Aerodynamic stability of self-anchored double deck suspension bridge

JOURNALOF

-',de; ' ;,'Dtl

i~I~iI~i~I Journal of Wind Engineering and Industrial Aerodynamics54/55 (1995) 25-34

ELSEVIER

Aerodynamic stability of self-anchored double deck suspension bridge S.D. K w o n *'a, S.P. C h a n g a, Y.S. K i m b, S.Y. P a r k c aDepartment of Civil Engineering, Seoul National University, Kwanak-Gu, Seoul, South Korea bHyundai Construction Co., Sejong-Ro 178, Chongro-Gu, Seoul, South Korea c Yooshin Engineering Co., Yoksam-Dong 832-40, Kangnam-Gu, Seoul, South Korea

Abstract In this paper wind tunnel test results and their interpretations, which were performed to study the aerodynamic stability of a self-anchored double deck suspension bridge, are presented. Section, tower and full models were tested under smooth and turbulent flow conditions. Even though the drag coefficient of the girder had high value, the amplitude of the lateral vibration was found to be very low. This may be due to the restraint provided by the horizontal curvature of the cables. Because the two natural frequencies of the tower, out-ofplane bending and torsional, were not well separated, coupled motions were observed in a wide range of wind velocity. It has to be emphasized that the vibration characteristics of the tower in the self-anchored suspension bridge may be very sensitive to the longitudinal boundary conditions of the girder at the supports. The effectiveness of corner cut, countermeasure to reduce the tower vibrations, was also studied.

1. Introduction Young-Jong bridge (provisional name) is now under construction at the western seaside of Korea. After completion, it will be a major part of the link connecting the new international airport at Young-Jong island with Inchun city. The total length of the bridge is 4420 m. Most of the spans are designed as steel truss structures with orthotropic deck. But there are three spans that are designed as a suspension bridge, with a center span 300 m long and two side spans 125 m long as shown in Fig. 1. This suspension bridge has several new features. The main girder of the suspension bridge as shown in Fig. 2 is a truss structure with double deck. The upper deck accommodates six traffic lanes and the lower deck four traffic lanes, along with two

* Corresponding author. 0167-6105/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0 1 6 7 - 6 1 0 5 ( 9 4 ) 0 0 0 2 6 - A

26

S.D. Kwon et al./J. Wind Eng. lnd. Aerodyn. 54/55 (1995) 25-34 5~ I0 @ 11.5

2.1 @

r ~

~ j ~

] '~__~__'+

,

--

I0 0 IZS

12,5

~%,~ ~LI,~ ~i~, ~

' ' + ++" 2oq'

'

+

P~"+;' "~

. 111 .

i

.

.

.

Fig. 1. Generalarrangement of the Young-JongBridge.

lu

"

ira,

Fig.

2. Deck

cross-section

and

tower.

railroad tracks. The main cables are anchored at the top steel deck directly without any anchorage block. The horizontal tension force of the cable is resisted by the axial stiffness of the girder. In ordinary suspension bridges, the cables have only a vertical curvature. In the Young-Jong bridge, however, the cables have both vertical and horizontal curvature.

S.D. Kwon et aL/J. Wind Eng. Ind. Aerodyn. 54/55 (1995) 25-34

27

Table 1 Natural frequencyand damping ratio of the free standing tower model Mode

out-of-plane bending in-plane bending torsional

Computed prototype frequency(Hz) 0.343 0.909 1.770

Measuredmodel frequency(Hz) (scaled)

Measureddamping ratio (% critical)

0.350 0.903 1.666

0.22 0.45 0.17

Fig. 3. Full bridge aeroelasticmodel. A series of wind tunnel tests were performed to investigate the aerodynamic stability and the serviceability of the girder under various wind conditions. The results of wind tunnel tests are presented along with the effectiveness of countermeasures employed for the reduction of the tower vibrations.

