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

net/publication/268402722

A Study on understanding of Coupled Flutter of Long-span Bridges

Article

CITATIONS READS

0 46

3 authors, including:

Hiroshi Katsuchi Pham Hoang Kien

Yokohama National University University of Transport and Communications
81 PUBLICATIONS   443 CITATIONS    6 PUBLICATIONS   0 CITATIONS   

SEE PROFILE SEE PROFILE

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

Modeling of aerodynamic loads on long span bridges by Volterra Series. View project

All content following this page was uploaded by Hiroshi Katsuchi on 16 September 2015.

The user has requested enhancement of the downloaded file.

A Study on understanding of Coupled Flutter of
Long-span Bridges
Hitoshi Yamada, Hiroshi Katsuchi and Pham Hoang Kien

Yokohama National University, Yokohama 240-8501, Japan

ABSTRACT: Flutter analysis has come to be used widely at the design stage of long-span
bridges. This is attributed to the study and the understand of nature of unsteady aerodynamic
forces, mechanism of mode-coupling and methodology of multi-mode flutter analysis. One of
the great contributions will be the study on the Akashi Kaikyo Bridge. A large scale
aeroelastic-model wind-tunnel test together with multi-mode coupled flutter was carried out
and various new findings were brought out. In this paper, methodology of flutter analysis is
first presented and then multi-mode coupling on flutter understood so far is reviewed based on
the results of the wind-tunnel test of the Akashi Kaikyo Bridge.

KEYWORDS: Flutter analysis, Mode coupling, Multi mode, Akashi Kaikyo Bridge

1 INTRODUCTION
Flutter is one of the utmost concerns for wind-resistant design of a long-span bridge.
Wind-tunnel test is typically employed for the purpose of investigation on wind-induced
response and decision of the final cross section. For such a purpose, a 2D section model test
reproducing the deck cross section is often employed. It has some advantages that a relatively
large scaled model can be used, it does not need a large wind tunnel nor does it cost too much.
However, in the case of a very long-span bridge such as the Akashi Kaikyo Bridge, 3D
characteristics prevail so that a large size aeroelastic full model test is required. Since a
full-model test required a large wind tunnel and cost much, an analytical prediction method of
flutter, what is called “flutter analysis”, was developed [1 and 2].
Because flutter analysis applies sectional unsteady aerodynamic forces to a 3D FE model
of a bridge, once sectional unsteady aerodynamic forces are measured, it very flexibly
predicts 3D response of the bridge numerically. It is very advantageous that flutter analysis
does not need a large scale full-model test but it yields much information quantitatively on
flutter.
When the Akashi Kaikyo Bridge was designed, a large scale aeroelastic model test and
comparative flutter analysis was carried out. Owing to that, the wind-resistant safety of the
Akashi Kaikyo Bridge was successfully secured. In addition, characteristics of coupled flutter
of a long-span suspension bridge were quantitatively elucidated [3 -5].
This paper presents first methodology of flutter analysis and then characteristics of
coupled flutter of a long-span bridge obtained from the large wind-tunnel test project of the
Akashi Kaikyo Bridge.

2 METHODOLOGY OF FLUTTER ANALYSIS

Since flutter is one of the vibration modes at the flutter onset wind speed, it can be obtained
by eigenvalue analysis for the equation of motion to which unsteady aerodynamic forces are
exerted. The unsteady aerodynamic force is expressed via non-dimensional flutter derivatives
as in Eq. (1). Multiplying by 2π, these are converted to the well-known flutter derivative
notation by Scanlan [2].

−141−
 ⎡ z z& y y& θ& ⎤
L = πρb 3 ω 2 ⎢ L ZR + L zI + L yR + L yI + LθR θ + LθI ⎥ (1a)
⎣ b bω b bω ω⎦
⎡ z z& y y& θ& ⎤
D = πρb 3 ω 2 ⎢ D zR + D zI + D yR + D yI + DθR θ + DθI ⎥ (1b)
⎣ b bω b bω ω⎦
⎡ z z& y y& θ& ⎤
M = πρb 4 ω 2 ⎢ M zR + M zI + M yR + M yI + M θR θ + M θI ⎥ (1c)
⎣ b bω b bω ω⎦

