You are on page 1of 7

The Fifth International Symposium on Computational Wind Engineering (CWE2010)

Chapel Hill, North Carolina, USA May 23-27, 2010

A study of effect of cable aerodynamic forces and deck static


deformation on flutter for cable-stayed bridges
Zhengqing Chena, Xugang Huab, Kejian Ouyangc
a

Wind Engineering Research Center, Hunan University, China, zqchen@hnu.cn


Wind Engineering Research Center, Hunan University, China, cexghua@hotmail.com
c
Wind Engineering Research Center, Hunan University, China, ouykj@yahoo.com.cn

ABSTRACT: Wind-induced flutter instability is a major concern in the design and construction of super long-span cable-stayed bridges. This kind of bridges is featured by
many long stay cables and the aerodynamic forces exerted on cables may not be neglected. The stiffening girder will also undergo appreciable static deformation at wind
velocity near flutter which affects the modal parameters of bridges and effective wind attack angle as well. Some influence factors that are generally regarded as of minor importance may contribute considerably to wind-induced flutter for cable-stayed bridges with
center span exceeding 1000m, and they are worthy for further exploration and revisit for
the sake of both accuracy in critical flutter speed prediction as well as refined wind tunnel experimental techniques. By taking the cable-stayed bridge of Sutong bridge with a
center span of 1088m, a multimode flutter analysis is carried out to determine the critical
flutter wind velocity. Emphases have been placed on studying the effects of unsteady
forces exerted on stay cables and the static deformation induced by mean wind loadings.
It is shown that aerodynamic force on cables is helpful to stabilize the bridge.
1 INTRODUCTION
The development of modern construction materials and construction techniques leads to
the design and construction of many modern cable-stayed bridges worldwide. Due to its
sensitive response to turbulent wind, wind-resistant performance should be carefully
studied for cables-stayed bridges. Flutter instability is an important index in windresistant performance and flutter instability should be prohibited due to its divergent oscillations and catastrophic consequence. Flutter is defined as at a threshold wind velocity
beyond which the bridge oscillates in a divergent way. Determination of the threshold
wind velocity can be done through wind tunnel experiments of sectional models of bridge
deck or by finite-element-based flutter analysis methods mostly in frequency domain (to
mention a few, e.g. Agar 1989, Chen 1994, , Diana et al. 1995, Katsuchi et al. 1999,
Chen et al. 2000, Ge and Tanaka 2000). The use of sectional model wind tunnel experiments implies that the aerodynamic contribution from other structural components other
than bridge deck is trial. Static deformation due to mean wind near flutter is not properly
accounted for in such tests. Most methods of flutter analysis also only consider the unsteady self-exited forces on bridge deck, and they are shown to be sufficient in most
cases.
Currently, Sutong bridge in mainland China and Stonecutters bridge in Hong Kong,
China are two bridges with a main span exceeding 1000m. These super-long span cablestayed bridges are featured by stay cables ranging from 200m to 600m in length. The projected area of all stay cables, perpendicular to the approach flow, exceeds that of bridge

The Fifth International Symposium on Computational Wind Engineering (CWE2010)


Chapel Hill, North Carolina, USA May 23-27, 2010

deck. For example, there are 270 stay cables in Sutong bridge and the average length and
diameter of these cables are 350m and 0.10m, the projected area of stay cables will be
9520m2; while the projected area of bridge deck is about 20884=8352m2. Therefore the
influence of cable aerodynamic forces on flutter instability may be non-trivial and should
be re-examined for accuracy in flutter prediction and refined wind tunnel experimental
techniques as well. Additionally, the static deformation induced by mean wind loadings
at a velocity near flutter may change appreciably the effective wind attack angle.
The present work mainly aims at studying the influences of cable aerodynamic
forces and static deformation on flutter instability. A multimode flutter analysis method
developed by the first author is first described. The salient feature of the method is the
avoidance of iteration with respect to frequency in the solution course of flutter problem
when structural damping is not included. To consider the aerodynamic forces on stay cables, their equivalent aerodynamic damping matrix and aerodynamic stiffness matrix are
derived from the principles of virtual work. By taking cable-stayed Sutong bridge as a
case study, aerostatic and flutter analysis are carried out by using the measured aerodynamic force coefficients and flutter derivatives from sectional model wind tunnel experiments. The influence of unsteady aerodynamic forces on cables and static deformation on
critical flutter velocity are studied.
2 MULTIMODE FLUTTER ANALYSIS
In the past years, a number of multimode flutter analysis methods have been developed.
In this study, the method developed by Chen (1994) is presented. This method is referred
to as the Multimode-participation and Single-parameter searching method (M-S method).
The most attractive merit of the M-S method is the avoidance of the iteration with respect
to vibration frequency to obtain flutter solution in the case when the structural damping is
ignored. Additionally, it also provides quantitative descriptions on modal participation
factors in terms of modal amplitude and modal energy. The main part of the method is
briefly described as follows.
2.1 M-S method
The equations of motion for a bridge in the smooth flow can be expressed as

