Vous êtes sur la page 1sur 12

Vol.1 No.

1 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATOIN June 2002


Article ID: 1671-3664(2002)01-0074-12
Modeling of cable vibration effects of cable-stayed bridges
S.H. Cheng

and David T. Lau

Department of Civil and Environmental Engineering, Carleton University, Ottawa, Ontario, Canada
Abstract: The analysis of dynamic responses of cable-stayed bridges subjected to wind and earthquake loads generally
considers only the motions of the bridge deck and pylons. The influence of the stay cable vibration on the responses of the bridge
is either ignored or considered by approximate procedures. The transverse vibration of the stay cables, which can be significant in
some cases, are usually neglected in previous research. In the present study, a new three-node cable element has been developed to
model the transverse motions of the cables. The interactions between the cable behavior and the other parts of the bridge
superstructure are considered by the concept of dynamic stiffness. The nonlinear effect of the cable caused by its self-weight is
included in the formulation. Numerical examples are presented to demonstrate the accuracy and efficiency of the proposed model.
The impact of cable vibration behavior on the dynamic characteristics of cable-stayed bridges is discussed.
Keywords: cable-stayed bridges; cable vibration; dynamics; finite elements; long-span structures
1 Introduction
In recent years, the innovations in the design of
long-span bridges have led to the construction of many
cable-stayed bridges of increasingly long span length.
The dynamic behavior of modern long span, flexible
cable-stayed bridge structures subjected to transient
dynamic loads, such as wind, earthquakes and traffic
loads, are very complex. The dynamic responses of a
cable-stayed bridge are the combinations of and
interactions between the motions of the bridge deck and
the pylons, as well as those of the stay cables. Under the
conditions of heavy traffic loads or unfavorable weather
conditions of wind accompanied by rain, the amplitude
of the stay cable vibration can be large. The observation
of significant rain-wind-induced cable vibration was
first reported by Hikami (1986) during the construction
of the Meiko-Nishi cable-stayed bridge in Japan, with a
maximum amplitude of 0.55m.

In previous analytical studies of cable-stayed bridges
(Fleming and Egeseli, 1980; Parvez and Wieland, 1987;
Wetyavivorn, 1987; Abdel-Ghaffar and Nazmy, 1991;
Nazmy and Abdel-Ghaffar, 1992), a common approach
in the modeling of the cable load carrying behavior is to
model the individual stay cable by a truss element in the


Correspondence to: David T. Lau, Dept. of Civil and
Environmental Engineering, Carleton University, 1125
Colonel By Drive, Ottawa, Ont., Canada, K1S 5B6.
Tel: (613) 520-2600 ext 7473, Fax: (613) 520-3951
Email: dtl@ccs.carleton.ca

Post-doctoral fellow,

Professor
Supported by: Natural Science and Engineering Research Council
of Canada
bridge finite element model. The sagging effect due to
the self-weight of the cable is taken into account by
using an equivalent Young's modulus. The single-truss
element model can give accurate results in static or
quasi-static analysis of cable-stayed bridges. For
dynamic response analysis, the truss model does not
account for the lateral vibration motions of the cable.
Thus, the interaction effects between the lateral
vibration behavior of the cables and the other parts of
the bridge superstructure, as well as the transverse
vibration of the stay cable itself, are ignored.
To address the interaction and vibration effects of
the cables, research works have been carried out to
improve the modeling of the behavior of stay cables.
Causevic and Screckovic (1987) have developed a
model using linear springs with lumped masses to
represent the stay cables in the analysis. Another
common approach is the multi-link model, which
discretizes a stay cable into multiple elements. This
modeling approach of the stay cables has been adopted
by several researchers to analyze the dynamic behavior
of cable-stayed bridges due to wind and seismic loads
(Baron and Lien, 1973; Yiu, 1982; Tuladhar and Brotton,
1987; Abdel-Ghaffar and Khalifa, 1991; Tuladhar et al.,
1995). However, this modeling scheme significantly
increases the number of degree-of-freedom of the
analytical model, especially when analyzing modern
multi-stay cable-stayed bridges. Although comparisons
have been made on the accuracy between the results
obtained from the single-truss model and the multi-link
model in some of the recent research (Abdel-Ghaffar,
1991; Tuladhar et al., 1995), the impact of cable
vibration effect on the dynamic responses of the cable-
No.1 S.H.Cheng et al. : Modeling of cable vibration effects of cable-stayed bridges 75
stayed bridges is still not clear.
The sensitivity of aerodynamic stability behavior to
the vibration of stay cables is an important issue for
consideration in the design of modern long-span cable-
supported bridges. With the trend of increasingly longer
span cable-stayed bridges, the length of the stay cables
is also getting longer and thus becomes more susceptible
to undergo vibration motions. The vibration of the
cables and their interaction with the motion of the
bridge superstructure become even more significant. It
is therefore essential to consider the effects and evaluate
the impact of the cable vibration on the dynamic
behavior of modern long-span cable-stayed bridges in
order to ensure safe and high performance design and
operation.
This paper presents a detailed investigation of the
cable vibration effect on the dynamic behavior of cable
stayed bridges. The formulation of a new and more
efficient cable model of using newly developed three-
node cable elements is presented. The new cable model
can accurately simulate the vibrational behavior of the
stay cables with significantly less number of degree-of-
freedom than the multi-link model. The interaction
behavior between the cables, the pylons and the bridge
deck are considered by adopting the dynamic stiffness in
the modeling of the cables. The accuracy and reliability
of the new cable model are demonstrated by numerical
examples and a case study of a long-span cable-stayed
bridge. The impact of considering cable vibration
behavior on the dynamic characteristics of the example
bridges is discussed in the paper.
2 Modeling of stay cable vibration
2.1 Transverse motion of stay cable
A flexible cable of length l fixed at both ends is
shown in Fig.1(a). The mass per unit length of the cable
is . For small amplitude vibration in the transverse
direction, it is assumed that the tension, S, in the cable
remains constant when the cable vibrates. The
equilibrium condition of an infinitesimal small segment
of the cable shown in figure 1(b) at time t can be
expressed as follows:

