TECHNICAL REPORT DOCUMENTATION PAGE

1. Report No. 2. Government Accession No. 3. Recipient's Catalog No.

TX -0-1400-1
4. Title and Subtitle: 5. Report Date:
"Investigation of Wind-Rain-Induced Cable-Stay Vibrations on August 2005
Cable-Stayed Bridges: Final Report"
6. Performing Organization Code:
TechMRT

7. Author(s): 8. Performing Organization

R.S. Phelan, K.C. Mehta, P.P. Sarkar & L. Chen Report No.: 0-1400-1

9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)
Texas Tech University
Department of Civil Engineering
Box 41023 11. Contract or Grant No.
Lubbock, Texas 79409-1023 Project 0-1400

12. Sponsoring Agency Name and Address Type of Report and Period
Texas Department of Transportation Sept 1998-Aug 2001
Research and Technology
P. 0. Box 5080 14. Sponsoring Agency Code
Austin, TX 78763-5080

15. Supplementary Notes

16. Abstract: In recent years, large-amplitude cable-stay vibrations have been observed on a number of bridges in
the U.S. and abroad during relatively low wind speeds with the presence of rain. Typically, the wind-rain-
induced vibration problem has not received adequate attention from bridge designers - resulting in the current
need for mitigation devices. Excessive vibrations accelerate fatigue of the cable-stays and cause distractions to
passing motorists.

Two bridges in Texas have experienced wind-rain-induced cable-stay vibration problems: The Fred Hartman
and Veterans Memorial Bridges located respectively, in Baytown and Port Arthur, Texas. Generally, vibration
has been observed in moderate to heavy rain at wind speeds of 15 m/s (34 mph) or less. This Report documents
the results ofthree years of study ofthe cable stay vibration problem on these two bridges by researchers from
Texas Tech University, under contract to TxDOT. The scope of the TTU effort includes background research,
wind tunnel tests, and field instrumentation and monitoring. Researchers from TTU have developed an
innovative aerodynamic ring as a mitigation for cable-stay vibration. Based on limited data, it appears the
circular rings work in mitigating wind-rain-induced cable-stay vibration.

17. Key Words 18. Distribution Statement

cable stay vibrations; cable stay aerodynamics; No restrictions. This document is available to
current mitigation devices the public through the National Technical
Information Service, Springfield, Virginia 2161
19. Security Classif (of this 20. Security Classif (of this 21. No. ofPages 22. Price
report): Unclassified page): Unclassified 95
Form DOT F 1700.7 (8-72)

-i-
 NOTICE

The United States Government and the State of Texas do not endorse products of
manufacturers. Trade or manufacturers' names appear herein solely because they are
considered essential to the object of this report.
 INVESTIGATION OF WIND-RAIN-INDUCED
CABLE-STAY VIBRATIONS ON
CABLE-STAYED BRIDGES: FINAL REPORT

by

R. Scott Phelan, Ph.D., P.E., TTU Assistant Professor

Kishor C. Mehta, Ph.D., P.E., TTU Hom Professor
Partha P. Sarkar, Ph.D., ISU Associate Professor
Lizhong Chen, TTU Ph.D. candidate

Project Number 0-1400

Report Number 0-1400-1

conducted for the

Texas Department of Transportation

by the

CENTER FOR MULTIDISCIPLINARY RESEARCH IN TRANSPORTATION

TEXAS TECH UNIVERSITY

August2005

-iii-
- iv-
 ACKNOWLEDGEMENTS

The authors would like to gratefully acknowledge the financial support of the Texas
Department of Transportation (TxDOT). Assistance by Keith Ramsey and Elton Brown of the
TxDOT Bridge Inspection Group and Stanley Shafer of the TxDOT Beaumont District was
especially helpful. In addition, comments and assistance from Randy Poston and Mike Lynch of
Whitlock, Dalrymple, Poston and Associates (WDP), Nick Jones of Johns Hopkins University,
Department of Civil Engineering and Karl Frank, Sharon Wood and Michael Kreger of the
University of Texas at Austin are appreciated. Special thanks go to Analog Devices for their
donation of the initial batch of accelerometers. Also, the authors would like to thank Carlos
Morales, a TTU civil engineering architecture graduate student, for rendering many of the
figures presented in this report.

A special acknowledgement goes to Thomas Gardner. A major portion of his TTU

master's thesis is referenced in this report. Mr. Gardner, under supervision of Drs. Mehta and
Sarkar, developed the 2-D force-damper apparatus used in the TTU wind tunnel tests. Also, Mr.
Gardner was responsible for most of the field instrumentation setup on the four TTU-
instrumented Veterans Memorial Cable-Stays. In addition, recognition goes to Dr. Zhongshan
Zhao for his contribution to the wind tunnel tests and preliminary field result analyses.

-v-
 IMPLEMENTATION STATEMENT

This research evaluates technologies potentially capable of mitigating wind-rain-induced

cable-stay vibrations for bridges. Wind tunnel tests are not likely to have commercial
implementation prospects. However, the potential for a full-scale cable-stay test site for
predicting vibrations is suggested by the researchers.

Based on the positive results from field and wind-tunnel tests, the proposed aerodynamic
circular rings, with slight modifications, could be ready as retrofits for implementation on
existing bridges that experience wind-rain-induced cable-stay vibrations. Any implementation
scheme would have to be custom-designed for a particular bridge and would be subject to
approval from the Texas Department of Transportation.

-vi-
Prepared in cooperation with the Texas Department of Transportation and the U.S. Department
of Transportation, Federal Highway Administration.

-vii-
-viii-
AUTHOR'S DISCLAIMER

The contents of this report reflect the views of the authors who are responsible for the facts and
the accuracy of the data presented herein. The contents do not necessarily reflect the official
view of policies of the Texas Department of Transportation or the U.S. Federal Highway
Administration. This report does not constitute a standard, specification, or regulation.

PATENT DISCLAIMER

Prior to the official commencement of this TxDOT Research Project 0-1400, TTU researchers
applied for two patents on circular ring cable-stay vibration mitigation. However, there was no
invention or discovery conceived or first actually reduced to practice including art, method
process, machine, manufacture, design, or composition of matter in the course of this project
which is or may be patentable under the patent laws of the United States of America or any
foreign country.

ENGINEERING DISCLAIMER

Not intended for construction, bidding, or permit purposes. The engineer in charge of the
research study was R. Scott Phelan, Texas Tech University, Lubbock, Texas.

TRADE NAMES AND MANUFACTURERS' NAMES

The United States Government and the State of Texas do not endorse products or manufacturers.
Trade or manufacturers' names appear herein solely because they are considered essential to the
object of this report.

- ix-
 Symbol Symbol

LENGTH LENGTH
in inches 25.4 millimeters mm mm millimeters 0.039 inches in
ft feet 0.305 meters m m meters 3.28 feet ft
yd yards 0.914 meters m m meters 1.09 yards yd
mi miles 1.61 kilometers km km kilometers 0.621 miles mi

AREA AREA
in1 square inches 645.2 square millimeters mml mm1 square millimeters 0.0016 square inches in2
ft' square feet 0.093 square meters m• m• square meters 10.764 square feet ft2
yd' square yards 0.836 square meters m• m2 square meters 1.195 square yards yd'
ac aaas 0.405 hectares ha ha hectares 2.47 aaas ac
mil square miles 2.59 square kilometers km' kml square kilometers 0.386 square miles mi2
VOLUME VOLUME
X' 29.57
tloz fluidounces milliliters ml ml mil&liters 0.034 fluid ounces floz
gal gallons 3.785 liters L l liters 0.264 gallons gal
ft' cubic feet 0.028 m~ m~ cubic meters 35.71 cubic feet ftJ
cubic meters
yd' cubic yards 0.765 cubic meters m~ m~ cubic meters 1.307 cubic yards yd'
NOTE: Volumes greater !han 1000 I shaH be shown in m~.

MASS MASS
oz ounoes 28.35 grams g g grams 0.035 ounoes oz
lb pounds 0.454 kilograms kg kg kilograms 2.202 pounds lb
T short tons (2000 lb) 0.907 megagrams Mg Mg megagrams 1.103 short tons (2000 lb) T
(or •metric ton") (or "t") (or ·n (or "metric ton")
TEMPERATURE (eX,Jict) TEMPERATURE (exact)
OF Fahrenheit 5(F-32)19 Celcius oc oc Celcius 1.8C + 32 Fahrenheit •F
temperature or (F-32)11.8 temperature temperature temperature

ILLUMINATION ILLUMINATION
fc foot-Qllldles 10.76 lux lx lx lux 0.0929 foot-candles fc
fl foot-Lamberts 3.426 candela/m 2 cdlm2 cdlm 2 candela/m 2 0.2919 foot-lamberts

FORCE and PRESSURE or STRESS FORCE and PRESSURE or STRESS

"
lbf poundforoe 4.45 newtons N N newtons 0.225 poundlorca lbf
lbflin2 poundforoe per 6.89 kilopascals kPa kPa kilopascals 0.145 poundloroe per lbflin 2
square inch square inch

.. . ...
• Sl is the symbol for the International System of Units. Appropriate
. . ~.. - . .
. .. . --· ~ -- - - (Revised September 1993}
 TABLE OF CONTENTS

TABLE OF CONTENTS .... ................ .......................... .................. .. .... ........ .... ......... xt
LIST OF FIGURES.................................................................................................... Xll
LIST OF TABLES ...................................................................................................... XIV

CHAPTER 1- INTRODUCTION........................................................................... 1
1.1 Overview..................................................................................................... 1
1.2 Background................................................................................................. 1
1.3 Fred Hartman Bridge ... .. ... ... .. ........ .... .... ............ .... .. .. ... .. .. .. .. .... .................. 2
1.4 Veterans Memorial Bridge.......................................................................... 5
1.5 Advisory Panel............................................................................................ 6

CHAPTER 2- CABLE-STAY VIBRATION BACKGROUND............................ 9

2.1 Chapter Overview....................................................................................... 9
2.2 Background................................................................................................. 9
2.3 Earlier TxDOT Field Observations............................................................. 11
2.4 Cable-Stay Aerodynamics ..... ........ ........ .... .... .. .. .... .... ........ .... ........ .... .. .. ... .. 12
2.5 Previous Reported Field Observations on Bridges..................................... 19
2.6 Mitigation Devices...................................................................................... 20

CHAPTER 3- TTU WIND TUNNEL TESTS........................................................ 25

3.1 Past Observations on Wind Tunnel Models................................................ 25
3.2 CSU Wind Tests......................................................................................... 27
3.3 TTU Wind Tunnel Section Model Tests..................................................... 37
3.4 Wind Tunnel Tests Summary .. ........ .. .... .... .. .... .... ... ... ... .. .. .. .. .. .. ...... ........ .... 50

CHAPTER 4- TTU FIELD TESTS AND RESULTS............................................ 51

4.1 Instrumentation Setup for Fred Hartman Bridge........................................ 51
4.2 Instrumentation Setup for Veterans Memorial Bridge................................ 52

CHAPTER 5- FIELD RESULTS............................................................................. 55

5.1 Overall Results for Veterans Memorial Bridge............................................ 55
5.2 Major Veterans Memorial Cable-Stay Accelerations Event........................ 56
5.3 Aerodynamic Ring Installation.................................................................... 58
5.4 Field Events after Ring Installation.............................................................. 59
5.5 Summary ...................................................................................................... 74
5.6 Conclusions.................................................................................................. 76

Xl
CHAPTER 6 -SUMMARY AND CONCLUSIONS................................................ 77
6.1 Wind Tunnel Results.................................................................................... 77
6.2 Field Site Results.......................................................................................... 77
6.3 Future Research............................................................................................ 78

REFEREN.CES............................................................................................................ 81

APPENDIX A.............................................................................................................. 83

APPENDIX B.............................................................................................................. 85

Xll
 LIST OF FIGURES

Figure 1.1 Fred Hartman Bridge............................................................................ 2

Figure 1.2 Veterans Memorial Bridge.................................................................... 5
Figure 2.1 Yaw, Inclination, Equivalent Yaw and Wind Attack Angles............... 10
Figure 2.2 Upper Rivulet on Cable-Stay................................................................ 10
Figure 2.3 Velocity-Restricted Behavior of Wind-Rain-Induced Vibration.......... 11
Figure 2.4 Various Aerodynamic Countermeasures Investigated by
Matsumoto et al.......................... .. .. .. .. .. .. .. .. .......... .. .. .... .. .. ..... .. .. .. .. .. .. .... 23
Figure 3.1 Wind-Tunnel Model Setup in the Meteorological Wind Tunnel
At CSU.................................................................................................. 28
Figure 3.2 Model at a 0° and ~o Showing an Artificial-Upper-Water Rivulet... 28
Figure 3.3 Aerodynamic Devices Tested............................................................... 28
Figure 3.4 (a), (b )Vertical Displacement Response (rms) of Secion Model
(D= 10.2cm).......................................................................................... 30
Figure 3.4 (c), (d), (e) Vertical Displacement Response (rms) of Section
Model (D= 10.2cm) .............................................................................. 31
Figure 3.5 Ht. for Cable With and Without Mitigation Devices............................ 33
Figure 3.6 Total Critical Damping at Different Wind Speeds............................... 34
Figure 3.7 Formation of upper water rivulet on a yawed and inclined Cable........ 35
Figure 3.8 Prevention of upper water rivulet with Circular Rings......................... 35
Figure 3.9 Directions of the Mean Aerodynamic Forces....................................... 35
Figure3.10 2DOF TTU Cable with Rings and Suspension System........................ 38
Figure 3.11 2DOF TTU Cable-Stay Setup in Wind Tunnel Section Model............ 39
Figure 3.12 Variation ofLift and Drag with Yaw Angle (bare cylinder)................ 41
Figure 3.13 Comparison of Responses of Bare Cylinder for each Yaw Angle........ 42
Figure 3.14 Response of Cylinder with Rivulet at P* 15°. .......... .. .. .......... .... .. ..... 43
Figure3.15 Response of Cylinder with Rivulet at P* = 25°.. ........... ..... ... .... ........ ... 44
Figure 3.16 Response of Cylinder with Rivulet at P* 35° ... .. .... ................ .. ......... 44
Figure 3.17 Response of Cylinder with Rivulet at P* 45°. ................ ................... 45
Figure3.18 Response of Cylinder with Rivulet at P* = 55° .................................... 45
Figure 3.19 Response with Covering of Sandpaper and No Rivulet at P* = 35°..... 46
Figure 3.20 Response with Covering of Sandpaper with Rivulet at P* = 35° ..... .... 46
Figure 3.21 Comparison of Responses of cylinder with and without Sandpaper
(no rivulet) ..... .... ................ .... .................. .... .............................. ........ ... 47
Figure 3.22 Response of Cylinder with Circular Rings Spaced at 4D, P* = 35° ..... 47
Figure 3.23 Response of Cylinder with Circular Rings Spaced at 2D, P* = 35° ..... 48
Figure 3.24 Variation of Lift and Drag with Yaw Angle......................................... 48
Figure 3.25 Aerodynamic Ring Effectiveness at Low Wind Speed with Rivulet.... 49
Figure 3.26 Aerodynamic Ring Effectiveness at High Wind Speed without
Rivulet................................................................................................... 49
Figure 4.1 Veterans Memorial: Instrumentation Scheme and Wind Angle........... 54

Xlll
Figure 5.1 Veterans Memorial 5 g Cable-Stay Vibration Event: Simultaneous Wind
Speed and Direction, Rainfall, and Stay Acceleration.......................... 57
Figure 5.2 Full-Scale Aerodynamic Ring Installation............................................ 58
Figure 5.3 B14HZ One-Minute RMS Acceleration vs. Wind Speed
(before rings)......................................................................................... 62
Figure 5.4 B 14HZ One-Minute RMS Acceleration vs. Wind Speed
(after rings)............................................................................................ 62
Figure 5.5 C14HZ One-Minute RMS Acceleration History vs. Wind Speed (Before
January 10, 2001 No Rings)............................................................... 63
Figure 5.6 C14HZ One-Minute RMS Acceleration History vs. Wind Speed (After
January 10, 2001- No Rings)............................................................... 63
Figure 5.7 B 14HZ One Minute RMS Acceleration Distribution with Wind Speed and
Direction (Before January 10, 2001 -Without Rings).......................... 66
Figure 5.8 B 14HZ One Minute RMS Acceleration Distribution with Wind Speed and
Direction (After January 10, 2001- With Rings)................................. 67
Figure 5.9 C14HZ One Minute RMS Acceleration Distribution with Wind Speed and
Direction (Before January 10, 2001- Without Rings)......................... 70
Figure 5.10 C14HZ One Minute RMS Acceleration Distribution with Wind Speed and
Direction (After January 10, 2001 Without Rings)........................... 71
Figure 5.11 Dominating Mode Distribution ofB14HZ (Before Ring Installation,
i.e. Before January 10, 2001) ................................................................ 72
Figure 5.12 Dominating Mode Distribution ofB14HZ (After Ring Installation,
i.e. After January 10, 2001) .................................................................. 73
Figure 5.13 Dominating Mode Distribution ofC14HZ (Before January 10, 2001, No
Rings)............................................................................................. ..... 73
Figure 5.14 Dominating Mode Distribution ofC14HZ (After January 10, 2001, No
Rings).................................................................................................... 74

XIV
 LIST OF TABLES

Table 1.1 Fred Hartman Bridge: Cable-Stay Geometry and Frequency

(WDP, 1999) .............. .... .. .. .. .. .. .. .. .. .. .. .. .. .. .......... .. ................ .. .. .. .. .... .. .. . 4
Table 1.2 Veterans Memorial Bridge: Cable-Stay Geometry and Frequency
(WDP, 2002)......................................................................................... 6
Table 3.1 Videotaped Vibration Events of the Fred Harman Bridge Stay
Cables.................................................................................................... 32
Table 3.2 Force Coefficients................................................................................. 36
Table 5.1 "Good Records" vs. Total Number of Records.................................... 55
Table 5.2 "Good Records": Accelerations Greater than 0.5g .............. .... ............. 56
Table 5.3 All Events (>2.5g) without Rings from August 1, 1999 to
January 10, 2001 ................................................................................... 59
Table 5.4 All Events (>2.5g) from January 10, 2001 to July 10, 2001 (B14 with
Rings, C14 without Rings).................................................................... 60
Table 5.5 Number of Good Records (>0.5g) for each Cable-Stay from August
1999 to July 2001.................................................................................. 61
Table 5.6 Maximum One Minute RMS Acceleration on All Channels Before and
After January 10, 200 1.. .... .. ... .... .. .. .. .. .... .. ... .... .... .. .. ..... .... .... .. .. .. ... .. ...... 64
Table 5.7 Acceleration RMS ofB14HZ (With Rain, With Wind Direction from
315° to 345° .......................................................................................... 66
Table 5.8 Acceleration RMS ofC14HZ (With Rain and Wind Direction Between
15° and 45°)........................................................................................... 69
Table 5.9 Cable-Stay Behavior Summary Table .................. .... .. .. .. .. .. .. .. .. ........ .... 75

XV
 CHAPTER I
INTRODUCTION

This report: ( 1) presents an overview of research conducted prior to this project on

the Fred Hartman and Veterans Memorial Bridges located at Baytown and Port Arthur,
Texas, respectively, (2) provides a background on current mitigation devices, (3)
summarizes wind tunnel tests performed at Colorado State University (CSU) and Texas Tech
University (TTU) on "passive" aerodynamic ring system, (4) describes field instrumentation,
and ( 5) summarizes field test results. Parallel complimentary studies of cable-stay vibrations
and their implications are continuing at Johns Hopkins University and the University of
Texas at Austin (see Section 1.5); their results are not included in this report.

1.1 Overview
This final report documents major findings concerning the cable-stay vibration
problem for bridges in Texas under investigation. Chapter l of the report provides a
background to the Fred Hartman and Veterans Memorial Bridges, along with a listing of the
Advisory Panel set up by the Texas Department of Transportation, TxDOT, to investigate the
problem. Chapter 2 provides a rather thorough background of cable-stay aerodynamics, field
observations and potential mitigation strategies. Chapter 3 documents wind tunnel tests
performed at Colorado State University and Texas Tech University and serves as a basis for
the field-testing of prototype aerodynamic rings. Chapter 4 presents the field instrumentation
used at each bridge. Chapter 5 presents field results TTU researchers obtained on the
Veterans Memorial Bridge, both before and after ring installation. The report concludes with
a summary of results and recommendations for future research in Chapter 6. The report also
has Appendices A and B. In Appendix A there is a Glossary of common Wind Engineering
terms. Appendix B presents criteria that can be used to evaluate closure of bridges to traffic
in case of high winds.

1.2 Background
In recent years, large-amplitude cable-stay vibrations have been observed on a
number of bridges in the U.S. and abroad during relatively low wind speeds in the range of 5
to 15 m/s (11-34 mph)-with and without the presence of rain. Wind-rain-induced cable-
stay vibration is an aerodynamic phenomenon that was relatively unknown until recently.
The proposed cause of the vibration problem is the change in cross-sectional shape of the
cable-stay that occurs when rain forms one or more beads, or rivulets, along the cable
surface. This modified cross section affects the aerodynamics of the cable-stay and, as a
result, large vibrations occur at wind speeds above the known vortex shedding wind speeds
for cylindrical bodies. Typically the wind-rain-induced vibration problem has not received
adequate attention from bridge designers. Excessive vibrations accelerate fatigue of the
cable-stays and cause distractions to passing motorists.
Organized efforts have been made by engineers under contract to TxDOT to
determine the vibration characteristic of cable-stays under ambient conditions. Generally,
vibration has been observed in moderate to heavy rain at wind speeds of 15 m/s (34 mph) or
less. When accompanied by rain, this phenomenon has been called "wind-rain-induced

Project 0-1400 Page 1

vibration", and has been observed in recent years on many cable-stayed bridges around the
world. Excessive cable-stay vibrations can distress the cable-stays themselves and subject
them to stress states for which they were not designed. Long-term fatigue damage to the
cable-stays is another concern. In addition, the safety perception to the public is an important
ISSUe.

The Fred Hartman Bridge near Baytown, Texas and the Veterans Memorial Bridge
near Port Arthur, Texas are long-span cable-stayed bridges. Large amplitude cable-stay
vibrations have been observed numerous times at both of these bridges, usually during rain
with relatively low winds. As a result of excessive cable-stay vibrations, a number of guide
pipes of the cable-stays have been damaged.
In addition to this report, other researchers have documented cable-stay aerodynamic
instabilities occurring during rain with wind events (Hikami et al., 1988). Occurrences of
extreme vibration also have been observed with wind only (Main et al., 1999; Matsumoto et
al. , 1995).

