Seracino).
Vibration based damage detection techniques were applied to an existing coastal bridge to
evaluate the ability of the methods to detect scour at the foundation as well as to determine
the practicality of using these testing methods in the field. A study of recent bridge failures
revealed that hydraulic damage is a leading cause of bridge failures in the United States.
Scour is a type of hydraulic damage that may result from a coastal storm. Scour occurs when
water current washes away compacted sediment from around bridge pilings. Methods have
been implemented in the field that visually inspect for scour or use underwater equipment. The true extent of damage caused by scour is not always visually observable. Also
underwater equipment can be laborious to install and operate and are prone to damage.
Vibration based damage detection techniques have been developed to successfully detect
damage in steel girders. This idea was translated into methods to evaluate an idealized
two-span steel bridge laboratory model for the presence and location of scour. The advantage of
using these testing methods is that the testing process can take place from the superstructure
and evaluate the response of the structure to determine the existence and location of damage.
Laboratory models cannot capture every feature of an existing structure, such as ambient
loading and various environmental conditions. Based on this, it was necessary for these
methods to be applied to an existing structure to evaluate the proposed methods ability to
detect scour.
The Northeast Creek Bridge in Jacksonville, N.C. was chosen to implement the field
monitoring techniques. A dive team visually observed scour at the foundation and
recommended the bridge be repaired. The bridge was monitored both before and after the
repair so the change in dynamic characteristics could be observed by applying the proposed
damage detection techniques. Accelerations of the bridge responding from an impact
deflection and the change in the curvature of the flexibility based deflected shape. From data
analysis, it was observed that the flexibility based deflection method most accurately
identified the existence and location of scour.
A finite element model of the bridge was created to guide the selection of testing equipment.
The damage features were applied to the finite element model for comparison. It was
observed that all of the proposed methods were successful in indicating scour for the finite
element model. Based on this, the assumption was made that the curvature based methods
perform better with theoretical data than field data.
A main advantage of using vibration based damage detection techniques is that field
monitoring can be implemented quickly and easily following a coastal storm. It is not
necessary to have a finite element model for this method, which would be a time consuming
and costly task. It is however necessary for data sets to be available for a healthy bridge for
the methods to be applied. It is recommended to continuously monitor scour critical bridges
before and after coastal storms to track changes in dynamic characteristics that can indicate if
© Copyright 2013 by Anna Katherine Harris Clark
by
Anna Katherine Harris Clark
A thesis submitted to the Graduate Faculty of North Carolina State University
in partial fulfillment of the requirements for the degree of
Master of Science
Civil Engineering
Raleigh, North Carolina
2013
APPROVED BY:
_______________________________ ______________________________
Dr. Rudolf Seracino Dr. Sami Rizkalla
Committee Chair
DEDICATION
This work is dedicated to my father and mother, who have always had the highest hopes for
me, to my Great Aunt Ruby who supported, encouraged and believed in me, and to Frank
BIOGRAPHY
Anna Katherine Harris Clark was born in Goldsboro, NC to Richard and Paula Harris. She
graduated from Greene Central High School as class valedictorian in 2006 and enrolled in the
North Carolina State University College of Engineering with much encouragement from her
N.C. State engineering alumni father and brother. During her undergrad career she
completed three co-op rotations working at ABB Power System division where she learned
the basics of the civil engineering design process. She also participated in Campus Crusade
for Christ and a Civil Engineering Study Aboard trip to Australia. She graduated Summa Cum Laude in December 2010 with a Bachelor of Science degree in Civil Engineering.
Following graduation, she began working towards her Master of Science degree in Civil
Engineering, specializing in Structural Engineering, under the guidance of Dr. Rudolf
Seracino. During graduate school, she worked as a research assistant. She plans to begin a
career as a Structural Engineer and eventually pursue her P.E. licenses following the
ACKNOWLEDGMENTS
I am thankful to God and the relationship I have with him through Jesus Christ. He has
shared many great blessings with me.
I would like to thank Dr. Rudolf Seracino for providing me with many great academic
opportunities and guidance. Also, I would like to thank Dr. Adel Elsaid for his time, support
and advice throughout this project. I would like to express my appreciation to Dr. Sami
Rizkalla and Dr. Mohammad Pour-Ghaz for their participation in my advisory committee.
I would like to acknowledge the support of the Department of Homeland Security Coastal
Hazards Center of Excellence as well as its staff and fellow researchers. I would like to
thank the NCSU Constructed Facilities Laboratory staff and students whom aided in the
experimental testing, particularly Mr. Greg Lucier, Mr. Jerry Atkinson, Mr. Jonathan
McEntire, Mr. Guanglong Qu and Mr. Omar Mohamedian. Also, I sincerely thank the N.C.
DOT Division 3 Bridge Maintenance Engineers who truly made this research project
possible, particularly Mr. Trevor Carroll.