2 2
2 2
0
y y
t S x
c c

c c
= (1)
where x is the longitudinal direction of the cable, and y
is the transverse direction normal to the length of the
cable. By separating the variables in the linear equation,
and assuming motion is harmonic with fixed-fixed
boundary conditions, the n transverse vibration mode
shape of the cable is expressed as follows:
th
2
( ) sin
n
n
f l n x
Y x
n l
t
= (2)
where
2
n
n
f S
l
=
th
n
( ) is the natural
frequency of the mode.
1, 2, 3, n =
Fig. 1 (a) Stay cable fixed at both ends; (b) equilibrium of an
infinitesimal small segment of cable
The transverse displacements at an arbitrary point
on a stay cable in three dimensions, as shown in Fig.2,
can be decomposed into two parts. The first part is the
linear displacements related to the nodal displacements
at the two ends of the cable, typically connected at one
end to the girder of the bridge deck and the other end to
the pylon. The second part is the displacements due to
the transverse vibration of the cable with the fixed
boundary conditions at both ends. The second part of the
displacements can be expressed as the summation of the
modal components given in Eq. (2). Thus, the total
transverse displacements of the cable can be expressed
as follows:
Fig. 2 Transverse vibration of a stay cable
EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol.1 76
_
+ + =
n
n y j i
L
x n
a v
L
x
1 v
L
x
x v
t
sin ) ( ) ( (3)
_
+ + =
n
n z j i
L
x n
a w
L
x
1 w
L
x
x w
t
sin ) ( ) ( (4)
where v , w are the displacements along the Y , Z axes
in the local coordinate system, L is the length of the
cable,
i
v ,
i
w ,
j
v and
j
w are the nodal displacements of
nodes and along the i j Y , Z axes in the local
coordinate system, and
yn
a ,
zn
a are the generalized
coordinates corresponding to the n modal shape
function terms. The number of modal coordinates to be
included in Eqs. (3) and (4) is determined by the
accuracy required in the analysis.
th
2.2 Longitudinal motion of stay cable
2.2.1 Static cable stiffness
The extensional behavior of the stay cable is another
important component of its dynamic motion. In practice,
the nonlinear static stretching stiffness, K , has the
form as follows:
c
g e c
1 1 1
K K K
+ = (5)
where K
e
=EA/L is the elastic stiffness due to the elastic
stretch of the cable, K is the
geometric stiffness due to the change in the cable
geometry, where E is the elastic Young's modulus, A
is the cross-sectional area of the cable, L is the chord
length of the cable, L
) /( 12
3
h
2 3
g
L W T =
h
is the length of the horizontal
projection of the cable, T is the mean tension in the
cable, and W is the weight per unit length.
2.2.2 Dynamic cable stiffness
When the bridge is subjected to dynamic loads, the
response of a stay cable is influenced by the movements
of the bridge deck and the pylon through its two end
supports. Thus, the dynamic behavior of the stay cable
is affected not only by the vibration of the cable itself,
but also by the motions of the bridge deck and the pylon.
The dynamic interaction between the stay cable and the
other parts of the superstructure should therefore be
considered in the dynamic response analysis. Further,
the longitudinal and the transverse motions in the plane
of the stay cable are dynamically coupled. If the tension
of the cable changes, the sag in the cable will change
which in turn will affect the transverse acceleration and
the inertia load in the cable. The geometric nonlinearity
of the stay cable resulted from the dynamic coupling
between the motions in the transverse and the
longitudinal direction is considered in the model by the
dynamic cable stiffness.
The dynamic extensional cable stiffness at frequency
of the structure has been derived by Davenport
(1994), which has the form as follows:
e
2
0 g e
2
0 e
) / ( ) / (
] ) / ( [1
) (
e e
e e
e
+

