You are on page 1of 11

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

net/publication/359115220

Wind-Induced Instability Mechanism of Old Tacoma Narrows Bridge from


Aerodynamic Work Perspective

Preprint · March 2022


DOI: 10.1061/(ASCE)BE.1943-5592.0001858

CITATIONS READS
0 59

3 authors:

hu Chuanxin Lin Zhao


Tongji University Tongji University
12 PUBLICATIONS   52 CITATIONS    84 PUBLICATIONS   541 CITATIONS   

SEE PROFILE SEE PROFILE

Yaojun Ge
Tongji University
145 PUBLICATIONS   1,172 CITATIONS   

SEE PROFILE

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

Wind-induced Fragility Analysis of Long-span Bridges View project

Mechanical analysis and experiments on transmission lines, wind loads and ice loads acting on transmission towers and conductors View project

All content following this page was uploaded by hu Chuanxin on 09 March 2022.

The user has requested enhancement of the downloaded file.


Wind-Induced Instability Mechanism of Old Tacoma Narrows
Bridge from Aerodynamic Work Perspective
Chuanxin Hu1; Lin Zhao2; and Yaojun Ge3

Abstract: Postflutter limit cycle oscillations (LCOs) are typical nonlinear aeroelastic phenomena for bridge girders. An H-shaped section
with a roughly 5:1 aspect ratio, treated as a simplified section of the Old Tacoma Narrows Bridge, was chosen for wind-induced instability re-
evaluation, considering its vibration amplitude-dependent aerodynamics characteristics and nonlinear structural damping effects from the
energy perspective of aerodynamic work. Forced vibrations at large torsional amplitudes in a wind tunnel were realized with the help of
an improved forced motion apparatus (FMA), and synchronous measurements of forces and displacements on the FMA were achieved.
Self-excited forces (SEFs) were extracted, and an energy map showing quantitative relationships between vibration amplitude, reduced ve-
locity, and aerodynamic work acting on the section were established. Furthermore, the postflutter LCOs phenomena originating from the
energy balance between nonlinear aerodynamic work input and energy consumption by structural damping effects were reillustrated, and
nonlinear structural damping effects of prototype bridges are therefore discussed. Moreover, the instability paths of the bridge were inves-
tigated by the energy map, considering its vibration amplitude-dependent aerodynamics and structural damping effects from an aerodynamic
work perspective. The results show that LCO velocity basically increases with vibration amplitude at different structural damping ratios
in torsional degree of freedom, characterizing postflutter LCOs phenomena. The structural damping ratio at the bridge’s collapse is also
re-estimated as about 0.0115 with better coincidence of on-the-spot observation and theoretical analysis. In addition, both the aerodynamic
damping ratio considering nonlinear characteristics of the SEFs and the structural damping ratio basically increasing with vibration amplitude
contribute to postflutter LCOs phenomena of the Tacoma Bridge. The instability paths of the bridge essentially rely on the competitive rela-
tionships between structural damping ratios, structural stiffness degeneration, and oncoming wind velocities in the process of collapse.
DOI: 10.1061/(ASCE)BE.1943-5592.0001858. © 2022 American Society of Civil Engineers.
Author keywords: Old Tacoma Narrows Bridge; Nonlinear aerodynamics; Nonlinear damping effects; Limit cycle oscillation;
Wind tunnel tests.

Introduction torsional vibration for about 70 min (Ammann et al. 1941), which
can be referred to as a typical nonlinear postflutter limit cycle oscil-
Since the collapse of Old Tacoma Narrows Bridge in 1940, the lations (LCOs) phenomena. Second, scholars have not drawn con-
aerodynamic and aeroelastic behavior of H-shaped sections have sistent conclusions on the actual structural damping ratio in torsion,
attracted persistent attention. It is believed that one degree of free- let alone its variations in the process of collapse, which is largely
dom (DOF) torsional flutter was responsible for the collapse responsible for the discrepancies in the estimation of onset flutter
(Scanlan and Tomko 1971; Larsen 2000). Although the H-shaped velocities of the bridge (Larsen and Walther 1997; Matsumoto
section has been the hotspot with the aim of improving understand- et al. 2003). Thus, it is necessary to revisit the collapse of the
ing of the aerodynamic behavior of bluff sections and the collapse Tacoma Bridge in the collapse process with current updated exper-
of the Tacoma Bridge, many controversial issues still remain. imental techniques and deepen understanding of the wind-induced
First, the collapse cannot be defined as a divergent type of flutter instability mechanism of the Tacoma Bridge, highlighting varia-
predicted by the linear small-amplitude assumption (Gao et al. tions of nonlinear aerodynamic loads and structural damping ratios.
2018); in fact, the bridge underwent large amplitude of antisymmetric Forced motion apparatus (FMA) wind tunnel tests have been
comprehensively employed to study the aerodynamic effects and
1
wind-induced performance of long-span bridges. In addition to
Assistant Professor, Dept. of Civil Engineering, Wuhan Univ. of Science simulating and reproducing unsteady structural vibration with
and Technology, Wuhan 430065, China. Email: chuanxinhoo@126.com
2 large amplitudes, large wind angles of attack, and multimodal
Professor, State Key Lab of Disaster Reduction in Civil Engineering,
Tongji Univ., Shanghai 200092, China; Key Laboratory of Transport In- coupling effects, the forced vibration method has several other ad-
dustry of Wind Resistant Technology for Bridge Structures, Tongji vantages: (1) easily realizing a wider range of reduced wind veloc-
Univ., Shanghai 200092, China (corresponding author). Email: zhaolin@ ities; (2) obtaining a higher signal-to-noise ratio of test signals;
tongji.edu.cn (3) directly measuring the self-excited forces (SEFs) acting on
3
Professor, State Key Lab of Disaster Reduction in Civil Engineering, the model and easily identifying aerodynamic parameters (Zhao
Tongji Univ., Shanghai 200092, China; Key Laboratory of Transport In- et al. 2020); and (4) avoiding additional wind-induced angle of at-
dustry of Wind Resistant Technology for Bridge Structures, Tongji tack, nonlinear stiffness, and nonlinear damping ratio introduced by
Univ., Shanghai 200092, China. Email: yaojunge@tongji.edu.cn
free vibration systems. Ukeguchi et al. (1966) first applied an FMA
Note. This manuscript was submitted on July 22, 2021; approved on
January 7, 2022No Epub Date. Discussion period open until 0, 0; to investigate aerodynamic characteristics of a bridge section. Since
separate discussions must be submitted for individual papers. This then, FMAs have been used for research on aerodynamic character-
paper is part of the Journal of Bridge Engineering, © ASCE, istics of bridge decks or bluff sections (Ukeguchi et al. 1966;
ISSN 1084-0702. Otsuki et al. 1974; Falco et al. 1992; Matsumoto et al. 1993;

