Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
i
THE UNIVERSITY OF SOUTH ALABAMA COLLEGE OF ENGINEERING
A SYSTEMS ENGINEERING APPROACH FOR EVALUATING COASTAL ROAD SYSTEM RELIABILITY USING CUMULATIVE CELERITY DISPERSION
FUNCTIONS
BY
Garland P. Pennison
A Dissertation
Submitted to the Graduate Faculty of the University of South Alabama
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
in
Systems Engineering
December 2020 Approved: Date: _______________________________________________________________________Chair of Dissertation Committee: Dr. Bret M. Webb ________________________________________________________________________ Committee Member: Dr. Robert J. Cloutier ________________________________________________________________________ Committee Member: Dr. Stephanie M. Smallegan ________________________________________________________________________ Committee Member: Dr. Eric J. Steward ________________________________________________________________________ Chair of Department: Dr. Robert J. Cloutier ________________________________________________________________________ Director of Graduate Studies: Dr. Robert J. Cloutier ________________________________________________________________________ Dean of the Graduate School: Dr. J. Harold Pardue
A SYSTEMS ENGINEERING APPROACH
FOR EVALUATING COASTAL ROAD SYSTEM RELIABILITY
USING CUMULATIVE CELERITY DISPERSION FUNCTIONS
A Dissertation
Submitted to the Graduate Faculty of the
University of South Alabama
in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy
in
Systems Engineering
by
Garland P. Pennison
B.S.C.E., Louisiana Tech University, 1979
M.S.C.E., Louisiana Tech University, 1993
December 2020
ii
ACKNOWLEDGMENTS
I will complete my 4th decade of engineering practice this November since
graduating from Louisiana Tech University in Ruston, LA with a Bachelor of Science in
Civil Engineering. I have learned so much over those years, that books could not capture
it all. But I am also of the age, that I humbly acknowledge the limits of my finite
understanding. This is not a complete work, but an opening chapter, in which we advance
epistemic knowledge by finding order in chaos, not previously identified by others.
The University of South Alabama provided a nurturing environment for my
doctoral education. Dr. Robert Cloutier (INCOSE SEBoK chief-editor), and Dr. Henry
Lester instructed and challenged, each from their unique systems perspectives. They each
embody a high standard of systems engineering excellence, which is humbling as a newly
anointed systems engineering professional. Dr. Stephanie Smallegan and Dr. Eric
Steward both challenged my understanding and conclusions as valiant members of my
committee, providing additional perspective to both encourage and manage research
expectations. The other professionals who encouraged and supported me while at South
include Dr. Kevin White, Dr. John Cleary, and Dr. Christy West.
The unexpected reward of this multidisciplined IN-CORE research project is the
extensive group of fellow professionals that I came to know personally through the
numerous research meetings, workshops, conferences, and online engagements. Those
iii
that I must recognize include Dr. Scott Douglass, Dr. Jamie Padgett, Dr. Ioannis Gidaris,
Dr. Yousef Darestani, Dr. John van de Lindt, Dr. Bruce Ellingwood, and Dr. Dan Cox.
My fellow students at South treated this senior engineer as an equal in learning, which
kept me young at heart, and made me one with them. The staff who encouraged me; Ms.
Shirell Dortch, Ms. Brenda Poole, and Ms. Ronda Girardeau; all helped me through the
trials and storms of the business side of academia with calm assurance.
Dr. Bret Webb, my trusted advisor and friend, applied some serious dynamic
forcing to get me to the other side of this academic journey and I am forever in his debt. I
am grateful for the opportunity to have learned under such a tremendously talented
coastal engineer and educator, who bridges practical and theoretical with the ease of a
master. This pursuit required the support and encouragement of my employer and
colleagues at HDR, who provided flexibility for me to make frequent trips to West
Mobile to pursue my dreams. To my family and friends who encouraged me in this
journey, I say thank you from the bottom of my heart. To my children Christy and Arin
Pennison, thanks for your love and support. To my wife and best friend forever, Carrie
Coxe Pennison, thanks for being my stalwart companion in this long journey.
Funding for IN-CORE study provided by Cooperative Agreement
70NANB15H044 between NIST and Colorado State University, and through a
subcontract from The University of North Carolina at Chapel Hill as part of the DHS
Coastal Resilience Center of Excellence. These sources of support gratefully
acknowledged and appreciated. All views expressed in this paper are those of the author
and do not necessarily reflect the views of the funding organizations or government
institutes or departments that provided research funding.
iv
TABLE OF CONTENTS
Page
LIST OF TABLES ............................................................................................................ vii
LIST OF FIGURES ......................................................................................................... viii
LIST OF ABBREVIATIONS, SYMBOLS AND UNITS ............................................... xii
ABSTRACT .......................................................................................................................xv
CHAPTER I COASTAL ROADS – SYSTEM OF INTEREST .........................................1
1.1 Transdisciplinary Systems Research ....................................................................... 1 1.2 Portfolio Review ..................................................................................................... 4
1.3 Increased Risks from Coastal Hazards ................................................................... 6 1.4 Assessing Resilience of Coastal Infrastructure ....................................................... 7 1.5 Applying Systems Models to Coastal Roads ........................................................ 11
1.6 Framing Coastal Road Systems Resiliency .......................................................... 15 1.7 Developing Coastal Road Reliability Functions ................................................... 18
CHAPTER II ASSESSING COASTAL ROADS RELIABILITY ...................................21
2.1 Failures, Froude, and Fragility Functions ............................................................. 21
2.2 Coastal Storm Models for Damage Assessment ................................................... 23 2.3 Framing the Failure Analysis Research Objectives .............................................. 28 2.4 Assessing Failure Mode and Effects ..................................................................... 30
2.5 Assessing System Reliability and Likelihood of Failure ...................................... 34 2.6 Considering CCD Function Uncertainties ............................................................ 40 2.7 Considering Cause and Effect ............................................................................... 45
CHAPTER III LOCAL COASTAL ROADS – NEXT GENERATION, ..........................49
3.1 Introduction ........................................................................................................... 50 3.2 Stakeholder Desirements ...................................................................................... 53 3.3 Functional Requirements ...................................................................................... 57 3.4 Conclusions ........................................................................................................... 60
v
CHAPTER IV TRANSDISCIPLINARY SYSTEMS THINKING: SUSTAINABILITY
OF COASTAL SYSTEMS ..........................................................................................62
4.1 Introduction ........................................................................................................... 63 4.2 Systems Approach Framework ............................................................................. 66 4.3 IN-CORE Model Development ............................................................................ 68 4.4 Emergent Knowledge ............................................................................................ 70 4.5 Transdisciplinary Systems Thinking..................................................................... 74
CHAPTER V COASTAL ROAD SYSTEM FAILURES: CAUSE AND EFFECT, ........75
5.1 Introduction ........................................................................................................... 75 5.2 Coastal Road System Failures .............................................................................. 77 5.3 Developing the Failure Model .............................................................................. 79
5.4 Identifying Failure Mechanisms ........................................................................... 82 5.5 Resilient Design and Construction........................................................................ 84
CHAPTER VI ASSESSING COASTAL ROAD SYSTEM RELIABILITY USING
CELERITY DISPERSION FUNCTIONS ...................................................................87
6.1 Coastal Road System Reliability .......................................................................... 87
6.1.1 Transdisciplinary Research ....................................................................... 88 6.1.2 Coastal Road System Risks ...................................................................... 89 6.1.3 Peak versus Cumulative Intensity Measures ............................................ 91
6.2 Correlating Coastal and Systems Engineering ...................................................... 93
6.2.1. Event and Study Area .............................................................................. 94
6.2.1.1 Hurricane Ike ................................................................................ 95 6.2.1.2. Hurricane Katrina ......................................................................... 95
6.2.2. Damage Characterization ......................................................................... 96
6.2.2.1 CR 257 Damage ............................................................................ 96 6.2.2.2 US 90 Damage .............................................................................. 97
6.2.3 Hydrodynamic Storm Models ................................................................... 97
6.2.3.1 Galveston Testbed Models ............................................................ 97
6.2.3.2. Hurricane Katrina Models ............................................................ 99
6.2.4 Model Verification and Validation ........................................................... 99
6.2.4.1 Hurricane Ike ................................................................................ 99 6.2.4.2. Hurricane Katrina ....................................................................... 104
vi
6.2.5 IM Data Extraction and Preparation ....................................................... 104
6.2.6 Cumulative IM Analysis Techniques ..................................................... 104
6.3 Cumulative Failure Functions ............................................................................. 105
6.3.1 Critical Physical Parameters ................................................................... 106 6.3.2 Coastal Hydrodynamic Model Output .................................................... 107 6.3.3 CCD Functions ....................................................................................... 108 6.3.4 Pseudo-Froude Functions ....................................................................... 119
6.4 Assessing System Reliability .............................................................................. 121
6.4.1 CCD Function ......................................................................................... 122 6.4.2 Pseudo-Froude Function ......................................................................... 124
6.4.3 Application of Predictive Functions ....................................................... 124
6.5 Conclusions ......................................................................................................... 125
CHAPTER VII ADAPTING TO CHANGE ...................................................................127
7.1. Key Findings ...................................................................................................... 127
7.1.1 Transdisciplinary Research Benefits ...................................................... 128 7.1.2 Coastal Transportation Systems Reimagined ......................................... 129
7.1.3 Cumulative Energy Dispersion Determines Failure ............................... 130 7.1.4 Proof of Concept ..................................................................................... 133
7.2 System Adaptation .............................................................................................. 136
7.2.1 Changing the Likelihood of Failure ........................................................ 137
7.2.2 Impact of Climate Change and Sea Level Rise ...................................... 138 7.2.3 Integrating Natural System Defenses ..................................................... 140
7.3 Continued Research Needs ................................................................................. 141
7.3.1 Refining Causal Loop Fragility Functions.............................................. 142 7.3.2 Experimental Data to Evaluate Failure Functions .................................. 144 7.3.3 Integrating Systems Engineering into Coastal Road Reliability ............ 156
REFERENCES ................................................................................................................160
BIOGRAPHICAL SKETCH ...........................................................................................177
vii
LIST OF TABLES
Table Page
1. CCD values for select CR 257 locations corresponding to image shown in Figure 33
(ordered from east to west). ...................................................................................... 116
viii
LIST OF FIGURES
Figure Page
1. IN-CORE Galveston Testbed Model, Task 3.2.4. Research Team. ............................. 3
2. Research Team attendees at Galveston Testbed Model meeting at Rice University in
Houston, TX (6/3/16). ................................................................................................... 3
3. Resilience metrics proposed by Ayyub in Figure 3 [6] and Figure 4.10 [7]. ............... 8
4. Local coastal road system domain diagram [16]. ....................................................... 13
5. Local coastal road conceptual system requirements and use cases [16]. .................... 14
6. Hurricane Ike Track and Inundation Depth (Courtesy of Harris County Flood Control
District v.01/03/2020) ................................................................................................. 25
7. Hurricane Ike Storm Surge Brazoria County (Courtesy of Harris County Flood
Control District v.01/03/2020) .................................................................................... 25
8. CR 257 data point locations for modeling Hurricane Ike damage.............................. 26
9. CR 257 damage assessment characterization post-Ike [55]. ....................................... 27
10. US Highway 90 data analysis locations extending from Bay St. Louis to Biloxi. ..... 27
11. Typical failed pavement section on CR 257 (Brazoria County, 9/15/08)................... 32
12. Typical failed pavement base material on CR 257 (Brazoria County, 9/15/08). ........ 33
13. Failed pavement road section in the most heavily damaged section of CR 257
(Brazoria County, 9/15/08). ........................................................................................ 33
14. Average wave height to stillwater depth ratio and water velocity direction changes at
the threshold failed limit state for the CCD function for Hurricane Ike storm event
and CR 257 damage. ................................................................................................... 46
15. NACCS coastal storm risk management framework. ................................................. 52
ix
16. Local coastal resilience priorities for East Boston and Charlestown. ......................... 57
17. Priority functional needs for local coastal roads based on subjective valuation......... 58
18. Priority weighted functional requirements for local coastal roads based on QFD
analysis. ....................................................................................................................... 58
19. Local coastal road next generation IDEF0 diagram for stakeholders’ priority
desirements. ................................................................................................................ 60
20. Transportation system resiliency systemigram. .......................................................... 72
21. Coastal road resiliency systemigram. ......................................................................... 73
22. Coastal road resiliency systemigram. ......................................................................... 82
23. Cumulative celerity dispersion (CCD) function. ........................................................ 83
24. Failure probability function for coastal road damage during Hurricane Ike, Galveston
Texas (from Dr. Ioannis Gidaris). ............................................................................... 93
25. Comparison of modeled (ADCIRC) and measured (NOAA) time-dependent water
levels during Hurricane Ike (2008). .......................................................................... 100
26. Direct comparison of modeled (ADCIRC) and measured (NOAA) water levels with a
line of perfect agreement and a linear regression equation for all data. ................... 101
27. Model-data comparison of significant wave height and peak wave period at the
NDBC locations. ....................................................................................................... 102
28. CR 257 cumulative celerity data for 67 data points extracted from Hurricane Ike
model......................................................................................................................... 112
29. CR 257 cumulative celerity data relative to distance from road to MLW shoreline for
67 data points extracted from Hurricane Ike model.................................................. 113
30. CR 257 cumulative celerity dispersion (CCD) function with data relative to hydraulic
gradient for 67 data points extracted from Hurricane Ike model. ............................. 114
31. Heat map showing relative damage mapped from CCD function values along CR 257
on Follet's Island in Brazoria County, TX (Bing Map Imagery, 2020). ................... 115
32. Post-Ike damage summary showing levels and extents of damage as mapped by Coast
& Harbor Engineering [55]. ...................................................................................... 115
33. Enlarged aerial image showing most significantly damaged locations along CR 257
with point numbers in Table 1 (Google Earth Imagery 9/13/2008).......................... 116
x
34. Aerial images showing damage states for points along east end of CR 257 on Follet’s
Island (Google Earth Imagery 9/13/2008). ............................................................... 117
35. Heat map showing relative damage mapped from CCD function values along US 90
from Bay St. Louis to Biloxi Bay in Harrison County, MS (Bing Map Imagery, 2020).
................................................................................................................................... 118
36. CR 257 and US 90 cumulative celerity dispersion (CCD) function with data relative
to hydraulic gradient for data points extracted from Hurricane Ike and Katrina
models. ...................................................................................................................... 119
37. CR 257 and US 90 cumulative Froude function for data points extracted from
Hurricane Ike and Katrina models. ........................................................................... 121
38. Average wave height to stillwater depth ratio and water velocity direction changes at
the threshold failed limit state for the CCD function for Hurricane Ike storm event
and CR 257 damage. ................................................................................................. 123
39. Photo taken at unidentified location along CR 257 showing multiple asphaltic
pavement and base course layers in the failed section (Brazoria County, 9/15/08).
................................................................................................................................... 131
40. CCD function for all data points and all AEP storm events evaluated including
Hurricanes Ike and Katrina. ...................................................................................... 132
41. Hurricane Ike CCD function cumulative gamma distribution. ................................. 134
42. Hurricane Ike CCD(WSE(h)) estimated probability of exceeding damage limit state
relative to distance along Follet's Island (plotted from east to west). ....................... 135
43. Sea change drivers in local coastal road system planning and design. ..................... 136
44. Coastal road damage CCD function causal loop diagram [17]. ................................ 143
45. Conceptual layout of proposed wave flume modeled section. ................................. 145
46. CMSI wave tank showing wave response in scarp face. .......................................... 147
47. CMSI wave tank showing wave response in scarp face and unwatering risks. ........ 147
48. CMSI wave tank showing wave response in scarp face and scour potential. ........... 148
49. CMSI wave tank showing wave response in scarp face and progression of subsurface
flow field. .................................................................................................................. 148
50. CMSI wave tank showing wave response in scarp face and continued progression of
subsurface flow field. ................................................................................................ 149
xi
51. CMSI wave tank showing wave response as flow field reaches air infused layer.
................................................................................................................................... 149
52. CMSI wave tank showing wave response as air field migrating to scarp face creating
failure surface............................................................................................................ 150
53. CMSI wave tank showing wave response as air field migrating to scarp face. ........ 150
54. CMSI wave tank showing potential for rapid drawdown slope failure. ................... 151
55. CMSI wave tank fully overtopping wave submergence creates failure surface with air
entrapment................................................................................................................. 151
56. CMSI wave tank showing large wave break on scarp face entraining significant air
bubble. ....................................................................................................................... 152
57. CMSI wave tank showing wave response as flow field collapses bubble with
subsurface flow field. ................................................................................................ 152
58. CMSI wave tank showing ebb current conditions creating changes in subsurface
profile. ....................................................................................................................... 153
59. CMSI wave tank showing ebb current conditions creating rising air bubbles in
subsurface profile. ..................................................................................................... 153
60. CMSI wave tank showing ebb current conditions as larger bubbles disperse into finer
bubbles. ..................................................................................................................... 154
61. Thin-film water surface response to wave strike on scarp surface as observed at CMSI
wave tank. ................................................................................................................. 155
62. Scarp water surface response to wave strike on face (dimensionless with surface wave
in blue). ..................................................................................................................... 155
xii
LIST OF ABBREVIATIONS, SYMBOLS, AND UNITS
ADCIRC Advanced Circulation Model for Oceanic, Coastal, and Estuarine Waters
AEP Annual Exceedance Probability
ASTAH ASTAH Systems Engineering Software
CAD Computer Aided Design
CAS Complex Adaptive System
CCD Cumulative Celerity Dispersion
CFD Computational Fluid Dynamics
CHL Coastal Hydraulics Laboratory
CHS Coastal Hazards System
CMSI Chicago Museum of Science and Industry
CoE Center of Excellence
CORETM Vitech Systems Engineering Software
COSI Coastal Storm Impulse
CR County Road
CSHORE Cross-shore Numerical Coastal Model
CSTORM-MS Coastal Storm Modeling System
CSU Colorado State University
CZS Coastal Zone Systems
DCR Massachusetts Department of Conservation and Recreation
DHS Department of Homeland Security
DRM Disaster Risk Management
EIS Environmental Impact Statement
ERDC U.S. Army Engineer Research and Development Center
ESE Ecological-Social-Economic
EU European Union
FDOT Florida Deparatment of Transportation
FHWA Federal Highways Administration
FMEA Failure Mode and Effects Analysis
HEC Hydraulic Engineering Circular
xiii
HWM High Water Mark
IDEF0 Integrated Definition for Function Modeling
IISE Institute of Industrial & Systems Engineers
IM Intensity Measure
IN-CORE Interdependent Networked - Community Resilience Modeling
Environment
INCOSE International Council on Systems Engineering
IRGC International Risk Governance Council
IWR Institute for Water Resources
LiDAR Light Detection and Ranging
LWT Linear Wave Theory
MCDA Multi-Criteria Decision Analysis
MDOT Mississippi Department of Transportation
MHHW Mean High High Water
MHW Mean High Water
MLLW Mean Low Low Water
MLW Mean Low Water
NAACS North Atlantic Coastal Comprehensive Study
NACCS North Atlantic Coast Comprehensive Study
NAVD88 North American Vertical Datum of 1988
NEPA National Environmental Policy Act
NIST National Institute of Standards and Technology
NNBF Natural and Nature based features
NOAA National Oceanic and Atmospheric Administration
NOS National Ocean Service
OO Object Oriented
PI Principle Investigator
QFD Quality function deployment
RFP Request for Proposal
RMSE Root-mean-square error
RSLR Relative Sea Level Rise
SAF Systems Approach Framework
SCE South Coast Engineers, LLC
SEBoK Systems Engineering Book of Knowledge
SPICOSA Science and Policy Integration for Coastal Systems Assessment
SSA Study Site Application
xiv
SSE Storm Surge Elevation
SWAN Simulating Waves Nearshore Coastal Model
SWL Still Water Level
SysML Systems Modeling Language
TAMU Texas A&M University
TEACR Transportation Engineering Approaches to Climate Resiliency
UMass University of Massachusetts
USA University of South Alabama
USACE United States Army Corps of Engineers
USDOT United States Department of Transportation
Vensim Systems Software by Ventana Systems, Inc.
Vol. Volume
XBEACH Deltares Nearshore Coastal Model
λ Poisson rate, wave length
Fr Froude Number
V velocity
a wave amplitude
C wave celerity
ϵ wave number times wave amplitude
g local gravitational field strength
h still water depth relative to reference datum
Hmo spectrally significant wave height
Hs statistically significant wave height
k wave number
Tp wave period
WSE water surface elevation
ω angular wave frequency
km kilometers
m meters
m/s meters per second
s seconds
xv
ABSTRACT
Pennison, Garland, P., Ph.D., P.E., University of South Alabama, December 2020. A
Systems Engineering Approach for Evaluating Coastal Road System Reliability using
Cumulative Celerity Dispersion Functions. Chair of Committee: Bret M. Webb, Ph.D.,
P.E.
This research applied systems engineering to coastal roads in evaluating system
architecture and reliability functions. The coastal road systems interfaces are quite
complex since system boundaries extend well beyond the system of interest. Because of
the interdependence of mobility and modality in the functioning of coastal road systems,
determining when failure occurs requires defining when the system no longer can meet
the systems functional and operational requirements. Without preserving connectivity to
users and other transportation systems, even if the system is physically intact, the coastal
road system may not meet its functional requirements.
Local coastal roads represent the system of interest for assessing system
engineering functional requirements. The local coastal road designation includes those
roads that primarily serve local communities and represent systems most vulnerable to
coastal hazards. A semi-empirical function that predicts likelihood of system failure
during extreme events provides an opportunity to reduce risk and improve coastal
transportation system resilience. A resilient system is a priority for transportation
agencies in managing changing climate risks.
xvi
Functional adaptation requires optimizing the engineered system’s ability to resist
system failure caused by changes in climate and extreme weather events. Using coastal
hydrodynamic intensity measure (IM) output from a dynamically coupled
ADCIRC+SWAN hindcast of Hurricane Ike (2008), we demonstrate that a Froude-based
cumulative celerity dispersion (CCD) function enables development of coastal road
system fragility functions that predict likelihood of damage for roads subjected to storm
surge and wave forcing.
These CCD functions, evaluated at random locations along County Road 257
(Brazoria, Texas, USA) relative to offset distance from shoreline; and using discrete
water surface elevation, wave period, and velocity hourly IM data; strongly predicted the
likelihood and relative degree of coastal road damage states resulting from Hurricane Ike
with R2 > 0.99. The CCD functions also validated US 90 road system damage from Bay
St. Louis to Biloxi Bay due to Hurricane Katrina’s landfall in Mississippi.
This work demonstrates that the likelihood and degree of road damage caused by
a significant coastal storm event is primarily a function of the cumulative wave celerity
dispersion at a given location. While component fragilities are important, component
failure mechanics appear to be secondary relative to cumulative celerity dispersion and
sediment transport as primary causal factors. The innovative CCD function can assess
and mitigate coastal transportation infrastructure risks to improve road system
functionality and resiliency and may have broader applications for describing damage to
the built and natural coastal environments during hurricanes and other extreme events.
1
CHAPTER I
COASTAL ROADS – SYSTEM OF INTEREST
The cumulative celerity dispersion (CCD) functions presented in this dissertation
strongly predicts likelihood of coastal road failure and consequentially suggests potential
solutions that reduce risk and improve reliability for coastal roads. A grant from the
Colorado State University (CSU) Center for Risk-Based Community Resilience Planning
to the University of South Alabama provided doctoral research funding.
1.1 Transdisciplinary Systems Research
Wikipedia describes a testbed as a platform for conducting rigorous, transparent,
and replicable testing of scientific theories, computational tools, and new technologies.
The term applies across many disciplines to describe experimental research and new
product development platforms and environments. The Colorado State University (CSU)
National Institute of Standards and Technology (NIST) funded Community Resilience
Center of Excellence (CoE) in Fort Collins, CO proposed using the significantly
impacted area of Galveston, TX, where Hurricane Ike made landfall on September 13,
2008; to understand and model community damage, loss, and recovery from hurricanes.
Hurricanes represent a multi-hazard problem with strong winds, storm surges, rain, and
related flooding that can persist for days.
2
During these natural hazards, there are cascading consequences that increase
damage and inhibit rescue and recovery efforts. The Galveston testbed team (see Figure 1
and Figure 2) identified housing recovery as a key indicator of community resilience,
including the infrastructure interdependencies among the building, roads, water supply,
wastewater, and electric power networks. The IN-CORE Model evaluates community
impacts during extreme natural disaster events and quantitatively assesses resilience of
these community infrastructure systems1. By accurately quantifying risks using a physics-
based model, communities can assess alternative resiliency measures for infrastructure
systems.
The University of South Alabama represented by the principal investigators of Dr.
Scott Douglass, Professor Emeritus; and, Dr. Bret Webb, Professor, teamed with Dr.
Jamie Padgett, Professor at Rice University and Dr. Ioannis Gidaris, Post-Doctoral
Researcher, in assessing fragility functions associated with transportation systems and
coastal road damage during Hurricane Ike. Dr. Yousef Mohammadi Darestani
subsequently advanced the research work.
1 https://ssa.ncsa.illinois.edu/isda/projects/in-core/
3
Figure 1. IN-CORE Galveston Testbed Model, Task 3.2.4. Research Team.
Figure 2. Research Team attendees at Galveston Testbed Model meeting at Rice
University in Houston, TX (6/3/16).
4
1.2 Portfolio Review
Chapter I presents coastal road systems as the system of interest from a systemic
thinking perspective. It includes a general overview of coastal hazards, resiliency,
systems modeling, and the systems engineering research objectives related to coastal road
systems. It also provides an overview of research that advanced a requisite knowledge
base for the system of interest. The original research objective intended to develop
fragility functions using probability theory. When the correlation of failed state to wave
celerity became evident, the research direction changed to proof of concept associated
with that finding.
Chapter II provides an overview of risk and reliability analysis more from an
interdisciplinary civil and coastal engineer’s perspective relative to failure modes
observed for CR 257. This chapter represents the study period in which correlations
between CCD functions relative to damage became evident and research focused on
assessing the validity and significance of that finding. It provides an overview of the
status of systems research relative to assessing resilience and reliability of coastal roads
subjected to coastal hazards damage from increasing storm frequency and intensity.
Chapter III presents an Institute of Industrial & Systems Engineers (IISE) 2018
conference paper [1], which proposes sociotechnical and system functional requirements
for the next generation of local coastal roads. An Integration Definition (IDEF0) function
model organizes decisions, actions, and activities for subsystems in evaluating the
integrated system architecture. Systematic analysis and functional decomposition provide
a next generation approach for planning, siting, and designing local coastal roads. This
5
paper won the Best Sustainability Division Conference Paper award at the IISE 2018
Annual Conference in Orlando, FL.
Chapter IV presents an IISE 2020 conference paper addressing the benefits of
applying transdisciplinary systems thinking in assessing coastal systems sustainability
and in developing the IN-CORE model. This paper also provides a discussion of the
challenges associated with integrating researchers from various disciplines and specialties
in working to create a coherent whole. It finds great applicability when systems
engineering is critical to the core research objective.
Chapter V presents an IISE 2020 conference paper addressing the benefits of
systems analysis to assess cause and effect for coastal roadway damage when impacted
by coastal storm surge and wave hazards. Evaluating the CCD function assists with
identifying probable damage failure mechanisms at critical damage limit states.
Identification of failure mechanisms facilitates design of mitigating features to reduce
damage likelihood. This paper won the IISE 2020 Annual Conference Best Conference
Division Conference Paper award.
Chapter VI presents a paper proposed for publication in a special issue of the
Coastal Engineering Journal entitled Coastal Hazards and Risks due to Tropical
Cyclones (abstract submitted). This paper presents detailed hydrodynamics associated
with development of the CCD and pseudo-Froude models and application to coastal road
damage with two major hurricane events on the Gulf Coast. This paper represents core
research that proposes at least 3 new findings related to coastal engineering systems and
the role of wave celerity functions in predicting the likelihood and degree of damage.