2. Test models A section model was employed in the test for the measurement of aerostatic coefficients and in the spring mounted model test. The geometric scale of the model was 1:100. An aeroelastic model of the tower with a 1:150 geometric scale was manufactured for free standing stage and completion stage tests. Natural frequencies and damping ratios of the free standing tower model are provided in Table 1. In the completion stage test, two coil springs were attached to the top of the tower for the simulation of restraint effects provided by the main suspension cables. A 5.5 m long full aeroelastic model of the bridge (Fig. 3) was constructed faithfully to the prototype. A scale factor of 1:100 was decided for convenience of transportation

28

S.D. Kwon et al./J.. Wind Eng. lnd. Aerodyn. 54/55 (1995) 25-34

Table 2 Natural frequency and damping ratio of the full bridge aeroelastic model Mode

Computed prototype frequency (Hz)

Measuredmodel frequency (Hz) (scaled)

Measured damping ratio (% critical)

vertical symmetric 1 vertical antisymmetric 1 torsional symmetric 1

0.454 0.750 1.223

0.479 0.781 1.221

0.49 0.21 0.47

and for the safety of the model. As can be seen in Table 2, measured frequencies of the full model were in good agreement with the calculated ones.

3. Experimental procedure Section and tower model tests were performed in the G6ttingen type wind tunnel at the Seoul National University. The test section is 1.37 m (W) x 0.95 m (H). A wind tunnel of the University of T o k y o was used for the full aeroelastic model tests. The test section of the wind tunnel is 16 m (W) × 1.9 m (H). Vibration responses were measured using a pair of noncontact optical displacement transducers. T w o kinds of turbulent flow were generated by two different types of wooden grid. The measured longitudinal turbulent intensities were 12% and 18%, respectively. Oil damper was used to adjust the damping ratio of the section model. Static wind forces of the girder under smooth flow were measured by force balance. The spring mounted section model was tested with an angle of attack varying from - 7 ° to 7 ° by steps of 1° to investigate directly the aerodynamic instability phenomena. Flutter derivatives were also derived from free oscillation tests for further study. Tower model tests were performed with the two different models; one for free standing and the other for completion stage. The model was exposed to two types of flow, smooth and turbulent, as in the section model test. The tests were repeated with various horizontal wind angles ranging from 0 ° to 90 °. The full aeroelastic model was tested under both smooth and turbulent flow. Vertical, torsional and lateral vibration responses of the girder were measured at half and quarter point of the main span. Bending and torsional vibration responses were measured at the mid height of the tower upper legs.

4. Response of the girder Spring mounted model tests were performed with the original section model and several versions of modified section. Based on the test results, a section which shows the best performance was selected as final section. Then aerostatic coefficients were measured with the selected section.

S.D. Kwon et al./J. Wind Eng. Ind..4erodyn. 54/55 0995) 25-34

29

80

- ~ odginal (vertical) original (torsion) | F 60modified (torsion) 1

~

/

"~ 40-

/,

/

o

!

II't

\-_

;I

/

20

"'..... ,... '%--~.....

Wind Angle Bou~lary of Severn I~clge ........-" .................. '"

0 -8

P -6

i

I -4

i -2

J 0

I 2

Angle of Attack (deg) Fig. 4. Vortex-induced vibration of the girder (section model test).

4

2

CD

0.4

CL

gill

ml g41 nil

CM

I

,

j~4 I*

01

" _ c

-2 '

-16

~

-8

I

0

D

,-

8

0

--

-0.2

16

Angle of Attack (deg) Fig. 5. Aerostatic coefficients of the girder.

In the spring mounted model tests, catastrophic instability phenomena were not observed up to the wind velocity of 100 m/s. However vertical and torsional vortex-induced vibrations occurred for the original section at a wind velocity less than the design velocity of 56 m/s as can be seen in Fig. 4. In order to stabilize vortex-induced vibration, the original section was modified by attaching fairing and spoiler to the upper deck of the model. In the section model test with the modified section, only vortex-induced torsional vibrations were observed occurring at a high attack angle as shown in Fig. 4. Shown in Fig. 5 are measured aerostatic coefficients of the girder obtained with the modified section model. The drag coefficient at the attack angle of 0 ° seems to be similar to that in the same type of double deck truss girder. But the gradient of the lift coefficient at 0 ° was considerably higher than in the usual case,

30

S.D. Kwon et al./J. Wind Eng. Ind. Aerodyn. 54/55 (1995) 25-34 1.5

I

-~-

1.2

--

measured (peak-peak) I computed (peak-peak) . . . .

E "" 0.9 "O

~. 0.6 E .< 0.3

.,~/" ° oo ooo~

eea I

b

20 40 Velocity (m/s)

60

Fig. 6. Buffeting response at ½point of the center span.