© ASCE 1 J. Bridge Eng.


Fig. 1. Geometrical sizes of reduced-scale bridge girder of the Tacoma Bridge (unit: mm).

Li 1996; Cigada et al. 2001; Diana et al. 2004; Li et al. 2016; section to the Tacoma Bridge. Therefore, the fluid structure interac-
Siedziako et al. 2017). tion can be investigated for a simplified cross section, that is, an
Recently, postflutter LCOs, also termed “nonlinear flutter”, H-shaped section, and in smooth flows, ignoring turbulence effects
have attracted increasing attention (Ge et al. 2018; Tang et al. (Nakamura and Nakashima 1986; Matsumoto 1999; Larsen 2000).
2019; Zhang et al. 2019; Zhou et al. 2019; Gao et al. 2020a, b; Based on this, an H-shaped section with an aspect ratio of roughly
Li et al. 2020; Wu et al. 2020; Zhang et al. 2020a, b). Considerable 5:1 under smooth flow conditions is considered, and a novel method-
research demonstrates that bridge girders may perform LCOs ology for estimation of a structural damping ratio for prototype brid-
after the critical state due to aerodynamic and/or structural nonlin- ges is proposed. Furthermore, both the structural damping ratio and
earities, referred to as postflutter LCOs. Unlike the LCO of vortex- the aerodynamic damping ratio in torsion are highlighted in the for-
induced vibrations (VIVs) observed in a lock-in region at a mation of the postflutter LCOs phenomena.
relatively lower and narrower wind velocity range (Hu et al. The remaining sections are arranged as follows. The section
2018, 2019, 2021), the postflutter LCO occurs in a higher and “Experimental Setups” introduces an improved FMA at Tongji
wider wind velocity range, and the LCO amplitude typically University, realizing large torsional amplitudes, and synchronous
grows continuously with increasing oncoming wind velocity measurements of forces and displacements on the FMA for an
(Zhang et al. 2019). Numerous studies have focused on the postflut- H-shaped section in TJ-3 wind tunnel. In the section “Aerodynamic
ter LCOs of bridge girders or bluff sections (Diana et al. 2004, Effects,” torsion-related SEFs are extracted, and nonlinear charac-
2008, 2010; Náprstek and Pospíšil 2011; Ge et al. 2018; Tang teristics are compared at different vibration amplitudes. Then, an
et al. 2019; Zhang et al. 2019; Zhou et al. 2019; Gao et al. energy map, showing relationships between vibration amplitude,
2020a, b; Li et al. 2020; Wu et al. 2020; Zhang et al. 2020a, b). reduced velocity, and aerodynamic work of the SEFs acting on
However, their research has several drawbacks: (1) nonlinear char- the section, are established in the section “Collapse Mechanism
acteristics of the SEFs have been highlighted, while effects of the of the Tacoma Bridge,” followed by a reconstruction of the col-
structural damping ratios on the postflutter LCOs have been ig- lapse of the Tacoma Bridge. Moreover, the structural damping
nored to some extent; (2) aerodynamic characteristics of SEFs ratio in torsion varying with vibration amplitude in the process of
under large vibration amplitude and related postflutter LCOs the collapse of the Tacoma Bridge is also estimated based on the
have not been fully revealed, mainly due to the limitation of tradi- updated methodology. Then, the instability path of the Tacoma
tional equipment in wind tunnel tests in terms of larger vibration Bridge is presented in detail, and some key influencing factors
amplitude, especially larger than 15°, regardless of the free vibra- are discussed. Finally, some conclusions are summarized in the
tion systems (Gao and Zhu 2015) and traditional FMAs (Ukeguchi final section
et al. 1966); (3) although the H-shaped sections of the simplified
bridge deck of the Tacoma Bridge have been comprehensively
taken as research objectives, there is still great difficulty in reveal- Experimental Setups
ing the nonlinear characteristics of the SEFs and related postflutter
LCOs phenomena characterizing ultra-large torsional amplitude as An improved FMA at Tongji University was applied to make mod-
high as 35° in the process of collapse, in addition to the nonlinear els vibrate in predesignated ways. The 3 DOF movement, namely
amplitude-dependent structural damping effects in that process. horizontal (X ), vertical (Y ), and torsional (RZ), are driven
Therefore, it is still necessary to illustrate the process and reveal separately. The amplitude limits of X, Y, and RZ are ±200 mm,
mechanisms of the postflutter LCOs phenomena observed in the ±100 mm, and ±360°, respectively. Moreover, the actual maximal
full-scale bridge on site, considering more factors such as involving amplitude of a suspended model depends on the capacity of the
amplitude-dependent aerodynamic loads at large torsional ampli- FMA and the mass or mass inertia moment of the suspended
tudes and related nonlinear structural damping ratios. model. Large amplitude limits facilitate research on wind-induced
The main objective of this study is to revaluate the collapse of the vibrations at large amplitude and nonlinear aerodynamics behavior
Tacoma Bridge with the help of an improved FMA, enabling large for bridge sections under extreme wind conditions. More details
torsional amplitude on the viewpoint of aerodynamic work and en- about the novel FMA are referenced in Zhao et al. (2020).
ergy concept. Particular attention is given to the energy map, consid- An H-shaped sectional model with an aspect ratio of roughly
ering nonlinear amplitude-dependent aerodynamics and structural 5:1, simplified section of the bridge deck of the Tacoma Bridge
damping effects, and then satisfactory consistent instability paths with a scale ratio of about 1:50, was selected, as shown in Fig. 1.
can be reillustrated compared with the whole collapse of the Tacoma The sectional model, composed of a web plate and two symmetrical
Bridge. Because of the complex geometrical shape of actual cross flange plates, is 0.700 m long (L), with a width (B) of 0.238 m, and
section of the Tacoma Bridge, the flow patterns around it and the re- a depth (D) of 0.048 m. The web plate is made of photosensitive
lated response are also complex. However, there are no obvious ef- resin and is 3D printed to create a honeycomb structure to form
fects of turbulence on the torsional flutter of the H-shaped section of the overall rigidity. The flange plates are made of lightweight
bluffness ratio B/D = 5 (Matsumoto et al. 1992), which is a similar wooden board. The model is designed to have sufficient stiffness

