Damage Detection of Tension Pendulums in Cable-Stayed

Bridges Using Structural Frequency Variance

Guang-Ming Wu 1; Ting-Hua Yi, A.M.ASCE 2; Dong-Hui Yang 3; and Hong-Nan Li, F.ASCE 4

Abstract: Structural health monitoring represents an efficient way to diagnose the condition of cable-stayed bridges. Most studies assessing
Downloaded from ascelibrary.org by UNB - Universidade de Bras?lia on 09/16/22. Copyright ASCE. For personal use only; all rights reserved.

the condition of cable-stayed bridges and detecting damage to them have been developed to monitor the longitudinal performance of bridge
bearings. However, few studies have aimed to detect the damage to tension pendulums, which are usually tie-down cables that resist girder
uplift from auxiliary piers. This study aims to develop an effective method for detecting the damage of tension pendulums in cable-stayed
bridges using structural modal frequencies. First, the sensitivity of the bridge frequency to tension pendulum damage is derived according to
the Rayleigh method. Then, a finite-element simulation of a cable-stayed bridge is conducted to validate the rationality of the frequency
index. Finally, a case study on tension pendulum damage is carried out for the cable-stayed Yonghe Bridge. The results indicate that the
frequency change is far more sensitive to the damage of the tension pendulums than to the damage of the stay cables. The failure of tie-down
cables will result in a substantial decrease in the bridge frequency, up to approximately 35% for the first vertical bending mode of the bridge.
In practice, the damage of tension pendulums can be effectively detected by a sudden decrease in the measured frequency of cable-stayed
bridges. DOI: 10.1061/(ASCE)CF.1943-5509.0001540. © 2020 American Society of Civil Engineers.
Author keywords: Bridge health monitoring; Damage detection; Uplift-restraint bearing; Auxiliary pier; Modal frequency.

Introduction To date, some studies have been performed on the damage de-
tection of bridge bearings (Ala et al. 2016; Fayyadh and Abdul
In recent decades, cable-stayed bridges have been constructed Razak 2012; Hoult et al. 2010). Webb et al. (2014) detected damage
worldwide because of their advantages, such as economical spans in degraded roller bearings using the relationship between the
and esthetics. Cable-stayed bridges represent critical connections in friction in bearings and the longitudinal expansion of girders.
the transportation networks of many countries; consequently, the Kromanis et al. (2016) demonstrated that transverse temperature
temporary closure or functional failure of such bridges can lead gradients induced the planar bending of girders, producing twists
to a disruption of transportation. The management and maintenance at roller bearings; the roller bearings ultimately suffered fatigue
of bridges have become considerable challenges for bridge owners failure owing to diurnal variations under ambient conditions.
and operators (Xu et al. 2019). For example, serious accidents like Brownjohn et al. (2015), while monitoring bearing displacement,
the collapse of a cable-stayed bridge may occur during its operation discovered that strong winds could double the displacement of
(Calvi et al. 2019; Malomo et al. 2020; Wolff and Starossek 2009), A-frame rocker bearings in a long-span bridge. Guo et al. (2016)
possibly because damage went undetected. Fortunately, the struc- found wear in sliding bearings in a bridge by analyzing the cumu-
tural health monitoring (SHM) of bridges has developed with ad- lative sliding displacement. Noade and Becker (2019) presented a
vanced sensing technology. SHM is an efficient way to monitor probabilistic framework to quantify the annual cyclic displacement
bridge performance and can provide reliable, real-time information demands on a bearing from diverse loadings using an idealized
on the structural integrity and conditions of a bridge during normal archetypal continuous concrete girder bridge. Wu et al. (2020) pro-
operation, even after extreme events. One of the applications of posed an approach to predicting the sliding life of sliding bearings
SHM is to detect damage to vulnerable bridge elements, such as using the dynamic monitoring data of bridges. However, the afore-
bridge bearings. mentioned studies focused mainly on longitudinal damage to
bridge bearings, whereas few studies have been performed to detect
1 vertical damage to uplift-restraint bearings using a SHM system.
Ph.D. Candidate, School of Civil Engineering, Dalian Univ. of Tech-
nology, No. 2, Linggong Rd., Ganjingzi District, Dalian 116024, China.
The uplift-restraint function of bearings is usually provided by
Email: [email protected] tension pendulums (as shown in Fig. 1), which are extensively used
2
Professor, School of Civil Engineering, Dalian Univ. of Technology, in asymmetrical cable-stayed bridges or cable-stayed bridges with a
No. 2, Linggong Rd., Ganjingzi District, Dalian 116024, China (corre- small ratio of side span to main span (Chen et al. 2014; Swiggum
sponding author). Email: [email protected] et al. 1994; Zhu et al. 2016). Tension pendulums, which are gen-
3
Associate Professor, School of Civil Engineering, Dalian Univ. of erally pretensioned tie-down cables, connect girders to auxiliary
Technology, No. 2, Linggong Rd., Ganjingzi District, Dalian 116024, piers and are used to resist girder uplift under normal operating
China. Email: [email protected] conditions (Svensson 2012). If a tension pendulum is damaged,
4
Professor, School of Civil Engineering, Dalian Univ. of Technology, the girder will be lifted from the auxiliary pier, and that uplift will
No. 2, Linggong Rd., Ganjingzi District, Dalian 116024, China. Email:
inevitably lead to a change in the structural characteristics of the
[email protected]
Note. This manuscript was submitted on March 7, 2020; approved on cable-stayed bridge. For example, a cable-stayed bridge with five
July 27, 2020; published online on October 24, 2020. Discussion period spans will become a structural system with three spans if the ten-
open until March 24, 2021; separate discussions must be submitted for in- sion pendulums are fully broken. Hence, the failure of these tie-
dividual papers. This paper is part of the Journal of Performance of Con- down cables could cause total bridge collapse (Lee et al. 2015). In
structed Facilities, © ASCE, ISSN 0887-3828. practice, some cases of bridge uplift in cable-stayed bridges have