=
K K 1
K
K (6)
where ( ) W T L / /
g 0
t e = is the fundamental frequency
of the taut wire.
Fig.3 Dynamic cable stretch stiffness
The variation of the dynamic cable stiffness is
plotted in Fig.3. The shaded area between the frequen-
cies and represents the region of the unstable
behavior of the stay cable, where
0
e e
1
) / ( 1
g e 0 1
K K + = e e .
Within this frequency range, the stiffness of the cable
becomes negative.
Studies on the massive guy cables (Davenport and
Steels, 1965) have shown that the dynamic stiffness
effect has a significant impact on the dynamic behavior
of guy cable structures. At certain frequencies when the
effective stiffness becomes negative, a dangerous
situation occurs that instead of applying a restoring
force to the mast pole, the guy cable tends to pull it over.
To the best of the author's knowledge, no detailed study
of the influence of the dynamic cable stiffness on the
behavior of the cable-stayed bridges has been reported
in the literature. It is of practical importance to know
how significant the effect of the dynamic stiffness is as
compared to the static stiffness, and the possibility of
when the effective stiffness of the cable becomes
negative.
In order to more accurately account for the cable
behavior in the dynamic response of the cable-stayed
bridges, the formulation of the dynamic cable stiffness
is included in the cable model employed in the current
No.1 S.H.Cheng et al. : Modeling of cable vibration effects of cable-stayed bridges 77
study. The impact of the dynamic stiffness is
investigated in detail in the case study of the Nanpu
Bridge.
A spring-mass-spring system, modified from the
model used in previous studies of dynamic behavior of
guy cable structures (Davenport and Steels, 1965), as
shown in Fig.4, is considered for the derivation of the
dynamic cable stiffness in the present investigation. For
convenience in the derivation of the longitudinal
displacement shape function of the cable, the cable mass
which participates in the longitudinal translational
motion is assumed to be lumped at the center point m of
the cable model. Appropriate constraints are applied to
the motion of the lumped cable mass when considering
the longitudinal motion of the stay cable. The stiffness
of the springs on the two sides of the lumped mass are
the geometric stiffness, K
g
, and the elastic stiffness, K
e
,
respectively. The new three-node cable model is shown
in Fig. 5. The in-plane displacement ( ) u x of an
arbitrary point x on the cable relative to the
displacement of the cable support point at the pylon can
be expressed as follows:
( ) = u x
2 ( )/ , 0 /2
2( )( )/ (2 )( )/ , /2
s s

+ s s

m j
m j i j
x u u L x L
L x u u L x L u u L L x L
(7)
where
i
u ,
m
u and
j
u are, respectively, the longitudinal
displacements at points i, m and j.
2.3 Stiffness matrix of stay cable element
During dynamic responses, the motion of the stay
cable includes components of translational movement in
the transverse direction and extension of the cable in the
longitudinal direction. For the derivation here, the first
three modal coordinates are considered in the modal
series in Eqs. (3) and (4) to express the transverse
displacements v and w along the Y and Z axes,
respectively. The displacements at an arbitrary point of
the cable element can be expressed in the matrix form as
follows:
Fig. 4 Spring-mass-spring system for modeling longitudinal
dynamic behavior of stay cable
Fig. 5 Displacement parameters of three-node stay cable model
| |
e
, , o o =
T
w v u
N N N (8)
where
=
e
o
T
3
3
2
2
1
1
(

z y z y z y
a a a a a a u w v u w v u
m j j j i i i
is the displacement parameter vector,
T
] , , [ w v u = o
is the displacement vector, N
u
, N
v
and N
w
are the shape
function matrices for displacements u , v and w, respe-
ctively:
N
u
=

s s
s s
L x L L x L L L x
L L L
2 / ], 0 0 0 0 0 0 / ) ( 2 0 0 1 0 0 / ) / 2 [(
/2 x 0 0], 0 0 0 0 0 ) / x (2 0 0 ) / x (-2 0 0 [0
N
v
= ) / sin( 0 0 / 0 0 ) / 1 ( 0 [ L x L x L x t ] 0 ) / 3 sin( 0 ) / 2 sin( 0 L x L x t t (9)
N
w
= ) / sin( 0 0 / 0 0 ) / 1 ( 0 [ L x L x L x t ] 0 ) / 3 sin( 0 ) / 2 sin( 0 L x L x t t
EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol.1 78
The stiffness matrix of the cable element of size
13 13 can be derived by calculating the strain energy
of the cable due to elongation and change in geometry.
The non-zero terms in the lower-triangular part of the
symmetric cable element stiffness matrix are given as
follows:
e 71 11
K K K = =
22 33 55 66 52 63
/ K K K K K K T = = = = = = L
L
) L
g 74 44
K K K = =
g e 77
K K K + =
2
88 99
/(2 ) K K T L t = =
2
10,10 11,11
2 / K K T t = =
2
12,12 13,13
9 /(2 K K T t = = (10)
2.4 Mass matrix of stay cable element
For the modeling of the longitudinal motion, the
translational mass related to the extensional behavior of
the cable is assumed to be lumped at the center point m.
The mass related to the transverse vibration are
determined from the kinetic energy of the transverse
motion, i.e., T
)
=
L
AL
0
) ( 2) / (1 x w v d ) (
2 2