6
Chapter VII provides an overview of the key research findings and other
information not included in previous chapters and provides recommendations for
continued research. It also provides recommendations for continued development and
research relative to CCD functions and application of systems engineering to coastal road
systems. Climate and cultural changes will require coastal road systems; land use
corridors; the natural environment; multimodal users; and, built infrastructure systems, to
all be reimagined. It closes by identifying future research needs related to this work.
1.3 Increased Risks from Coastal Hazards
Climate change and increased populations increase economic and life safety risks
along coastlines. While impacts of coastal hazards reasonably discourage further
development in at-risk environments, development of properties and built infrastructure
in coastal environments continues with correspondingly increased risks [2]. Enhanced
resiliency of coastal infrastructure and structures reduces risks and expedites recovery.
Multihazard risks require a systematic approach in defining and evaluating system risk,
resiliency, vulnerability, and sustainability. Improved systems analysis of coastal hazard
systems with reduced uncertainties also assists with moving the engineering design
profession towards greater use of risk-based design methodologies.
Coastal hazards include storm surge, waves, currents, and storm duration from
significant coastal events such as hurricanes, cyclones, typhoons, super storms, and
nor’easters. Coastal transportation systems incur significant damage during such events
[3-5]. Risk evaluation requires developing models that systematically identify, quantify,
and evaluate resiliency associated with fragility and recovery curves for environments,
7
populations and infrastructure exposed to these risks. Since resiliency includes the ability
to prepare for and adapt to changing conditions, correctly evaluating failure modes of
coastal transportation systems and components is essential to progressing coastal hazard
risk models.
1.4 Assessing Resilience of Coastal Infrastructure
Defining resilience with consistent terminology and quantifiable metrics is a
critical first step in advancing a systematic approach to resiliency studies. Complex and
diverse resiliency models result in ambiguous definitions applied to widely varying and
complex multihazard systems. Review of literature shows widely varying resiliency
system models. Ayyub [6] suggested a universal definition of resiliency that includes
“…the ability to prepare for and adapt to changing conditions and withstand and recover
rapidly from disruptions.”
Ayyub suggested that monotone measures provide a broader model framework
than probability theory for resiliency metrics by replacing additive property of probability
with the weaker property of monotonicity. A monotone measure requires that if A is a
subset of B, then the measure of A is less than or equal to the measure of B and the
measure of the empty set must be 0. Each increasing and decreasing subset sequence
must be semicontinuous when evaluated from below and above.
Measurable resiliency model functions failure and recovery profiles with
resilience as a function of these performance profiles and time. Ayyub theorizes that
impacted systems respond and recover differently based on failure type and infrastructure
8
performance trajectories before and after recovery as shown in Figure 3. Brittle, ductile,
and graceful categorize different potential failure modes.
Figure 3. Resilience metrics proposed by Ayyub in Figure 3 [6] and Figure 4.10 [7].
Distinct quantifiable events relate post-event recovery to pre-event conditions,
recognizing that varying rates of system performance recovery are critical components of
an effective model. Degradation terms model system deterioration relative to design life,
9
or alternately, positive improvements made during recovery. Performance segregation
metrics estimate various levels of performance breakdown during an event relative to
overall performance at a system level.
Economic valuation and benefit-cost analysis assesses the effectiveness of system
enhancement alternatives in improving resiliency. Valuation includes savings in potential
direct and indirect losses, as well as costs of Ayyub’s model gained wide acceptance in
generally defining resilience metrics in evaluating system performance (Q) with aging
effects and failure occurrence estimated with a Poisson process and rate (λ) [7]. A
significant problem exists with monotonic model functions since uncertainties extend
well beyond time and space boundaries for most systems evaluated in coastal resiliency
models. System boundaries delineating between complex infrastructure, natural features,
and climatological or weather systems lack definition in coastal environments.
Wu, Lo, & Wang [8] alternately suggested that a reliability index more effectively
ranks likelihood of failure, since probability remains problematic when evaluating built
infrastructure risks for natural hazards. Current challenges in reliability analysis include
lack of failure data; difficulty in validating or calibrating calculated probability and
reliability functions; and, scarcity of statistical data for system modeling. Authors noted
that uncertainties modeled in a stochastic cost-benefit analysis assist with ranking coastal
infrastructure projects, including resiliency enhancements. System risks increase in
response to stronger coastal forces with resultant damages. As uncertainty of the future
increases with climate change effects along coastlines, future adaptation plans become
more unreliable.
10
Consequence analysis assumes critical importance when both immediate and
future consequences depend on system vulnerabilities and critical infrastructure design
decisions. Reliability index methods provide some advantages over failure probability
methods in comparing alternatives, primarily because analysis does not require full
knowledge of probability density functions. While relative reliability methods and
stochastic cost-benefit analysis reduce uncertainty for comparing risk reduction options,
uncertainties remain related to the probability of system failures.
Wamsley et al. present a methodology for assessing coastal vulnerability metrics
using a comprehensive risk assessment and vulnerability model [9]. This model was used
by U.S. Army Engineer Research and Development Center (ERDC) and Institute for
Water Resources (IWR) to analyze the North Atlantic Coastal Comprehensive Study
(NAACS) area, to quantify the vulnerability of populations, infrastructure, and resources
at risk; and, to identify methods to improve system resilience to coastal storm damage
(see Figure 15 in Chapter III) [10].
The procedures and methodology described in the NAACS methodology and
linked tools are quite extensive and provide a way to systematically evaluate geospatially
risk and vulnerability associated with coastal infrastructure. An assessment for Long
Beach Island, NJ illustrates the suggested approach for evaluating potential impacts of
storms and sea level change and identification of coastal storm risk strategies.
The Federal Highways Administration (FHWA) has developed similar guidance
for coastal highways [11]. FHWA authorized numerous climate change resiliency pilot
projects and has published extensive documentation relative to the results of those
studies. The studies pilot-tested the FHWA Climate Change and Extreme Weather
11
Vulnerability Assessment Conceptual Model, and FHWA created a Vulnerability
Assessment Framework based on feedback from these pilot projects [12, 13].
An outcome of the assessment model involved South Coast Engineers (SCE)
evaluating the previously installed sheetpile cutoff wall and gabion countermeasures
constructed by the Florida Department of Transportation and Development (FDOT) to
mitigate overwash scour for the Barrier Island Roadway Overwashing from Sea Level
Rise and Storm Surge: US 98 on Okaloosa Island, Florida project [14]. For that event,
the overwash was inland and the damage was to the landward side of the road due to
weir-flow damage as overtopping occurred with flow into Choctawhatchee Bay from the
Gulf of Mexico during storm surge. The SCE study included developing an adaptation
decision matrix for engineering considerations and alternative evaluations.
1.5 Applying Systems Models to Coastal Roads
This research applied systems engineering to coastal roads in evaluating system
architecture and reliability functions. The coastal road systems interfaces are quite
complex since system boundaries extend well beyond the system of interest. Because of
the interdependence of mobility and modality in the functioning of coastal road systems,
determining when failure occurs requires defining when the system no longer can meet
the systems functional and operational requirements. Without preserving connectivity to
users and other transportation systems, even if the system is physically intact, the coastal
road system may not meet its functional requirements.
Local coastal roads represent the system of interest for assessing system
engineering functional requirements. The local coastal road designation includes those
12
roads maintained by municipalities, counties, or secondary state roads that primarily
serve local communities. It represents systems not intended for high-volume traffic and
likely more vulnerable to coastal hazards. CR 257 in Brazoria County along Follet’s
Island is one such example. System functional requirements were initially developed
using CORETM software [15].
Local coastal roads were modeled with ASTAH SysML software and proposed
the domain model shown in Figure 4 using model-based systems engineering (MBSE)
methodology [16]. Domain model captures major stakeholders and systems within the
local coastal road system of interest. Figure 5 illustrates local coastal road conceptual
system requirements and use cases developed for the object-oriented (OO) SysML
decomposition model. These figures illustrate the complexity associated with modeling
complex built-systems in a high-hazard coastal environment setting. Key subsystems
within the SysML model include:
• Road Structure;
• Road Corridor Enhancements;
• Modality Interconnects;
• Coastal Road Defenses; and,
• Travel Control/Safety.
Within that SysML model related to coastal hazard vulnerabilities are systems
involving road structure and coastal road defenses. Model assumes a representative level
of use cases for local coastal roads. Because of the complexity of the modeled system, the
level of definition developed for the model illustrates the challenge involved in applying
SysML to passive infrastructure systems.
13
Figure 4. Local coastal road system domain diagram [16].
14
Figure 5. Local coastal road conceptual system requirements and use cases [16].
15
Challenges remain as to how to define and utilize use cases and activity diagrams
for complex infrastructure systems with so many external system interfaces. External
interfaces like weather systems dynamically and continually change relative to local
coastal roads in magnitude, intensity, frequency, flow properties, values, and constraints.
Application of risk models using geospatial analysis represents one such methodology
attempting to more accurately assess and refine model-based systems engineering.
The systems model explored various components of the SysML model initiated at
a very high level and demonstrates feasibility of decomposing infrastructure systems into
subsystems models. SysML models are potentially viable for evaluating coastal road
systems if integrated and suitably refined to facilitate the simulation of vulnerabilities.
External interfaces and forces dynamically and continually change relative to local
coastal road storm hazards magnitude, intensity, frequency, flow properties, values, and
constraints. The previous studies were further expanded by applying soft systems
methodologies (SSM) and systems thinking tools to the resilience of coastal road systems
and system architecture, using previous object-oriented SysML modeling [17].
1.6 Framing Coastal Road Systems Resiliency
Failure analysis requires considering the resiliency and reliability of a system.
Coastal system resilience represents the integrated capacity and capability of a coastal
system to recover quickly to pre-event conditions when subjected to hazardous events
such as hurricanes, coastal storms, and flooding, rather than simply reacting to impacts.
Since passive infrastructure systems generally lack dynamic functionality to respond in
an adaptive capacity manner, this creates a problem with defining resilience for coastal
16
road systems. Adaptive capacity represents the ability of a system to adapt along with the
environment where the system exhibits changes as emergent behavior.
A system with highly adaptive capacity exerts complex adaptive emergent and
rapidly responding behavior in a changing environment. Correspondingly, a passive
infrastructure system with a low adaptive capacity responds with almost predictable
emergent behavior (usually resulting in failure) in response to changed environmental
forcings. A passive infrastructure system requires expanding system boundaries to
evaluate potential feedback loops and interaction with adjoining systems, including
external environmental weather or climate system forcings, in order to identify features
that can potentially improve system resilience.
Batouli & Mostafavi [18] proposed an interconnected logic block diagram
framework involving systems stressors, physical network, and user agencies as
representative of the dynamic mechanism involved with infrastructure transformation of
coastal road systems when exposed to sea-level rise hazards. The resultant complex
adaptive system (CAS) framework suggests that the adaptive physical network state
depends on actions directed by either the agency or stressors as subsystems, which
changes the system state as identified by state variables or state parameters. Evolution of
future states depends on predictive functions determined by current states. State is
dependent on combined physical and functional conditions of networked system assets
with dynamic mechanisms abstracted to describe resulting state transformations.
The CAS model’s strengths are that the computational framework captures and
simulates underlying dynamic mechanisms and complex interactions among climatic
stressors, physical networks, and human decision makers. The model’s weakness is that it
17
includes a significant number of assumptions in the functions that require extensive effort
to update and recalibrate; otherwise outputs rapidly lose validity as systems evolve.
Wainwright et al. suggest a modeling framework which integrates geological,
engineering, and economic approaches for assessing climate change driven economic risk
to coastal developments [19]. A joint probability analysis based on a storm erosion and
hazard model developed by Callaghan et al., was applied to shoreline transects [20-22].
The coastal risk systems analysis framework utilizes the most conventional methodology
in engineering analysis to define risk. These include:
• determination of the extent or severity of identified hazards for a range of
exceedance levels (i.e. the probabilities or likelihoods); and,
• determination of the potential losses for the extents and/or severities identified
(i.e., consequences).
Developing Bayesian network models to assess coastal systems and failure
probabilities is increasingly gaining prominence for coastal system failure models.
Coastal flood assessments increasingly propose networks of interlinked elements using a
Bayesian network model [23]. A Bayesian Network based on Gaussian copulas to
generate the equivalent of 500 years of daily stochastic boundary conditions for a coastal
watershed in Southeast Texas near Galveston probabilistically represented compound
floods caused by riverine and coastal interactions [24]. Davies et al. present a framework
for probabilistic modelling of non-stationary coastal storm event sequences using a novel
mixture of parametric percentile bootstrap and Bayesian techniques to quantify
uncertainties [25].
18
Application of Bayesian models typically requires modeling systems that can
apply stochastic analysis to large data sets using either artificial intelligence or machine
learning software. Machine learning models based on evidence-driven disaster risk
management (DRM) using many different data types, information sources, and types of
models are found to be effective for risk management analysis [26].
Community resilience planning requires methods to assess and measure a
community’s current resilience level and the risks and benefits of plans for its social and
economic institutions and physical systems [27]. NIST is making significant investments
in developing comprehensive risks and recovery models to improve community
resiliency models [28]. This includes funding for development of the IN-CORE Model.
The Python based IN-CORE model advances opportunity for machine learning by
additionally integrating physics-based functionality.
1.7 Developing Coastal Road Reliability Functions
Research applies systems engineering in advancing the development of risk and
reliability models for coastal transportation systems. Application of systems engineering
in developing stochastic functions that reduce uncertainties in predicting coastal hazard
damage and improve infrastructure systems reliability, advances understanding needed to
manage community transportation risks.
Research proposes a cumulative celerity function based in coastal theory that
supports developing coastal hazards fragility functions. The IN-CORE model allows
users to simulate hazards to optimize community disaster resilience planning and develop
post-disaster recovery strategies using these physics-based hazard simulations. The IN-
19
CORE model represents a highly detailed and effective transdisciplinary systems model
to evaluate risk. Research supports application of systems engineering to an engineered
coastal road system exposed to tropical cyclone or hurricane hazards.
Hazards represent a largely unpredictable and extreme wind and water storm
system evidenced by high speed winds, extensive coastal storm surge, and significant
water wave heights. Catastrophic damage often results as evidenced by Bolivar Island,
where only a few building structures remained standing after Hurricane Ike’s passage.
Research demonstrates that a systems approach to an engineered system identifies critical
hydrodynamics and mechanisms that cause complete or partial failure when subjected to
extreme climatological system hazard forcings.
By evaluating coastal modeling data that demonstrated strong correlation to
damage states as described in Chapter II with additional detail in Chapter VI, nearshore
hydrodynamics generate intensity measures (IMs) for cumulative celerity functions,
correlating probabilistic analysis to wave and storm surge mechanics and identifying
system variables that most significantly impact failure probabilities. Identifying these
critical system variables that largely determine failure, adaptation of the system
components are possible through systematic planning and design, which provides
opportunity to develop coastal road infrastructure that mitigates potential damage.
While research benefits from understanding a transportation system’s functional
requirements associated with resilience, the focus of this research is to develop fragility
functions based on modeled storm data that predicts the likelihood of system failure.
Failure evidenced by loss of system functionality potentially limits post-disaster recovery
and transportation system accessibility.
20
Reliable transportation systems are critical to maintaining community robustness
and recovery. Understanding which components are most likely to fail and why provides
opportunity to improve reliability of coastal road systems when subjected to extreme
natural hazard threats or incidents.
This risk is particularly amplified when the road network provides access along a
barrier island, with limited points of ingress and egress for evacuation and emergency
response. Applying systems engineering to assess risk and predict reliability of
engineered coastal systems has been an active and growing area of research for several
decades [29-36]. Applying system engineering theories to quantifying uncertainties
associated with multivariate probabilities of engineered systems with multiple or not-
mutually exclusive failure modes engages mathematicians, physicists, scientists, and
engineers alike in pursuit of developing models that resolve all uncertainties.
This potentially creates added risk with repetitive application or misapplication of
probabilistic methods or models that potentially limit identifying and resolving system
failures. Research utilizes means and methods appropriate to the system of interest in
applying systems engineering and analysis in a consistent manner. The goal is to develop
and present research findings that are not only scientifically defensible, but also broadly
interpretable, implementable, and intuitive.
21
CHAPTER II
ASSESSING COASTAL ROADS RELIABILITY
Author hypothesizes that the probability and likelihood of coastal road failure is
strongly correlated to cumulative overtopping water level, wave height, and wave period
evaluated using cumulative celerity dispersion (CCD) functions, relative to cumulative
water velocity kinetic head and distance from shoreline, and cumulative values
differentiate the likelihood and degree of relative damage states. Furthermore, since the
CCD function is founded in theory relative to nearshore coastal hydrodynamics as
evidenced by the pseudo-Froude function, it identifies adaptable metrics to improve
reliability and mitigate risks.
2.1 Failures, Froude, and Fragility Functions
Coastal roads are increasingly damaged due to storm surge, waves, erosion, and
sea level rise. Improving the resilience of coastal roads subjected to sea level rise and
extreme storm events reduces the likelihood of catastrophic damage to critical
transportation infrastructure [5, 12, 37-39]. FHWA provides technical guidance and
methods for assessing the vulnerability of coastal transportation facilities to extreme
events and climate change [11].
22
There is limited research investigating potential failure characteristics that
correlate other system components to the likelihood of coastal road failure [40, 41],
particularly when applied to disparate events and locations. The prevailing theory
attributes coastal road damage primarily to storm surge overwash, impacting waves, and
“weir-flow” scour as described in Chapter 8 of HEC-25 Vol. 1 [42] and Section 3.3.2 of
Vol. 2 [11] . Coastal weir-flow characteristically includes road pavement damage on the
landward side and not seaward side of the road. Overtopping weir-flow is considered the
critical road failure mechanism since the road pavement, functioning as a broad-crested
weir structure, creates scour that undermines the embankment section, which supports the
pavement structure [11, 14, 42-44].
For shallow water waves, the Froude number (Fr) is a dimensionless number
defined by the ratio of inertial forces quantified as water velocity, to gravitational forces
quantified as gravity wave celerity or phase speed as shown in Equation (1).
𝐹𝑟 = ∑
𝑉
√𝑔ℎ (1)
Channel flow propagating across a unit width of channel with Fr < 1 represents
subcritical flow conditions; Fr = 1, critical flow; and, Fr > 1, supercritical flow. The
Froude number defining transition between non-breaking and breaking waves typically
ranges between 1.3 and 1.6 [45, 46]. Clopper notes that weir-flow across highway
embankments creates supercritical flow zones over the embankment crest and on the
downstream slope of the embankment [43, 44].
23
Douglass and Krolak make similar observations of sandy coastal road
embankments that confirm similar flow characteristics during ebb flow conditions over
coastal road sections [42]. Douglass reports investigating the coastal weir-flow damage
mechanism at a prototype-scale in a FHWA-funded laboratory study conducted jointly by
the University of South Alabama (USA) and Texas A&M University (TAMU) in June
2005 [42].
The Galveston Test Bed Model task group generated fragility models for built
infrastructure and socioeconomic systems using hindcasted Hurricane Ike storm data
[47]. After pounding Galveston, TX with waves that overtopped the seawall on
September 12, 2008, Hurricane Ike made landfall over Galveston Island on September
13, with maximum sustained winds reaching 110 mph and extending outward nearly 120
miles from the eye [48]. Ike also produced a pronounced forerunner that generated
onshore storm surge and extensive flooding for nearly a day before impact [49, 50].
2.2 Coastal Storm Models for Damage Assessment
Extensive studies evaluated Hurricane Ike coastal waves, surge, and inundation in
the Gulf of Mexico2. These rigorous coastal storm numerical models evaluated
Hurricanes Rita and Ike that impacted the northwestern Gulf of Mexico in 2008 [50-54].
A dynamically coupled version of ADCIRC and SWAN (PADCSWAN v51.52.34) was
applied to compute Hurricane Ike coastal hazard incremental intensity measures (IM)
[47]. Coastal model output included coastal hazard IMs for storm surge elevations, water
velocity, and wave characteristics at each node location at hourly increments.
2 https://comt.ioos.us/projects/tropical_inundation
24
The ADCIRC model predicts the spatiotemporal distribution of water levels as a
result of all astronomical and meteorological forcing supplied, including resultant
contributions from wave action predicted by the SWAN model. These water level
predictions report on an hourly basis at every node in the model mesh. Reference terms
include water surface elevations (WSE), still water levels (SWLs), or sometimes even
storm surge elevations (SSEs).
In addition to the prediction of water levels at every mesh node, the ADCIRC
model also provides an estimate of the water velocity due to all forcings, including the
excess momentum due to wave action. Water velocity components (easting, northing / U,
V) report on an hourly basis at each node. The IM data report the maximum water
velocity in terms of a magnitude, as well as component-based form. Reported velocities
do not include orbital wave velocities.
The primary wave characteristics extracted from the model report on an hourly
basis at each node include the spectrally significant wave height, peak wave period, and
wave direction. The spectrally significant wave height is a property of the wave energy
density spectrum and is not necessarily equal to statistically significant wave height.
Wave direction data refers to direction from which waves propagate. A wave direction
value of 180 degrees means that waves are propagating south to north.
Hurricane Ike caused extensive damage to Brazoria County Road 257 (CR 257),
often referred to as the Bluewater Highway, which runs from Surfside to Galveston along
Follet’s Island, crossing over to Galveston Island at San Luis Pass. Figure 6 shows the
track of Hurricane Ike and inundation depths. Figure 7 shows Brazoria County and
Follet’s Island, located to the southwest of Galveston, TX and San Luis Pass.
25
Figure 6. Hurricane Ike Track and Inundation Depth (Courtesy of Harris County Flood
Control District v.01/03/2020)
Figure 7. Hurricane Ike Storm Surge Brazoria County (Courtesy of Harris County Flood
Control District v.01/03/2020)
26
Analysis evaluated WSE, water velocity, and wave data incremental IMs for
overtopping flows at sixty-seven (67) randomly spaced locations along CR 257. Model
data filtered out IM values when still water levels were below the road grade. We added
thirty (30) additional point locations as a test data set to validate the CCD functions
ability to identify the damaged sections. Locations in Figure 8 include undamaged to
completely destroyed road sections for which damage data were available [55, 56].
Figure 9 shows the damage characterized by Coast & Harbor Engineering [55]
based on extensive field damage assessments with supporting drawings and other
documentation. CSHORE and XBEACH models were used to evaluate Hurricane Ike and
Follet’s Island by evaluating multiple transects along Follet’s Island [57, 58]. Coastal
hydrodynamic and beach morphology models provided additional data [50, 52, 53].
Figure 8. CR 257 data point locations for modeling Hurricane Ike damage.
27
Figure 9. CR 257 damage assessment characterization post-Ike [55].
A Hurricane Katrina coastal model evaluated the CCD function along U.S.
Highway 90 (US 90), which parallels the Gulf Coast shoreline between Bay St. Louis and
Biloxi Bay. Hurricane Katrina damaged the coastal road when it made landfall on August
29, 2017 as a Category 3 storm. Hurricane Katrina produced higher storm surge with
shorter duration (8/29 – 8/30/2005) than Hurricane Ike. Analysis evaluated 88 data points
(Figure 10) relative to the CCD functions capability to identify the damaged areas.
Figure 10. US Highway 90 data analysis locations extending from Bay St. Louis to
Biloxi.
28
Hurricane Katrina created one of the most significant water surface elevations
ever recorded at 27 feet in Bay St. Louis [11]. A hindcast ADCIRC + SWAN simulation
of Hurricane Katrina (2005) provided data for the coastal hazards test bed study;
however, terrain data for the coarse grid mesh failed to accurately model elevations along
the road centerline. An XBEACH model utilized 2004 LiDAR terrain elevations with 4-
meter grid spacing to model the US 90 corridor to refine WSE and velocity data.
ADCIRC + SWAN model provided spatiotemporal forcings for XBEACH model
simulations. ADCIRC + SWAN model also provided better wave data at node locations.
2.3 Framing the Failure Analysis Research Objectives
Research examines coastal road systems subjected to storm surge and wave
hazards during extreme events. Each of the data points along CR 257 in Texas and along
US 90 in Mississippi effectively provides an independent data set for testing and
validating the proposed CCD function. Research examined coastal roadway system
failures by exploring damage mechanisms and empirical models using physics-based
relationships. Initial analysis broadly examined correlations between various modeled
system component granularly defined binary damage states based on record data, relative
to incremental IM data extracted from coastal storm models.
The purpose is to identify correlation between coastal road damage states to
modeled or physical parameters. Modeled data are not independent of physical data, since
physical data generate and calibrate numerical coastal models that simulate measured
environmental response data for a significant storm event. Modeling introduces
relationships between random variables as artifacts of algorithms used to generate data;
29
however, these relationships are verifiable utilizing broadly accepted numerical models
and validated using measured data. Hurricane Ike is an ideal candidate for evaluation
since it represents one of the most intricately studied hurricane events along the Gulf
Coast based on the publications reviewed.
Research identifies the engineered system of interest (nearshore coastal road),
identifies forcings (hurricane coastal storm modeled data), and proposes the underlying
relationship for developing fragility functions. While all data are not yet available, with
more physical modeling data needed to truly develop defensible fragility functions, the
CCD functions represent the engineered system’s unique capability at a specific location
based on the specific environmental setting (elevation and offset distance from shoreline),
to resist failure up to an approximate limit state when subjected to storm forces. While
CCD functions represent complex nearshore wave mechanics, explaining these apparent
relationships remains a challenge and provides opportunity for additional research.
Engineered system of interest involves regional road transportation infrastructure,
namely coastal road systems located near shorelines. Research originally considered
evaluating fragility of coastal bridge structures subjected to extreme waves, but
involvement in the Galveston Test Bed project redirected research focus towards coastal
road systems. We are currently supporting research at Rice University in the development
of fragility functions that utilize peak IM data to assess road fragility for coding in the
IN-CORE model. Dr. Jamie Padgett, a subject matter expert in fragility functions, has led
this effort with her post-doctoral graduate students. Since the CR 257 coastal road system
provides the greatest amount of records relative to damage extents and locations, it
provides needed data to correlate modeled storm data to damage states.
30
2.4 Assessing Failure Mode and Effects
CR 257 on Follet’s Island located between Christmas Bay and the Gulf of Mexico
in Brazoria County Texas experienced the most significant damage to a linear road
system along a 20-kilometer section during Hurricane Ike in early September, 2008. The
damage was not continuous, and included a range of damage states, so the road system
provided opportunity to verify a failure function to recorded damage states.
The typical road system assessed by the IN-CORE model consists of an
engineered coastal road of varying horizontal and vertical alignment relative to the
shoreline, asphaltic or rigid pavement surfacing, base material, compacted subbase soils,
ditching, and subsurface drainage pipes. Engineered coastal defenses may include
seawalls, floodwalls, barriers, levees, shoreline armoring, engineered dunes, surge
barriers, groins, and breakwaters. Natural coastal defenses typically include beach, living
shorelines, native dunes, mangroves, marshes, reefs, or natural rock armoring.
CR 257 consists of an elevated asphalt pavement road section setback at varying
distances from the shoreline with a gently sloping beach and dunes of varying elevation.
Beach increases in steepness to the southwest near Surfside where breakers occur more
frequently due to the steeper nearshore slope as the name implies. A section of shoreline
in Surfside protects the local roads with rock armoring [55]. Assessment of failure mode
and effects also included Hurricane Katrina damage to US 90 in Harrison County, MS.
Coastal defenses include a seawall and concrete barriers so fragilities vary.
Road functionality depends on the transportation route remaining accessible for
traffic. After CR 257 failed, temporary earthen road bypasses provided temporary access
for emergency response and recovery equipment.
31
To provide perspective of the road system conditions, pre-Ike conditions were
described as follows for CR 257 [59]:
CR 257 serves as a direct link between Galveston Island and south Brazoria
County, including the City of Surfside, City of Freeport, City of Angleton, and various
communities on Follett’s Island. CR257 was constructed as a paved two-lane roadway in
the 1950’s. Over time, various layers of base material and hot-mix asphalt pavement
were added to raise the overall profile grade of the roadway to its current elevation.