© ASCE 04020126-1 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2021, 35(1): 04020126

 cable and ith tension pendulum, respectively; EAci and Lci = axial
Girder
stiffness and length of ith stay cable, respectively; aci = angle be-
tween ith stay cable and girder; n = number of stay cables; EAbi and
Sliding bearings Lbi = axial stiffness and length of ith tension pendulum, respec-
Tension pendulums
tively; and φj ðxci Þ and φj ðxbi Þ = vibration amplitudes of girder
at positions of ith stay cables and ith tension pendulum, respec-
Auxiliary pier
tively.Similarly, the maximum kinetic energy of the bridge consists
of the kinetic energies of the girder T G, the towers T T , the stay
cables T C , and the tension pendulums T B :
Fig. 1. Configuration of uplift-restraint bearings on auxiliary piers. Z
ω2j l
TG ¼ m̄ ϕ2 ðxÞdx ð5Þ
2 0 G j
Downloaded from ascelibrary.org by UNB - Universidade de Bras?lia on 09/16/22. Copyright ASCE. For personal use only; all rights reserved.

been reported (Chen et al. 2011; Huang and Lai 2016) because the Z
ω2j X
2 hi
uplift constraints of the bridges were easily broken under the effects TT ¼ m̄T ξ 2ij ðyÞdy ð6Þ
of heavy corrosion and fatigue. Therefore, it is essential to develop 2 i 0
a method to detect damage to tension pendulums.
To detect damage to tension pendulums in a cable-stayed bridge, ω2j X
n

the sensitivity of the bridge frequency to damage to tension TC ¼ ρA L ϕ2 ðx Þ ð7Þ

6 i¼1 ci ci j ci
pendulums is derived according to the Rayleigh method. Then a
numerical model of a cable-stayed bridge is developed, and damage
scenarios of tension pendulums are carried out to validate the ω2j X
4
TB ¼ ρA L ϕ2 ðx Þ ð8Þ
rationality of the frequency index. Finally, a cable-stayed bridge 6 i¼1 bi bi j bi
is used as a case study to detect damage to tension pendulums using
structural modal frequencies. where ωj ¼ jth bending frequency of bridge; m̄G and m̄T = girder
mass and tower mass, respectively, divided by their individual
lengths; ρ = material density of cables; and Aci and Abi = areas
Sensitivity of Structural Frequency to Damage of stay cables and tension pendulums, respectively. For the vibra-
to Tension Pendulums tion mode of cable-stayed bridges, the maximum potential energy
must be equal to the maximum kinetic energy, V max ¼ T max . With
In this section, the sensitivity of the structural frequency to damage the exception of extradosed cable-stayed bridges, the potential
to tension pendulums is established using the Rayleigh method. energy of a cable-stayed bridge is stored mainly in the cables, and
According to the law of conservation of energy, the energy in a the kinetic energy of the bridge is stored primarily in the girder and
freely vibrating system must remain constant if no damping forces towers. Therefore, the frequency of the cable-stayed bridge derived
act to absorb it (Clough and Penzien 1975). This energy includes from the equation of energy can be approximated by Eq. (9):
the kinetic energy of the mass and the potential energy of the struc-
Pn EAci sin2 aci ϕ2j ðxci Þ P EAbi ϕ2 ðxbi Þ
tures. The maximum potential energy of the bridge consists of the
i¼1 þ 4i¼1 Lbij
ω2j
Lci
potential energies of the girder V G, the towers V T , the stay cables ¼ Rl P2 R hi ð9Þ
2 2
V C , and the tension pendulums V B , whose expressions are de- 0 m̄G ϕj ðxÞdx þ i 0 m̄T ξ ij ðyÞdy
scribed by Eqs. (1)–(4), respectively:
Z To study the sensitivity of the frequency to the damage to the
1 l cables, a damage ratio D of the cable area is introduced into
VG ¼ E I ðxÞϕj0 02 ðxÞdx ð1Þ
2 0 G G Eq. (9). Then the frequency is expressed by Eq. (10):
Z Pn EAci ð1−Dci Þsin2 aci ϕ2j ðxci Þ P EAbi ð1−Dbi Þϕ2j ðxbi Þ
1X 2 hi þ 4i¼1
ET I T·i ðyÞξ ij0 02 ðyÞdy
i¼1
VT ¼ ð2Þ ω2j ¼ Rl
Lci
P2 R hi
Lbi
ð10Þ
2 i¼1 0 m̄ ϕ 2 ðxÞdx þ m̄ ξ 2 ðyÞdy
0 G j i 0 T ij