+ , where
is the mass density. The total mass matrix

c M of the
cable element can be obtained by combining the
contributions to the mass effect from all the cable
vibration motion components. The non-zero terms in the
lower-triangular part of the symmetric cable element
mass matrix are given as follows:
22 33 55 66
/ 3 M M M M AL = = = = :
25 36
/ 6 M M AL = =
77
1 M =
88 99 10,10 11,11 12,12 13,13
/ 2
M M M M M M
AL
= = = = =
=
82 85 93 96
/ M M M M AL t = = = =
10,2 10,5 11,3 11,6
/(2 ) M M M M AL t = = = =
12,2 12,5 13,3 13,6
/(3 ) M M M M AL = = = = t (11)
3 Numerical examples
To verify the validity and reliability of the
proposed three-node cable model, the numerical
example of a three-span cable-stayed bridge used in a
previous study by Tuladhar et al.(1995) is presented in
this section. Modal analysis of the bridge is carried out.
The dead load of the bridge is calculated based on the
unit weight of the bridge deck provided in the reference.
In the present study, the emphasis is placed on the
comparison of the different modeling schemes of the
transverse motion of the cable, and the influence of the
cable vibration effect on the dynamic modal properties
of the bridge.
The geometry and details of the mesh of the
example bridge are shown in Fig.6. The bridge has a
symmetric layout with a main span of 371m and two
side spans of 159m each. The bridge has a steel deck
and concrete pylons.
Fig. 6 (a) Geometry of example bridge; (b) Mesh details of
example bridge
nalytical mode Three different a ls are employed in
the
model
Co from the two
diff
comparison study. The bridge deck and pylons in all
three models are modeled by three-dimensional beam
elements. The stay cables are modeled by the proposed
new three-node cable elements, the multi-link elements
and the single-truss elements in separate analyses,
respectively. The tension forces in the cables are
obtained from the dead load condition. The motion of
the bridge is assumed to be restricted in the plane of the
structure, i.e., only three degrees-of-freedom (two
translations and one rotation) are considered at each
node of the finite element model of the bridge.
3.1 Comparison between the three-node
and the single-truss model
mparing the results obtained
erent models, numerous additional coupled cable
vibration modes are identified by the three-node model.
These cable vibration modes show up as additional
vibration modes between the frequencies of the primary
structural modes of the bridge structure. Shown in Fig. 7
are the mode shapes of some typical corresponding
modes obtained from the new three-node cable model
and the single-truss cable model. The impact on the
mode shapes due to the vibrational behavior of the
cables is clearly demonstrated. The mode shapes
obtained from the three-node cable model include the
motions not only of the deck and the pylons, but also of
the stay cables. In the case of the single-truss cable
model, only the motions of the deck and the pylons are
considered. The cables in the mode shapes by the single-
truss model remain straight as the vibrational behavior
of the cables is ignored. Fig. 8 shows the mode shapes
No.1 S.H.Cheng et al. : Modeling of cable vibration effects of cable-stayed bridges 79
of the additional vibration modes involving cable
vibrations given by the three-node cable model. The
(a) (b)
Fig. 7 Vibrat tayed bridge ion mode shapes of example cable-s
(a) Single-truss cable model; (b) Proposed three-node cable model
odal frequencies obtained by the proposed three-node m
cable model and the single-truss cable model are plotted
against the modal order in Fig.9. The relative flat
regions on the curve of the three-node cable model
result represent the additional vibration modes involving
cable vibrations.
Comparing the frequencies of the cable vibration
modes with the frequencies obtained from the cable
vibration formula
2l
n
n
f S = ( 1, 2, 3, n = ), as
shown in Table 1, it th modal
frequencies of the stay cables of the example bridge are
similar to the analytical values, which indicates that the
additional cable vibration modes are developed without
significant movement or deformation in the other parts
of the bridge superstructure.
For the example bridge, w
is noted that e first three
hich has a steel deck, it is
noted that the fundamental frequencies of the stay cables
occurred between 0.484 Hz to 1.19 Hz are within the
frequency range of the first ten modes of the bridge,
which is between 0.349 Hz and 1.97 Hz. This overlap in
the frequency range illustrates the importance of
considering the cable vibration behavior in the dynamic
response analysis of the example cable-stayed bridge.
Depending on the external load conditions, the impact
of cable vibration effects on the bridge response can be
significant. For example, the total damping property of
the bridge is the combination of structural damping and
aerodynamic damping effects. The cable vibration
modes are important for consideration of the wind
stability of the structure, as well as in suppressing the
bridge vibration movements through contributions to the
aerodynamic damping effect.
Fig.8 Additional vibration modes involving
cable vibration effects
EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol.1 80
Fig. 9 Impact of cable vibration effect on dynamic characteristics
of example bridge
3.2 Comparison between the three-node model
e efficiency of the
pro
pu Bridge
ayed bridge
loc
and the multi-link model
To further demonstrate th
posed three-node cable model, the example bridge is
also analyzed by using multi-link elements to model the
cables. Each cable is divided into two links, four links
and eight links, respectively, in three separate analyses.
The fundamental cable frequencies obtained from the
analytical solution, the three-node model and the multi-
link models are presented in Table 2. Results show that
for the example here, each cable should be divided into
at least eight links in order to achieve accurate results.
The cable vibration effect can be modeled accurately by
the eight-link model. The first twenty-two natural