Finally, I thank my family and Frank who have always supported and encouraged me
TABLE OF CONTENTS
List of Tables ... vii
List of Figures ... viii
Chapter 1. Introduction ...1
1.1 Structural Health Monitoring ...2
1.2 Research Significance ...3
1.3 Research Scope and Methodology ...4
1.4 Organization of the Thesis ...4
Chapter 2. Literature Review ...7
2.1 Introduction ...7
2.2 Review of Bridge Failures ...7
2.2.1 Storm Surge Bridge Failures ...9
2.2.2 Scour Damaged Bridges ...14
2.3 Methods for Monitoring Bridges for Scour ...21
2.3.1 Streambed Elevation Measuring Methods ...22
2.3.2 Fiber-Optic Sensors Bridge Monitoring ...25
2.3.3 Robot Assisted Bridge Monitoring ...29
2.3.4 Tiltmeters Monitoring ...30
2.4 Vibration Based Superstructure Monitoring Techniques ...32
2.5 Summary ...41
Chapter 3. Bridge Field Monitoring...42
3.1 Introduction ...42
3.2 Bridge Geometry ...43
3.3 Selecting Accelerometers ...45
3.4 Selecting Accelerometer Spacing ...47
3.5 Accelerometer Test Setup ...54
3.7 Impact Load Test Setup ...60
3.8 Damaged Bridge Test ...62
3.9 Bridge Retrofit ...64
3.10 Repaired Bridge Test ...66
Chapter 4. Analysis of Field Data ...68
4.1 Introduction ...68
4.2 Frequency Response Functions ...68
4.3 Mode Shapes and Natural Frequencies ...72
4.4 Changes in Curvature of Mode Shapes ...87
4.5 Changes in Flexibility Deflection...94
4.6 Changes in Flexibility Curvature ...100
4.7 Damage Features Summary ...106
Chapter 5. Finite Element Model and Analysis of the Northeast Creek Bridge ...108
5.1 Introduction ...108
5.2 Finite Element Model Description ...108
5.3 Collection of Accelerations, Natural Frequencies and Horizontal Mode Shapes .112 5.4 Changes in Horizontal Mode Shape Curvature Damage Feature ...119
5.5 Changes in Flexibility Based Deflection ...125
5.6 Changes in Flexibility Based Curvature ...130
5.7 Damage Features Summary ...135
Chapter 6. Conclusions ...137
6.1 Summary of Findings ...137
6.2 Future Work Recommendations ...145
References ...147
LIST OF TABLES
Table 2.1: Types of bridge failures from 1989 to 2000 [Wardhana et al., 2003] ...10
Table 4.1: Natural frequency in Hz for the first six mode shapes for each span and the
average in the healthy and scoured bridge condition ...80
Table 5.1: Finite element model natural frequencies for the repaired and scoured conditions
and percent change between both conditions ...118
Table 5.2: Percent change between repaired and scoured natural frequencies for the finite
element model and the field testing ...118
Table 5.3: Percent change between field testing and the finite element model natural
frequencies for the first 6 horizontal mode shapes ...119
LIST OF FIGURES
Figure 2.1: Number of failed bridges distributed by year [Wardhana et al., 2003] ...8
Figure 2.2: Interstate-10 bridge failure resulting from Hurricane Ivan [Douglass et al., 2004]
...11
Figure 2.3: Soil washed away from the Bob Sykes Bridge abutment due to Hurricane Ivan
[Douglass et al., 2004] ...12
Figure 2.4: Little Lagoon bridge deck damage due to Hurricane Ivan [Douglass et al., 2004]
...12
Figure 2.5: Escambia County Road 191 Bridge repaired after Hurricane Ivan [Douglass et al.,
2004] ...13
Figure 2.6: Collapse of US 90 Bridge deck over Biloxi Bay Mississippi [Chen et al., 2009] 14
Figure 2.7: Localized scour at a pile due to Hurricane Ike [FEMA, 2009] ...15
Figure 2.8: Scour depth in a sand-bed stream over time [FHWA, 2001] ...17
Figure 2.9: Horseshoe vortex around the base of a piling causing local scour [FHWA, 2001]
...18
Figure 2.10: US 90 bridge in Bay St. Lewis experienced up to 4.5 meters of scour during Hurricane Katrina [Robertson et al., 2007] ...20
Figure 2.11: Echo sound transducer measuring streambed depth [Mueller et al., 1999] ...23
Figure 2.12: Fiber-optic sensor connected to rod implanted in riverbed [Zarafshan et al.,
2012] ...25
Figure 2.13: Barrier rod installation in streambed [Zarafshan et al., 2012] ...26
Figure 2.14: Scour levels recorded from sensor located near a bridge pier [Zarafshan et al.,
Figure 2.15: Sensors on a bridge pier and the cross section of a sensor [Manzoni et al., 2011]
...28
Figure 2.16: UMV used to inspect Rollover Pass Bridge called a Sea-RAI USV [Murphy et al., 2011] ...29
Figure 2.17: Tiltmeter instrumentation plan [Briaud et al., 2011] ...31
Figure 2.18: Idealized two-span steel bridge laboratory model [Elsaid, 2011] ...34
Figure 2.19: CDF and MCDF values for the finite element model idealized two-span steel model [Elsaid, 2011] ...36
Figure 2.20: CDF and MCDF values for the idealized two-span steel laboratory model [Elsaid, 2011] ...37
Figure 2.21: Idealized two-span finite element model change in flexibility deflection for 24 inches of symmetrical scour [Elsaid, 2011] ...38
Figure 2.22: Idealized two-span laboratory model change in flexibility deflection for 24 inches of symmetrical scour [Elsaid, 2011] ...39
Figure 2.23: Change in curvature of flexibility based deflection between the finite element model with 24 inches of symmetrical scour and the healthy model [Elsaid, 2011] ...40
Figure 2.24: Change in curvature of flexibility based deflection between the laboratory model with 24 inches of symmetrical scour and the healthy model [Elsaid, 2011] ...40
Figure 3.1: Bridge selected for testing over the Northeast Creek in Jacksonville, NC ...42
Figure 3.2: Section of bridge deck and girders ...43
Figure 3.3: Details of the pier located in the Northeast Creek ...44
Figure 3.4: Vertical scour depths at each pile ...45
Figure 3.5: Triangular impact load used in the finite element model ...46
Figure 3.7: Spacing of measurement points versus average mutual information of points
[Trendafilova et al., 2001] ...50
Figure 3.8: Three test setups of varying sensor spacing [Trendafilova et al., 2001] ...50
Figure 3.9: Histogram values for the first ten nodes ...52
Figure 3.10: Average mutual information values based on the finite element model ...53
Figure 3.11: Girders of the two spans at the center pier support ...54
Figure 3.12: Sensor spacing near center pier ...55
Figure 3.13: Accelerometer locations ...56
Figure 3.14: NCDOT Hydra-Platform truck used during testing ...57
Figure 3.15: (a) NI PXI-4496 High Channel Industrial Platform Module and (b) NI PXI-1033 5-Slot Chassis ...58
Figure 3.16: Computer and Data Acquisition System during testing ...59
Figure 3.17: Impact hammer model PCB 086D50 ...60
Figure 3.18: Impact locations ...61
Figure 3.19: Impact hammer used during testing ...62
Figure 3.20: Impact load (a) between sensors 6 and 7 and the resulting accelerations (b) at sensor 7 as recorded during the first test at this location on the first span during damaged testing ...64
Figure 3.21: Scoured piles exposed to the creek water ...65
Figure 3.22: Base board attached to the footing (a) jackhammer boards into the soil (b) attach vertical boards to the base board (c) to create formwork for the cementitious non-excavatable flowable fill ...66
Figure 4.1: Single input and single output system ...69
Figure 4.3: MATLAB plot of the absolute value of the tfestimate function for the fourth test
on the first sensor on the second span of the damaged bridge with an impact between the
fourth and fifth sensor ...72
Figure 4.4: FRF versus frequency for the 10 sensors used on the fourth test of the second
span of the damaged bridge while impacting between the fourth and fifth sensor ...73
Figure 4.5: Dynamic amplification factors at different frequency ratios and damping ratios
[Zhou, 2006] ...75
Figure 4.6: FRF magnitude and phase angle versus frequency for sensors 3, 4 and 5 used on
the fourth test of the second span of the damaged bridge while impacting between the fourth
and fifth sensor ...77
Figure 4.7: Mode shapes 2 through 6 found from each span test for the scoured bridge ...82
Figure 4.8: Mode shapes 2 through 6 of each span extracted for the repaired bridge ...84
