294 - Snaebjoernsson - Rev1
294 - Snaebjoernsson - Rev1
294 - Snaebjoernsson - Rev1
www.ndt.net/app.EWSHM2016
Abstract
Wind and traffic induced vibrations of a 446 m long suspension bridge in a complex terrain
are studied, based on full-scale data. The full-scale data consists of wind velocity recorded by
sonic anemometers mounted above the bridge deck and acceleration records from accelerome-
ters placed on either side of four different sections of the steel bridge box-girder. In addition,
a GNSS base-rover system monitors the displacements at the center span. Ambient tempera-
ture data is also available. The analysis focuses on the identification of modal parameters and
the evaluation of the buffeting response. The modal parameters are affected by ambient tem-
perature, the magnitude of excitation and/or amplitude of response. Similarly, the response is
induced by either wind or traffic or both. For numerical modelling verification, it is necessary
to isolate the different sources of excitation. The variability in modal parameters and buffet-
ing response is investigated as such information is necessary for establishing the boundaries of
normal operations for the bridge.
1 INTRODUCTION
Norway’s national road network along the west coast has been very dependent on ferry transport
across the many wide and deep fjords that cut through the Norwegian coastline. The Norwegian
Parliament has agreed on a transport plan to realize a ferry free road (E39) between Kristiansand
in the south and Trondheim in the north by 2030 [1]. To achieve this goal, new standards need
to be set in long span bridging and/or tunnelling, as the fjords that need crossing are between
1.6 and 5 km long and up to 500 m deep. A development of this type requires research on
many levels. One part is to investigate the loading and response of suspension bridges. For
this purpose, a full-scale monitoring program was initiated on the Lysefjord suspension Bridge,
near Stavanger in Norway. The establishment of a full-scale suspension bridge laboratory is
important for future development of fjord crossings, as relatively few studies of full-scale bridge
in complex terrains such as fjords are available.
In traditional design of long span bridges, there are many sources of uncertainty. For in-
stance:
• Simplified assumptions regarding terrain conditions, topography and local wind condi-
tions.
1
Snæbjörnsson et al. Performance evaluation of a suspension bridge
3 THE INSTRUMENTATION
The wind conditions along the bridge are monitored using eight sonic anemometers mounted on
hangers 8, 10, 16, 18, 20, 24 and 30, at the South-West side of the bridge. They are placed either
on the hangers or on the main cables about 6 m above the deck, except at hanger 8 where there
are two anemometers mounted on the hanger at 6 and 10 m above the bridge deck respectively.
The anemometer on hanger 10, is a Vaisala weather transmitter WXT520, providing ambient
temperature and barometric pressure as well as wind velocity and direction, while the others are
3D WindMaster Pro sonic anemometers from Gill Instrument Ltd. The frequency resolution
2
Snæbjörnsson et al. Performance evaluation of a suspension bridge
x
Accelerometer
Measuring GNSS
Reference GNSS y z
Sonic anemometer
Fig. 2. Sensors installed along the bridge deck. Anemometers are represented by grey triangles; GNSS
sensors are visible as one blue and green dot; accelerometers are depicted as red rectangles.
for the WindMaster is 32 Hz and 4 Hz for the Vaisala weather station. The distance between
hangers is 12 m, so the distance between anemometers varies from 24 m to 264 m.
The bridge response is monitored by four couples of tri-axial accelerometers located inside
the bridge deck, near hangers 9, 18, 24, 30, and one additional pair is placed at the top level of
the North tower (figure 2). The accelerometers are placed on each side of the tower and the deck
to be able to detect lateral, vertical and torsional motion. The dynamic displacement response
of the bridge is obtained based on the acceleration data through integration in the frequency
domain using discrete Fourier transform (DFT).
Real-Time Kinematic-Global Positioning System is installed at midspan on the deck of the
Lysefjord Bridge to measure directly the total displacement response of the bridge deck in
the East, North and vertical directions. The system is comprised of a set of Trimble BD930
GNSS receivers coupled to Trimble AV33 GNSS antennas, in a base-rover setup. The individual
sensors can handle data sampling at a frequency of 20 Hz with an accuracy of ± 8mm + 1ppm
for the horizontal displacement and ± 15mm + 1ppm for the vertical displacements. The base-
rover combination will increase the measurement accuracy, as it is the relative displacement
between the "fixed" base station and the "moving" rover station that is being recorded.
All sensors are linked to interconnected data acquisition units (CUSP-3/CUSP-M/CUSP-
Me) from Canterbury Seismic Instruments. A GPS timestamp is used to synchronize the data.
All data is sampled at 50 Hz and then decimated down to 20 Hz during data processing. A 3G
3
Snæbjörnsson et al. Performance evaluation of a suspension bridge
router enables a wireless data access and transfer via a mobile net.