frequencies obtained from the three-node model and the
eight-link model, including the bridge frequencies and
all the cable fundamental frequencies, are presented in
Table 3. There is good agreement between these two
sets of results. The total number of degree-of-freedom
of the three-node cable model is 171, whereas that of the
eight-link model is 387. Typical of modern multi-stay
cable-stayed bridges, there can be more than a hundred
stay cables in one bridge. If the multi-link element is
employed to model the stay cables, the number of
degree-of-freedom of the analytical model is therefore
significantly higher than that of the proposed cable
model. It not only requires more effort in data
preparation, but also increases significantly the cost in
computer time and storage.
4 Case study of the Nan
Nanpu Bridge is a long-span cable-st
ated in Shanghai, China. It has a center span of 423
m, as shown in Fig.10. The bridge girder has a
composite deck cross-section comprised of a prestressed
concrete slab supported by two I-shaped steel beams.
The impact of cable vibration effect on the vibration
characteristics of the bridge is studied. The influence of
the dynamic cable stiffness on the extensional behavior
Fig. 10 Structural layout of Nanpu Bridge,
China (symmetric half)
of the stay cables is investigated by comparing the
behavior of the longest and the shortest stay cables in
the bridge.
4.1 Modeling of the bridge
Two different numerical models are employed to
study the cable vibration effect on the dynamic
characteristics of the Nanpu Bridge. In these two models,
the main girder of the bridge is modeled by the triple-
beam finite element model (Cheng, 1999), which can
take into account the warping torsional stiffness
behavior of the bridge deck cross-section.
The mass and stiffness properties of the bridge deck
cross-section are modeled by a central beam and two
side beams in the triple-beam finite element model.
Each of these three beams are then further divided into
ninety-four 3D beam elements in the longitudinal
direction separated by the cable support points, the
supports of the piers and the pylon locations. Each of
the two H-shaped pylons is modeled by sixty-nine 3D
beam elements. The difference between the two models
considered here is due to the difference in the modeling
of the stay cables. In the first model, the stay cables are
modeled by single-truss elements, whereas in the second
model, the cables are modeled by the proposed new
three-node cable elements.
4.2 Modal analysis
The modal analysis results obtained from these two
models are presented in Table 4. The results show that
by considering the cable vibration in the dynamic
behavior of the Nanpu Bridge, additional cable vibration
modes as well as new cable-deck coupled modes are
identified. The new cable-deck coupled modes involve
the transverse motions of some stay cables in the cable
plane and the torsional behavior of the main girder in
the center span. Comparing the two sets of results, it is
noted that the natural frequencies of some
corresponding pairs of bridge structural modes are close,
with those obtained from the three-node cable model
slightly lower due to increases in the modal masses of
the bridge system contributed from the vibrating cables.
No.1 S.H.Cheng et al. : Modeling of cable vibration effects of cable-stayed bridges 81
Table 1 First three modal frequencies of stay cables in the example bridge
/ Hz
1 2 3
f
0.7601 0.7526 (8) 1.5202 1.5226 (27) 2.2803 2.2833(38) 2
0.7588 (9)
3 1.1332 1.1342 2.2 3 2.271 (37) 3.3995 3.4004 (48)
0.5117 0.5091 (5) 1.0234 1.0240 (16) 1.5351 1.5372 6
7 0.4853 0.9706 0.9707 (15) 1.4559 1.4558
1.1402 1.1410 2.2 3 2.286 (39) 3.4205 3.4217 9
10 1.1849 1.1847 (21) 2.3697 2.3719 3.5 6 3.555 (51)
0.5122 1.0243 1.0250 (17) 1.5365 1.5372 12
(19) 66 8
4 1.1913 1.1913 (22) 2.3826 2.3866 (42) 3.5739 3.5736 (52)
0.8218 0.8206 (11) 1.6435 1.6439 (32) 2.4653 2.4618 (43) 5
2.4678 (44)
(29)
0.5147 (6) 1.5454 (30)
0.4888 (4) (25)
0.7622 0.7527 (3) 1.5244 1.5259 (28) 2.2865 2.2860 (39) 8
0.7588 (9) 2.2915 (40)
(20) 80 0 (49)
2.2915 (40)
(41) 54 6
0.8212 0.8131 (10) 1.6423 1.6417 (31) 2.4635 2.4618 (43) 11
0.8239 (12) 2.4678 (44)
0.5091 (5) (29)
0.5147 (6) 1.5454 (30)
e modal ord
f / Hz f / Hz
Cable No.
Analytical Three-node Analytical Three-node Analytical Three-node
1 0.4838 0.4844 (3) 0.9676 0.9677 (14) 1.4515 1.4514(24)
Note: Numbers in the brackets indicate th er.
Table 2 Comparison of the fundamental frequencies of the stay cables in the example bridge
Multi-link element /Hz Analytical Three-node Cable No.
cab z
2 links 8 links
/Hz le element /H
4 links
2
3
4
5
6
7
8
9
10
11
12
0.7601
1.1332
1.1913
0.8218
0.5117
0.4853
0.7622
1.1402
1.1849
0.8212
0.5122
1.1342
1.1913
0.8206
0.4888
1.1410
1.1847
0.5091-0.5147
0.6827-0.6860
1.0208
0.7391-0.7415
0.4589
0.6827-0.6860
1.0208-1.0267
1.0653-1.0714
0.7391-0.7415
0.4623
0.7364
1.1623
0.8013
0.4633-0.4810
0.7408
1.1554
0.7991
0.7496
1.1277
1.1842
0.8163
0.4813-0.4876
0.7546
1.1347
1.1775
0.8139
1 0.4838 0.4844
0.7527-0.7588
0.5091-0.5147
0.7527-0.7588
0.8131-0.8289
0.4348-0.4364
1.0653-1.0714
0.4348-0.4364
0.4633-0.4810
1.1013-1.1083
0.4975-0.5029
1.1013-1.1176
0.4975-0.5029
0.4813-0.4876
0.5067-0.5125
0.5067-0.5125
EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol.1 82
Table 3 Comparison of three-node cable model and multi-link cable model modal analysis results
Natural frequency /Hz
Mode
Proposed thre link element
Mode shape
No.
e-node Multi-
cable element (8 links per cable)
1
8-10
14-17
19-22
0.3481
0.4844-0.5147
0.7526-0.8131
0.9677-1.0250
1.1342-1.1913
0.3465
0.4813-0.5125
0.7496-0.8139
0.9530-0.9997
1.1277-1.1842
Heave (girder, Sym.)
H .)
Hea
H
Hea
H
2
3-6
7
11
12
13
18
0.4615
0.5892
0.8206
0.8289
0.9240
1.1017
0.4584
0.5895
0.8259
0.8306
0.9204
1.1110
eave (girder, Anti-Sym
Pure cable motion
ve (girder, Anti-Sym.)
Pure cable motion
eave (girder, Sym.)
Pure cable motion
ve (girder, Anti-Sym.)
Pure cable motion
eave (girder, Sym.)
Pure cable motion
Table 4 Impact of cable vibration behavior on the dynamic characteristics of Nanpu Bridge
Single-
c
truss Proposed three-node