1X n
1X n
EAci sin2 aci ϕ2j ðxci Þ The sensitivity of the frequency to the damage to the tension
VC ¼ ΔFj·ci · ΔLj·ci ¼ ð3Þ pendulums is expressed by Eq. (11), where κ represents
2 i¼1 2 i¼1 Lci P
∫ l0 m̄G ϕ2j ðxÞdx þ 2i ∫ h0i m̄T ξ 2ij ðyÞdy. From Eq. (11), the frequency
1X 4
1X 4
EAbi ϕ2j ðxbi Þ has a negative relationship with the damage to the tension pendu-
VB ¼ ΔFj·bi · ΔLj·bi ¼ ð4Þ lums. The more serious the damage, the smaller is the frequency:
2 i¼1 2 i¼1 Lbi
∂ðω2j Þ −EAbi ϕ2j ðxbi Þ
where φj = girder vibration shape of jth bending mode of bridge; ¼ ð11Þ
∂Dbi Lbi · κ
ϕj0 0 = second derivative of φj ; ξ ij = tower flexural shape of jth
bending mode of bridge; ξ ij0 0 = second derivative of ξ ij ; EG I G ðxÞ For the case where the girder is attached to the auxiliary pier, the
and l = flexural stiffness of girder at longitudinal position of x bending vibration shape of the girder ϕj ðxÞ can be assumed to be
and the length of the girder, respectively; ET I T·i ðyÞ and hi = flexu- ϕaj ðxÞ, as shown in Fig. 2(a), where ϕaj ðxbi Þ equals zero. Conse-
ral stiffness of ith tower at vertical position of y and height of ith quently, the frequency sensitivity also equals zero, which means
tower, respectively; ΔFj·ci and ΔFj·bi = amounts of change in axial that the frequency of the cable-stayed bridge does not change when
force of ith stay cable and ith tension pendulum, respectively, in- damage occurs to a tension pendulum. In other words, the damage
duced by jth bending mode of bridge; ΔLj·ci and ΔLj·bi , induced to tension pendulums cannot be detected using the bridge fre-
by jth bending mode = amounts of change in length of ith stay quency index when the girder is attached to the auxiliary pier.

© ASCE 04020126-2 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2021, 35(1): 04020126

Downloaded from ascelibrary.org by UNB - Universidade de Bras?lia on 09/16/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Detectable case of damage to tension pendulums.

Fig. 2. Mode shape diagram of cable-stayed bridge: (a) attached; and continuous uplift of a girder is easily neglected and difficult to
(b) lifted. be detected by manual inspection on site. In addition, damage
to a tension pendulum, such as corrosion or broken wires, is gen-
erally obscured by its sheath; such damage is also undetectable by
visual inspection. In these abnormal situations, the proposed ap-
In cases where the girder is lifted from the auxiliary pier, ϕj ðxÞ proach will be a helpful tool in damage detection.
can be assumed to be ϕuj ðxÞ, as shown in Fig. 2(b). Then the sen-
sitivity of the frequency to the damage to the tension pendulums is
rewritten as shown in Eq. (12). As shown in Eq. (12), the sensitivity Simulation of Detectability of Tension Pendulum
has no clearly defined amplitude. To determine the order of mag- Damage
nitude of the sensitivity, a comparison is made between the sensi-
tivity of the frequency to the damage to the tension pendulums and Following the preceding theoretical analysis, the feasibility of the
to the damage to the stay cables. The sensitivity to the damage to proposed frequency-based damage detection method needs to be
the stay cables is expressed by Eq. (13): validated using a numerical model and case study. To make the case
study more convincing, a numerical model is developed first using
∂ðω2j Þ −EAbi ϕ2uj ðxbi Þ the bridge used for the case study, in which real tension pendulum
¼ ð12Þ damage has occurred.
∂Dbi Lbi · κ

∂ðω2j Þ −EAci sin2 aci ϕ2uj ðxci Þ Bridge Profile

¼ ð13Þ
∂Dci Lci · κ The bridge in this study is the Yonghe Bridge, one of the earliest
cable-stayed bridges in mainland China. The bridge has five spans:
In general, the length of the tension pendulum Lbi is only sev- a 260-m main span and two side spans, each with lengths of 99.85
eral times the height of the girder H G, with Lbi ≈ ð2 − 4ÞHG . and 25.15 m, in each direction. An elevation profile of the bridge is
However, the length of the longest stay cables Lci is approximately shown in Fig. 4. At the side spans, the bridge has two auxiliary
100–220 times the height of the girder H G according to the ratio piers (P1 and P4), which include uplift-restraint bearings. These
of the span to the height of the girder (Ministry of Transport of uplift-restraint bearings consist of pot bearings and tension pendu-
the People’s Republic of China 2007). Thus, Lbi is substantially lums, as shown in Fig. 5, which resist the downward deflection and
smaller than Lci . The average angle aci is assumed to be approx- uplift of the girder, respectively. The tension pendulums are
imately 45°, resulting in sin2 aci ¼ 1=2. In addition, the area of the tie-down cables consisting of 139 steel wires with a diameter of
stay cable Aci is several times the area of the tension pendulums 5 mm. The standard strength of the steel wires is 1,600 MPa, and
Abi . However, there are certain vibration shapes of a bridge where the length of the tie-down cables is 6 m. The swinging distance of
ϕuj ðxci Þ is not much smaller than ϕuj ðxbi Þ. The first several bending the tie-down cables is 25 cm. More details regarding the bridge
shapes of the bridge will generally satisfy requirements. Substitut- construction materials and properties can be found in the literature
ing these parameters into Eqs. (12) and (13), it is evident that the (Li et al. 2014). After 19 years in service, the bridge cracked at the
sensitivity of the frequency to the damage to tension pendulums is bottom of the midspan girder. The bridge was repaired during 2005
at least one order of magnitude larger than the sensitivity to the and 2007 and then reopened for traffic at the end of 2007 (Li et al.
damage to stay cables in a cable-stayed bridge. This high sensitivity 2010). However, the tension pendulums of the bridge were fully
guarantees that the damage to tension pendulums can be easily de- damaged in July 2008.
tected using changes in the frequency. During the repair of the bridge, a SHM system was designed and
A detectable case is illustrated in Fig. 3. In most cases, there is a implemented by the Center of Structural Monitoring and Control
high probability that a girder will experience uplift intermittently (SMC) at the Harbin Institute of Technology. Detailed information
or continuously from auxiliary piers under heavy traffic once about the SHM system is introduced in Li et al. (2014). In this
the tension pendulum is heavily damaged. Slight intermittent or study, only the accelerometers on the girder are given, as shown