Before Hurricane Ike, CR 257 consisted of two 12’ travel lanes with 2’ shoulders for a
total width of 28’. The roadway was essentially the high point of the island with drainage
sheet flowing off the right-of-way either to the beach side or to the bay side of the road.
Except in limited cases, no roadside ditches were present to convey lateral drainage. In
addition to CR 257, Brazoria County owned and maintained several paved beach access
roads which provided access directly from CR 257 to the beach. These roads consisted of
hot-mix asphaltic pavement on approximately 12” of base material.
The same report describes post-Ike conditions as follows:
CR 257 sustained catastrophic damage resulting from the tidal surge associated
with the approach of Hurricane Ike. The damage ranged from partial failure of the edge
of pavement to the complete obliteration of the pavement structure and embankment
within the right-of-way. It appears that much of the damage occurred when the tidal
surge began to recess and the water flow accelerated toward the gulf side of the island.
Much of the pavement material deposited on the beach side of the roadway. The total
length of roadway that received partial damage was approximately 18,480 feet. The total
length of roadway that was completely damaged was approximately 12,140 feet.
32
Photographic damage records indicate that failed sections include locations with
multiple layers of pavement and base material as shown in Figure 11. The base material
appears to be an original Texas shell/sand mix base road described by Doran in 1965 [60]
with multiple courses of asphaltic concrete surfacing as shown in Figure 12. These
components of the road pavement and base structures are potentially key determinants in
determining the sensitivity of fragility functions. The lack of experimental data in this
research effort limits developing fragility functions in considering a functional
decomposition of subsystem fragilities, but provides additional opportunity for continued
research and function refinement by others. Figure 13 shows the most heavily damaged
section of road based on the CCD values at Point 4.08 looking east.
Figure 11. Typical failed pavement section on CR 257 (Brazoria County, 9/15/08).
33
Figure 12. Typical failed pavement base material on CR 257 (Brazoria County, 9/15/08).
Figure 13. Failed pavement road section in the most heavily damaged section of CR 257
(Brazoria County, 9/15/08).
34
Initial assessment assumed that likely damage mechanisms were attributable to
weir overtopping and toe scour; rapid drawdown affecting slope stability; saturated soil
seepage undermining pavement section; rigid/flexible pavement structural resiliency;
dunes failing to reduce wave energy; increased sediment transport and flux; and, extreme
beach scouring in response to storm events. While each of these subsystem components
are likely to affect resiliency of the coastal road in determining degree of damage, these
component reliability functions appear secondary to energy dispersion and sediment
transport as causal factors based on CCD functions.
Damage assessment treated road damage as a binary damage state with the failed
state including partial or complete damage to the road such that continued use results in
an unacceptable risk to the user or vehicle due to the severity of damage. Partial scour at
the edge of the road that did not undermine the pavement section or compromise the
system functional use ranked as no-damage. Future laboratory tests will assist with
quantifying forcings necessary to verify likely failure mechanisms.
2.5 Assessing System Reliability and Likelihood of Failure
Defining system boundaries for built infrastructure in nearshore coastal settings is
challenging. They sometimes lack definition when identifying system boundaries
between infrastructure, user, and environmental interfaces. These increase the complexity
of system decomposition and subsystem risk and reliability analysis. Improvements to
increase resiliency, such as protective dunes, often impact other systems and are resisted
by a local community due to perceived negative impacts. System boundary for the coastal
storm model [47] includes the Gulf of Mexico with a refined mesh for the Galveston area.
35
System boundary evaluated for CR 257 along Follet’s Island extends
approximately 20-kilometers from the southwest end of Galveston Island at San Luis
Pass to Surfside Beach. System boundary for US 90 extends approximately 40-kilometers
from Bay St. Louis at Pass Christian to near Interstate Highway 110 interchange in Biloxi
Mississippi. CR 257 does overwash at some locations as storm surge flows into the back
bays. The high ground along US 90 from Bay St. Louis to Biloxi Bay does not overwash.
Reliability is the probability of a structure or system performing its required
function adequately for a specified period under stated conditions. While reliability
analysis finds wide acceptance in structural and industrial engineering, there are fewer
applications of fragility functions in coastal infrastructure system design; however, these
types of models are rapidly gaining acceptance for complex systems of interest,
particularly for coastal bridge and breakwater designs. Many coastal functions rely on
probability theory or stochastic methodologies in assessing risks and uncertainties
associated with extreme events and natural hazards.
Coastal road failures demonstrate that extended duration of extreme event likely
increases the probability of failure. The CCD function developed in this research as
detailed in Chapter VI strongly predicts damage likelihood. It enables development of
coastal road system fragility functions describing system capacity to resist failure when
impacted by extreme tropical storm system forcings of varying intensity and duration.
Each road node location provides an independent data set within the models. The coastal
hazard intensity measures generated by the coastal models are unique to each
geographical location and the road characteristics are unique at each location.
36
Coastal storm models, including ADCIRC + SWAN and XBEACH, provide
incremental discrete surge still water elevation, wave height, wave period, current
velocity, and current vector output data. The CCD functions are computed at each time
step using incremental IM data and then summed over the storm duration to generate a
cumulative celerity value with units of meters per second. The cumulative velocity head
(V2/2g) is divided by the road offset distance from the shoreline to create a dimensionless
kinetic energy gradient factor.
Filtering excludes IM data at each hourly time step if wave height at the node
location is less than 0.3 meters or if the road is not overtopped. The cumulative sum
provides a representative empirical value for cumulative shallow water celerity or phase
speed modified by the cumulative kinetic energy gradient factor. CCD functions do not
represent continuous integrations of functions, but do assist with developing fragility
functions that predict likelihood of damage for future events.
Coastal model data supports time-dependent reliability evaluation since road
failure strongly correlates to duration of extreme event overtopping. A regression
equation model predicts road failure or risk of failure when CCD exceeds a critical
threshold range. The composite function includes terms for cumulative celerity and
current velocity head relative to offset distance of road from shoreline. CCD data for CR
257 indicate threshold values indicating damage states are exceeded shortly after velocity
vectors redirect water velocities in a southwesterly vector direction, indicating ebb flow
conditions as storm surge abates and winds redirect after the hurricane moves onshore.
Failure mechanism appears to be partially attributable to weir overflow during
storm surge ebb flow, but overwash during surge flood flow does not typically produce
37
corresponding failures landward of the road slab. Review of cumulative CCD plots also
show that the failed limit state typically occurs when depth-limited irregular sea-state-
dependent wave breaking combined with ebb flow conditions exist at the road section.
One potential failure mechanism may result from breaking waves collapsing air pockets,
which in turn generate transient pressure forces that transmit into saturated subsoils.
A significant failure mechanism appears to depend on the interaction of strong
ebb currents backflowing over the slab at critical depth, interacting with the irregular sea
state with depth-limited breaking plunging waves. These forces create significant
dynamic loads on the pavement, base material, and underlying embankment. Flood flow
with storm surge moving onshore elongates wave lengths. Ebb currents conversely
reduce wave lengths and steepen wave breaking with opposing wave-current interactions.
The time variant reliability assessment involves integration of complex functions
of correlated stochastic random variables with respect to sea states, forces, and
geostructural road system limit states. The IM values represent incremental output data
from complex numerical models simulating complex extreme events. Since the CCD
functions aggregate incremental hourly IM data output at specific geospatial coordinates
for the duration of the storm, research identified significance of time-dependent reliability
for the engineered system. Coastal defenses like seawalls and other structural barriers
appear to temporally reduce the likelihood or degree of failure, which requires
consideration when developing fragility functions to assess system response to extreme
storm events.
Applying research to reliability models benefits from resolving stochastic
distributions of time-dependent variables. Hazard systems models, CCD functions, and
38
linear equations derived to date suggest that a semi-empirical tool is potentially feasible,
even though the system is spatiotemporally varied. The CCD functions are partially
deterministic using coastal storm models and application of incremental IM data, but
empirical in assessing the likelihood and relative degree of damage. Research sought to
resolve the extent with which stochastic models are applicable to other locations in
predicting the likelihood of damage by advancing a pseudo-Froude function.
The CCD function requires numerical coastal hydrodynamic models to generate
the incremental IM extreme storm hazard data. All models used for this research assumed
fixed bed bathymetry and surface elevation data. Since research focuses on reliability of
coastal road system, research utilized storm events that have been extensively modeled
by others [47]. Generating ensemble averaged quantities requires multiple realizations of
data sequences with correct probability distribution and correlations.
IM data relies extensively on ensemble averaged data incorporated into numerical
coastal storm models that predict the spatiotemporal distribution of water levels as a
result of modeled astronomical and meteorological forcings, including resultant energy
contributions from wave action. Research also evaluated extreme event load data for
storms with annual exceedance probabilities (AEP) of approximately 50, 100, and 200-
year storms for the CR 257 coastal road system; however, corresponding damage or load
response data is not available for these events.
Previous research reports that the proximity of the road to the shoreline provides a
strong indicator of failure likelihood [11, 40, 61, 62]. Likelihood of failure for the CCD
function does not exclusively depend on distance of road from shoreline. Application of
the CCD function to Hurricane Katrina and US 90 suggests that the likelihood and type
39
of damage for the short duration storm events are more dependent on the cumulative
kinetic energy than the cumulative wave celerity terms in the CCD function, which is
why the CCD function having its basis in the inertia to gravity forces ratio described by
Froude is important. The level of damage and type of failures observed were
significantly less for US 90 than for CR 257.
Roads positioned closer to the shoreline are typically more likely to experience
extreme forcings represented by incremental IM values from hydrodynamic model output
data. Similar detailed damage data as was available for CR 257 for previous road failures
when subjected to extreme hurricane storm surge and wave events are not readily
available for additional model validation. Hurricane Michael is likely to provide
comparable data to Hurricane Ike for evaluating CCD functions with respect to
documented road damage in future research by others.
The coastal road system represents a composite engineered structure with
multiple-failure mechanisms. Coastal roads often integrate into a coastal levee system
with elevated embankment that creates a storm surge barrier or with other types of coastal
defenses. System components experience varying degrees of damage in response to
forcings of various intensities where partial damage does not necessarily result in
catastrophic failure or loss of system function.
Research suggests that the strong correlation of CCD, computed using either
cumulative IM water surface elevations or wave periods, would likely qualify wave
energy dispersion in shallow water as the top event of fault tree analysis. The CCD
functions provide a methodology for predicting the likelihood of failure that reduces but
does not eliminate the significance of other probabilities. The difficulty of utilizing fault
40
tree analysis involves assessing system failure due to multiple coastal structures with not-
mutually exclusive failure modes in a natural environmental setting. Future research will
determine whether CCD functions can evolve into a defensible time-dependent system
reliability model using Bayesian methods to evolve fragility functions for use with risk
and reliability models using machine learning and artificial intelligence programming.
2.6 Considering CCD Function Uncertainties
Analysis of errors for coastal hazard intensity measures determined by the coastal
storm model were summarized in a report for the research associated with the Galveston
Test Bed Model [47]. Hurricane Ike coastal storm simulation used the dynamically
coupled version of ADCIRC and SWAN. Epistemic uncertainties are minimal due to the
advancement of numerical coastal modeling and increased computing power. The
potential for uncertainties that introduce errors are extensive, but the error range is small
when compared to observed data. These potential uncertainties include the following:
• Meteorological forcings
• Model forcings
• Model mesh
• Model configuration
• Boundary conditions
Validation of ADCIRC and SWAN models utilized a traditional model-data
comparison by evaluating their hindcast predictions of water levels and wave
characteristics against measured and observed data during Hurricane Ike. Simulation
involved no additional calibration or validation of the models. By validating to measured
41
data, the validation assessment provides an opportunity to quantify the potential model
errors attributed to the storm surge and wave hazards.
The modeled data was compared to measured data and the differences quantified
as root-mean-square error (RMSE) and as a percentage (error) of the incremental value as
described in Chapter VI. Wave characteristics predicted by the model are the spectrally
significant wave height (Hmo) and the incremental wave period (Tp). This value is
different than the statistically significant wave height (Hs). While the two measures are
approximately the same in deep water, they diverge in shallow water by a maximum ratio
of about 1.7, with Hs being greater.
The coastal hydrodynamic model reproduced about 90% of the measured values
with a tendency to overestimate wave height magnitudes by about 11%. Like wave height
and water level predictions, the coupled coastal models tend to overestimate wave period.
These sources of potential error in the modeling process identify the difficulty in
estimating uncertainties when simulating real-world problems. There are increased
uncertainties in progressing any type of predictive model.
Since the CCD function effectively integrates the incremental hourly IM values
over the event duration, the variables that influence the predictive outcome for likely
damage mechanisms are key. The correlation between celerity computed using either
wave period or water surface elevation (measured relative to zero datum) significantly
reduces uncertainty in the model by providing two different functions with numerically
quasi-independent variables validated by observed events, so these values effectively
validate each other.
42
There is some uncertainty with the measured distance to the shoreline. As noted in
the post-event report by Coast & Harbor Engineering [56], the distance to the MHHW
line relative to the road centerline changed during and after the event [57, 58, 63]. The
distance scaled from aerial images for the CCD function is approximate and not highly
refined. Shoreline offsets from MLW migrate during storm events. Beach scour changes
the MHHW during the event, so there is a temporal and spatial variation associated with
changing bathymetry and beach profiles due to scour and sediment transport. Fixed
geospatial elevations are not evaluated in the coastal hydrodynamic models or reflected in
output incremental IM data in this research.
There is some uncertainty with the input mesh surface model elevations used to
with the coastal hydrodynamic models. The model resolution does not reflect a high-
degree of accuracy relative to the minor variations of road alignment, profile, and width.
The model does not account for increased current outflow as significant scour occurs
forming backflow channels, wave scour features, overwash deposits, and dune breaches
[64, 65]. The roadway pavement and embankment section often degrade by meters over
the course of the event, once again introducing uncertainty. Since the WSE used by the
CCD function is measured relative to datum and not road grade, the uncertainty is
believed to be small but cannot be further evaluated without physical modeling data.
While the model is generating output for the incremental hourly IM values at a
specific coordinate, it is not considering the variation in that incremental IM value at each
computational time step. The CCD function represents a discrete summation of hourly
data. There are also uncertainties and sensitivities associated with the design features of
the road and risk reduction features (dunes, armoring, seawalls, barriers, curbs, etc.).
43
Incorporating uncertainty requires evaluating the range of results relative to the
damage observed and using a probabilistic function like those generated for the other
Galveston Test Bed Model features that includes uncertainties in those parameterizations.
Some have used coupling of models evaluated over a wide range of conditions to
quantify both aleatoric and epistemic-type uncertainties; however, there are flaws with
that approach with limited site-specific data [66].
Reliability theory can assess performance of existing structures and quantify
uncertainties associated with new ones by using either the probability of failure or
probability of survival of the structure. In flood risk assessments, the approach is
typically to assess the probability of failure by defining a reliability function. Since there
is uncertainty in strength and/or load variables, random variables characterize these
uncertainties using probability distribution functions. Probability of failure computes
using a joint probability density function for resistance and load [67]. These functions
effectively provide the basis of developing fragility curves.
Uncertainty is also associated with the frequency of occurrence. Analysis showed
Hurricane Ike generated variable impact recurrence intensities along the length of Follet’s
Island. The incremental IMs vary by location based on the height of roadway and the
storm magnitude relative to surge, waves, and overtopping currents. Salvadori et al. [68]
examined the traditional univariate approach of estimating probability of structural
failures in terms of Return Period and Design Quantile. The study highlights the
weakness of traditional approaches which assumes design variables are independent,
neglecting the dependence of variables and potentially underestimating risks when using
probabilistic methods for design of coastal or offshore structures.
44
Fundamental differences exist between univariate and multivariate framework,
with multiple design value pairings possible with the latter. Isolines for probability
functions at defined return periods range from a large significant wave height with a
shorter storm duration to a smaller significant wave height with a longer storm duration.
Salvadori et al. recommend alternate most-likely (likelihood based) and wave height
conditional (failure criticality based) design realizations to evaluate and compare risks.
Determining multivariate design pairs based on significant differences resulting
from these alternate design realizations highlights critical underlying issues in risk-based
design. Multivariate analysis inherent in the CCD function indicates that both short- and
long-duration events can result in the same resultant CCD values with equivalent
likelihood and degree of damage.
While this approach holds potential promise in applying the CCD function for
variable storm frequencies, intent for this effort is to resolve uncertainties by validating
the model results using different events and locations. Risk, uncertainty, and reliability
analysis guidance provided by USACE Water Resources Support Center Institute for
Water Resources (IWR) Engineer Research and Development Center (ERDC) support
developing fragility curves [69]. Misrepresenting the CCD function accuracy due to
aleatoric uncertainty at an early state would be problematic.
Identifying the sensitivities of each variable is also a likely outcome of future
research efforts. Tornado or swing diagrams have recently been used in seismic responses
to assess the sensitivities of different random variables on one or more response
parameters [70]. This technique holds potential promise in resolving which uncertainties
are the most relevant in future research.
45
2.7 Considering Cause and Effect
Data points were located on CR 257 plan-profile sheets to determine damage
extents post-Ike. Damage states assignments utilized a 5-point rating scale as follows:
• 0 –No Damage
• 1 – No Road Damage - Beach Erosion near Embankment
• 2 – Edge of Road Scour Seaward - Pavement Damage < Half-Width
• 3 – Road Damage - Full Width
• 4 – Road and Embankment Washout
For preliminary analysis, damage states 0–1 represent collectively ‘no road
damage’; whereas, damage states 2–4 represent collectively ‘road damage’. Data include
Hurricane Ike hydrodynamic model IM data from September 11–15, 2008. Fragility
based on peak IMs describe the failure probability of a coastal road during a coastal
storm event based on wave height, inundation duration, and distance to shoreline data.
Dr. Ioannis Gidaris developed the failure probability function shown in Figure 24
in Chapter VI while evaluating the Galveston Test Bed Model in coordination with the
research team. The plot illustrates average values of spectrally significant wave height
with respect to other variables. Publication of these and other findings in coordination
with the Rice University research team is pending.
While larger wave heights are more likely to contribute to increasing damage with
vertical wave impacts on buildings or dune/beach escarpments, it does not appear to be as
significant a contributor to increasing damage for horizontal structures such as roadway
46
pavement surfaces. The contribution appears to become more significant as the distance
from road to shoreline decreases.
Figure 14 shows the correlation between the failed limit state spectrally
significant wave height to depth ratio. Data represent average values of all 67 base data
points. Data show significant correlations when damage likely occurs during ebbing
storm surge conditions with the change in flow direction evident.
Figure 14. Average wave height to stillwater depth ratio and water velocity direction
changes at the threshold failed limit state for the CCD function for Hurricane Ike storm
event and CR 257 damage.
47
Distance to the shoreline is a critical variable based on observed performance and
data analysis. The cumulative velocity head for the cross-shore current relative to the
offset distance of road to shoreline effectively provides a driving energy gradient
transporting the sediment away from the road and dunes. Distance is not a variable with
impacts independent of the velocity head based on extensive data analysis.
Wave celerity provides a cumulative measure of energy dispersed through
shallow-water wave breaking. The velocity head gradient relative to offset distance
creates an energy gradient for the current advection component. Inundation duration is
extremely critical since a short event will not likely produce the aggregate energy
dispersion necessary to produce critical failures in the road section.
By grouping data into two states, damage and no damage, comparisons between
average values assess which random variables were most likely to distinguish the two
states. Two correlations were eventually determined to have potential relevance. As noted
in the CR 257 damage report, location of the road within 145 meters of the shoreline
created a strong likelihood for road damage [56].
The post-event assessment report noted that damage to primary dunes, secondary
dunes, and road was dependent on the distance from the mean higher high water
(MHHW) line. Major damage occurred within ±15 meters of an average offset of 137
meters from the Post-Ike MHHW line. Landward of that distance, little to no damage
occurred. Similar probability of failure results when location of the road is within 75
meters of the marsh edge, generally defined by ground elevations less than 0.5 meter.
A strong relationship does not appear to exist relative to degree of damage and
likelihood when correlating dune height relative to shoreline road offset distance
48
normalized at 145 meters. At low dune heights, minor damage occurs at greater distances
from the shoreline as expected without the wave mitigation feature this physical barrier
provides. Initial evaluation did not consider time to failure since data were not available
to assess progressive or even cascading failures. Aerial imagery suggests increased
likelihood of washouts at existing or abandoned avulsions.
Other variables that likely change the threshold values to reduce the likelihood of
failure include risk reduction features such as improved dune sections, breakwaters,
seawalls, hardened structural pavement sections, and rock armoring, like post-event
measures implemented by Brazoria County when repairing CR 257. These measures
effectively dampen the response from wave celerity by allowing energy to disperse
without causing damage. Living shorelines, beach nourishments, marshes, mangroves,
and other natural and nature based features (NNBF) systems provide alternative
considerations in assessing whether these can also be used to mitigate damage [39, 71].
An alternate consideration is evaluation of the reliability of a road location
relative to potential range of external conditions that are likely to produce failure. Rather
than assessing the likelihood of failure for a design event on a section of road, an
alternate approach is to determine the range of external stressors that will produce failure,
then assign the likelihood of failure to a corresponding probability of occurrence in
assessing the limits for feasible coastal road alignment locations and profile elevations.
49
CHAPTER III
LOCAL COASTAL ROADS – NEXT GENERATION3,4
Coastal road systems provide a vital transportation infrastructure asset within
local coastal communities that are being challenged relative to resiliency and
sustainability. Increasing demand on coastal infrastructure with increased development
densities in coastal areas requires providing coastal road systems that meet changing
needs. This study considers the growing disparity between uniformity in design standards
with conflicting desirements of local communities. System engineers should approach
this challenge from a divergent perspective. There is growing difference of opinion with
“next generation” desirements. Next generations question whether development should
continue in high risk coastal areas while choosing to live there. Physical and economic
damages impacting coastal roads require reimagining fundamental system requirements
for coastal infrastructure.
Voices from a diverse group of stakeholders are considered in applying system
models to Local Coastal Roads. This analysis led to different desirements and priorities
than provided in current unified technical guidance. Next generation priorities of framing
systems analysis are changing system requirements. Providing systems analysis tools for
3 Paper received 2018 IISE Conference Sustainable Development Best Track award. 4 G. P. Pennison, R. J. Cloutier, and B. M. Webb, "Local Coastal Roads-Next Generation," in 2018
Institute of Industrial and Systems Engineers Annual Conference and Expo, IISE 2018, Orlando, FL,
2018: Institute of Industrial and Systems Engineers.
50
local communities to reimagine connectivity and environments associated with local
coastal road systems increases opportunity to meet desirements of the next generation.
Changing conditions make systems analysis, planning, siting, and architecture design
essential and beneficial for continued resiliency and sustainability.
3.1 Introduction
Coastal road systems play a strategic role for continued development within the
United States. According to the National Oceanic and Atmospheric Administration
(NOAA) National Ocean Service (NOS), counties directly on shorelines account for 39
percent of total population occupying less than 10 percent of total land area (not
including Alaska) [72]. Coastal counties and parishes population increased by almost 40
perent from 1970 to 2010. Population is projected to increase an additional 8 percent by
2020. Coastal population density is many times greater than corresponding inland
counties and increasing population in at-risk areas at significantly faster rates.
Increased coastal development places greater numbers of engineered structures at
risk for damage from coastal hazards. This includes coastal infrastructure associated with
local roads, as well as state and federal highway systems. Coastal hazards (e.g., storm
surge, hurricane force winds and waves, tidal and riverine flooding) place both
population and coastal infrastructure at risk. Risk is greatest when measures are not taken
to mitigate these natural hazard risks.
One significantly increasing hazard is the result of relative sea level rise (RSLR).
Storm events are also increasing coastal hazards and frequency of coastal road flooding
and damage. RSLR combines risk of local subsidence, glacial melt, gravitational
51
redistribution, and sea level rise. When combined with increased atmospheric energy, the
result has been increased levels of damage resulting from higher tides, greater intensity
storms, natural wetland buffer loss, and beach erosion. Even in the absence of RSLR or
other climate change impacts, risk in the coastal environment is exponentially increasing
because of increased property development. The predictable response is to protect
developed areas and coastal infrastructure at great expense without regard to robustness
and sustainability.
Local communities and governmental transportation agencies struggle with policy
decisions regarding changing requirements for planning, designing, constructing,
maintaining, and operating coastal roads. Risk requirements vary locally, as do
community desirements and levels of service provided by local coastal roads.
Desirements is a term meant to convey stakeholders needs and desires, usually captured
in client interviews, public hearings, and "voice of the customer" needs assessment, and
often evolve to become system or functional requirements as the project matures.
Coastal roads and highways are an invaluable infrastructure asset for coastal
communities in providing:
• Property access for private, business, commercial and governmental entities.
• Shoreline access for marine related industries and recreational activities.
• Critical egress for hurricane and nor’easter life-safety evacuation.
• Emergency response, property protection, and critical services restoration access
post-event.
• Access corridor for utilities, pedestrians, cyclists, multi-modal transport.
• Connectivity to public spaces, shoreline features, aesthetic vistas, and marinas.
52
Coastal communities and coastal road infrastructure are challenged relative to
resiliency and sustainability with changing climate conditions and increased hazard risks.
Increased demand on infrastructure due to increasing development in coastal areas
requires providing coastal road systems that meet changing needs. Coastal system
engineers, planners, environmental and social scientists are challenged to provide systems
that meet changing desirements and functional requirements. Regulatory guidance is
increasingly available with technical publications for planning, siting and designing
coastal infrastructure and road systems. The United States Army Corps of Engineers
(USACE) North Atlantic Coast Comprehensive Study (NACCS) Resilient Adaptation to
Increasing Risk final report is one such example as shown in Figure 15 [73], which can
be applied to transportation or other system types.
Figure 15. NACCS coastal storm risk management framework.
53
3.2 Stakeholder Desirements5
Private and community property developments, including required infrastructure
support systems, increasingly require stringent assessment relative to potential impacts on
community socioeconomics and natural environments relative to sustainability. There is a
growing disparity between regulatory requirements for project development and local
community perspectives relative to how road projects should integrate and evolve as
coastal risks increase. The National Environmental Policy Act (NEPA), signed into law
on January 1, 1970, requires federal agencies to evaluate environmental, social, and
economic effects of proposed actions [74]. Because of inherent coastal risks, Federal
Highway Administration (FHWA) developed Hydraulic Engineering Circular No. 25,
Volumes 1 and 2, regarding planning and design of highways in the coastal environment
[11, 75]. These manuals provide comprehensive design guidance from a coastal and
transportation planning/engineering perspective.
System adaptability challenges from an analytical perspective include managing
the following risks [76].
• Transportation system change occurs slowly (typically discontinuous and
reactive) and coastal ecosystem change (typically continuous and responsive to
some stressor). Transportation systems may not adapt at rates necessary to keep
up with increased sea levels and storminess.
• Identifying infrastructure that is exposed (now or in the future) and vulnerable to
both RSLR and increased storminess is complicated and a potentially expensive
process for local and state transportation agencies.
5 A coined word to differentiate stakeholder wants and needs, from system requirements.
54
• Physical structures are vulnerable to RSLR, which is likely to result in increased
costs for maintenance, repair, replacement of facilities and materials, and eventual
adaptation.
• Function of linked, regional transportation systems may be vulnerable to
disruption if a RSLR-vulnerable link (e.g., a coastal highway) fails.
• Infrastructure and living system adaptations will need to occur to avoid a
wholesale change in the marshes, estuarine systems, low-lying urban areas, and
exposed highway infrastructure along the U.S. coast. (In many cases
transportation infrastructure, because of its static nature, is an impediment to
ecosystem resilience, e.g., barrier island rollover during overwash, marsh
migration with SLR, etc.).
In August, 2014, the City of Boston and partners announced a “Designing with
Water” juried competition to develop competing ideas for creative and innovative
climate-change resilient design solutions for three at-risk waterfront sites in Boston[77,
78]. One of those at-risk sites is Morrissey Boulevard and Columbia Point in the
Dorchester neighborhood of south Boston. The University of Massachusetts (UMass) is
located on Columbia Point. Morrissey Boulevard completed in 1924 provides access to
the UMass campus and is owned and maintained by the Massachusetts Department of
Conservation and Recreation (DCR). It is increasingly flooded during storms and high
tides.