4
Snæbjörnsson et al. Performance evaluation of a suspension bridge
HS1 HA1
1 1
0.5 measured 0.5
0 Benchmark 0
−0.5 Alvsat −0.5
−1 −1
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
VA1 VS1
1 1
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
TS1 TS2
1 1
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Fig. 3. Identified mode shapes and natural frequencies for the first two lateral (top), vertical (middle)
and torsional (bottom) modes of vibration.
structural damping is close to the predicted one. However, the evolution with the mean wind
velocity is slightly non-linear and the measured damping is less than the predicted one. This
may lead to a computed response larger than expected [4]. For larger suspension bridges, the
non-linearity of the vertical damping may be more important and less predictable, which would
justify the application of OMA to trace its evolution.
5
Snæbjörnsson et al. Performance evaluation of a suspension bridge
HS1 HA1
4 4
ζ (%)
2 2
0 0
0 5 10 15 20 0 5 10 15 20
VA1 VS1
4 4
ζ (%)
2 2
0 0
0 5 10 15 20 0 5 10 15 20
TS1 TA1
4 ζ + σζ 4
ζ
ζ (%)
2 Computed 2
0 0
0 5 10 15 20 0 5 10 15 20
U (m/s) U (m/s)
Fig. 4. The modal damping ratio as a function of the mean wind velocity for the first two lateral (top),
vertical (middle) and torsional (bottom) modes of vibration.
long term fluctuations (several years), seasonal fluctuations (several months) or daily fluctua-
tions (several days). For the Lysefjord Bridge, the seasonal periodicity and the daily periodicity
have been investigated.
The influence of seasonal temperature variations on the modal parameters of a suspension
bridge is not well documented, although such analysis have been carried out for a concrete
arch bridge [14, 15] where significantly higher eigenfrequencies were recorded during winter
time. The Lysefjord study suggests that the first lateral, veritcal and torsional eigen-modes
clearly display a temperature dependent "seasonal trend" in line with the temperature depen-
dancy demonstrated in Fig. 5.
The daily fluctuations of the natural frequencies were studied for a period of ten days, using
data recorded in September 2015 (Fig. 6). The natural frequency of the first horizontal symmet-
ric lateral mode of vibration, HS1, fluctuates between 0.135 Hz for diurnal data and 0.14 Hz for
nocturnal data. These frequency fluctuations are relatively small compared to those for VA1,
where the natural frequency ranges from 0.217 at day time to almost 0.23 during the night. For
the first symmetric torsional mode of vibration, TS1, the natural frequency fluctuates between
1.225 and 1.25 Hz. As indicated by Kim et al. [16], the natural frequency variations may be
magnified by the influence of traffic. At night time, the lower temperatures and the reduced
traffic load leads to higher natural frequencies, whereas during the day time, the increased tem-
perature and traffic load leads to lower natural frequencies. The relative importance of the traffic
load is however surely much less than the temperature effects.
6
Snæbjörnsson et al. Performance evaluation of a suspension bridge
0.145 0.46
mean + RMS value
HA1 (Hz)
HS1 (Hz)
0.14 mean value 0.45
0.135 0.44
0.13 0.43
0 5 10 15 20 0 5 10 15 20
0.23 0.3
VA1 (Hz)
VS1 (Hz)
0.225
0.295
0.22
0.215 0.29
0 5 10 15 20 0 5 10 15 20
1.26 2.2
TA1(Hz)
TS1(Hz)
1.24 2.18
1.22 2.16
0 5 10 15 20 0 5 10 15 20
T (◦ ) T (◦ )
Fig. 5. Evolution of the first two lateral (top), vertical (middle) and torsional (bottom) natural
frequencies as a function of temperature.
0.145
HS1 (Hz)
0.14
0.135
0.13
0.23
VA1 (Hz)
0.225
0.22
0.215
1.25
TS1 (Hz)
1.24
1.23
1.22
20/09 22/09 24/09 26/09 28/09 30/09
time (DD/MM)
Fig. 6. Evolution of the natural frequency for the first lateral (top), vertical (middle) and torsional
(bottom) mode of vibration for a period of 10 days and nights in September 2015.
7
Snæbjörnsson et al. Performance evaluation of a suspension bridge
0.14
Mean + RMS value
HA1 (Hz)
HS1 (Hz)
0.45
Mean value
0.135 0.44
0.13 0.43
0 5 10 15 20 0 5 10 15 20
0.23 0.3
VA1 (Hz)
VS1 (Hz)
0.225
0.295
0.22
0.215 0.29
0 5 10 15 20 0 5 10 15 20
2.22
1.26
TA1 (Hz)
TS1 (Hz)
1.24 2.19
1.22
2.16
0 5 10 15 20 0 5 10 15 20
U (m/s) U (m/s)
Fig. 7. Evolution of the first two lateral (top), vertical (middle) and torsional (bottom) natural
frequencies as a function of wind velocity.