Figure 4.9: Complete 6 mode shapes for the center two spans of the scoured and repaired
bridge ...86
Figure 4.10: CDF calculated considering comparisons of horizontal mode shapes 1, 2 and 3
...89
Figure 4.11: CDF calculated considering comparisons of most scour sensitive horizontal
mode shapes 1, 3 and 5 ...89
Figure 4.12: CDF calculated considering comparisons of horizontal mode shapes 1, 2, 3, 4
and 5 ...90
Figure 4.13: CDF calculated considering comparisons of horizontal mode shapes 1, 2, 3, 4, 5 and 6 ...90
Figure 4.14: MCDF calculated considering comparisons of horizontal mode shapes 1, 2 and 3
...92
Figure 4.15: MCDF calculated considering comparisons of the most scour sensitive
Figure 4.16: MCDF calculated considering comparisons of horizontal mode shapes 1, 2, 3, 4
and 5 ...93
Figure 4.17: MCDF calculated considering comparisons of horizontal mode shapes 1, 2, 3, 4,
5 and 6 ...93
Figure 4.18: Flexibility deflection for the scoured and repaired bridge as well as the absolute
difference between the two conditions considering the mass-normalized horizontal mode
shapes 1, 2 and 3 ...96
Figure 4.19: Flexibility deflection for the scoured and repaired bridge as well as the absolute
difference between the two conditions considering the most sensitive to scour
mass-normalized horizontal mode shapes 1, 3 and 5 ...97
Figure 4.20: Flexibility deflection for the scoured and repaired bridge as well as the absolute
difference between the two conditions considering the mass-normalized horizontal mode
shapes 1, 2, 3, 4 and 5 ...98
Figure 4.21: Flexibility deflection for the scoured and repaired bridge as well as the absolute
difference between the two conditions considering the mass-normalized horizontal mode
shapes 1, 2, 3, 4, 5 and 6 ...99
Figure 4.22: Flexibility deflection curvature for the scoured and repaired bridge as well as the
absolute difference between the two conditions calculated from the mass-normalized
horizontal mode shapes 1, 2 and 3 ...102
Figure 4.23: Flexibility deflection curvature for the scoured and repaired bridge as well as the
absolute difference between the two conditions calculated from the mass-normalized
horizontal mode shapes most sensitive to scour 1, 3 and 5 ...103
Figure 4.24: Flexibility deflection curvature for the scoured and repaired bridge as well as the
absolute difference between the two conditions calculated from the mass-normalized
Figure 4.25: Flexibility deflection curvature for the scoured and repaired bridge as well as the
absolute difference between the two conditions calculated from the mass-normalized
horizontal mode shapes 1, 2, 3, 4, 5 and 6 ...105
Figure 5.1: Three dimensional view of the scoured bridge finite element model with filled objects ...110
Figure 5.2: Center scoured pier of the finite element model ...111
Figure 5.3: Bridge deck view of the finite element model ...111
Figure 5.4: Three dimensional view of repaired bridge finite element model with outlined objects ...112
Figure 5.5: Node and impact locations ...113
Figure 5.6: Horizontal impact load used in the finite element model ...113
Figure 5.7: FRF magnitude and phase angle versus frequency for nodes 6, 7 and 8 on the second span of the scoured bridge finite element model showing a peak near 40 Hz ...115
Figure 5.8: The first six horizontal mode shapes extracted for the repaired and scoured bridge finite element models ...116
Figure 5.9: CDF values for horizontal mode shapes 1, 2 and 3 ...120
Figure 5.10: CDF values for horizontal mode shapes 1, 3 and 5 ...121
Figure 5.11: CDF values for horizontal mode shapes 1, 2, 3, 4 and 5 ...121
Figure 5.12: CDF values for horizontal mode shapes 1, 2, 3, 4, 5 and 6 ...122
Figure 5.13: MCDF values for horizontal mode shapes 1, 2 and 3 ...123
Figure 5.14: MCDF values for horizontal mode shapes 1, 3 and 5 ...123
Figure 5.15: MCDF values for horizontal mode shapes 1, 2, 3, 4 and 5 ...124
Figure 5.17: Flexibility deflection for the scoured and repaired finite element model as well
as the absolute difference between the two for horizontal mode shapes 1, 2 and 3 ...126
Figure 5.18: Flexibility deflection for the scoured and repaired finite element model as well
as the absolute difference between the two for horizontal mode shapes 1, 3 and 5 ...127
Figure 5.19: Flexibility deflection for the scoured and repaired finite element model as well
as the absolute difference between the two for horizontal mode shapes 1, 2, 3, 4 and 5 ...128
Figure 5.20: Flexibility deflection for the scoured and repaired finite element model as well
as the absolute difference between the two for horizontal mode shapes 1, 2, 3, 4, 5 and 6 ..129
Figure 5.21: The curvature of the flexibility based deflection for the scoured and repaired bridge and the absolute difference between the two for mode shapes 1, 2 and 3 ...131
Figure 5.22: The curvature of the flexibility based deflection for the scoured and repaired
bridge and the absolute difference between the two for mode shapes 1, 3 and 5 ...132
Figure 5.23: The curvature of the flexibility based deflection for the scoured and repaired
bridge and the absolute difference between the two for mode shapes 1, 2, 3, 4 and 5 ...133
Figure 5.24: The curvature of the flexibility based deflection for the scoured and repaired
Chapter 1. Introduction
Damage detection techniques that are accurate and easy to implement are vital to the health
of bridges in the United States following a meteorological event. The U.S. Department of
Transportation Bureau of Transportation Statistics reported in 2010 that 11.5% of all
highway bridges were structurally deficient and 12.8% were functionally obsolete.
Structurally deficient bridges are in need of repair because the expected load cannot be
carried, while functionally obsolete bridges are in need of immediate repair because their
intended design requirements cannot be met, according to The Association of American State Highway and Transportation Officials (AASHTO) 2009 Bottom Line Report. Wardhana et
al. [2003] completed a comprehensive study of bridge failures from 1989 to 2000 and
observed that hydraulic damage was a leading cause of bridge failures in the U.S.
representing nearly 53% of all bridge failures studied. Hydraulic damage is a common result
of coastal storms. Storm surges may occur during a storm due to a rapid rise in the level of
water. Strong wind-driven waves in combination with a storm surge apply heavy loads to a
bridge deck resulting in connection damage and possible bridge deck uplift [Douglass et al.,
2004]. The resulting strong currents also cause sediment to displace at the streambed level
resulting in scour of the pilings. The scour process is cyclic because sediment is removed
during a storm surge and as the water recedes, scour holes refill with displaced sediment
[FHWA, 2001]. This makes field monitoring for the existence and magnitude of scour
damage very difficult. The holes refilled with sediment are not compacted enough around
the pilings for the structure to receive the required support; therefore, measuring streambed
elevations may not be significant in capturing the full extent of scour [De falco and Mele,
2002]. Also the streambed elevation measurements can be altered due to the presence of
storm debris. Visual inspections may overlook the true magnitude of the damage if scoured
holes have been refilled with sediment. Monitoring methods that involve underwater
techniques that avoid these common monitoring issues. This need has led to development of
methods that examine the change in vibration characteristics between damaged and healthy
structures [Doebling et al., 1996]. This study explores vibration based damage detection
techniques applied to a coastal bridge as a means to identify the existence of scour at the
pilings.
