You are on page 1of 5

Long-Span Cable-Supported Bridges Performance

Improvements in Extreme Winds: Five Case Studies


Jon D. Raggett, PhD, SE, Pres., West Wind Laboratory, Inc., Monterey, CA, USA. Contact: raggett@westwindlaboratory.com
DOI: 10.2749/101686616X14480232444487

Abstract existing bridges. In all cases, the bridge


deck geometries were changed to elim-
Long-span cable-supported bridges are very flexible and are susceptible to a inate the aerodynamic mechanisms
dynamic, instability in strong windsflutter. Discussed in this paper are five case that caused the anticipated excessive
studies where the anticipated critical flutter wind speeds were low in the original motions.
design, but increased dramatically by making small changes to the geometry of
the deck cross-section. In none of the cases were excessive motions observed; they For all cases, static aerodynamic coef-
were only anticipated in the design phase. Four of the five aerodynamic improve- ficients and aeroelastic flutter deriva-
ments described in this paper were designed and have been implemented. The tives were obtained in a wind tunnel
improvements to the Golden Gate Bridge, the fifth, have been designed and are using large-scale section models of the
expected to be installed soon. bridge deck and towers. The turbulent
wind flow field was generated analyti-
Keywords: long-span bridges; flutter wind speeds; aerodynamic improvements. cally. Bridge response time histories
were then generated analytically using
the static and dynamic aerodynamic
coefficients obtained experimentally
Introduction trouble-causing aerodynamic mecha- based on turbulent wind speed flow
nism. Geometric changes to the bridge field,1 using the dynamic response
Long-span cable-supported bridges deck are the typical changes made. characteristics of the bridge structure,
are very flexible and, as a consequence, and a numerical simulation procedure
can experience large wind-induced Buffeting motions are difficult to
reduce through changes in the bridge in the time domain similar to that
motions. These motions are typically described in Ref. [2] (except that the
caused by buffeting from random tur- deck geometry alone. A bridge deck
section can be streamlined to reduce flutter derivatives were used instead of
bulence in the wind and wake, caused the impulse response functions used in
by periodic vortices shed into the lateral sway. However, that stream-
lining can make the section more the cited reference).
wake, or caused by aeroelastic instabil-
ity because of the shape of the bridge airfoil-like, which may increase
deck cross-section. This last aerody- the vertical and torsional buffeting Airfoil Behavior
namic mechanism was responsible for response. Aeroelastic instabilities and
the catastrophic failure of the Tacoma vortex-induced motions are more eas- Consider first the behavior of an air-
Narrows Bridge in 1940, and is the ily reduced by making small changes in foil (a wing) in a horizontal wind with
subject of this paper. the bridge deck geometry. speed U. Assume that the airfoil has
been disturbed and is moving down-
Motion can be reduced by increas- Discussed in this paper are five case
ward with a speed of dw/dt. The appar-
ing the bridge stiffness, by increas- studies performed in the last 37 years
ent angle of incidence of the wind with
where potential motions in extreme
ing mechanical damping (often with respect to the airfoil is a = tan1(dw/
tuned mass dampers), or by making winds were identified, and then elimi-
dt/U). The aerodynamic lift per unit
small changes to the bridge geom- nated by making small changes to
length of the airfoil will be:
etry that can eliminate the expected, the bridge deck geometry. In no case
were actual wind-induced motions Lift = (1/2rU2)(2pa)(c) (1)
experienced. The potential for wind-
where r is the density of air, and c is
This paper is being published posthu- generated problems was identified in
the chord of the airfoil.
mously. Dr. Jon D. Raggett passed away the design phase of new bridges, or in
on 26 September 2015 shortly after having the design phase of modifications to This is shown schematically in Fig. 1.
completed the revised version of the paper,
which was subsequently accepted for Lift
publication. The Editorial Board wishes to
express its sincere condolences to his family,
and would like to recognise Jon's contribu-
tions to the journal and to the profession as
a whole. U


dw dw
Peer-reviewed by international ex- dt dt
perts and accepted for publication
by SEI Editorial Board
C
Paper received: April 16, 2015
Paper accepted: September 30, 2015 Fig. 1: Flow over airfoil