Sea level in Boston has risen by a foot over the last century. It is projected to rise
another two to six feet by the end of this century. As sea levels rise, and chronic flooding
becomes the “new normal,” cities are moving to more flexible, resilient solutions. The
55
competition represented Boston’s effort to explore innovative solutions, allowing defined
areas to flood or contain water to prevent damage to other inland areas. These “next
generation” ideas expand and potentially challenge concepts of how a coastal road should
be designed in accordance with federal guidance.
These stakeholder desirements are not unique to Boston and vary based on local
communities and cultures. While roads provide facilities for basic transport functions,
desirements are changing from utilitarian functions determined by engineers, to include
resiliency and sustainability components driven by local community aesthetic, cultural,
and socioeconomic needs. The Living with Water competition teams reimagined
requirements and needs for changing conditions in the Morrissey Boulevard area at
Columbia Point in response to local stakeholder concerns, desirements and needs.
Presentations included an extensive narrative with graphical depictions of concepts.
These presentations represent a broad range of concepts for reimagining role and function
of coastal roads in local communities.
DCR subsequently issued a Request for Proposals (RFP) for Morrissey Boulevard
soliciting consultants to develop plans and complete engineering design for
improvements necessary to improve road resiliency for the next 50 years. These
requirements represent the voice of the stakeholder from the perspective of the owner and
managing agency. While the RFP allows new ideas to be included in proposals, overall
desirements are more utilitarian and practical than those generated in the competition.
This disparity is increasing in coastal transportation system planning. Significant
differences are apparent between what is feasible for implementing without much public
56
resistance, and what is imagined for future needs, based on changing environmental
conditions and diverse socioeconomic settings.
Because desirements varied, concepts were prioritized to assess which were most
important for the competing teams in envisioning system requirements to satisfy
stakeholder desirements. Using the top ten prioritized requirements, a Pugh decision
matrix was used to evaluate concepts that generically represent different functional
approaches as to “how” these objectives can be achieved. Providing flood risk reduction
for coastal flooding is perceived as a mandatory requirement to be incorporated in the
system architecture planning without being specified.
The priority desirement is for blue-green parkway corridors, referring to
combined blue water and green open space features in a linear parkway corridor, whether
natural or engineered. The second priority desirement is to regulate risky coastal land
developments through limited access to local coastal roads. This creates potential
conflicts with individual property rights; however, the right to own property does not
implicitly or explicitly guarantee the right to infrastructure. Costs of maintaining access
to at-risk properties is challenging budgets of many transportation agencies.
Dr. Derek Hitchins defines stakeholders as “those entities that stand to gain, or
particularly, to lose, from the successful implementation of a project or process [79].”
Since local stakeholder desirements represented in the competition were filtered by the
competing teams, a study recently completed for East Boston and Charleston in which
400 residents participated in the planning process was compared to the results from the
design competition [80]. The concerns, desirements and strategies are shown in Figure 16
and compare well to priorities presented in the design competition.
57
Figure 16. Local coastal resilience priorities for East Boston and Charlestown.
3.3 Functional Requirements
By providing efficient road networks, continued coastal development is
encouraged, but at what risk? Both the perspectives of stakeholders presented in the
Boston design competition and engineers represented in coastal highway design guidance
are considered as a system to determine required components and functions. Quality
function deployment (QFD) is a method to help transform customer needs (desirements)
into engineering characteristics for a product or service. The priority functional needs for
local coastal roads based on subjective valuation are shown in Figure 17. Priority
weighted functional requirements based on QFD analysis are ranked in descending order
in Figure 18 to address the question “How?” coastal roads can be better planned, sited,
and designed.
Concern
s
Coastal flooding threatening safety, property & livelihoods
Coastal risks and accessibility limits mobility, housing, open spaces & waterfront access
Flood insurance and housing costs are rapidly escalating
Local economic opportunities for residents/businesses
Connectivity with open public spaces not functional or integrated into community
Central urban district connectivity preservation
De
sire
me
nts
Effective & long-term solutions for coastal flooding
Safe & reliable transportation system for jobs/healthcare
Highway & transit infrastructure connectivity enhanced
Expanded safe system for biking & pedestrian connectivity
Water transit options, publicly accessible waterfront
Affordable neighborhoods for broader demographics
Open spaces for diverse recreational/passive use
Enhanced views, social spaces, ecological features
Str
ate
gic
Pla
ns
Elevated waterfront parks and plazas with soft features
Elevated waterfront pathways
Docks and other in-water features for water access
Nature-based marshes, living shorelines, terraces, beaches
Mobility enhanced with network of pedestrian, biking & water routes integrated with complete streets improvements
Elevated roadways and deployable floodwalls
Seafood & maritime industries integrated with mixed use developments for smaller built space footprints
58
These QFD functional requirements reflect what is expected and typical relative
to current planning and engineering guidance. With the current concept of restoring to
previous conditions after impact from major storms in coastal areas, sustainability of
maintaining these restoration practices in providing coastal resiliency as prioritized above
requires assessing whether this practice is functional or sustainable. System boundaries
lack definition in coastal environments to delineate between complex infrastructure
systems. Interactions between natural and manmade features increase these complexities
in risk and reliability analysis. Improvements to increase resiliency often impact other
systems and often resisted by a local community due to perceived negative impacts. For
Traffic Volume
Capability
Traffic Load Capacity
Service Levels
Acceptability
Maintenance and Costs Minimized
Right-of-way Easement
Restrictions
Increased Wave Energy
Resiliency
Rising Sea Levels
Resiliency
Strategic Planning Coastal
Resiliency
Route Selection to
Mitigate Risks
Elevated Causeway Wetland
Crossings
Integrated Systems Hazard Analysis
Combined Highway -
Levee System
Resilient Bridge
Structures
Route Realignment to Mitigate
Risks
Figure 17. Priority functional needs for local coastal roads based on subjective valuation.
Figure 18. Priority weighted functional requirements for local coastal roads based on
QFD analysis.
59
example, beach dune features to reduce wave energy and improve wildlife habitat are
often rejected by property owners because of obstructed views [81].
To facilitate a systematic evaluation of local coastal road requirements, functions,
and components; a basic system model was initiated using CORETM Spectrum UE 9.0
(SP 15) system analysis software provided by Vitech Corporation. An Integration
Definition (IDEF0) function modeling diagram organizes decisions, actions, and
activities for subsystems so these can be understood and improved. A systematic analysis
and functional decomposition provide a next generation approach for planning, siting and
designing local coastal roads. Based on stakeholder’s functional requirements in an urban
setting, seven system requirements are identified as priorities in Figure 19.
These requirements effectively identify what the next generation perceives as its
priorities for planning, siting, developing, and utilizing local coastal road systems located
in relatively high-risk environments. It recognizes that efficiencies of road systems
designed for transportation and flood risk reduction should be balanced with quality-of-
life experiences, modes of transportation other than automobiles, and environmental
considerations. Since regulatory agencies, land use planners, landscape architects,
engineers, environmental scientists, et al. associated with infrastructure development
work most efficiently in a structured approach, the outputs shown in Figure 19 provide a
conceptual framework for systematic planning and design of next generation local coastal
roads.
60
Figure 19. Local coastal road next generation IDEF0 diagram for stakeholders’ priority
desirements.
3.4 Conclusions
System engineers should approach coastal systems analysis from the perspective
of integrating local desirements with the functional and practical system requirements for
improving coastal infrastructure sustainability and resiliency. Increased physical and
economic damages impacting coastal roads, require reimagining fundamental system
requirements for coastal roads and associated infrastructure. Using the creative work of
61
others from the Boston competition, voices from a diverse group of stakeholders were
considered in applying systems analysis to local coastal roads. This analysis led to
different desirements and priorities than provided in current technical guidance.
While technical requirements of applying engineering principles in design do not
change, next generation priorities of framing systems analysis are consistently changing
functional requirements. The IDEF0 function modeling diagram provides reasonable
system requirements, inputs, outputs, and components for a systems framework, which
can be used in modeling local coastal road systems. Providing systems analysis tools for
local communities to reimagine connectivity and environments associated with local
coastal road systems, increases opportunity to meet desirements and needs of the next
generation. Changing climates, socioeconomic, and environmental conditions will make
local coastal roads systems analysis, planning, siting, and architecture design and
modeling both essential and beneficial for resiliency and sustainability planning.
62
CHAPTER IV
TRANSDISCIPLINARY SYSTEMS THINKING:
SUSTAINABILITY OF COASTAL SYSTEMS6
Sustainable development requires integrating economic, social, cultural, political,
and ecological factors into evaluation and decision-making processes. Sustainability
benefits from transdisciplinary systems thinking, since the integrated global coastal system
collectively and continuously evolves with natural and human subsystems competing for
resources. Disparity between multidisciplinary models in socio-economic and physical
sciences introduces unique challenges when applying transdisciplinary methodologies to
complex coastal systems. Systems science training and early adoption of a collaborative
systems approach framework are critical to success. Transdisciplinary research provides
opportunity for integrating and advancing a posteriori knowledge. The IN-CORE v1.0
model released on January 8, 2020 utilized a transdisciplinary systems approach in
development.
6 Pennison, G.P., and Webb, B.M. 2020. Transdisciplinary Systems Thinking: Sustainability of Coastal
Systems. Proceedings of the 2020 Industrial and Systems Engineering Conference. L. Cromarty, R.
Shirwaiker, P. Wang, eds. 6 pp.
63
4.1 Introduction
Sustainable engineered coastal systems benefit from integrating systems
engineering with coastal and other transdisciplinary sciences. External forcings are
constantly changing and increasing for engineered and natural coastal systems subjected to
increasingly extreme storm hazards. Defining logical and physical system architecture for
complex coastal systems is problematic with complex external interfaces and variable
boundary conditions. Engineered, natural, and meteorological coastal systems evolve
spatially and temporally during extreme storm events.
Increased coastal development places greater numbers of engineered structures at
risk of damage from coastal hazards. Functional sustainability and adaptability of human
development within these at-risk coastal zones benefit from transdisciplinary systems
thinking and modeling. Even in the absence of relative sea level rise (RSLR) or other
equally significant climate change impacts, developed coastal system sustainability is at
greater risk from increasing property development, coastal population growth, and
dependence on coastal and marine related economies.
Risk is the cumulative product of failure probability and consequences. Risk can
change in these complex systems even if the hazard probabilities remain fixed, due to the
potential for changing consequences associated with increased development in, and/or
reliance on, coastal areas and resources. The predictable stakeholder threat response is to
protect developed areas and coastal infrastructure at great expense without regard to
robustness and sustainability [1].
Sustainable development requires integrating economic, social, cultural, political,
and ecological factors into the evaluation and decision-making process [82, 83]. UNCED
64
Article 21 (§17.15) encourages integrated planning and decision-making for sustainable
development of coastal zones and the contiguous marine environment. Sustainability
priorities include systematic assessment of critical coastal areas, eroded zones, physical
processes, development patterns, user conflicts, environmental hazards, climate change
stressors, and sea level rise impacts.
A key difficulty with sustainable development of engineered coastal systems in a
natural system environment involves the inability to control or regulate an open system. A
second difficulty involves definition. Gallopin [83] notes that output and state variables are
typically synonymous when considering a “sustainable” system, in which preservation of
the system itself is the desired objective. Conversely a “functionally sustainable” system,
where the desired objective is system performance or output sustainability, produces
differing output and state variables. Since human development fundamentally changes
coastal systems, sustainability requires either improving or transforming built-components
to adapt and preserve functionality or adapting the system architecture to maintain
functionality.
Sustainable development benefits from transdisciplinary systems thinking, since
the changing state of natural and human subsystems collectively and continuously evolve.
Rapidly increasing external stressors created by extreme storm events impact system
functionality and threaten continued development and livability within coastal zones,
forcing system component interactions into catastrophic and often unrecoverable damage
states. When the system model includes socio-economic and political variables, an
optimized or incremental performance outcome that guarantees sustainable development
over extended generations and territories becomes highly unlikely. The challenge becomes
65
how best to evaluate and adapt engineered coastal systems with a sustainable development
approach in an integrated systems framework, given the alternative paradigms for
sustainability objectives defined by Schellnhuber [84].
Coastal systems resiliency assessments typically include some minor
considerations relative to sustainability. Typical coastal risk evaluation methodologies
applied by the United States Army Corps of Engineers (USACE) [85] include one or more
of the following tools: (1) multi-criteria decision analysis (MCDA) [86]; (2) scenario
analysis [87, 88]; (3) risk analysis [89]; and, (4) engineering for climate change and other
emergent conditions [87, 90]. Transdisciplinary sustainability considerations typically
emerge as independent system evaluations synthesized together in final work products.
Environmental Impact Statements (EIS) conforming with the National Environmental
Policy Act (NEPA) assess socio-economic and environmental impacts for proposed
engineered systems during the project development phase [74]. Independent consultants
typically execute this task independent of the project design team to limit bias in decision
making. The USACE North Atlantic Coast Comprehensive Study (NACCS) [73] and
Systems Approach to Geomorphic Engineering (SAGE)7 programs both represent a
changing trend within the coastal practice community. The United States Department of
Transportation (USDOT) Federal Highways Administration (FHWA) is applying
transdisciplinary policies to the NEPA process through its relatively new Eco-Logical
framework8.
7 http://www.sagecoast.org/ 8 https://www.environment.fhwa.dot.gov/env_initiatives/eco-logical.aspx
66
Complex sustainability problems for diverse and interconnected systems of interest
requires transdisciplinary systems thinking. Transdisciplinary research projects typically
involve three phases: problem transformation, interdisciplinary integration, and
transdisciplinary integration [91, 92]. Brink et al. critically examined various
transdisciplinary urban development and coastal adaptation post-research efforts in
Sweden to identify actionable sustainability measures that proved useful to social actors
and satisfied rigorous scientific quality criteria [91]. This review of transdisciplinary
research projects found that epistemic integration would greatly benefit from establishing
an initial systems framework to navigate and integrate the often-disparate transdisciplinary
approaches, not only among the researchers, but also among the community and
governmental stakeholders directly impacted by sustainability issues.
4.2 Systems Approach Framework
The SPICOSA (Science and Policy Integration for Coastal Systems Assessment)
was a 4-year European Union (EU) project (2007-2011) purposed to stimulate systematic
research restructuring and stimulate integration of new knowledge and methods throughout
the European Region9 [34]. The resultant SAF (Systems Approach Framework) research
product provided an organic holistic research approach for continued integrated assessment
of diverse complex systems. The project involved 18 Study Site Applications (SSAs); 22
countries; 54 research institutes, universities, and small enterprises; and, multidisciplinary
linking of ecological, economic, social, and governance sectors [32]. The program purpose
9 http://www.spicosa.eu/
67
was to create a collaborative system for scientific and transdisciplinary practitioners to
improve the Coastal Zone Systems (CZS) sustainability decision-making process.
This intensive research and planning effort identified problems with integrating
natural and social systems. The system model framework envisioned an improved coastal
zone feedback loop by implementing systematic Ecological-Social-Economic (ESE)
assessments. Reducing delay in feedback responses between economic or social changes
in local communities in response to loss of natural system goods and services could
potentially mitigate irreversible impacts on the natural system, and thereby improve system
sustainability [32]. Reviewers identified lack of researcher training in applying systems
science to the various disciplines as a consistent and fundamental research project
weakness.
Disparity between multidisciplinary models in socio-economic and physical
sciences also led to uncertain and sometime infective outcomes in applying
transdisciplinary methodologies to complex coastal systems. Innovative alternative
solutions for sustainability issues sometimes created controversy. Identifying and viewing
coastal systems as a relational network with complex emergent properties can generally be
beneficial when applying system thinking to developing effective policies that address
local and regional sustainability issues [33]. As system boundaries grow and become more
complex, effectiveness and enforceability of policies often conflict with self-preservation
interests.
Hopkins et al. [33] report that integrating systems thinking into research and policy
making within the SAF research program generally resulted in expanding single-issue
studies into multi-issue studies; expanding static to dynamic indicators; better defining
68
boundaries between system-dependent and system-independent problems; and,
establishing a quantitative basis for collaborative decision making. A key challenge
involves consideration of scale interdependency and need for global scale-free networks to
maximize integrated sustainability of human increasing dependency on coastal systems,
particularly when facing growing risk factors from climate change combined with growing
population density, development, and dependency on coastal at-risk environments.
4.3 IN-CORE Model Development
Evaluating sustainability for diverse natural and social systems when exposed to
extreme natural hazard events challenges system sustainability models developed for non-
extreme events. Resiliency systems models seek to incorporate sustainability by treating
these extreme events as risk and recovery resilience paradigms [6]. Extreme events often
cause irreversible damage to the fundamental environment and socio-economic systems,
effectively creating an unsustainable system for both the environment and human actors.
Quantifying resiliency and response of a system to an extreme event facilitates adapting
built infrastructure to improve sustainability.
Predicting system vulnerability and failure limit states typically involves complex
stochastic and empirical models. Modeling community disaster resilience requires
transdisciplinary experts to collaborate in modeling how physical, economic, and social
infrastructure systems within a real community interact and affect recovery efforts. A
transdisciplinary research team from 10 universities evaluated diverse engineered and
socioeconomic systems in assessing hurricane hazards for the Galveston Test Bed Model,
along with other types of natural disasters as subsystem models within the parent
69
Interdependent Networked Community Resilience Modeling Environment (IN-CORE) all-
hazards model10. The model evaluates community impacts during extreme events and
quantitatively measures resilience11.
The Colorado State University (CSU) Center for Risk-Based Community
Resilience Planning integrates engineering, social sciences, and economic disciplines in
comprehensively modeling community resilience. Systems that are essential for the
recovery and vitality of a community (technological, financial, social/political support,
healthcare delivery, education, and public administration) integrate within the IN-CORE
model by simulating natural hazards and geospatially applying system fragility functions
to publicly available asset inventories and databases. The model provides a quantitative
and science-based approach to assess community resilience at the local and regional levels
in response to natural disasters of varying intensities. The goal is to make local
communities more resilient, and in doing so, improve the likelihood of long-term
sustainability [27]. Developing the underlying systems science is challenging.
Marchese et al. [93] examine the complex interrelation of resiliency and
sustainability terminology with a thorough literature review. They note that there is
significant opportunity to develop sustainability practices that improve consistency and
integration with resiliency methods. They suggest framing sustainability as a critical
function of a project, policy, or system, with that functionality maintained during and after
an event. They recommend modeling those critical functions as a combination of
environmental, social, and economic indicators. Integration of these two mutually inclusive
10 http://resilience.colostate.edu/index.shtml 11 https://ssa.ncsa.illinois.edu/isda/projects/in-core/
70
system frameworks into a combined transdisciplinary approach potentially makes policy
implementation more palatable, and risk mitigation measures more defensible.
Development of these models stimulates the subsequent challenge in applying
systems thinking when integrating resiliency and risk assessment models into sustainability
model system frameworks. Just as with the SPICOSA EU effort [33], the challenge in
model development is to account for the core methodology’s long-term sustainability and
adaptability through research integration into policy and practice. If these transdisciplinary
models and system frameworks are not resilient and robust in application, then developing
and adapting these complex systems models becomes challenging and potentially
ineffective.
4.4 Emergent Knowledge
The system design process should assess system vulnerabilities in evaluating
sustainability. Sustainability infers the ability of a system of interest to be maintained at a
given level or state. Functional sustainability infers the ability to maintain a certain rate or
level of system functionality for the system of interest, while the system itself evolves and
adapts. This requires either providing additional system inputs or expanding the system
boundaries that allows the system to adapt with new emergent behavior properties while
maintaining a consistent system functionality or output. Extreme vulnerability and lack of
resiliency decreases a system’s ability to adapt and demonstrate sustainability.
Transdisciplinary systems thinking considers various measures of adaptability and
integration in assessing system sustainability, unlike independent disciplinary approaches.
Assessing emergent behavior of integrated subsystems is extremely complex relative to
71
subsystem and component functions, just as with transdisciplinary systems thinking.
Coastal design concepts most often focus on spatial definition and performance of
subsystems, with limited integrated system response analysis, particularly in response to
extreme events and natural disasters. Prevailing reasoning predicts that all disasters
generate catastrophic failure forcings for complex interrelated systems, such that the
coastal system response is unpredictable and most likely unsustainable. While a coastal
defense system may protect the built environment, resisting storm surge or wave forcings
may cause consequential damage to the natural environment.
Direct epistemic benefits encourage continued integration of diverse disciplines for
resiliency and sustainability research efforts, since literature shows a posteriori knowledge
consistently emerges from most transdisciplinary research projects. Assessing coastal road
system resiliency functions for the IN-CORE Galveston Test Bed Model included
identifying critical system components shown in the Figure 20 systemigram.
72
Figure 20. Transportation system resiliency systemigram.
Probabilistic systems analysis applied to ADCIRC+SWAN coastal model data for
Hurricane Ike that hit the Galveston area in 2008 [2], identified strong correlations between
road system damage relative to nodal site characteristics and coastal storm model data.
Initial analyses compared various incremental intensity measure (IM) data from coastal
model output to other measured and modeled attributes of the storm event, engineered, and
natural system components. Output data for the Hurricane Ike hindcast model included
wind fields, currents, water surface elevations, flooding depths, and wave characteristics.
Other transdisciplinary research teaming partners collaboratively and concurrently
assessed storm model data and pre- and post-event records, including detailed damage
73
assessments, in developing fragility function models [2]. These geospatial damage and
recovery models evaluate economic, social, building, utility, and transportation
infrastructure systems.
By analyzing incremental IM hourly output data from coastal storm numerical
hindcast models, functional decomposition and coastal engineering disciplinary research
identified cumulative celerity dispersion (CCD) functions that strongly predict likelihood
and degree of coastal highway damage during extreme storm events. The coastal road
system model responds to extreme storm events based on several key components as shown
in Figure 21.
Figure 21. Coastal road resiliency systemigram.
74
4.5 Transdisciplinary Systems Thinking
Integration of systems thinking in a transdisciplinary team research effort
provides opportunity to develop sustainable solutions for complex problems associated
with any system-of-interest. The IN-CORE model research program is making significant
advancements in defining resiliency and sustainability functions for complex natural and
engineered systems. By evaluating engineered system failure during extreme events,
numerical model data evaluated using both systems and coastal engineering analytics, led
to significant findings regarding coastal road system failures. Correlating cumulative
energy dispersion with likelihood and degree of failure will improve coastal road
systems.
75
CHAPTER V
COASTAL ROAD SYSTEM FAILURES:
CAUSE AND EFFECT12,13
Engineered and natural coastal systems benefit from integrating systems
engineering with coastal and other transdisciplinary sciences. By analyzing incremental
intensity measure (IM) hourly output data from coastal storm hindcast models, functional
decomposition identified cumulative celerity dispersion functions that strongly predict
likelihood and degree of coastal highway damage for extreme storm events. Discrete time
integration of these functions provides additional information related to identifying the
likely damage or failure mechanism. Identifying these provides opportunity to improve
both design and construction methods that enhance coastal road reliability and
sustainability.
5.1 Introduction
External forcings are constantly changing and increasing for engineered and
natural coastal systems subjected to increasingly extreme storm hazards as a result of
climate change. Coastal populations and developments in coastal zones are increasingly
12 Paper received 2020 IISE Conference Construction Division Best Track award. 13 Pennison, G.P., and Webb, B.M. 2020. Coastal Road System Failures: Cause and Effect. Proceedings
of the 2020 Industrial and Systems Engineering Conference. L. Cromarty, R. Shirwaiker, P. Wang,
eds. 6 pp.
76
at risk. Defining logical and physical system architecture for complex coastal systems is
problematic given the many external interfaces and variable boundary conditions.
Engineered, natural, and meteorological coastal systems also evolve both spatially and
temporally during extreme storm events.
Coastal hazards (e.g., storm surge, hurricane force winds and waves, tidal and
coastal flooding) place both population and coastal infrastructure at increased risk.
Increased levels of development damage results from higher tides, greater intensity
storms, natural wetland buffer loss, and beach erosion. Even in the absence of relative sea
level rise (RSLR) or other equally significant climate change impacts, developed coastal
system sustainability is at greater risk from increasing property development, coastal
population growth, and dependence on coastal and marine related economies. The
predictable stakeholder threat response is to protect developed areas and coastal
infrastructure at great expense without regard to robustness and sustainability [94].
Applying systems engineering when assessing risk and predicting reliability of
engineered coastal systems has been an active and growing area of research for several
decades. The inevitable question in post-event analysis for catastrophic failures is “What
caused the system to fail?” Identifying failure mechanisms for coastal road systems in
nearshore environments can reduce risk of future occurrences. While research benefits
from understanding a transportation system’s functional requirements associated with
resilience; fragility functions based on modeled storm output data potentially reduces
system damage and critical failures. Reliable transportation systems are essential to
community resiliency and recovery. Understanding which components are likely to fail
and why, improves reliability of coastal road systems when subjected to extreme threats
77
or incidents. Applying system engineering theories in quantifying uncertainties associated
with multivariate probabilities; or, not-mutually exclusive system failure modes; engages
many in pursuit of developing the perfect predictive model that resolve all uncertainties.
This approach potentially creates additional risk with consistent application or
misapplication of probabilistic methods or models potentially limiting identifying and
resolving system failures. Research objective is to avoid such pitfalls by utilizing means
and methods appropriate to the system of interest in applying systems engineering and
analysis in a consistent manner. The desirement is to develop reliable hazard functions,
not only in a manner that is scientifically defensible, but also broadly interpretable,
implementable, and discernable by the system actors and stakeholders.
5.2 Coastal Road System Failures
Transportation system risk and resiliency models in coastal zones provide various
system frameworks for assessing transportation network reliability [18, 76, 95-97].
FHWA has developed extensive design guidance for developing resilient coastal
highways [11, 39, 42]. Regional resiliency studies propose various vulnerability and risk
assessment frameworks [5, 14, 37]. Montoya et al. [40] propose a vulnerability model for
barrier island roads based on morphological data in relation to the roadway with three
vulnerability indicators measured along shorenormal transects: island width < 305 m;
dune crest elevation < 3 meters above the highway; and, edge of pavement within 70
meters of the ocean shoreline. Anarde et al. [41] propose a coupled hydrodynamic,
geomorphic, and engineering reliability model for assessing vulnerability.
78
When a coastal road, protective dune, armoring, or seawall fails along a coastal
road, communities flood and local coastal roads typically experience catastrophic
damage, precluding life-safety access, and delaying evacuation. Coastal system resilience
represents the integrated capacity and capability of a coastal system to recover quickly to
pre-event conditions when impacted by natural hazards such as hurricanes, coastal
storms, and flooding, rather than simply reacting to impacts. Since passive infrastructure
systems generally lack dynamic functionality to respond in an adaptive capacity manner,
this creates a problem in defining resilience and adaptive capacity for coastal road
systems.
Damage from overtopping weir flow of pavement, base, and embankment
routinely occurs seaward of coastal roads during ebb flow as storm surge retreats,
overtopping or breaching barrier islands. This failure mechanism at pavement sections is
well documented, particularly for riverine floods overtopping a broad-crested weir
pavement section [43, 44, 98]. Evaluation of overtopping storm surge for Highway 82 in
Southwest Louisiana utilized CFD flow simulation and acoustic analysis to assess flood
flow from storm surge and resulting differential pressure flow fields [99, 100].
Significant damage to Brazoria County Coastal Road 257 (CR 257) occurred
during Hurricane Ike in September 2008. Post-Ike road conditions included extensive
damage to pavement, base, and embankment [59]. CR 257 serves as a direct link between
Galveston Island and south Brazoria County. Damage ranged from partial failure at the
edge of pavement to complete breaching and destruction of both pavement and
embankment. Damage assessment teams observed that damage appeared to occur when
ebbing storm surge accelerated seaward as the hurricane moved inland, reversing
79
nearshore wind directions. Photographic records of damage indicate that many deep
asphaltic overlay sections failed [55]. While pavement materials and typical sections do
not appear to be critical components in determining the likelihood of failure, these are
important for sensitivity analysis and damage reduction considerations.