© ASCE 2 J. Bridge Eng.


Table 1. Parameters of high-precision dynamic balance
Component Fx (N) Fy (N) Fz (N) Tx (N · m) Ty (N · m) Tz (N · m)
Range ±65 ±65 ±200 ±5 ±5 ±5
Precision 0.1% 0.1% — — — 0.1%

Table 2. Parameter range of wind tunnel experiments


Property Value
Motion type Simple harmonic motion
DOFs Torsion
Mean angle of attack (°) 0
Amplitude (°) 3, 6, 9, 12, 15, 20, 25, 30, 35, 40, and 45
Frequency of forced vibrations 4.2, 2.1
(Hz)
(a) Forced vibration
Wind velocity (m/s) 2–15
Reynolds number 3.54 × 104–2.65 × 105

industrial automation, were used. Trial parameters of the high-


precision balance are listed in Table 1.
Synchronous acquisition equipment was applied to synchronize
the force signal and displacement signal. The instrument model is
the INV3062V synchronous acquisition instrument produced by
Beijing Oriental Institute of Vibration and Noise Technology.
The device is equipped with a built-in parallel sampling digital-
to-analog conversion chip, model ADS1278, produced by Texas
Instruments. The chip is able to convert the force signals (a total
of 12-channel force signals) and the displacement signals
(four-channel displacement signals) into voltage signals, synchro-
nizing both signals in this way. The maximal time delay between
(b)
any channel is about 15.6 ns. In this test, a sampling frequency
of 512 Hz was used to collect the time-varying balance-based
force and displacement signals. Thus, synchronous measurements
of balance-based forces and displacements of the model can be
achieved based on these techniques.
All experiments were conducted in TJ-3 boundary layer wind
tunnel, which is a vertically arranged low-speed closed-circuit
wind tunnel. The test section is 15.0 m wide by 2.0 m high. The
wind velocity can be adjusted continuously from 1.0 to 17.6 m/s.
It should be noted that wind velocity at the location of the sus-
pended model is calibrated, which is about 1.1 times that of the on-
coming wind velocity may be due to the Venturi effect and/or
wind-blocking effects. The model’s mass m is 0.516 kg. Only
single-frequency harmonic vibrations in torsion at 0° mean angle
of attack are considered. The Reynolds (Re) numbers are defined
as Re = UB/v, where U is wind velocity at the location of the sus-
(c) pended model and v is kinematic viscosity, ranging from 3.54 ×
104 to 2.65 × 105. The experiment parameters are listed in Table 2.
Fig. 2. Schematic diagram of FMA and measured model: (a) sectional Time-varying torsional displacement of the section prescribed
model; (b) overall view; and (c) detail view. in the forced vibrations (α(t)) are expressed as
α(t) = α0 sin (2πfv t + φ0 ) (1)
and avoid local deformation or vibration during the test. Local
views of the sectional model are shown in Fig. 2. where fv = frequency of the forced vibration; α0 = maximal ampli-
The sectional model was installed on the FMA by two tude of the forced vibration; and φ0 = initial phase lag.
supporting connectors, with two laser displacement sensors and
two high-precision dynamic balances at each end. In addition,
thin elliptical end-plates were fixed to each side of the sectional Aerodynamic Effects
model to improve the two-dimensional (2D) flow pattern, as
shown in Fig. 2. HL-C235CE-W-type laser displacement
Extraction of SEFs
transducers, produced by Panasonic, were used for displacement
measurements, with a measuring range of 350 ± 200 mm, a resolu- Owing to the limitations of the FMA acceleration capacity, the time
tion of 8 μm, and a maximal linearity error of less than ±0.04%. history of the input data is divided into a divergent segment,
High-precision balances with six-component, produced by ATI stable segment, and attenuation segment, as shown in Fig. 3(a).

© ASCE 3 J. Bridge Eng.


(a)

(b)

(c)

Fig. 3. Signal calibration by extraction of extra aerodynamic loads without wind: (a) original signal; (b) modified signal by phase lag; and (c) in-
tercepted signal.