1.3 Fred Hartman Bridge

The Fred Hartman Bridge (Figure 1.1 ), also known as the Baytown Bridge, is a cable-
stayed bridge that crosses the Houston Ship Channel and connects Loop 201 in Baytown with
State Highway 225 in LaPorte, Texas. The total length of the bridge is 754 m (2475 ft) with
five spans including a central span of381 m (1,250 ft).

Figure 1.1 Fred Hartman Bridge

Project 0-1400 Page 2

 Each of the bridge's twin decks is 24m (78ft) wide and is supported by two planes of
inclined cable-stays arranged in a fanned configuration on each side of two diamond-shaped
towers. Thus, there are eight planes of cables for the two decks with a total of 192 cables;
each plane of cables consisting of 24 cables of variable lengths and diameters. Each cable-
stay has a number of strands with cement grout housed inside a Polyethylene (PE) pipe. PE
pipes from 11 to 18 em (4.3 to 7.1 inches) in diameter are wrapped by weatherproof tape.
The two concrete towers, supporting the twin decks side-by-side, are joined at the deck level.
The composite deck consists of concrete roadway slabs that are stiffened by steel girders and
transverse steel beams. The bridge was opened to traffic in 1995. The longest cable-stays
are approximately 198m (650 feet) in length. Cable-stay vibrations at the Fred Hartman
Bridge have been observed to reach amplitudes of five times the cable diameter, or more.
The main parameters of the cable-stays that are of direct relevance to cable vibration
and mitigation, such as cable length, cable-stay design tension and the fundamental
frequency, are given for the Fred Hartman bridge in Table 1.1. The frequencies of higher
modes are theoretically integer multiples of the fundamental frequency (i.e. first mode) and
are not listed in this table.

Project 0-1400 Page3

 Table 1.1 Fred Hartman Bridge: Cable-Stay Geometry and Frequency (WDP, 1999)

Cable-stay Length (ft) Design Fundamental

Tension (kips) Frequency (Hz)
1 565.5 1147 0.668
2 559.4 1174 0.687
3 548.9 1002 0.655
4 493.6 729 0.720
5 449.4 828 0.810
6 406.1 699 0.864
7 364.1 440 1.032
8 323.9 585 1.104
9 286.3 456 1.255
10 252.5 405 1.355
11 223.0 346 1.794
12 197.1 341 1.700
13 195.3 371 1.944
14 220.9 359 1.545
15 251.1 427 1.523
16 286.5 485 1.263
17 326.0 538 1.096
18 368.4 614 1.000
19 412.7 720 0.944
20 458.3 753 0.838
21 505.1 861 0.770
22 552.5 797 0.700
23 600.6 963 0.654 •

24 649.1 1018 0.585

Project 0-1400 Page4

1.4 Veterans Memorial Bridge
The Veterans Memorial Bridge (Figure 1.2) is located near Port Arthur, Texas. It
provides for crossing over the Neches River on State Highway 87. The main structure of the
bridge consists of five spans with a total length of 451 m (1 ,480 ft) including a central span
of 195 m (640 ft). The bridge deck is a precast concrete box girder that is supported by four
planes of cable-stays, each of which is arranged in a vertical harped configuration and
anchored to the concrete tower. The bridge was opened to traffic in 1991. The design
tension and the fundamental frequency for each of the 14 cable-stay lengths are given in
Table 1.2

Figure 1.2 Veterans Memorial Bridge

Project 0-1400 Page 5

 Table 1.2 Veterans Memorial Bridge: Cable-Stay
Geometry and Frequency (WDP, 2002)

Cable-stay Length (ft) Design Fundamental

Tension (kips) Frequency (Hz)
01 36.9 545 10.9
02 58.9 548 7.0
03 80.6 543 5.1
04 102.0 531 4.0
05 123.6 515 3.3
06 145.2 495 2.9
07* 166.8 478 2.5
08* 188.4 457 2.2
09 210.0 431 1.9
10 231.6 411 1.7
11 253.2 396 1.6
12 274.8 383 1.4
13 296.4 374 1.3
14* 318.0 369 1.2
Note: Cable Stays C07, B08, Bl 4 and CJ 4 (Band C planes, see Figure 4.5), as indicated with an *
in this table, were instrumented by TTU with two 2-axis accelerometers each.

1.5 Advisory Panel

Concerns by TxDOT regarding the observed large amplitude vibration of cable-stays
at both the Fred Hartman Bridge and Veterans Memorial Bridge prompted an evaluation of
the in-service performance of these bridges. An advisory panel of university experts and
consultants was assembled to assist TxDOT in conducting this evaluation in 1998. In
addition to TxDOT, the advisory panel consists of:
• Whitlock, Dalrymple, Poston & Associates, Inc. (WDP).
• University of Texas at Austin (UT).
• Texas Tech University (TTU),
• Johns Hopkins University (JHU).

Project 0-1400 Page 6

1.5.1 WDP & Associates. Inc.
WDP was charged with the responsibility of the field evaluation of damage caused by
large amplitude cable-stay vibrations, as well as with overall program management and
coordination duties. Prominent contributions by WDP are as follows:
• Vibration testing of stays to assess damage and characterization of cable-stay
dynamics.
• Preparation of repair plans and specifications for a temporary restrainer system and
repair of fatigue damage to guide pipes to Fred Hartman.
• Design, implementation and evaluation of prototype viscous dampers for cable-stay
vibration mitigation at Fred Hartman and Veterans Memorial Bridges.
• Evaluation of a Freyssinet proprietary mechanical damper on selected stays at Fred
Hartman.
A final report documenting the major accomplishments of research performed by
WDP and JHU was submitted to TxDOT by WDP, April30, 2001 (WDP 2001). TTU results
presented in this 0-1400 final report compliment results presented in the WDP report.
Retrofit installation tasks remain for WDP.

1.5.2 University of Texas at Austin

The principal contribution of the UT-Austin researchers has been in assessing the
implication oflarge amplitude wind-rain-induced vibrations on the cable-stay grout and
potential fretting and fatigue damage of the 7-wire strands that comprise an individual cable-
stay. Several salient milestones from the UT work are as follows:
• Analytical study of the bending stresses and potential fatigue stress damage induced
by large amplitude vibrations.
• Instrumentation and monitoring of grout and polyethylene pipe of selected stays at the
Fred Hartman Bridge to assess potential damage and analytical model calibration.
• Development of a full-scale laboratory test to assess fatigue and fretting resistance of
stays to large amplitude vibrations.

1.5.3 Texas Tech University

The principal areas of contribution by TTU researchers has been the aerodynamic
characterization of the wind-rain-induced cable-stay vibration phenomenon, development of
meteorological instrumentation for field studies, and the exploration of aerodynamic
damping devices. Prominent milestones achieved from the TTU work are as follows:
• Review ofliterature to develop background on cable-stay vibration.
• Wind tunnel testing to characterize wind-rain-induced cable-stay vibrations.
• Wind tunnel evaluation and development of aerodynamic damping strategies for
cable-stay vibration mitigation.
• Development of a 22-channel field instrumentation system for the Veterans Memorial
Bridge and system monitoring.
Project 0-1400 Page 7
 • Evaluation of multiple "aerodynamic rings" as a prototype damper.
• Monitoring and analysis of the behavior of cable-stays both before and after ring
installation.

Results of these tests and analyses of data are presented in this report.

1.5.4 Johns Hopkins University

A principal contribution of JHU has been the development, implementation, and
long-term monitoring of vibration instrumentation at both the Fred Hartman and the Veterans
Memorial Bridges. Salient milestones from the JHU work are as follows:
• Development, implementation and maintenance of a 64-channel and a 24-channel
system for vibration monitoring ofFred Hartman and Veterans Memorial Bridges,
respectively.
• Continuous monitoring of field data acquisition systems for a period of over 3 years.
• Database development and management of more than 5,000 five-minute records from
the Fred Hartman Bridge and of more than 3,200 five-minute records from the
Veterans Memorial Bridge.

Project 0-1400 Page 8

 CHAPTER2
CABLE-STAY VIBRATION BACKGROUND

2.1 Chapter Overview

This chapter presents a thorough background of cable-stay aerodynamics and
parameters relevant to the wind-rain-induced cable stay vibration phenomenon. The chapter
includes initial TxDOT field observations of the Fred Hartman and Veterans Memorial
Bridges taken in 1996. Cable-stay aerodynamic theory is presented, followed by field
observations for a number of international cable-stayed bridges. The chapter concludes with
an overview of currently available cable-stay vibration mitigation devices. Aerodynamic
theory utilizes its own terms that are not typically common to bridge designers; a glossary of
aerodynamic terms is given in Appendix A.

2.2 Background
Cables used as bridge stays are usually made of high strength steel. They function as
tensile structural members and are very flexible. Their inherent low structural damping in
transverse oscillation, typically 0.1% of critical (Davenport, 1995), is easily overcome by
aerodynamic influences. This combination of flexibility and low damping make the
ordinarily smooth-surfaced circular cable-stays in cable-stayed bridges prone to excessive
cross-wind vibrations. Contrary to what one might initially expect, it is not always the
longest stays that have the most vibration problems. For a given plane of stays only two or
three stays may vibrate, which are not necessarily adjacent to one another. The most extreme
visible vibration occurs in one of the first three modes.
Hikami and Shiraishi (1988) tested a rigid model in simulated wind-rain-induced
vibration condition to investigate the role of rain in cable vibrations. A 2.6 m long model
with the same surface material (polyethylene pipe) and diameter (140 mm) as the original
was used. Without simulated rain, cables with inclination angle, a= 45° and yaw angle, P=
- 45° or 45° (see Figure 2.1) were found to be stable. 1 With simulated rain, P=- 45° was
stable and P= 45° was unstable. Vibrations with a steady amplitude of 11 em (4.3 in) were
observed for the latter in the wind speed range of 9 to 13 m/s (20 to 29 mph). Matsumoto et
al. (1995) demonstrated the aerodynamic similarity between an inclined/yawed cable (a,/])
and a yawed cable with fJ*. They also argued that the axial f1 ow in the wake of a yawed
cable played an important role in cross-wind cable vibration. They concluded that in order
for galloping instability to occur, the yaw angle must be greater than 25°.

1
For an explanation of angles, refer both to Figure 2.1 and to Section 2.4.1.

Project 0-1400 9
 Vertical

Horizontal
Plane

Wind Vibration
Plane
Stagnation
Point

Figure 2.1. Yaw(~), Inclination (a), Equivalent Yaw(~*) and Wind Attack (y)
Angles.

The effect of artificial rain was also studied by Matsumoto et al. (1995) employing an
artificial rivulet. The location of the artificial rivulet was characterized by an angle(}
measured from the stagnation point (line). This is illustrated in Figure 2.2.

Rivulet
Wind

Stagnation

Figure 2.2 Upper Rivulet on Cable-Stay

Parametric study of an artificial rivulet location indicated that the aerodynamic

responses of a yawed cable were very sensitive to the location of the upper rivulet.
Comparisons were also made with the responses of a bare cable. Cables, yawed or non-
yawed, were found aerodynamically unstable for certain locations of the upper rivulet. The
Scruton number and the turbulence level in the incident flow were reported to affect the
cable's dynamic responses. Three types of cable responses were identified: a) "divergent",
b) "velocity-restricted", and c) a combination of these two as shown in Figure 2.3.

Project 0-1400 10
 A A A

(a) divergent
v
1\ v (c) combination
v
(b) velocity-restricted

Figure 2.3 Velocity-Restricted Behavior of Wind-Rain-Induced Vibration (V; velocity, A;

amplitude)

2.3 Earlier TXDOT Field Observations

Prior to the commencement of this research project, TxDOT began to document the
cable-stay vibration problem. Analysis of several vibration events videotaped by TxDOT at
the Fred Hartman Bridge from April 1 to April 4, 1996 indicated that the cable-stay
vibrations typically occurred in the first three modes. A summary of observations is cited in
WDP (1998). The frequencies were estimated by counting the number of cycles during a
fixed period of time. The limited information provided from the earlier field observations
strongly suggested that large amplitude vibrations were due to the combined action of wind
and rain. Information concerning wind speed, wind direction and precipitation was deemed
essential to a more precise understanding of the environmental conditions when specific
cables experienced large oscillations.
The fact that only a few neighboring cables experienced large amplitude vibration in
a wind-rain-induced vibration condition at any one time seems to indicate that the formation
ofrivulet(s) governed the aerodynamic forcing on the cables within the wind speed range
observed. This implication is supported by the following facts: (1) the neighboring cables
are similar in terms of diameter, inclination, natural frequency and damping characteristics;
(2) the location of the upper rivulet on the windward side is dependent on the balance of
wind pressure, gravity, friction and surface tension acting on the rivulet.
Specifically, the wind speed and direction, the curvature/diameter and surface
condition of cables are important factors in the formation and location of the upper rivulet. It
is known that for a particular inclined cable, a continuous upper rivulet would form under a
certain combination of all of these parameters.
Full-scale studies of the actual cause of the vibration are expensive, time consuming,
inconvenient, and inevitably involve uncontrolled, randomly varying environmental factors.
For these reasons, it is difficult to develop a full understanding of the mechanism of vibration
based solely on full-scale studies. TTU researchers have investigated both full-scale studies
and wind tunnel tests. The focus of the wind tunnel portion of this study was to observe the
behavior of cable-stays in a more controlled wind tunnel environment by using section
models.

Project 0-1400 11
2.4 Cable-Stay Aerodynamics
Theories and explanations on the mechanisms of cable-stay vibration put forward by
researchers generally address three issues. They are (1) the aerodynamics of inclined cables,
(2) the destabilizing role of axial flow in the wake of an inclined/yawed cable, and (3) the
effect of the upper rivulet formation on the windward surface of the cable. In addition, a)
high-speed vortex shedding, b) modal frequencies, c) dominant modes of vibration, d) low
Scruton numbers, and e) aerodynamic damping and response parameters are considered
responsible for some observed vibrations. Each is discussed in the following subsections.

2.4.1 Aerodynamics of inclined cables

Yawed and inclined cables generally can be considered (a) to be aerodynamically
unstable and (b) to satisfy galloping criteria as a result of the generation of axial flow (along
the axis of the cable) in the wake of the cable (Matsumoto et al., 1995). These criteria can
occur in wind without accompanying rain.

Inclination andYaw Angles

Two basic angular parameters, the cable-stay inclination angle a and the yaw angle
f3 (see Figure 2.1) are defined. The inclination angle a is the angle the cable forms with the
horizontal plane. The yaw angle /)is defined as the angle between the cable's projection on
the horizontal plane and the vertical plane normal to the wind direction. Yaw angle /)is
equal to 0° when the wind is perpendicular to the vertical plane containing the cable; f3 is
greater than 0° when the wind is in the direction of cable descending as shown in Fig. 2.1.
Matsumoto et al. (1995) went further to study section model responses for yaw angles
of 0° to 45° and found that the onset critical velocity that causes vibration is almost
independent of yaw angle. In addition, the response of the cylinders increased with yaw
angles up to 45°. From testing of inclined and yawed cylinders and comparing the results to
yawed-only cylinders, Matsumoto et al. (1995) found that each case yielded essentially the
same response. The equivalent yaw angle, actual yaw angle, and inclination angle, can be
related geometrically as follows in Equation 2-1. Note that for a= 0, /3= /]*.
f3" =sin_, (sin f3 cos a) [2-1]

where f3 * = equivalent yaw angle on a horizontal plane,

f3 = actual yaw angle, and
a= inclination angle.

In uniform flow, the aerodynamics of an inclined and yawed cable (a, fJ) is
equivalent to that of a yawed cable (0°, /]*) having an inclination angle of 0° and a yaw angle
of/]*.

Project 0-1400 12
Wind Angle ofAttack
The wind angle of attack y, as defined in Figure 2.1, can be shown to be:

r tan _ 1 [ - sin a tan¢cosa ]

+---'---.,-- [2-2]
tan(900-/J) sin(90°-/J)

where ¢is the angle that the mean wind vector makes with the horizontal plane.
a, /]are defined in Equation 2-1
Typically, ¢is approximately equal to zero.

Reduced Velocity
For accurate wind tunnel analysis, the reduced velocity of the cable-stay section must
be calculated. Reduced velocity, RV, is defined as follows:

RV= U [2-3]
nD

where n = system natural frequency,

D =characteristic dimension (diameter), and
U velocity ofthe oncoming flow.
Though it appears that the reduced velocity is simply the inverse of the Strouhal
number, precise determination of the vortex shedding frequency is difficult. Therefore, since
large responses are associated with vortex shedding during lock-in, where the vortex
shedding frequency and system natural frequency match, the system natural frequency is
used to define reduced velocity.

Summary ofRelevant Recent Research

Matsumoto et al. (1995) determined that an essential aerodynamic component of
cable-stay vibration is coupling between vertical and horizontal vibration. This was based on
full-scale observations as well as wind tunnel section model studies. A section model was
restrained to vibrate along an inclined plane rather than the vertical plane, and an unstable
response was observed.
Intuitively, as the yaw angle increases, so does the magnitude of axial flow in the
wake of a cylinder. This axial flow interrupts the natural vortex shedding that occurs.
Matsumoto et al. (1995) suggested that a possible mechanism of aerodynamic instability in
cable-stays is the axial flow in the wake of a yawed cable (yawed with respect to the wind
flow direction). It is suggested that the axial flow acts as a splitter plate that, when inserted
in the wake of a yawed circular cylinder, causes an unstable response. The splitter plate
(axial wind flow in the case of cable-stays) produces unsteady circular flow on the leeward
side of the cylinder. This unsteady circular flow occurs above the splitter plate during
downward motion of the cylinder and below the splitter plate during upward motion. The
decreased pressure associated with the unsteady circular flow produces alternating positive

Project 0-1400 13
and negative lift cycles on the cylinder. The inclusion of a water rivulet on the upper
windward portion of the cylinder further increases the separation of flow, thus supplementing
the already occurring unsteady flow. This type of flow produces an effect similar to that of
wake galloping, which has been observed in transmission lines (Hikami, 1988).
2.4.2 Destabilizing role of axial-flow
Galloping instability occurs when the sum total of the inherent system damping and
the aerodynamic damping produced by fluid flow about a body is negative, thereby
producing a response that grows until the limit state of the nonlinear system is reached. This
occurs for cables with or without an upper rivulet. The axial-flow velocity increases with
oncoming velocity at a certain rate depending on the yaw angle P(Fig. 2.1 ). Galloping
occurs when the axial flow reaches more than 30% of the approaching velocity. Galloping
instability has been observed for P> 25°. The instability appears to take place at RV 2 70,
irrespective of P The worst response occurs at a= 0° and P= 45° and decreases for lesser or
higher values of p.

2.4.3 Rivulet effect

The upper rivulet generally is an excitation factor, whereas the lower rivulet is a
damping factor. In other words, the upper rivulet causes destabilization or stabilization of
response depending on its position whereas the lower rivulet generally is stabilizing. (}> 60°
is considered critical for P= 45° (Matsumoto et al., 1995). Model tests by Bosdogianni and
Olivari (1996) found that the shape of the rivulet did not change the response significantly
and the position of the lower rivulet played an insignificant role. Flamand (1993) conducted
wind-tunnel tests on a rigid model of diameter 16 em (6.3 inches) covered with standard
polyethylene (PE) casing. In order to simulate the rain formation and wind-rain-induced
cable vibration, clean PE casing and PE casing with soot surface conditions were tested. In
the wind speed range tested, 6 to 13 m/s (13 to 29 mph), the upper rivulet was observed to
form only with PE casing plus soot at the wind speed of about 10 m/s (22 mph), and
significant vibration with limited amplitude occurred. The soot was generated from fuel-oil
combustion. It was designed to reproduce the deposition of atmospheric pollutants on cable-
stays. The sooted cable surface was no longer water-repellent and therefore facilitated rivulet
formation. The formation of an upper rivulet was observed only in the wind speed range of 7
to 12 m/s (16 to 27 mph).
In general, the maximum amplitude response of the velocity-restricted type tends to
decrease and the onset velocity of galloping instability increases with increasing Scruton
number (Matsumoto, 1998). Large amplitude galloping vibration can occur at Scruton
numbers as high as 200, which implies that very high structural damping might be required to
effectively suppress cable-stay vibrations, with or without rain, of prototype cable-stays at
high natural wind speeds.
Hikami et al. (1988) studied a section model in the wind tunnel and simulated rain by
using spray nozzles. Hikami et al. (1988) used the same wind speed and Reynolds number as
the prototype, but the Scruton number (see Section 2.4.7) for the wind tunnel model was half
that of the prototype due to differences in mass and damping.
Yaw and inclination angles in this study were ±45 degrees each. The natural
frequency of oscillation for this section model and suspension system was 2 Hz. As will be

Project 0-1400 14
described in Chapter 3, an unstable response was found for the case with rain where the wind
was in the direction of declination, and a stable response was found for the case with rain
where the wind was in the direction of inclination. The adverse response occurred in the
wind speed range of9 to 13 m/s (20 to 29 mph). 2 Hikami et al. (1988) also found that a more
adverse response occurred for systems with frequencies of approximately 1 Hz. It was
observed that at wind speeds below the onset velocity of cylinder vibration, the rain water
droplets ran down around the circumference of the cable section forming a rivulet on the
lower, leeward side of the cable. Above the onset velocity, a drag force acts upon the
individual water droplets that overcome friction and gravity and an upper rain rivulet forms
in addition to the lower rivulet. Focusing on the formation of the rain rivulets, Hikami et al.
(1988) found that when the stay declines in the direction of the wind within a limited wind
speed range, the upper windward rivulet forms in addition to the lower leeward rivulet. The
aerodynamic force generated with the upper rivulet acted to excite the cable while the
aerodynamic force generated with the lower rivulet acted to dampen the motion of the cable.
As the stay vibrates due to upper and lower rivulet formation, the upper and lower rivulets
oscillate with the same frequency as the cable motion (Hikami et al., 1988).