Some pavement failures and material displacement observations suggest that
rapidly varying hydrodynamic pressure gradients between the subgrade and surface may
generate moments that initiate rotational failures in pavement slabs [101, 102]. It is
known that the pressure oscillations for pore water under spillway chutes (underpressure)
propagate instantaneously without damping effects, sometimes producing a hydraulic
resonance in the slab [103]. Protective measures constructed along CR 257 now include
large dunes and overwash scour protection armoring that extends deep into the sand
embankment, similar to a system that constructed in Surfside after Hurricane Rita [55].
Research evaluated pre- and post-event using plans and damage assessment records to
confirm damage extents along CR 257. Damage severity ratings characterized damage
states from 0 (no damage) to 4 (road/embankment washout). Initial binary damage states
assumed that 0–1 collectively represents ‘no damage’ and 2–4 represents ‘damage’.
5.3 Developing the Failure Model
Evaluating system reliability for diverse natural and built coastal systems exposed
to extreme natural hazard events challenges system models developed for normal design
forcings. Extreme events often cause irreversible damage to the natural environment and
built systems. Quantifying resiliency and response of a built system subjected to an
extreme event facilitates adapting infrastructure to improve reliability. Predicting system
80
vulnerability and failure limit states typically involves complex stochastic and empirical
models. Modeling community disaster resilience requires transdisciplinary experts to
collaborate in modeling how physical, economic, and social infrastructure systems within
a real community interact and affect recovery efforts.
A transdisciplinary research team from 10 universities evaluated diverse
engineered and socioeconomic systems in assessing hurricane hazards for the Galveston
Test Bed Model, along with all other types of natural disasters as subsystem models for
the parent Interdependent Networked Community Resilience Modeling Environment (IN-
CORE) all-hazards model14. Model evaluates community impacts during extreme events
and quantitatively measures resilience15.
The Colorado State University (CSU) Center for Risk-Based Community
Resilience Planning integrates engineering, social sciences, and economic disciplines in
comprehensively modeling community resilience [2, 104]. Systems that are essential for
the recovery and vitality of a community (technological, financial, social/political
support, healthcare delivery, education, and public administration) integrate within the
IN-CORE model by simulating natural hazards and geospatially applying system fragility
functions to publicly available asset inventories and databases. The model provides a
quantitative and science-based approach to assess community resilience at the local and
regional levels in response to natural disasters of varying intensities. The goal is to make
local communities more resilient, and in doing so, improve the likelihood of long-term
sustainability [27]. Developing the underlying systems science is challenging.
14 http://resilience.colostate.edu/index.shtml 15 https://ssa.ncsa.illinois.edu/isda/projects/in-core/
81
FHWA developed guidance for hydrodynamic modeling in the coastal
environment [105]. Hydrodynamic models numerically simulate coastal storms to assess
resultant impacts on coastal highways due to flooding, wave damage, and scour.
Probabilistic systems analysis applied to ADCIRC+SWAN coastal model data [2] for
Hurricane Ike that struck Galveston in September 2008, identified strong correlations
between road system damage relative to nodal site characteristics and coastal storm
model data. Initial analyses compared various incremental intensity measure (IM) data
from coastal model output to other measured and modeled attributes of the storm event,
engineered, and natural system components. Systemigram in Figure 22 identifies critical
functional relationships between subsystems and forcings.
82
Figure 22. Coastal road resiliency systemigram.
5.4 Identifying Failure Mechanisms
Output data for the Hurricane Ike hindcast model included wind fields, currents,
water surface elevations, flooding depths, and wave characteristics. Other
transdisciplinary research teaming partners collaboratively and concurrently assessed
storm model data and pre- and post-event records, including detailed damage
assessments, in developing fragility function models [2]. These geospatial damage and
recovery models evaluated economic, social, building, utility, and transportation
infrastructure systems. Hurricane simulations for different event frequencies impacting
83
the same area validated subsystem assumptions and models for incorporation into the
parent IN-CORE model.
By analyzing incremental intensity measure (IM) hourly output data from coastal
storm numerical hindcast models, functional decomposition and coastal engineering
disciplinary research identified cumulative celerity dispersion (CCD) functions that
strongly predict likelihood and degree of coastal highway damage during extreme storm
events as shown in Figure 23. The coastal road system reliability model responding to a
significant hazard event also includes several key external rate controllers determined by
the storm events intensity, duration, and frequency.
Figure 23. Cumulative celerity dispersion (CCD) function.
84
Storm duration affects the cumulative aggregate values within the CCD function
such that the longer the duration of storm event for overtopped road sections, the greater
the damage. As storm intensity increases, generally water surface elevation increases due
to the energy of the extreme weather system with increased wave periods frequency
(shorter waves), along with increased maximum (significant) wave heights. Damage
mitigation measures conversely typically reduce the degree of road damage, which
reduces the resultant damage functions (i.e., improves reliability function).
Subsequent analysis identified cumulative celerity, distance to shoreline, and
cumulative flow velocities as critical variables in determining likelihood and degree of
damage during a major storm event. Systems analysis applying coastal nearshore
hydrodynamics determined that the CCD function strongly predicts likelihood and degree
of damage for coastal roadways impacted by coastal storm surge and wave hazards.
Evaluating the cumulative function with time-step integration assists with identifying
probable damage failure mechanisms at critical damage limit states.
5.5 Resilient Design and Construction
The CCD function, using cumulative overtopping water surface elevation and
cumulative wave period hourly incremental intensity measures (IMs) for overtopping
flows, strongly correlate in predicting road damage along CR 257 on Follet’s Island in
Brazoria County, Texas. The model strongly predicts road failure when cumulative
modified celerity dispersion function exceeds a critical threshold value or limit state.
Discrete time integration of the CCD function finds that it reaches the threshold or limit
state with significant road damage likely occurring shortly after flow velocity vectors
85
redirect seaward (ebb flow) across the barrier island, which is consistent with other event
observations. Analysis of incremental IM data also shows that failure limit state coincides
with depth-limited irregular sea state wave breaking depths, when wave heights reach
critical wave breaking depth relative to overtopping water surface depths. Breaking wave
impacts likely create rapidly varying pressure gradients under the pavement, increasing
trapped air uplift dynamic forces, displacing pavement upward, and subbase material
seaward (i.e., like compressing bellows).
Other critical parameters include duration of event which affects the magnitude of
cumulative CCD values. The distance to shoreline also strongly influences likelihood of
failure, correlating to previous research findings by others. Cumulative velocity is also
critical to failure since it provides the mechanism for sediment scour and transport away
from the damaged site. Model application to assess US Highway 80 damage during
Hurricane Katrina subsequently confirmed the underlying hydrodynamic coastal theories
with publication pending in coastal science journals.
While these functions predict the likelihood of coastal road failure during extreme
storm events, they also provide opportunity to improve resiliency. In identifying the key
variables that determine the likelihood of failure, some can be adapted in design to reduce
system vulnerability and to improve reliability. The critical hazard variables associated
with the storm event provide opportunity to mitigate impacts during design based on a
design probability of recurrence, including considering climate change impacts such as
sea level rise over the system’s design life cycle. As one example, determining the
optimal road elevation at a given location by evaluating the CCD function over a range of
extreme storm events, provides opportunity to minimize damage risks for hurricane
86
hazards. Elevating the entire roadway along a barrier island restricts overtopping flow,
thereby causing unintended consequential damage to other built and natural systems.
Lowering road elevations farther away from the shoreline facilitates overtopping flow at
depths and velocities that deposit a sand layer that protects, but does not damage, the
pavement structure.
System modeling tendencies are to integrate system component functions to a
point that the complexity of engineered systems placed into a natural system potentially
fails to identify the most likely failures. By evaluating engineered system failure during
extreme events, storm model output data evaluated using both systems and coastal
engineering analytics, led to significant findings regarding catastrophic coastal road
system failures. A systematic approach has identified many critical correlations not
previously described, with coefficients of determination 0.98 or greater for damage
likelihood models. Understanding these critical failure functions provides opportunity to
consider those risk factors in both design and construction, and in so doing improve
coastal road system reliability and sustainability.
CHAPTER VI
ASSESSING COASTAL ROAD SYSTEM RELIABILITY
USING CELERITY DISPERSION FUNCTIONS
6.1 Coastal Road System Reliability
Coastal road system reliability requires that the transportation system meets or
exceeds functional and performance design requirements when subjected to extreme
weather-related events. Coastal hazards include storm surge, waves, and currents;
impacting coastal roads for an extended storm duration. Significant coastal events such as
hurricanes, cyclones, typhoons, super storms, and nor’easters generate extreme forcings.
Nearshore coastal road systems often incur significant damage during extreme events
ranging from minor overtopping scour to catastrophic destruction of pavement and
embankment infrastructure.
Extreme forcings increasingly result in significant damage to nearshore coastal
roads. Improving the resilience of coastal roads subjected to sea level rise and extreme
storm events reduces the likelihood of catastrophic damage to critical transportation
infrastructure [5, 11, 12, 37-39]. The prevailing theory attributes coastal road damage
primarily to storm surge overwash, impacting waves, and ebb flow scour. Overtopping
weir-flow is considered the critical damage mechanism since the roadway embankment
88
section effectively creates a broad-crested weir structure as storm surge ebbs [11, 14, 42-
44].
There is limited research investigating potential failure characteristics that
correlate system components to the likelihood of coastal road failure [40, 41], particularly
when applied to disparate events and locations. A cumulative celerity dispersion (CCD)
function presented in this paper strongly predicts damage likelihood. It facilitates
development of coastal road system fragility functions that describe the system’s capacity
to resist forcings and limit failure when impacted by extreme tropical storm system
forcings of varying intensity and duration. Continuum mechanics and nearshore
hydrodynamics provide foundational theory for the CCD function basis as a cumulative
pseudo-Froude function.
6.1.1 Transdisciplinary Research
A transdisciplinary research team from 10 universities evaluated diverse
engineered and socioeconomic systems in assessing hurricane hazards for the Galveston
Texas Testbed Model, in addition to other types of natural disasters, as subsystem models
for the Interdependent Networked Community Resilience Modeling Environment (IN-
CORE) all-hazards model16. The IN-CORE Model proposes to assess community impacts
during extreme natural disaster events and quantitatively assess resilience17. By
accurately quantifying risks using a physics-based model, communities can assess
16 http://resilience.colostate.edu/index.shtml 17 https://ssa.ncsa.illinois.edu/isda/projects/in-core/
89
alternative resiliency measures for infrastructure systems. This research contribution
assesses the likelihood of coastal road failure during extreme tropical events.
Increased coastal development places greater numbers of engineered structures at
risk of damage from coastal hazards. Even in the absence of relative sea level rise (RSLR)
or other equally significant climate change impacts, sustainability of coastal system
developed infrastructure is at greater risk from increasing property development, coastal
region population growth, and dependence on coastal and marine related economies.
Research demonstrates that a systems approach to a coastal road system
beneficially identifies critical hydrodynamics and components that will likely fail when
subjected to extreme climatological system forcings. By determining critical modeling
parameters with strong correlation to failed damage states, nearshore hydrodynamics
generate functions correlating probabilistic analysis to water wave mechanics in
identifying variables that increase failure probabilities. Identifying these variables provides
opportunity to design coastal road infrastructure that mitigates the likelihood of damage.
6.1.2 Coastal Road System Risks
Applying systems engineering to assessing risk and predicting reliability of
engineered coastal systems has been an active and growing area of research for several
decades [29-36]. Coastal zone transportation system risk and resiliency models provide
various system frameworks for assessing transportation network reliability [18, 76, 95-
97]. FHWA has developed extensive design guidance for developing resilient coastal
highways [11, 39, 42]. Regional resiliency studies propose various vulnerability and risk
assessment frameworks to improve system reliability [5, 14, 37].
90
Proximity of coastal roads to shoreline increases the likelihood of failure [11, 40,
61, 62]. Montoya et al. [40] propose a vulnerability model for barrier island roads based
on morphological data in relation to the roadway with three vulnerability indicators
measured along shorenormal transects: island width < 305 m; dune crest elevation < 3
meters above the highway; and, edge of pavement within 70 meters of the ocean
shoreline.
Anarde et al. [41] propose a coupled hydrodynamic, geomorphic, and engineering
reliability model for assessing coastal transportation system vulnerability. Wainwright et
al. suggest a modeling framework that integrates geological, engineering, and economic
approaches for assessing climate change driven economic risk to coastal developments
[19]. That research utilizes a joint probability storm erosion and hazard model analysis
developed by Callaghan et al. for shoreline transects [20-22] in predicting the likelihood
of damage.
Seaward damage along coastal roads caused by overtopping weir flow of
pavement, base, and embankment results as storm surge retreats, often overtopping or
breaching barrier islands. Weir flow scour failure at edge of pavement is well researched
for riverine floods overtopping a broad-crested weir type pavement section [43, 44, 98].
Evaluation of overtopping coastal storm surge for Highway 82 in Southwest Louisiana
utilized CFD flow simulation and acoustic analysis to assess scour from storm surge
flood flow that creates differential pressure flow fields [99, 100].
Some pavement failures and material displacement observations suggest that
rapidly varying hydrodynamic pressure gradients between the subgrade and surface may
generate moments that initiate rotational failures in pavement slabs [101, 102]. Pressure
91
oscillations for trapped pore water associated with saturated soils under spillway stilling
basin slabs (underpressure) propagate instantaneously without damping effects,
sometimes producing a significant hydraulic resonance in the slab [103].
6.1.3 Peak versus Cumulative Intensity Measures
A fragility function represents the conditional probability of exceeding a system
component limit state or a building damage state given an intensity measure or vector.
Basco and Mahmoudpour [106] proposed integrating storm surge height, wave
conditions, and storm duration to predict damage likelihood for transportation
infrastructure (roads, bridges, railroads, etc.) using fragility functions. The Modified
Coastal Storm Impulse (COSI) parameter quantifies risk and resilience fragility curves
based on water surface elevation and wave conditions (height, period, direction)
integrated over the storm duration.
Combined wind-wave-surge hurricane-induced building damage functions
developed for the IN-CORE Galveston Texas Test Bed Model evaluate hourly intensity
measures (IMs) to assess incremental multihazard damage fragilities [2]. Incremental
hourly IMs extracted from the tropical storm coastal hydrodynamic model predict
multihazard building damage states at each discrete time step for the event duration.
Critical storm hazard parameters for structures include 3-second gust wind speed,
stillwater flood depth, and significant wave height. Final building damage state
probabilities use time dependent reliability functions to predict the likelihood of spatial
building damage resulting from the tropical storm event.
Advanced hydrodynamic coastal models reflecting local storm characteristics are
critical for multihazard damage prediction or loss assessment. Models account for storm
92
surge level and wave height determined by the direction and magnitude of wind velocity,
terrain roughness, and effects of local topography on wind, current, and wave stress.
While event peak IMs provide critical information reflecting a storm’s severity and
intensity, these measures do not necessarily reflect the extent and degree of damage
resulting from a storm. Locally variable wind, water, and wave incremental measures
during the event; relative to intensity, sequence, and duration; are collectively significant
in predicting the order of failure modes and resultant building damage states.
Coastal road damage functions describing the damage likelihood of a coastal road
during a major storm event suggest correlations between peak wave height, inundation
duration, and distance to shoreline. Dr. Ioannis Gidaris developed a failure probability
function shown in Figure 24, while evaluating the Galveston Test Bed Model in
coordination with the research team. Subsequent research by Yousef et al. propose a
fragility function for estimating likelihood of roadway failure as a function of road offset
distance from shoreline and inundation duration (research publication pending).
93
Figure 24.Failure probability function for coastal road damage during Hurricane Ike,
Galveston Texas (from Dr. Ioannis Gidaris).
6.2 Correlating Coastal and Systems Engineering
Research supports application of systems engineering to an engineered coastal
road system when exposed to tropical cyclone hazards, which represents a largely
unpredictable extreme climatological wind and water storm system evidenced by high
speed winds, extensive coastal storm surge, and significant water wave heights. Research
demonstrates that a systems approach to an engineered system helps identify critical
hydrodynamics and mechanics that result in complete or partial failure when subjected to
extreme climatological system hazard forcings. By determining critical modeling
parameters with significant correlation, nearshore hydrodynamics generate semi-
empirical functions correlating probabilistic analysis to wave and storm surge mechanics
in identifying variables that most significantly impact failure probabilities.
94
Research examines coastal roadway system failures by exploring damage
mechanics and semi-empirical models using physics-based relationships. Initial analysis
broadly examined correlations between various system component parametrics relative to
modeled hourly incremental intensity measures with expectations of identifying
correlations related to physical parametrics. Subsequent examination determined that
CCD functions strongly predict the likelihood of failure and supports a likely cause-effect
relationship as to when and how damage occurs.
CCD and Froude functions utilize common variables when computing cumulative
values over the storm’s duration at a specific node location along the coastal road
centerline. Each node location defined by geospatial coordinates provides an independent
data set within the coastal hydrodynamic models. The forcing parameters generated by
the coastal models and the road pavement/environmental siting characteristics are unique
at each node location. Coastal storm models provide incremental discrete stillwater
elevation, wave height, wave period, current velocity, and current vector data for
evaluating failure functions. These functions effectively represent the cumulative
dispersive energy impacting and potentially damaging the coastal road system.
6.2.1. Event and Study Area
Engineered system of interest involves regional road transportation infrastructure,
namely coastal road systems located near the shoreline. Galveston Test Bed modeling
efforts directed this research towards evaluating nearshore coastal road systems.
95
6.2.1.1 Hurricane Ike
Approximately 20 kilometers of County Road (CR) 257 on Follet’s Island,
located between Christmas/Drum Bays and the Gulf of Mexico in Brazoria County
Texas, experienced extensive and significant damage during Hurricane Ike in early
September, 2008. Damage ranged from edge of road scour to complete destruction of
road and embankment sections with avulsions along the barrier island [55]. Since the
post-event damage assessment generated detailed records for the extensively damaged
road system relative to location and extents, these records facilitate correlation of storm
and damage data [59]. CR 257 consists of an elevated coastal asphalt pavement road
section setback at varying offset distances from the shoreline with a gently sloping beach
and protective dunes of varying elevation. The beach increases in steepness to the
southwest near Surfside, where breakers occur more frequently due to the steeper
nearshore slope as the name implies.
6.2.1.2. Hurricane Katrina
Research assessed whether CCD function applies to other locations and events by
assessing Hurricane Katrina’s impacts on US 90 in Harrison County, MS. US 90 is the
nearshore multilane coastal highway that parallels the approximately 40 kilometers
shoreline from Bay St. Louis to the Biloxi Bay bridges. Hurricane Katrina impacted the
Mississippi coastline in August, 2005 with extensively documented and researched major
coastal bay bridge failures [3, 11, 42, 107-118]. Relatively minor road damage occurred
along US 90 with little or no damage data records available from the Mississippi
Department of Transportation (MDOT). Coastal defenses along US 90 include a largely
buried shoreline stepped seawall and concrete curbs/barrier rails. Location of US 90
96
along the more steeply sloped and elevated mainland shoreline creates different storm
surge and wave breaking characteristics than Follet’s Island.
6.2.2. Damage Characterization
Hurricane Ike represents a major storm with extended duration (9/11 – 9/15/2008)
and extensive overtopping flow of varying depth along the barrier island road. Hurricane
Katrina represents an equally significant storm of shorter duration (8/29 – 8/30/2005),
with deeper overtopping flood flow depths at the roadway section.
6.2.2.1 CR 257 Damage
Sixty-seven (67) data points categorize relative post-Ike damage states along CR
257 using plans and aerial/site photos. Damage does not include road sections with sand
or debris overwash requiring clearing post-event, if pavement section remains intact.
Damage states utilize a 5-point rating scale:
• 0 –No Damage
• 1 – No Road Damage - Beach Erosion near Embankment
• 2 – Edge of Road Scour Seaward - Pavement Damage < Half-Width
• 3 – Road Damage - Full Width
• 4 – Road and Embankment Washout
For preliminary analysis, damage states 0–1 collectively represent ‘no road
damage’; whereas, damage states 2–4 collectively represent ‘road damage’. It appears
that critical damage occurs with ebb flow as storm surge retreats across the barrier island.
Broken asphaltic pavement material deposited some distance seaward suggesting
significant material transport forces. Length of roadway partially damaged was
approximately 5.6 km; length destroyed was approximately 3.7 km. Photographic records
97
from post-event damage assessment show failed road sections included multiple layers of
base and pavement from road overlays, representing a significant and deep asphaltic
pavement and stabilized base typical section demolished by the storm event.
6.2.2.2 US 90 Damage
Eighty-eight (88) data points categorize relative post-Ike damage states along CR
257 using aerial/site photos showing pavement overlays for damage repairs. Substantially
less damage occurred to US 90 as compared to CR 257. Thick deposits of sand and debris
covered US 90, protecting underlying pavement from further damage. MDOT quickly
contracted repair work for US 90 to restore vehicular access in support of restoration
efforts without a post-event record to document damage extents.
6.2.3 Hydrodynamic Storm Models
Previous studies document the use of the ADCIRC + SWAN model to generate
incremental intensity measures for Hurricane Ike coastal hazards research, including
development of the Hurricane Ike Galveston Test Bed hindcast model [2, 41, 50, 119,
120]. All hydrodynamic models assume fixed bed elevations over the event duration.
6.2.3.1 Galveston Testbed Models
A hindcast ADCIRC + SWAN simulation of Hurricane Ike (2008) provided data
for the coastal hazards test bed study as documented in previous papers [2, 47]. Data
boundaries established for test bed model captured the counties of (in whole or in part)
Chambers, Galveston, Harris, and Brazoria. Hurricane Ike simulation used the
dynamically coupled version of ADCIRC and SWAN (PADCSWAN v51.52.34). The
coupled ADCIRC + SWAN model simulation generated incremental time series of flood
depth, significant wave height, and currents at select locations during Hurricane Ike.
98
Relevant tidal constituents force the ADCIRC model along an open ocean
boundary of the model mesh along with external meteorological forcing supplied in terms
of wind and pressure fields. Dynamic coupling between ADCIRC and SWAN occur
every 10 minutes in the simulation, incorporating effects of changing water levels and
waves frequently throughout the model run. Two different types of data validate the
model: time-dependent measurements of water levels along the coast and waves further
offshore; and measured high water marks (HWMs) determined after the storm event by
subject matter experts.
Wave characteristics predicted by the model are the spectrally significant wave
height (Hmo) and peak wave period (Tp). The statistically significant wave height (Hs)
represents the average of the largest one-third of wave records for wave height
measurements. These two representations of significant wave height are nearly equal in
deep water, but diverge in shallow water with the ratio of Hs to Hmo approaching a
maximum value of about 1.7 [121].
The ADCIRC model predicts spatiotemporal distribution of water levels as a
result of all astronomical and meteorological forcings, including resultant contributions
from wave action predicted by the SWAN model. These water level predictions output as
data on an hourly basis at every node in the model mesh with data referenced as water
surface elevations (WSE), still water levels (SWLs), or sometimes even water surface
elevations (SSEs). All represent the same value referenced to NAVD88 vertical datum.
In addition to prediction of water levels at every mesh node, the ADCIRC model
also provides an estimate of the water velocity due to all forcing including the excess
momentum due to wave action. Output data records the water velocity components
99
(easting, northing / U, V) at one-hour intervals for each node. The primary wave
characteristics extracted from model results includes spectrally significant wave height,
peak wave period, and wave direction.
6.2.3.2. Hurricane Katrina Models
Hurricane Katrina created one of the most significant water surface elevations
ever recorded at 27 feet in Bay St. Louis [11]. A hindcast ADCIRC + SWAN simulation
of Hurricane Katrina (2005) provided data for the coastal hazards test bed study;
however, terrain data for the coarse grid mesh failed to accurately represent pre-event
elevations along the road centerline. An XBEACH model utilized 2004 LiDAR terrain
elevations with 4-meter grid spacing to model the US 90 corridor refining the WSE and
velocity data. ADCIRC + SWAN model provided spatiotemporal forcings for XBEACH
model simulations. Model also provided better wave data at node locations.
6.2.4 Model Verification and Validation
Verification assesses whether a system model accurately represents the
developer’s requirements. Validation evaluates whether the model provides an accurate
representation of the real-world system relative to its intended uses.
6.2.4.1 Hurricane Ike
A total of 109 high water mark (HWM) measurements, made by FEMA
assessment teams, were available within the project study area to validate model data
(NAVD88 vertical datum). Figure 25 compares modeled and measured water levels
referred to as storm surge hydrographs, and their resulting errors for six tide gage
locations. Only three of the six gages contained complete time series and those that failed
did so on the rising limb of the surge hydrograph, subsequently missing the peak surge
100
event. The ability of the model to reasonably reproduce water levels before, during, and
after the storm event is evident in these plots. The RMSE for these comparisons range
from 0.11 meters to 0.25 m, and the corresponding peak errors (RMSE divided by the
maximum water surface elevation) are generally less than 11%.
Figure 26 shows a more direct comparison of the model-data errors associated
with time-dependent water levels as a plot of the ordered pairs of modeled and measured
water levels at each of the six gages during the storm hindcast. As shown by the linear
regression and corresponding equation, the model was able to accurately predict
measured water levels with a goodness of fit coefficient of nearly 0.95 out of 1.0, and a
regression slope of approximately 0.94. Model errors are generally proportional to the
magnitude of the water surface elevations. Lower water levels have lower errors, while
higher water levels have higher errors.
Figure 25. Comparison of modeled (ADCIRC) and measured (NOAA) time-dependent
water levels during Hurricane Ike (2008).
101
Figure 26. Direct comparison of modeled (ADCIRC) and measured (NOAA) water levels
with a line of perfect agreement and a linear regression equation for all data.
Validation of primary wave characteristics (height, period, and direction) utilizes
time histories of measured wave characteristics retrieved from available NDBC buoy
records in the Gulf of Mexico. Two buoys (42001, 42002) are in very deep water in the
central Gulf of Mexico basin, while three (42019, 42020, 42035) are in shallow water
landward of the continental shelf break. Waves at these latter three locations are more
influenced by bathymetry and their properties will change due to forcing (wind),
dissipation (friction, wave-wave interactions), and transformations (shoaling, refraction,
etc.). Modern wave models struggle to capture all interactions.
For this validation, model-data comparisons only include wave height and period
as most buoys did not contain reliable wave direction measurements. The five sub-panels
of Figure 27 compare modeled (ADCIRC + SWAN) and measured (NDBC) time series
102
of significant wave height and peak wave period. The wave heights at these locations
generally ranged from 1 meter to over 8 meters during the storm event, with wave periods
varying between 6 s and 17 s.
As shown in these figures, the model was able to reproduce the time histories of
wave height and period with some skill, though errors associated with magnitudes and
phase (timing) are evident, particularly at the nearshore buoy locations (42019, 42020,
40235). The RMSE for wave height ranged from about 0.5 meters to 0.75 meters with
corresponding peak error ranging from 7% - 12%. The RMSE for wave period ranged
from 1.5 s - 2.5 s with a peak error range of 8% - 16%.
Figure 27. Model-data comparison of significant wave height and peak wave period at the
NDBC locations.
Model validation compared modeled and measured wave height and period. For
the wave height comparison, goodness of fit is nearly 0.9 out of 1.0 and the slope of the
103
regression is approximately 0.89. The model reproduces about 90% of measured values
with a tendency to overestimate wave height magnitudes by about 11%. Comparison of
wave periods shows slightly worse agreement, with a goodness of fit of about 0.77 and a
regression slope of 0.88. Like the wave height and water level predictions, the model
tends to overestimate wave period.
Model validation also compares modeled maximum water surface elevations to
spatially discrete high-water marks measured after the storm event. Use of the high-water
mark observations is a helpful but often imperfect tool for validating a model. They are
helpful because they are often the only points of comparison available away from the
coast and/or shoreline, where tide gages are located.
The maximum predicted still water elevation extracted from the model results at
each of the discrete post-event HWM locations, converted to a consistent vertical datum
(NAVD88), and directly compared with the corresponding published 109 high water
mark (HWM) measurements verified by FEMA. The model tends to overestimate the
actual values about 70% of the time.