Taking the divergent segment or attenuation segment in the dis- can be obtained point by point as
placement time history as reference segment, the input data under
wind or no-wind conditions can be moved back and forth to Mse = Md − Md0 (2)
make them overlap, realizing the synchronization of input data where Md and Md0 = dynamic force in torsional direction acting on
under wind and no-wind conditions. The signal after translation the model under wind and no wind conditions, respectively.
is shown in Fig. 3(b). Furthermore, only the stable segments with Fig. 4 shows the time history of dynamic forces under wind con-
time interval of 40 s taken from the time interval of 72 s can be ditions, no-wind conditions, and torsion-related SEFs in typical
used for subsequent analysis, as shown in Fig. 3(c). cases. Ratios of the SEF to dynamic force, depending on wind veloc-
It is clear that the general force acting on the model and mea- ity, torsional amplitude, and vibration frequency, are over 20% in
sured by the high-precise dynamic balances mainly comprises the most cases. Furthermore, Fig. 5 shows the normalized power spectral
inertial force, initial installation force, and dynamic force, involv- density (PSD) of the non-wind-induced aeroelastic force. It is obvi-
ing the wind-induced aeroelastic force and the non-wind-induced ous that there are only even high-order harmonic components due to
aeroelastic force due to the interaction between the vibrating doubly symmetrical aerodynamic forces of the H-shaped section.
model and the surrounding air.
As mentioned previously, synchronous measurements of
Aerodynamic Work
balance-based forces and displacements of the model have been re-
alized. Only torsion-related aerodynamic force is considered and The torsion-related SEFs, mainly leading to the 1DOF torsional
analyzed. Thus, the measured SEFs, namely torsion-related SEFs, flutter, are expressed in the form of a nondimensional coefficient

© ASCE 4 J. Bridge Eng.


(a)

(b)

Fig. 4. Time history of SEFs with or without wind: (a) wind velocity 6.6 m/s, torsional amplitude 3°, vibration frequency 4.2 Hz; and (b) wind ve-
locity 16.5 m/s, torsional amplitude 35°, vibration frequency 2.1 Hz.

Table 3 compares Strouhal number calculated with that from pre-


vious results, and they agree well with each other. the Strouhal num-
ber (St) here is defined as St = fvortexB/U, where fvortex = frequency of
the vortex shedding from the trailing edge of the section, which is
also called the Karman vortex. However, the component correspond-
ing to vortex shedding in the wake is too small to be observed at a
larger vibration amplitude and higher reduced velocity, implying
that the component can be ignored at large vibration amplitude.
Therefore, the torsion-related SEFs can be rewritten as

n
Mse (t) = Ai sin (2πifv t + φi ) + n(t) (4)
i=1

Fig. 5. Normalized PSD of dynamic force without wind (torsional am- where i = order of the harmonics; Ai and φi = amplitude and initial
plitude 35°, vibration frequency 2.1 Hz). phase lag of the ith harmonic component of the torsion-related
SEFs respectively; and n(t) = noise error.
Nondimensional work in a vibration period can be defined as
(CMse (t)) as

Wnon = (5)
Mse (t) 1 2 2
CMse (t) = (3) ρU B L
1 2 2 2
ρU B L
2 where T = vibration period; Wα = aerodynamic work in a vibration
where ρ = air density, 1.225 kg/m3. circle and can be written as
Figs. 6 and 7 show time history and corresponding normalized T
PSD of the torsion-related SEFs at different vibration amplitude Wα = Mse (t)dα (6)
(U/fvB = 6.6), respectively. The abscissa in Fig. 7 stands for re- 0
duced frequency (fB/U). There are obviously high-order harmonic
components in addition to the fundamental one with a reduced fre- Ignoring noise in the SEFs in Eq. (4), the work in a vibration
quency of 0.15 in all cases, which reflects the nonlinear character- period can be determined through Eq. (1); that is,
istics of the SEFs. Ratios of the third harmonic component to the Wα = α0 A1 π sin (φ1 − φ0 ) (7)
fundamental one at amplitude 3° are larger than those at other
vibration amplitudes. Furthermore, there is also a component with where, φ1 − φ0 = phase lag between the general SEFs and model
a reduced frequency of 0.57 at vibration amplitude 3°, which is gen- motion at forced vibration frequency. Therefore, only the funda-
erated by vortex shedding in the wake around the mode. mental components of the SEF work in a vibration period, and

© ASCE 5 J. Bridge Eng.


Fig. 6. Time history of torsion-related SEFs at different vibration amplitude (U/fvB = 6.6).

Table 3. Comparison of Strouhal numbers of the Tacoma Bridge cross


section
Farquharson Larsen and Schewe (1989)
(1952) Walther (1997) Wind tunnel
Wind tunnel DVMFLOW (scale ratio about
Reference (scale ratio 1:36) (prototype) 1:220) Present
Value 0.60 0.57 0.575 0.57

(a)

(b)
Fig. 8. Nondimensional work contour map.

Table 4. LCO velocity (structural damping ratio in torsion ζ s = 0)


Reference Vibration amplitude LCO velocity
Larsen (2000) 15° 3.8–4.0
Billah and Scanlan (1991) Small amplitude About 2.2
Larsen and Walther (1997) 3°–30° 3.7–4.5
Present 3°–45° 3.5–4.4

(c) on the section when the phase lags are in the range of 0°–180°
and negative work when they are in the range of 180°–360°.
Fig. 7. Normalized PSD of torsion-related SEFs at different vibration If all amplitudes are considered, the nondimensional work (Wnon)
amplitude (U/fvB = 6.6): (a) vibration amplitude 3°; (b) vibration ampli- contour map in a vibration period, which is the function of vibration
tude 15°; and (c) vibration amplitude 35°. amplitude and reduced velocity, also termed energy map, can be ob-
tained with 187 cases included, as shown in Fig. 8. The reduced ve-
locity at which the SEF does no work can be referred to as
high-order harmonic components do no work on the model in a vi- energy-balanced velocity or LCO velocity. LCO velocity is in the
bration period as a whole. The work done by the SEFs in a vibration range of 3.5–4.2 when vibration amplitude ranges from 3° to 45°, in-
period depends on the amplitudes of the fundamental components dicating that LCO velocity depends on vibration amplitude.
of the SEFs and the phase lag between the general SEFs and motion Furthermore, Table 4 compares LCO velocities with those from
at forced vibration frequency. The general SEFs do positive work previous conclusions when structural damping ratio in torsion

© ASCE 6 J. Bridge Eng.


