluate the measurements of a structural health monitoring system of the
KayaThisrailway
paperbridge
aims into Korea
evaluatewiththehigh-speed
measurements trains.
of A non-destructive
a structural health monitoring system of using
the
4. Summary and Conclusions
accelerometers
Kaya railway bridgeand strain sensors
in Korea with is high-speed
designed totrains.
monitor A the performance monitoring
non-destructive of the bridge underusing
system new
development
accelerometers
This paper aimstrain
andtospeeds
strain of 400
sensors
evaluate theand 406 km/h.
ismeasurements
designed to The static
monitor
of and
the dynamic
performance
a structural behavior
health of the under
ofmonitoring
the bridge bridge areof the
new
system
analyzed
development and discussed.
train speeds Accordingly,
of 400 and the
406 following
km/h. The remarks
static and
and
Kaya railway bridge in Korea with high-speed trains. A non-destructive monitoring system conclusions
dynamic are
behavior drawn:
of the bridge areusing
- The
analyzed mathematical
and discussed. correlation
Accordingly,models the describing
following the overall
remarks andstructural behavior
conclusions
accelerometers and strain sensors is designed to monitor the performance of the bridge under new are of the
drawn: bridge can
- be
Theobtained with the
mathematical support of
correlation the health
models monitoring
describing system.
the overall structural behavior of the bridge can
development train speeds of 400 and 406 km/h. The static and dynamic behavior of the bridge are
- The torsionalwith
be obtained response shows of
the support a higher effect
the health than the vibration
monitoring system. response on the bridge deck.
analyzed and discussed. Accordingly, the following remarks and conclusions are drawn:
- The effective
torsionaltorsion
responsetest should
shows be assessed
a higher effectat thethe
than endvibration
points ofresponse
the bridge.
on the bridge deck.
- - The mean value
effective control
torsion testchart for
should the
be acceleration
assessed at and
the endstrain can
points
The mathematical correlation models describing the overall structural behavior be
of applied
the bridge.for bridgeofmonitoring
the bridge can
- and for thevalue
The mean earlycontrol
warning of any
chart abnormal
for the behavior.
acceleration and strain can be applied for bridge monitoring
be obtained with the support of the health monitoring system.
and for the early warning of any abnormal behavior.
- The torsional response shows a higher effect than the vibration response on the bridge deck.
- The effective torsion test should be assessed at the end points of the bridge.
Appl. Sci. 2016, 6, 24 12 of 13

- The mean value control chart for the acceleration and strain can be applied for bridge monitoring
and for the early warning of any abnormal behavior.
- The dynamic factor calculation shows that the static behavior increases with train
speed developments.
- The statistical analysis of cubic and quadratic fits shows that the cubic fitting in monitoring strain
is better to predict the dynamic behavior of the strain contents.
- The comparison between the results of acceleration and strain dynamic contents shows that the
dynamic evaluation of acceleration is better and effective to assess the vibration of the bridge, but
the strain dynamic contents can be used to decrease the monitoring cost.
- The fatigue performance of the bridge deck satisfies the requirement of infinite-fatigue-life design
method, and the highest cycles occur in a close region of stress mean and amplitude. Therefore,
the bridge-deck fatigue is safe under current trains’ speeds.
- The frequency calculation of the acceleration and strain measurements shows that the strain
measurements are enough to estimate the static and dynamic behaviorin frequency domain.
- Comparing the first mode contents of the measurements and the FEM calculations shows that the
dynamic behavior of the bridge is safe with development speeds of trains.
- The increase of PSD with train speeds indicates that the simple beam girders of steel bridges are
very sensitive to train induced vibrations, and, therefore, may be not suitable for increased speed
of train traffic.
- Based on the vibration, torsion, fatigue and frequency contents of the bridge, it is concluded
that the bridge is safe under the development speed with a recommendation not to increase
the train speed because the torsion performance is critical at 406 km/h at the entrance and exit
monitoring points.

Acknowledgments: This research was supported by the Basic Science Research Program through the National
Research Foundation of Korea (NRF) funded by the Ministry of Science, information and communication
technology (ICT), and Future Planning (Grant No. 2013R1A2A2A01068174).
Author Contributions: The authors have contributed equally to this work.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Lee, J.; Kim, S.; Kwark, J.; Lee, P.; Yoon, T. Dynamic characteristics of high-speed railway steel bridges.
Trans. Korean Soc. Noise Vib. Eng. 2007, 17, 632–637. (In Korean).
2. Xia, H.; de Roeck, G.; Zhang, N.; Maeck, J. Experimental analysis of a high-speed railway bridge under
Thalys trains. J. Sound Vib. 2003, 268, 103–113. [CrossRef]
3. Ding, Y.; Sun, P.; Wang, G.; Song, Y.; Wu, L.; Yue, Q.; Li, A. Early-Warning Method of Train Running Safety
of a High-Speed Railway Bridge Based on Transverse Vibration Monitoring. Shock Vib. 2015, 2015, 518689.
[CrossRef]
4. Al Shaer, A.; Duhamel, D.; Sab, K.; Foret, G.; Schmitt, L. Experimental settlement and dynamic behavior of
a portion of ballasted railway track under high speed trains. J. Sound Vib. 2008, 316, 211–233. [CrossRef]
5. Kim, E.; Park, J.; Sim, S. Dynamic Behavior of Composite Steel Girder Bridge Exceeding Train. J. Korea Acad.
Ind.Cooperation Soc. 2013. (in Korean). [CrossRef]
6. Liu, K.; Reynders, E.; de Roeck, G.; Lombaert, G. Experimental and numerical analysis of a composite bridge
for high-speed trains. J. Sound Vib. 2009, 320, 201–220. [CrossRef]
7. Majka, M.; Hartnett, M.; Bień, J.; Zwolski, J. Experimental and numerical dynamic analysis and assessment
of a railway bridge subjected to moving trains. In Proceedings of the 3rd International Conference on Bridge
Maintenance, Safety and Management-Bridge Maintenance, Safety, Management, Life-Cycle Performance
and Cost, Porto, Portugal, 16–19 July 2006; pp. 713–714.
8. Koh, H.M.; Kim, S.; Choo, J.F. Recent Development of Bridge Health Monitoring System in Korea. In Sensing
Issues in Civil Structural Health Monitoring; Springer: Heidelberg, Germany, 2005; pp. 33–42.
 Appl. Sci. 2016, 6, 24 13 of 13