&& + CX
& + KX = F
(1)
MX
ae
where M, C and K are the global mass, damping and stiffness matrices, respectively; X ,
&& represent the nodal displacement, velocity and acceleration vectors, respec& and X
X
tively; and Fae denotes the vector of nodal aeroelastic forces.
The motion-dependent forces distributed on unit length of bridge deck are expressed
as a linear function of nodal displacement and nodal velocity, as
* h&
&
1
2
* B&
2 *
2 * h
* p
2 * p
(2a)
L = U (2 B ) KH
+ KH
+K H +K H
+ KH
+K H
2 U
3
4 B
5U
6 B
ae 2
1U

&
* p&
1
2
* B&
2 *
2 * p
* h
2 * h
(2b)
D = U (2 B ) KP1
+ KP2
+ K P3 + K P4
+ KP5
+ K P6
ae 2
U
U
B
U
B

&
&
1
2
2 * h
* B&
2 *
2 * h
* p
2 * p
(2c)
M
= U 2 B KA1
+ KA2
+ K A3 + K A4 + KA5
+ K A6

ae 2
U
U
B
U
U

( )

The Fifth International Symposium on Computational Wind Engineering (CWE2010)


Chapel Hill, North Carolina, USA May 23-27, 2010

where is air mass density; U is wind velocity; B is the width of bridge deck; K = B/U
is the reduced circular frequency; h, p and are the vertical, lateral and torsional displacements, respectively; Ai* , H i* and Pi* (i = 1, L, 6 ) are flutter derivatives which
are expressed in terms of reduced wind velocity = U/(fB); and f is the natural frequency. The aeroelastic forces on bridge deck are illustrated in Fig. 1.
Lae

M ae

Dae

Wind attack angle

B
Fig. 1

Aeroelastic forces acting on bridge deck

By converting the distributed aeroelastic forces on element e of bridge girder into


equivalent nodal loadings at member ends, one obtains the equivalent nodal loadings for
that element, as

Faee = 2 A e X e + 2 B e

&e
X

(3)

where Ae and Be are referred to as elemental aerodynamic force matrices. It is worthy to


note that both the matrices Ae and Be are expressed in terms of flutter derivatives which
are functions of only the reduced wind velocity. The global aeroelastic forces may be assembled as
&
X
(4)
Fae = 2 AX + 2 B

where A and B are global aeroelastic stiffness and damping matrices, respectively.
Substituting Eq. (3) and Eq. (4) into Eq. (1) and then using modal approach give the
representing modal-motion equations, as
&
~
~&
~Y
&& + C
(5)
Y
Y + Y = 2 AY + B

~
~
stiffness matrix; B = T B is the
where A = T is the generalized aeroelastic
~
generalized aeroelastic damping matrix; C = diag (211 2 2 2 L 2 m m ) is the
generalized damping matrix; and = diag 12 22 L m2 is the diagonal matrix of
eigenvalues; m is the number of modes used for flutter analysis.
The bridge vibration becomes harmonic with a single frequency at a threshold wind
velocity since other frequency components tends to dampen out. As a consequence, the
modal response related to this undamped frequency is of interest and is written as
(6)
Y = Y0 exp(it )
Substituting Eq. (6) into Eq. (5) leads to

~
~ 1 ~
(7a, b)
or
Y0 = MY0
Y0 = 2 I + A + i B C Y0

The Fifth International Symposium on Computational Wind Engineering (CWE2010)


Chapel Hill, North Carolina, USA May 23-27, 2010

~
~ 1 ~
where M = I + A + i B C with the order of mm; I is the mm identity matrix;