© ASCE 04020126-3 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2021, 35(1): 04020126

 25.15 99.85 260.00 99.85 25.15

Acceleration sensor (14)

South Unit: m North

P1 (2) P2 (2) (2) (2) (2) (2) P3 (2) P4

56.00 58.85 59.00 56.00 56.00 59.00 58.85 56.00

Fig. 4. Bridge elevation profile and layout of accelerometers.

Downloaded from ascelibrary.org by UNB - Universidade de Bras?lia on 09/16/22. Copyright ASCE. For personal use only; all rights reserved.

of the towers and girder are 3.55 × 1010 and 3.61 × 1010 N=m2 ,
Girder
respectively, and the material densities of the girder and towers
are 2,545.5 and 2,550 kg=m3 , respectively. The concrete transverse
6000

Pot bearing beams are individually simplified as the mass element MASS21
every 2.9 m along the bridge. The mass of a single concrete trans-
Tie-down cable
(139 5) verse beam is 6,932.8 kg. The cables are modeled using LINK10
nonlinear elements. The updated elastic modulus and material den-
Auxiliary pier
sity of the cables are 2.0 × 1011 N=m2 and 7,850 kg=m3 , respec-
tively. The main girder floats on the main tower, and all of the
Fig. 5. Layout of tension pendulums (unit: millimeters). towers are fixed to the ground. The longitudinal restrictions at
the abutments and auxiliary piers are simulated using COMBIN14
linear elastic spring elements. To simulate the tie-down cables, a
in Fig. 4. A total of 14 accelerometers are installed longitudinally at total of four LINK10 elements are added to connect the main girder
seven cross sections of the girder. At each cross section, two of the and the auxiliary piers. The updated elastic modulus of the tie-down
accelerometers are fixed on the downstream and upstream sides. cables is 2.0 × 1011 N=m2 . Moreover, the fixed restraints of the no-
The sampling frequency of the accelerometers is 100 Hz. des linking the fishbone model and the auxiliary pier are modified
as free to simulate the girder uplift scenario. Only the bottom nodes
of the tie-down cables are fixed, as shown in Fig. 6.
Numerical Model and Damage Scenarios
To simulate the broken-wire effect of the tie-down cables in-
To validate the detectability of damage to a tension pendulum, a duced by corrosion and fatigue, three damage scenarios are simu-
simulation is implemented for the damage to the tension pendulum lated in the FE analysis. The first scenario is the single-damage case
of a cable-stayed bridge. As shown in Fig. 6, the cable-stayed in which the area ratio of the damaged tension pendulum to the
bridge is modeled as a fishbone finite-element (FE) model with undamaged pendulum is reduced from 100% to 0% at 5% intervals.
updated structural parameters using the FE program ANSYS This decrease is applied to just one tie-down cable. The second
version 15.0. The FE model is available on the SMC (Li et al. scenario is a double-damage case in which the decrease in the area
2014). The towers and concrete girder of the bridge are simulated ratio is the same as in the first scenario but for both tie-down cables
by the elastic beam element BEAM44. The updated elastic moduli at one auxiliary pier. The third scenario is a multiple-damage case

Lifted

Attached

Fig. 6. Finite-element model of cable-stayed bridge.

© ASCE 04020126-4 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2021, 35(1): 04020126

Downloaded from ascelibrary.org by UNB - Universidade de Bras?lia on 09/16/22. Copyright ASCE. For personal use only; all rights reserved.

40

Fig. 7. Single-damage case: frequencies of (a) first vertical bending

mode; and (b) second vertical bending mode.
Fig. 9. Multiple-damage case: frequencies of (a) first vertical bending
mode; and (b) second vertical bending mode.

in which the simulation is applied to all four tie-down cables; the

area proportion is equal to zero, which represents the full fracture of
the tie-down cable. Under these damage scenarios, a modal analysis
of the bridge is carried out to study the effect of tension pendulum
damage on the dynamic performance of the bridge.

Bridge Frequency Results

The first several vertical bending frequencies of the bridge are most
commonly used in practice. Consequently, the frequencies of only
the first several vertical bending modes are obtained through modal
analysis. The variation patterns of these frequencies are similar.
Therefore, only the frequencies of the first two bending modes
are demonstrated in this study for conciseness. The frequency of
the numerical bridge model with intact tie-down cables is called
the initial frequency. The initial frequencies of the first two vertical
bending modes are 0.417 and 0.594 Hz, and the measured frequen-
cies of the real bridge are 0.409 and 0.598 Hz, respectively; thus,
the initial frequencies are 102% and 99% of the measured frequen-
cies. This indicates that the updated FE model and the frequency
results are reliable. The variation trends of the first two frequencies
in the single-damage, double-damage, and multiple-damage cases
are shown in Figs. 7–9, respectively, where the frequency ratio is
the ratio of the frequency of the bridge with damaged tie-down
cables to the initial frequency.
If the girder is lifted from the auxiliary piers, the frequencies
decrease with a decrease in the area ratio in the three damage cases.
The decrement in the frequency of the first vertical bending mode is
Fig. 8. Double-damage case: frequencies of (a) first vertical bending
slightly larger than that of the second vertical bending mode. For
mode; and (b) second vertical bending mode.
example, in the multiple-damage case, the frequency ratios of the