where L, D and M are unsteady aerodynamic lift, drag and pitching moment, respectively, ρ is
air density, B is deck width, K (= ω B/U) is reduced frequency, ω is circular frequency, U is
mean wind speed. LzR LzI, …, MθI are called unsteady aerodynamic force coefficients or flutter
derivatives of the deck cross section which are a function of K.
Then, aerodynamically influenced equation of motion can be written as
&& + C′u& + K ′u = 0
Mu (2)
where C′ = C − Fv , K ′ = K − Fd , and M, C and K are mass, damping and stiffness matrices,
respectively. u is a displacement vector. Fv and Fd are motion-dependent unsteady
aerodynamic force matrices associated with velocity and displacement, respectively..
Applying modal analysis and assuming a sinusoidal motion for the generalized coordinate
vector X,
X(t ) = X exp(λt ) ， λ = λR + iλI (3)
a flutter condition can be deduced as follows:
~ ~ ~
(
det λ2 M + λC + K = 0 ) (4)
~ ~ ~
where M , C and K are a generalized mass, damping and stiffness matrices, respectively.
Eq. (4) can be solved as a complex eigenvalue problem in which the circular frequency ω
and damping ratio to critical ζ can be calculated from the eigenvalue λ（= λR + i λI） obtained
(Fig. 1(1)).
ω = λ2R + λ2I , ζ = λR λ2R + λ2I (5), (6)
Instead of the eigenvalue analysis, the flutter condition can be obtained by directly
solving Eq. (4) [6]. That is, the flutter condition is given at the point that the determinant
vanishes. Since the determinant equation is complex and includes two unknown variables, K
and ω, the condition must be satisfied requiring that both the real and imaginary parts of the
determinant are simultaneously zero. Then, the flutter frequency is obtained from ω and the
flutter speed can be calculated from K (= ωB/U) and ω (Fig. 1(2)).
Original Cross Section, α = 0 deg.
0deg
0.6 1.20
Circular Frequency: ω

1.00 Mode 13
Damping in log

Mode 12
0.4 0.80 Mode 11
0.60 Mode 10
0.2
0.40 Mode 2
0.20 Real Roots Mode 1
0
Imaginary Roots
0 50 100 150 200 0.00
-0.2 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Wind Speed (m/s) Reduced Frequency: K

(1) Aerodynamic damping vs. wind speed (2) Determination of flutter condition
Figure 1 Output example of flutter analysis by modal analysis

−142−
 The procedure described above has an advantage of shorter calculation time because of
application of modal analysis. However, the key to success of the analysis is inclusion of
adequate number of modes. Therefore, unlike the modal analysis method, a direct flutter
analysis method in which an eigenvalue problem is directly applied to Eq. (2) was developed
[7]. Previous studies showed that the flutter analysis by the modal method with adequate
number of modes coincided with the direct flutter analysis (Fig. 2).

Figure 2 Change of aerodynamic damping by Direct Flutter Analysis

3 COUPLE FLUTTER OF LONG-SPAN SUSPENSION BRIDGE

When the aeroelastic model test of the Akashi Kaikyo Bridge was carried out, characteristics
of coupled flutter of a long-span suspension bridge were examined by flutter analysis. Firstly,
analytical reproduction of flutter observed in the test was tried as shown in Fig. 3. Using only
unsteady aerodynamic lift and pitching moment associated with vertical and torsional degrees
of freedom in flutter analysis was common understanding at that time, however the analysis
could not reproduce the model response. Finally, inclusion unsteady aerodynamic drag (Pi*)
could successfully reproduce the model response. This is one of the notable findings at the
aeroelastic-model test of the Akashi Kaikyo Bridge. However, it was concluded that this was
significant in the case of a long-span truss-stiffened deck such as the Akashi Kaikyo Bridge.
Large torsion of the deck due to wind loading made effects of Pi* significant [3-5].
Mode coupling was investigated by flutter analysis. As described earlier, mode selection
is the key factor in the flutter analysis with modal analysis. Effects of difference in the
number of modes included on flutter were investigated as shown in Fig. 3. It shows that six
modes (1st symmetric lateral [1], vertical [2] and torsion [12], 2nd symmetric vertical [10] and
lateral [11], and 3rd symmetric vertical [13]) can almost replace the 25-mode analysis (1-25
modes) [4 and 5].