2
and = gives the eigenvalue at the critical flutter state.
Eq. (7) represents a parameterized and generalized complex eigenvalue formulation
for the coupled wind-bridge system in the modal coordinate, where the entries of matrix
M are expressed in terms of the reduced wind velocity and vibration frequency. The flutter condition is as follows: at a certain value of reduced wind velocity and vibration frequency, the damping part of one complex eigenvalue is zero. In recognizing that the
~
second term of imaginary part in matrix M is zero if C =0 (no structural damping), a single-parameter sweep of reduced wind velocity within a predefined range is sufficient to
determine the critical flutter wind velocity. When structural damping is included, an additional frequency iteration procedure is needed to obtain the exact flutter frequency. By
assuming that one eigenvalue f has zero real part at some reduced wind velocity f, the
flutter frequency and flutter critical wind velocity are obtained as
~
U ff B
(8a, b)
f = f ,
Uf =
2
2.2 Aerodynamic force matrix on stay cables
The aerodynamic force matrix of stiffening girder has been well documented in the
literature (e.g. Ge and Tanaka 2000). For the traditional range of reduced wind velocity
for the bridge deck, the corresponding reduced velocity for the stay cables, Uf/D, will be
larger by a factor of B/D than that of bridge deck, and usually have values of several
hundreds or southlands. Experimental identification of aerodynamic derivatives for the
extremely high reduced velocity by forced vibration technique is underway and the results will be elsewhere when available. However, it is well know that the aerodynamic
forces tend to be steady beyond a particular value of reduced velocity, as a result of
quasi-steady assumptions. In this pilot study, the aerodynamic forces on cables are evaluated by quasi-steady approach. Following this approach, the expression of flutter derivatives of stay cables are given as follows
1
2
(9)
H 1* = C D
P5* = C D
K
K
where subscript bar represent the flutter derivatives for stay cables. C D is drag coefficients of cables and is taken as 0.7 in this study. The remaining flutter derivatives for stay
cables are zero.
After obtaining the flutter derivatives for stay cables, the distributed forces on stay
cables may be converted to nodal loads by applying principle of virtual works. It should
be noted that numerical integration is needed to obtain the aerodynamic matrix for stay
cables as wind velocity varies along the cable length.

3 APPLICATION TO SUTONG BRIDGE


The Sutong bridge crosses the Yangtze River has a total length of 6km and it open to
public in June of 2008. The main bridge is a cable-stayed bridge with a center span of
1088m and two back spans of 500m, as shown in Fig. 2. Two auxiliary and one transi-

The Fifth International Symposium on Computational Wind Engineering (CWE2010)


Chapel Hill, North Carolina, USA May 23-27, 2010

tional piers are erected in each back span. The cable-stayed bridge has two inverted Y
concrete towers reaching about 300m high from the tower bottom. The superstructure is a
streamlined, flat, closed steel box girder with a width of 41m and a height of 4.0m, as
shown in Fig. 3. The stay cables are arranged in double included cable planes with a
standard spacing of 16m in the center span and 12m near the end of the side spans. The
wind tunnel experiments including sectional models and aeroelastic models were carried
out in wind tunnel of Tongji University, China, and the measured aerodynamic force coefficients and flutter derivatives are used in this study (SLDRCE 2002).
A 3D frame finite element (FE) model is developed in the self-developed package
NACS (Chen and Agar 1993), and it consists of 866 nodes and 1122 elements. The modal parameters for the first 30 modes are computed. The first lateral, vertical and torsional frequencies are found to 0.0958Hz, 0.1712Hz and 0.5213Hz. The results agree
well with published results.

Fig. 2.

Span arrangement of the main bridge (m)

Fig. 3.

Cross section of the deck (m)

3.1 Flutter analysis with/without considering cable aerodynamic forces


Flutter analysis without considering aerodynamic forces on stay cables and static deformation is first carried out. The measured flutter derivatives of the deck are taken from
the research report of Tongji University, China (SLDRCE 2002), and only 8 flutter derivatives are available for analysis. The modal damping ratio is assumed to 0.5%. The
analysis results of flutter velocity and frequency are shown in Table 1, and the sectional
model and full aeroelastic model experimental results are also provided for comparison.
The numerical results agree very well with sectional model experimental results. The
flutter shape is characterized by a coupled vertical and torsional displacement. There is a
notable difference in flutter velocity between sectional model test and aeroelastic model
test, which is suspected to be caused by static deformation due to mean wind. Note that

The Fifth International Symposium on Computational Wind Engineering (CWE2010)


Chapel Hill, North Carolina, USA May 23-27, 2010

the static deformation cannot be effectively taken into account for sectional model experiments.
Table 1 Comparison of flutter analysis results and experimental results
Attack
angle

w/o cable aerodynamic forces

w cable dynamic forces

Velocity (m/s) Frequency (Hz) Velocity (m/s) Frequency (Hz)

sectional model Aeroleastic model


(m/s)
(m/s)