© ASCE 04020126-5 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2021, 35(1): 04020126

Downloaded from ascelibrary.org by UNB - Universidade de Bras?lia on 09/16/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Influence lines of reaction force of Auxiliary Piers P1 and P4.

first and second vertical bending modes decrease to 64.4% and out after completion of the bridge’s construction. The prestress,
69.8%, respectively, when the tie-down cables break (correspond- 50–100 kN, ensured the tightness of each component of the bear-
ing to a decrease in the area ratio of 100%). The 35.6% reduction in ing. The tie-down cables were barely stressed after the construction
the frequency of the first vertical bending mode is larger than the of the bridge was complete. In general, the girder was attached to
30.2% reduction in the frequency of the second vertical bending the auxiliary pier. The influence lines of the reaction force for
mode. For the first vertical bending mode, when the decrease in Auxiliary Piers P1 and P4 are shown in Fig. 10. The minimum
the area ratio is equal to 100%, the frequency ratios decrease to coefficient of the reaction force was −0.84, where the negative co-
93.3%, 75.2%, and 64.4% for the single-damage, double-damage, efficient indicates that the direction of the reaction force was
and multiple-damage cases, respectively. The frequency reduction downward. Derived from the imposed pressure and the coefficient
in the multiple-damage case is the largest, and the frequency reduc- of the reaction forces of the auxiliary piers, the allowable vehicle
tion in the double-damage case is larger than that in the single- weight of the bridge was approximately 834 kN. In other words, if
damage case. The fracture of the tie-down cables evidently leads the weight of a vehicle exceeds 834 kN, the girder would be lifted
to a substantial reduction in the frequency of the cable-stayed from the auxiliary pier when the vehicle passes through the main
bridge. span. It is worth noting that the bridge was subjected to substantial
For the same reduction in the area ratio, representing the same overloading. As shown in the loading data from the SHM system,
degree of damage to the tie-down cables, the frequency is equal to the average daily number of vehicles weighing more than 50 kN
the initial frequency if the bridge is attached to the auxiliary piers. was 742 from January to May 2008 (Li 2010). An average of
However, the frequency decreases suddenly if the condition of 46 vehicles per day weighed more than 1,000 kN; these are referred
the bridge changes to the uplift condition. This indicates that to as heavy vehicles. Several of these heavy vehicles weighed up to
the damage to the tie-down cables is undetectable when the bridge 2,670 kN, which far exceeds the actual allowable vehicle weight.
is attached but detectable when the bridge experiences uplift. A For this bridge, the compression imposed on the auxiliary piers was
decrease in the minimum frequency can reflect a decrease in the wholly insufficient to balance the uplift force of the girder. There-
effective area of a tie-down cable. In a practical scenario, because fore, in the random overloading situation, the bridge would often be
the tie-down cables of a bridge operate in the same environment, lifted from the auxiliary piers during its operation.
the tie-down cables are assumed to experience the same amount of When the bridge experienced uplift, the only connections be-
damage, which corresponds to the multiple-damage case. Damage tween the girder and the auxiliary piers were the tie-down cables.
to the tie-down cables can be detected when the decrease in the area At the moment of uplift, the performance of the tie-down cables
ratio is equal to 30% if the threshold for the frequency change is set would affect the dynamic properties of the bridge. In general,
to 5% in this study. the frequency of the bridge was sensitive to the support condition.
If the effective area of the tie-down cables decreased because of
corrosion and fatigue-induced fracturing, the frequency of the
Damage Analysis of Tension Pendulums for bridge would decrease significantly. Conversely, a sharp decrease
Cable-Stayed Bridge
in the frequency of the bridge would imply damage to the tie-down
As described previously, damage to the tension pendulums of a cables when the bridge experienced uplift.
cable-stayed bridge is detectable only when the bridge is lifted from
its auxiliary piers. Therefore, an uplift analysis of a real bridge Damage Analysis Using Bridge Modal Frequencies
should be implemented.
The accelerometers recorded the acceleration of the bridge from
January to July 2008. However, only 12 days of acceleration data
Uplift Analysis of Bridge Girder could be obtained from the SMC website. Therefore, only data for
As indicated by the results of the loading experiment following these 12 days were utilized to analyze the tension pendulum dam-
completion of bridge construction, the measured pressures imposed age in this study. The frequencies of the bridge on these 12 days
on Auxiliary Piers P1 and P4 were only 627 and 664.9 kN, respec- were automatically identified using 1-h acceleration data by the
tively. The direction of the pressure was downward, but the actual automated eigensystem realization algorithm (Yang et al. 2019).
pressures on Piers P1 and P4 were approximately only half of the The frequencies of the first four bending modes of the bridge
design values (1,369.3 and 1,368.2 kN, respectively). Moreover, are shown in Fig. 11. The maximum measured frequencies of these
installation and prestressing of the tie-down cables were carried four modes were 0.409, 0.598, 0.952, and 1.105 Hz. To illustrate

© ASCE 04020126-6 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2021, 35(1): 04020126

Downloaded from ascelibrary.org by UNB - Universidade de Bras?lia on 09/16/22. Copyright ASCE. For personal use only; all rights reserved.