2.4.4 High-speed vortex shedding

Matsumoto et al. (1995) considered high speed vortex shedding as a phenomenon that
could cause excessive vibration. During high speed vortex shedding, non-coherent von
Karman vortex sheets shed at different frequencies (n 1, n2) along the yawed cylinder length.
At random intervals, a single von Karman vortex sheet is shed coherently along the entire
length of the cylinder whose frequency is lower than that of the smaller, non-coherent
vortices (ns = n1-n2). Neither axial flow or formation of a rivulet is a requirement for this to
occur. The intermittent 2-D von Karman vortices that are shed at a frequency ns = n 1 - n2
correspond to a reduced velocity (RV) of 40, or any integral multiple of 40.
Perhaps contrary to the work ofHikami (1988) and Matsumoto (1995), Verwiebe
(1998) emphasized the importance of the rain water rivulet in changing the forces on the
cable-stay. Verwiebe (1998) presented evidence that the effective shape of the cross section
of the cable-stay depends on the momentary location of the rain water rivulet(s). The
rivulet(s) oscillates around the cable-stay surface and changes shape continuously, thus
modifying the cross sectional shape in a time dependent manner. As the cross-sectional
shape varies with time, so do the pressure distribution on the cylinder surface and the
resulting forces on the cable-stay. If the wind force incident upon the cable is varying with
the same frequency as the cable oscillation and with the same sign, then positive work is
done and vibration of the cable system ensues. Verwiebe (1998) suggests that initial
oscillation of the cylinder is required before the positive work done by the rain and wind
interaction can cause an increase in the cable-stay oscillations. The range of wind speeds
where oscillation occurs is about 5 to 20 m/s (11 to 45 mph) -which includes the speeds
reported in previous experiments ofHikami et al. (1988) and Matsumoto et al. (1995). Note
that Verwiebe (1998) performed experiments at yaw angles of0° and 90° with inclination of
30° only and the frequency of vibration was 8.9 Hz, which is well above the frequency of2
Hz or less used by Hikami et al. (1988) and Matsumoto et al. (1995). Verwiebe (1998)

2
Actual wind speeds, rather than the non-dimensional reduced velocity (RV), are used when precise
data of the testing appamtus is not available.

Project 0-1400 15
argues that the motion of the rivulets is paramount to the excessive vibration of the cable
system and points out that studies on fixed cross sections may only yield instantaneous
results and show galloping instability which is not the actual phenomenon.
2.4.5 Modal frequencies
In the field, natural vibration modal frequencies and damping values of the cable-
stays generally are determined by manually exciting the cable-stay and analyzing the
resulting vibration data taken from accelerometers attached to the cable-stay. Reported
natural frequency values vary from 0.6 Hz for the longest Fred Hartman cables and up to 11
Hz for the shortest Veterans Memorial cables (Whitlock, Dalrymple, Poston, and Associates,
1999). Damping values in modes 1 to 3 ranged from 0.15% of critical for the longest stays to
0.6% of critical for the shortest stays. Flexibility and low inherent damping combine to make
the cable-stays vulnerable to vibration caused by exterior influences such as wind and rain.
The natural vibration occurring during a pure wind condition appears to differ from
the von Karman vortex oscillation phenomenon. The frequency of vibration is well below
the critical frequency of von Karman vortex shedding for a cylindrical body (Hikami et al.,
1988). It has also been observed that most of the large amplitude stay vibration occurs when
the wind is in the direction of declination of the stays (Hikami et al., 1988; Main et al., 1999).
Significant vibrations also occur for winds in the direction of stay inclination, but do so
during instances of very heavy rainfall (Main et al., 1999). Hikami et al. (1988) observed
that at low wind speeds during rainy conditions, the water droplets that contact the stay
collect and form a rivulet on the lower surface of the stay. This rivulet oscillates
circumferentially with the same frequency as the stay vibration. Within a certain wind speed
range, 8-15 m/s (18-34 mph), two rivulets may form on the upper and lower portions of the
stay. Large amplitude vibration during rain and wind events occurred in the same wind
speed range (Hikami et al., 1988). This "velocity-restricted" response was also evident in the
full-scale measurements ofMain et al. (1999). Most studies to date have suggested that the
predominant vibration occurs transverse to the oncoming flow; however, full-scale data
reported by Main et al. (1999) have shown that the stay oscillations under rainfall actually
have a degree of two-dimensionality. That is, the stays vibrate in an elliptical pattern rather
than a predominately vertical pattern. As will be discussed in the next chapter, TTU
researchers developed a 2-D force damper apparatus precisely to capture this elliptical
pattern in the wind tunnel model analysis (see Section 3 .2).

2.4.6 Dominant modes of vibration

Main and Jones have instrumented the Fred Hartman, where significant wind-rain-
induced vibrations have occurred (Main, et al., 1999). "Triggered" events are recorded when
a predetermined acceleration and/or wind speed threshold is exceeded. It has been noted that
each individual cable seems to vibrate at a particular lower-mode shape. For example, a long
Fred Hartman cable-stay, 183m (601ft) in length with a fundamental frequency of0.65 Hz,
vibrates predominately in the 3rd mode, though not in the first two. Similarly, a mid-length
Fred Hartman cable-stay, 87 m (286ft) in length with a fundamental frequency of 1.2 Hz,
has been found to vibrate predominately in the 2nd mode, though not in the first.

Project 0-1400 16
 Higher modes of vibration in the cables also have been found on both the Fred
Hartman and the Veterans Memorial. It is generally accepted-though unproven-that
cables vibrating in lower modes cause more damage than cables vibrating in higher modes,
since lower-mode vibrations generally cause larger displacements. However, it is entirely
possible that higher mode vibrations occur often enough and produce greater deflection
curvature, to produce significant fatigue loadings, and possibly even greater loadings, on the
cable-stays due to cycles of reversed stressing.
Considering the physics of the rivulet formation, it is difficult to conceive that the
rivulet is consistently located at the most critical location along the full cable length. It is
possible that the rivulet primarily causing the vibration at the lower wind speeds forms at the
critical location only over a partial cable length. This could explain why there is a preference
for certain lower-modes to vibrate.

2.4.7 Scruton number

The Scruton number is an important parameter in model tests. It is a non-dimensional
parameter that reflects the a) relative effects of excitation force, b) mass/stiffness property
and c) damping characteristics of the dynamic responses of a cable-stay in wind.
Researchers use the Scruton number to relate wind tunnel tests of scale section models to
full-scale cable behavior. The Scruton number is defined in this report as follows:
4nm~
Sc = ~m
pD~
[2-4]
where m =mass per unit length of the cylinder,
(m critical damping ratio (mechanical),
D = diameter of the cylinder, and
p= air density.
In general, the maximum amplitude response of the velocity-restricted type tends to
decrease and the onset velocity of galloping instability tends to increase with an increasing
Scruton number (Matsumoto, 1998). Large amplitude galloping vibration can occur at
Scruton numbers as high as 200, which implies that very high structural damping might be
required to effectively suppress cable vibrations, with or without rain, of prototype cable-
stays at high natural wind speeds.

2.4.8 Aerodynamic damping and response parameters

The Glauert-Den Hartog criterion for galloping is based upon static-force
coefficients. However, the galloping phenomenon can be also described as an aeroelastic
instability that is similar to the single-degree-of-freedom (SDOF) flutter found in the
linearized domain; true for incipient motions only. Thus, the critical speed for the onset of
galloping can be precisely found by the flutter theory.

3
Sometimes the Scruton number is defined without the 471: term shown in Equation 2-4.

Project 0-1400 17
 The cause of the SDOF flutter can be explained by an overall, or total, damping <;;T of
a structural system. It is a summation of the mechanical damping <;;m and the aerodynamic
damping <;;aero of the system, and can be written as:

[2-5]

In Equation 2-5, t;m is measured at zero wind speed and <;;aero is dependent on a
reduced velocity, RV. The aerodynamic damping (t;aero) is a function of the cross-sectional
shape of the structure and is often plotted as a non-dimensional number H 1* (RV), known as
a flutter derivative. A more detailed description of flutter derivatives can be obtained in
Simiu and Scanlan (1996). Eight flutter derivatives including the H 1* were used in the
analysis of bridge-deck flutter (Sarkar, 1992). H1 * (RV) can be defined as:

[2-6]

where pis the density of air,

m is the mass per unit length of the structure,

B is the representative across-wind or along-wind dimension of the

cross section.

Other terms in Equation 2-6 are defined in Equation 2-5. A positive H 1* (RV) is an
indication ofaeroelastic instability or SDOF flutter. The critical reduced velocity, RVer, is
the value ofRV at which t;aero(RV) is negative so as to nullify t;m (i.e., r;r = 0). RVer gives the
value of critical wind velocity Ucr at which aeroelastic instability or initiation oflarge
vibration occurs.
While Ht * can be obtained from wind-tunnel tests using section models, the
aerodynamic damping in any particular mode of vibration of the prototype cable can be
predicted by:

_pD2 n;Jif/2(x)dr [2-7]

2mL 0

where m is the mass per unit length of the cable,

L length of the cable
D = diameter of the cable, and
\jl(x) is the mode shape of the cable.

Project 0-1400 18
 Since the mode shapes of a cable are sinusoidal, the value of the integral in Equation
2-7 is L/2. The total damping of the cable can be calculated using Equation 2-5.

2.5 Previously Reported Field Observations on Bridges

Though field observations of the cable-stay vibration problem have become
somewhat more common recently, traditionally, documented cases have been relatively rare.
The following documents major recorded cable-stay vibrations divided into appropriate
subsections.

2.5.1 Velocity-restricted vibrations

Cable-stay vibrations on the Brotonne Bridge, with a central span of320 m (1050 ft),
over the Seine River in France have occurred at a wind speed of about 15 m/s (34 mph) at
20° to 30° with respect to the bridge axis (Wianecki 1979). Also on the Meikonishi Bridge,
with a main span of 405 m (1330 feet) in Japan, cables that experienced no significant
vibration under wind action alone were observed to experience large-amplitude oscillations
in wind-rain-induced vibration conditions (Hikami and Shiraishi 1988). Wind-rain-induced
vibrations were at much lower frequencies with much higher amplitudes than that of the
vortex-induced vibrations. It seems that there was a range of wind speeds, 7 to 15 m/s (15 to
34 mph) and a narrow range of wind angles that induced significant vibration for individual
cables. The observed frequency range was 1 to 3Hz. Typical diameter of the cables was 14
em (5.5 in). Vibrations occurred when cables were declined in the direction of wind. Only
PE (polyethylene) tube lapped cables with 14 to 20 em (5.5 to 7.9 inch) diameters
experienced wind-rain-induced vibrations, Matsumoto et al. (1992). Wind-rain-induced
vibrations occurred at a reduced wind velocity, RV, in the range of20 to 90, i.e. 6 to 18 m/s
(13 to 40 mph). At the time oflarge-amplitude vibrations, wind was blowing skewed to the
bridge axis. Also, cables that were located on the leeward side of the bridge pylons vibrated.
Vibrations were almost in-plane. Maximum amplitude was up to 2m (6.6 ft). At the
Tenpozan Bridge in Japan, Matsumoto et al. (1995) observed that the longest cable of 186m
(610ft) experienced wind-rain-induced vibration in the first and second modes in the
velocity range of8 to 12 m/s (18 to 27 mph). The amplitude of the second mode vibration
was lower, approximately 1/4 that of the first.

2.5.2 Single-moded vs. multiple-moded vibration

Cable vibrations up to the third mode were observed on the Brotonne Bridge,
Wianecki (1979). Also, eye-witness accounts, observed from approximately 100m (328ft)
from the bridge, reported vibrations of30 em (12 in) in single amplitude in the first mode.
On the Meikonishi Bridge, cable-stay vibrations were mostly of a single mode, and there
were only a few occasions that two or three modes were involved (Hikami and Shiraishi
1988). Power spectrum density diagrams of the response of the longest cable on the Higashi
Kobe Bridge indicated multiple-moded vibration at reduced velocities of approximately 400,
200, 120, 80 and 40 (Matsumoto et al. 1995). Cable-stay vibrations were usually multiple-
moded, non-stationary and exhibited wave-propagation type oscillations (Matsumoto et al.,
1998).

Project 0-1400 19
2.5.3 High wind speeds without rivulet formation
At a mean wind speed of 40 m/s (89 mph) on the Higashi-Kobe Bridge, the inclined
cables having 12 axial protuberances, were observed to experience vibrations similar to
wind-rain cable vibrations. The axial protuberances were designed to suppress wind-rain-
induced vibrations. In this particular case, rivulets did not form because of the surface
treatment. It was not certain if the response was velocity-restricted or not, Matsumoto
(1998).

2.5.4 Effect of rivulet

At the Meikonishi Bridge in Japan, cable-stays that were stable in windy conditions
vibrated significantly in wind-rain conditions (Hikami and Shiraishi, 1988). Water rivulet
formation on the lower surface of the cable-stays was observed in the field. The lower
rivulet was shifted to the leeward side and oscillated in the circumferential direction of the
cable-stays. Later laboratory tests on a rigid cable model having the same diameter and
surface conditions as the prototype revealed the formation of another rivulet along the upper
windward surface of the cable when the cable was descending in the direction of wind.
According to Matsumoto et al. (1995), inclined cables in the leeward side of a bridge pylon
can have both an upper and a lower rivulet. The balance of the gravity force, wind pressure
and water-surface tensile force acting on the rivulet determined the location of the upper
rivulet. Upper rivulets at certain locations were aerodynamically destabilizing.

2.5 .5 Effects of turbulence

Matsumoto et al. (1992) noted that all the five bridges on which cable vibrations of
large amplitude occurred were located near or facing the sea, i.e., the turbulence level in the
approaching wind was quite low. The turbulence could play different roles in either reducing
the amplitudes (stabilizing effect) or reducing the critical-onset velocity (destabilizing effect)
depending on the yaw angle, presence of rivulet, etc.

2.5.6 Low-damping in lateral oscillation of cable-stays

Hikami and Shiraishi (1988) reported that a critical damping ratio of 0.1% to 0.4%
was measured for the cable-stays of the Meikonishi Bridge. A damping ratio of 0. 5% or
above might be needed to effectively suppress cable vibration of the velocity-restricted type
(Geurts et al., 1998). Much higher damping might be needed to reduce cable vibrations at
high wind speeds (Matsumoto, 1998).

2.5.7 Reynolds number effect

Many cable vibrations have been observed in the range of Reynolds number of
60,000 to 200,000, i.e. the sub-critical range. The combination of the wind speed range of 7
to 15 m/s (15 to 33 mph) and the cable diameter range of8 to 20 em (3 to 8 in) might be
associated with the importance of the Reynolds number to velocity-restricted vibrations at
relatively low wind speeds (Matsumoto, 1998).

2.6 Mitigation Devices

Currently, cable-stay oscillations caused by wind-rain-induced aerodynamic forces
are controlled by one, or a combination of the following methods: (1) single-point

Project 0-1400 20
mechanical dampers, typically at the base of each cable, (2) restraining cable devices
connecting adjacent cables at various locations along the length of the cable, resulting in a
reduced effective length for each cable and/or (3) aerodynamic damping approaches such as
grooves, protuberances or circular rings. The former method is considered a "concentrated"
damping mechanism, while the latter two are considered "distributive."

2.6.1 Single-point mechanical dampers

The traditional method-and the currently selected Fred Hartman retrofit solution-is
a combination of passive single-point mechanical dampers and restraining devices.
Mechanical dampers generally are linear viscous mechanisms, somewhat similar to an
automobile shock absorber. However, they also can be non-linear, computer-controlled
mechanisms. Mechanical dampers are a proven technology. However, they generally: (1)
can be expensive to install, (2) may need periodic maintenance, and (3) typically require
substantial cable-stay displacements to occur before the damping mechanism becomes
functional.

2.6.2 Restrainers
Restrainers are employed to tie adjacent cable-stays together at discrete points along
the cable. Restrainers are effective solutions because one cable adjacent to another
oscillating cable generally will not be oscillating. (However, this may not hold true for
parallel cable systems like the Veterans Memorial.) When adjacent cables do oscillate
together, many times they will vibrate out of phase or in different modes from each other. In
these typical cases, restrainers are able to utilize the stiffness of adjacent cables to prevent a
particular cable from oscillating. If the restrainer is unable to prevent oscillations, it
continues to be considered beneficial in that it causes the cable-stay to vibrate at higher
modes, with less deflection amplitude, as it "fixes" intermediate nodal points. Again, though
a higher mode vibration is visually less dramatic, significant fatigue loadings can occur.
Restrainers also are a proven technology. However, they are fairly difficult to install-
particularly at cable-stay heights generally required. Also, restrainers have had problems due
to failure through loosening of the attachments to the cable-stays.

2.6.3 Aerodynamic devices

While mechanical dampers and/or restrainers are effective, they require routine
maintenance and can be costly to implement. Also, they can decrease the aesthetic appeal of
a cable-stayed bridge. Further, mechanical dampers typically can be optimized for a few
modes of vibration only. Therefore, it may be desirable to determine a solution to the cable-
stay vibration problem through aerodynamic means.
Aerodynamic devices potentially have certain advantages. They a) can be effective
over a wide range of wind speeds, and may perform even better at higher wind speeds, if
properly designed; b) are generally cost-effective and demand little maintenance efforts, thus
they can function reliably; c) can function over a wide range of vibration modes, d) can be
designed to be aesthetically pleasing; and e) can reduce the effect of the aerodynamic forces
before the cable-stay begins to vibrate substantially, where mechanical devices must dissipate
energy of the cables that are already vibrating.

Project 0-1400 21
 Various forms of aerodynamic solutions to the vibration of smooth-surfaced, circular
cables have been sought. While some can be adopted only at the design stage, others are
feasible for retrofitting as well. Aerodynamic countermeasures usually modify the surface of
the cable-stay cross section to improve its aerodynamic performance in terms of reducing the
excitation from the moving air or increasing the aerodynamic damping. Matsumoto et al.
(1995) listed three types of cable-stay surface/cross section modifications: 1) surface
dimpling, 2) parallel axial protuberances, and 3) elliptical plates. A variation of the
elliptical plate is helical strak:e which has been used successfully on chimneys to reduce
vortex-induced vibrations. Figure 2.4 shows some aerodynamic countermeasures
investigated by Matsumoto et al. (1997).
In practical applications of aerodynamic countermeasures, helical wires have been
used at the Normandy Bridge; axial protuberances were successfully designed to reduce the
cable vibrations at low wind speeds on Higashi Kobe Bridge; dimpled surface treatment has
been made on the Tatara Bridge according to Matsumoto ( 1998). On the other hand, rigid
model tests conducted by Matsumoto et al.(1997) found that elliptical plates could be very
effective in reducing the dynamic response of a cable. In a model test by Flamand (1993),
wires of diameter l.Smm were introduced to a model cable of 100 mm at intervals of30° to
disrupt the circumferential motion of the rivulet. Spiral (helical) wires were also tested.
Combinations of different wire diameters and pitches were studied. The wire of 1.3 mm
wound at 0.3 m pitch was found to be the most effective. Double helix was also tested.
These aerodynamic devices are designed to prevent the formation of a continuous
upper water rivulet, which was considered to make the cross section aerodynamically
unstable, or to interrupt the axial flow in the wake of an inclined cable. The axial flow was
considered responsible for instability of cables at high wind speeds without the rain.

Project 0-1400 22
 D-S4mm
(a)Rigid Cable Model(RS4)
(b)Rigid Cable Model
with Axial Protuberances(R54AP)

(d)Risid cable Model

with Elliptical Plates
(RSSE)

(c)Rigid Cable Model

with Dimples(R.SSD)

Fig. 2.4 Various aerodynamic countermeasures investigated by Matsumoto et al.

Project 0-1400 23
 CHAPTER3
TTU WIND TUNNEL TESTS

The TTU wind tunnel studies conducted in Colorado State University and Texas Tech
University wind tunnels investigated helical strak:es, elliptical rings and circular rings. Wind
tunnel tests were conducted on elastically supported section models of the cable-stays. The
section models were allowed either one-degree (vertical oscillation), as reported in Sarkar et.al.
(1998), or two-degrees of freedom (vertical and torsional oscillation), as reported in Sarkar and
Gardner (2000). The focus of this portion of the research project has been to observe the
characteristics of cylinder oscillation under the influences of rain and wind and to determine
the mechanism of flow about the cylinder that causes adverse vibration. In addition, based on
these wind tunnel studies, methods to control the vibrations of the cable-stays have been
proposed. To simulate varying wind directions incident upon the cable-stays, the section
models were tested with various yaw angles.
The primary goal of the wind-tunnel tests has been to study the effectiveness of cable-
stay mitigation devices, i.e., helical strake, elliptical ring and circular ring, in reducing the
amplitude of vibration to an acceptable level. A parametric study of these devices was
performed (diameter, pitch, etc.) to find optimum configurations. A secondary goal has been
to identify the aerodynamic damping with and without mitigation devices in order to quantify
the amplitude of vibration and the critical speed for the onset of vibration in the prototype.
Experimental designs, wind tunnel tests conducted in CSU and TTU wind tunnels, and
results are reported in this chapter.

3.1 Past Observations on Wind Tunnel Models

A great deal of information was known prior to this study-some of which was verified
during the current TTU wind-tunnel tests. It is a well known fact that yawed or yawed and
inclined (i.e. yawed-inclined) cables (without rain) have a tendency to exhibit galloping
beyond a critical wind speed because of the axial flow that is generated in the wake of the
cable. These cables exhibit only divergent-type response at higher wind speeds. In addition,
an upper rivulet can form at relatively low wind speeds to make the cable-stay prone to large-
amplitude vibration.
Both elastically supported rigid models (or section models) and flexible models of
cable-stays have been used in wind-tunnel tests to assist in understanding the fundamental
features of cable vibration. Most researchers have employed rigid models. In the case of rigid
model tests, yawed models and yawed-inclined models have been utilized. These two types of
models are aerodynamically similar. The advantage of rigid models is that they are easier to
build and set up. They have been useful to reveal the most essential features of wind-rain-
induced cable vibration. The main disadvantage is their incapability of realistically
reproducing coupled vibrations of more than one mode. This is a limitation in the modeling as
slender cables contain multiple lateral modes of vibration and can be excited to vibrate in
combinations of different modes.