able element cable element
Mode No. ency/Hz Mode No. ency/Hz
Mode shape
Frequ Frequ
1 0.1216 1 0.1216 Floating (girder, Anti-Sym.)
2 0.3433 2 0.3422 Heave (girder, Sym.)
3 0.3687 3 0.3679 Sway (girder, Sym.)
4 0.4255 4 0.4246 Heave .)
6-12 0.4758-0.4767 Pure cable motion (# C22)
14-20 0.4890-0.4901 Pure cable motion (# C21)
5 0.4992
Be .)
23-30 0.5256-0.5262
0.5392
32-38 0.5407-0.5419 Pure cable motion (# C20)
40-47 0.5464-0.5474 Pure cable motion (# S21)
49-54 0.5571-0.5577 Pure cable motion (# C19)
56-63 0.6048-0.6058 Pure cable motion (# S20)
8 0.6106
(girder, Anti-Sym
5 0.4696 New mode *
13 0.4829 New mode *
21 0.4951 Torsion (girder, Sym.)
6 0.5069 22 0.5042 nding (pylon, Anti-Sym
Pure cable motion (# S22)
7 31 0.5376 Bending (pylon, Sym.)
39 0.5448 New mode *
48 0.5537 New mode *
55 0.5599 New mode *
64 0.6073 Torsion (girder, Anti-Sym.)
* w mode: Domina vibration couple th torsion of ma at the center spa Ne nt cable d wi in girder n.
No.1 S.H.Cheng et al. : Modeling of cable vibration effects of cable-stayed bridges 83
Fig. 11 shows the influence of the cable vibration
effect on the dynamic characteristics of the Nanpu
Bridge in terms of modal order. The frequencies of the
first 20 modes obtained from the single-truss cable
model and the first 200 modes obtained from the three-
node cable model are plotted against the mode number.
It can be observed that in the same frequency range of
0-0.8 Hz, there are 14 modes extracted from the single-
truss cable model, whereas more than 200 modes are
identified by the three-node cable model. Thus, by
considering the cable vibration behavior in the modal
analysis, 186 new modes have been extracted in the
frequency range below 0.8 Hz.
Fig. 11 Impact of cable vibration effect on dynamic
characteristics of Nanpu Bridge
The mode shapes of some typical coupled cable-
deck vibration modes obtained by using the three-node
cable model are shown in Fig.12. The interactions of the
cable vibration behavior with the motions of the main
girder and the pylons are clearly shown.
Fig. 12 Mode shapes of the Nanpu Bridge using
three-node cable model
For the case of pure cable vibration modes, the
freq
sets of results is observed. The difference between the
analytical and the modal analysis results is due to the
Fundamental frequency/Hz
uencies of these modes in the center and side spans
of the Nanpu Bridge are compared with the analytical
values, as shown in Table 5. Altogether ten stay cables,
five in the center span and another five in the side spans,
are studied. Fairly good agreement between these two
difference in the boundary conditions. The analytical
solution is based on the fixed-fixed end boundary
conditions of the cables, whereas in the modal analysis,
the two ends of the cable move with the deck and the
pylons.
Table 5 Fundamental frequencies of some longest
stay cables in the Nanpu Bridge
Cable
e-node
cable element
No.*
Analytical Proposed thre
C22
C21
C20
0.4762
0.5418
0.4758-0.4767
C19
C18
S22
S21
S20
S19
S18
0.4894
0.5593
0.6202
0.5278
0.5495
0.6078
0.6102
0.6574
0.4890-0.4901
0.5407-0.5419
0.5571-0.5577
0.6171-0.6203
0.5256-0.5262
0.5465-0.5474
0.6048-0.6058
0.6105-0.6120
0.6587-0.6599
sest to . The sym denotes the cabl
ile 'S' den e cables in the si
* The numbering sequence of the cables starts from the one
clo the pylon bol 'C' es in
the center span, wh otes th de
span.
According to the analytical solution. the fundamental
frequency of the longest cable in the Nanpu Bridge is
0.476 Hz, whereas that of the shortest is 1.58 Hz. The
cable frequencies overlap with the frequency range of
mode 5 to 45 of the bridge structure, i.e., between 0.426
Hz and 1.61 Hz. For dynamic motions of the bridge in
this frequency range, the vibration of the cables and
their interaction with the motions of the deck and the
pylons can be significant, which can affect the dynamic
response of the bridge.
4.3 Dynamic cable stiffness
The dynamic cable stiffness accounts for the
interaction effect between the motion of the cables and
that of the other parts of the bridge superstructure. It
influences the extensional behavior of the cable by
modifying its longitudinal stiffness. To get an insight on
the significance of the interaction effect in the case of
the Nanpu Bridge, the dynamic cable stiffness of the
longest and shortest cables of the Nanpu Bridge are
determined for the bridge modes of natural frequencies
at 0.1 Hz, 0.2 Hz, 0.5 Hz and 1.0 Hz. the cable tensions
are obtained under the dead load condition, The results
are presented in Table 6.
For the case of the longest cable, there is no
significant change in the extensional stiffness of the
cable by including the interaction effect. For the
extensional stiffness of the cable to reach the critical
EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol.1 84
stat
Shortest cable
e of zero, the corresponding bridge frequency for the
longest cable is evaluated to be 7.76 Hz, and for the
shortest cable is 105.6 Hz. The higher frequency is
likely beyond the bridge frequency range that is of most
concern to dynamic response analysis of bridges due to
wind. The results seem to indicate that the impact of
dynamic stiffness on the extensional behavior of the stay
cables in cable-stayed bridges is not as significant as
that pointed out by Davenport and Steels (1965) for
massive guy cables. However, in reality, the stay cables
may be subjected to deterioration due to corrosion and
fatigue problems. This may affect the dynamic stiffness
behavior of the stay cables. There are a number of cases
reported that cable replacements are required due to
tension loss caused by corrosion and fatigue damage in
the cables. These two phenomena happen simulta-
neously and reinforce each other. The tension in the
cable can drop significantly due to damage to the wires
inside the stay cable. Under such conditions, the
dynamic cable stiffness may have a significant impact
on the longitudinal behavior of the cable.
Table 6 Effect of dynamic stiffness on the extensional stiffness
of the longest and shortest cables in the Nanpu Bridge
Item Longest cable
) m /(N
1