1.1 Structural Health Monitoring
The application of different field methods to monitor structures for scour is referred to as structural health monitoring. Structural health monitoring is defined as the process of
determining and tracking structural integrity and assessing the nature of damage [Chang et
al., 2003]. Global health monitoring methods are used to determine if damage is present in
the entire structure, while non-destructive evaluations are used as local health monitoring
methods to identify the exact location and extent of damage [Chang et al., 2003]. These two
types of methods are both vital to identifying damage in a structure; however, oftentimes
present a challenge when applying the methods because of attempts to identify local damage
by observing the global response of a structure [Mosavi, 2010]. Adams [2007] describes the
non-destructive evaluations as involving the identification of the loads on the structure,
resulting mechanical damage, the growth of the damage and the future performance of the structure’s components as the damage accumulates. These qualities can be assessed by
modeling the structure, obtaining measurements, data analysis and forming predictions.
Diagnostics is the identification of damage and aims to detect, locate and quantify damage.
This can be done through signal processing and feature extraction of the models measured
variability [Adams, 2007].
Damage alters a structures performance because of a permanent change in its mechanical
the basis of using vibration based damage detection techniques. Gathering and analyzing the
changes in dynamic characteristics such as natural frequencies, damping ratios, mode shapes
and flexibility between damaged and healthy structures can be used to quantify the existence
and extent of damage [Elsaid, 2011]. There is an added challenge in that data from an
undamaged structure is necessary for comparisons with the damaged structure. Also the
modal characteristics collected must be accurate for the damage detection features to
successfully identify potential damage. Other aspects can alter vibration responses of a
structure such as varying temperature, moisture and loading [Mosavi, 2010].
1.2 Research Significance
This study is focused on the accuracy and practicality of using vibration based damage
detection methods to identify the existence and location of scour on an existing coastal
bridge. Monitoring for scour was chosen as a focus in this study because hydraulic damage
is a common cause of bridge failures and is not simple to identify. Vibration based damage
detection methods would require obtaining modal characteristics by accessing the
superstructure above water. No underwater investigations would be necessary which would
ease operator labor as well as protect equipment from possible water damage. Vibration
based monitoring would be able to assess the condition of the structure itself instead of
simply evaluating the condition of the damage. This is important since scour related damage
cannot always be visually observable. Vibration based methods have been applied to
theoretical finite element models and could successfully identify the existence and location of
damage; however, field application presents a variety of variables that cannot be captured in
a theoretical model, such as environmental conditions and ambient loading. It is necessary
for these methods to be applied to an existing structure in order to validate their ability to
detect the presence and location of scour. The output of this research aims to lead to a
remaining strength of a bridge. Quickly identifying damage is essential for bridges that are
on critical evacuation or emergency response routes.
1.3 Research Scope and Methodology
In order to apply the vibration based damage features to an existing bridge, the dynamic
characteristics must be known of the healthy bridge as well as the dynamic characteristics of
the scoured bridge. A bridge was chosen for this study that had existing scour at the piles.
The damage was discovered by an underwater dive team. The inspectors recommended the bridge repair be made priority. The dynamic characteristics were collected from the bridge
while the piles were scoured and once again after the bridge had been repaired. The damage
features were applied based on the change in dynamic characteristics between the two
conditions. Accelerations resulting from an impact hammer were collected through field
monitoring and signal processing was used to determine the dynamic characteristics. The
dynamic characteristics analyzed include the natural frequencies, the curvature of the
horizontal mode shapes, the flexibility deflection of the structure and the curvature of the
flexibility deflection. A maximum in the change of dynamic characteristics should identify
the location of damage. A finite element model of the bridge was developed to assist in
selecting the equipment used to perform the monitoring. Damage features were extracted
using data from the finite element model. This was used to compare the effectiveness of the
damage features extracted from theoretical data compared to using data from an existing
coastal structure.
1.4 Organization of the Thesis
This thesis consists of six chapters describing the application of the proposed damage
Chapter 2 discusses the current conditions of bridges in the United States and common
causes of failure. Scour is presented as a serious and difficult to identify type of bridge
damage that often results from coastal storms. The advantages and disadvantages of the
field application of different types of scour monitoring techniques are discussed. This
includes discussion of the development of vibration based damage techniques that can be
easily and quickly performed on a coastal structure following a storm.
Chapter 3 describes the bridge selected for field monitoring. The process of developing the
testing setup including the number and spacing of accelerometers is described. The selection of the accelerometers, impact hammer and data acquisition system is discussed. A
description of the repairs made to the bridge scour is presented. Discussion and comparison
of the two testing days on the scoured and repaired bridge is also included.
Chapter 4 presents the analysis results of the accelerations collected from the field
monitoring. The chapter also describes how the damage features were calculated. The raw
accelerations are processed through MATLAB and Excel to gather the frequency response
functions that correspond to natural frequencies and horizontal mode shapes of the structure.
The change in the horizontal mode shape curvature, the flexibility deflection and the
curvature of the flexibility deflection is demonstrated for the scoured and repaired bridge.
Discussion on the accuracy of the damage features is included.
Chapter 5 discusses the creation of the finite element model of the bridge. The process of
gathering accelerations and applying the damage features presented in Chapter 4 is discussed.
The challenges of building an accurate finite element model for existing bridges are
presented. This testing does not require a finite element model, so the related advantages are
Chapter 6 concludes the findings from this study. This includes evaluation of the
effectiveness of the damage features ability to detect scour as well as the practicality of
implementing this type of testing on an existing structure. Also the finite element model and
related challenges are discussed. Future research recommendations are made as well in this
Chapter 2. Literature Review
2.1 Introduction
Following coastal storms, scour of pilings is a major concern for coastal bridges in the United
States. This type of damage can be difficult to detect because it is not always visually
observable. Much research has been focused on developing scour monitoring techniques that
can be easily implemented in the field after a coastal storm. Vibration based damage
detection techniques have been developed to detect damage in various structures and could potentially be used to detect scour at the pilings of coastal structures. This chapter will
highlight scour as a critical concern to coastal bridges, explore different developed scour
monitoring techniques and consider vibration based damage techniques as an option for scour
monitoring.
2.2 Review of Bridge Failures
Proper detection of damage is critical to the health of bridges in the United States due to the
current condition of many bridges. The U.S. Department of Transportation Bureau of
Transportation Statistics reported in 2010 that there were 604,460 highway bridges in the
country. Of these bridges, 69,220 bridges were considered structurally deficient and 77,412
bridges were functionally obsolete being 11.5% and 12.8% of the total bridges, respectively.