Fig. 9. LCO velocity of under various torsional amplitudes and struc-
tural damping ratios. Fig. 10. LCO velocity contour map.

equals 0 (ζs = 0). Present results coincide well with those from Larsen (1952), from wind tunnel tests of the full bridge aeroelastic
and Walther (1997). However, there are considerable discrepancies model, probably due to three-dimensional (3D) effects. In addition,
between the present results and those of Billah and Scanlan (1991); the present results coincide well with those from Larsen and
the latter are closer to the reduced velocity of 1.8 and onset reduced Walther (1997) (obtained from DVMFLOW) when vibration am-
velocity of VIVs due to the Karman vortex in the wake when the plitudes are lower than 10°, but there are large differences at vibra-
Strouhal number takes a value of 0.57. The possible reason may be tion amplitudes of about 30°. Therefore, the results in this paper are
due to the SEFs at small vibration amplitude being more prone to Kar- reliable compared with previous research.
man vortex in the wake, as is discussed above. The LCO velocity basically increases with vibration amplitude
at different structural damping ratio in torsion, which is a typical
characteristic of the postflutter LCOs phenomena. However, the
Collapse Mechanism of the Tacoma Bridge slope of the LCO velocity–vibration amplitude depends on the
structural damping ratio in torsion. It decreases as the structural
Self-Sustained Oscillations damping ratio in torsion increases, implying that the oscillation be-
came self-sustained instead of violently diverging when the struc-
According to Scanlan and Tomko (1971), for 1DOF torsional tural damping ratio in torsion ascends from 0.003 to 0.02. Thus,
vibration, the torsional flutter equation can be expressed as there are obviously self-sustained oscillations, depending on the vi-
 
1 Bα̇ bration amplitude and the structural damping ratio in torsion.
I(α + 2ξα ωα α̇ + ω2α α) = ρU 2 (2B2 ) KA*2 + K 2 A*3 α (8) In conclusion, the present results can be reillustrated by previ-
2 U
ous results. Furthermore, the more complex development of the
where I, ξα , and ωα = mass inertia moment, structural damping ratio aerodynamic performance about amplitude-dependent aerodynam-
in torsion, and the circular frequency of oscillations, respectively; ics and the energy balance between SEFs and structural damping is
ωα = 2πfv ; K = reduced circular frequency; K = Bωα /U ; A*2 and formulated quantitatively in detail.
A*3 = nondimensional flutter derivatives.
The flutter derivative A*2 , which is associated with torsion-
related SEFs caused by torsional velocity, characterizes 1DOF tor- Structural Damping Ratio
sional flutter or 2DOF coupled flutter, corresponding to positive A*2 It is of interest to replot Fig. 9 in order to illustrate relationships be-
or negative A*2 , respectively (Matsumoto et al. 1997). Relationships tween the vibration amplitude and structural damping ratio in tor-
between the flutter derivative A*2 and aerodynamic work that the sion (ζs), as shown in Fig. 10. The number on the contour line
SEFs do in a vibration period can be calculated as denotes reduced velocity at the flutter state, which is referred to
Wα as LCO velocity here. According to Ammann et al. (1941), the
A*2 = (9)
πρB4 ωα α20 Tacoma Bridge collapsed in a 19 m/s gale on November 7, 1940
with corresponding reduced velocity of about 7.9, while twisting
In effect, flutter derivative A*2 can be taken as normalized non- at maximal amplitude of approximately 35° at 1/4 midspan.
dimensional work, as for its physical significance. Then flutter de- The correction coefficient of the maximal torsional amplitude of
rivative A*2 can be obtained by virtue of Eq. (9). the 2D sectional model and the prototype bridge, which is 3D,
According to Billah and Scanlan (1991), the LCO velocity of can be approximately taken as 4/π here, considering the effects of
the Tacoma Bridge can be obtained by determining the critical the 3D structural and vibrational behavior of prototype bridge.
A*2 , that is, It is important to note that the correction coefficient here is obtained
(A*2 )cri = 14.48ζ s (10) by taking the self-sustained oscillations as VIVs (Zhu 2005).
It must be realized that the correction coefficient depends on several
Based on A*2 and Eq. (10), the LCO velocity of the Tacoma factors; for example, the aerodynamic damping ratio, the mode
Bridge at different structural damping ratios in torsion and vibra- shape, the structural damping, and the nonuniformity of the span-
tion amplitude can be obtained, as shown in Fig. 9, which also wise flow, some of which has been ignored to some extent. The
lists previous results for comparison. It is obvious that the present equivalent 2D vibration amplitude of the Tacoma Bridge is then
results are generally consistent with those from Billah and Scanlan about 27°. Thus, the structural damping ratio in torsion (ζs) of
(1991) at small vibration amplitudes, but there are considerable dis- the first antisymmetric torsional mode of the bridge girder can be
crepancies between present results and those from Farquharson estimated as roughly 0.0115, indicated by Point C in Fig. 10,

© ASCE 7 J. Bridge Eng.


which is obviously smaller than 0.014 by Larsen and Walther
(1997), ignoring 3D effects of the prototype bridge.
It was also found that there are obviously multiple stable points
at large structural damping ratios and vibration amplitudes. Taking
the contour line with value of 10.30 as an example, it shows that
there are three stable points at the vibration amplitude larger than
30°, when the structural damping ratio in torsion takes a value of
0.015, as indicated by the dotted line, and Points K, J, and I on
the line. Multiple stable points actually represent multiple critical
states of flutter with a total damping ratio of zero. It can then be in-
ferred that both the aerodynamic damping ratio and the structural
damping ratio vary nonlinearly with the amplitude and in a dy-
namic equilibrium state, which are responsible for multiple stable
points at the same reduced velocity. Furthermore, the slope of the
vibration amplitude tends to infinity at large amplitudes, indicating Fig. 11. Energy map at 0° mean angle of attack.
that sudden instability is very likely to occur at large amplitudes,
termed hard flutter. Therefore, the H-shaped section is prone to
Table 5. Amplitude-dependent structural damping ratio in torsion
nonlinear dynamic phenomena, even sudden instability, that is,
hard flutter at large wind velocities and vibration amplitudes. Initial structural damping ratio a b
0.003 0.03103 −0.00332
0.004 0.02291 0.00050
Instability Path and Possible Path 0.005 0.01895 0.00237
The total damping ratio (ζ) in torsion, comprising aerodynamic and 0.006 0.01521 0.00413
structural damping effects, can be expressed as 0.007 0.01192 0.00568
0.008 0.00883 0.00714
ζ = ζα + ζs (11) 0.009 0.00590 0.00852
0.010 0.00323 0.00978
where ζ α = aerodynamic damping ratio in torsion, and is
defined as
characteristics of the SEFs and structural damping ratio varying
ρB4 * with vibration amplitude are responsible for the self-sustained os-
ζα = − A (12) cillations of the Tacoma Bridge.
2I 2
It can be assumed that the structural damping ratio can be ex-
When the total damping ratio (ζ) equals to zero, self-sustained pressed as a linear function of vibration amplitude, as indicated
oscillation occurs. Then structural damping ratio in torsion ζs can by the dotted line from Point B to Point C; that is,
be obtained as