Project 0-1400 25
3 .1.1 Response of elastically supported rigid models
A rivulet at an angle Bfrom the stagnation point from the given wind direction was
shown previously in Figure 2.2. At an upper rivulet angle of approximately B= 70°, the cable
can vibrate severely. Though lower rivulets also form, these do not appear to produce
instability. Considering the elliptical cable cross-section along the wind direction, combined
with the "bulge" of the upper rivulet, one can see that an "airfoil shape" becomes the leading
edge into the wind. Thus, for a declining cable-stay, lift is produced.
Sketches shown in Figure 2.3 suggest that cable responses can be divergent and/or
velocity-restricted. For bare cables, divergent responses were observed only for yawed angles
greater than 25°. Depending on the location of the artificial rivulet, models might exhibit both
types of response or no significant response at all. While the introduction of an artificial
rivulet is convenient and instrumental in the understanding of the role of the rivulet in
prototype cable vibrations, it must be understood that rivulets form on real cables only under
certain combinations of wind direction, wind speed, cable inclination, cable surface condition,
etc. Further, rivulets on prototype cables were observed to be oscillating in the circumferential
direction, and thereby interacting with the cable motion, whereas the artificial rivulet is
stationary. Therefore, caution must be exercised when interpreting the wind-tunnel results
from model tests using an artificial rivulet. 4
A possible explanation of the velocity-restricted response as observed in the full-scale
monitoring of the vibrations at low wind speeds follows. The scenario is wind accompanied by
rain. For commonly encountered inclined cables with diameters varying between 8 to 20 em (3
to 8 inches), as wind speed increases beyond 7 m/s (15 mph), an upper rivulet can form on
cables that are descending in the wind direction. The mean location of the rivulet shifts away
from a stationary point with increasing wind speed. When the rivulet is formed at certain
locations, the cable shows a tendency to have a divergent response and exhibits galloping with
large amplitude motion. The large-amplitude motion remains limited because the cable has a
non-linear stiffness. Not all cable-stays in a cable-stayed bridge will exhibit this behavior at
the same time as they have different attitudes toward the wind. As the wind speed increases
further, the upper rivulet shifts further away from the stagnation point and is eventually blown
off the cable at a wind speed greater than 15 m/s (33 mph). Thus, the mechanism that caused
the aeroelastic instability disappears. The result is a velocity-restricted response within 7-15
m/s wind speed. Beyond 15 m/s of wind speed, the rivulet is out of the scenario and some
inclined and yawed cables will still have a tendency to exhibit large amplitude vibration, but
only beyond a critical speed. The large amplitude motions will not reduce with increasing
wind speed as occurred in the lower-wind speed range. The actual motion of the cable is along
an inclined plane depending on the angle of attack and is more complex than SDOF galloping-
type response because of either the interaction of the oscillation of the rivulet with the cable
motion or the cable's richness in vibration modes.

3 .1.2 Response of flexible models

Results from wind-tunnel tests on a flexible cable model are reported by Matsumoto
(1998). Depending on: a) the wind speed, b) different modes, or c) combinations of the two,

4
Nonetheless, when one compares field results presented in Chapter 5 of this report with the wind-
tunnel results presented in this chapter, strong similarities in cable-stay behavior are observed.

Project 0-1400 26
the cable model can be excited. The general tendency is that higher modes are excited at
higher wind speeds. Wave-propagation type cable vibrations also have been observed.
3.2 CSU Wind Tunnel Tests
The primary goal of the wind-tunnel tests was to study the effectiveness of the
mitigation devices, i.e., helical strake, elliptical ring and circular ring, in reducing the
amplitude of vibration to an acceptable level. A parametric study of these devices was
performed (diameter, pitch, etc.) to find optimal configurations. The secondary goal was to
identify the aerodynamic damping with and without the mitigation devices in consideration in
order to quantify the amplitude of vibration and the critical speed for the onset of vibration in
the prototype.

3.2.1 Experimental Design and Wind-Tunnel Tests

The test setup (Fig. 3 .I) consisted of a yawed (in horizontal plane) section model of a
cable (10.2 em or 4" diameter and 0.97 m or 3'-2" length) that could vibrate in a vertical plane
along a single degree of freedom.
To simulate three-dimensional effects ofthe wind-model interaction, the model was
tested without any end plates. The model was suspended elastically (with springs) from a pair
of force balances fixed to two separate small frames. This force balance-spring-model
combination was fixed to an outside frame 1.73m x 1. 73m (5'8" x 5'8") in cross section and
1.37m or 4'6" in depth (along-wind dimension). The dimensions of the outside frame were
chosen to fit inside the 1.83 m x 1.83 m (6' x 6') Meteorological Wind Tunnel (MWT) at
Colorado State University (CSU).
All dynamic tests were performed with the model fixed at a = 0° and ~ 3 7° in smooth
flow without any rain (Fig. 3.1 ). This was the maximum value of~ obtained with the
experimental setup which was close to the worst case as reported in earlier studies (a = oo and
~ = 45°). The effect of the upper rivulet on the stability of vibration was studied by fixing an
artificial rivulet (semi-elliptical cross section, 10 mm in width and 5 mm in height) on the
upper side of the model (Fig. 3.2) Two locations with angles of8 = 65° and 70° (Fig. 2.1) were
tested to verifY the effect of the location on the stability. The model was tested with and
without any mitigation devices. The effect of the Scruton number on the amplitude was tested
as well. The first set of experiments were done with a low Scruton number, Sc = 7.6
(m=3.02/kg/m, sm 0.25%, D = 0.102 m, n 1.0 Hz), and the second set of experiments were
performed with a high Scruton number, Sc 54.6 (m 7.85 kg/m, sm 0.69%, D = 0.102 m,
n 1.0 Hz). The Scruton number for cable A23 of the Fred Harman bridge (next in length to
the longest cable A24) is estimated as Sc 44.7 (m 74.7 kg/m, assuming the first-mode
damping s = 0.15%, D 0.16 m).
The experimental setup was chosen as a single-degree-of-freedom (vertical) model,
similar to the majority of past researchers. Although this simple setup does not capture all the
features of the motion of the prototype cables under wind and rain, it does capture the
aerodynamic phenomena causing instability and the effectiveness of the devices in damping
out the motions.

3.2.2 Aerodynamic Devices Tested

The aerodynamic devices tested are (i) Aerodynamic Plate-Damper (Fig. 3.3a), (ii)
Helical Strake (Fig. 3.3b), (iii) Elliptical Ring (Fig 3.3c), and (iv) Circular Ring (Fig. 3.3d).

Project 0-1400 27
 Fig. 3.1 Wind-Tunnel Model Setup in the Meteorological Wind Tunnel at CSU
(viewing downstream)

Fig. 3.2 Model at a =0° and ~=0° showing an artificial-upper-water rivulet

a. Aerodynamic Plate Damper b. Helical Strake (\orientation)

J
c. Elliptical Ring (\ orientation) d. Circular Ring

Fig. 3.3 Aerodynamic devices tested

Project 0-1400 28
The width of the plate damper was 3D with 0.5D spacing between the center of the plates and
the model. Wires ofD/20 diameter were fixed on the plates as shown in Fig. 3.3a to prevent
the formation of water rivulets. The elliptical rings were made out of wires ofD/20 diameter
placed at 45° angle to the axis ofthe model at 1.5D spacing (Fig. 3.3c). These were inclined
counterclockwise if viewed in the along-wind direction(\). The reverse inclination(/) was also
tested. Helical strakes were made of wires that were helically wound around the model (Fig.
3.3b). Two different wires of diameters D/20 and D/8 placed at a pitch of 1.5D, 3D, and 5D
were tested. The helical strakes were also tested in both orientations (\ and I as viewed in the
along-wind direction). The inclination angle (<p) of the helical strake with respect to the axis of
the model can be calculated as <p [tan (Circumference ofthe model/Pitch of helical
strake)]. Fig. 3.3c shows a D/8 helical strake at a pitch of 3D(\ orientation). The angle <p for
this case can be calculated as tan -I (n/3) 46.3°.

3.2.3 RMS Response

The root mean square (RMS) values of the response of the section model at different wind
speeds were calculated from the time histories of the displacement along the vertical degree of
freedom (60-sec record sampled at 50Hz). These are shown in Fig. 3.4 for all cases
considered.
The interpretation of the wind-tunnel RMS response tests with conclusions follow:
• The model (a= 0°, ~ 37°) without any rivulet showed galloping instability at
U/nD::: 80 (Fig. 3.4a).
• The model with artificial rivulet (upper) at 8 = 70° showed galloping instability at
U/nD :::60 and velocity-restricted response (a crest in the response curve) at a
reduced velocity, U/nD, slightly below 40, similar to the one observed by
Matsumoto et al. (1995) (Fig. 3.4a). The artificial rivulet (upper) at 8 = 65°
stabilized the response compared to that of the model without any rivulet (Fig.
3.4a). Thus, the location of the upper rivulet is important. This was also observed
by Matsumoto et al. (1995).
• The aerodynamic plate-damper produced the anticipated aerodynamic damping
(positive) like an airfoil along the vertical degree of freedom. However, its
effectiveness reduced with increasing angle of attack (y). Further, it showed
instability along torsional degree of freedom beyond a certain U/nD for a range of
angles of attack not including y = 0. The implication is that the aerodynamic plate-
damper cannot be used in the case of a cable that is relatively flexible in torsion and
where there can be large variations in the angle of attack.
• The Scruton number affected the response of the model as depicted in Fig. 3.4b.
The model with a lower Scruton number produced a higher response than the one
with higher Scruton number. This behavior is typical of velocity-restricted
response such as Karman-vortex-shedding type, although the aerodynamic
phenomenon producing the large amplitudes in this case is different.
• The diameter of the wires constituting the helical strake did not seem to have any
significant influence on the response (Fig. 3.4c). Hence, a smaller diameter wire

Project 0-1400 29
 such as D/20 can be used. The pitch influenced the response but only beyond a
certain critical value. The 0/20 helical strake with pitch of values 1.50 and 30
performed differently (Fig. 3.4c ). The combination of the diameter and the pitch of
the helical strake is important.

General Comparison

1600 .00
--No Rivulet
1500.00
1400.00
1300.00 -e- Rivulet at 65
1200.00
1100.00
-o---- Rivulet at 70
~ 1000.00
E
~ 900.00 ~
0 800.00
0 - - Helical Strake(0/20,1.50) i
0 I
..... 700 .00
)(
600 .00
Vi ---- Elliptic all
:!E 500.00
a::: Ring(0/20, 1.50)
): 400 .00
300.00 --Elliptical
200.00 Ring(Reverse,0/20,1.50)
100.00
-+- Circular Ring(0/8,1.50)
0.00
0 50 100 150 200 250' - - - -·---- ~- -----~~------
K ( = U/nO)

(a) Sc =7.6

Effect of S c ru ton N u m be r

400 . 00
E 350 . 00 -circular
~
0 300 . 00 Ring(0/8,1.50)
0
0 250.00 Low Sc
)(
200.00
150 . 00 - circular
Vi
:!E 100 . 00 Ring(0/8,1.50)
a::: 50 . 00 High S c
): 0 . 00
0 50 10 0 150 200 2 50

K ( = U /n 0)

(b) Low Sc=7 .6, High Sc=54.6

Fig. 3.4 (a), (b) Vertical displacement response (rms) of

section model (0= 10.2 em)

Project 0-1400 30
 EffectofPitch and Diameteron HelicaiStrake
--Helical
e
~
400.00
350.00 S tra k e (D /2 0, 3 .0 D )
0 300.00
0
0 250.00 ----- H elica I
"'><"'
-
( /)
:i
200.00
150.00
100.00
Strake(D/20,1.50)

--6---Helical
-
0::
>
50.00
0.00
0 50 100 150 200
S tra k e (D /8,3 .0 D)

~Helical
K ( = U /n D) Strake(D /8,5.00)

(c) Sc=7.6

Comparison ofHelicaiStrake and Circular Ring

e 500.00

-
0
0
0
(,)
400.00

300.00
---.-Helical Strake
(0/8,1.50,\)
..-
>< 200.00 --Helical
'iii Strake(0/8,1.50, /)
:E
-
0::
>
100.00

0.00
0 50 100 150 200 250
---fr- C ire u Ia r
Ring(0/8,1.50)
K ( = U In 0)
(d) Sc=54.6

Effect of Pitch on Circular Ring

300.00
e
~
g 200.00 +----····················---------;;d-----1 --circular
0 Ring(0/8,1.50)
"'><"'
--Circular
Ui 100.00
:i Ring (0 /8,3 .0 0)
0:::
>
0 50 100 150 200 250

K ( = U/n D)

(e) Sc=54.6
Fig. 3.4 (c), (d), (e) Vertical displacement response (rms) of section model (D=l 0.2 em)

Project 0-1400 31
 • The orientation of the helical strake influenced the response. Both orientations (/ or
\)produced acceptable or visibly low responses (Fig. 3.4d) up to U/nD=250 which
correspond to 61 mph for the first mode of vibration and 122 mph for the second
mode of vibration of cable A23 of the Fred Hartman bridge.
• The circular ring performed better than the helical strake (in both orientations) if the
entire range ofU/nD from 0 to 250 is considered (Fig. 3.4d). A pitch of 1.5D and
3D for the circular ring did not produce any significant difference in response (Fig.
3.4e).
• The elliptical ring performed well in one orientation(\) but produced instability at
U/nD > 20 in the reverse orientation(/) (Fig. 3.4a). Since the wind angle of attack
can vary significantly, this aerodynamic device is not feasible.
3.2.4 Measurement and Prediction of Aerodynamic Damping
The damping of a section model can be estimated from its decayed response in free
vibration along a single degree of freedom. The damping of the model includes contribution
from aerodynamic effect that varies with the wind speed. To measure the damping, the section
model in this experiment was displaced in the vertical plane and suddenly released to vibrate
freely. This procedure was repeated at different wind velocities. The free vibration response
(in the direction of lift) was recorded at each wind speed for 20 seconds at 50 Hz sampling rate.
The non-dimensional number H 1 * (Eq. 2-6) which gives a measure of the aerodynamic
damping, is plotted against the reduced velocity (U/nD) in Fig. 3.5. For the model without any
mitigation devices, H 1 * starts increasing beyond a reduced velocity of 40. At a critical
reduced velocity of 80, H 1 * becomes a positive number, which suggests that the yawed cable
has negative aerodynamic damping and is susceptible to galloping vibrations. The
corresponding critical reduced velocity for the cable with the rivulet for the onset of SDOF
flutter or galloping is 60. With the circular rings or helical strake attached, H 1 * decreases
(becomes a larger negative number) in the same region of the reduced velocity, i.e., the circular
rings or helical strake has stabilized the yawed cable.
To show how much aerodynamic damping the circular ring and the helical streak will
add to one ofthe prototype cables, the following computation is performed. Cable A23 of the
Fred Hartman Bridge, which was reported to vibrate in the second mode at relatively low wind,
was chosen as an example (Table 3.1 ). The total critical damping ratio (~T) of cable A23 of the
Fred Hartman Bridge in the second fundamental mode of vibration was calculated assuming a
mechanical damping ratio (~m) of 0.1% (using Eqs. 2-5 and 2-7). In this computation, it is
assumed that the equivalent yaw angle (i.e. ~*) of this yawed and inclined cable is 3 7° for a
particular wind direction (can occur with wind at 33° with respect to the bridge axis). The
value ofH 1 *measured in the wind tunnel for~= 37° and a 0° (Fig. 3.5) was used in this
calculation. In this calculation, one value of wind speed was used throughout the length of the
cable. In reality, the wind speed is expected to vary along the length of the cables at the
prototype site. The mean wind speed at the top of Cable A23 is expected to be almost 10%
5
higher than the mean wind speed at its base .

5
A more refined calculation to account for this variation is desirable.

Project 0-1400 32
 H1* (~=37° and a.=0°) versus U/nD

50.00
25.00
--- Rivulet at 70
0.00
-25.00
--- No Rivulet
-50.00
""
"""" -75.00
:X: --Helical Strake
-100.00
-125.00
(0/8, 1.50,\)
-150.00 -+-Circular Ring
-175.00 - - - - - - - - - - --- (0/8,3.00)
-200.00
0.00 50.00 100.00 150.00 200.00

U/nD

Fig. 3.5 H 1• for cable with and without mitigation devices

The plot of the total critical damping ratio as a function of wind speed is shown in Fig. 3.6.
The total critical damping becomes zero or negative indicating instability in the second
vibration mode for this cable without rivulet at 40 mph wind speed or greater. The
corresponding value for this cable with rivulet is 35 mph. The same cable with circular ring or
helical strake shows a trend of increasing damping up to 70 mph. Also note that these devices
increase the damping of the cable to 0.5% and greater for wind speeds of 25 mph or greater,
when the problem of vibration was observed to occur. Earlier in this report, it was mentioned
that a minimum of0.5% damping is required to damp out large amplitudes at low wind speeds.

Table 3.1 Videotaped vibration events of the Fred Hartman Bridge stay-cables

Event Date Time Cable Mode Frequency Estimated Location of Remarks

(Hz) Amplitude Amplitude
peak-peak
em (in)
1 411 9:00pm 9 2 2.1 38.1 (15) middle rain, wind
2 411 4:00pm 24 ? high 7.6 (3) ? wind
3 4/3 11:15am 1,2,3 1 0.8 88.9 (35) middle rain, wind
4 4/3 11:20 am 10,11 2 >1.5 10.2 (4) 1/3 from rain, wind
top
5 4/4 6:28am 15,16 1 1.0 30.5 (12) middle rain, wind
6 4/4 6:41am 23,24 2 1.2 91.4 (36) middle rain, wind
7 4/4 6:55am 24 3 1.8 106.7 (42) 113 from Gusts up to 50
top mph

Project 0-1400 33
 Total Critical Damping vs Wind Speed

2.00
--
~
0
1.50 --+--Rivulet at 70
C )::"'
t:: co
·- CJ
Q. ·-
E s:: --.- No Rivulet
co co 1.00
c"fi (I)

~:!: --Helical Strake

_
...
·-

u .
0.50
0
~;:;::::
(0/8, 1.50,\)

-
n;-
0
1-
0

0.00
10 20 30 40 50 60 70 8
- - Circular Ring
(0/8,3.00)

-0.50
Wind Speed (mph), n=1.3 Hz (2nd Mode)
m=1.59 slugslft, 0=6.3", ~=37° and a=0°

Fig. 3.6 Total critical damping at different wind speeds

3.2.5 Effectiveness in Prevention of Water Rivulets

The effectiveness of the circular rings to prevent the formation of a continuous upper
rivulet was tested on the same section model in simulated rain. The inclined/yawed section
model was fixed at a = 37° and ~ = 37° and rain was simulated inside the wind tunnel over the
entire length of the model. In Fig. 3. 7 the formation of an upper rivulet on the bare model is
observed at a wind speed of20 mph. Once the circular rings were mounted as in Fig. 3.8, the
rivulet disappeared. Rain-drops that precipitated between two rings on the model were
intercepted by the lower ring and fell off from underneath the model. The circular rings reduce
the cable length below a critical value that is needed for the raindrops to form a continuous
upper rivulet. To explain this further, consider a very short cable that has the same yaw and
inclination angles as the section model and a length that is equivalent to the pitch (30) at which
the circular rings were placed. This short cable, if subjected to rain, will not have any rivulet
formation because there is not enough length for the raindrops to organize into a rivulet. A
single helical strake is expected to behave similar to the circular rings.

Project 0-1400 34
 Fig. 3.7 Formation of upper water rivulet on a yawed and inclined cable

Fig. 3.8 Prevention of upper water rivulet with circular rings

.····
.Fn

Fig. 3.9 Directions of the mean aerodynamic forces

Project 0-1400 35
3.2.6 Force Coefficients
For force measurement, the section model was rigidly fixed at a= 0° and p 0° (model axis
is normal to the wind direction and on a horizontal plane) and three force coefficients, C n
(drag), CL (lift), and CM (moment) were measured.
The mean values of Co, CL, and CM, as measured in the wind tunnel at separate wind
speeds (Umax = 80ft/s or 24.4 m/s, maximum Reynolds number= 1.67 x 10 5) and normalized
with the model diameter D for forces and D2 for moment, are listed in Table 3.2.

Table 3.2 Force coefficients

Case Orientation Parameter a p Co CL eM

Helical \at 64.5° with D/20 Wire at oo oo 1.33 -0.16 0.37
Strake model axis 1.5D Pitch
Helical \at 46.3° with D/20 Wire at oo oo 1.34 -0.14 0.34
Strake model axis 3D Pitch
Elliptical \at 60° with D/20 Wire at oo oo 1.28 -0.03 0.49
Ring model axis 1.5D Pitch median

The Co and CL of a cable (smooth surface) without any aerodynamic devices are 1.2 and 0,
respectivell, for sub-critical Reynolds number (equivalent to about U=42 mph for A23 cable
of Fred Hartman Bridge), beyond which Co drops to a lower value (see Simiu and Scanlan,
1996). The coefficients Co and CM of a cable with circular rings are expected to be marginally
different from those of the bare cable. Based on the model tests with elliptical rings, Co with
circular rings is expected to be slightly lower than 1.28 (5% increase beyond the Co of a bare
cable) and CM much lower than 0.49. The CL with circular rings is close to zero. The force
coefficients of a cable with circular rings is reported later7 . The mean aerodynamic forces per
unit length (FL: Lift Force, F0 : Drag Force, M: Moment) on the prototype cable can be
calculated as

where pis the density of air (1.21 kg/m 3 or 0.0024 slugs/ft3 at standard temperature and
pressure), U is the components of the mean-hourly wind speed normal to the axis of the cable
and Dis the diameter of the cable. The directions of these forces are given in Fig. 3.9, where y
is the angle of attack as defined earlier.
3.2.7. Explanation of Possible Vibration Mechanisms
A great deal of information was known prior to commencement of this study and some of
which was verified during the current wind-tunnel tests. For example, it is a well known fact

6
See Figure 3.12
7
See Figure 3.24

Project 0-1400 36
that yawed or yawed and inclined cables (without rain) have a tendency to exhibit galloping
beyond a critical wind speed because of the axial flow that is generated in the wake of the
cable. These cables exhibit only divergent-type response at higher wind speeds. The upper
rivulet forms at relatively lower wind speeds to make the cable prone to large-amplitude
vibration. A possible explanation of the velocity-restricted response as observed in the full-
scale monitoring of the vibrations at low wind speeds follows. The scenario is wind
accompanied by rain. The flowing hypothesis is supported by full-scale measurements and
observations. For commonly encountered inclined cables with diameters varying between 80
mm to 200 mm, when wind speed increases beyond 5m/s, an upper rivulet forms on cables
descending in the wind direction. The mean location of the rivulet continues to shift away
from the stationary point with increasing wind speed. When the rivulet is formed at certain
locations, the cable shows a tendency to have divergent response and exhibits galloping with
large amplitude motion. The large-amplitude motion remains limited as the cable has a non-
linear stiffness. Not all stay-cables in a cable-stayed bridge would exhibit this behavior at the
same time as they have different attitudes. As the wind speed increases further, the upper
rivulet shifts further away from the stagnation point and is eventually blown off the cable at a
wind speed greater than 15 m/s. Thus, the mechanism that caused the aeroelastic instability
disappears. The result is a velocity-restricted response within 5-15 m/s wind speed. Beyond
15 m/s of wind speed, the rivulet is out of the scenario and some inclined and yawed cables
continue to have a tendency to exhibit large amplitude vibration beyond a critical speed. In the
range of high wind speeds, the vibrations are limited and non-divergent due to the non-liner
characteristic of the cable. The large-amplitude motions do not reduce with increasing wind
speed as occurred in the lower-wind speed range. The actual motion of the cable is along an
inclined plane depending on the angle of attack (Fig. 2.1) and is more complex than SDOF
galloping-type response due to either the interaction of the oscillation ofthe rivulet with the
cable motion or the cable's richness in vibration modes.