The Association of American State Highway and Transportation Officials (AASHTO) 2009
Bottom Line Report defines structurally deficient bridges as structures that cannot carry the
expected load and are in need of repair. Functionally obsolete bridges are structures unable
to fulfill their design and require immediate repair. AASHTO is concerned with the age of
the majority of bridges in the US. Nearly half of all bridges are more than 40 years old and
The most recent comprehensive study of bridge failures in the US was completed by
Wardhana et al. [2003] who examined 503 failed bridges occurring from 1989 to 2000 across
the country. The study classified failed bridges as structures or components of structures that
can no longer perform the function for which it was designed. The average age of the failed
bridges was 52 years and the bridges age ranged from under construction to 157 years. The
majority of the bridges studied were steel, concrete or timber. The most common category of
failures was beam/girder failures that made up 37.2% of the failures. Truss failures were also
common representing 23.1% of the total failures. Figure 2.1 shows the number of bridge
failures according to the year the failure occurred.
Figure 2.1: Number of failed bridges distributed by year [Wardhana et al., 2003]
Figure 2.1 shows a peak in bridge failures in 1993 with 112 failed structures. The peak is
attributed to flooding of the Mississippi and Missouri rivers and tributaries that caused bridge failures across Iowa, Minnesota and Missouri. The second highest peak in Figure 2.1
occurred in 1996 and is also attributed to flooding. Six of the ten states with the most failed
bridges were coastal states comprising of 36% of the total bridge failures. The National 55
46
31 28 112 33 45 59 18 33 26 17 0 20 40 60 80 100 120
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
Oceanic and Atmospheric Administration (NOAA) states that flooding is often a result of
coastal storms. Flooding causes more damage in the US than any other severe weather
related event and the damage due to flooding has increased greatly since the early 1990’s.
Table 2.1 lists the most common types of bridge failures from the Wardhana et al. [2003]
study. Hydraulic damage caused the majority of failures representing 52.88% of the total
failures with 32.8% of failures being related directly to flooding and 15.51% of failures from scour. Scour is defined as “erosion or removal of streambed or bank material from bridge
foundations due to flowing water, usually considered as long-term bed degradation, contraction and local scour” by the Federal Highway Administration. Based on this
definition, flooding and scour are closely related issues.
According to NOAA coastal counties only represent 17% of the nation’s land, but are home to half of America’s population. Coastal areas are at high risk of coastal storms, flooding,
coastal erosion and land subsidence. These hazards are a threat to infrastructure whose
function is critical for evacuations and emergency response during a coastal storm.
2.2.1 Storm Surge Bridge Failures
Douglass et al. [2004] studied bridge and road failures after Hurricane Ivan passed between
Florida and Alabama in 2004. The researchers observed that the majority of damage was due
to a combination of storm surge and waves on top of the storm surge. A storm surge is
defined as a rapid rise in the level of water that occurs during a storm. Figure 2.2 is the failed
portions of the bridge deck on Interstate-10 crossing the Escambia Bay arm of Pensacola
Bay. The damage is the result of a storm surge that raised the water level to the height of the
bridge deck so that large wave loadings impacted the bridge spans. The force impacting the
Table 2.1: Types of bridge failures from 1989 to 2000 [Wardhana et al., 2003]
Failure causes and events Number of occurrences Percentage of total
Hydraulic 266 52.88
Flood 165 32.80
Scour 78 15.51
Debris 16 3.18
Drift 2 0.40
Others 5 0.99
Collision 59 11.73
Auto/truck 14 2.78
Barge/ship/tanker 10 1.99
Train 3 0.60
Other 32 6.36
Overload 44 8.75
Deterioration 43 8.55
General 22 4.37
Steel deterioration 14 2.78
Steel-corrosion 6 1.19
Concrete-corrosion 1 0.20
Fire 16 3.18
Construction 13 2.58
Ice 10 1.99
Earthquake 17 3.38
Fatigue-steel 5 0.99
Design 3 0.60
Soil 3 0.60
Natural Disaster 2 0.40
Miscellaneous/other 22 4.37
Figure 2.2: Interstate-10 bridge failure resulting from Hurricane Ivan [Douglass et al., 2004]
Figure 2.3 shows the approach to the Bob Sykes Bridge abutments where the embankments
have been eroded due to exposure to storm surge and waves. From inspection of Figure 2.3
the former soil line is visible before the abutment soil washed away causing scour. Figure
2.4 is of the Little Lagoon bridge deck with damage caused by storm surge currents and
waves. Figure 2.5 is of the repaired Escambia County Road 191 Bridge that suffered from embankment erosion and deck damage due to storm surge current and wave forces.
Douglass et al. [2004] notes that each of these bridges are located very close to the ocean and
likely received a much higher combined force from the storm surge current and waves than
inland bridges experienced from flooding of rivers. Coastal bridges are at the frontline for
Figure 2.3: Soil washed away from the Bob Sykes Bridge abutment due to Hurricane Ivan
[Douglass et al., 2004]
Figure 2.5: Escambia County Road 191 Bridge repaired after Hurricane Ivan [Douglass et al.,
2004]
One year after Hurricane Ivan, Hurricane Katrina made landfall in the same region of the
country causing additional damage. Combined, the two storms cost an excess of 1 billion
dollars to rebuild coastal bridges [Chen et al., 2009]. Much like Hurricane Ivan, the failure
of most bridges resulted from waves in combination with storm surge current. Chen et al.
[2009] describes the increased force resulting from water rising to a level with strong winds
that generate waves over a short distance to strike the bridge superstructure causing both an
uplift force and horizontal force. The uplift force can exceed the weight of the bridge deck and the horizontal force can exceed the resistance of the connections of the deck to the piling
caps. This causes the deck to detach from the substructure. This is shown in Figure 2.6 in a
photograph of the failure of the US 90 bridge over Biloxi Bay in Mississippi during
Figure 2.6: Collapse of US 90 Bridge deck over Biloxi Bay Mississippi [Chen et al., 2009]
Coastal storms are an obvious threat to coastal bridges, but sometimes the damage due to a
storm is less apparent. Identification of damage that is not clearly visible is essential in
assessing the condition of coastal structures after a meteorological event.
2.2.2 Scour Damaged Bridges
result, soil becomes loosened and suspended by the flow of the water and washes away from
the site. The effect of scour is related to the velocity and turbulence of water as well as the
type of soil surrounding the foundation [FEMA, 2009]. Scour is often the resulting damage
due to a storm surge. Figure 2.7 shows localized scour at a foundation pile after Hurricane
Ike.