6 Scientific Paper Structural Engineering International Nr. 1/2016


For small angles of incidence, a is Lift
Translation
Lift
approximately dw/dt/U. Therefore, the
magnitude of the lift will be approxi-
U
mately proportional to the vertical
velocity of the airfoil, and it will be
directed opposite to the vertical veloc-
ity of the airfoil. The lift will tend to
resist the vertical airfoil motion. In a Force = (two lift)
horizontal wind, the airfoil will act
as an aerodynamic damper, adding Moment = (two lift)(r)
damping to the mechanical damping, Lift
to resist downward vertical motions.
Using the same arguments, in a hori- U
zontal wind, the airfoil is equally effec-
tive as a damper that resists upward Lift
airfoil motions. There will be a down- Rotation
ward lift tending to resist the upward r
vertical motion of the airfoil. For
upward or downward airfoil motions Fig. 2: Winglet pair
in a horizontal wind, the airfoil will act
as an aerodynamic damper tending to The design for re-doing the deck of specify a critical flutter wind speed cri-
resist the airfoil motion. the Westbound Bridge included two terion during construction lower than
If airfoils are placed at the edges of changes, the closing of grilled openings the critical flutter wind speed criterion
a bridge deck in clear air, as shown in the road deck, and the addition of for the completed bridge. The critical
in Fig. 2, they will act as an aerody- solid traffic barriers. Before the deck flutter wind speed criterion during
namic damper to resist both vertical modifications were made, the critical construction was 34 m/s with a mean
and torsional deck motions.3 If the flutter wind speed of the bridge was 46 recurrence interval (MRI) of approxi-
deck moves down in a wind with speed m/s. Both changes degraded the aero- mately 1000 years. This specification is
U, both airfoils will generate a lift up, dynamic performance of the bridge. typical. What was not typical was that
resisting the downward deck motion With the modifications made to the the construction wind speed criterion
a vertical aerodynamic damper. If the deck (closing of the deck openings and of 34 m/s was greater than the criti-
deck is rotating counterclockwise in a the addition of solid barriers), the criti- cal flutter wind speed for the existing
wind U, the windward airfoil will go cal flutter wind speed was predicted to bridge, that is, 33 m/s. In order for the
down (generating a lift up), and the be only 27 m/s. At this location (across contractor to begin any construction,
leeward airfoil will go up (generat- the Chesapeake Bay), at the bridge the existing bridge had to be enhanced
ing a lift down). Those two opposing deck elevation, it was determined that aerodynamically to bring its critical
lifts multiplied by the moment arm the critical flutter wind speed criterion flutter wind speed up to the construc-
between them will generate an aerody- for a new bridge would be 69 m/s. It tion stage criterion.
namic moment that resists that coun- was proposedand implemented in
The existing bridge had three lanes
terclockwise deck motiona torsional the designto add 1.37 m wide winglet
of traffic and no dedicated walk-
aerodynamic damper. aerodynamic dampers below the road-
ways. There were narrow walkways
way, in the truss space. These aerody-
Presented below is a discussion of namic dampers are shown in Fig. 3.
five case studies where the princi-
pals described in this section have to The winglets on the windward side
be used to improve the performance are in clear air and will be effective in
of the bridge design in strong winds, resisting both upward and downward
by making small changes in the deck deck motions. The winglets on the lee-
geometries that increase aerodynamic ward side are in disturbed air, but mov-
damping (which in turn increases the ing air nonetheless, and are somewhat
critical flutter wind speed of the bridge effective in resisting both upward and
1.37 m
dramatically). downward motions, as described in
the previous section. The pair of wing-
4.68 m

lets will act as vertical and torsional


William Preston Lane aerodynamic dampers, as described in
Westbound BridgeBaltimore, Ref. [3].
Maryland, USA Aerodynamic damper