3.3 TTU Wind Tunnel Section Model Tests

Overview
Based on the positive results of the single-degree of freedom wind tunnel tests
performed at CSU, TTU researchers, under consultation and agreement with TxDOT, began
additional wind-tunnel tests using a two-degree of freedom model (Sarkar and Gardner, 2000).

2DOF Suspension System

For this portion of the research project, TTU researchers designed and assembled a
two-degree-of-freedom (2DOF) elastic suspension system (vertical and horizontal motions)
that enabled the section model to behave similarly to actual cable-stays. The suspension
system used is designed to test yawed and inclined section models and is shown in Figure 3.1 0.
Also shown in Figure 3.10 are the circular rings placed on the cable-stay.

Project 0-1400 37
 Figure 3.10 2DOF TTU Cable with Rings and Suspension System

Section model suspension systems commonly use leaf springs as a means of restraining
the translational motion of the model in a given direction. Leaf springs have very low inherent
damping but are difficult to use for yawed and/or inclined cases of vibration, as was necessary
for this research. In addition, the range of motion of a model attached to a leaf spring forms a
circular arc that is acceptable for small displacements (ignoring the small circular motions) but
may excessively restrict system response when displacements become large, thus potentially
producing errant results near the system limit.
Because of the inclined and yawed nature of bridge cable-stays, it was necessary to
construct an elastic suspension system capable of supporting a section model in any
configuration of yaw and/or inclination. In addition, as observed in full-scale measurements by
several researchers , the nature of vibration of the cable-stays is two-dimensional, i.e. , the
vibration tends to be elliptic rather than constrained along a particular axis . To allow elliptic
motions of the model in varying yaw and inclination configurations, the elastic suspension
system developed for this study is able to vibrate along two perpendicular axes-the
horizontal, or "streamwise", and the vertical. In addition, the system has low inherent damping
to properly duplicate field conditions.

Project 0-1400 38
TTU wind tunnel characteristics
The closed circuit, vertically oriented wind tunnel used in this research has sufficient
power to produce wind speeds in excess of 45 m/s (I 00 mph). The TTU wind tunnel has
separate aerodynamic and atmospheric boundary layer test sections. The aerodynamic test
section extends for a length of2.4 m (8 feet) beyond the contraction exit where the flow is least
turbulent. The atmospheric boundary layer test section is 122 m (50 feet) downstream of the
aerodynamic test section. Each section has a glass viewing window and a large access door.
Presented experiments were conducted in the aerodynamic test section that is 1.1 m high x 1.8m
wide (3'9"H x 5'9"W) following an inlet contraction of 4.5: 1. The velocity profile was
measured at this location and was found to vary only ±0.2 m/s (±0.5 mph) across the wind
tunnel cross-section. Turbulence intensity was calculated to be 1.15% with two turbulence
reducing screens. Experiments were conducted in the range of 2-36 m/s (5-80 mph). The
2DOF section model with suspension system is shown inside the TTU wind tunnel in Figure
3.11.

Figure 3.11 2DOF TTU Cable-Stay Setup in Wind Tunnel Section Model

The cylinder model used in the 2DOF portion of the study was made of7.6 em (3.0
inches) l.D. schedule 40 polyvinyl chloride (PVC) pipe. This pipe has an outside diameter of
8.9 em (3.5 inches) and a length of 112 em (44 inches) . The same cylinder model was used in
all 2DOF test cases. The cylinder model is hollow with a 1.3 em (0.5 inch) diameter, two piece
rod that can slide in and out of the ends of the section model to allow re-positioning of the
section model without changing the model length and/or mass.
To allow the section model to be positioned in inclination, the entire system ends
(vertical and horizontal supports) were mounted to aluminum plates that were in tum mounted
to two 3.2 em ( 1.25 inch) diameter steel pipes, each via U-bolts. For yaw positioning, each end
mount assembly can be moved relative to one another along the wind tunnel walls by
relocating the aluminum plates.

Project 0-1400 39
 Damping values were 0.25 %of critical for the vertical motion and 0.33% of critical for
the horizontal motion. Taking into account the mass and damping of this system, the test
section model Scruton number was 31.7 based on vertical damping. This value compares with
the full-scale value ofSc 44.7 (as discussed in Section 2.4.7 of this report).

Results
RMS o(Response
The RMS (root-mean-square) values of the displacement response of the section model
at several wind velocities were calculated from the acquired time histories along the vertical,
horizontal, and combined degrees of freedom. The combined response is the square root of the
sum ofthe squares ofvertical and horizontal responses (Equation 3.1). Generally, the RMS
value referred to in this report is the RMS of the displacement time histories.

= ~ (Vertical RMS) + (Horizontal RMS)

2 2
Combined RMS [3 .1]

Considering Matsumoto et al.' s ( 1995) equivalent yaw angle definition and suggestion
that a yawed-inclined cylinder behaves similarly to a yawed-only cylinder, it was decided to
test the section models in yaw only. Due to the size limitation of the Texas Tech wind tunnel,
the current model suspension system was capable of inclination angles only up to 18°, which
was not sufficient to properly study the effects of inclination. However, yawed angle
variations were not limited by the testing apparatus.

Response due to Various Yaw Angles (without rings)

The rivulet location that produced the most extreme response changed with yaw angle.
An unstable response was not found beyond the yaw angle of 40° for any rivulet location,
therefore the ()= 73° location (Figure 2.2) was used for these yaw angles. When the rivulet
was located at different angles above or below the point that generated the most extreme
response, a more stable response occurred.

Drag and Lifi Coefficients

For force measurement, the section model was rigidly fixed at a= 0° and j3 0° (model
axis is normal to the wind direction and on a horizontal plane) and force coefficients, including
CD (drag), and CL (lift) were measured. Lift and drag coefficients are defined, respectively, as:

c - FL
L- I u2 A [3.2]
2p

[3.3]

where FL = lift force,

FD = drag force,
p = air density,

Project 0-1400 40
 U wind speed, and
A = D x Lcosf/,
D cylinder diameter
L =length of the yawed cylinder
p· =yawed cable angle (see Figure 2.1)

The CD and Cr of a cable (smooth surface) without aerodynamic devices are 1.2 and 0,
respectively, for sub-critical Reynolds number (equivalent to about U 42 mph for the A23
cable of the Fred Hartman Bridge), beyond which CD drops to a lower value (see Simiu and
Scanlan, 1996). The drag coefficient decreased with increasing yaw angle with and without a
rivulet (Figure 3 .12). This is due to the change from cross-flow to axial flow and the decrease
in projected area with increasing yaw angle.