Figure 2.7: Localized scour at a pile due to Hurricane Ike [FEMA, 2009]
Difficulties arise when estimating the magnitude of scour occurring on underwater
foundations. This is due to the cyclic nature of the process that causes scour. Scour can be
very deep at the time of flooding and then as waters recede the scour holes can refill with
displaced sediment [FHWA, 2001]. Shallow foundations suffering from scour are subject to
collapse. Scour of deeper foundations may collapse the structure or experience uneven
vertical settlement of the structure along with lateral displacement from lateral loads. Also,
the unbraced length of the pilings is increased resulting in an increase of the bending moment
The three main types of scour affecting most bridges include long-term degradation of the
river bed, general contraction scour at the bridge and local scour at the piers or abutments
[FHWA, 2001]. These three types of scour are independent from one another. Long-term
degradation refers to natural or man-made changes in the trend of the stream or watershed
that causes sediment to be displaced overtime and the elevation of the streambed is changed.
This type of scour can be identified by monitoring streambed elevations overtime. General
scour and local scour can be one of two conditions, clear-water scour or live-bed scour.
Clear-water scour occurs when no bed sediment material is transported except locally around
pilings. Live-bed scour is the transfer of bed material and is often cyclic because the scour hole develops during a flood and later refills as the storm surge water elevation falls. During
flooding, streams with coarse-bed material typically have clear-water scour when discharge,
the volume of water passing a point, is low. Live-bed scour happens at higher stream
discharges and clear-water discharge occurs as the flood subsides and the stream discharge
falls. The condition of scour can be determined based on the particle size of the streambed as
well as the velocity of the flow of the water. In sand-bed streams, most commonly found
near the coast, live-bed scour fluctuates between scour and refill with a maximum scour
depth about 30% higher than the equilibrium depth experienced between scour and refill.
For the same sand-bed, clear-water scour increases steadily overtime and will generally reach
a maximum of 10% greater scour depth than the live-bed equilibrium depth. This change in
Figure 2.8: Scour depth in a sand-bed stream over time [FHWA, 2001]
Contraction scour occurs during a flood when the flow area of the body of water is decreased
resulting in an increase in average flow velocity. An increase in water velocity causes
sediment to be removed and relocated. Contraction scour may occur because of bridge
abutments or piers obstructing the area of flow. Contraction scour involves sediment
removal for typically all or most of the channel width. Live-bed contraction scour is
typically cyclic because soil washes out and falls back into place; therefore, it is difficult to
measure after a meteorological event. Clear-water scour often occurs locally around the piles
or abutments that are obstructing the flow of water. The acceleration of water around the
obstructions removes soil at the base by creating vortices. The type of vortex created around
the obstruction is known as the horseshoe vortex. The rate of transport of sediment from
around the base of the obstruction is greater than the transport rate in the surrounding flow
causing a scour hole to form. This action is demonstrated in Figure 2.9. As the size of the
after the hole is formed will decrease creating equilibrium [FHWA, 2001]. Suspended
sediment may settle into the scour hole.
Figure 2.9: Horseshoe vortex around the base of a piling causing local scour [FHWA, 2001]
Local scour depth at piers and abutments is mainly governed by the velocity and depth of
flow, size and shape of the structure obstructing flow, size and gradation of the bed material,
angle between the flow and obstruction object, bed configuration and possible jams or debris
[FHWA, 2001]. Based on the many factors affecting scour, identifying the extent of scour is
difficult to determine or predict due to a coastal storm.
Researchers from the Naval Research Laboratory [Teague et al., 2006] placed moorings off
the coast in the Gulf of Mexico and took measurements of the change in sediment levels after
Hurricane Ivan passed over the location of six of the moorings. They observed a displacement of over 100 million cubic meters of sediment from the 35x15 km region
wind stress and recorded depths of 360 mm and 320 mm of scour. This was caused by wave
velocities and wind currents occurring at the ocean floor that generated a vortex causing
sediment suspension and removal. Therefore, the forces from storm surge currents and
waves that threaten collapse of bridge decks are the same forces that act at the sediment level
and displace soil surrounding the substructure.
Robertson et al. [2007] observed that in Hurricane Katrina the majority of structures were
damaged due to uplift, excessive lateral loading, restraint failure, debris effects and scour.
The researchers observed extensive levels of scour around bridge abutments and piers in the Gulf Coast area that led to collapse or partial collapse of many of the structures. The type of
scour that affected most of the structures is called liquefaction-induced scour that occurred as
a result of a storm surge. As the name implies, the soil behaves like a viscous liquid and is
easily transported with the flow of water. Saturated sandy soil is subject to local
liquefaction-induced scour due to wave action. Figure 2.10 is the US 90 bridge in the town
of Bay St. Louis, about 30 miles from the Biloxi Bay Bridge in Figure 2.6. The Bay St.
Louis Bridge experienced extensive scour along the Mississippi coastline due to Hurricane
Katrina at a maximum depth about 4.5 meters. The approach to the abutments as well as the
Figure 2.10: US 90 bridge in Bay St. Lewis experienced up to 4.5 meters of scour during
Hurricane Katrina [Robertson et al., 2007]
Just as scour can occur on coastal structures due to flood water, it can also be caused by tidal
fluctuation. Scour that occurs due to tide changes may occur in two directions, as opposed to
one main direction for scour occurring due to a storm surge. Structures in waterways
exposed to tidal change may experience increase in soil discharge from the foundations as the
waterway area increases with the tide [FHWA, 2001].
Hydraulic damage is the most frequent cause for bridge failures representing nearly half of
all failures [Wardhana et al., 2003]. Both uplift of bridge decks as well as scouring of
Monitoring for scour was chosen as a focus for the Department of Homeland Security
because it is an issue that is not always visually observable. Proper identification of scour
damage immediately following a meteorological event is essential for emergency response
and recovery.
2.3 Methods for Monitoring Bridges for Scour
Much research is focused on developing methods to monitor scour because accurate
assessment of scour damage is difficult to determine. Flood water accelerates around an object blocking its flow, such as a pile, and removes sediment from around the base creating
scour. As the scour hole increases in size, the flow in the vortex decreases causing the
amount of sediment being displaced to decrease [FHWA, 2001]. The suspended sediment
may settle into the scour hole and will not be compacted enough to provide adequate
confinement for the pile. This action shows that simple measurements of sediment levels
after a flood may not be significant in fully capturing the extent of scour [De falco and Mele,
2002].
The US Geological Survey (USGS), the Federal Highway Administration (FHWA) as well as
various states Department of Transportation (DOT) organizations have invested in scour
related research. The USGS in collaboration with the FHWA have created a database of
scoured bridges across the US and have drafted documents and training materials with
recommendations for engineers facing scour related design problems. In addition to
government research, there have been many academic research programs related to scour
2.3.1 Streambed Elevation Measuring Methods
Mueller et al. [1999] collaborated with the USGS to investigate scour monitoring methods
using sounding weights and echo sounders to measure the streambed elevation changes
during and after a flood. These two methods represent common portable streambed elevation
measuring devices. The streambed elevation is measured by observing the distance between
a known datum and the streambed.