(a) (c)

(b) (d)

Fig. 11. Change in frequencies of bridge: (a) first mode; (b) second mode; (c) third mode; and (d) fourth mode.

the entire process of tie-down cable damage with these limited fre- worst cases when the bridge was subject to random overloading;
quency data, the tie-down cables were roughly divided into the best boundary condition occurred when the girder was sitting
four conditions, labeled C1, C2, C3, and C4. Each condition was (at least partially) on the pot bearings, while the worst boundary
roughly defined by the maximum or minimum frequency, which condition occurred when the girder was lifted from the pot bear-
reflected the best or worst boundary conditions, respectively, of ings. As in the analysis of the numerical results, even if the tension
the bridge during the random overloading situation. pendulum had experienced serious damage, the numerical fre-
C1: Condition 1 mainly spanned the period from January to quency was equal to the initial frequency of the healthy bridge when
February. The frequencies were equal to the frequencies of the the girder was sitting entirely on the pot bearings. When the girder
bridge in a healthy state. This indicates that the girder sat on the was uplifted, the frequencies dropped substantially. In a random
pot bearings. At this moment, the bridge frequencies could not re- overloading situation, the frequencies of the bridge randomly fluc-
flect the damage condition of the tie-down cables, although the tie- tuated between the two conditions. In addition, the decreasing trend
down cables were heavily damaged. of the minimum frequency could represent a gradual decrease in the
C2: The frequencies decreased substantially by approximately effective area of the tie-down cable.
10%–20% for the four bending modes of the bridge. It is possible C4: The frequencies decreased to their minimum values,
that the girder experienced uplift due to overloading and other fac- 60%–65% of the frequencies of the healthy bridge. The tie-down
tors. The only links between the girder and the auxiliary piers were cables broke completely in July, which was determined by an onsite
the remaining tie-down cables. Moreover, the effective area of the detection of the bridge. After all the tie-down cables were fractured,
tie-down cables decreased substantially due to the corrosion and the girder was lifted a few centimeters from the pot bearing, as
fatigue of the tie-down cables. The corrosion condition of the tie- shown in Fig. 12(b). The final constraint condition and the decrease
down cables can be estimated from Fig. 12(a). in the frequency of the studied case were the same as the results of
C3: The frequencies further decreased relative to the frequencies the simulation for the multiple-damage case.
in C2, indicating that the tie-down cables deteriorated further due to The failure of the uplift-restraint bearing is shown in Fig. 12.
additional damage. This deterioration resulted from the fatigue From the onsite detection, as shown in Fig. 12(a), the effective
damage of the tie-down cables under substantial long-term over- area of the tie-down cables decreased by more than 40% owing
loading. The girder experienced uplift during this time. The to 20 years of corrosion (Li 2010). According to conservative es-
frequencies fluctuated up and down during the periods from C1 timates, in January 2008, the effective area of the tie-down
to C2 and from C2 to C3 but had an overall decreasing trend. cables had decreased by approximately 40% because of the com-
The reason for these random fluctuations may be because the boun- bined action of corrosion and fatigue. This indirectly validates the
dary condition of the bridge always changed between the best and hypothesis that the tie-down cables had been damaged heavily in

© ASCE 04020126-7 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2021, 35(1): 04020126

 substantially when the tension pendulums were fully broken in
this case. This substantial decrease (up to approximately 35%)
is far larger than the frequency fluctuation induced by environ-
mental factors. This is not a special case. Derived from structural
mechanics, the failure of tension pendulums and the uplift of the
girder is equivalent to removing auxiliary piers from cable-stayed
bridges. Therefore, the effect of removing auxiliary piers on the
frequency is equivalent to the effect of the failure of the tension
pendulums. For the cable-stayed bridge model presented in a
study by Kuihua et al. (2017), the frequency of the first vertical
bending mode decreased from 0.167 to 0.115 Hz when the aux-
iliary piers were removed from the bridge. The reduction in the
Downloaded from ascelibrary.org by UNB - Universidade de Bras?lia on 09/16/22. Copyright ASCE. For personal use only; all rights reserved.

frequency of the first vertical bending mode reached up to 30%.

For the cable-stayed bridge model described in a study by Xing
et al. (2014), the frequency of the first vertical bending mode de-
creased from 0.182 to 0.146 Hz, and that of the second vertical
bending mode decreased from 0.224 to 0.180 Hz, constituting an
approximately 20% reduction in the frequencies. For other cable-
stayed bridge models (Huang and Liu 2007; Miao and Guo 2016;
Zhou 2013), the frequency of the first vertical bending mode
(a) would experience a reduction of at least 30% when the auxiliary
piers were to be removed from the bridges. Consequently, if the
tension pendulums of a cable-stayed bridge are fully broken, the
bending frequency of the bridge will experience a substantial re-
duction. In summary, detecting the damage to tension pendulums
of a cable-stayed bridge is entirely feasible in practice using the
structural frequency.