8
Snæbjörnsson et al. Performance evaluation of a suspension bridge
σx /Iu (m)
0.6 computed 0.6
measured
0.4 0.4
0.2 0.2
0 0
6 8 10 12 14 16 6 8 10 12 14 16
0.8 0.8
σz /Iw (m)
σz /Iw (m)
0.6 0.6
0.4 0.4
0.2 0.2
0 0
6 8 10 12 14 16 6 8 10 12 14 16
·10−3 ·10−3
σt /Iw (◦ )
σt /Iw (◦ )
2 2
1 1
0 0
6 8 10 12 14 16 6 8 10 12 14 16
V x (m/s) V x (m/s)
Fig. 8. RMS of the bridge deck response for a N-NE wind (left panels) and a S-SW-wind (right panels)
as a function of the mean wind velocity normal to the deck V x .
wind field is assumed stationary and homogeneous along the bridge deck. The modal coupling
induced by the wind load as well as the non-symmetry of the structure about the centre of the
bridge span is neglected. The wind load is calculated using a co-coherence function which is
fitted to measured data. The PSD of the buffeting response is computed for the resonant part
only since the accelerometers are not designed to measure accurately the background response,
and the lowest frequency measured is consequently set to 0.08 Hz.
The standard deviation of the measured bridge displacement at mid-span is compared to the
computed one in Fig. 8. The RMS of the displacement has been normalized by the corre-
sponding turbulence intensity to reduce the spreading of the data. A good overall agreement is
found between the measured and computed RMS values for S-SW wind. However, the com-
puted response underestimates the bridge displacements for N-NE exposure, which is generally
characterized by a larger turbulence intensity and a larger yaw angle. The mountainous envi-
ronment affects particularly the yaw angle that fluctuates between 20◦ and 45◦ . For large yaw
angles, larger bridge displacement are recorded than predicted by the "cosine rule" [19] which
is applied in the present study.
For the data studied, the RMS of the measured bridge displacement does not display any
sudden variations at low wind speeds, which indicates that the traffic induced vibrations were
not strongly present during the period studied.
9
Snæbjörnsson et al. Performance evaluation of a suspension bridge
·10−2 ·10−2
6 with traffic 6
without traffic
σ z (m/s2 )
σ z (m/s2 )
4 4
2 2
0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
U (m/s) U (m/s)
Fig. 9. RMS of vertical bridge acceleration at mid span for S-SW wind using the criteria τ (left) and
criteria λ (right).
2. The application of a criterion (τ ) based on the time varying standard deviation of the
vertical bridge acceleration response as a ratio of the RMS of vertical wind velocity.
3. The application of a criterion (λ) based on the relative importance of the high frequency
components of the acceleration response, evaluated as a ratio of the RMS of vertical
response at 4 to 10 Hz and 0 to 4 Hz respectively.
The selection of nocturnal records is straightforward, but may be too harsh. The analysis of
several samples showed that buffeting response from nocturnal and diurnal samples did not
show strong differences when the wind velocity was above 12 m/s. The proposed criterion τ ,
may efficiently detect variation of the acceleration response due to heavy vehicles, but is best
suited for low and moderated wind speeds. Criterion λ, tends to overestimate the number of
samples dominated by traffic-induced vibration at low wind speed. The two criterions τ and
λ are compared on Fig. 9. For criterion τ , some samples displaying a high RMS at low wind
speed are not counted as dominated by traffic-induced vibration. One reason may be that the
threshold value for τ is too low. For strong winds, criterion λ detects almost no traffic-induced
vibration, as is to be expected. The criterion τ may be more adapted for low wind speeds, while
λ seems more reliable for high wind speeds. Alternative forms of these criteria will be further
explored.
6 CONCLUSIONS
Wind and traffic induced vibrations of a 446 m long suspension bridge in a complex terrain have
been studied in full scale.The focus has been on modal parameter characterization of a full-
scale suspension bridge by the use of an automated SSI-COV procedure. This step is central to
properly study the buffeting response of a full-scale bridge.
One of the challenges in the data analyzing is to differentiate between the two main excitation
sources, wind and traffic. Because traffic-induced vibrations affect every mode of vibration
of the bridge, direct filtering cannot be applied. A simple approach, based on evaluating the
relative importance of the high frequency part of the acceleration bridge response over the low-
frequency part, has been proposed to isolate records dominated by wind-induced vibrations
from those dominated by traffic.
Environmental effects were observed, such as the daily and seasonal fluctuations of the
eigen-frequencies. The large data scatter obtained, especially at low wind speed, may be due to
influence from traffic-induced vibrations.
10
Snæbjörnsson et al. Performance evaluation of a suspension bridge
7 ACKNOWLEDGEMENTS
The authors would like to gratefully acknowledge the support of the Norwegian Public Roads
Administration to the measurement campaign, as well as the installation and maintenance of
the monitoring system.