The sounding weights are lowered and raised by a bridge crane or truck-mounted boom on the bridge deck to calculate the streambed depth. The weights used are typically 45 to 136
kilograms. This method could be used during a flood because the heavy weights will be able
to sink to the bottom despite strong water currents. The Red River flooded in May of 1990
and sounding weights were used to measure changes in the streambed. The weights were
able to successfully free-fall and measure the depth, while echo sounders were unsuccessful
because of extreme water turbulence and air entrainment. There are several significant
disadvantages to using this method. During a flood or shortly afterwards, operating a crane
on the bridge deck may be dangerous. During high currents, the weights may be swept
downstream while being lowered causing inaccurate measurements. It is critical for the
weights to drop with vertical precision for accurate measurements to be made. A general rule
for operators of this method is to disregard using weights if the depths are greater than 10 m
and velocities of flow are greater than 3 m/s. Also, debris can snag the line as the weight
descends. If the line breaks, the operators are at a safety risk and the weight may become
lost. This method is only able to capture the depth of scour, not the size of the scour hole.
The disadvantages of this system are significant, so the Mueller et al. explored using more
advanced echo sounders to measure the streambed depths [Mueller et al., 1999].
reflect off of the streambed and back to the transducer. The researchers used modified water
skis to encase and navigate the transducer around the bridge pier as shown in Figure 2.11.
Figure 2.11: Echo sound transducer measuring streambed depth [Mueller et al., 1999]
The test setup performed well during three major floods because it created a small amount of
drag and was easy to control in turbulent water. An advantage of the echo sounders is that
measurements can be quickly recorded at many different points so the entire scour hole
measurements can be collected. Echo sounders are able to measure the depth to sediment for
depths larger than 3 m deep and water velocities less than 4 m/s. Echo sounders did not
perform well making continuous measurements during an intense flood because the signal
was vulnerable to high levels of turbulence and air entrainment. Also, echo sounders may
record faulty data if there is a large amount of suspended sediment in the water affecting the
recorded signal. The recorded data would also be altered by possible storm debris collected
around the pilings. As mentioned before, sediment can refill scour holes and not be
would measure to where the sediment has settled and would not be able to provide any data
regarding the compaction of the sediment.
The magnetic sliding collar (MSC) system is an alternative method to examine the depth of a
streambed examined by researchers Nassif et al. [2002] in collaboration with the New Jersey
Department of Transportation. The MSC system is comprised of a steel pipe that is placed
vertically in the streambed and a sliding collar moves with the change in sediment elevation.
Magnets were attached to the collar and the resulting magnetic fields were monitored to
determine the depth of the collar in the sediment. This device may be used during a flood to measure the maximum depth of scour experienced. If sediment falls back into the scour hole,
the collar will be buried. The steel pile should be embedded into the sediment a minimum of
1 meter below the expected scour depth. The researchers implemented MSC devices at
several locations on the Matawan Bridge in New Jersey to monitor change in the sediment
levels over a time span of three and a half years. The bridge is subject to tidal changes and
possible storm surge from the Raritan Bay. No significant storms occurred over the three and
half years of monitoring the bridge; therefore, no change occurred in streambed elevations
indicating scour damage. Some disadvantages acknowledged by the researchers were that
the magnetic arrays used in the MSC had a spacing that limited the measurements to be
recorded to the nearest 6 inches. This is a large interval and on a smaller structure significant
scour levels may be overlooked. The researchers describe the installation of the MSC system
as extensive. While it can be installed from the superstructure, it requires a pneumatic post
driver to embed the steel pipe vertically 1 meter into the streambed. Aggravating scour
critical sediment may have negative consequences. Turbulent water current could make the
steel pipe installation more difficult considering the rod should be installed normal to the
steam bed surface. Also, the results recorded from the MSC could be altered by any floating
2.3.2 Fiber-Optic Sensors Bridge Monitoring
Zarafshan et al. [2012] developed a method to monitor bridge piers and abutments for scour
by observing vibration characteristics of a sensor rod embedded into the streambed. The
method relies on the interaction between the sensor rod and the streambed material. There is
a relationship between the free length of the rod and the rod’s natural frequency; therefore, as
scour occurs the natural frequency of the rod will change. The fiber-optic Bragg grating
(FBG) dynamic sensor was used to determine the natural frequency of the rod. The FBG
sensor was chosen because it can be safely exposed to water as opposed to electrical sensors. FBG sensors were beneficial in this study because the sensors were small in size, immune to
electrical and electromagnetic noise, and recorded high resolution signals. Figure 2.12 shows
the rod, FBG sensor and fiber-optic cable implanted in the riverbed.
Figure 2.12: Fiber-optic sensor connected to rod implanted in riverbed [Zarafshan et al., 2012]
The researchers used this method to monitor a bridge over Salt Creek River in Illinois. The
bridge abutments and piers face upstream flow at a skewed angle and are at an increased risk
of the piers and abutments before monitoring began. There were 14 sensors distributed
around the piers and abutments and each rod was embedded to a calibrated length. The
entire monitoring system included the sensors connected to a rod and a fiber-optic cable
routed to an optical interrogation unit that converted the signals of the sensor to vibration
responses, a computer to control the interrogation unit, data acquisition system for the
processing, as well as a cellular wireless modem connected to the field computer to
wirelessly transfer the real-time data to the laboratory. There was a concern for protecting
the rods in the streambed, so steel barrier rods were embedded around the sensor rod in an
effort to protect the sensors from floating debris. Figure 2.13 shows the barrier rods being installed.
Figure 2.13: Barrier rod installation in streambed [Zarafshan et al., 2012]
The scour levels were measured in real-time over a ten day period and it was observed that
the sensors were capable of monitoring scour as well as accumulation of the sediments.
Figure 2.14: Scour levels recorded from sensor located near a bridge pier [Zarafshan et al.,
2012]
The sensors were able to successfully track the changes in sediment level, but it should be
noted that accumulation of sediment on a scoured bridge pier does not create additional
structural support to the bridge. Scour is dangerous because it alters the pier’s ability to
provide support to the structure, so it is important when monitoring scour to evaluate the
structural support provided by the compaction of sediment around the pilings. While the
barrier rods protected the sensors from major debris, smaller pieces of debris such as leaves
collected around the sensors and altered the recorded signal. The sensor rods had to be
regularly inspected and cleaned of any collecting debris for the recording signal to be
accurate. Regularly clearing debris would add labor to the monitoring process. Installing the
sensor rods underwater to a very specific depth would also be a laborious task. As discussed
earlier, a vortex can form around a water flow obstruction where the sediment is locally
removed. The sensor rod, barrier rods and any debris between the rods could act as a water
flow obstruction and experience local scour unrelated to scour created around the bridge pier.
Also, if a flood had occurred during the monitoring time frame, there would have been likely
could have dislodged the sensor rod and the barriers. The sensors washing away would be a
costly mishap.