The studied bridge is one of a pair of Lions Gate Bridge


suspension bridges, each with a main ReconstructionVancouver,
span length of 485 m. Both are truss- British Columbia, Canada
stiffened suspension bridges. In the
Westbound Bridge, the roadway is The reconstruction (total deck and Modified section with solid barrier,
on top of the stiffening trusses; in the hanger replacement) of the Lions Gate solid deck, and winglet
Eastbound Bridge, the roadway passes Bridge posed an interesting prob- Fig. 3: Westbound William Preston lane
through the stiffening trusses. lem to the contractors. It is typical to bridge

Structural Engineering International Nr. 1/2016 Scientific Paper 7


on both sides of the bridge, and to and leeward sides of the bridge deck. would be much more airfoil-like. A
protect the workers using them, tem- Again the pair of walkway covers rounded corner would be most effec-
porary railings were added between acted exactly as the pair of wing- tive to allow the upward angled flow to
the walkways and the traffic lanes. It lets were described previously and in attach to the upper deck surface, creat-
was recommended that the tempo- Ref. [3]. Both covered walkways were ing an upward lift. However, the bev-
rary railings be extended upward, and effective as aerodynamic dampers for eled corner was found to be effective
that a 1.22 m wide cover be installed vertical and torsional motions in both enough, and easier to construct. With
over the walkways. It rains frequently directions. that bevel, for all reasonable wind
in Vancouver, so the walkway covers speeds, the torsional aerodynamic
As the existing deck was replaced, the
were a welcome addition for weather damping on the bridge deck was posi-
walkway covers (the winglet pair aero-
protection. The true intended purpose tive, eliminating the possibility of tor-
dynamic dampers) were replaced with
of the walkway covers was to act as sional flutter instability. The originally
a new deck section, which met all con-
an aerodynamic damper that would designed sharp corner and the beveled
struction and final critical flutter wind
resist vertical, and most importantly, corner design are shown in Fig. 5. The
speed criteria.
torsional motions. The torsional aero- bridge, with that modified edge, was
dynamic damping became negative aerodynamically stable for wind speeds
damping for wind speeds >33 m/s. With in excess of the criterion of 95 m.
the addition of the walkway covers
Kap Shui Mun Bridge Shortly after the Kap Shui Mun Bridge
(winglets), the torsional aerodynamic Hong Kong was completed, it did experience,
damping was found to be positive for without a problem, the full force of a
The Kap Shui Mun Bridge is a three-
all expected wind speeds. With the typhoon.
span, steel and concrete cable-stayed
winglets installed over 176.8 m in the
bridge with a main span length of
middle of the 473 m main span, the
430 m and a wide (32.3 m) and deep
increased aerodynamic damping did
raise the critical flutter wind speed
(7.2 m) boxsection deck. There are Ironton Russell Bridge
traffic lanes and rail tracks for high- Russell, Ohio, USA
of the bridge during construction to
speed trains within the box.
35 m/s, greater than the criterion dur-
The Ironton Russell Bridge is a three-
ing construction of 34 m/s. The section The specified critical flutter wind
span concrete cable-stayed bridge with
of the bridge deck, with and without speed criterion was very high at 95 m/s.
a main span length of 274 m and a nar-
the winglets, is shown in Fig. 4. The original design had a sharp cor-
row deck width of only 13 m. Narrow
ner at the upper deck surface and the
The walkway covers were in rela- bridges are particularly prone to aero-
large cable anchor covers. The original
tively clear air, both on the windward elastic instabilities. The original design
design also had a critical torsional flut-
had wide edge girders (1.35 m) with a
ter wind speed of 71.9 m/s, well below
12.19 m pointed exterior where the point was
the criterion of 95 m/s.
at mid-height. For this configuration,
The winglet pair discussed previously the critical flutter wind speed was
is an extremely powerful aerodynamic found to be only 28.8 m/s, well below
damper, for both vertical and torsional the critical flutter wind speed criterion
motions. One single winglet located of 38.3 m/s.
windward is also a powerful damper,
The effectiveness of the modifications
for both upward and downward
made to the Kap Shui Mun Bridge,
motions. Would one winglet by itself,
Pre-reconstruction typical deck section resisting downward motions of the lead-
that is effective as an aerodynamic
ing edge only, influenced the proposed
damper for downward motions only, be
modification to the Ironton Russell
a powerful enough damper to increase
Bridge. It was proposed to move the
the critical flutter wind speed above
1.22 m point of the edge beam down to 16.5%
the criterion? In order to increase the
Covered walkway of the height, to smoothen and accel-
(aerodynamic
critical flutter wind speed of the pro-
erate the flow under the edge girder,
damper) posed Kap Shui Mun Bridge to the
making it airflow-like. When the lead-
criterion of 95 m/s, it was proposed to
ing edge of the bridge rotated upward
modify the upper corner of the bridge
(from any undefined disturbance) in a
deck section, to make the flow up over
horizontal wind, the apparent angle of
the leading edge to be more airfoil-
attack of the flow would act downward
like, for downward motions only of the
generating a lift downward, thereby
leading edge.
resisting the initial upward motion.
It was proposed (and was included in The originally designed, and the modi-
the final design) to bevel that upper fied edge configurations are shown in
corner with a 0.484 0.484 m bevel. Fig. 6. For this design, the critical flut-
If a strong wind disturbance should ter wind speed was increased to 37.6
make the leading edge of the bridge m/s (a 30% increase in the critical flut-
deck rotate downward, the apparent ter wind speed because of the minor
angle of attack of that wind would be geometric change in the exterior pro-
upward, and with the beveled lead- file of the edge girder). A slightly more
Covered walkway detail ing edge corner, the flow up over the open traffic barrier was used in the
Fig. 4: Lions Gate Bridge corner and top surface of the bridge final design that increased the critical