ρB4 * ξα = aα0 + b (14)


ζs = A (13)
2I 2
where a and b = parameters, depending on the initial structural
If flutter derivative A*2 at all amplitudes and reduced velocities damping ratio in torsion at a small vibration amplitude. Table 5
are considered, the energy balance point at different amplitude lists the parameters a and b within the initial damping ratio range
and reduced velocity constitutes an energy map according to of 0.003–0.01, which is identified by the least square method. It
Eq. (13). The energy map at 0° mean angle of attack is shown in can be used for reference to recognize variations of damping ratios
Fig. 11. Points D, E, and F are in the contour line with a value of for prototype bridges.
−0.020. Every point in the figure indicates the aerodynamic damp- Possible instability paths of the Tacoma Bridge are discussed in
ing ratio, in which aerodynamic work counterbalance energy con- the following, assuming it has sufficient wind-resistant resilience.
sumption by structural damping. It can be assumed that the According to Ammann et al. (1941), along with the failure of
structural damping ratio in torsion before violent torsional motion local structural members, such as loosening of the central cable
is 0.005 here. According to Ammann et al. (1941), the motion of clamp, complex cracks appeared and propagated in the material
the Tacoma Bridge suddenly transited from vertical motion to tor- of the bridge deck, implying that structural damping ratio increased
sional motion in a 12 m/s gale on November 7, 1940 with corre- nonlinearly in the process of collapse. Moreover, structural stiff-
sponding reduced velocity of about 4.3, indicated by Point A in ness declined in that process, validated by the fact that the first
Fig. 11. Torsional amplitude increased abruptly to a large ampli- antisymmetric torsional frequency of the bridge dropped from
tude in a short time, indicated by the dotted line from Point A to 0.233 Hz initially to 0.20 Hz eventually. Therefore, in addition to
Point B. Lastly, the oscillations became self-sustained when re- oncoming wind velocity, there are obvious variations of structural
duced velocity increased from 4.3 to 7.9 until collapse, as indicated damping ratio and structural stiffness with vibration amplitude in
by the dotted line from Point B to Point C. It is of interest to find the process of collapse. When decrease of structural stiffness
that the instability path starting at Point B would be a contour and/or increase of wind velocity are predominant over increase of
line with a value of −0.005, rather than the dotted line from structural damping ratio, structural response increases rapidly as
Point B to Point C, not considering the cracking of the concrete the reduced wind speed increases, as shown by the dashed line
deck and variations of the structural damping ratio with vibration from Point C to Point D in Fig. 11; when increase of structural
amplitude in the process. Moreover, it can be seen that the slope damping ratio are predominant over decrease of structural stiffness
of the contour with a value of −0.005 is obviously larger than and/or increase of wind velocity, structural response decreases dra-
that of the dotted line from Point B to Point C, implying that matically as the reduced wind speed increases, as shown by the
both aerodynamic damping ratio originating from nonlinear dashed line from Point C to Point E in Fig. 11. When they reach

© ASCE 8 J. Bridge Eng.