Conclusions

Tension pendulums are components of uplift-restraint bearings that

resist the uplift of a bridge from auxiliary piers. To detect the dam-
age to tension pendulums in cable-stayed bridges, a frequency in-
dex is derived by the Rayleigh method. A simulation validation is
carried out, and a case study of the Yonghe Bridge is utilized to
demonstrate the detectability of the damage. The main conclusions
that can be drawn from this study are as follows:
• Frequency change is far more sensitive to damage to tension
(b) pendulums than to damage to stay cables in a cable-stayed
bridge when the bridge is lifted from auxiliary piers. This rel-
Fig. 12. Failure of bearing: (a) fracture of tie-down cable (reprinted atively high sensitivity guarantees an easy detectability of ten-
with permission from Li 2010); (b) uplift of girder from pot bearing sion pendulum damage. However, such damage is undetectable
[reprinted from Kaloop and Hu 2015 under Creative Commons-BY- when the bridge is attached to auxiliary piers.
3.0 license (https://creativecommons.org/licenses/by/3.0/)]. • For the same area ratio of tie-down cables, the bridge frequency
is equal to the initial frequency if the bridge is attached to
auxiliary piers, and the frequency of the bridge experiences a
sudden decrease if the bridge is uplifted. Furthermore, with a
decrease in the area ratio, the bridge frequency decreases non-
January, although the frequencies showed no change in Stage C1. linearly. Finally, the fracture of tie-down cables results in a
In addition, the allowable weight of a single vehicle is 834 kN, significant decrease in the bridge frequency of approximately
but the measured single vehicle weight reached up to 2,670 kN. 35% for the frequency of the first vertical bending mode in
In such a serious overloading situation, tension pendulum damage the simulation.
will be accelerated. As illustrated earlier, the tension pendulums • A substantial decrease in structural frequency is caused by
were fully damaged during a period of just 8 months owing to the the complete failure of tension pendulums in a real bridge
combined effect of the initial damage and substantial overloading. and in the numerical simulations of some cable-stayed bridges.
This indicates that detecting damage to tension pendulums on
cable-stayed bridges is entirely feasible in practice using the
Empirical Effect of Tension Pendulums on Structural
structural frequency variance.
Frequency
In this study, the sensitivity of bridge frequency is not assigned
In the case study of the Yonghe Bridge, the effect of structural specific parameters. The frequency index is used only to qualita-
temperature on frequency is ignored because the change in the tively detect damage to tension pendulums. For quantitative detec-
frequency induced by the structural temperature is much smaller tion, more empirical parameters for cable-stayed bridges will be
than 35%. In contrast, the frequency of the bridge decreased considered in future work.

© ASCE 04020126-8 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2021, 35(1): 04020126

 Data Availability Statement Lee, Y. J., S. H. Lee, and H. S. Lee. 2015. “Reliability assessment of
tie-down cables for cable-stayed bridges subject to negative reactions:
All the data, models, or code generated or used during the study are Case study.” J. Bridge Eng. 20 (10): 4014108. https://doi.org/10.1061
available from the corresponding author by request. /(ASCE)BE.1943-5592.0000717.
Li, H. 2010. “Fracture cause analysis of tension rocker bearing cables
at subsidiary piers in a prestressed concrete cable-stayed bridge.”
Acknowledgments Supplement, Eng. Mech. 27 (1): 210–215.
Li, H., S. Li, J. Ou, and H. Li. 2010. “Modal identification of bridges under
This research work was jointly supported by the National Natural varying environmental conditions: Temperature and wind effects.”
Science Foundation of China (Grants Nos. 51625802, 51978128, Struct. Control Health 17 (5): 495–512. https://doi.org/10.1002/stc
and 51708088), the LiaoNing Revitalization Talents Program .319.
(Grant XLYC1802035), and the Foundation for High Level Talent Li, S., H. Li, Y. Liu, C. Lan, W. Zhou, and J. Ou. 2014. “SMC structural
Innovation Support Program of Dalian (Grant No. 2017RD03). health monitoring benchmark problem using monitored data from an
Downloaded from ascelibrary.org by UNB - Universidade de Bras?lia on 09/16/22. Copyright ASCE. For personal use only; all rights reserved.

actual cable-stayed bridge.” Struct. Control Health 21 (2): 156–172.