Direct comparison of modeled and measured data finds that the model
consistently overestimated the HWMs for elevations greater than 3 meters (NAVD88).
The goodness of fit of the linear regression is about 0.64 with a slope of 0.66. Analysis of
errors shows the relative and cumulative frequencies between the under/over estimations
ranging from -1.5 meters to +1.5 m. Nearly 50% of the HWM comparisons fall within the
-0.5 meters to +0.5 meters error bands (N=109). When expanded, 90% of the HWM
comparisons fall within the -1.0 meters to +1.0 meters error bands (N=109).
104
6.2.4.2. Hurricane Katrina
The XBEACH model for the Bay St. Louis to Biloxi Bay Mississippi Gulf Coast
shoreline utilized previously validated ADCIRC + SWAN model data to provide forcing
data for 4 area cross-shore grids, each approximately 10 kilometers in length along the
US 90 corridor. WSE data from the XBEACH model reflected refined mesh 2004 LiDAR
data along the US 90 corridor with improved model output relative to the ADCIRC +
SWAN model. CCD analysis utilized wave data from the ADCIRC + SWAN coupled
model.
6.2.5 IM Data Extraction and Preparation
WSE values extract as elevations relative to datum and not flow depths. Water
depth at any node requires subtracting ground elevations from the WSE. Cumulative
functions evaluate incremental hourly data output from the model for the node coordinate
locations along the coastal road centerline. The spatiotemporal data extraction for
incremental IMs includes hourly data for evaluating CCD and pseudo-Froude functions at
nodes. Output IM data extracted from the coastal model using MATLAB routines include
WSE, wave height (Hmo) and period (Tp), and water velocity values (U, V).
6.2.6 Cumulative IM Analysis Techniques
Model data output from the coastal hydrodynamic models represents incremental
hourly IMs and/or vector data extracted at each node location along the coastal road
centerline. While data evaluated by the model represents an integrated continuous
dataset, each discretized data set generated by the hydrodynamic model represents an
independent data set that is spatiotemporally unique at that time and location along the
coastal road. Because these data are independent, cumulative functions that strongly
105
predict the likelihood of damage with a coefficient of determination R2 ≈ 1 have a greater
significance.
Cumulative values presented in this study assume aggregation of hourly output
data or functions, since the resultant integrated values of discretized data assumes “per
hour” temporal units without specification. This research uniquely assesses cumulative
incremental intensity measures at different node locations to predict the likelihood of
damage resulting from significant storm events.
Hazard research for coastal transportation systems generally recognizes that storm
duration correlates directly to the likelihood and degree of damage. Like earthquake
fragility functions, event magnitude also directly correlates to damage likelihood. The
limit state at which catastrophic damage occurs is particularly critical when evaluating
integrated hazard functions.
Study utilized different modeled storm output data representing 50-, 100- and
205-year return periods at the same CR 257 node locations as Hurricane Ike to verify
continuity and sensitivity of CCD and pseudo-Froude functions for different storm
intensities. Damage data is not available for these other modeled events. Study then
added 30 new nodes to verify the CCD functionality in predicting likelihood and degree
of damage from Hurricane Ike. Study also applied tool to a different storm and coastal
topology data set by evaluating US Hwy 90 damage from Hurricane Katrina.
6.3 Cumulative Failure Functions
A fragility function or semi-empirical tool that limits uncertainty and predicts
likelihood of coastal road damage when subjected to storm forcings also identifies
106
adaptable mechanisms that can reduce risk and improve system resilience. It helps
identify those variables that improve reliability. While investigating fragility functions
for hazards mapping functions associated with coastal roads associated with the IN-
CORE model, research identified several findings that justify continued development.
1. Gravity wave celerity dispersion functions computed using cumulative water
surface elevation and cumulative wave period hourly incremental IMs for
overtopping flows are strongly correlated for event duration. Incremental IMs
facilitate cumulative discrete data integration.
2. Cumulative functions vary based on distance between CR 257 and shoreline
when approximately measured at mean low water using pre-event aerials.
Including setback distance as a gradient variable improves damage grouping.
3. Cumulative flow velocity creates a kinetic energy gradient and sediment
scouring forcing with overtopping flow or road section. Cumulative celerity
dispersion (CCD) function that accounts for this energy gradient relative to
distance of road from shoreline strongly predicts likelihood and degree of
damage. Evaluating the cumulative function progression during an event
facilitates validating likely damage failure modes.
6.3.1 Critical Physical Parameters
Previous research and data assessments by others identified that distance of
coastal road from the shoreline measured at mean low water provides a critical parameter
in determining the likelihood of damage during a significant event. Road offset distance
of 145 meters or less from shoreline resulted in damage for CR 257, regardless of dune
107
height. Similar probability of failure results when location of the road is within 75 meters
of the back-bay edge of marsh.
Shoreline armoring, seawalls, barrier rails, dunes, armored pavement shoulders,
rigid pavement slabs, etc. critically protect road infrastructure during significant events.
The seawall in Galveston protected the island road systems and reduced the likelihood of
major damage. Features such as concrete barrier rails modify overland flow depths and
wave breaking characteristics, but not typically considered in model meshes.
6.3.2 Coastal Hydrodynamic Model Output
Several criteria are applicable when evaluating CCD functions based on findings.
1. Sensitivity analysis finds filtering or excluding data aggregation when wave
heights are less than 0.3 meters does not substantively change cumulative
values, but does reduce data scatter when evaluating dispersion functions.
CCD values in this study apply that filter.
2. Cumulative IM values for velocity, WSE, and waves exclude data if the WSE
IM is at or below the pre-event road centerline elevation (i.e., no road
overtopping at that time step).
3. WSE used in evaluating CCD functions are measured relative to zero datum
elevation (NAVD88) and not relative to overtopping depth at the road
centerline.
4. CCD(WSE) computations include both WSE (h+Hmo) terms in cumulative
functions. CCD functions are linearly correlated with or without inclusion of
wave height in computations, but data scatter decreases by including wave
height in the celerity function.
108
6.3.3 CCD Functions
Shallow water wave celerity according to Linear Wave Theory (LWT) represents
the frequency dispersion of gravity waves on the water surface when the shallow water
depth h is less than or equal to 0.05 times the wavelength λ. Munk [122] noted that an
oscillatory wave moving into shallow water may be represented as a solitary wave.
Shallow water surface waves slow as the square root of the depth when waves begin to
interact with the bottom. In shallow water, phase velocity slows to equal group velocity
such that the product of wave period and wave celerity equals wavelength λ. Equation (2)
provides a reasonable approximation of shallow water celerity for a solitary wave, where
h is water depth (measured relative to zero datum) and Hmo is the wave height [123].
𝐶 = √𝑔(ℎ + 𝐻𝑚𝑜) (2)
Equation (3) expresses the wave celerity dispersion relationship from Hedges
equation (23), where 𝜖 ~ 𝑘𝑎 (k is wave number, a is wave amplitude) [124, 125].
Equation (4) reformulates the expression by substituting angular wave frequency 𝜔 with
2𝜋/𝑇.
𝐶 =
𝑔
𝜔[tanh[𝑘ℎ] + 𝜖] (m/s) (3)
𝐶 =𝑇
𝜋[𝑔
2[tanh[𝑘ℎ] + 𝜖]] (m/s) (3)
109
For shallow water depths and the modeled wave steepness associated with
Hurricane Ike, the value of tanh[𝑘ℎ] approaches 0, and 𝜖 appears to approach the
asymptotic value of 2/𝑔 or 0.2. Kirby and Dalrymple accepted the Hedges basic
dispersion equation (23) as agreeing with Stokes for the case of 𝜖 = 0.2 [124, 125].
While some relationships evident in the CCD functions represent artifacts of the
underlying numerical hydrodynamic modeling functions programmed into the ADCIRC
+ SWAN models, the strong correlation of CCD functions in predicting the likelihood of
damage based on actual storms suggests that the following relationships in equations (5)
and (6) are true for Hurricane Ike and the modeled wave steepness along CR 257.
[Tanh(𝑘ℎ) + 𝜖]~
2
𝑔~ 0.2 (5)
[𝑔
2[tanh[𝑘ℎ] + 𝜖]]~1 (6)
Wave period 𝐶𝐶𝐷(𝑇𝑝) functions for Hurricane Katrina and US 90 result in
slightly different relationships shown in equations (7) and (8). Since tanh[𝑘ℎ] → 0 in
shallow water and 𝜖 ~ 𝑘𝑎, then substituting and solving for approximate wave amplitude,
𝑎 ~ (𝜆/𝑔𝜋) for Hurricane Ike and 𝑎 ~ (𝜆/𝑔) for Hurricane Katrina. Expressed as a
relationship between wave length λ and wave height H, 𝜆/𝐻~ (𝜋)(𝑔/2)~ 15.4 for
Hurricane Ike and 𝜆/𝐻~ (𝑔/2)~ 4.9 for Hurricane Katrina, with 𝜋 being the
110
differentiator. The underlying hydrodynamic significance and validation of these
relationships warrants additional investigation through future research.
[tanh(𝑘ℎ) + 𝜖]~
2𝜋
𝑔~ 0.64 (7)
[𝑔
2[𝑡𝑎𝑛ℎ[𝑘ℎ] + 𝜖]]~ 𝜋 (8)
The CCD formulae are shown in Equations (9) and (10). Calculations for CCD
functions use either cumulative WSE or Tp incremental IM values with a resultant
coefficient of determination approaching 1. The variable Distance represents the
approximate measured distance between pre-event MLW and the road centerline. The
velocity term V, represents the cumulative water velocity IMs overtopping the pre-event
road grade elevation at that node.
Velocity term aggregates absolute IM scalar values at the node regardless of
direction. The “velocity head” term relative to road distance from shoreline represents a
dimensionless kinetic energy gradient important for predicting the relative degree of
damage at those locations. A first-order equation reasonably describes the correlation,
with a variable coefficient depending on whether celerity includes both WSE and wave
height as shown in Equation (2).
111
𝐶𝐶𝐷(𝑊𝑆𝐸) =
[
[ ∑𝑉
2
2𝑔
𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒
]
[√𝑔 ∑√ℎ + 𝐻𝑚𝑜]
]
(m/s) (9)
𝐶𝐶𝐷(𝑇𝑝) =
[
[ ∑𝑉
2
2𝑔
𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒
]
[∑ 𝑇𝑝
𝜋[𝑔
2[tanh[𝑘ℎ] + 𝜖]]]
]
(m/s) (10)
Figure 28 uniquely shows that cumulative celerity computed at each distinct road
centerline location using wave period strongly correlates to cumulative shallow water
celerity dispersion values computed using WSE and wave height. The equation slope
varies depending on whether data evaluation includes both WSE and Hmo. Aggregation
filters out incremental IM data when wave height is less than 0.3 meters or in the absence
of road overtopping.
Though strongly correlated with damage data points clustered at higher values,
damage data lack separation from no damage points. Linear trendline shows flatter slope
for damage data than no damage data, indicating that at lower values of cumulative
celerity, frequency dispersion attributed to storm surge and wave height exceeds that
attributed to wave period. As cumulative values increase and slope flattens, longer wave
periods during the more intense storm events increase 𝐶𝐶𝐷(𝑇𝑝) relative to 𝐶𝐶𝐷(𝑊𝑆𝐸),
along with resulting damage levels.
112
Figure 28. CR 257 cumulative celerity data for 67 data points extracted from Hurricane
Ike model.
Figure 29 incorporates distance from road to shoreline measured at mean low
water into the cumulative celerity function. This results in a gradient celerity measure
relative to unit distance from shoreline, but does not clearly separate damage and no
damage values. Several damage node locations along CR 257 that are setback farther
away from the shoreline plot have smaller values due to the increased setback distance.
113
This formula does not account for the kinetic energy and sediment scour/transport
component created by the water velocity.
Figure 29. CR 257 cumulative celerity data relative to distance from road to MLW
shoreline for 67 data points extracted from Hurricane Ike model.
Figure 30 accounts for cumulative velocity head relative to distance, effectively
accounting in addition to the cumulative celerity dispersion values for cumulative kinetic
energy unit gradient between shoreline and road over the storm duration. This plot shows
114
that damage points clearly separate from no damage points and the value of the CCD
function clearly distinguishes degree of damage as well. The order of damage points
strongly correlates to observed and documented damage along CR 257.
Figure 30. CR 257 cumulative celerity dispersion (CCD) function with data relative to
hydraulic gradient for 67 data points extracted from Hurricane Ike model.
Figure 31 shows the heat map representation of CCD function values for CR 257
in Brazoria County. Heat maps include data for the additional 30 test node locations used
to verify the semi-empirical tool results from the original 67 points. The values for CCD
115
functions computed using WSE or Tp incremental IM values are relatively equal. Figure
32 shows a map developed by Coast & Harbor Engineering [55] showing the relative
extents of damage along the same reach of CR 257. The CCD values in Figure 31
strongly correlate to damage locations, levels, and extents recorded post-event as shown
in Figure 32 [55].
Figure 31. Heat map showing relative damage mapped from CCD function values along
CR 257 on Follet's Island in Brazoria County, TX (Bing Map Imagery, 2020).
Figure 32. Post-Ike damage summary showing levels and extents of damage as mapped
by Coast & Harbor Engineering [55].
Figure 33 shows an enlarged aerial image of the most significantly damaged
extents recorded along CR 257 post-Ike. Values of CCD functions shown in Table 1 also
show the percent difference between CCD functions computed using both WSE and Tp.
116
Figure 33. Enlarged aerial image showing most significantly damaged locations along CR
257 with point numbers in Table 1 (Google Earth Imagery 9/13/2008).
Table 1. CCD values for select CR 257 locations corresponding to image shown in Figure
33 (ordered from east to west).
Figure 34 shows the relative damage for an extended reach of CR 257 with node
colors representing relative damage states for each node. These are confirmed by the
post-event records summarily shown in Figure 32 [55, 56].
Point # CCD(WSE) CCD(T p ) % Diff.
4.05 138 109 21%
4.06 224 236 -6%
1.21 221 190 14%
4.07 303 336 -11%
4.08 419 412 2%
1.22 221 190 14%
117
Figure 34. Aerial images showing damage states for points along east end of CR 257 on
Follet’s Island (Google Earth Imagery 9/13/2008).
US 90 from Bay St. Louis to Biloxi Bay experienced much less damage than CR
257 with little or no records to document damage extents other than aerial images
showing overlaid sections post-event. The event produced deeper overtopping water
surface elevations of shorter duration. Hurricane Ike generated longer wave periods than
Hurricane Katrina. Resultant CCD values shown in the Figure 35 heat map reasonably
correspond to post-event aerial images showing reconstructed pavement sections.
Damage along US 90 significantly impacted the west end and the Pass Christian area.
118
Figure 35. Heat map showing relative damage mapped from CCD function values along
US 90 from Bay St. Louis to Biloxi Bay in Harrison County, MS (Bing Map Imagery,
2020).
Figure 36 shows CCD data for both Hurricane Ike and Katrina. Data includes all
data points for both hurricanes, including the 30 added data verification points along CR
257. Data show strong correlation of the CCD function for all data values from both
events. CCD data trendlines show steeper gradients at smaller values than at larger
values. This shows that wave period celerity dispersion becomes more significant for
intense storms of greater intensity and longer duration in producing catastrophic road
damage, relative to shallow water dispersion due to depth of overtopping flow. The
damage along US 90 was minor relative to CR 257.
119
Figure 36. CR 257 and US 90 cumulative celerity dispersion (CCD) function with data
relative to hydraulic gradient for data points extracted from Hurricane Ike and Katrina
models.
6.3.4 Pseudo-Froude Functions
Continuum mechanics and nearshore hydrodynamics provide foundational theory
for the CCD function basis as a cumulative pseudo-Froude function. Equation (11)
represents the function for computing the dimensionless cumulative 𝐹𝑟𝑊𝑆𝐸 value.
120
𝐹𝑟𝑊𝑆𝐸 = ∑
𝑉
√𝑔(ℎ + 𝐻𝑚𝑜) (11)
Equation (12) shows the simplified function used for computing the
dimensionless cumulative 𝐹𝑟𝑇𝑝 value for Hurricane Ike. Function for 𝐹𝑟𝑇𝑝
applied to
Hurricane Katrina shown in Equation (13) excludes π like the 𝐶𝐶𝐷(𝑇𝑝) function.
𝐹𝑟𝑇𝑝
= ∑𝑉
(𝑇𝑝
𝜋)
(12)
𝐹𝑟𝑇𝑝= ∑
𝑉
𝑇𝑝 (13)
Figure 37 shows pseudo-Froude data for both Hurricane Ike and Katrina. Data
includes all data points, except for the 30 additional data verification points for CR 257.
Plot shows same continuity evident with the CCD function in Figure 36 for very different
road locations and storm events. The coefficient of determination remains very strong for
the pseudo-Froude function, but data scatter increases. The pseudo-Froude function does
not square the velocity term and excludes the distance variable. Additional research
requires further evaluation and understanding of the pseudo-Froude function relative to
predicting the likelihood and degree of damage states.
121
Figure 37. CR 257 and US 90 cumulative Froude function for data points extracted from
Hurricane Ike and Katrina models.
6.4 Assessing System Reliability
A model or tool that strongly predicts the likelihood and degree of damage for
coastal roads impacted by extreme storm events beneficially provides opportunity to
mitigate damage by modifying system components where feasible to improve reliability
and improve resiliency.
122
6.4.1 CCD Function
Research finds that the likelihood and degree of damage caused by a significant
coastal storm event is primarily a function of the cumulative energy dispersed at a coastal
road node using incremental hourly IM data extracted from a coastal storm model. While
component failure mechanisms likely affect resiliency of the coastal road in determining
degree of damage, these component reliability functions appear secondary to energy
dispersion and sediment transport as causal factors based on CCD functions.
Model analysis primarily treated road damage as granular binary data with failed
damage state including partial or complete damage to the road pavement section such that
its continued use for access results in unacceptable risk due to the nature of the hazard.
The CCD function suggests that the likelihood of extreme damage is highly predictable
based on a limit state value. Coastal road failures demonstrate that extended duration of
an extreme event likely increases the probability of failure. Since data analysis supports a
limit state, a short duration of extreme intensity could similarly reach a failed state if the
aggregate IM values evaluated by the CCD function exceeds the critical limit state.
Time integrated data from IMs for CR 257 indicate threshold damage state values
typically occur after velocity vectors redirect storm surge in a southerly and westerly
direction, indicating ebb current conditions as storm surge abates and winds redirect after
the hurricane moves onshore (Figure 38). Failure mechanism appears to be partially
attributable to weir overflow during storm surge ebb flow.
123
Figure 38. Average wave height to stillwater depth ratio and water velocity direction
changes at the threshold failed limit state for the CCD function for Hurricane Ike storm
event and CR 257 damage.
A critical failure mechanism appears to be dependent on the interaction of strong
ebb currents backflowing over the slab at critical depth, interacting with the irregular sea
state with depth-limited breaking plunging waves eccentrically loading the slab and
ejecting underlying base material with a bellows effect. Flood currents, as storm surge
moves landward, elongate wave lengths; whereas, ebb currents, as storm surge retreats
seaward, reduce wave lengths and steepen wave breaking. This likely failure mechanism
requires testing and validation in future wave flume experiments.
124
6.4.2 Pseudo-Froude Function
The pseudo-Froude function supports the underlying continuum mechanics and
nearshore hydrodynamics that provide the basis of the CCD function. Discrete analysis of
pseudo-Froude function values for incremental IMs contradict the likelihood of failure
from short-term peak or extreme hydrodynamic values. While the pseudo-Froude
function strongly correlates with linear trendlines like CCD functions, the likelihood and
degree of damage lacks the definition provided with the CCD functions. Because of the
wave-current interaction during ebb flow as a likely key failure mechanism, application
of the pseudo-Froude function requires additional research.
6.4.3 Application of Predictive Functions
Evaluating data from other storms with different annual exceedance probability
(AEP) or recurrence intervals confirms that the impact of a storm along a coastal
shoreline does not result in a uniform storm intensity. By evaluating the gamma
distribution for multiple event data at each point, the AEP for the data points along
Follet’s Island and CR 257 showed probabilities ranging from about a 230-year event to
approximately a 330-year event in the most heavily damaged locations, with an average
287-year event for the study limits. This varying AEP distribution becomes critical when
evaluating the likelihood and degree of potential damage during a future event.
As coastal storm risk models continue to aggregate storm data in developing
synthetic gridded data for AEP models, modeling of data to assess risk and reliability for
existing features will ultimately require simply extracting data from a geospatial data set
for an area of interest. The CCD function advances the ability to then plan and design a
resilient coastal road system, because adaptable system variables include only setback
125
distance and elevation. Coastal or road defenses designed to mitigate the risk from
extreme events are challenging in design and typically not cost effective to construct.
Elevating the entire road system above the critical water surface elevation
produces the undesired results of blocking flow to and from back bays, consequentially
increasing beach, and dune erosion. Accommodating overwash of flood and ebb flow at
strategic locations with lower road elevations at greater setback distances likely mitigates
risk of catastrophic damage by keeping the CCD function below the critical ‘failed’ limit
state, as evidenced along CR 257 at multiple ‘no-damage’ locations. This provides
opportunity to improve resiliency and reliability within coastal transportation system.
6.5 Conclusions
Development of fragility functions requires fully understanding the multivariate
forcings and system load responses for coastal road systems. This includes both
environmental siting and shoreline/dune/road protective measures. US 90 in Harrison
County, MS benefits from the extensive largely buried stepped seawall system, wide
median section, and incidental pavement features, such as curbs and concrete barriers.
These features modify flow depths and wave-current interactions, effectively modifying
wave breaking characteristics and decreasing damage at pavement sections.
Research using physical models are necessary to fully understand the role of
cumulative celerity dispersion in effectively determining the likelihood and degree of
potential damage to critical coastal road systems resulting from extreme storm events.
The strong correlations of CCD functions reinforce the need to verify and validate semi-
empirical functions that depend on coastal model incremental IM output data over the
126
storm’s duration. By verifying consistency in application across disparate locations and
storm events, future research needs include verifying applicability off CCD functions at
other coastal road locations for other named storm events, such as Hurricane Michael.
Additional research must confirm the failure mechanisms suggested by these data
and the time integration of the CCD function relative to critical threshold values. The
consistency of exceeding critical threshold values when water velocities change flow
direction from a flood flow to an ebb flow vector orientation, suggests that wave-current
interactions at the road section requires physical modeling to understand the forcings.
The end use of these findings includes coding and integration into fragility
functions for use in the IN-CORE model. Additional data refinement and physical
modeling will allow development of fragility functions using the CCD functions. The
importance of the CCD function in predicting the likelihood and degree of failure is not
insignificant; neither is the strong correlation between the cumulative celerity functions
evaluated using either WSE or Tp incremental IM values. It provides opportunity as
climate change and relative sea level rise impacts increase, to evolve Bayesian CCD
functions in future research, which will likely prove practical and invaluable for
evaluating these complex geospatially integrated coastal infrastructure hazard models.
CHAPTER VII
ADAPTING TO CHANGE
7.1. Key Findings
Research intended to develop a reliable method of quantifying coastal road
system architecture and fragility functions developed for subsequent use with the IN-
CORE all-hazards model. This research included participation since 2016 with the Rice
University team with Dr. Jamie Padgett as a Principal Investigator (PI). Research
developed fragility curves using peak IMs for coastal roads and bridges (research paper
pending publication). Research partners have included Dr. Ioannis Gidaris [108] and Dr.
Yousef Mohammadi Darestani. Programming of the IN-CORE model currently includes
Tier 1 road and bridge fragility models, with Tier 2 research effort recently authorized
and active under the continued leadership of Dr. Bret Webb in his role as a PI.
The CCD function presented in the body of this work represented by portfolio
provides a strongly correlated semi-empirical tool for predicting the likelihood and
degree of coastal road damage related to coastal hydrodynamics, which has potential of
reducing epistemic uncertainty by identifying relationships not previously described.
While much work remains, the analysis presented provides sufficient evidence for
continued investigatory work by application to other events with detailed records of road
damage, such as Hurricane Michael’s impact on Mexico Beach and Port St. Joe, FL.
128
7.1.1 Transdisciplinary Research Benefits
Research has benefited from the transdisciplinary approach pursued in this body
of work. Transdisciplinary systems thinking considers various measures of adaptability
and integration in assessing system sustainability, unlike independent disciplinary
approaches. The IISE 2020 conference paper included in Chapter IV focuses on the
complexities and benefits of pursuing a systems engineering research effort using a
transdisciplinary research team approach. Findings presented in this work are based on
systems, environmental, coastal, and diverse civil engineering knowledge and experience
acquired through 40-years of advanced study and applied engineering experience.
Applying systems engineering to assessing risk and predicting reliability of
engineered coastal systems has been an active and growing area of research for several
decades. Applying system engineering theories in quantifying uncertainties associated
with multivariate probabilities, to engineered systems with multiple or not-mutually
exclusive failure modes, engages mathematicians, physicists, scientists, and engineers
alike in pursuit of developing models that resolve all uncertainties. This potentially
creates additional risk in that consistent application or misapplication of probabilistic
methods or models can potentially limit identifying and resolving system failures.
Initial evaluation assumed that likely damage mechanisms were attributable to
weir overtopping and toe scour; rapid drawdown affecting slope stability; saturated soil
seepage undermining pavement section; rigid/flexible pavement structural resiliency;
dunes failing to reduce wave energy; increased sediment transport and flux; and, extreme
beach scouring in response to storm events. Additional complexity included coastal
defenses along the shoreline/road, dunes, and property developments.
129
While each of these system components impact coastal road resiliency in
determining degree of damage, these component reliability functions appear secondary to
energy dispersion as a causal factor based on CCD functions. Functional decomposition
assisted with assessing primary failure mechanisms and evaluating modifications that
potentially reduce failure probability. The CCD function strong correlation in predicting
likelihood and degree of significant damage clearly shows that a failure mode and effects
analysis (FMEA) can expand this body of work, if supported by physical modeling to
quantify failure mechanics and assess criticality of components.
7.1.2 Coastal Transportation Systems Reimagined
Body of work included modeling coastal highway systems in systems engineering
course practicums [15-17, 126]. Paper presented at IISE 2018 Orlando, FL conference
included in Chapter III won Best Sustainability Division Conference Paper award [1]. It
makes the case that system engineers should approach coastal systems analysis from the
perspective of integrating local desirements with functional and practical system
requirements for improving coastal infrastructure sustainability and resiliency. Increased
physical and economic damages impacting coastal roads, require reimagining
fundamental system requirements for local coastal roads and associated infrastructure.
Coastal road systems are a vital transportation infrastructure asset within coastal
communities and progressively challenged relative to resiliency and sustainability.
Increasing demand on coastal infrastructure with greater coastal development densities
requires providing coastal road systems that meet changing needs. There are changing
dynamics due to next-generation connectivity desirements, increasing coastal hazards,
and natural and nature-based features incorporated for risk mitigation. Climate change
130
adaptation and disaster risk management require understanding changing requirements
that place roads and communities at risk.
Physical and economic damages require addressing adaptive system requirements
for coastal road infrastructure. Systems analysis allow engineers to reimagine
connectivity and environments associated with coastal roads to meet next-generation’s
needs. Changing conditions make systems analysis, planning, siting, and architecture
tools essential for resilient and sustainable design of coastal road systems.
Engineering for function is essential; designing a resilient coastal road corridor
for connectivity is excellence. HDR’s TRANSCON: Create 2030 internal Omaha, NE
transportation conference held in April 2019 included two technical sessions with a
tailored presentation entitled Coastal Roads: Next-Gen Challenges. Presentation included
a discussion regarding the changing paradigm to data driven design utilizing big data,
such as integrated coastal and transportation network models, in computational designs
by applying machine learning and artificial intelligence to optimize system design data.
7.1.3 Cumulative Energy Dispersion Determines Failure
This body of work makes the case that CCD functions strongly predict the
likelihood and degree of coastal road damage when impacted by a major storm event.