0 degree

142

0.4179

145

0.4132

144

115.6

+3 degree

94

0.4681

97

0.4681

89

88.4

Then flutter analysis considering the aerodynamic forces on stay cables are then
conducted. The results are also presented in Table 1. There is about 2%~3% increases in
critical flutter velocity for the two wind attack angles when aerodynamic forces on cables
are considered, which appears insignificant for the bridge in study. It is seen that cable
aerodynamic forces creates additional damping to the cable and therefore to the bridge.
The vibration of stay cables thus dissipate energy and stabilize the bridge system. However in this study local vibration of stay cables is not considered as each cable is modelled with a single truss element. Such effect will be studied in the future.
3.2 Flutter analysis considering static deformation
Next flutter analysis considering static deformation is carried out. First the static deformation under a particular wind velocity is calculated and changes in wind attack angle
due to the deformation of bridge deck are obtained. The flutter derivatives for effective
wind attack angle, accounting for the static deformation, are evaluated by interpolation
approach. The modal parameters are updated as well based on the deformed structure.
The aerodynamic force coefficients in structural axis for the deck are as follows:
CH=0.911, CV=-0.1077 and CM=0.0283 for zero degree wind attack angle; CH=0.9791,
CV=0.1471 and CM=0.0642 for +3 degree wind attack angle. A value of 0.8 and 1.8 are
used for the drag force coefficients for stay cables and bridge tower, respectively.
Fig. 4 shows the rotation of bridge deck for 0 and +3 degree wind attack angle at
two selected wind velocities. The maximum rotation of deck along the longitudinal axis
at the designated wind condition is about 0.0106 (or 0.61o) and 0.0173 (or 0.99o), respectively. The torsional modal frequencies associated with the deformed bridge are calculated as 0.5163Hz and 0.5151Hz for 0 and +3 degree wind attack angles, respectively,
and they are very close to the original bridge.
0.018

Rotation Rxx(rad)

0.014

+3 wind attack angle and Udeck=100m/s

0.012
0.01
0.008
0.006

+3 wind attack angle and Udeck=100m/s

0.004

0.002
0
-1000

zero wind attack angle and Udeck=120m/s

Lateral displacement (m)

0.016

0 wind attack angle and Udeck=120m/s

-800

-600

-400

-200

Location (m)

200

400

600

800

1000

-1
-1000

-800

-600

-400

-200

200

400

600

800

1000

Location (m)

Fig. 4. Rotation and lateral displacement of bridge deck (a) rotation; (b) lateral displacement

By using the effective wind attack angles and updated modal parameters, flutter
analysis is performed for 0 and +3 degree wind attack angles. The flutter velocity is com-

The Fifth International Symposium on Computational Wind Engineering (CWE2010)


Chapel Hill, North Carolina, USA May 23-27, 2010

puted to 128.4m/s and 92.8m/s. The decrease in flutter velocity is mainly contributed by
the change of wind attack angle rather by decrease of torsional modal frequencies.
4 CONCLUDING REMARKS
In this study, a multimode flutter analysis considering deck static deformation and
cable aerodynamic forces is described. The aerodynamic forces on stay cables are approximately evaluated by quasi-steady approach. By taking cable-stayed Sutong bridge
as an example, the influence of static deformation and cable aerodynamic forces are studied. It is shown that the inclusion of cable aerodynamic forces help to increase the flutter
velocity in analysis, but the increase is not significant. The change in wind attack angle
due to mean wind loadings may considerably change (or decrease in this study) the flutter
velocity and needs to be considered for refined accuracy in prediction and improved experimental wind tunnel techniques.
5 ACKNOWLEDGEMENTS
The present work was supported by National Science Foundation of China under the
grant Nos. 50738002 and 50808079, which are acknowledged.
6 REFERENCES
Agar, T. J. A., 1989. Aerodynamic flutter analysis of suspension bridges by a modal technique,
Engineering Structures, 11, 75-82
Chen, X., Matsumoto, M., and Kareem, A., 2000. Aerodynamic coupling effects on flutter and
buffeting of bridges, Journal of Engineering Mechanics, ASCE, 126, 17-26.
Chen, Z. Q., and Agar, T. J. A. 1993. Geometric nonlinear analysis of flexible spatial beam structures, Computers and Structures, 49, 1083-1094
Chen, Z. Q. 1994. The three dimensional analysis and behaviors investigation on the critical flutter state of bridges, Proceedings of the International Symposium on Cable-Stayed Bridges,
Shanghai, China, 10-13
Diana, G, Falco, M., Bruni S., Cigada, A., Larose, G.L., Davenport, A.G., Collina, A. 1995.
Comparison between wind tunnel tests on a full aeroelastic model of the proposed bridge over
Stretto Di Messina and numerical results, Journal of Wind Engineering and Industrial Aerodynamics, 54/55, 101-113.
Ge, Y. J., and Tanaka, H. 2000. Aerodynamic analysis of cable-supported bridge by multi-mode
and full-mode approaches. Journal of Wind Engineering and Industrial Aerodynamics, 86,
123-153
Katsuchi, H., Jones, N. P., and Scanlan, R. H. 1999. Multimode coupled flutter and buffeting
analysis of the Akashi-Kaikyo Bridge, Journal of Structural Engineering, ASCE, 125, 60-70
State-key Laboratory for Disaster Reduction in Civil Engineering (SLDRCE) 2002. Investigations on wind-resistant behavior of Sutong cable-stayed bridge-sectional model testing. Research Report, Bulletin of Laboratory of Wind Tunnel, Tongji University, Shanghai, China (in
Chinese).

You might also like