8 Scientific Paper Structural Engineering International Nr. 1/2016


Beveled edge,
0.484 m 0.484 m
U
7.2 m U

Original deck cross-section Modified deck cross-section with beveled edge


Fig. 5: Kap Shui Mun Bridge

1.37 m
U U
0.23 m

Lowered point

Original edge girder section Modified edge girder section


Fig. 6: Ironton Russell bridge

flutter wind speed to 48 m/s, well above west, this is a wind speed with a mean these modifications, air flows are fairly
the criterion of 38.3 m/s. recurrence interval (MRI) of approxi- smooth over the leading edge, acceler-
mately 100 years. The strongest mean ating over the walkway, and generating
The aerodynamic damping would have
wind speed ever recorded in San a lift that resists the original downward
been stronger yet if the lower edge
Francisco in modern times is only about motion. The original section at an
beam shape had been curved instead
27 m/s. It was proposed to make modi- upper corner and the proposed modi-
of beveled. However, the curved shape
fications to the Golden Gate Bridge to fied section are shown in Fig. 7. These
would have been more expensive to
increase its critical flutter wind speed modifications are capable of increas-
cast, and the added effectiveness was
to a wind speed with an MRI of 10 000 ing the critical flutter wind speed from
not needed.
years (the typical standard for new 30 to 45 m/s, and a wind speed with an
bridges). A wind speed of 30 m/s has MRI of 10 000 years.
Golden Gate Bridge an MRI of approximately 10 000 years
The aerodynamic enhancements will
for winds from the east. However, for
San Francisco, California, USA winds from the west, a mean wind
be installed soon.
The Golden Gate Bridge (GGB), built speed, at the bridge deck elevation of
in 1938, is a suspension bridge with 70 m, with an MRI of 10 000 years is Conclusions
a main span length of 1280 m. It is approximately 45 m/s. Modifications
27 m wide, with a roadway on top of to the bridge deck geometry were pro- All the potential problems in the cases
a 7.62 m deep stiffening truss. It was posed, to the west side of the bridge described were identified from wind
originally built without a horizon- deck only, to increase the critical flut- tunnel tests using large-scale models of
tal stiffening truss across the bottom, ter wind speed of winds from the west, a section of the bridge deck. The use
between the bottom chords of the stiff- from 30 to 45 m/s. of large-scale section models allows
ening trusses. The bridge axis is almost very small changes in the bridge deck
It was proposed to make slight modi-
exactly northsouth. geometry (as mentioned) to be mod-
fications to the deck section, on the
eled accurately. The solutions were
In 1952, in winds with gust wind speeds west side of the bridge only, to make
also designed from these wind tun-
of 31 m/s, the bridge experienced exces- the flow over the leading edge more
nel tests. Again, in none of the cases
sive torsional motions. Shortly there- airfoil-like (like a single winglet wind-
were unacceptable wind-induced
after, a horizontal stiffening truss was ward, and for downward motions only
motions ever observed; they were only
installed across the bottom, effectively of the leading edge). To make the
anticipated and the potential for their
creating a stiff, trussed box-section. flow more airfoil-like, a small fair-
occurrence was eliminated before any
ing (with a radius of 0.3 m) placed on
of the motions occurred.
Even with the torsionally stiffened the outboard edge of the west walk-
deck, the bridge is quite flexible and way was proposed. Furthermore, it Four of the five projects have been
could experience large torsional was proposed to streamline the flow completed (in the last 36 years) and
motions at a relatively low mean wind through the pedestrian railing, on one of the projects will be constructed
speed of 30 m/s. For winds from the the west side of the bridge only. With soon.