The underlying hydromechanics prove challenging since it presumes that frequency
dispersion through wave phase velocity or celerity in nearshore shallow water depths;
however, shallow water is not frequency dispersive since shallow water wave celerity is
determined by depth, and not by wave period.
The coupled coastal models used to generate incremental IMs for this research
assume a fixed terrain model that does not change with time. It does not consider that as a
131
road is damaged and eventually breached, that the elevation of the road or scour hole at
that breach may change by 3 meters or more over the duration of the storm. The CCD
functions also do not account for materials of construction; protective measures; design
criteria; or, other environmental/infrastructure system characteristics.
A photograph taken along CR 257 shown in Figure 39 illustrates the complexity
of evaluating such a sensitivity. It also illustrates the significance of the CCD function in
predicting the likelihood of failure along CR 257 because of the varying depths of
pavement. It suggests that the likelihood of damage is primarily determined by
cumulative wave and current energy impacting the road for an extended storm duration.
Figure 39. Photo taken at unidentified location along CR 257 showing multiple asphaltic
pavement and base course layers in the failed section (Brazoria County, 9/15/08).
132
Because of these random variables and uncertainties, the CCD functions are
particularly significant and important for continued research, since the highly refined
correlation to observed damage and representation of the relative degree of damage
supports the validity of the semi-empirical tool. Figure 40 shows all data points evaluated
in this research effort and the strong coefficient of determination between CCD evaluated
using both WSE and Tp. Note that by excluding the wave height component from
CCD(WSE), correlation between the two CCD functions approaches unity.
Figure 40. CCD function for all data points and all AEP storm events evaluated including
Hurricanes Ike and Katrina.
133
The correlation shown in Figure 40 represents 185 different data points at
different locations evaluated for different storm events of varying intensities. Each data
point represents a unique and independent location with unique IM data sets. Initial
findings included presentations at multiple conferences with a peer reviewed technical
journal publication pending (Chapter VI).
Presentations included the Predicting Coastal Roadway Damage using Modified
Dispersion Functions speaker presentation at the ICCE 2018 conference in Baltimore.
Poster presentation at the April 2018 NIST research project team meeting at CSU in Fort
Collins, CO, won the first-place award for doctoral and post-doctoral poster submittals.
Presentations included Using Failure to Improve Resiliency speaker presentation at the
ASBPA October 2018 conference meeting in Galveston, TX. Poster presentations
included Using Coastal Road Failures to Improve Resiliency at the TRB Transportation
November 2019 Resilience Conference in Washington, DC.
Pending presentations include Transdisciplinary Systems Thinking: Sustainability
of Coastal Systems (Chapter IV) and Coastal Road System Failures: Cause and Effect
(Chapter V) at the IISE 2020 Virtual Conference in October, 2020. The latter paper was
awarded the Best Construction Division Track Conference Paper for IISE 2020.
7.1.4 Proof of Concept
Research demonstrates that the probability and likelihood of coastal road failure is
strongly correlated to cumulative overtopping water level, wave height, and wave period
CCD functions, relative to cumulative water velocity kinetic head and distance from
shoreline, and critical threshold cumulative values delineate the likelihood and degree of
relative damage states. Furthermore, since nearshore coastal hydrodynamics as
134
evidenced by the pseudo-Froude function provides a basis for the CCD function, failure
mode and effects analysis (FMEA) potentially improves system reliability. Figure 41
shows the gamma distribution for CCD functions computed using both WSE and Tp for
Hurricane Ike and CR 257 IM data sets.
Figure 41. Hurricane Ike CCD function cumulative gamma distribution.
135
Figure 42 shows Ike’s AEP (287-year average) along CR 257 computed using the
other AEP storm data and a gamma distribution at each point location. The AEP
probabilities at each location for a CCD limit state value of 37 are also plotted (reference
Figure 38). Plot shows that significantly damaged sections (red text near x-axis)
correspond to the sections where Ike’s localized storm intensity (dashed line) exceeds the
limit state probability (solid blue). The plot shows that differential probabilities (red line)
strongly predict both the location and degree of recorded damaged sections. Plot also
illustrates the complexity associated with evaluating likelihood of failure pre-event, since
the storm does not impact the road with a uniform design storm intensity.
Figure 42. Hurricane Ike CCD(WSE(h)) estimated probability of exceeding damage limit
state relative to distance along Follet's Island (plotted from east to west).
136
7.2 System Adaptation
As uncertainty of the future increases with climate change effects increasingly
evident along coastlines, uncertainties increase correspondingly making future adaptation
plans unreliable. Changing coastal transportation system risks include relative sea level
rise; increasing storm intensity and frequency of extreme events; loss of wetlands,
beaches, barrier islands, and wave attenuation marshes; and, increased coastal
development with shoreline encroachment, particularly in third-world countries.
Sustainable coastal road system challenges include damages and recurring
recovery costs; conflicting stakeholder interests; problem with maintaining the status quo
for coastal stakeholders; and, increased population and property vulnerabilities. The
changing climate forces a changing paradigm; changing paradigm promulgates changing
regulatory and design requirements; and, increased regulatory pressure and design
restrictions stimulates an adverse reaction among the coastal inhabitants, encouraging an
inevitable cultural transformation.
Figure 43. Sea change drivers in local coastal road system planning and design.
137
7.2.1 Changing the Likelihood of Failure
Proximity between shoreline and road increases the cumulative value of the CCD
functions and increases the likelihood of failure, confirming previous research findings
by others. Cumulative velocity is also critical to failure since it provides the forcing for
scour and sediment transport. It creates a strong kinetic energy gradient when the road
overtops. Higher velocities typically result with ebb flow when high water elevations
decrease as the storm center moves inland.
As flow depth increases with road overtopping, water velocity typically decreases
with increased conveyance section. The occurrence of critical wave breaking depths also
decreases, as still water depth increases relative to the wave heights. Some sections of CR
257 and US 90 obviously benefit from lower road grades that allow sand drifts to deposit
across the road surface during a major storm event. While this creates a cleanup nuisance
and temporary traffic hazard during the storm recovery period, it can also be beneficial in
reducing the final CCD value and likelihood of damage and consequential repair costs.
The WSE used in determining the resultant CCD value is relative to zero datum
and not road grade. CCD values do not increase with lower road elevations. The shorter
duration event decreased CCD values significantly for US 90 and Hurricane Katrina;
however, the increased depth of WSE also decreased IM velocities along much of the
road. The largest CCD values for US 90 occur where water velocities increase with
longshore flow orienting parallel to and along the road alignment. By identifying the key
variables that determine the likelihood of failure, some road features can be adapted in
design to reduce system vulnerability and to improve reliability.
138
Determining the optimal road elevation at a given location by evaluating the CCD
function over a range of extreme storm events, provides opportunity to minimize damage
risks for hurricane hazards. Elevating the entire roadway along a barrier island restricts
overtopping flow, causing unintended damage to other built and natural systems.
Lowering road elevations farther away from the shoreline facilitates overtopping flow at
depths and velocities that deposit a sand layer that protects, but does not damage, the
pavement structure.
A FHWA Transportation Engineering Approaches to Climate Resiliency
(TEACR) project study was executed for the Barrier Island Roadway Overwashing from
Sea Level Rise and Storm Surge: US 98 on Okaloosa Island, Florida [14]. For that event,
the overwash was inland and the damage was to the landward side of the road due to
weir-flow damage as overtopping occurred with flow into Choctawhatchee Bay from the
Gulf of Mexico during storm surge. Whether the CCD function predicts likelihood of
damage due to extended duration of overtopping currents but without significant wave
energy warrants investigation in future research. That study included an adaptation
decision matrix for use when evaluating coastal road system failures.
7.2.2 Impact of Climate Change and Sea Level Rise
Fragility surfaces developed using CCD functions would likely change
(qualitatively) as a function of climate change (i.e., relative sea level rise), beach
nourishment, or other adaptive management measures. Change adaptation would require
quantifying system modifiers using a Bayesian network to model the logic and
dependencies between the system variables that define the system.
139
Effectively this adaptation generates a causal relationship like the one presented
in Figure 44. Bayesian methods update the probabilities gained with new information
relative to the impact on dependent variables created by relative sea level rise; damage
mitigation measures (seawalls, living shoreline features, dunes, armoring, beach
nourishment, etc.). New data modify the probability density function (PDF), which then
propagates the impact of changing the hazard or survivability function as a perturbation
that reflects throughout the Bayesian network. In order to incorporate the impacts of
climate change and other external modifiers of the system response, change metrics
integrate into the reliability function. The Bayesian approach establishes an efficient
network, to evolve the model as additional data become available.
There are alternate methods for approaching modeling and simulation of
interdependent critical infrastructure systems, including modeling how reliability varies
with changes in stressors [127]. The USACE developed risk assessment methods to
verify that project and system have an acceptable level of risk for the anticipated hazards.
Since natural disasters are more frequent, unpredictable, and consequentially more
severe, resilience (in addition to risk management) quickly has become a priority for
federal agencies. The International Risk Governance Council (IRGC) Resource Guide on
Resilience provides online guidance. For Tier 3 analysis, potential tools for assessing
changes in resiliency metrics include Bayes nets, agent-based modeling, network science,
or system dynamics.
Potential variabilities and nature-based adaptive management measures are
addressed in a recent FHWA white paper [39]. Engineer Research and Development
Center (ERDC) and Institute for Water Resources (IWR) recommends a more qualitative
140
methodology in developing metrics for assessing vulnerability to coastal storms subject
to the impacts of climate change. Coastal-forcing metrics in terms of nearshore forcings
and other measures are considered in a methodological approach to assess vulnerability
[9]. Previous assessments include impacts of climate change on Gulf Coast transportation
systems [5, 128, 129].
7.2.3 Integrating Natural System Defenses
From a qualitative approach, relative sea level rise significantly impacts the
continued viability of barrier islands and accelerates beach transformations, including
loss of coastlines in many locations. The impact of climate change on resiliency of
coastal roads due to extreme events appears to be less due to RSLR than to storm
magnitudes, frequency, and duration. Nature-based solutions propose to reduce the
transformation rate by providing buffering features that force wave breaking before
reaching the shoreline. This decreases the dispersive energy reaching a coastal road
pavement section.
The CCD function suggests while natural and nature-based features (NNBF)
potentially delay likely failure, solutions must significantly absorb and disperse energy
effectively without causing physical damage to infrastructure; otherwise, these features
are not likely to reduce the transformational geomorphological responses of climate
change long-term.
The CCD function suggests that the most likely methodology for modifying long-
term risk is to manage the coastal road alignment, both vertically and horizontally. For
sections of coastal road nearer the shoreline, roadway would elevate above the probable
storm surge elevation to mitigate risk of overtopping. To allow flood and ebb storm surge
141
to move landward or seaward without causing catastrophic damage, road alignment
would shift inland. At locations sufficient distance away from the shoreline, road profile
would lower, allowing surge to safely overtop without causing catastrophic damage.
Alternatively, crossings might utilize low-level bridge crossings. At low-level road
crossings, shoulders require armoring to stabilize road sections for scour protection.
The CCD function suggests that a model that establishes the PDF for the variables
on a 3D axis, such that the combined probabilities determine what the likelihood of
failure is relative to shoreline offset horizontally, and profile relative to storm surge
elevation vertically. These type of sinuous quasi-embayments formed by altering the
horizontal-vertical alignment of coastal roads would likely be successful in dissipating
storm surge energy as suggested by small-scale breakwaters for Pocket Beach in
Yorktown, VA [39]. Storm surge dissipation in a sinuous small embayment created with
storm surge is very effective at reducing storm surge wave energy through diffraction and
refraction.
7.3 Continued Research Needs
Research needs associated with continued development of the IN-CORE model
necessarily requires continued coordination with the universities included in that
monumental research effort. The directors, principle investigators, and graduate students
involved in this effort have made this research effort a truly rewarding and memorable
experience. The transdisciplinary nature of the team represents the future of research
where boundaries and disciplines are routinely blurred, for this is the environment in
which major advances in quantifying problems and developing solutions dwell. An
142
inevitable key partner absolutely required for that effort is the systems engineer who
purposefully works to span that gap by engaging the other disciplines.
This research opens the door to continued investigation because it validates many
of the previous discussions by Hedges, Dalrymple, Kirby, et al. regarding celerity
dispersion equations and epistemic uncertainties in the turbulent wave breaking
environment. The hope is that additional research will better explain a very complex
system of energy and wave frequency dispersion during extreme conditions, especially
when shallow water by theory is not dispersive. While highly refined numerical coastal
models producing validated simulations of real-world storms are evident in the IM data;
simplified CCD functions predicting likelihood and degree of damage for a complex road
system with relative accuracy hopefully stimulates additional research and discussion.
7.3.1 Refining Causal Loop Fragility Functions
The causal loop shown in Figure 44 potentially provides a tool to quickly assess
the likelihood of failure based on input data. This basic model uses Vensim software by
Ventana Systems, Inc., which a systems dynamic stochastic modeling tool. These visual
modeling tools facilitates simulations by assembling components with probabilistic
stimulus-response fragility functions. As input data into the system vary, output
responses dynamically adjust using the systems modeling functions.
143
Figure 44. Coastal road damage CCD function causal loop diagram [17].
The diagram illustrates at a simplified level the weaknesses in applying the CCD
functions at a broad level since a complex coastal storm model generates the IM data. As
storm simulations output data aggregate within the ERDC Coastal Hazards System
(CHS) database18, the Coastal Storm Modeling System (CSTORM-MS) models managed
by the Coastal Hydraulics Laboratory (CHL)19, will provide much of these data. These
stored metadata from thousands of simulations will provide opportunity to develop
gridded assessments of the CCD function as a geospatial overlay in assessing the
18 https://chswebtool.erdc.dren.mil/ 19 https://www.erdc.usace.army.mil/Locations/CHL/
144
likelihood of failure relative to damage limits states for coastal road systems. The slider
functions prove to be the most challenging as previously researched by Basco and
Mahmoudpour in association with proposed COSI functions [106].
The causal loop also shows the benefits of applying Bayesian methods to
advancing understanding of the underlying coastal hydromechanics associated with the
CCD functions. The simulation model shows that as stimulus-response data evolve, the
model can easily be adapted to reflect the revised data using a Bayes model [23, 24].
7.3.2 Experimental Data to Evaluate Failure Functions
Understanding the failure functions requires a physical model and experimental
data to characterize the forcings. While initially proposed for inclusion in this research
effort, the overly ambitious plan scaled back to focus on verifying the CCD functions
using numerical model data. Proposed experiment should evaluate the response of a scale
model slab to forcings created by onshore waves interacting with strong offshore ebb
currents at water depths where wave breaking occurs at the slab. Typical section shown
in Figure 45 provides scaled approximation of typical beach shoreline along CR 257 if
experimental setup utilizes the wave flume at South Alabama.
Measuring instruments include a triaxial load cell mounted under the slab;
pressure sensors mounted above and under the slab; wave gauges in front and behind the
slab; velocity flow meter to record velocities over the slab; and a high-speed camera to
record images during peak forcings. Adding isopropyl alcohol to the circulated water will
enhance wave breaking characteristics since this is problematic in small-scale testing. A
pump will circulate water within the flume towards the wave paddle to simulate ebb
currents while the wave generator will create waves moving onshore to generate breaking
145
waves on the slab. Dampening features can be evaluated by adding systems to simulate
protective measures.
Figure 45. Conceptual layout of proposed wave flume modeled section.
Experimental setup is based on evaluating failure conditions predicted by the
CCD function. Expectations are that the slab will experience significant displacement
with resultant rapid pressure changes underneath during wave breaking. These pressure
forcings will include air bubbles that likely contribute to failure. These forces must be
significant to produce the type of failure evidenced along CR 257 and US 90 during
Hurricane Ike and Katrina respectively. Data analysis will evaluate force, wave, velocity,
and pressure measurements during peak events. Observing conditions during peak
forcings is the focus of this future research exercise. Observations will include generating
extensive data available for additional analysis.
Anticipated results include developing fragility functions for coastal road systems
with potential application to similar coastal structures. It is likely that the CCD function
146
similarly has application with modifications to predict failure at other locations for
different structure types or features. Fragility functions potentially incorporate natural
resonance frequency of the structure based on results of laboratory testing.
A series of pictures (Figure 46 - Figure 60) follow that were taken at the Chicago
Museum of Science and Industry wave tank demonstration exhibit with a scarp beach
profile. Dr. Dan Cox with Oregon State University (OSU) is a coastal engineer and
principle investigator on the IN-CORE team who helped design the exhibit. It generates
various wave forms to visually compare runup and wave breaking on two different beach
forms. The insert is fiberglass and the images are based on looking at the thin-film space
between the preformed model and the glass walls. There are some obvious differences
between comparing a fixed bed exhibit and live-bed forcings at a sand based coastal road
during a storm event, but there are some similarities as well.
The notations on the photos excerpted illustrate some potential failure
mechanisms observed while there. The photographs and notations suggest that there are
multiple failure mechanisms at work. The interaction of these mechanisms requires
laboratory testing to fully understand the various forcings. Since this research effort does
not include experimental testing, these records provide documentation of recorded
observations relevant to the forcings on the system of interest, which may be helpful in
future research.
147
Figure 46. CMSI wave tank showing wave response in scarp face.
Figure 47. CMSI wave tank showing wave response in scarp face and unwatering risks.
148
Figure 48. CMSI wave tank showing wave response in scarp face and scour potential.
Figure 49. CMSI wave tank showing wave response in scarp face and progression of
subsurface flow field.
149
Figure 50. CMSI wave tank showing wave response in scarp face and continued
progression of subsurface flow field.
Figure 51. CMSI wave tank showing wave response as flow field reaches air infused
layer.
150
Figure 52. CMSI wave tank showing wave response as air field migrating to scarp face
creating failure surface.
Figure 53. CMSI wave tank showing wave response as air field migrating to scarp face.
151
Figure 54. CMSI wave tank showing potential for rapid drawdown slope failure.
Figure 55. CMSI wave tank fully overtopping wave submergence creates failure surface
with air entrapment.
152
Figure 56. CMSI wave tank showing large wave break on scarp face entraining
significant air bubble.
Figure 57. CMSI wave tank showing wave response as flow field collapses bubble with
subsurface flow field.
153
Figure 58. CMSI wave tank showing ebb current conditions creating changes in
subsurface profile.
Figure 59. CMSI wave tank showing ebb current conditions creating rising air bubbles in
subsurface profile.
154
Figure 60. CMSI wave tank showing ebb current conditions as larger bubbles disperse
into finer bubbles.
Data points shown in Figure 46 were extracted using CAD for the wave and thin-
film water surfaces observed at the CMSI wave tank with points plotted in Figure 61.
Figure 62 shows the dimensionless representation of thin-film water surface response to
the surface wave strike. This failure phenomenon is representative of the real-world
response in a sandy beach scarp as evidenced by physical systems.
It shows that the phreatic subsurface response mirrors the surface waveform but
with less amplitude. The observed dynamic response is likely to cause failure at beach
scarps. The failure surface, bubble creation, bubble migration, bubble evolution, and
other forcings evidenced in the images likely contribute to, but do not generate, the
magnitude of forcings to produce the extent of damage observed on CR 257. The
downward wave force on the surface of the pavement exacerbates beach erosion but is
not likely to be the leading cause of pavement failure in terms of forcing.
155
Figure 61. Thin-film water surface response to wave strike on scarp surface as observed
at CMSI wave tank.
Figure 62. Scarp water surface response to wave strike on face (dimensionless with
surface wave in blue).
y = 0.0005x3 - 0.0006x2 + 0.0689x - 0.0005
R² = 1
y = 3E-06x6 - 0.0007x5 + 0.0288x4 - 0.485x3 + 4.1547x2 - 17.635x + 29.71R² = 1
y = 124.7x3 - 4367.5x2 + 50995x - 198497R² = 1
y = -0.0008x3 + 0.0444x2 - 0.8326x + 5.2529R² = 0.9998
-0.5
0
0.5
1
1.5
2
2.5
0 2 4 6 8 10 12 14 16 18 20
y = -0.8827x3 + 1.4458x2 + 0.4435xR² = 0.9998
y = 0.0012x3 + 0.0848x2 + 0.914x + 2E-14R² = 1
0
0.2
0.4
0.6
0.8
1
0.0 0.2 0.4 0.6 0.8 1.0
Series1
Series2
Poly. (Series1)
Poly. (Series2)
156
The collapsing air bubble likely amplifies pressure fluctuations in the subsurface
by transmitting shock waves into the subsurface through the porous asphalt pavement.
The impacting wave forces likely contribute to differential forces on a fractured
pavement section, causing uplift pressure and slab rotation that ejects subgrade material.
Thin-film observations along the glass walls of the wave flume setup shown in Figure 45
potentially provides opportunity to record responses to wave impacts using high-speed
imaging systems.
The CCD function suggest that the dynamic energy gradient slope created by the
ebb current velocity head relative to the distance of the road from the shoreline; and, the
dispersive energy created by wave breaking at the road location are the two primary
determinants and predictors of failure. The time integrated function demonstrates that
failure occurs with ebbing storm surge and depth-limited wave breaking conditions occur
at the coastal road location. Once the pavement breaks apart and exposes the sandy
embankment section underneath, ebb current scour becomes the likely dominant mode of
failure. The pavement surfacing acts as a form of armoring that when removed, facilitates
flow-induced scour conditions.
7.3.3 Integrating Systems Engineering into Coastal Road Reliability
While the causal loop in Figure 44 simplifies the relationship of failure identified
in the CCD function, it also illustrates the inherent need and benefits for engineers and
scientists associated with other specialty interests to integrate systems engineering
analysis and tools into applied research and design efforts. If the very complex system
does not simulate effectively at a macro-level scale, it is not likely to achieve desired
results when details model at the micro-level scale.
157
Systemigrams presented in Figure 20 and Figure 21 show the soft-systems
engineering diagrams related to failure modes. Coastal hazard system models potentially
range from less complex models for broader application of infrastructure resiliency
metrics to more complex models that facilitate detailed risk analysis using hybrid fragility
functions for individual components. Assessing which model best represents measured
data will be determined based on modeling results, initial analysis, and consultation with
other expert team members.
Improved systems analysis of coastal hazard systems with reduced uncertainties
potentially encourages greater use of systems application to risk-based design
methodologies. Developing a systems model for predicting coastal road failure provides
opportunity for developing functional solutions to reduce risk and improve reliability,
which effectively improves system resilience.
To improve resilience of coastal roads, risk reduction measures such as living
shorelines, seawalls, dunes, armoring, and other forms of coastal defenses require
integration into the system design. While resilience can be a systems property, total
resilience of a single project improves if considered as a subsystem of a composite
coastal projects system. Resilience effectively incorporates risk analysis into a framework
that uses adaptation and mitigation strategies to improve risk management [130].
Research supports application of systems engineering to an engineered coastal
road system exposed to extreme natural storm hazards. Research demonstrates that a
systems approach can identify critical mechanisms that cause complete or partial failure
of the engineered system. By determining critical modeling parameters with significant
correlation, nearshore hydrodynamics provide functions correlating probabilistic analysis
158
to wave mechanics in identifying variables that most significantly alter failure
probabilities.
While research benefits from understanding a transportation system’s functional
requirements associated with resilience, research must develop fragility functions based
on modeled storm data output, which strongly predict the likelihood of structural failure
that potentially limits recovery post-disaster. Reliable transportation systems are critical
to community robustness and recovery. Understanding which components are most likely
to fail and why, improves reliability of coastal road systems.
Emergent behavior requires understanding relationships between subordinate and
superior systems. The porosity or impenetrability of the system boundary requires
identification. This includes the ability of the system to absorb, adapt, and reconfigure to
changed inputs, as well as the ability of the system to release and reset where possible.
Depending on the system response, the system and subsystem properties, relationships
and processes are likely to change or evolve. Anticipating and leveraging the system
response is often a critical objective in assessing emergent behavior.
A passive infrastructure system such as a coastal road limits opportunity for
emergent behavior if the system boundaries do not extend beyond the immediate system
of interest. Even so, during a storm event, the potential to modify the system to produce a
different outcome are practically non-existent. A storm surge barrier that protects a
coastal transportation network is one example of an adaptive composite system that can
close in advance of an approaching storm to provide a measure of risk reduction for
properties within the barrier’s system boundaries.
159
Evaluating these response metrics in object-oriented systems modeling of a real-
world system becomes increasingly difficult by introducing these potential complexities.
GIS models increasingly introduce the opportunity to include a verticality component for
considering the responsiveness of aggregate systems. These introduce probability
functions as mapped shapefiles that can aggregate probabilities to assess risk and
resiliency for a sequence of events impacting a complex system. For an unsteady state
condition with extreme number of variables, such as a coastal road system response to a
major storm system, predicting the emergent behavior or system response relies on using
probabilistic models with a high-degree of uncertainty.
Fully understanding multiplicity of potential outcomes for emergence is likely the
most challenging task in considering coastal road system resiliency. The CCD function
relies on only a few cumulative state variables generated by a coastal model to predict the
likelihood and level of damage, which provides potential opportunity to reduce
complexity and uncertainties within in assessing the coastal road system damage
response. Integrating systems engineering into coastal road reliability functions requires
reimagining system functions for coastal road as a system, including the public-use
corridors and environmental settings in which they are located.
REFERENCES
160
REFERENCES
[1] G. P. Pennison, R. J. Cloutier, and B. M. Webb, "Local Coastal Roads-Next
Generation," in 2018 Institute of Industrial and Systems Engineers Annual
Conference and Expo, IISE 2018, Orlando, FL, 2018: Institute of Industrial and
Systems Engineers.
[2] H. Masoomi, J. W. van de Lindt, M. R. Ameri, T. Q. Do, and B. M. Webb,
"Combined wind-wave-surge hurricane-induced damage prediction for
buildings," Journal of Structural Engineering, vol. 145, no. 1, p. 04018227, 2018.
[3] J. Padgett et al., "Bridge Damage and Repair Costs from Hurricane Katrina,"
Journal of Bridge Engineering, vol. 13, no. 1, pp. 6-14, January 1 2008.
[4] PB, ICF, and SCE, "Gulf Coast Study, Phase 2—Task 1: Assessing Infrastructure
for Criticality in Mobile, Alabama," 2011.
[5] PB, ICF, and SCE, "Impacts of Climate Change and Variability on Transportation
Systems and Infrastructure: The Gulf Coast Study, Phase 2 Task 3.2: Engineering
Assessments of Climate Change Impacts and Adaptation Measures," 08/2014
2014.
[6] B. M. Ayyub, "Systems Resilience for Multihazard Environments: Definition,
Metrics and Valuation for Decision Making," Risk Analysis, vol. 34, no. 2, pp.
340-355, 2014.
161
[7] B. M. Ayyub, Risk analysis in engineering and economics. CRC Press, 2014.
[8] W. Wu, S.-C. R. Lo, and X. H. Wang, "An Objective Approach to Reliability-
Based Risk Assessment and Mitigation for Coastal Infrastructure Development,"
Labour and Management in Development, vol. Vol. II, pp. 0-16, 2011.
[9] T. V. Wamsley, Z. A. Collier, K. Brodie, L. M. Dunkin, D. Raff, and J. D. Rosati,
"Guidance for Developing Coastal Vulnerability Metrics," Journal of Coastal
Research, pp. 1521-1530, 2015.
[10] USACE, "North Atlantic Coast Comprehensive Study: Resilient Adaptation to
Increasing Risk," USACE, Washington, DC 2015.
[11] S. L. Douglass, B. M. Webb, and R. Kilgore. (2014). Highways in the Coastal
Environment: Assessing Extreme Events, Hydraulic Engineering Circular 25
(Volume 2), FHWA NHI-14-006.
[12] FHWA, "FHWA Climate Change Resilience Pilot Projects Peer Exchanges,"
Baltimore, MD July 16-17, 2014 2014.
[13] FHWA, "2013-2015 Climate Resilience Pilot Program: Outcomes, Lessons
Learned, and Recommendation," July, 2016 2016.
[14] FHWA, "TEACR Engineering Assessment: Barrier Island Roadway Overwashing
from Sea Level Rise and Storm Surge: US 98 on Okaloosa Island, Florida,"
09/2016 2016.