a..J
eo
c
-
c~ 1.4 ••--_..,_ I
'00 1.! ·~ .
~·
c:- j-+-cn
ca II)
.r:::-c
·- CD
0.6+ .-11-CL
I

~~~~~~~3~~=~3
..J ·-
~e
ca CD
0.2 L
-0.2
I • • ___...,

tiS 0 0 20 40 60
0
Yaw Angle, degrees (fi*)

Figure 3.12. Variation of Lift and Drag with Yaw Angle (bare cylinder).

Notice the negative lift for jJ* = 20° to 40°. Some of this negative lift is attributed to the
effect of wind flow over the cylinder ends and interaction of the wind with other components
of the elastic suspension system. Error in the lift coefficient calculations was ± 0.11 and error
in the drag coefficients was± 0.24. However, the same trend exists for all cases studied and
further study is required before this apparent lift can be accounted for as simply due to the
effects of interaction of the wind flow with the suspension system components. For each case
of cylinder configuration with and without the upper rivulet, the most negative lift occurred at
p* = 35°. This is important because the most extreme dynamic response also occurred at p* =
35°. Also note the increase in lift with rivulet at f/ oo and the trend thereafter that is similar
to the case without rivulet. The lift is expected to increase at tl = 0° with rivulet due to the
increased flow separation region above the cylinder created by the upper rivulet.
If the lift shown in Figure 3.12 is due to end effects, one would expect a continuously
decreasing trend. This is because as the cylinder model is yawed more, a larger surface area is
available for the wind flow to interact with, resulting in greater applied forces. In contrast to
the findings of Matsumoto et al. (1995) where the most extreme vibration case in single degree

Project 0-1400 41
of freedom for a smooth cylinder occurred at 45°, the current TTU research finds the most
tJ
extreme response to occur at = 35° in two degrees of freedom.
An important distinction between the current work and the work of other researchers is
TTU' s use of a two-degree-of-freedom system rather than a single-degree-of-freedom system.
The ability of the two-degree-of-freedom system to move along two axes appears to produce
results different from those produced by a vertical motion only suspension system. In addition,
the difference in results can be attributed to different Scruton numbers (Sc 1. 0 in Matsumoto
et al., 1995, experiments versus Sc 31.7 for the present 2DOF experiment). However, for the
yaw angles where an extreme response was observed, the response was consistent with
galloping instability where the displacement continued to grow to the limit of the suspension
system. Figure 3.13 shows the responses of the smooth cylinder for the different yaw angles.

~ 1.5 ,--------------····----······--,
·--
~
"-"' -+- Yaw=1SO
~
·-a 1 -Yaw=2SO
-+- Yaw=3SO
1

t
>0.5 Yaw=4SO
~
0 1_._ Yaw=5SO
r/J.

~ 0
0 100 :;ro 300 400
Rtrl:ml Velocity( UnD)
Figure 3.13 Comparison of Responses of Bare Cylinder for Each Yaw Angle.

To determine the maximum yaw angle where an unstable response may occur with bare
cylinders, i.e. when the RMS displacement is greater than 7.6 em (3 inches), single-degree-of-
freedom (vertical vibration only) studies were performed in an attempt to match the results
produced by Matsumoto et al. (1995) in which an unstable response was found for p* = 45°. 8
In the present research, the case of p* 35° and p* 40°, yielded an unstable response, but a
stable response was found for p* = 45°. Similar results were observed for the same values of p*
in the two-degree-of-freedom tests. Since no unstable response was found for p* > 40°, it is
concluded that the unstable response can be expected for smooth cylinders with yaw angles
from 15° to 40° in two-degrees-of-freedom. Unstable responses were found for all cases of
yaw angle with an upper rivulet in the most sensitive location and, again, the most extreme
response occurred at p* 35°.
Interestingly, in all cases of dynamic response with rivulet, the reduced velocity for the
onset of vibration occurred in the range ofRV 100 to 200. As the wind speed increased
8
Beyond anRMS displacement= 7.6 em (3 in.), the response of the cylinder section model grew
seemingly without limit until reaching the end stops of the elastic suspension system.

Project 0-1400 42
beyond RV = 300, the predominant response was in the vertical direction. To compare with
reported results at reduced velocities above 300, the system was restrained to vibrate only in
the vertical direction. This was required because the drag in the horizontal direction became
large enough to push the section model to the suspension system limits. However, up to the
RV = 300 limit, the horizontal response was smaller compared to the vertical response. Figure
3.14 shows the response at p* = 15°.

~
S'4-,-----------------.,.---,
5
8(!) 3 -r--------="'*"------~·-~-1 -+-Vertical
0
~
~ 2 +----~~---~~~~··--~ ! -a-- Horizontal
0"" 1 +~-----~-=~~~~~~···--~
-A- Combined

~
00 0 +----~!!..___--,-----,----.,-

~ 0 50 100 150 200 250

Reduced Velocity U/nD)

Figure 3.14 Response of Cylinder with Rivulet at p* = 15°.

tJ
At = 15°, although not evident in Figure 3.14, an unstable response was actually
reached for wind speeds beyond RV 220. Because the response grew to the system limits
beginning at RV = 220, it was decided not to record runs for one minute at higher wind speeds
to prevent damage to the suspension system. A peculiar observation is that the case of p* = 15°
yielded a response similar to that of the smooth cylinders where a divergent response is evident
at high reduced velocities (RV > 200). In fact, a similar response (not shown) occurred at
yaw = 0° at even lower wind speeds. Figure 3.15 shows a similar response at f] = 25°.

Project 0-1400 43
 3.5 . . . , - - - - - - - - - - - - - - ,
~

.s 3 +--------.t.----------------1
""'-"'
5 25
8 2 + - - - - t - - - - - - '.....------.~----j -+- VtrticalI I
g 1.5 -t-----+---=------'lo&-~~\\--------·1 -Hriza1all
-c..
-~ 1+---~--~~~~~~~--~ -+- G:niJired •
0
~ 0.5 + - - - - 1 + - - - - - - - - \ : -~-~
0
~ 0~~~~--~-~----

~ -0.5 ~-iff) ~~~f+---7'11

RedmlVelocity( UnD)
Figure 3.15 Response of Cylinder with Rivulet at f/ 25°

Notice in Figures 3.14 and 3.15 that the response tapers off gradually above the critical
wind speed ofRV 150. This result is in sharp contrast to the response at f/ 35°, shown in
Figure 3.16.

,-.4.-------------------,

·-'S 3
d
"-"'

gs2
!"''' - - - - - - - - ,

~
·- 1+----H~.--~-------~
~
t+-;

~ oLI~~~----~~~~~~-
~ -l_L______ _ _ _ __

lhlml \tiaity(= l.Yrq

Figure 3.16 Response of Cylinder with Rivulet at p* 35°.

Notice the sharp drop in amplitude after RV 200 at p* = 35°. This demonstrates the
velocity-restricted response reported by Matsumoto et al. (1995). In Figures 3.17 and 3 .18, it
is seen that the magnitude of the oscillation is significantly less at fJ* = 45° and /3* = 55°,
though the range over which the oscillations build is the same as in other yaw angles.

Project 0-1400 44
 0.6 ~----------------,
.s
"-"
0.5 - 1 - - - - - - - - - - l l , . _ __ _ _ _ _ _--i
d
~ 0.4 1-+-Verti~
g 0.3 - + - - - - - - - 1 - - - f - - - - _ , _ : w - - - - - · · · - - - i
-a0.2
0 ..
1- Hxizoota1
~-.-Combined I
~
o O.l
tZl
t-.nr _ __.........,......._.._.I!;;;B-----i

~ 0 +-"-'---,----,-------r····--
0 100 200 300 400
Reductrl Velocity (=U/nD)
Figure 3.17 Response of Cylinder with Rivulet at p* = 4S 0 •

. 0.35 ···,----···---··---··-------,
:§, 0.3 -1---···----··---------i

d11) 0.25 +-------1--...--··---··~--i

s
8 0.2 -1-------f-~ ---u.~~..-------c
!-+-Vertical I
co::l
]. 0.15 ~----J.-----lll-------i
i --- Hri2oota1l
·-e 0.1 -f-. ---;-~"'F=-J!r"'--..,.~;;w~...,..-----,
1-+- Combined I
0 ! ~. . . . . .
...

~ 0.05
~ 0 -1---,------,----,--------1
0 100 200 300 400
Reduced Velocity ( U/nD)
Figure 3.18 Response of Cylinder with Rivulet at p* SS 0 •

Maintaining the p* 3S 0 case as the yaw angle with the most extreme response,
modification of the surface roughness was made to study the effect on the dynamic response.
To facilitate this, 1SO-grit sandpaper was applied to the cylinder surface. As can be seen in
comparing the "Yaw = 3 S0 " case in Figure 3.13 with the "Combined" case in Figure 3.19, the
response without rivulet changed from a divergent type response (smooth cylinder) to a

Project 0-1400 4S
velocity-restricted type response. There was little effect when a rivulet was added except that
the onset velocity for vibration was reduced compared to the other cases (Figure 3.20).

-S'= 0.5 -,--------~········----------

5e 0.4 -+----------__,.___.,,..--...... --___, r-~~~~____,

8 o.3
~
.------------1---\l•--- -.-vertical
]. 0.2 +------------..f----~~~ --Horizontal
0
<.;.....
0
0•1 +-.F"iF-.I'""'Ifiliiiilli~~~~~~~=--.J~IIIII=i*! -lr- Combt' ned

0 100 200 300

Reduced Velocity(= U/nD)

Figure 3.19 Response with Covering of Sandpaper and No Rivulet at p* = 35°

-=
c 3
=
e
Q)
2.5
Q)
0
2 __.__Vertical
~
0.. 1.5
·-
0
rJ'l
1
<.;..... 0.5
--Horizontal
-1r- Combined
0
r.n 0
~ 0 100 200 300
Reduced Velocity(= U/nD)

Figure 3.20 Response with Covering of Sandpaper with Rivulet at tJ = 3 5°

Figure 3.21 shows a comparison of the responses of the cylinder at fJ* 35° for the
cases with and without the sandpaper covering. These results suggest that the size of the
roughness is important If the formation of a rainwater rivulet is actually the cause of the
cable-stay vibration, then the protuberance of the roughness from the cylinder surface should
be sufficient to trap the rain water incident upon the cylinder. If this is achieved, the net effect,
with or without rain, is an increase in the cylinder surface roughness, which increases local
turbulence and stabilizes the response of the cylinder (Matsumoto et al., 1995).

Project 0-1400 46
 -+-Combined,
Smooth Cylinder

-11- Combined,
Cylinder with
Sandpaper
~ 0. ~ -f-JIItll:lltltfl!ltia!III':!~~W--_j
0 100 200 300 400
Reduced Velocity (=U/nD)

Figure 3.21 Comparison of Responses of Cylinder

with and without Sandpaper (no rivulet)

Response due to various yaw angles with rings

In an effort to mitigate the unstable response of a cylinder with rivulet, circular rings
were placed on the cylinder at a spacing of two and four times the cylinder diameter, beginning
at the cylinder centerline (see Figure 3.10). It was not necessary to use an artificial rivulet in
conjunction with the rings as the circular rings would interrupt the continuity of the cylinder
surface which is necessary for a uniform rivulet to form (Sarkar et al., 1998).9 The circular
rings were very effective in limiting the response of the yawed cylinder. The spacing of the
rings had an effect on the degree to which the response was limited (Figures 3.22 and 3.23).

~ Q35~--------~----··-,

~ Q3+----------~~--~
1:l
8 0.25 . .......-:::-l
8 02 -~-.----~..---- ~~~----------1 1

-+- Vertical •
ct:l
"a 0.15 -c-.-----J~--\..-.---F<----·----- •- Hriz.aial.
a
~
r:l.l
o.1 tt:M:h.-:tt~~.,.....~ -~
0.05 +.L::>~~i.::;ML---------------,
..... I-.-Ccntimil
~ 00
100 400
Rtrlml Velccity(= UnD)

Figure 3.22 Response of Cylinder with Circular Rings Spaced at 4D, p* = 35°

9
However, at times, the artificial rivulet was left on with the circular rings. The rings continued to
mitigate the cable-stay vibrations!

Project 0-1400 47
 .-
ci
:,.=., 0.25 ,--------·····~-----------,

eg0.150.2
= +-------- -+-Vertical
]. 0.1 h~~h~itl:~~ts::li=---- -Ibizontal
6 0.05 +---&::--.~~!:__~:!.___ _ _ _ __ ......... Cmbi.ned
<+....
0 0+----~---~---~--~
fZl

~ 0 100 200 300 400

Reduced Velocity(= UlnD)

Figure 3.23 Response of Cylinder with Circular Rings Spaced at W, rJ 35°

Figure 3.24 shows the variation oflift and drag to yaw angles with rings attached. The
trend of the lift and drag coefficients compares with the case of the bare cylinder shown in
Figure 3.12. Notice again in this case the dip in the lift coefficient at p* = 35°.

-
c
a..J
t!O
"CCO
Q
~

L;f= • ~
I I: ~~I
"• •
c-
t'ISJ!j 0.6 .
!1: c
..J .!1! ~
·~20 • •
u u 0.2.
·-IE I
•

E 8 -0.2 0
cno 40 60
Yaw Angle, degrees {ft*)

Figure 3.24 Variation of Lift and Drag with Yaw Angle

(cylinder with circular rings at 2D spacing)

Low wind speed with rivulet (in the wind tunnel)

Figure 3.25 dramatically shows the potential of the rings in that the cable displacement
with a rivulet (artificial) is reduced from an RMS displacement from over 5.5 em (2.2 in) to 0.8
em (0.3 in) for a 4D ring spacing, and to 0.5 em (0.2 in) for a Wring spacing.

Project 0-1400 48
 2.5
---+--With Upper
c
-
0..
tn 1.5
2

-11-
Rivulet Only

With Rivulet
c.... and Rings@:
0 2D .
(f)
0.5
:E
~
0
0 100 200 300

Reduced Velocity (=U/nD)

Figure 3.25 Aerodynamic Ring Effectiveness at Low Wind Speed with Rivulet

The reduced velocity shown in the figure is dimensionless. For the longest cable-stays
on both the Fred Hartman and Veterans Memorial bridges, a reduced velocity of 125 that gives
the maximum oscillation, corresponds to a wind speed U of 13.8 m/s (31 mph) and 15.2 m/s
(34 mph), respectively, for fundamental frequency n, and cable-stay diameter D of these
cables. As shown in the figure, the vibration is velocity-restricted. At a reduced velocity of
either below 100 or above 175 there is very little vibration.
High wind speed without rivulet (in the wind tunnel)
Though not the primary focus of the research to date, several high-speed wind tunnel
tests without a rivulet (i.e. without rain) have been performed on the cable-stay section. As
shown in Figure 3.26, the vibration of a bare cylinder begins to increase-----apparently without
bound-at a reduced velocity above 250. With the addition of aerodynamic rings, however,
these large RMS displacements at high wind speeds are essentially cancelled-at least in the
wind tunnel.

-·=.
-.
-Q. 3
4
.t--- Bare Cylinder
.!
c 2 · --+-Rings @ 40
'to-
0 1
rn • Rings@ 20
:i 0
0::
0 100 200 300 400

Reduced Vel. (= U/nD)

Figure 3.26 Aerodynamic Ring Effectiveness at High Wind Speed without Rivulet

Project 0-1400 49
3.4 Wind Tunnel Tests Summary
Yawed circular cylinders exhibit divergent oscillatory behavior when subjected to wind
speeds above the known vortex shedding wind speeds in the sub-critical Reynolds number
5
range(< 2 x 10 ) at high reduced velocities (RV > 300). This may be due to axial flow in the
wake of the cylinder that produces a fluctuating pressure system above and below the cylinder,
thereby increasing the vertical and horizontal response.
When the cross-sectional shape of a cylinder is modified by an artificial rain rivulet
located 65° to 75° from the stagnation point, an unstable velocity-restricted type response
occurs in the critical reduced velocity range of 100-200 for any yaw angle up to 45°. An upper
rainwater rivulet presets the separation of flow, resulting in a greater flow separation region
(lower pressure) above the cylinder. Within the critical reduced velocity range, an increased
response of the cylinder occurs in both the horizontal and vertical directions.
Also, it was demonstrated that increased surface roughness of the cable-stay, limits the
velocity-restricted response that typically occurs for a bare cylinder. The divergent response
that occurs for a bare, yawed cylinder at high-reduced velocity (RV > 300) is not evident in the
increased surface roughness case. However, due to the overall similarity of responses for a
cable-stay, with and without surface roughness, it is expected that a divergent response will
occur for the roughened cylinder-though at higher wind speed than for the bare cylinder case.
When a rivulet is added, a velocity-restricted response occurs, but over a more narrow range
than occurs with a smooth cylinder with rivulet. Thus, the size of the surface roughness is
important. It should be sufficiently large to trap the rainwater and prevent continuous rivulet
formation.
A number of different aerodynamic damping devices were tested in the wind tunnel by
TTU researchers. The installation of a helical strake, elliptical rings or circular rings resulted
in: (a) interruption of the axial wind flow and (b) modification of the cable-stay cross section;
and (c) disruption of the formation of a continuous water stream along the cable-stay. Both
single-degree of freedom (SDOF) and two-degree of freedom (2DOF) wind tunnel experiments
proved the effectiveness of circular aerodynamic rings.
A two-dimensional force-damper apparatus was developed for additional tests in the
wind tunnel. The addition of the horizontal degree of freedom had the effect of delaying the
onset velocity of large response of the cylinder section model. The response of the system,
however, was similar in magnitude to that of a single-degree-of-freedom system. The response
in the horizontal direction is important up to RV = 280 for the cylinder section model. Above
that value, very little vibration occurs in the horizontal direction.
The velocity-restricted nature of wind-rain-induced cable-stay vibration was
demonstrated as both low-wind speed and high-wind speed mitigation behavior of the circular
rings were tested. Circular rings attached to the cylinder reduced the system response by up to
90% compared to the bare cylinder case at high reduced velocities. Circular rings with an
outside cross-section diameter of D/14, where Dis the diameter of the cylinder, were attached
at regular intervals along the cylinder. The vibration of the cylinder decreases more as the
spacing between the circular rings decreases. Currently, a circular ring thickness, t, between
D/20 and DII 0 has been found most effective. The damping effect increases with a larger
value oft. As will be discussed in the next chapter, full-scale prototype installation on three
cable-stays occurred in January 2001 by the Texas Department of Transportation ( TxDOT).

Project 0-1400 50
 CHAPTER4
FIELD INSTRUMENTATION

Notwithstanding the wind tunnel tests, both bridges, Fred Hartman Bridge and
Veterans Memorial Bridge, were instrumented with meteorological and response measuring
instruments. Meteorological instruments measured wind speed, wind direction, barometric
pressure, temperature and rainfall. Response measuring instruments were slightly different
in the two bridges. In the Hartman Bridge accelerometers and displacement sensors
measured cable-stay response, accelerometers measured deck vibration and load cells
measured mechanical damper force. In the Veterans Memorial Bridge only accelerometers
were used to measure cable-stay response. Details of the instrumentation on the two bridges
are provided below.

4.1 Instrumentation Setup for Fred Hartman Bridge

The full scale measurement system installed on the Fred Harman Bridge monitors the
meteorological condition at the bridge site and the dynamic response of the bridge to ambient
excitations. Measurements taken at the bridge include: stay cable vibration, deck vibration,
damper force, and meteorological conditions.
Measurement of Stay Cable Vibration
Acceleration
In-plane and lateral accelerations of stay cables are measured in g' s by tri-axial
accelerometers (Crossbow Technologies, www .xbow. com/Products/Accelerometers.htm,
Model CXL04LP3) with a ± 4g range. The sensitivity of the accelerometers is 500 m V/g
and the noise level is 10 mg. The accelerometers are located on the cables at about 20 feet
above the deck level.
Instrumented stay cables are: ASI, AS3, ASS, AS9, AS16, AS18, AS20, AS22, AS23,
AS24, AN24, BSI, BS8, BS16, BS18, BS24, CS16, CS24, DSI, and DS24.
Displacement
Displacement of stay cables is measured in inches by string pot displacement transducers
installed at the deck rail level, near the anchorages of the cables.
Four stays were originally instrumented and displacement was measured in the in-plane
direction only. Instrumented stays at this stage were: AS9, AS20, AN24 and BS24.
The system was expanded to 4 other cables when dampers were installed on these cables
in March, 1999. Since then, both in-plane and lateral displacements are measured at the
location of the damper-cable connection. Instrumented stays at this stage are: AS16, AS23,
AS24, and BS16.
Measurement ofDeck Vibration
Acceleration
Accelerations of the bridge deck are measured at the windward edge of the east deck at
mid-span and at location near the anchorage of stays AS9, AS16, AS19 and CS19.

Project 0-1400 51
 Originally, both vertical and lateral accelerations of the deck were measured in g' s by the
same tri-axial accelerometers as those used for the stay cables. After November 3, 1999, the
tri-axial accelerometer at mid-span and those near the anchorage of stays AS9, AS16, and
AS19 were replaced by uniaxial accelerometers with higher resolution and only vertical
acceleration has been measured since.
Measurement of Damper Force
The force exerted in the dampers is measured in pounds by load cells installed in line
with the dampers at stays AS16 and AS23.
Measurement ofMeteorological Conditions
Wind Speed and Direction
Wind speed and direction are measured at three locations: the top of the south tower, the
deck level at mid-span and the deck level at stay AS18. Anemometers used at the deck level
are Gill UVWs which is capable of measuring wind speed up to 35m/sand has a threshold of
0.3 m/s. The one used at the tower top is a propeller-vane anemometer which can measure
wind speed up to 60 m/s and has a threshold of lm/s and a resolution of 0/.3 m/s.
Rainfall
The amount of rainfall is measured in inches by two rain gauges installed at two
locations: the deck level at stay AS18 and the top of the south tower. The resolution of the
rain gauges is 0.01 inches.
Atmospheric Pressure
The atmospheric pressure at the bridge is measured by a barometer installed at the deck
level at stay AS18.
Temperature
The temperature is measured in Fahrenheit by a temperature probes installed at he deck
level at stay AS18.
All of the transducers used for the measurements were connected to a PC-based data
acquisition system located inside the southeast tower. The data received from the transducers
are amplified and low-pass filtered using 4 pole Bessel Filters set at 10 Hz before being
digitized and continuously monitored by the system. Data files are recorded automatically
when threshold wind speed or acceleration levels are exceeded, ensuring that events with
large oscillations of the bridge, or those associated with significant meteorological conditions
are appropriately captured.

4.2 Instrumentation Setup (or Veterans Memorial Bridge

The Veterans Memorial bridge deck is a precast concrete box girder supported by
eight single planes of 14 cable-stays. Labeled "A" through "H" in Figure 4.1, each plane of
cable-stays is arranged in a vertical, harped configuration and anchored to a central concrete
tower. Researchers from Texas Tech University instrumented 4 of the 112 Veterans
Memorial Bridge cable-stays. This instrumentation is complementary to the instruments

Project 0-1400 52
previously installed by researchers of Johns Hopkins University. Texas Tech University
research personnel in Lubbock, Texas retrieve this data via telephone connections10. When a
certain acceleration threshold is exceeded, that particular data segment is retained for
analysis at a later time.
Two 2-axis accelerometers were attached to each of the four cables-one at the lower
approximately L/3 and one at the lower approximately L/4 locations to ensure capture of at
least the first 11 modes of vibration. In addition, one 3-axis anemometer, one temperature
gauge and two precipitation gauges were installed at the deck level with the desire to capture
simultaneous wind speed, wind direction and rain. A total of 22 channels of data were
collected at 40 MHz.
The remote processing station continuously records 20-minute data files. In addition
to weather data, each data file contains 16 acceleration data sets. A "trigger" occurs within a
20-minute timeframe when: a) the instantaneous acceleration of any of the four cable-stays
exceeds a given threshold or b) the total 3-D wind speed is greater than a pre-determined
value. If either of these two thresholds is exceeded during one of the 20 minute time periods,
the previous and subsequent 20-minute data files are saved to disk. Otherwise the previous
data file is deleted automatically. Four large capacity hard drives are installed at the remote
station. In addition, two modem lines are installed for remote data monitoring and collection.

Data files recorded to disk (i.e. those not deleted by the DAQ software), are referred
to as "records". Many data files, i.e. records, saved to disk were later determined to not have
meaningful data due to a variety of reasons. For example, lightning strikes, power surges,
power outages, and accelerometer failures were common at the site. When at least one of the
acceleration data channels for a cable-stay could be determined to be good, the 20-minute
data file, or record, is categorized as a "good record". When either a "good record" or
continuous series of"good records" occurred for a given cable stay, the single good record or
the series of records is categorized as an "event".

10
High speed internet connections were unavailable on the bridge at the beginning of this project.
Project 0-1400 53
 195m
r - - - - -..-C._p_la_ne_ _ _ (640ft)
"D" plane "8" plane
(cable stays C01 thru C14) (cable stays 801 thru 814)
C07 C14 accelerometer B14 BOB

anchorage
Note: Four instumented
rower B0814
cEille stays labeled

LB14

SECTION A-A

cable stay cEille stay

C07 C14
--"D"plane

A tower

---"H"plane _ _,_,_____ _ ~-~ -~ ~--a~;~: ------"E"plane---

PLAN VIEW

Figure 4.1 Veterans Memorial: Instrumentation Scheme and Wind Angle

Project 0-1400 54
 CHAPTERS

FIELD RESULTS

Instrumentation described in the previous chapter was connected to three data

acquisition systems, (1) Fred Hartman Bridge data acquisition system monitored by John
Hopkins University, (2) Veterans Memorial Bridge data acquisition system monitored by
Johns Hopkins University and (3) Veterans Memorial Bridge data acquisition system
monitored by Texas Tech University.

Johns Hopkins University implemented and maintained a 64-channel data acquisition

system of Fred Hartman Bride and a 24-channel system on the Veterans Memorial Bridge.
Analysis and results of these data were provided to TxDOT through a WDP Progress Report
(WDP, 2001). Johns Hopkins University is continuing to monitor these data acquisition
systems in association with the Center for Transportation Research, University of Texas-
Austin. Analysis and interpretation of these data are not finalized; they will be available in a
CTR, UT-Austin final report.

Analysis and interpretation of data monitored and analyzed by Texas Tech University
are presented here. Specifically of interest are the results of cable-stay vibrations prior to and
following installations of aerodynamic 'ring' dampers.

5.1 Overall Results for Veterans Memorial Bridge

Several site visits were required over the two-year field test time period. Frequently,
accelerometers had to be replaced due to the harsh operating environment. However, over
8500 records were obtained on (at least one of) the four TTU-instrumented cable-stays at the
Veterans Memorial Bridge from August 1999 to July 2001 as shown in Table 5.I. These
files were evaluated to determine ifthe "trigger'' had actually recorded a "good record". As
discussed, an event is defined as either (1) a single good record, or (2) a series of continuous
good records, where a cable-stay is continuously vibrating. Due to typical potential field
measuring errors (e.g. random periodic instrumentation failure, etc.), a majority of the
"recorded" data files did not constitute being categorized as a "good record".

Table 5.1 "Good Records" vs. Total Number of Records

Before 1/10/01 After 1110/01 *

Number of Records 6706 1885
Number of Good 274 224
Records

*Note, Jan 10, 2001 was the installation date ofthe aerodynamic rings
on Cable-Stays Al4, BOB and Bl4..

0-1400 55
 In total, approximately 500 "good records" were obtained. Of these, 252 events were
identified with accelerations greater than 0.5 g on at least one of the sixteen acceleration data
channels. Table 5.2 summarizes the events found for particular cable-stay acceleration
thresholds, both before and after the ring installation. This table shows that more total
weather events were recorded on all four cable-stays in the 6 months after January, 2001 than
were recorded during the previous 1Yz years of monitoring. However, for Cable-Stay B 14,
after the rings were installed, the number of occurrences with maximum instantaneous
accelerations greater than 1.0 g decreased, in general. One of the instantaneous 5 g events
listed in Table 5.2 is highlighted in Figure 5.1.

It is important to emphasize that "good events" included in Table 5.2 are actual.
recorded data files. As only four cable-stays out of the total 112 were instrumented, it is
possible that other cable-stays may have vibrated more than the instrumented cable-stay
results presented in this table. In addition, as discussed previously, a portion of the
instrumentation was "out" for certain periods during the two-year monitoring period. Thus,
values presented in the table can be considered to indicate a reasonable minimum number of
occurrences that can be expected for the four cables. Nevertheless, the number of records
indicated in the table signifies that a potential fatigue situation is present in these cable-stays.

Table 5.2. "Good Records":-Accelerations Greater than 0.5 g

Maximum Total B14 only B14 with Rain

Instantaneous Before After Before After Before After
Accelerations
>=0.5 and <1.0 31 123 28 53 4 34
>=1.0 and <1.5 26 21 26 14 5 3
>=1.5 and <2.0 5 9 2 5 0 0
>=2. 0 and <2. 5 1 0 0 0 0 0
>=2.5 and <5.0 3 17 1 1 1 0
>=5.0 3 13 3 0 3 0
Note: Only for a short period oftime, e.g. 3 months, was the acceleration threshold set
to 0.5 g before January 10, 2001. The threshold was set to 1.0 g during the other 15 months
ofthis initial time period (i.e. before ring installation). After January 10, 2001 (i.e. after
the rings were installed), the acceleration threshold was reset to 0.5 g.

5.2 Major Veterans Memorial Cable-Stay Acceleration Event

Figure 5.1 represents a major accomplishment of this research effort. The parameter
plots shown in the figure were obtained from particular instrumentation selected and installed
on the bridge by TTU researchers-based on results obtained from the background research
discussed in Chapter 2 of this report and on TTU wind tunnel test results presented and
discussed in Chapter 3 of this report.

0-1400 56
 Figure 5.1 shows plots of four parameters-all occurring simultaneously on the
Veterans Memorial Bridge on October 8, 1999. The total time for each of the four plots (i.e.
Plots A, B, C and D) is 2400 seconds, or 40 minutes. The time of interest is from about 200
to 1700 seconds. Plot A shows the wind speed at the bridge site to be a fairly constant 7 to 9
m/s (15-20 mph) during the time of interest.

814 Low and Z-Direction (08 Oct 99, from 9:10am to 9:50am)

i~
5:: E 0 500 1000 1500 2000
-A
I::[
8
:1- - :

200 400
:

600
\ :

800
D"mU~o of~:fal --~ :
1000 1200 1400 1600 1800
: :
2000 2200
l Plm

2400
8

~~ -c
0 500 1000 1500 2000

5
2
0 1 - - - -.... PlaiD
-2
-5 ~----~-----_L _ _ _ _ __ L_ _ _ _ _ ~---~

0 500 1000 1500 2000

Seconds

Figure 5.1 Veterans MemorialS g Cable-Stay Vibration Event:

Simultaneous Wind Speed and Direction, Rainfall, and Stay Acceleration

Plot B shows the cumulative rainfall to be increasing during the time of interest-
indicating that rain is occurring. (When the line in the plot has a slope, it is raining. When
the line is horizontal, it is not raining.) The rainfall rate is 0.45 mm/minute, indicating a
heavy, steady rainfall. The precipitation gage is rated at 50 mm capacity. As shown in the
0-1400 57
plot, at approximately 750 seconds into the recorded time period, the gage dumped the
accumulated water and reset itself Rainfall was continual for 1500 seconds, or
approximately 25 minutes.

Plot C offigure 5.1 shows the angle ofthe wind direction to remain in a range of
300° to 340° during the time of interest.'' As referenced in Figure 4.1, such a wind direction
comes in the "declining" direction of Cable-Stay B 14, where B 14 is the longest cable in the
"B" plane of 14 cables on the Veterans Memorial. A "declining" direction refers to when the
cable declines, from the tower to the anchorage, in the same direction as the wind. Plot 0 of
Figure 5.1 shows the non-oscillating Cable-Stay Bl4--when time is less than 300 seconds-
to begin instantaneous accelerations of 5 g, almost immediately after the rain begins. The
cable-stay continues to oscillate during the time of interest. When the rain ceases, the cable-
stay soon returns to its non-oscillating state. In summary, Plots A through 0 in Figure 5.1
show an actual field event: 1) developing as predicted, 2) oscillating under prescribed
conditions, and 3) ceasing to oscillate when contributing conditions vary outside certain
ranges. These four plots, combined with wind tunnel results presented in Chapter 3 of this
report, suggest the factors causing the wind/rain induced vibration phenomenon now are
reasonably well understood.

5.3 Aerodynamic Ring Installation

Aerodynamic rings were installed on two of the four TTU-instrumented cables, B 14
and B08, and on one JHU-instrumented cable, A 14, in mid-January, 2001. Installation of the
rings on one of the cable-stays is shown in Figure 5.2. Data was collected and analyzed for
one and a half years prior to the ring installation, and for 6 months after. After the rings were
installed, there was a total of 14 days having moderate to heavy rains. To assist in the
capture of events after ring installation, the trigger acceleration threshold was reduced to
0.5 g.

Figure 5.2 Full-Scale Aerodynamic Ring Installation

11
Refer to the Plan view in Figure 4.1 for the defined wind angle direction orientation.
0-1400 58
5.4 Field Events after Ring Installation
Two sets of acceleration events greater than 2.5 g on the four TTU-instrumented
cable-stays (i.e. B 14, B08, C14, C07) are presented in this report. The first set consists of
events recorded before rings were installed on B08 and B 14, from August 01, 1999 to
January 10, 2001 (approximately 1.5 years). The second set consists of events recorded after
rings were installed on B08 and B 14, from January 10, 2001 to July 10, 2001 (approximately
0.5 years). Each set is presented using the following three comparisons: (1) one-minute
RMS acceleration histories vs. one-minute wind speed (Figures 5.3-5.6), (2) one-minute
RMS acceleration history distributions vs. a one-minute wind speed and direction (Figures
5.7-5.10), and (3) dominating vibration modes (Figures 5.11-5.14). As will be discussed
subsequently, these comparisons appear to indicate that aerodynamic rings effectively
suppress wind-rain-induced vibration of cable-stays in the field.

Tables 5.2 through 5.5 summarize the recorded "good records" and "events" in
tabular form. Using the same recorded data and dividing each event into 1 minute RMS data
points, Figures 5.3 through 5.10 present the cable-stay behavior in graphical form.

5.4.1. Field events with maximum instantaneous acceleration >2.5 g

Before ring installation, four events with instantaneous accelerations >2.5 g occurred
on four separate days. These events are listed in Table 5.3. as shown in the table:
• A maximum instantaneous acceleration of 5.8 g, with a maximum wind speed of 12
m/s (26 mph), occurred on Cable B14 with rain (as shown in Fig. 5.1).
• Two of the four recorded events having an instantaneous acceleration> 2.5 g
occurred on Cable B14. The other two events presented occurred on Cable C14.
Neither cable had rings installed at this time.
• All four recorded events occurred with rain.
• Almost all of the events occurred when the wind speed was between 10 and 12 m/s
(23 and 26 mph), with one recorded event reaching 25 m/s (57 mph).

Table 5.3 All Events (>2.5 g) without Rings from August 1, 1999 to January 10,2001

MAXIMUM MAXIMUM
INSTANTANEOUS WIND SPEED
NO. DATE CABLE ACCELERATION (2) (mph) RAIN?
1 09-21-2000 B14 5.6 23 YES**
2 01-27-2000 C14 4.9 25 YES
3* 10-08-1999 B14 5.8 26 rns
4 08-03-1999 C14 3.0 57 YES
Note: The third entry, marked *, is based on the event presented previously in this report (Figure 5.1). The
rain event, marked **, was determined to have rain present based on NOAA data.

0-1400 59
 The number of"events" shown in Table 5.3 is small. Again, it is important to note
that due to the occurrence of several equipment shutdowns during the monitoring period,
additional events were likely missed by the DAQ system. After the ring installation on
Cable-Stay B14, a total of three system-wide events triggered (i.e. any event triggers from
Cable-Stays B08, B14, CO? or C 14), each having an instantaneous acceleration >2.5 g. As
shown in Table 5.4, two of these events occurred on Cable-Stay C14 (without rings), while
one occurred on Cable-Stay B 14 (with rings). The event forB 14 was not wind-rain-induced
vibration. The only event recorded with rain during this six-month period was a two-day-
long event that occurred on (non-ringed) Cable-Stay C14.

Table 5.4 All Events (>2.5 g) from January 10, 2001 to July 10, 2001
(B14 with Rings, C14 without Rings)

MAXIMUM MAXIMUM
INSTANTANEOUS WIND SPEED
NO. DATE CABLE ACCELERATION {g) {mph) RAIN?
1 05-26-2001 C14 2.9 20 NO
2 03-27-2001 C14 6.5 28 YES
3 03-11-2001 B14 2.7 19 NO
Note: All entries were determined to have rain or not have rain based on NOAA data.

Both precipitation gauges functioned flawlessly during the first 1112 year monitoring
period, i.e. "before 1/10/01". Unfortunately, after analysis of the "after 1/10/01" data, it
became apparent that neither precipitation gauge installed at the bridge performed properly
during the final six-month monitoring period. Neither gauge recorded significant rainfall
during several major rain events, though, based on additional data obtained from the National
Oceanic and Atmospheric Administration, NOAA, collected at the nearby Port Arthur, Texas
Airport, it appears that rain did occur in the area on March 27, 2001 (NOAA 2001). From
Table 5.4, the following can be concluded:

• All three events had a maximum wind speed somewhere between 19~28 mph.
• A maximum instantaneous acceleration of6.5 g occurred on Cable-Stay C14 with
rain during a maximum wind speed of 28 mph.
• Two of the three recorded events occurred on Cable-Stay C14, on which rings were
not installed. One of these events occurred without rain.
• Only one event occurred on Cable-Stay B14, on which rings were installed. This
event was not wind-rain-induced.
• For a 20 mph wind without rain, Cable-Stay C14 experienced a maximum
instantaneous acceleration of2.9 g. For a similar 19 mph wind without rain, Cable-
Stay B14 experienced a maximum instantaneous acceleration of2.7 g.
• For a 28 mph wind with rain, Cable-Stay C14 experienced a maximum instantaneous
acceleration of 6.5 g. Unfortunately, a similar wind speed with rain was not recorded
0-1400 60
 for Cable-Stay B14. This could indicate that such an event did not occur in the
monitoring period-or it could indicate that the rings were effective in preventing the
wind-rain-induced vibrations from occurring. Conservatively, with such a limited
data set, one can conclude only that the rings may prevent vibrations from occurring.

Table 5.5 shows the number of all "good records" with accelerations greater than 0.5
g for each cable. For example, before January 10,2001 (i.e. before installation of rings on
Cable-Stay B14), there are 4 separate 20-minute "good records" in which at least one channel
of Cable-Stay B 14 has an instantaneous acceleration greater than 2.5 g.

Table 5.5 Number of"Good Records" (>0.5 g) for each Cable-Stay from 8/99 to 7/01

MAXIMUM INSTANTANEOUS BEFORE JAN. 10, 2001 AFTER JAN. 10,2001

ACCELERATION B14 C14 B08 C07 B14 C14 B08 C07
;::::o.s g and <1.0 g 28 7 18 17 53 29 37 78
;::::1.0 g and <1.5 g 18 1 2 0 14 4 2 4
;::::1.5 g and <2.0 g 2 3 1 2 5 1 0 5

+t~
;::::2.0 g and <2.5 g 0 0 2 0 0 4
~2.5 g 4 2 1 29 0 21

Note: The numbers in this table differ slightly from those shown in Table 5.2, as Table 5.2 lists every separate
"good record" while this table counts a continuous series of simultaneous "good records" as one "good
record". Also, the four (before) Bl4 "good records" shown above correspond to the two (before) Bl4
"events" shown in Table 5.3.

On Cable-Stay B 14, only one event with an acceleration greater than 2.5 g occurs
after January 10, 2001. Before January 10,2001, all four "good records" for Cable-Stay B14
with accelerations greater than 2.5 g are with rain. After January 10,2001, the greater than
2.5 g "good record" for Cable-Stay B 14 is without rain.

5.4.2. One-minute RMS acceleration history vs. one-minute wind mean speed
In order to evaluate the effectiveness of rings, the relationships between one-minute
RMS acceleration history and one-minute mean wind speed, before and after ring
installation, are plotted for every channel in Figures 5.3, 5.4, 5.5, and 5.6 for Cable-Stays
B 14 and C 14. One-minute RMS accelerations for every identified "good record" and every
channel have been calculated. The data of one record is 20 minutes long, so it contains 20
one-minute RMS data points of acceleration. Figure 5.3 shows the one-minute RMS
acceleration history of B 14HZ 12 vs. wind speed before ring installation, i.e. from August 01,
1999 to January 10, 2001. The maximum RMS acceleration is approximately 1.7 g. Figure
5.3 contains 5400 one-minute RMS points. After ring installation, the one-minute RMS
acceleration history of B 14HZ is shown in Figure 5.4. From this figure we can see the
maximum RMS acceleration decreases to 1.3 g. More importantly, no wind-rain-induced

12
The "B 14" refers to Cable-Stay B 14. The "H" refers to the "high" or L/3 position. The "Z" refers to vertical
direction ofthe cable-stay (with respect to acceleration). See Figure 4.1, "Section A-A"
0-1400 61
vibration events were found after the rings were installed. Figure 5.4 contains 4400 one-
minute RMS points.

B14HZ (Before Rings, i.e. before Jan 10, 2001)

..... '
2-
•814HZ ../ wind-rain vibration events
1.8
,~,,-----,~ approx. 330° wind direction (i.e.
§ 1.6- ', t • ) the declining direction for 814).
~u 1.4- , , -'\...~~---,-,
,~- ',
~ 1.2 I '

~ • • : low-speed flutter vibration events due

~ 1- ' • _, to 0° wind direction, generally without
~ 0.8. rain
c:
~ 0.6-
Q.,
0c: 0 .4 -
0.2
0___.... •
0 5 10 15 20 25 30 35 40 45 50
One-Minute RMS Wind Speed (mph)

Figure 5.3 B14HZ One-Minute RMS Acceleration vs. Wind Speed (Before Rings)

814HZ (After Rings, i.e. after Jan 10, 2001)

Ci 2
-:- 1.8 low-speed flutter vibration events due to
Gi
u 1.6
u 0° wind direction, without rain
< 1.4
U) 1.2
::E
0:: 1
no wind-rain vibration events
s:I
0.8
c: 0.6
~ 0.4
Q.,
c: 0.2
0
0
0 5 10 15 20 25 30 35 40 45 50
One-Minute RMS Wind Speed (mph)

Figure 5.4 B14HZ One-Minute RMS Acceleration vs. Wind Speed (After Rings)

Figures 5.5 and 5.6 show the one-minute RMS acceleration histories before and after
January 10, 2001 respectively for C14HZ. In contrast to Bl4, no rings were installed on
Cable-Stay Cl4 after January 10, 2001. As indicated in the figures, a significant increase in
recorded vibration events occurred on C14 after Jan 10,2001. The maximum RMS
acceleration ofC14HZ increased dramatically from 0.8 g to 1.4 g after January 10,2001.

0-1400 62
 C14HZ (Before Jan 10, 2001)

1.6
wind-rain vibration events
~ 1.4 no vibration events due to 0° wind approx. 30° wind direction
s direction, without rain, recorded (i.e. the declining direction for
Qi 1.2 (Note: C14 instrumentation was
CJ
CJ
C14).
<( down during 0° 814 events) /
(/)
:::E
~~~
-----.~~~~
0:: 0.8
,,
~
'
•
Q) \
\
~ 0.6 I
I
I
I
~