Table 3 Structural properties of the bridge Property

Prototype

Model

mass per unit length mass moment of inertia per unit length

45.4 t/m 7068.0t m2/m

4.54 Kg/m 7.09 Kg cmZ/cm

In the full model test performed under smooth flow with velocity less than 70 m/s, no harmful vibrations of the girder were observed. Because of the sharp gradient of the lift coefficient, excessive vertical oscillations under turbulent flow were expected. Nevertheless, the test results showed that the angle of attack does not affect the magnitude of the response significantly. Buffeting response was investigated to check the serviceability of the bridge but it was found to be acceptable from the engineering point of view. In Fig. 6, it can be seen that the measured buffeting response in the middle of the center span is in good agreement with the computed one. Scanlan's method was used to compute the buffeting response. F r o m a separate analytical study, it was found that the longitudinal boundary condition at the supports of the girder does not affect buffeting response even in the self-anchored suspension bridge. The horizontal rms response of the center span under turbulent flow was only 5 cm at the design wind speed. This may be attributed to the restraint of the horizontal movement due to the horizontal curvature of the cables.

5. Response of the tower The response characteristics of the tower were investigated in both tower model and full model tests. Tower model tests were performed with two different models; one

S.D. Kwon et al./J. Wind Eng. lnd. Aerodyn. 54/55 (1995) 25-34

e

31

out-of-plane bending (peak-peak)

•-~,.- torsion (peak-peak)

I ~

~ 0.8

3 t~ "O

__=

0.6 E <

2

E <

1

._o

,,..~ ~" m

/

0.2

0 F--

00

80

Velocity (m/s) Fig. 7. Response of the free standing tower under smooth flow.

-4

1.2 -e-

•

vE

out-of-plane bending

tor, iona,

O~ "10

3 v

• 0.9 "O

O

E o.e

<

<

c

t.-

~ 0.3

1 ._o

IXI

O I--

0

0

10

20

30

40

0 50

Wind Angle (deg)

Fig. 8. Relation between wind angle and response of the free standing tower. simulating free standing stage and the other the completion stage. The velocity ratios of the model and prototype were derived from Froude number scaling. Results of free standing tower tests performed in smooth flow are presented in Figs. 7 and 8. Galloping was not observed within design velocity. Bending and torsional responses in function of wind velocity are given in Fig. 7. It was observed that the bending mode is dominant at low velocity, then the torsional mode dominates near the design velocity. There were sharply defined velocity bands where each mode was excited. It was observed that 1.0% of the structural damping ratio might be sufficient to suppress the bending vibrations. Fig. 8 shows the relation between horizontal wind angle and vibration amplitude. The torsional vibration amplitude of the tower takes its maximum value when wind blows perpendicular to the bridge axis. But the maximum response of out-of-plane bending vibrations takes place at the horizontal wind angle of 5 °.

32

S.D. Kwon et al./J. Wind Eng. Ind. Aerodyn. 54/55 (1995) 25-34 2.5 .,...."

2 "~

vz

....J ........................................................... ~t ..'~'~

1.5

O c(1) "~ O"

ii

i iiiiiii ii iiii i

-

/

_hi!!~!! in~_d.~!), iii

bending (full model)

1

LL

~

0.5

-

~ ~

bending

-

.......... torsional designed

0

1E0

I

~

I

~

I

=

I

1E1

1E2

1E3

1E4

1E5

1E6

1E7

1E8

Spring Constant (t/m) Fig. 9. Frequency of the tower model (completion stage).

.......... tower model (peak-peak) 0.8

1

full model (peak-peak)

E

bending

O.6

torsional

i

-o.4E <

0.2

"-'"/ ' " • ...............I~

0

......

.... ....................... bending![ torsional . . . . . . . . . .

20

40

60

80

100

Velocity (m/s) Fig. 10. Tower response of the full model and tower model (completion stage).

To simulate the completion stage using the tower model, two coil springs were attached to the tower top. Fig. 9 shows the relation between computed natural frequencies of bending and torsional mode and spring constant. As the spring constant increases, the natural frequency of the bending mode increases rapidly and the mode shape changes to the 2nd mode. But the natural frequency of the torsional mode is not much affected. This indicates that the natural frequency of the bending mode of the tower in a self-anchored suspension bridge may be very sensitive to the longitudinal boundary conditions. It has to be pointed out that the two natural frequencies, bending and torsional, are closely spaced in both the completion stage tower model and full model. In Fig. 10, the tower response from the full aeroelastic model is compared with the tower model simulating the completion stage. Both tests were performed under smooth flow conditions. In the tower model test, bending mode vibrations are

S.D. Kwon et aL/J. Wind Eng. Ind. Aerodyn. 54/55 (1995) 25-34 0.5 vE

.....".,.