[15] G. Pennison, "SE601 Systems Engineering Fundamentals Practicum, Local
Coastal Roads-Next Generation," University of South Alabama Class
Assignment, Mobile, AL 2016.
162
[16] G. Pennison, "SE 606 Systems Architecture Practicum: Local Coastal Road
Model – SysML," University of South Alabama, Mobile, AL, Practicum
4/28/2017 2017.
[17] G. Pennison, "SE 610 - Systems Thinking Coastal Road System Resilience,"
University of South Alabama 6/24/2018 2018.
[18] M. Batouli and A. Mostafavi, "Multiagent Simulation for Complex Adaptive
Modeling of Road Infrastructure Resilience to Sea-Level Rise," Computer-Aided
Civil and Infrastructure Engineering, vol. 33, no. 5, pp. 393-410, 2018.
[19] D. J. Wainwright et al., "Moving from deterministic towards probabilistic coastal
hazard and risk assessment: Development of a modelling framework and
application to Narrabeen Beach, New South Wales, Australia," Coastal
Engineering, vol. 96, pp. 92-99, 2015/02/01/ 2015.
[20] D. P. Callaghan, R. Ranasinghe, and D. Roelvink, "Probabilistic estimation of
storm erosion using analytical, semi-empirical, and process based storm erosion
models," Coastal Engineering, vol. 82, pp. 64-75, 2013/12/01/ 2013.
[21] D. P. Callaghan and D. Wainwright, "The impact of various methods of wave
transfers from deep water to nearshore when determining extreme beach erosion,"
Coastal Engineering, vol. 74, pp. 50-58, 2013/04/01/ 2013.
[22] F. Li, P. van Gelder, J. Vrijling, D. Callaghan, R. Jongejan, and R. Ranasinghe,
"Probabilistic estimation of coastal dune erosion and recession by statistical
simulation of storm events," Applied Ocean Research, vol. 47, pp. 53-62, 2014.
163
[23] S. Narayan, D. Simmonds, R. J. Nicholls, and D. Clarke, "A Bayesian network
model for assessments of coastal inundation pathways and probabilities," Journal
of Flood Risk Management, vol. 11, pp. S233-S250, 2018.
[24] A. Couasnon, A. Sebastian, and O. Morales-Nápoles, "A Copula-Based Bayesian
Network for Modeling Compound Flood Hazard from Riverine and Coastal
Interactions at the Catchment Scale: An Application to the Houston Ship Channel,
Texas," Water, vol. 10, no. 9, p. 1190, 2018.
[25] G. Davies et al., "Improved treatment of non-stationary conditions and
uncertainties in probabilistic models of storm wave climate," Coastal
Engineering, vol. 127, pp. 1-19, 2017/09/01/ 2017.
[26] GFDRR, "Machine Learning Language for Disaster Risk Management,"
International Bank for Reconstruction and Development, Washington, DC2018.
[27] T. McAllister, C. Clavin, J. W. van de Lindt, B. Ellingwood, D. Mizzen, and F.
Lavelle, "Data, Information, and Tools Needed for Community Resilience
Planning and Decision-Making," 2019.
[28] NIST, "Critical assessment of lifeline system performance: Understanding
societal needs in disaster recovery," ed, 2016.
[29] P. B. Sayers et al., Risk, performance and uncertainty in flood and coastal
defence - a review. DEFRA/EA R&D Technical Report FD2302/TR1 (HR
Wallingford Report SR587) January, 2002.
[30] F. A. Buijs, J. Hall, P. Sayers, and P. Van Gelder, "Time-dependent reliability
analysis of flood defences," Reliability Engineering & System Safety, vol. 94, no.
12, pp. 1942-1953, 2009.
164
[31] A. Burzel, D. R. Dassanayake, and H. Oumeraci, "Spatial Modeling of Tangible
and Intangible Losses in Integrated Coastal Flood Risk Analysis," Coastal
Engineering Journal, vol. 57, no. 01, p. 1540008, 2015.
[32] T. S. Hopkins, D. Bailly, and J. Støttrup, "A systems approach framework for
coastal zones," Ecology and Society, vol. 16, no. 4, p. Art. 25, 2011.
[33] T. S. Hopkins, D. Bailly, R. Elmgren, G. Glegg, A. Sandberg, and J. G. Støttrup,
"A systems approach framework for the transition to sustainable development:
potential value based on coastal experiments," Ecology and Society, vol. 17, no. 3,
2012.
[34] A. Newton, "A systems approach for sustainable development in coastal zones,"
Ecology and Society, vol. 17, no. 3, p. 41, 2012.
[35] J. D. Rosati, K. F. Touzinsky, and W. J. Lillycrop, "Quantifying coastal system
resilience for the US Army Corps of Engineers," Environment Systems and
Decisions, vol. 35, no. 2, pp. 196-208, 2015.
[36] K. Touzinsky, J. Rosati, C. Fox-Lent, A. Becker, and A. Luscher, "Advancing
coastal systems resilience research: improving quantification tools through
community feedback," Shore & Beach, vol. 84, no. 4, p. 30, 2016.
[37] K. Bosma, E. Douglas, P. Kirshen, K. McArthur, S. Miller, and C. Watson,
"MassDOT-FHWA pilot project report: Climate change and extreme weather
vulnerability assessments and adaptation options for the Central Artery,"
Retrieved from Boston, MA, 2015.
165
[38] G. Filosa, A. Plovnick, L. Stahl, R. Miller, and D. H. Pickrell, "Vulnerability
Assessment and Adaptation Framework, Third Edition," (in English), Tech
Report 2017.
[39] B. M. Webb, S. L. Douglass, B. Dix, and S. Asam, "White Paper: Nature-Based
Solutions for Coastal Highway Resilience," ICF, South Coast Engineers 2018.
[40] L. V. Montoya, E. J. Sciaudone, E. Smyre, and M. F. Overton, "Spatiotemporal
vulnerability of coastal roadways based on morphological indicators and shoreline
change predictions," Natural Hazards Review, In Review 2020.
[41] K. A. Anarde et al., "Impacts of Hurricane Storm Surge on Infrastructure
Vulnerability for an Evolving Coastal Landscape," Natural Hazards Review, vol.
19, no. 1, 2018.
[42] S. L. Douglass and J. Krolak, Highways in the Coastal Environment, Hydraulic
Engineering Circular 25, Second Edition, FHWA NHI-07-096. Washington, D.C.:
Federal Highway Administration, 2008.
[43] P. E. Clopper, "Minimizing embankment damage during overtopping flow," 1988.
[44] G. R. Powledge, D. C. Ralston, P. Miller, Y. H. Chen, P. E. Clopper, and D. M.
Temple, "Mechanics of Overflow Erosion on Embankments. II: Hydraulic and
Design Considerations," Journal of Hydraulic Engineering, vol. 115, no. 8, pp.
1056-1075, 1989.
[45] P. Lubin and H. Chanson, "Are breaking waves, bores, surges and jumps the same
flow?," Environmental Fluid Mechanics, vol. 17, no. 1, pp. 47-77, 2017.
166
[46] T. Okamoto and D. R. Basco, "The Relative Trough Froude Number for initiation
of wave breaking: Theory, experiments and numerical model confirmation,"
Coastal Engineering, vol. 53, no. 8, pp. 675-690, 2006/06/01/ 2006.
[47] B. M. Webb, "Coastal Hazard Intensity Measures," in "2017_INCORE1-
CoastalHazardIntensityMeasures-040117-1521-2192," University of South
Alabama, Mobile, AL2017.
[48] T. E. E. Service, "Hurricane Ike Impact Report," 2009.
[49] A. Kennedy et al., "Origin of the Hurricane Ike Forerunner Surge," Geophysical
Research Letters - GEOPHYS RES LETT, vol. 38, 04/28 2011.
[50] M. E. Hope et al., "Hindcast and validation of Hurricane Ike (2008) waves,
forerunner, and storm surge," Journal of Geophysical Research: Oceans, vol. 118,
no. 9, pp. 4424-4460, 2013.
[51] P. Kerr et al., "US IOOS coastal and ocean modeling testbed: Evaluation of tide,
wave, and hurricane surge response sensitivities to mesh resolution and friction in
the Gulf of Mexico," Journal of Geophysical Research: Oceans, vol. 118, no. 9,
pp. 4633-4661, 2013.
[52] P. Kerr et al., "US IOOS coastal and ocean modeling testbed: Inter‐model
evaluation of tides, waves, and hurricane surge in the Gulf of Mexico," Journal of
Geophysical Research: Oceans, vol. 118, no. 10, pp. 5129-5172, 2013.
[53] L. Zheng et al., "Implications from the comparisons between two‐and three‐
dimensional model simulations of the Hurricane Ike storm surge," Journal of
Geophysical Research: Oceans, vol. 118, no. 7, pp. 3350-3369, 2013.
167
[54] Y. Huang, R. H. Weisberg, L. Zheng, and M. Zijlema, "Gulf of Mexico hurricane
wave simulations using SWAN: Bulk formula‐based drag coefficient sensitivity
for Hurricane Ike," Journal of Geophysical Research: Oceans, vol. 118, no. 8, pp.
3916-3938, 2013.
[55] J. Carter and H. Bermudez, "CR257 Shoreline Stabilization Project Post-Ike
Damage Assessment and Preliminary Solution Design PowerPoint Presentation,"
2/11/2009 2009.
[56] I. Coast & Harbor Engineering, "CR 257 (Bluewater Highway) Shoreline
Stablization Project; Post-Hurricane Ike Damage Assessment, Coastal
Engineering Analysis, and Preliminary Design," 02/27/2009 2009.
[57] C. Harter, "The impact of Hurricane Ike on the geomorphology of Follett's Island,
Texas-Short and long term effects," 2015.
[58] C. Harter and J. Figlus, "Numerical modeling of the morphodynamic response of
a low-lying barrier island beach and foredune system inundated during Hurricane
Ike using XBeach and CSHORE," Coastal Engineering, vol. 120, pp. 64-74,
2017.
[59] C. H. Engineering, "CR 257 (Bluewater Highway) Shoreline Stabilization Project:
Post-Hurricane Ike Damaage Assessment, Coastal Engineering Analysis and
Preliminary Design," Brazoria County02/27/2009 2009.
[60] E. Doran, "Shell Roads in Texas," Geographical Review, vol. 55, no. 2, pp. 223-
240, 1965.
168
[61] K.-H. Kim, H.-S. Yoo, and N. Kobayashi, "Mitigation of beach erosion after
coastal road construction," Journal of Coastal Research, vol. 27, no. 4, pp. 645-
651, 2011.
[62] Y. Onat, O. P. Francis, and K. Kim, "Vulnerability assessment and adaptation to
sea level rise in high-wave environments: A case study on O'ahu, Hawai'i," Ocean
and Coastal Management, Article vol. 157, pp. 147-159, 5/1/1 May 2018 2018.
[63] U. Gravois, "Rapid Response Measurements of Hurricane Waves and Storm
Surge," AGU Fall Meeting Abstracts, 12/01 2010.
[64] M. K. Potts, Forms and Distributions of Hurricane Ike Backflow and Scour
Features: Bolivar Peninsula, Texas: Master's thesis. College Station, TX: Texas
A&M University, 2010.
[65] D. J. Sherman, B. U. Hales, M. K. Potts, J. T. Ellis, H. Liu, and C. Houser,
"Impacts of Hurricane Ike on the beaches of the Bolivar Peninsula, TX, USA,"
Geomorphology, vol. 199, pp. 62-81, 2013.
[66] Palemón-Arcos, A. Torres-Freyermuth, A. Pedrozo-Acunã, and P. Salles, "On the
role of uncertainty for the study of wave-structure interaction," Coastal
Engineering, vol. 106, pp. 32-41, 2015.
[67] A. Carter, V. Magar, J. Simm, B. Goulby, and M. Wallis, "Accounting for
Multivariate Probabilities of Failure in Vertical Seawall Reliability Assessments,"
in ICE2013 Conference Paper, Edinburgh, UK, 2013, pp. 1-9: ResearchGate.
[68] G. Salvadori, G. R. Tomasicchio, and F. D'Alessandro, "Practical Guidelines for
Multivariate Analysis and Design in Coastal and Off-Shore Engineering," Coastal
Engineering, vol. 88, pp. 1-14, 2014.
169
[69] M. T. Schultz, B. P. Gouldby, J. D. Simm, and J. L. Wibowo, "Beyond the Factor
of Safety: Developing Fragility Curves to Characterize System Reliability,"
Geotechnical and Structures Laboratory7/2010 2010.
[70] A. Calabrese and C. G. Lai, "Sensitivity analysis of the seismic response of
gravity quay walls to perturbations of input parameters," Soil Dynamics and
Earthquake Engineering, vol. 82, pp. 55-62, 2016/03/01/ 2016.
[71] C. Fox-Lent, J. D. Rosati, and K. T. Touzinsk, "USACE Resilience Assessment
Methodologies - Engineering and Infrastructure."
[72] "What percentage of the American population lives near the coast?," NOAA NOS,
Accessed on: October 10Available:
https://oceanservice.noaa.gov/facts/population.html
[73] "North Atlantic Coast Comprehensive Study: Resilient Adaptation to Increasing
Risk," USACE, Washington, DC2015.
[74] "What is the National Environmental Policy Act?," Accessed on: January
24Available: https://www.epa.gov/nepa/what-national-environmental-policy-act
[75] S. L. Douglass, Krolak, Joe. (2008). Highways in the Coastal Environment,
Hydraulic Engineering Circular 25, Second Edition, FHWA NHI-07-096.
[76] F. M. Shilling, J. Vandever, K. May, I. Gerhard, and R. Bregoff, "Adaptive
Planning for Transportation Corridors Threatened by Sea Level Rise," Journal of
the Transportation Research Board, vol. No. 2599, pp. 9–16, 2016.
[77] ""Designing with Water" In Boston and Around the World Press Release,"
Accessed on: August 1Available:
https://www.cityofboston.gov/news/Default.aspx?id=14745
170
[78] "Boston Living with Water International Design Competition," ed, 2014.
[79] D. K. Hitchins, Advanced Systems Thinking, Engineering and Management.
Norwood, MA: Archtech House, Inc., 2003.
[80] "Coastal Resilience Solutions for East Boston and Charleston Final Report," City
of Boston, Boston, MAOctober 2017.
[81] J. R. Weggel, "Visibility over Shorefront Sand Dunes: An Ocean View," Drexel
University Libraries, Philadelphia, PA1998.
[82] UNCED, "Agenda 21," in United Nations Conference on Environment and
Development (UNCED), Rio de Janerio, Brazil, 1992: United Nations.
[83] G. Gallopín, A systems approach to sustainability and sustainable development.
United Nations Publications, 2003.
[84] H.-J. Schellnhuber, "Discourse: Earth System analysis—The scope of the
challenge," in Earth System Analysis: Springer, 1998, pp. 3-195.
[85] C. W. Karvetski, J. H. Lambert, J. M. Keisler, and I. Linkov, "USACE
Infrastructure Investments with Integration of Climate Change, Sea-Level Rise,
and Other Scenarios," presented at the Environment, Energy Security and
Sustainability Symposium and Exhibition, 2010.
[86] V. Belton and T. Stewart, Multiple criteria decision analysis: an integrated
approach. Springer Science & Business Media, 2002.
[87] C. W. Karvetski, J. H. Lambert, J. M. Keisler, and I. Linkov, "Integration of
decision analysis and scenario planning for coastal engineering and climate
change," IEEE Transactions On Systems, Man, And Cybernetics-Part A: Systems
And Humans, vol. 41, no. 1, pp. 63-73, 2010.
171
[88] G. Montibeller and L. A. Franco, "Raising the bar: strategic multi-criteria decision
analysis," Journal of the Operational Research Society, vol. 62, no. 5, pp. 855-
867, 2011.
[89] Y. Y. Haimes, "On the complex definition of risk: A systems‐based approach,"
Risk Analysis: An International Journal, vol. 29, no. 12, pp. 1647-1654, 2009.
[90] M. L. Parry, O. F. Canziani, J. P. Palutikof, P. J. Van Der Linden, and C. E.
Hanson, "IPCC, 2007: climate change 2007: impacts, adaptation and
vulnerability. Contribution of working group II to the fourth assessment report of
the intergovernmental panel on climate change," Cambridge Uni-versity Press,
Cambridge, UK, 2007.
[91] E. Brink et al., "On the road to ‘research municipalities’: analysing
transdisciplinarity in municipal ecosystem services and adaptation planning,"
Sustainability Science, journal article vol. 13, no. 3, pp. 765-784, May 01 2018.
[92] T. Jahn, M. Bergmann, and F. Keil, "Transdisciplinarity: Between mainstreaming
and marginalization," Ecological Economics, vol. 79, pp. 1-10, 2012/07/01/ 2012.
[93] D. Marchese, E. Reynolds, M. Bates, H. Morgan, S. Clark, and I. Linkov,
"Resilience and sustainability: Similarities and differences in environmental
management applications," The Science of the total environment, vol. 613-614,
pp. 1275-1283, 09/25 2017.
[94] G. Pennison, R. J. Cloutier, and B. M. Webb, "Local Coastal Roads-Next
Generation," Orlando, FL, 2018, vol. Proceedings of the 2018 Industrial and
Systems Engineering Conference, p. 6: University of South Alabama.
172
[95] H. Demirel, M. Kompil, and F. Nemry, "A framework to analyze the vulnerability
of European road networks due to Sea-Level Rise (SLR) and sea storm surges,"
Transportation Research Part A, Article vol. 81, pp. 62-76, 11/1/November 2015
2015.
[96] X. Xu, A. Chen, S. Jansuwan, K. Heaslip, and C. Yang, "Modeling transportation
network redundancy," Transportation Research Procedia, vol. 9, pp. 283-302,
2015.
[97] X. Xu, A. Chen, and C. Yang, "An optimization approach for deriving upper and
lower bounds of transportation network vulnerability under simultaneous
disruptions of multiple links," Transportation research procedia, vol. 23, pp. 645-
663, 2017.
[98] P. A. Gilbert and S. P. Miller, "A Study of Embankment Performance During
Overtopping," Vicksburg, MS1991.
[99] N. Zhang, Z. Li, A. Davis, P. Li, and A. He, "Numerical and Experimental
Investigation of Wave-Structure Interactions on Coastal Highways and Levees,"
in ASME 2012 Fluids Engineering Division Summer Meeting collocated with the
ASME 2012 Heat Transfer Summer Conference and the ASME 2012 10th
International Conference on Nanochannels, Microchannels, and Minichannels,
2012, pp. 197-202: American Society of Mechanical Engineers.
[100] N. Zhang, Z. Li, S. Klemetson, and S. Yadagiri, "CFD and Acoustical Analyses
for Coastal Highway Erosion," in ASME 2010 International Mechanical
Engineering Congress and Exposition, 2010, pp. 223-227: American Society of
Mechanical Engineers.
173
[101] K. L. Hendrickson, "Navier-Stokes Simulation of Steep Breaking Water Waves
with a Coupled Air-Water Interface," Doctor of Science Dissertation, Department
of Ocean Engineering, Massachusetts Institute of Technology, 2005.
[102] M. González-Betancourt, "Uplift force and momenta on a slab subjected to
hydraulic jump," Dyna, vol. 83, no. 199, pp. 124-133, 2016.
[103] A. Bellin and V. Fiorotto, "Direct dynamic force measurement on slabs in
spillway stilling basins," Journal of Hydraulic Engineering, vol. 121, no. 10, pp.
686-693, 1995.
[104] N. Attary, H. Cutler, M. Shields, and J. W. van de Lindt, "The economic effects
of financial relief delays following a natural disaster," Economic Systems
Research, pp. 1-27, 2020.
[105] B. Webb, "A Primer on Modeling in the Coastal Environment," 2017.
[106] D. R. Basco and N. Mahmoudpour, "The Modified Coastal Storm Impulse (COSI)
Parameter and Quantification of Fragility Curves for Coastal Design," in
Proceedings of the Coastal Engineering Conference, 2012.
[107] M. Hayatdavoodi and R. Ertekin, Review of Wave Loads on Coastal Bridge
Decks. 2016, pp. 030802-030802.
[108] I. Gidaris et al., "Multiple-Hazard Fragility and Restoration Models of Highway
Bridges for Regional Risk and Resilience Assessment in the United States: State-
of-the-Art Review," Journal of Structural Engineering, vol. 143, no. 3, p.
04016188, 2017.
174
[109] S. L. Douglass, Q. Chen, J. Olsen, B. Edge, and D. Brown, "Wave forces on
bridge decks," Coastal Transportation Engineering Research and Education
Center, Univ. of South Alabama, Mobile, Ala, 2006.
[110] Q. Chen, L. Wang, H. Zhao, and S. L. Douglass, "Prediction of storm surges and
wind waves on coastal highways in hurricane-prone areas," Journal of Coastal
Research, pp. 1304-1317, 2007.
[111] R. L. McPherson, "Hurricane induced wave and surge forces on bridge decks,"
Texas A & M University, 2010.
[112] A. Guo, Q. Fang, and H. Li, "Analytical solution of hurricane wave forces acting
on submerged bridge decks," Ocean Engineering, vol. 108, pp. 519-528, 2015.
[113] A. Guo, Q. Fang, X. Bai, and H. Li, "Hydrodynamic experiment of the wave force
acting on the superstructures of coastal bridges," Journal of Bridge Engineering,
vol. 20, no. 12, p. 04015012, 2015.
[114] X. Guoji and C. S. Cai, "Wave Forces on Biloxi Bay Bridge Decks with
Inclinations under Solitary Waves," Journal of Performance of Constructed
Facilities, Article vol. 29, no. 6, pp. 1-13, 2015.
[115] N. Ataei and J. E. Padgett, "Probabilistic modeling of bridge deck unseating
during hurricane events," Journal of Bridge Engineering, vol. 18, no. 4, pp. 275-
286, 2012.
[116] A. Spiller and C. Arnold, "Statistical analysis of coastal bridge vulnerability based
on empirical evidence from Hurricane Katrina AU - Padgett, Jamie E," Structure
and Infrastructure Engineering, vol. 8, no. 6, pp. 595-605, 2012/06/01 2012.
175
[117] N. Ataei and J. E. Padgett, "Limit state capacities for global performance
assessment of bridges exposed to hurricane surge and wave," Structural Safety,
vol. 41, pp. 73-81, 2013.
[118] N. Ataei and J. E. Padgett, "Fragility surrogate models for coastal bridges in
hurricane prone zones," Engineering Structures, vol. 103, pp. 203-213, 2015.
[119] A. Kennedy et al., Inundation and destruction on the bolivar peninsula during
hurricane ike. 2011.
[120] A. Sebastian, J. Proft, J. C. Dietrich, W. Du, P. B. Bedient, and C. N. Dawson,
"Characterizing hurricane storm surge behavior in Galveston Bay using the
SWAN+ADCIRC model," Coastal Engineering, vol. 88, pp. 171-181,
2014/06/01/ 2014.
[121] E. F. Thompson and C. Vincent, "Significant wave height for shallow water
design," Journal of waterway, port, coastal, and ocean engineering, vol. 111, no.
5, pp. 828-842, 1985.
[122] W. H. Munk, "The Solitary Wave Theory and its Application to Surf Problems,"
Annals of the New York Academy of Sciences, vol. 51, no. 3, pp. 376-424, 1949.
[123] USACE, "Coastal engineering manual," ed: US Army Engineering Research and
Development Center Vicksburg, Mississippi, 2002.
[124] J. T. Kirby and R. A. Dalrymple, "An approximate model for nonlinear dispersion
in monochromatic wave propagation models," Coastal Engineering, vol. 9, no. 6,
pp. 545-561, 1986.
[125] T. Hedges, "An empirical modification to linear wave theory," Proceedings of the
Institution of Civil Engineers, vol. 61, no. 3, pp. 575-579, 1976.
176
[126] G. Pennison, "SE 602 - Coastal Hazards Fragility Functions," 2016.
[127] M. Ouyang, "Review on modeling and simulation of interdependent critical
infrastructure systems," Reliability Engineering & System Safety, vol. 121, pp. 43-
60, 2014/01/01/ 2014.
[128] CCSP, "Impacts of Climate Change and Variability on Transportation Systems
and Infrastructure: Gulf Coast Study, Phase I," in "U.S. Climate Change Science
Program Synthesis and Assessment Product 4.7: Report by the U.S. Climate
Change Science Program and the Subcommittee on Global Change Research,"
Washington, DC, USA03/2008 2008.
[129] A. Clark and M. Savonis, "Transportation and Climate Change in the Gulf Coast:
The Need is Now! Presentation," 2008, p. 24: SSPEED – Storm Prediction and
Global Climate Impact.
[130] T. P. Bostick, T. H. Holzer, and S. Sarkani, "Enabling Stakeholder Involvement in
Coastal Disaster Resilience Planning," Risk Anal, vol. 37, no. 6, pp. 1181-1200,
Jun 2017.
BIOGRAPHICAL SKETCH
177
BIOGRAPHICAL SKETCH
Name of Author: Garland P. Pennison
Place of Birth: Morgan City, Louisiana
Date of Birth: March 12, 1959
Graduate and Undergraduate Schools Attended:
University of South Alabama, West Mobile, Alabama
Louisiana State University, Baton Rouge, Louisiana
Louisiana Tech University, Ruston, Louisiana
Louisiana Tech University, Ruston, Louisiana
Louisiana State University, Alexandria, Louisiana
Degrees Awarded:
Masters of Science in Civil Engineering, 1993, Ruston, Louisiana
Bachelors of Science in Civil Engineering, magna cum laude, 1979, Ruston,
Louisiana
Awards and Honors:
Graduate Research Assistant, Department of Civil, Coastal, and Environmental
Engineering, 2017-2019
Tau Beta Pi Engineering Honor Society, 1978
Chi Epsilon Civil Engineering Honor Society, 1978
Publications:
G. P. Pennison, "Microbial rock plant filters in Louisiana wastewater treatment,"
Louisiana Tech University Master’s Thesis, 1993.
Pennison, Garland P. "Structural Deflections from Wave Forces." Coastal
Structures and Solutions to Coastal Disasters 2015: Resilient Coastal
Communities. Reston, VA: American Society of Civil Engineers, 2017. 610-618.
178
Pennison, G.P. 2018. Predicting Coastal Roadway Damage using Modified
Dispersion Functions. Center of Excellence for Risk-Based Community
Resiliency Planning, First Place Award in the Graduate and Post-Doctoral
Competition; Awarded 4 May 2018.
Pennison, G.P., Cloutier, R.J., and Webb, B.M. 2018. Local Coastal Roads—Next
Generation. Proceedings of the 2018 Industrial and Systems Engineering
Conference. K. Baker, D. Berry, C. Rainwater, eds. 6 pp.
Pennison, G.P., and Webb, B.M. 2018. Coastal Roads: Using Failure to
Strengthen Resiliency. 2018 National Coastal Conference, American Shore and
Beach Preservation Association (ASBPA 2018) Resilient Shorelines for Rising
Tides, Galveston, TX.
Pennison, G.P., Webb, B.M., Padgett, J., and Gidaris, I. 2018. Predicting Coastal
Roadway Damage using Modified Dispersion Functions; Pennison. 36th
International Conference on Coastal Engineering (ICCE 2018), Baltimore, MD.
Pennison, G.P., and Webb, B.M. 2019. Using Coastal Road Failures to Improve
Resiliency. 2nd International Conference on Transportation System Resilience to
Natural Hazards and Extreme Weather (Transportation Research Board of The
National Academies of Sciences, Engineering, and Medicine), Washington, DC.
Pennison, G.P., and Webb, B.M. 2020. Coastal Road System Failures: Cause and
Effect. Proceedings of the 2020 Industrial and Systems Engineering Conference.
L. Cromarty, R. Shirwaiker, P. Wang, eds. 6 pp.
Pennison, G.P., and Webb, B.M. 2020. Transdisciplinary Systems Thinking:
Sustainability of Coastal Systems. Proceedings of the 2020 Industrial and Systems
Engineering Conference. L. Cromarty, R. Shirwaiker, P. Wang, eds. 6 pp.