Changes of aerodynamic damping for critical mode

1.5
●: Experiment
Damping Ratio: ζ (%)

1.0

0.5 ○: Analysis
*
（with P i , i = 2, 3, 5 & 6）
0.0 △: Analysis
*
（without P i , i = 2, 3, 5 & 6）
-0.5
0 20 40 60 80 100
Wind Speed: U (m/s)
Figure 3 Comparison between flutter analysis and aeroelastic-model test of Akashi-Kaikyo Bridge

−143−
 Table 1 shows that mode contributions of the six modes to flutter. Mode contribution
ratio and change of flutter speed are shown. It can be seen that the contributions of modes 1
and 11 are small, and mode 12 is essential for flutter; that is, flutter did not occur when mode
12 was excluded. However flutter ultimately occurred even when mode 2 was excluded. The
flutter speed fluctuates among the cases but remains close to the six-mode case. It is
noteworthy that even the two-mode case (modes 2 and 12), which is classical theory, gave a
result very close to the six-mode case. Modes 10 and 11 played a destabilizing role while
mode 13 stabilized the system. It is also noteworthy that no significant differences in the
eigenvectors can be seen among different mode combinations [4 and 5].
1.0
Magnitude of ξi vector

0.8 Flutter Speed

25-Mode : 94.1 m/s
0.6 6-Mode : 95.8 m/s

25-Mode Analysis
0.4 6-Mode Analysis

0.2

0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Mode
Figure 4 Comparison of mode participation between 25- and 6-mode analyses

Table 1 Mode Contribution to Flutter

Mode Flutter Speed
Case 1 2 10 11 12 13 (m/s)
1 ― 0.270 ― ― 1.0 ― 82.0
2 ― 0.233 0.191 ― 1.0 ― 76.3
3 ― 0.223 0.183 0.050 1.0 ― 74.9
4 ― 0.256 0.217 0.069 1.0 0.234 80.4
5 0.069 0.245 0.224 0.064 1.0 0.244 79.1

4 CONCLUDING REMARKS
Methodology of flutter analysis and mode coupling on flutter was briefly reviewed. Owing to
numerical method of coupled flutter instead of wind-tunnel test, a great deal of information on
coupled flutter can be obtained. This can make us understand better mechanism of coupled
flutter of a long-span bridge.

5 REFERENCES
1 Bleich, F., McCullough, C. B., Rosecrans, R. and Vincent, G. S., The Mathematical Theory of Vibration in
Suspension Bridges, Bureau of Public Roads, Dept. of Commerce, USA, 1950.
2 Scanlan, R.H., The Action of Flexible Bridges under Wind, I: Flutter Theory, J. of Sound and Vibration, 60(2),
1978, pp.187-199.
3 Miyata, T., Tada, K., Sato, H., Katsuchi, H. and Hikami, Y., New Findings of Coupled-flutter in Full Model
Wind Tunnel Tests on the Akashi Kaikyo Bridge, Proc. of Cable-Stayed and Suspension Bridges, Deauville,
France, AFPC, Vol.2, 1994, pp.163-170.
4 Katsuchi, H., Jones, N. P., Scanlan, H. R. and Akiyama, H. : A Study of Mode Coupling in Flutter and
Buffeting of the Akashi-Kaikyo Bridge, Structural Eng./Earthquake Eng., JSCE, Vol.15, No.2, 1998,
pp.175s-190s.
5 Katsuchi, H., Jones, P. N. and Scanlan, H. R. : Multi-Mode Coupled Flutter and Buffeting Analysis of the
Akashi-Kaikyo Bridge, J. of Struct. Engrg., ASCE, Vol.125, No.1, 1999, pp60-70.
6 Jain, A., Jones, N. P. and Scanlan, R. H., Coupled Flutter and Buffeting Analysis of Long-span Bridges, J. of
ST, ASCE, Vol. 122, No.7, 1996, pp.716-725.
7 Miyata, T. and Yamada, H., Coupled flutter estimate of a suspension bridge, Proc. of International Colloquium
on Bluff Body Aerodynamics and its Application, Kyoto, 1988, pp.485-492.

−144−

View publication stats