a balance, the structural response remains unchanged, as shown by Billah, K. Y., and R. H. Scanlan. 1991. “Resonance, tacoma narrows bridge
the dashed line from Point C to Point F in Fig. 11. Therefore, it can failure, and undergraduate physics textbooks.” Am. J. Phys. 59 (2):
be concluded that possible instability path of structural response at 118–124. https://doi.org/10.1119/1.16590.
collapse fundamentally depends on the competitive relationships Cigada, A., M. Falco, and A. Zasso. 2001. “Development of new systems to
measure the aerodynamic forces on section models in wind tunnel test-
between the structural damping ratio, structural stiffness, and on-
ing.” J. Wind Eng. Ind. Aerodyn. 89 (7–8): 725–746. https://doi.org/10
coming wind velocity.
.1016/S0167-6105(01)00075-7.
Diana, G., F. Resta, and D. Rocchi. 2008. “A new numerical approach to
reproduce bridge aerodynamic non-linearities in time domain.”
Conclusions J. Wind Eng. Ind. Aerodyn. 96 (10–11): 1871–1884. https://doi.org/10
.1016/j.jweia.2008.02.052.
An H-shaped section with an aspect ratio of roughly 5:1, a simpli- Diana, G., F. Resta, A. Zasso, M. Belloli, and D. Rocchi. 2004. “Forced
motion and free motion aeroelastic tests on a new concept dynamomet-
fied section of the bridge deck of the Tacoma Bridge, was taken as a
ric section model of the Messina suspension bridge.” J. Wind Eng. Ind.
research objective. The forced vibrations in wind tunnel at large
Aerodyn. 92 (6): 441–462. https://doi.org/10.1016/j.jweia.2004.01.005.
torsional amplitudes were realized with help of an improved Diana, G., D. Rocchi, T. Argentini, and S. Muggiasca. 2010.
FMA, which enabled synchronous measurements of force and dis- “Aerodynamic instability of a bridge deck section model: Linear and
placement response. Furthermore, nonlinear characteristics of SEFs nonlinear approach to force modeling.” J. Wind Eng. Ind. Aerodyn.
at different vibration amplitude were analyzed, and then an energy 98 (6–7): 363–374. https://doi.org/10.1016/j.jweia.2010.01.003.
map concept was established. Based on this, postflutter LCO phe- Falco, M., A. Curami, and A. Zasso. 1992. “Nonlinear effects in sectional
nomena at the collapse of the Tacoma Bridge were revealed, and model aeroelastic parameters identification.” J. Wind Eng. Ind.
then possible instability paths were discussed. Some main conclu- Aerodyn. 42 (1–3): 1321–1332. https://doi.org/10.1016/0167-6105(92)
sions were reached: 90140-6.
1. There are obvious nonlinear self-sustained oscillations, depend- Farquharson, F. B. 1952. Aerodynamic stability of suspension bridges.
ing on the vibration amplitude and the structural damping ratio Seattle: Univ. of Washington Experiment Station.
Gao, G., L. Zhu, J. Li, W. Han, and B. Yao. 2020a. “A novel
in torsion in the process of collapse of the Tacoma Bridge. The
two-degree-of-freedom model of nonlinear self-excited force for cou-
LCO velocity basically increases as vibration amplitude at dif- pled flutter instability of bridge decks.” J. Sound Vibr. 480: 115406.
ferent structural damping ratios in torsion, which is a typical https://doi.org/10.1016/j.jsv.2020.115406.
characteristic of postflutter LCO phenomena. Gao, G., L. Zhu, F. Wang, H. Bai, and J. Hao. 2020b. “Experimental inves-
2. The structural damping ratio in torsion at collapse is estimated to tigation on the nonlinear coupled flutter motion of a typical flat closed-
be about 0.0115, considering 3D effects. Both aerodynamic box bridge deck.” Sensors 20 (2): 568. https://doi.org/10.3390
damping ratio in torsion originating from nonlinear characteris- /s20020568.
tics of the SEFs, and structural damping ratio varying with vi- Gao, G. Z., and L. D. Zhu. 2015. “Nonlinearity of mechanical damping and
bration amplitude, are responsible for the self-sustained stiffness of a spring-suspended sectional model system for wind tunnel
oscillations of the Tacoma Bridge. tests.” J. Sound Vibr. 355: 369–391. https://doi.org/10.1016/j.jsv.2015
3. A possible instability path of structural response in the process .05.033.
Gao, G. Z., L. D. Zhu, W. S. Han, and J. W. Li. 2018. “Nonlinear post-
of collapse fundamentally depends on the competitive relation-
flutter behavior and self-excited force model of a twin-side-girder
ships between the structural damping ratio, structural stiffness, bridge deck.” J. Wind Eng. Ind. Aerodyn. 177: 227–241. https://doi
and oncoming wind velocity. .org/10.1016/j.jweia.2017.12.007.
The study has revealed postflutter LCOs phenomena of the col- Ge, Y. J., J. L. Xia, L. Zhao, and S. Y. Zhao. 2018. “Full aeroelastic model
lapse of the Tacoma Bridge mainly in terms of energy effects of testing for examining wind-induced vibration of a 5,000 m spanned sus-
SEFs. Furthermore, a novel methodology is proposed for estimat- pension bridge.” Front. Built Environ. 4: 20. https://doi.org/10.3389
ing the structural damping ratios at collapse. This will aid in estima- /fbuil.2018.00020.
tion of these parameters for prototype bridges and extend the deep Hu, C. X., L. Zhao, and Y. J. Ge. 2018. “Time-frequency evolutionary
understanding of nonlinear aerodynamics and structural effects. characteristics of aerodynamic forces around a streamlined closed-box
girder during vortex-induced vibration.” J. Wind Eng. Ind. Aerodyn.
182: 330–343. https://doi.org/10.1016/j.jweia.2018.09.025.
Hu, C. X., L. Zhao, and Y. J. Ge. 2019. “Mechanism of suppression of
Data Availability Statement vortex-induced vibrations of a streamlined closed-box girder using ad-
ditional small-scale components.” J. Wind Eng. Ind. Aerodyn. 189:
All data, models, or code that support the findings of this study are 314–331. https://doi.org/10.1016/j.jweia.2019.04.015.
available from the corresponding author upon reasonable request. Hu, C. X., L. Zhao, and Y. J. Ge. 2021. “A simplified vortex model for the
mechanism of vortex-induced vibrations in a streamlined closed-box
girder.” Wind. Struct. 32: 309–319.
Larsen, A. 2000. “Aerodynamics of the Tacoma narrows bridge - 60 years
Acknowledgments later.” Struct. Eng. Int. 10 (4): 243–248. https://doi.org/10.2749
/101686600780481356.
The authors gratefully acknowledge the support of the National Larsen, A., and J. H. Walther. 1997. “Aeroelastic analysis of bridge girder
Natural Science Foundation of China (52108471 and 52078383) sections based on discrete vortex simulations.” J. Wind Eng. Ind.
and Independent Subject of State Key Lab of Disaster Reduction Aerodyn. 67–68: 253–265. https://doi.org/10.1016/S0167-6105(97)
in Civil Engineering (SLDRCE19-B-11). 00077-9.
Li, M., S. Li, H. Liao, J. Zeng, and Q. Wang. 2016. “Spanwise correlation
of aerodynamic forces on oscillating rectangular cylinder.” J. Wind
Eng. Ind. Aerodyn. 154: 47–57. https://doi.org/10.1016/j.jweia.2016
References .04.003.
Li, Q. C. 1996. “Measuring flutter derivatives for bridge sectional models in
Ammann, O. H., T. Von Karman, and G. B. Woodruff. 1941. The failure of water channel - closure.” J. Eng. Mech. 122 (4): 390–390. https://doi
the Tacoma narrows bridge. Washington, DC: Federal Works Agency. .org/10.1061/(ASCE)0733-9399(1996)122:4(390).

© ASCE 9 J. Bridge Eng.