9. Koh, H.-M.; Lee, H.-S.; Kim, S.; Choo, J.F. Monitoring of Bridges in Korea. In Encyclopedia of Structural Health
Monitoring; John Wiley & Sons: Chichester, England, 2009.
10. Arangio, S.; Bontempi, F. Structural health monitoring of a cable-stayed bridge with Bayesian neural
networks. Struct. Infrastruct. Eng. 2015, 11, 575–587. [CrossRef]
11. Zhang, Q. Statistical damage identification for bridges using ambient vibration data. Comput. Struct. 2007,
85, 476–485. [CrossRef]
12. Zhan, J.; Xia, H.; Chen, S.; de Roeck, G. Structural damage identification for railway bridges based on
train-induced bridge responses and sensitivity analysis. J. Sound Vib. 2011, 330, 757–770. [CrossRef]
13. Yam, L.; Yan, Y.; Jiang, J. Vibration-based damage detection for composite structures using wavelet transform
and neural network identification. Compos. Struct. 2003, 60, 403–412. [CrossRef]
14. Sartor, R.; Culmo, M.; DeWolf, J. Short-Term Strain Monitoring of Bridge Structures. J. Bridg. Eng. 1999, 4,
157–164. [CrossRef]
15. Xia, H.; Zhang, N.; de Roeck, G. Dynamic analysis of high speed railway bridge under articulated trains.
Comput. Struct. 2003, 81, 2467–2478. [CrossRef]
16. Li, Z.; Chan, T.; Zheng, R. Statistical analysis of online strain response and its application in fatigue assessment
of a long-span steel bridge. Eng. Struct. 2003, 25, 1731–1741. [CrossRef]
17. Ding, Y.; Wang, G.; Sun, P.; Wu, L.; Yue, Q. Long-term structural health monitoring system for a high-speed
railway bridge structure. Sci. World J. 2015, 2015, 250562. [CrossRef] [PubMed]
18. Kaloop, M. Bridge safety monitoring based-GPS technique: case study Zhujiang Huangpu Bridge.
Smart Struct. Syst. 2012, 9, 473–487. [CrossRef]
19. Kaloop, M.; Hu, J. Stayed-Cable Bridge Damage Detection and Localization Based on Accelerometer Health
Monitoring Measurements. Shock Vib. 2015, 2015, 102680. [CrossRef]
20. Elnabwy, M.; Kaloop, M.; Elbeltagi, E. Talkha steel highway bridge monitoring and movement identification
using RTK-GPS technique. Meas. J. Int. Meas. Confed. 2013, 46, 4282–4292. [CrossRef]
21. Lorieux, L. Analysis of Train-Induced Vibrations on a Single-Span Composite Bridge. Master Thesis,
The Royal Institute of Technology (KTH), Stockholm, Sweden, 2008.
22. Gao, Q.; Wang, Z.; Jia, H.; Liu, C.; Li, J.; Guo, B.; Zhong, J. Dynamic Responses of Continuous Girder Bridges
with Uniform Cross-Section under Moving Vehicular Loads. Math. Probl. Eng. 2015, 2015, 951502. [CrossRef]
23. Maljaars, J.; Waarts, P.; Leenderts, J.; Hoogvelt, R. Dynamic increment factor in modular expansion joints of
bridges under heavy traffic loading. In Proceedings of the 7th International Symposium on Heavy Vehicles
Weights and Dimensions, Delfet, The Netherland, 16–20 June 2002; pp. 289–302.
24. Li, S.; Li, H.; Liu, Y.; Lan, C.; Zhou, W.; Ou, J. SMC structural health monitoring benchmark problem using
monitored data from an actual cable-stayed bridge. Struct. Control Health Monit. 2014, 21, 156–172. [CrossRef]
25. European Committee for Standardization. Eurocode 3: Design of Steel Structures, Part 1–9: Fatigue; BS EN
1993-1-9:2005; European Committee for Standardization: Brussels, Belgium, 2005.

© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons by Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).

View publication stats