e
a
K
7
1.0119 10
7
3.0943 10
)
1
g
a
K m /(N
8
3.8815 10
10
2.3517 10
) m /(N
1
0

K
a
7
0.9862 10
7
3.0902 10
0.1 2 e t =
7
0.9862 10
7
3.0902 10
0.2 2 e t =
7
0.9862 10
7
3.0902 10
( )
d
K e
) m /(N
1
0.5 2 e t =
1.0 2 e t =
7
0.9861 10
7
0.9812 10
7
3.0902 10
7
3.0902 10
A
3 2
/
0
)
K
c s
/
a
e
K E =
d
, 12 [( ) ]
b
g h
K T A L =
3
, /(
c
e g e g
K K K K K = +
2 2
th e
L ,
e
To stu ss on the
dynamic behavior of the stay cables, the dynamic
stiffness of e longest and the shortest cables of th
Nanpu Bridge are investigated for the cases of t nsion in
the
ith a f quency 2 Hz,
ss
ue of
0.8
a
as fatigue related damage which results in
ten
vibration effect on the dynamic
ayed bridges is studied in this paper.
ble element has been developed.
The
strength steel cables may lie within
the
( ) ( ) /[ ( )]
e g e g
K K K m K m e e = +
dy the effe t of the ten ion lo
e
cable taken as
DL
0.75T ,
DL
0.50T and
DL
0.25T res-
pectively, where
DL
T is the initial cable tension caused
by the dead load. The results are shown in table 7.
If the remaining cable tension is
DL
0.50T , when the
bridge vibrates w re of 0 the
extensional stiffne of the longest cable considering the
dynamic interaction may drop by 15.2% to a val
.
nted in 3710
7
N/m. From the data prese table 7, it
can be noticed that the value of the dynamic cable
stiffness depends more on the tension in the cable than
on the external exciting frequency of the bridge motion.
A negative value of the extensional stiffness of the
longest cable occurs under the condition when the
remaining tension force is
DL
0.25T and the oscillating
frequency of the bridge is 1.0 Hz. Physically, this
negative stiffness indicates that instead of resisting the
elongation of the cable, the deform tion tend to increase.
The cable then is under an unstable condition. No such
negative stiffness has been found in the case of the
shortest cable for the same combination of the cable
tensions and the bridge frequencies. The effect of the
dynamic interaction between the cables and the other
parts of the superstructure on the extensional stiffness of
the short cables is significantly less than that on the long
cables.
The results show that in the dynamic analysis of
cable-stayed bridges, if the cable is significantly
corroded or h
sion loss in the cable, the dynamic stiffness effect
should be considered, especially for the long stay cables
in cable-stayed bridges.
5 Conclusions
The impact of cable
behavior of cable-st
A new three-node ca
new cable model can accurately model the
transverse motion of the cable, and the interaction
between the motions of the cables and the other parts of
the superstructure by adopting the dynamic stiffness in
the longitudinal direction. The new cable model is
computationally more efficient than the multi-link
approach, especially in the analysis of modern multi-
stay cable-stayed bridges. The validity and accuracy of
the proposed model have been verified by numerical
examples and the case study analysis of the Nanpu
Bridge in China.
Results show that the natural frequencies of the
cable vibration modes in cable-stayed bridges with
conventional high
frequency range of the lower bridge structural modes.
The transverse vibration motions of the cables can
influence the dynamic response of the bridge and thus
should be considered in the dynamic response analysis.
By considering the cable vibration behavior, the natural
frequencies, mode shapes and modal order of the cable-
stayed bridges are significantly altered. A large number
of additional modes involving cable vibrations as well
as new cable-deck coupled modes are identified by
using the proposed three-node cable model. These
modes cannot be predicted by the conventional single-
truss cable model. The additional cable vibration modes
can have a significant impact on the overall damping
properties of the bridge system by contributing to the