Li, W., S. Laima, X. Jin, W. Yuan, and H. Li. 2020. “A novel long short- Siedziako, B., O. Oiseth, and A. Ronnquist. 2017. “An enhanced forced vi-
term memory neural-network-based self-excited force model of limit bration rig for wind tunnel testing of bridge deck section models in ar-
cycle oscillations of nonlinear flutter for various aerodynamic configu- bitrary motion.” J. Wind Eng. Ind. Aerodyn. 164: 152–163. https://doi
rations.” Nonlinear Dyn. 100 (3): 2071–2087. https://doi.org/10.1007 .org/10.1016/j.jweia.2017.02.011.
/s11071-020-05631-5. Tang, Y., X. G. Hua, Z. Q. Chen, and Y. Zhou. 2019. “Experimental inves-
Matsumoto, M. 1999. “Vortex shedding of bluff bodies: A review.” J. Fluids tigation of flutter characteristics of shallow π section at post-critical re-
Struct. 13 (7–8): 791–811. https://doi.org/10.1006/jfls.1999.0249. gime.” J. Fluids Struct. 88: 275–291. https://doi.org/10.1016/j
Matsumoto, M., Y. Daito, F. Yoshizumi, Y. Ichikawa, and T. Yabutani. .jfluidstructs.2019.05.010.
1997. “Torsional flutter of bluff bodies.” J. Wind Eng. Ind. Aerodyn. Ukeguchi, N., H. Sakata, and H. Nishitani. 1966. “An investigation of
71: 871–882. https://doi.org/10.1016/S0167-6105(97)00213-4. aeroelastic instability of suspension bridges.” In Suspension
Matsumoto, M., N. Shiraishi, H. Shirato, K. Shigetaka, and Y. Niihara. Bridges Symp. Lisbon.
1993. “Aerodynamic derivatives of coupled/hybrid flutter of fundamen- Wu, B., Q. Wang, H. Liao, and H. Mei. 2020. “Effects of vertical motion on
tal structural sections.” J. Wind Eng. Ind. Aerodyn. 49 (1–3): 575–584. nonlinear flutter of a bridge girder.” J. Bridge Eng. 25 (11): 04020093.
https://doi.org/10.1016/0167-6105(93)90051-O. https://doi.org/10.1061/(ASCE)BE.1943-5592.0001637.
Matsumoto, M., H. Shirato, and S. Hirai. 1992. “Torsional flutter mecha- Zhang, M., F. Xu, and Y. Han. 2020a. “Assessment of wind-induced
nism of 2-d h-shaped cylinders and effect of flow turbulence.” nonlinear post-critical performance of bridge decks.” J. Wind Eng.
J. Wind Eng. Ind. Aerodyn. 41 (1–3): 687–698. https://doi.org/10 Ind. Aerodyn. 203: 104251. https://doi.org/10.1016/j.jweia.2020
.1016/0167-6105(92)90480-X. .104251.
Matsumoto, M., H. Shirato, T. Yagi, R. Shijo, A. Eguchi, and H. Tamaki. Zhang, M., F. Xu, T. Wu, and Z. Zhang. 2020b. “Postflutter analysis of
2003. “Effects of aerodynamic interferences between heaving and tor- bridge decks using aerodynamic-describing functions.” J. Bridge
sional vibration of bridge decks: The case of Tacoma narrows bridge.” Eng. 25 (8): 04020046. https://doi.org/10.1061/(ASCE)BE.1943
J. Wind Eng. Ind. Aerodyn. 91 (12–15): 1547–1557. https://doi.org/10 -5592.0001587.
.1016/j.jweia.2003.09.010. Zhang, M., F. Xu, Z. Zhang, and X. Ying. 2019. “Energy budget analysis
Nakamura, Y., and M. Nakashima. 1986. “Vortex excitation of prisms with and engineering modeling of post-flutter limit cycle oscillation of a
elongated rectangular, h and perpendicular-to cross-sections.” J. Fluid bridge deck.” J. Wind Eng. Ind. Aerodyn. 188: 410–420. https://doi
Mech. 163: 149–169. https://doi.org/10.1017/S0022112086002252. .org/10.1016/j.jweia.2019.03.010.
Náprstek, J., and S. Pospíšil. 2011. “Post-critical behavior of a simple non- Zhao, L., X. Xie, Y.-Y. Zhan, W. Cui, Y.-J. Ge, Z.-C. Xia, S.-Q. Xu, and
linear system in a cross-wind.” Eng. Mech. 18 (3/4): 193–201. M. Zeng. 2020. “A novel forced motion apparatus with potential appli-
Otsuki, Y., K. Washizu, H. Tomizawa, and A. Ohya. 1974. “A note on the cations in structural engineering.” J. Zhejiang Univ.-Sci. A 21 (7): 593–
aeroelastic instability of a prismatic bar with square section.” J. Sound 608. https://doi.org/10.1631/jzus.A1900400.
Vibr. 34 (2): 233–248. https://doi.org/10.1016/S0022-460X(74)80307-X. Zhou, R., Y. Ge, Y. Yang, Y. Du, S. Liu, and L. Zhang. 2019. “Nonlinear
Scanlan, R. H., and J. J. Tomko. 1971. “Airfoil and bridge deck flutter de- behaviors of the flutter occurrences for a twin-box girder bridge with
rivatives.” J. Eng. Mech. Div. 97 (6): 1717–1733. https://doi.org/10 passive countermeasures.” J. Sound Vibr. 447: 221–235. https://doi
.1061/JMCEA3.0001526. .org/10.1016/j.jsv.2019.02.002.
Schewe, G. 1989. “Nonlinear flow-induced resonances of an h-shaped sec- Zhu, L. D. 2005. “Mass simulation and amplitude conversion of bridge sec-
tion.” J. Fluids Struct. 3 (4): 327–348. https://doi.org/10.1016/S0889 tional model test for vortex-excited resonance.” [In Chinese.] Eng.
-9746(89)80015-5. Mech. 22 (5): 204–208.

© ASCE 10 J. Bridge Eng.

View publication stats

You might also like