0.4

"0

==

\

1.5

1

q It

"''0

-B- bending(peak-peak) / -B- bending(RMS) / -0" torsional(peak-peak) /

................ .! ......to.on.. s S

0.3

33

...Ij....... • " ''J

-1.2 "O

0.9

O.

E

<¢

E

0.2

0.6 <

0.1

0.3 "~

¢-

e--

"0

o

i

0.0 0.0

t 0.1

.......... ~ . . . . ~ d l .

~.£........., ............

012 0'3 01,

O

F-

018 07

Corner Gut Size (m) Fig. 11. Relation between corner cut size and amplitude (full model).

initiated at low velocity and gradually change to torsional vibrations without clear separation. The gradual transition of mode with wind velocity is also observed in full model test results. In the velocity zone where the gradual mode transition occurs, the tower response is characterized by a coupled motion of two modes. It may be caused because the two natural frequencies are not well separated as mentioned before. The main difference in the responses of the tower model and full model is the shifting of the velocity zone where a particular mode is excited. This shifting of velocity zone may be mainly due to the improper modeling of spring properties and partly due to model imperfection. In the full model test performed under turbulent flow, the vibration of the tower was governed by the bending mode in contrast to the coupled motion under smooth flow. Large vertical vibration of the girder might be one of the causes of preventing the torsional mode from being excited. The amplitude of the tower vibration may be mitigated by aerodynamic or/and mechanical devices. In this study, both measures were investigated to reduce vibration of the tower under smooth flow. Corner rounding and attachment of aerodynamic devices such as a deflector attached to the column of the tower were tried to suppress vibration, but there was little reduction in amplitude. Finally, corner cut was attempted and Fig. 11 shows the results. When the ratio of the cutting size to width of the column was around 1:10, the amplitude of the bending and torsional vibration were reduced respectively, by 50% and 60%. But a cutting ratio larger than 1:10 aggravated the bending vibration. Fig. 12 shows the relation between structural damping and peak amplitude. As can be seen, high damping may be needed to suppress especially the torsional response.

6. Concluding remarks The aerodynamic stability of the Young-Jong bridge has been studied by a series of wind tunnel tests. The shape of the girder which has the best aerodynamic stability

34

S.D. Kwon et al./J. Wind Eng. Ind. Aerodyn. 54/55 (1995) 25-34 • 0.3 [

E

•

bending(peak-peak)]

-~-

Iorsional (peak-peak) ]

v

1

-0.8

"O

v

e',

0.6

E 0.2

<

E

<

t¢.. ~

!.

'10 t'"

"....

n"t

"... ....~.

0.4 .o_ P O

.....

t---

0.1

0.2 0

1

2

3

4

5

Damping Ratio (% critical) Fig. 12. Effect of structural damping and amplitude (full model).

was selected based on the section model test. Good performance of the selected section was confirmed in the full bridge aeroelastic model test. In the tower test simulating completion stage and in the full model test, the velocity zone where the bending mode was excited was not well separated from that of the torsional mode. Tower vibration may be reduced by either cutting the corners of the tower legs or by increasing structural damping. It was observed that the longitudinal boundary condition at the supports of the girder has great influence on the vibration mode of the tower in the case of the present self-anchored suspension bridge. Since the actual longitudinal boundary conditions could not be known in advance, the vibration characteristics of the tower and the onset wind velocity may not be predicted accurately. Therefore the uncertainties mentioned above may have to be considered in the design of the bridge.

Acknowledgement The authors would like to express their deepest gratitude to Professor Y. Fujino and the members of the Bridge Engineering Laboratory of the University of Tokyo for allowing using the large wind tunnel and for helping during the experiment.

References [1] Honshu-Shikoku Bridge Authority, Wind tunnel testing for Akashi Straits Bridge (1990). [2] R.H. Scanlan, State-of-the-art methods for calculating flutter, vortex-induced and buffeting response of bridge, Federal Highway Administration, FHWA/RD-80/050 (1981). [3] Japan Highway Association, Wind resistant design for Highway Bridge (1991). [4] K. Okawaet al., Aerodynamic stability of tower ofalong-spannedcabte-stayedbridge, in:Proc. 1st Int. Colloq. on Bluff body aerodynamics and its applications, Kyoto, 1988, pp. 501-510.