aerodynamic damping behavior of the system. The
analysis results also show that the significance of the
dynamic cable stiffness effect depends on cable tension.
If there is no significant tension loss due to corrosion or
fatigue problems of the cables, the influence of the
dynamic cable stiffness from the longitudinal behavior
of the cables may be negligible and not as significant as
suggested by previous studies of guy cables.
No.1 S.H.Cheng et al. : Modeling of cable vibration effects of cable-stayed bridges 85
Table 7 Impact of cable tension v
Tension
ariation the dynamic cable stiffness
/ (10
7
Nm
-1
)
on
Dynamic cable stiffness ( ) K e e
-1
/ N / ( rad s )
Longest cable Shortest cable
0.1 2t 0.9530 3.0847
0.2 2t 0.9529 3.0847
0.9524 3.0847
0.9507 3.0847
0.8370 3.0620
0.5 2t
0.75
DL
T
1.0 2t
0.1 2t
0.2 2t 0.8365 3.0620
0.8323 3.0620
0.8156 3.0620
0.3766 2.8538
0.5 2t
0.5
DL
0T
1.0 2t
0.1 2t
0.2 2t 0.3689 2.8538
0.3092 2.8538
-0.3964 2.8538
0.5 2t
0.2
DL
5T
1.0 2t
43.
Acknowledgements
by the National Sciences and
uncil of Canada to the second
aut
, A.M. and Khalifa, M.A., (1991),
Cable Vibrations in Dynamics of Cable-
of Cable-stayed Bridges,
Bridge, Computers and Structures,
Bridge Cables: Effects on Bridge
Analysis of Long-Span Cable-Stayed Bridges,
nd,
r of Massive Guy Cables, Journal of Stru-
cture Division, ASCE, 91, (ST2):
Fleming, J.F. and Egeseli, E.A., (1980), Dynamic
f Cables in Cable-
ffects
d
al Dynamics, 21: 1-20.
Cable-stayed Bridges,
able-stayed Bridges,
ingdom.
The support provided
Engineering Research Co
hor for this research gratefully acknowledged.
References
Abdel-Ghaffar
Importance of
stayed Bridges, Journal of Engineering Mechanics,
ASCE, 117(11): 2571-2589.
Abdel-Ghaffar, A.M. and Nazmy, A.S., (1991), 3-D
Non-linear Seismic Behavior
Journal of Engineering Mechanics, ASCE, 117(11):
3456-3476.
Baron, F. and Lien, S., (1973), Analytical Studies of a
Cable-stayed
3:443-465.
Causevic, M.S. and Sreckovic, G., (1987), Modeling of
Cable-stayed
Vibrations, Proceedings of International Conference
on Cable-stayed Bridges, Bangkok, Thailand, pp. 407-
420.
Cheng, S.H., (1999), Structural and Aerodynamic
Stability
Ph.D.Thesis, Department of Civil and Environment
Engineering, Carleton University, Ottawa, Canada.
Davenport, A.G., (1994), A Simple Representation of
the Dynamics of a Massive Stay Cable in Wi
Proceedings of International Conference on Cable-
Stayed and Suspension Bridges, Vol. 2, Deauville, pp.
427-438.
Davenport, A.G. and Steels, G.N., (1965), Dynamic
Behaviou
Behavior of a Cable-stayed Bridge, Earthquake
Engineering and Structural Dynamics, 8(1):1-16.
Hikami, Y., (1986), Rain Vibration o
stayed Bridges, Journal of Wind Engineering, 27(3):
23-34.
Nazmy, A.S. and Abdel-Ghaffar, A.M., (1992), E
of Ground Motion Spatial Variability on the Response
of Cable-stayed Bridges, Earthquake Engineering an
Structur
Parvez, S.M. and Wieland, M., (1987), Earthquake
Behaviour of Proposed Multi-span Cable-stayed Bridge
over River Jamuna in Bangladesh, Proceedings of
International Conference on
Bangkok, Thailand, pp. 479-489.
Tuladhar, R. and Brotton, D.M., (1987), A Computer
Program for Nonlinear Dynamic Analysis of Cable-
stayed Bridges Under Seismic Loading, Proceedings of
International Conference on C
Bangkok, Thailand, pp. 315-326.
Tuladhar, R., Dilger, W.H. and Elbadry, M.M., (1995),
Influence of Cable Vibration on Seismic Response of
Cable-stayed Bridges, Canadian Journal of Civil Engi-
neering, 22: 1001-1020.
Wethyavivorn, B., (1987), Dynamic Behavior of
Cable-stayed Bridges, Ph.D. Thesis, Department of
Civil Engineering, the University of Pittsburgh,
Pittsburgh, PA.
Yiu, P.K.A., (1982), Static and Dynamic Behaviour of
Cable Assisted Bridges, Ph.D. Thesis, Department of
Civil Engineering, the University of Manchester, Man-
chester, United K

Vous aimerez peut-être aussi