https://doi.org/10.1002/stc.1559.
References Malomo, D., N. Scattarreggia, A. Orgnoni, R. Pinho, M. Moratti, and G. M.
Calvi. 2020. “Numerical study on the collapse of the Morandi Bridge.”
Ala, N., E. H. Power, and A. Azizinamini. 2016. “Predicting the service life J. Perform. Constr. Facil. 34 (4): 4020044. https://doi.org/10.1061
of sliding surfaces in bridge bearings.” J. Bridge Eng. 21 (2): 04015035. /(ASCE)CF.1943-5509.0001428.
https://doi.org/10.1061/(ASCE)BE.1943-5592.0000767. Miao, Y., and X. Guo. 2016. “Analysis of the influence of auxiliary piers to
Brownjohn, J. M. W., K. Y. Koo, A. Scullion, and D. List. 2015. “Opera- the coupled vibration of vehicle and long span cable-stayed bridge with
tional deformations in long-span bridges.” Struct. Infrastruct. Eng. hybrid girder.” J. Railway Sci. Eng. 13 (5): 876–881.
11 (4): 556–574. https://doi.org/10.1080/15732479.2014.951857. Ministry of Transport of the People’s Republic of China. 2007. Guidelines
Calvi, G. M., M. Moratti, G. J. O’Reilly, N. Scattarreggia, R. Monteiro, D. for design of highway cable-stayed bridge. JTG/T D65-01-2007.
Malomo, P. M. Calvi, and R. Pinho. 2019. “Once upon a time in Italy:
[In Chinese.] Beijing: China Communications Press.
The tale of the Morandi Bridge.” Struct. Eng. Int. 29 (2): 198–217.
Noade, B. M., and T. C. Becker. 2019. “Probabilistic framework for life-
https://doi.org/10.1080/10168664.2018.1558033.
time bridge-bearing demands.” J. Bridge Eng. 24 (7): 4019065. https://
Chen, C.-C., W.-H. Wu, F. Shih, and S.-W. Wang. 2014. “Scour evaluation
doi.org/10.1061/(ASCE)BE.1943-5592.0001430.
for foundation of a cable-stayed bridge based on ambient vibration mea-
surements of superstructure.” NDT & E. Int. 66 (Sep): 16–27. https://doi Svensson, H. 2012. Cable-stayed bridges. Weinheim, Germany: Wiley.
.org/10.1016/j.ndteint.2014.04.005. Swiggum, K. E., S. D. Anderson, and J. S. Russell. 1994. “Case study of
Chen, Y., K. Chen, and J. Lv. 2011. “Analysis of construction control for Burlington cable-stayed bridge.” J. Constr. Eng. Manage. 120 (3):
rehabilitation of tension rocker bearings of subsidiary piers in PC 649–666. https://doi.org/10.1061/(ASCE)0733-9364(1994)120:3(649).
cable-stayed bridge.” [In Chinese.] J. Highway Transp. Res. Dev. Webb, G. T., P. J. Vardanega, P. R. A. Fidler, and C. R. Middleton. 2014.
28 (4): 79–83. “Analysis of structural health monitoring data from Hammersmith fly-
Clough, R. W., and J. Penzien. 1975. Dynamics of structures. New York: over.” J. Bridge Eng. 19 (6): 5014003. https://doi.org/10.1061/(ASCE)
McGraw-Hill. BE.1943-5592.0000587.
Fayyadh, M. M., and H. Abdul Razak. 2012. “Condition assessment of Wolff, M., and U. Starossek. 2009. “Cable loss and progressive collapse in
elastic bearing supports using vibration data.” Constr. Build. Mater. cable-stayed bridges.” Bridge Struct. 5 (1): 17–28. https://doi.org/10
30 (May): 616–628. https://doi.org/10.1016/j.conbuildmat.2011.12 .1080/15732480902775615.
.043. Wu, G.-M., D.-H. Yang, T.-H. Yi, H.-N. Li, and H. Liu. 2020. “Sliding life
Guo, T., J. Liu, and L. Y. Huang. 2016. “Investigation and control of prediction of sliding bearings using dynamic monitoring data of
excessive cumulative girder movements of long-span steel suspension bridges.” Struct. Control Health 27 (5): e2515. https://doi.org/10
bridges.” Eng. Struct. 125 (Oct): 217–226. https://doi.org/10.1016/j .1002/stc.2515.
.engstruct.2016.07.003. Xing, C., H. Wang, A. Li, and Y. Xu. 2014. “Study on wind-induced
Hoult, N. A., P. R. A. Fidler, P. G. Hill, and C. R. Middleton. 2010. “Long- vibration control of a long-span cable-stayed bridge using TMD-type
term wireless structural health monitoring of the Ferriby Road Bridge.” counterweight.” J. Bridge Eng. 19 (1): 141–148. https://doi.org/10
J. Bridge Eng. 15 (2): 153–159. https://doi.org/10.1061/(ASCE)BE .1061/(ASCE)BE.1943-5592.0000500.
.1943-5592.0000049. Xu, X., Q. Huang, Y. Ren, D.-Y. Zhao, J. Yang, and D.-Y. Zhang. 2019.
Huang, P., and X. Liu. 2007. “Dynamic characteristics analysis of cable- “Modeling and separation of thermal effects from cable-stayed bridge
stayed bridges with single tower under various boundary conditions.”
response.” J. Bridge Eng. 24 (5): 4019028. https://doi.org/10.1061
Highway (8): 1–5.
/(ASCE)BE.1943-5592.0001387.
Huang, X., and M. Lai. 2016. “Analysis on the tension pressure bearing
Yang, X.-M., T.-H. Yi, C.-X. Qu, H.-N. Li, and H. Liu. 2019. “Automated
disease of PC cable-stayed bridge.” J. Wuhan Inst. Shipbuild. Technol.
eigensystem realization algorithm for operational modal identification
(3): 21–23.
Kaloop, M. R., and J. W. Hu. 2015. “Stayed-cable bridge damage detection of bridge structures.” J. Aerosp. Eng. 32 (2): 4018148. https://doi.org/10
and localization based on accelerometer health monitoring measure- .1061/(ASCE)AS.1943-5525.0000984.
ments.” Shock Vib. 2015: 1–11. https://doi.org/10.1155/2015/102680. Zhou, Q. 2013. “The impact of auxiliary pier tension-compression bearing
Kromanis, R., P. Kripakaran, and B. Harvey. 2016. “Long-term structural on cable-stayed bridge with unequal height pylons dynamic character-
health monitoring of the Cleddau bridge: Evaluation of quasi-static istics and seismic response.” [In Chinese.] Ph.D. dissertation, School of
temperature effects on bearing movements.” Struct. Infrastruct. Eng. Civil Engineering and Architecture, Changsha Univ. of Science and
12 (10): 1342–1355. https://doi.org/10.1080/15732479.2015.1117113. Technology.
Kuihua, M., S. Shengjiang, J. Guoqing, and S. Yamin. 2017. “Static and Zhu, J., W. Zhang, K. F. Zheng, and H. G. Li. 2016. “Seismic design of
dynamic mechanical properties of long-span cable-stayed bridges using a long-span cable-stayed bridge with fluid viscous dampers.” Pract.
CFRP cables.” Adv. Civ. Eng. 2017 (4): 1–11. https://doi.org/10.1155 Period. Struct. Des. Constr. 21 (1): 04015006 . https://doi.org/10.1061
/2017/6198296. /(ASCE)SC.1943-5576.0000262.

© ASCE 04020126-9 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2021, 35(1): 04020126