Structural Engineering International Nr. 1/2016 Scientific Paper 9


Opened railing

New fairing
Radius = 0.32 m

U
U

Existing walkway detail Modified walkway detail


Fig. 7: Golden Gate Bridge

In this paper, details of the bridge Manager, Golden Gate Bridge Highway and [2] Chen X, Matsumoto M, Kareem A. Time
deck geometry are identified that Transportation District (Golden Gate Bridge); domain flutter and buffeting response analysis
Dr Steven P. Stroh, VP, URS Corporation of bridges. J. Eng. Mech. 126(1): 110.
can be highly beneficial in increasing
(Ironton Russell Bridge and the Kap Shui
expected critical flutter wind speeds [3] Raggett, JD. Stabilizing winglet pair for slen-
Mun Bridge); Ronald Crockett, Vice President der bridge decks. Proceedings of the 6th Annual
of long-span, cable-supported bridges. Engineering, American Bridge Company
In essence, this paper is the beginning ASCE Structural Division Structures Congress,
(Lions Gate Bridge Reconstruction); and August 1720, 1987.
of a database of geometric details for William Pines, Division of Engineering
improving the performance of new or and Construction Management, Maryland [4] Anderson JD. Fundamentals of
existing bridges in extreme winds. Transportation Authority (William Preston Aerodynamics. McGraw-Hill, 2007.
Lane Bridge). [5] Batchelor GK. An Introduction to Fluid
Mechanics. Cambridge, 1967; 467471.
Acknowledgements References
[6] Simiu E, Scanlan RH. Wind Effects on
The author would like to thank the following [1] Cao Y, Xiang H, Zhou Y. Simulation of sto- Structures, 3rd edn. John Wiley & Sons: New
individuals for their support in making this chastic wind velocity field on long-span bridges. York, 1996.
paper possible: Denis J. Mulligan, General J. Eng. Mech. 2000; 126(1): 16.

.FNCFSTIJQ#FOFUT
XXXJBCTFPSHBQQMJDBUJPO
t8PSMEXJEFOFUXPSLJOSFTFBSDIBOEQSBDUJDF
t'SFFTVCTDSJQUJPOUPRVBSUFSMZKPVSOBM4&*
t'SFFBDDFTTUPFMFDUSPOJDBSDIJWFPG4&*TJODF
t'SFFTVCTDSJQUJPOUP&CPPLT 4USVDUVSBM&OHJOFFSJOH%PDVNFOUT 4&%

t3FEVDFEQSJDFTGPS*"#4&1VCMJDBUJPOT
t3FEVDFESFHJTUSBUJPOGFFTGPS*"#4&$POGFSFODFT
t1PTTJCJMJUZUPKPJO5FDIOJDBM(SPVQT
t0QQPSUVOJUZUPKPJOBDUJWJUJFTPG/BUJPOBM(SPVQT
t'SFF+PC"EWFSUJTJOHPOXXXJBCTFPSH
t"DDFTTUP.FNCFST"SFBPOXXXJBCTFPSHJODMVEJOH.FNCFST%JSFDUPSZ

10 Scientific Paper Structural Engineering International Nr. 1/2016

You might also like