. -...
\

~ 0.4 '',,,. .. • ~
0 0.2 l • ~~~~-#~---------~~~ •

oil'~-··· • •
0 5 10 15 20 25 30 35 40 45 50
One-Minute RMS Wind Speed (mph)

Figure 5.5 C14HZ One-Minute RMS Acceleration History vs. Wind Speed
(Before January 10, 2001-No Rings)

C14HZ {After Jan 10, 2001)

1.6
•C14HZ wind-rain vibration events
~ 1.4 approx. 30° wind direction
s (i.e. the declining direction for
Qi 1.2
CJ
CJ
<(
C14)
(/)
:::E 0.8 vibration events due to low-
0::
Q)
-; 0.6 speed flutter, approx. 0°
c
~ wind direction, without rain
Q, 0.4
c
0 0.2

0
0 5 10 15 20 25 30 35 40 45 50
One-Minute RMS Wind Speed (mph)

Figure 5.6 C14HZ One-Minute RMS Acceleration History vs. Wind Speed
(After January 10, 2001-No Rings)

Rings were placed on Cables A14, B14 and B08. Though Cable Stay A14 was
instrumented, the instruments did not provide useful data. Data for B08 and B 14 are similar
to each other, with B 14 results always more extreme than B08. Therefore, only analyzed
data from Cable-Stay B 14 is presented in this report. Though the results appear promising,
with the limited data collected to date, firm conclusions cannot be made.

0-1400 63
 Table 5.6 lists the changes of maximum one-minute RMS accelerations for alll6
channels before and after January 10, 2001. As indicted in the table, the "After/Before
Ratios" of all channels on cables installed with rings (i.e. B 14 & B08) after January 10 are
less than 1, with an overall average of0.5. In contrast, for the channels on cables without
rings installed (i.e. Cable Stays C 14 & C07), the "After/Before Ratios" are greater than 1,
with an overall average of2.2.

Table 5.6. Maximum One-Minute RMS Acceleration

on All Channels Before and After January 10, 2001

MAX MAX AVERAGE WITH RINGS

CHANNEL RMSACC RMSACC AFTER/ BEFORE OF "AlB" AFTER
NO. BEFORE AFTER RATIO RATIO 1110/2001?
B14HX 0.8 0.3 0.4
B14HZ 1.8 1.3 0.7
0.6 YES
B14LX 0.7 0.4 0.5
B14LZ 2.0 1.6 0.8
C14HX 0.5 1.2 2.3
C14HZ 0.8 1.4 1.7
2.6 NO
C14LX 0.5 1.5 3.1
C14LZ 0.9 2.7 3.1
B08HX 0.3 0.2 0.5
B08HZ 1.2 0.2 0.2
0.4 YES
B08LX 0.4 0.2 0.4
B08LZ 1.3 0.3 0.3
C07HX 0.4 0.6 1.8
C0 7HZ 0.8 1.8 2.3
1.9 NO
C07LX 0.3 0.5 1.5
C07LZ 0.9 1.8 1.9

It is noted that very few wind-rain-induced or low-speed flutter 13 events were

recorded from Cable-Stay C14 before January 10, 2001. There are several possible
explanations for this. Though a wind-rain-induced event may have occurred, it may have
happened when at least one component of the DAQ instrumentation system had
malfunctioned. Another possibility is that only a few wind-rain-induced events actually
occurred from a direction capable of"energizing" Cable-Stay C14 prior to January 10,2001.
Similarly, the absence of wind-rain-induced events for Cable-Stay B 14 after January
10, 2001 is most likely to be due to either: 1) no favorable weather conditions occurred after
January 10, 2001, or 2) the rings worked- based on the data collected for wind-rain
conditions.

13
"Low-speed flutter" is used to refer to galloping in this document and it is a velocity-restricted response. It
generally occurs without rain.
0-1400 64
 Again, though the available field data is limited, it appears somewhat reasonable to
conclude that the aerodynamic rings have suppressed the vibration of cable-stays during
wind-rain-induced vibration events. For "without rain" cases, the rings may not be effective
at low-wind speeds, as indicated in both wind tunnel and field tests. The effective aeroelastic
damping from the circular rings becomes functional at higher wind speeds. At these higher
wind speeds, the divergent cable-stay response is likely to occur without the rings. (Note
these higher wind speed cases were not recorded during the field monitoring period, so the
expected effectiveness of the rings at the higher field-site wind speeds could not be
confirmed.)

5.4.3. One-minute RMS acceleration distributions vs. wind direction and speed
As noted earlier, wind direction has a major effect on the vibration response of cable-
stays. When the wind direction angle is oo (i.e. perpendicular to the cable surface as
indicated in Figure 4.1) and without rain generally, the wind-induced vibration is low-speed
flutter, i.e. a divergent type. In contrast, when the wind direction angle is approximately
±30° (30° for Cl4, 330° for Bl4), combined with rain, the cable vibration is usually
categorized as wind-rain-induced vibration, i.e. a velocity-restricted type, which is shown in
Figures 5.7 through 5.10. In order to take wind direction into consideration, all one-minute
RMS accelerations, both before and after ring installation, have been plotted about their
corresponding directions. Figure 5.7 shows the one-minute RMS acceleration distribution of
Bl4HZ vs. wind direction before ring installation. These 5400 points correspond to points
shown in Figure 5.3. The maximum recorded RMS acceleration is approximately 1.5 g with
zero degree wind direction (i.e. low-speed flutter) and 6.7 m/s (15 mph) wind speed. With
rain, a 330° wind direction, and a 9.4 rnls (21 mph) wind speed, an RMS acceleration of 1.7 g
is recorded as indicated in the figure. The former vibration is categorized as low-speed
flutter, and the latter is categorized as wind-rain-induced vibration. For this cable-stay, the
wind-rain-induced vibration and low-speed flutter typically cause a larger acceleration
response than those vibrations caused by vortex shedding.
Figure 5.8 shows the one-minute RMS acceleration distribution ofB14HZ after ring
installation. These points correspond to points shown in Figure 5.4. In Figure 5.8, low-speed
flutter continues to occur. However, the wind-rain-induced vibration appears to have
disappeared. Again, as discussed previously, though the amount of data is limited, the
preliminary indication is that the aerodynamic rings may be suppressing the wind-rain-
induced cable-stay vibration. As expected, the circular rings cannot suppress the low-speed
flutter of the cable-stays. However, this low-speed flutter typically does not cause extremely
large vibrations.
Table 5.7 lists B14HZ RMS acceleration results with rain and with wind directions
between 315° and 345°, both before and after the aerodynamic rings were installed. Based
on results shown in the table, when the wind speed is between 9-10 m/s (20-22 mph), before
the rings were attached, a maximum RMS acceleration of 1. 7 g occurred. In contrast, after
rings were installed, a maximum RMS acceleration of only 0.08 g occurred. As this appears
to be an order of magnitude reduction, the aerodynamic rings appear to mitigate the wind-
rain-induced vibration. Again, this conclusion is subject to previously discussed reservations
concerning the limited amount of recorded data.

0-1400 65
 Table 5.7. Acceleration RMS of 814HZ
(With Rain, With Wind Direction from 315° to 345°)

Before Rings After Rings

WIND SPEED (Before January 10, 2001) (After January 10, 2001)
NO.
(mph) RMSACC RMSACC
TOTAL TOTAL
FROM TO FROM TO
1 5~10 1 0.03
2 10~15 1 0.05
3 15~20 13 0.05 1.7
4 20-22 2 1.7 1.7
5 23-25 6 0.07 0.08
6 25-29 16 0.06 0.09

0-1400 66
 - ..

..
·.
•,
I 1.5g
·,;····--+-+---· ~-=-------.:------ - -- - . - - - -
·. .·
.... . ,•
,•

' ,•
' ... ~

.
,•
' ,•
: ' ,•
\
•'
, ,- e.-&-~----.......
.. '
.·
' ,•
''
0~5 \ •'

~;il"' 15mph, L5g.)

I
I
§ I
A
8c{
I
1- - - -

~ -0~5 2
0:::

_,.... ··· , ·-~\

·· ·· ····················· ·· ···· Low-Speed
.,
f ._.......···· ···· ··
............:'::\.......
.... '·····, ...
Flutter Events
Before Rings
.... 'J. . _···...... ,.......-,..- :
, ·. ··... , Wind/Rain Events
'• .... .,JI/··. ···... ./"'
Before Rings

,
... .... ·- ~
·..

1.0g
- - - --~1- · - - - ' - • -
·····.~-~
..•
,"""'·.
·· ...
•• •'
················ ...
.,. ...•'

(330°wind, 2 1mph, l.7g,

10-08-1 999 , 9: 10-9:50 AM)
- - - -.- »1 .5- 1.5g • . - .. -
RMS ACC (g)

0 0-1 0 mph, no rain !:>. I 0-20 mph , no rain 0 20- mph, no rain
• 0-10 mph, rain A I 0-20 mph, rain e 20- mph, rain
---
Figure 5.7. B14HZ one minute RMS acceleration distribution
with wind speed and direction (Before 1anuary 10, 2001-Without Rings)

0-1400 67
 .
-
..
/ •,

: '. • 1.0g
: 1.0g ··.'

.. 0.3
_,___
Low-Speed
•~
. Flutter Events

0.5g:

~
u
u
<(
(/)
::!:
0:::
0
-~ -d.s 1.5

········..t .

. ······- .............•.

...
~

-8.3 \

·. \7
I·.
•, :. </~No Wind/Rain Events
.. ..
• -/ ............ -·· . • Recorded(After Rings)

·'
,•.
.... / •• 300° ..
---~--------

RMS ACC (g)

0 0-10 mph, no rain 6 I 0-20 mph, no rain 0 20- mph, no rain

• 0-1 0 mph, rain
·- - - ·
A I 0-20 mph, rain e 20- mph, rain

Figure 5.8. Bl4HZ one minute RMS acceleration distribution

with wind speed and direction (After January 10, 2001-With Rings)

0-1400 68
 The RMS acceleration distribution vs. the wind direction for Cl4HZ, where no rings
were installed, is shown in Figures 5.9 and 5.10, before and after January 10, 2001,
respectively. Figures 5.9 and 5.10 correspond to Figures 5.5 and 5.6, respectively. Figure
5.9 shows no low-speed flutter events before January 10, 2001, and relatively few wind-rain-
induced events for Cable-Stay C 14. In contrast, Figure 5.10 shows an extremely high
number ofwind-rain-induced events after January 10, 2001. C14HZ was expected to
experience the same low-speed flutter as B 14HZ did before and after January 10, 2001, since
they have the same geometric and material properties. However, low-speed flutter events on
C 14HZ did not occur as often, nor were they as large as the low-speed flutter events on
B14HZ (shown in Figure 5.8) that occurred during the same time period. Further
investigation has revealed that the absence of recorded C 14 low-speed flutter events during
the initial monitoring period was likely due to equipment malfunctions during these time
periods.

Table 5.8 Acceleration RMS ofC14HZ

(With Rain and Wind Direction Between 15° and 45°)

Before Jan 10, 2001 After Jan 10, 2001

WIND SPEED (No Rin2s) (No Rin2s)
NO.
(mph) RMSACC RMSACC
TOTAL TOTAL
FROM TO FROM TO
1 5~10 3 0.03 0.04 1 0.07
2 10~15 1 0.06 125 0.02 0.7
3 15~20 395 0.04 1.3
4 20~25 1 0.41 50 0.06 1.2

Similar to Table 5.7, Table 5.8lists C14HZ RMS acceleration results, with rain and
with wind directions between 15° and 45°. Again, though the amount of data is limited, it
appears that without the rings installed, Cable-Stay Cl4 remains prone to wind-rain-induced
vibrations.
In Tables 5.7 and 5.8, it is evident that only a few events recorded before and after
ring installation had similar wind conditions. Though preliminary results appear favorable,
further data collection is necessary to prove the effectiveness (or non-effectiveness) of the
rings. However, it appears the rings have no measurable beneficial effect on mitigating the
low-speed flutter induced from a oo wind direction. Nevertheless, again based on limited
(perhaps incomplete) field data, it appears probable that the passive rings are effective
against wind-rain-induced cable-stay vibrations.

0-1400 69
 ....... -- . - .......
1.0g

-o-.9
Wind/Rain Events

<~ ?::<01110/01
:
..
..
:
• - -~5

• 0.5g
;< •
..
....
....
..
: ..
§ 0.3 - . '
u
u
<(
. '
. ................ .
~
Wind/Rain Events
Before 0 III 010 I
II)
:::!:
a::

-0.2 0.8

. .. No Low-Speed Fluuer
.... Events

. '"1- Before 01110101

-0.3
..
I
.... -.-
. /
..
,
..
.
:.·

. -.
300°

RMSACC (g)

0 0-10 mph, no rain !:::. I 0-20 mph, no rain 0 20- mph, no rain
• :__o_-_l_O_m_,p'-h-,'_r_ai_n_ _ _ _--=
- --= A=-1_0_-_ 20
-'--~ph, rain _____
. _2_0_-_ m_,_p_h-'--,_ra_in_ _ __

Figure 5.9. Cl4HZ one-minute RMS acceleration distribution

with wind speed and direction (Before January 10, 2001-Without Rings)

0-1400 70
 ..
..
-9.9

.
---~'-------------

~ ·· \
.
~ ·.
-·------+--

.·

§
u
u
<(
If) 0.5g:
:::!:
0::

-~ -~5

Flutter Events
..
"(
After 01 / 10/01 1.0g 1.5g;

-0.3

..
.. /
/ .•• 300°
.t

----+-----------~~~---
'

RMS ACC (g)

- ------- - - - ------------
0 0-10 mph, no rain I 0-20 mph, no rain
f.,. 0 20- mph, no rain
• 0-10 mph, rain A I0-2_9 mph, rain e 20- mph, rain

Figure 5.10. Cl4HZ one-minute RMS acceleration distribution

with wind speed and direction (After January 10, 2001-Without Rings)

0-1400 71
5.4.4. Dominating vibration modes before and after ring installation
Interestingly, aerodynamic rings appear also to have slightly altered the dominating
vibration modes of the two "B" cables. Figure 5.11 shows the occurring percentage of each
mode as the dominating mode before the rings are installed. Figure 5.12 shows the same
relationship after the rings were installed. Before ring installation the three most dominating
modes are the 13 1h (39% of all events), ih (18%) and 6th (10%). After ring installation, the
status changes with the three most frequently occurring modes being the 1oth (31% ), 4th
( 17%) and 7th ( 10%). Also, the 2nd mode begins to appear more often after ring installation,
while the 3rd, 6th, Iih and 15th modes disappear. The dominant modes of vibration after ring
installation are lower in general than before ring installation. Due to higher curvatures in the
vibration pattern, higher modes could induce higher stresses in the cable, assuming
comparable overall cable deflections. Thus, the vibration behavior appears to be somewhat
altered after installation of the rings on Cable-Stay B 14.

(/) 45%
1-
ifi 40%
[j 35%
:j 30%
<
u.. 25%
0
w 20%
C!)
~ 15%
ifi 10%
u
0:::
w
D.
5%
0%
2 3 4 5 6 7
• - I
8 9 10 11
·-
12 13
Ill_ I_·
14 15 16
MODE NO

Figure 5.11 Dominating Mode Distribution of B 14HZ

(Before Ring Installation, i.e. Before January 10, 2001)

0-1400 72
 35%
(/)
5; 30%
w
iii-1 25%
-1
ct 20%
u.
0
w 15%
~
!z 10%
~
w 5%
D.

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
MODE NO

Figure 5.12 Dominating Mode Distribution ofB 14HZ

(After Ring Installation, i.e. Mter January 10, 2001)

In contrast, for Cable-Stay C 14HZ, which is shown in Figures 5.13 and 5.14 and which did
not have the rings installed, the dominant modes appear to have remained essentially the
same, as expected. The most popular dominating mode is the same, i.e. the 13th mode, both
before and after January 10, 2001. The second most fopular dominating mode changes
slightly from the gth before January 10, 2001 to the 7t after January 10, 2001. Such a small
change could have been caused by slightly varied weather conditions during the time period.

35%

~ 30%
z
w
iii 25%
::l
ct 20%
u.
0
~ 15%
~
iii 10%
li
~ 5%

m,.,l,.,
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
MODE NO

Figure 5.13 Dominating Mode Distribution ofC14HZ

(Before January 10, 2001, No Rings)

0-1400 73
 !Z
50%

45%
~ 40% •
l
..I 35%
..I
c 30%
u.
0 25%

I- -· · ·~. ._,_
w
~ 20%

~
tii 15%

~ ~: - •- . - ~1- -.
, +J.....
..__,_,

2 3
.-...J•--...-----,---,..,
4 5 6 7 8 9 10 11 12 13 14 15 16
MODE NO

Figure 5.14 Dominating Mode Distribution of C 14HZ

(After January 10,2001, No Rings)

5.5 Summary
Based on data collected from the Veterans Memorial Bridge, severe wind-rain-
induced vibration did not occur on either Cable-Stays Bl4 or BOS after ring installation,
which occurred on January 10, 2001. Wind-rain-induced vibration often occurred on these
two cables prior to the ring installation. In contrast, a number of severe wind-rain-induced
vibrations occurred on Cable-Stays Cl4 and C07, where no rings were installed before and
after January 10, 2001. Thus, it appears that aerodynamic rings may be suppressing the
wind-rain-induced vibrations. However, with the limited number of field data points
collected, firm conclusions concerning the effectiveness of the aerodynamic rings cannot be
made at this time. Also, as expected, the circular aerodynamic rings do not mitigate cable-
stay vibrations due to low-speed flutter (where the wind direction is 0°, or perpendicular to
the plane of stays). 14
Though the number of recorded wind-rain-induced events after the rings were
installed was less than desired, TTU researchers have made an extensive effort to evaluate
the data collected in the final six months of the research project. Comparisons of the cable-
stays before and after the rings were installed are presented in Figures 5.3 5.14. Table 5.9
serves as a summary of the findings.