References
[1] Norwegian Ministry of Transport and Communication. National Transport Plan 2014 –
2023. Report to Storting (White Paper) Summary. 2012.
[2] Etienne Cheynet, Jasna Bogunović Jakobsen, and Jónas Snæbjörnsson. Wind character-
istics on a suspension bridge at the inlet of a fjord. IN-VENTO 2014: XIII Conference of
the Italian Association for Wind Engineering. 2014.
[3] Jasna Bogunović Jakobsen et al. Assessment of Wind Conditions at a Fjord Inlet by
Complementary Use of Sonic Anemometers and Lidars. Energy Procedia 80 (2015).
12th Deep Sea Offshore Wind R&D Conference, EERA DeepWind’2015, pp. 411 –421.
[4] Etienne Cheynet, Jasna Bogunović Jakobsen, and Jónas Snæbjörnsson. Buffeting re-
sponse of a bridge at the inlet of a fjord. ICWE14: 14th International Conference on
Wind Engineering. 2015.
[5] Dionysius M Siringoringo and Yozo Fujino. System identification of suspension bridge
from ambient vibration response. Engineering Structures 30.2 (2008), pp. 462–477.
[6] JMW Brownjohn, Filipe Magalhães, Elsa Caetano, and Alvaro Cunha. Ambient vibration
re-testing and operational modal analysis of the Humber Bridge. Engineering Structures
32.8 (2010), pp. 2003–2018.
[7] Ragnar Sigbjörnsson and Erik Hjorth-Hansen. Along-wind response of suspension bridges
with special reference to stiffening by horizontal cables. Engineering Structures 3.1
(1981), pp. 27–37.
[8] Einar N. Strømmen. Structural Dynamics. Cham: Springer International Publishing, 2014.
Chap. Eigenvalue Calculations of Continuous Systems, pp. 89–159.
11
Snæbjörnsson et al. Performance evaluation of a suspension bridge
[9] Yong Xia, Bo Chen, Shun Weng, Yi-Qing Ni, and You-Lin Xu. Temperature effect on
vibration properties of civil structures: a literature review and case studies. Journal of
Civil Structural Health Monitoring 2.1 (2012), pp. 29–46.
[10] YouLiang Ding and AiQun Li. Temperature-induced variations of measured modal fre-
quencies of steel box girder for a long-span suspension bridge. International Journal of
Steel Structures 11.2 (2011), pp. 145–155.
[11] KY Koo, JMW Brownjohn, DI List, and R Cole. Structural health monitoring of the
Tamar suspension bridge. Structural Control and Health Monitoring 20.4 (2013), pp. 609–
625.
[12] Robert Westgate, Ki-Young Koo, and James Brownjohn. Effect of solar radiation on sus-
pension bridge performance. Journal of Bridge Engineering 20.5 (2014), p. 04014077.
[13] Robert Westgate, Ki-Young Koo, and James Brownjohn. Effect of vehicular loading on
suspension bridge dynamic properties. Structure and Infrastructure Engineering 11.2
(2015), pp. 129–144.
[14] Hoon Sohn et al. An experimental study of temperature effect on modal parameters of the
Alamosa Canyon Bridge. Earthquake Engineering & Structural Dynamics 28.8 (1999),
pp. 879–897.
[15] Filipe Magalhães, Àlvaro Cunha, and Elsa Caetano. Online automatic identification of
the modal parameters of a long span arch bridge. Mechanical Systems and Signal Pro-
cessing 23.2 (2009), pp. 316 –329.
[16] Chul-Young Kim, Dae-Sung Jung, Nam-Sik Kim, Soon-Duck Kwon, and Maria Q Feng.
Effect of vehicle weight on natural frequencies of bridges measured from traffic-induced
vibration. Earthquake Engineering and Engineering Vibration 2.1 (2003), pp. 109–115.
[17] Allan Garnett Davenport. Buffeting of a suspension bridge by storm winds. Journal of
the Structural Division 88.3 (1962), pp. 233–270.
[18] R.H. Scanlan. The action of flexible bridges under wind, II: Buffeting theory. Journal of
Sound and Vibration 60.2 (1975), pp. 201 –211.
[19] J. Xie, H. Tanaka, R.L. Wardlaw, and M.G. Savage. Buffeting analysis of long span
bridges to turbulent wind with yaw angle. Journal of Wind Engineering and Industrial
Aerodynamics 37.1 (1991), pp. 65 –77.
[20] Etienne Cheynet, Jasna Bogunović Jakobsen, and Jónas Snæbjörnsson. Full scale mon-
itoring of wind and traffic induced response of a suspension bridge. MATEC Web of
Conferences 24 (2015), p. 04003.
12