An alternate fiber-optic sensor method was developed by Manozi [2011] that used FBG
sedimeters to evaluate the temperature change along the depth of a bridge pier. The
temperature of sensors in the flowing water should be lower because of thermal resistance
compared to the sensors buried in the river bed. Figure 2.15 is a diagram of the sensors along
an underwater bridge pier.
Figure 2.15: Sensors on a bridge pier and the cross section of a sensor [Manzoni et al., 2011]
Laboratory tests proved that there was a temperature variant between the sensors in flowing
water and the sensors locating in the saturated soil. Plans are being made to apply the
monitoring method in the field. The researchers acknowledge there will be a challenge related to the vulnerability of the system when exposed to flowing water [Manzoni et al.,
2011]. As with the previously mentioned fiber-optic sensor method, it would be a costly
Methods of evaluating scour above water will avoid most of these issues related to debris
interference or underwater damage to expensive equipment. Also, it is labor intensive to
properly install equipment underwater.
2.3.3 Robot Assisted Bridge Monitoring
Unmanned marine vehicles (UMV) were developed as an option for post-disaster bridge
inspections [Murphy et al., 2011]. The researchers evaluated the performance of a UMV’s
ability to inspect for damage on the Rollover Pass Bridge over the Galveston Bay three months after Hurricane Ike. The UMV should be able to inspect the bridge substructure for
damage and map the debris field by taking photographs. The results were compared to
inspections done previously by a dive team. The UMV used to inspect the bridge is shown in
Figure 2.16.
Figure 2.16: UMV used to inspect Rollover Pass Bridge called a Sea-RAI USV [Murphy et
The UMV is operated by three people who each take a task to pilot the UMV, gather images
through the attached cameras or to act as a safety officer. Other engineers are involved by
collecting and observing data output. The UMV was successful in identifying debris and did not detect scour damage at the pilings. The UMV results were similar to the dive team’s
field inspection results. There are advantages to using the UMV compared to investigations
completed by a dive team. After a storm, divers inspecting a damaged structure may be at a
high danger risk which could be avoided by using a UMV. Also, human error is avoided as
the UMV is less likely to miss important details. Disadvantages observed by the researchers
include vulnerability of the UMV to extreme environmental conditions such as changes in tide or current. Large debris or changes in topography may act as an obstacle for the UMV
navigating the water. It is necessary for near-by boat ramps to be available and undamaged
for the UMV to be used [Murphy et al., 2011]. Similar to issues discussed for the fiber-optic
sensors, there are disadvantages to underwater investigations because equipment is subject to
damage or loss because of dynamic water leading to a very costly consequence. Also, visual
inspections may not be able to fully capture the damage of a structure. Connection damage
or scour vortices refilled with loose sediment are all dangerous forms of damage that are not
visually observable.
2.3.4 Tiltmeters Monitoring
Briaud et al. [2011] investigated an above water bridge monitoring system using tiltmeters to
measure changes in the angle of the bridge deck due to damage. The researches implemented
this testing method on a Guadalupe River Bridge in Texas that had existing scour due to a
flood of the river in 1998. The monitoring is aimed to identify if the condition was
worsening and the bridge was losing stability. Figure 2.17 shows the instrumentation plan on
Figure 2.17: Tiltmeter instrumentation plan [Briaud et al., 2011]
Tiltmeters were installed in two directions to measure the angle change along the axis of
water flow and along the axis in the direction of traffic. A consistent reading from the
tiltmeters corresponds to a stable bridge. If the calculated threshold value was exceeded by
the tiltmeters, the bridge was at risk for losing stability and should be closed [Briaud et al.,
2011]. There are several advantages to monitoring scour above water, such as avoiding
equipment damage due to water exposure or harsh environmental conditions. Also,
accessing and installing equipment is less laborious if done from the bridge deck than below
water on the substructure. Tiltmeters are inexpensive, simple and reliable. Damage other
than scour may be identified by observing the tilt of the bridge deck. A concern with this
method is that once damage has been identified, the bridge deck has already displaced and
the structure is at risk for failure. A successful monitoring method should be able to identify
2.4 Vibration Based Superstructure Monitoring Techniques
Tracking streambed elevation measurements, performing underwater investigations and
measuring tilt of the bridge deck are all scour monitoring methods that have previously been
implemented in the field. Most of the methods were considered successful in measuring the
amount of soil dislodged, but disadvantages exist in the field monitoring process. Ideally, a
field monitoring system should be developed that could determine if the structure is damaged
due to the presence of scour instead of quantitatively measuring the amount of scour. The
goal of monitoring is to make an assessment of the health of the structure. Furthermore, the field monitoring system should be easy to implement by an end-user, especially in the event
of an emergency. Underwater investigation techniques require specialized training that may
make monitoring immediately following an emergency event difficult. Monitoring the
bridge from the superstructure alone benefits the field monitoring operator. Also, it protects
the scour critical sediment surrounding the pilings from additional aggravation. Damage
detection techniques based on the dynamic characteristics resulting from vibrations of the
superstructure of the bridge is a developing scour monitoring system that satisfies the test
setup ideals. Zarafshan et al. [2012] performed an investigation of using fiber-optic sensor
rods embedded into the sediment around scour critical piles which uses the same vibration
response concept, but as previously discussed, it is difficult to implement underwater.
Considering the vibration response of the superstructure alone avoids underwater
interactions. Vibration based damage detection techniques were able to identify damage
successfully in theoretical and in some laboratory models and should be further investigated
by implementing in the field.
Mosavi [2010] investigated using vibration based damage detection techniques to identify the
location of damage in steel girders. He created an idealized two-span steel laboratory model
locations. Random vibrations were recorded by 15 sensors along the length of the beam.
The collected vibrations were used to evaluate damage diagnosis patterns based on statistical
pattern recognition using time series models. The damage diagnosis patterns were based on
comparison between healthy beams and damaged beams. The Fisher criterion was applied to
the healthy and damaged beam vibrations and theoretically the highest Fisher criterion value
corresponded to the sensor most closely located to the damage in the beam. Mosavi found
that the methods were successful in identifying the general region of damage in the
laboratory tested beams. The change in natural frequency and vertical mode shapes between
the damaged and repaired beams was also investigated. Mosavi also observed that the natural frequencies gathered from the vibration response of the beam generally decreased for
damaged beams. Monitoring natural frequencies could identify the presence of scour;
however, it was observed these results were altered by temperature variations.
Based on Mosavi’s [2010] findings that vibration responses could be used to detect damage
in steel beams, Elsaid [2011] explored implementing vibration based damage detection
techniques to identify and locate scour in bridges. Elsaid created an idealized two-span steel
bridge laboratory model. The support between the two spans was extended to represent
different levels of scour as well as unsymmetrical scour. The model was mounted with
evenly spaced sensors and vibrations were recorded in response to an impact hammer.