14
Note that the maximum accelerations are not as critical for this wind case at low speeds compared to wind-
rain-induced vibration cases.
0-1400 74
 Table 5.9 Cable-Stay Behavior Summary Table

B14 B14 C14 C14

Before Rings MterRings No Rings No Rings
Before 1/10/01 After 1/10/01 Before 1/10/01 Mter 1/10/01
WIND Small number of Large number
vs. Wind-rain (3 30°) No wind-rain
wind-rain (30°) ofwind-rain (30°\
RMSACCEL
Low-speed flutter Low-speed flutter Low-speed flutter Low-speed flutter
events occurred events occurred did not occur * events occurred
DOMINANT 13th lOth 13th 13th
MODES Pattern altered after rings installed Similar before and after 0 Ill 0/2001

*Note: The lack of low speed flutter events for C 14 before 1/10/01 is most likely due to temporary equipment
downtimes.

As indicated in Table 5.9, after the rings were installed, no wind-rain-induced

vibration events with acceleration >2.5 g were recorded for Cable-Stay B14. In addition, no
rain event from an approximate 330° angle was recorded. Also, as indicated in Table 5.9, the
dominant vibration modes changed after the rings were installed on Cable-Stay B 14. The
dominant vibration modes remained the same on Cable-Stay C14, where the rings were not
installed. The rings, as expected, did not mitigate low-speed flutter.
Additional possibilities for the improved Cable-Stay B 14 behavior include: (1)
possible short-term equipment downtimes during unrecorded events15 and (2) the randomness
of particular events occurring at any given time, etc. However, it must be noted that it is also
possible that the rings worked-the rings were effective in suppressing vibrations in the wind
tunnel. Considering the possibility that the rings were effective, the actual recording of the
data to prove this effectiveness is difficult to obtain as the ring effectiveness generally brings
the cable-stay accelerations below an acceptable recording threshold. 16 Certainly, the rings
have not been proven to not work. Had any wind-rain-induced vibration events been
recorded for Cable-Stay B 14 after the rings were installed, one could assume the rings did
not work. 17

Unfortunately, the DAQ system was not designed to trigger based on wind
direction--only due to wind speed and/or cable acceleration. Thus, evaluation of any "after
ring" events was both difficult and time consuming. (A full scale prototype evaluation of the
aerodynamic rings was not anticipated at the time the instrumentation and the DAQ system
were being developed and installed.) Thus, unfortunately, the overall effectiveness of the

15
However, it is actually felt by the researchers that the equipment performed remarkably well over the two-
year monitoring period, given the relative harsh local environmental conditions at the site and the initially
planned six-month testing period.
16
It must be stressed that wind-rain events were recorded for Cable-Stay B14 before the rings were installed
and these same events were noticeably absent after the rings were installed.
17
Again however, even in this case, it should be noted that a suboptimal ring thickness and/or ring spacing
could have been chosen for this prototype evaluation.
0-1400 75
aerodynamic rinWs cannot be stated with complete confidence at this time, though they do
appear to work. 1

5.6 Conclusions
1. Wind-rain-induced vibration events occur often for these cable-stays.
2. Wind-rain-induced vibrations appear when a) the wind is combined with rain, b)
wind speeds are between 20-30 mph, and c) the wind comes from a direction of
either approximately 30° for Cable-Stay C14 or approximately 330° for Cable-Stay
B14.
3. Low-speed flutter is often triggered when wind speeds are between 7-11 m/s (15-25
mph) and from either a 0° or 180° wind direction. These are velocity-restricted in
response and are not as critical as wind-rain-induced cases.
4. Aerodynamic rings appear to decrease maximum overall cable-stay RMS acceleration
values.
5. Aerodynamic rings appear to effectively mitigate wind-rain-induced cable-stay
vibration in the field. However, they are unable to suppress low-speed flutter events
(i.e. 0° wind), as expected from wind-tunnel tests.
6. Aerodynamic rings appear to somewhat change the dominating vibration mode of the
cable-stays, possibly making them favorable for lower stresses.
7. Further study and additional field data collection is needed to conclusively prove the
effectiveness of the rings on Cable-Stays B14 and B08.

18
Results of the ratio of the maximum RMS accelerations before and after 1/10/01 shown in Table 4.5 suggest
that the rings work.
0-1400 76
 CHAPTER6
SUl\'IMARY AND CONCLUSIONS

Two highway bridges under the jurisdiction of the Texas Department of Transportation,
TxDOT, have experienced a wind-rain-induced cable-stay vibration problem. These two
bridges are the Fred Hartman and Veterans Memorial, located in Baytown and Port Arthur,
Texas, respectively. This report documents the results of three years of study of the cable stay
vibration problem on these two bridges by researchers from Texas Tech University, under
contract to TxDOT. The scope of the TTU effort includes background research, wind tunnel
tests, and field instrumentation and monitoring. In addition, TTU researchers developed an
innovative aerodynamic ring as a mitigation device for the cable-stay vibration problem.
Results from a field prototype application of the rings on the Veterans Memorial Bridge are
presented. The rings appear to be effective in eliminating the rivulet formation on the cable
stay, while at the same time, adding aerodynamic damping to the system. Complimentary
research at Johns Hopkins University and at the University of Texas-Austin on cable-stay
vibrations and their implications is continuing; the results of these additional research efforts
are not included here.

6.1 Wind Tunnel Results

Wind tunnel tests using single degree of freedom (SDOF) and two degrees of freedom
(2DOF) cylinders demonstrated vibration characteristics with yawed angles of attack and led to
the logical development of an aerodynamic damping device.
A two-dimensional force-damper apparatus was used to study wind-rain-induced cable-
stay vibration. The rain rivulet was simulated with a small plastic, smooth circular cord
attached longitudinally to the cylinder. Velocity-restricted vibration was observed in a reduced
velocity (dimensionless quantity) range of 100 to 175 (see Figure 3.25). This reduced velocity
range translates into wind speeds between 7 to 15 m/s (15 to 33 mph) depending on the
diameter and natural frequency of vibration of a cable-stay. Circular rings with a cross-
sectional dimension of D/14 (where Dis the diameter of cable-stay) placed at a spacing of2D
and 4D suppressed the cable-stay vibrations. These successful wind tunnel results prompted
installation of prototype rings on the Veterans Memorial Bridge cable-stays for a field study.

6.2 Field Site Results

6.2.1 Before ring installation on Cable-Stays B08 and B 14

TTU researchers instrumented and monitored four of the Veterans Memorial Bridge
cable-stays. Several significant events were recorded over a one and one-half year period. In
particular, an instantaneous acceleration event of 5g was recorded on October 8, 1999. The
plot of this event, shown in Figure 5.2, demonstrates that under favorable conditions, the cable-
stay will vibrate and will continue to vibrate until one or more of the parameters causing the
favorable conditions ceases. In this case, the vibration began and continued due to a) rain with
wind, b) wind speed in the velocity-restricted region, and c) wind from a critical angle for the
cable-stay. In this case, the vibration ended once the rainfall ceased. For Cable-Stays B 14 and

Project 0-1400 77
B08, the critical wind direction angle is approximately 330°. For Cable-Stays C07 and C 14,
the critical angle is approximately 30°19 .

6.2.2 After ring installation on Cable-Stays B08 and B 14

Prototype aerodynamic circular rings, or bands, were manufactured and placed on three
Veterans Memorial cable-stays. TTU researchers monitored Cable-Stays B08 and B14 for six
months, while an additional cable-stay (A14) with rings was monitored by Johns Hopkins
University. Unfortunately, the number of data points collected in the six months after the rings
were installed was not large.
Though any conclusions based on this limited data set must be considered with caution,
the data appears to suggest that the rings performed as expected in mitigating the wind-rain-
induced cable-stay vibrations in "field" (i.e. real world) conditions?0 That is, it appears the
rings were effective in mitigating cable-stay vibrations during wind-rain-induced events, as no
wind-rain-induced events occurred on cable-stays with rings during the six-month monitoring
period. Nonetheless, as expected, the passive circular rings appear ineffective in mitigating
vibrations due to low-speed flutter events (i.e. winds at 0°).
A distinction is made in this report between "wind-rain-induced" vibrations and "low-
speed flutter" response. Low-speed flutter referred to in this report should not be confused
with typical flutter, where the response is divergent. Low-speed flutter can have large
vibrations, but typically produces less than 2.0 g accelerations. In this document, "low-speed
flutter" is used to refer to galloping and is a velocity-restricted response. It generally occurs
without rain. In addition, low-speed flutter is not divergent.

In summary, as stated previously, any conclusions stated based on the current field site
data are made with reservation, as the number of data points collected to date are few.
Nevertheless, though based on limited data, it appears the circular rings work in mitigating
wind-rain-induced cable-stay vibrations.

6.3 Future Research

6.3.1 Optimized ring geometry, surface and placement

Future work should be directed at determining an optimal cross-sectional size and shape
of the circular ring, with aesthetics of the cable-stayed bridge in mind. Several potential
changes in the ring and/or cable-stay surface area parameters should be investigated. Smaller
rings can be used to decrease drag. The prototype used a ring thickness of D/8, whereas D/1 0,
D/12 or even D/20 may be sufficient. Also, a wider spacing of rings may be adequate. For
example, the prototype used a 3D spacing, whereas a 4D or 5D spacing may also be acceptable.
Additionally, for the three prototype cable-stays, the rings were distributed fully from
the top of the cable-stay to the lower anchorage. It may be possible that only the lower portion
(say L/3) of the cable-stay would require being retrofitted with the aerodynamic rings. Finally,

19
See Figure 4.1 for wind direction orientation.
20
Thought the limited data collected is insufficient to confirm the effectiveness of the rings in the field, certainly,
the data collected to date does not suggest that the rings do not work. See Section 5.5 for a more complete
discussion of field results.

Project 0-1400 78
the surface treatment of the cable-stay should be investigated for mitigation effectiveness. If
mechanical dampers are to be used in conjunction with an aerodynamic mitigation technique,
the optimum magnitude and attachment location of the mechanical damper should be
determined. This is also important for use in the development of an analytical model for such a
system.

6.3.2 Full-scale testing

Full scale testing of the circular rings in natural wind, along with other vibration
mitigation devices, should be performed when possible. To accomplish this task, a controlled
full-scale test site should be developed. Such a field site must have a large tower with the
ability to record actual wind, rain and other meteorological data. One such site is the Wind
Engineering Research Field Laboratory in Lubbock. Full-scale testing, using the TTU 200m
(650 foot) tower, would allow confirmation or rejection of numerous, conflicting claims from a
variety of mitigation device manufacturers. Knowledge gained from such full-scale tests
would allow TxDOT and other government agencies to make better decisions when developing
cable-stay vibration mitigation strategies.

6.3.3 Flexible cable-stay section model for wind tunnel tests

The results of the testing presented in this report could be used in the development of a
flexible aeroelastic wind tunnel model. That is, a flexible scale model of a larger portion of the
cable-stay should be subjected to the same yaw, inclination, and rain conditions in the wind
tunnel as has been done in the section model studies to date. From these more refined wind
tunnel model studies, analytical models can be developed that take into account the fluctuating
wind forces on the cylinder caused by rivulet formation and oscillation. If flutter is determined
to be a problem, analytical models using a flutter derivative approach with time domain models
can be used.
In addition, the following items should be studied to properly determine the mechanism
of vibration of the stay-cables:
• Effect of Scruton number, upstream turbulence, intensity of rain and wind
combinations and rivulet oscillation.
• Precise flow visualization and flow measurements in the wake.
• Measurement of flutter derivatives.

6.3.4 Active control

Mitigation of cable stay vibrations appears to be an excellent candidate for active
control technology. Unlike typical building structures designed to resist seismic forces, active
control of cable stays should have much less constraints, as massive forces and associated
power requirements are not needed.
Two scenarios are possible. An "active only" system would simply sense vibrations in
a particular cable stay and energize rings along the cable-stay to damp the oscillations. This
active system can be considered "reactive." An active "smart" system could sense the wind-
rain-induced aerodynamic forces on the cables in real time using appropriate sensors, calculate
the acceleration magnitude and mode of vibration of the cable-stay, and selectively energize

Project 0-1400 79
specific "smart'' rings to quickly dissipate the vibration energy. This second type of active
system can be considered "proactive."
Successful implementation of either of the proposed active systems offers the potential
for: (a) superior damping, i.e. less fatigue, for the cable stays, (b) a reduction in the number of
required aerodynamic rings per cable, (c) elimination of ice buildup on the cable stays, and (d)
innovative aesthetic treatments to the overall bridge structure.

Project 0-1400 80
 REFERENCES

Baker, C.J., 1987. Measures to control Vehicle Movement at Exposed Sites

During Windy Periods, Journal of Wind Engineering and Industrial
Aerodynamics, Vol. 25.

Bosdogianni, A and Olivari, D., 1996. Wind- and rain-induced oscillations

of cables of stayed bridges, Journal of Wind Engineering and Industrial
Aerodynamics, Vol. 64, pp 171-185.

Crossbow Technologies, www.xbow.com/Products/Accelerometers.htm,

Model CXL04LP3, accessed March 2004.

Davenport, A.G., 1995. The dynamics of cables in wind, International

Symposium on Cable Dynamics, Liege, Belgium: AI.M., 1-12 Insert.

Flamand, 0., 1993. Rain-wind induced vibration of cables, Proceedings of

the First IAWE European & African Regional Conference, pp 471-479.

Geurts, C., Vrouwenvelder, T., Staalduinen, P., and Reusink, J., 1998.
Numerical modeling of rain-wind-induced vibration: Erasmus Bridge,
Rotterdam, Structural Engineering International, 2/1998, pp 129-135.

Hikami, Y. and N. Shiraishi, "Rain-Wind Induced Vibrations of Cables in

Cable Stayed Bridges," Journal ofWind Engineering and Industrial
Aerodynamics 29 (1988): 409-418.

Matusmoto, M., Shiraishi, N., and Shirato, H., 1992. Rain-wind induced
vibration of cables of cable-stayed bridges, Journal of Wind Engineering
and Industrial Aerodynamics, Vol. 41-44, pp 2011-2022.

Matsumoto, M., 1998. Observed behavior of prototype cable vibration and

its mechanism, Proceedings of the International Symposium on Advances in
Bridge Aerodynamics, Copenhagen, Denmark, May 10-lJ.

Matsumoto, M., Y. Daito, T. Tanamura, Y. Shigemura, S. Sakuma and H.

Ishizaki, 1997. Wind-induced Vibration of Cables of Cable-stayed Bridges.
The Second East European & African Conference on Wind Engineering,
Genova, Italy, 1791-1798.

Matsumoto, M., J. Aoki, N. Shiraishi, and M. Yamagishi, "Various

Mechanism oflnclined Cable Aerodynamics," in Proceedings of the Ninth
International Conference on Wind Engineering, (New Delhi, India, 1995),
759-770.

Main, J. and N. Jones, "Full-scale Measurements of Stay-cable Vibration,"

in Proceedings of the Tenth International Conference on Wind Engineering,
(Copenhagen, Denmark, 1999).

0-1400 81
Sarkar, P. and T. Gardner (2000). "Model Tests to Study Rain/Wind-
Induced Vibration of Stay Cables", Proceedings of ASCE Structures
Congress, Philadelphia, PA, May 2000.

Sarkar, P., K.C. Mehta, T. Gardner, and Z. Zhao (1998). "Aerodynamic

Solutions to Cable Vibrations," Proceedings the Texas Section ASCE Fall
1998 Meeting, Dallas, Texas, September 9-11, 1998.

Sarkar, P. P., 1992. New identification methods applied to the response of

flexible bridges to wind, Ph.D. dissertation submitted to The Johns Hopkins
University, Baltimore, MD.

Simiu, E., and Scanlan, R.H., Wind Effects on Structures: Fundamentals

and Applications to Design, 3'd ed., (New York: John Wiley & Sons, 1996),
150-155.

Verweibe, C., "Exciting Mechanisms ofRain-Wind-Induced Vibrations,"

Structural Engineering International (1998), 112-117.

White, F., Fluid Mechanics, 3'd ed.,(New York: McGraw-Hill, 1994), 411-
413.

Whitlock, Dalrymple, Poston, and Associates, Inc. (1998). Progress Report

Number 1, Evaluation and Repair of Stay-Cable Vibrations: Fred Hartman
Bridge, Veterans Memorial Bridge. Austin: Texas Department of
Transportation.

Whitlock, Dalrymple, Poston, and Associates, Inc. (1999). Progress Report

Number 2, Evaluation and Repair of Stay-Cable Vibrations. Austin: Texas
Department of Transportation (Feb 12, 1999).

Whitlock, Dalrymple, Poston, and Associates, Inc. (200 1). Progress Report
Number 3 (Final), Evaluation and Repair of Stay-Cable Vibrations. Austin:
Texas Department of Transportation.

Whitlock, Dalrymple, Poston, and Associates, Inc. (2002). Fax submittal

from Scott Withoft ofWDP and Associates, February 8, 2002.

Wianecki, J., 1979. Cables wind excited vibrations of cable-stayed bridges,

Proceedings of the 5th International Conference on Wind Engineering,
Colorado 1979, pp 1381-1393. Oxford-New York, Pergamon Press.

Young Instruments, 2004, www.youngusa.com, Model 27005, accessed

March 2004.

Young Instruments, 2004, www.youngusa.com, Model 05305, accessed

March 2004.

0-1400 82
 Appendix A

Aerodynamic Study of fluid flow and resulting forces on a body

Aerodynamic damping Friction of air (fluid) opposes vibration of large amplitude

Divergent Amplitude continues to increase with each cycle of vibration

Drag force Along-wind force

Flow separation region Area or region of a body behind the location of separation of
flow

Flutter One or two-degree-of-freedom aeroelastic instability involving rotational motion

Force coefficient Along wind non-dimensional coefficient

Galloping Single-degree-of-freedom translational aerolastic instability

Lift force Cross-wind force, usually but not necessarily, vertical

Reduced velocity Non-dimensional number using the ration of velocity to diameter and
system natural frequency

Reynolds number Ratio of inertial forces to viscous forces in fluid flow

Rivulet A small stream or brook

Scruton number A non-dimensional parameter incorporating the ration of structural

mass to fluid mass, and structural damping, which is a measure of the propensity of a
structure to resonant dynamic response

Splitter plate A plate placed along the longitudinal direction of a cylinder (for cable-
stays it is axial flow along the cylinder)

Stagnation point Point on a body where the approaching flow is brought to rest

Strouhal number Non-dimensional vortex-shedding frequency

Sub-critical range Low Reynolds number where the drag coefficient for a cylindrical
shape is high

Turbulence Fluctuations in fluid flow

Von Karman vortex sheet Same as vortex shedding but as a sheet rather than eddies

0-1400 83
Vortex shedding The periodic shedding of eddies formed from the rolling-up of the
boundary shed from a bluff body

Wake The region oflow velocity and turbulent flow in the region downstream of a body

0-1400 84
 AppendixB

Bridge Closure Criteria in High Wind Events

Suggested Criteria for Bridge Closure

(a) In case of a named tropical storm approaching the Houston area, we carefully
monitor that storm with the help of the Hurricane Forecast Center. If the strike
probability of the storm making landfall in the Houston vicinity is high (say
25%), the bridge should be closed 4-5 hours ahead of the expected time of
landfalL Of course a judgment has to be made in conjunction with the local
emergency management officials depending on when they issue an order for the
evacuation and how long it takes to evacuate Baytown. However, the next
criteria should always get precedence.
(b) When the 3-sec gust recorded at the bridge site on the deck level exceeds 50 mph
consecutively three times within a 10 min. period, the TxDOT officials should be
informed and immediate action should be taken to close the bridge as soon as
possible (it can be rehearsed to time this exercise where TxDOT has been
informed of a potentially dangerous wind speed situation on the bridge deck and
it takes an action to implement the closure of the bridge).

Accident Wind Speeds in Cross Winds

In the UK., at exposed locations such as embankments or high river bridges, wind-
induced high-sided vehicle accidents are a common occurrence. Over 370 wind-
induced accidents were reported in the UK. during one extreme wind event alone in
1990. Accidents are caused either by the vehicle being completely blown over, or by
the vehicle deviating significantly from its original path. Mainly three types of
accidents can be expected to occur in regular winds that exceeds a certain wind speed:
(1) overturning accidents, (2) sideslip accidents, (3) rotation accidents. The critical
wind speed for wind-induced accidents depend on (a) the type of accident listed
above, (be) the parameters of the vehicle such as type (car, tractor-trailer, van,
coaches), mass and mass inertia, road tire friction coefficients, and aerodynamic
coefficients, (c) speed of vehicle, (d) wind direction or yaw angle with respect to the
vehicle axis. Curves were obtained for different vehicles showing the variation of
accident wind speed (w) with vehicle speed (u) and the angle between the direction of
travel and the wind vector (p). The following conclusions were obtained.

Standard Car: Lowest accident wind speed w is 25 m/s (56 mph) corresponding to
u = 5m/s (11 mph). The vehicle is most at risk for winds normal to the direction of
travel. Reducing u did not increase w. The conclusion was that speed restrictions in
windy periods at exposed sites may well increase the risk of accident

Standard Coach: Lowest accident wind speed w is 35 m/s (78 mph) corresponding
to u = 5 m/s (11 mph). Both overturning and rotation accidents can occur depending
on p.

0-1400 85
 Standard Large Van: Only overturning accidents can occur for different
combinations ofu and~· Front windward wheel will be the first to lose contact with
the ground. Lowest accident wind speed w is 21 m/s (47 mph) corresponding to u =
25 m/s (56 mph). Accident speed is about 31 m/s (69 mph) ifu is 5 m/s (11 mph).
Hence, slowing down does help.

Articulated Tractor-Trailer Combination: Only overturning accidents can occur

for different combinations of u and ~. Rear windward wheel will be the first to lose
contact with the ground (different from the large vans). Lowest accident wind speed
w is 21 m/s (47 mph) corresponding to u 25 m/s (56 mph). Accident speed is about
28 m/s (63 mph) ifu is 5 m/s (11 mph). Hence, slowing down does help.

Some parameters of the vehicles are:

Coach

Based on the above discussion and study by Baker, traffic control on the bridge can
be a combination of the following:

• Close the bridge if a named tropical storm or a hurricane is approaching. The

time of closure will depend on factors such as estimated time of landfall,
evacuation plans, etc.

• Close the bridge if wind gusts on the bridge deck exceeds 50 mph at least
three times in a 10 minute period.

• Put a warning sign for high-sided vehicles (trucks, large vans, motor homes)
to slow down to 25 mph if wind gusts on the bridge deck exceed 42 mph at
least three times in a 10 minute period.

0-1400 86