University of Southern California Dissertations and Theses

A framework for comprehensive assessment of resilience and other dimensions of asset management in metropolis-scale transport systems

(USC Thesis Other)

A framework for comprehensive assessment of resilience and other dimensions of asset management in metropolis-scale transport systems

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content A FRAMEWORK FOR COMPREHENSIVE ASSESSMENT OF RESILIENCE AND OTHER DIMENSIONS OF ASSET MANAGEMENT IN METROPOLIS-SCALE TRANSPORT SYSTEMS by Eyüphan Koç A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY CIVIL ENGINEERING December 2021 Copyright 2021 Eyüphan Koç ii Acknowledgements I am extremely grateful, first and foremost, for my Ph.D. advisor Dr. Lucio Soibelman. His approach to our intellectual and professional relationship ensured that I felt ownership and responsibility of the work I conducted at USC, not only in research but also in teaching and service. His willingness to raise independent thinkers while he takes a position of consistent support throughout (slow and fast) episodes of research is an approach I will adopt in my career. Dr. Burcin Becerik-Gerber, Dr. Nan Li, Dr. Ryan Qi Wang, Dr. Dan Wei, Dr. Adam Rose, Dr. Azad Madni, Dr. Konstadinos Goulias and Dr. Ertugrul Taciroglu are other faculty members at USC and elsewhere who contributed to this work greatly either as collaborators, mentors and members on my qualification and thesis defense committees. I would also like to express my gratitude to all faculty members at my alma mater, Bogazici University, for inspiring and supporting me since my first day as a freshman there in 2010. Working with Dr. Emre Camlibel as an undergraduate assistant was particularly instrumental in my decision to pursue a doctorate. To this day, I am not entirely sure whether my family—mom, dad, and sister—ever fully agreed with my decision to come to California, 7000 miles away from home. Yet I am sure that they are proud. They were with me through the toughest of times and helped me push forward even when I doubted myself. I lost the kindest person on earth, my grandfather and namesake Eyup Koc, only one year into my graduate studies and could not attend his funeral. As I finish this thesis, I remember him once more as my guide in life. Thanks to my dear friends and colleagues in Los Angeles and around the world; Murat Uzun, Rasit Mete Esrefoglu, Evangelos Pantazis, Ashrant Aryal, Joao Pedro Carneiro, Arsalan Heydarian, Pouyan Hosseini, Meida Chen, Gerry Rizzo, Jinwoo Im, Becca Peer, Victor Pourcel, Fernando Lucchini, Salih Tugay Gurler, Omer Mesci, Cem Okucu, Baransel Soysal, Vedat Ozan Ozkan, Utku Gumusay, Suat Yigitoglu, Alp Kiremitci, Baris Seven, Yalim Acar, Kuba Graczyk, Jerfi Firatli and Meryem Sengul; I was always saved just-in-time from writing boredom and episodes of graduate school angst. Thanks to the Nautical Science Program at USC, 1969 schooner “Atlantas” and Captain Lars for incredible times in the classroom and out in the Pacific. Among my role models, one man holds a special place and I would like to conclude by remembering him and his legacy: “My people are going to learn the principles of democracy the dictates of truth and the teachings of science. Superstition must go. Let them worship as they will, every man can follow his own conscience provided it does not interfere with sane reason or bid him act against the liberty of his fellow men.” —Mustafa Kemal Atatürk (1881 – ∞). iii Table of Contents Acknowledgements ii List of Figures vi List of Tables viii Abstract ix CHAPTER 1 INTRODUCTION 1 1.1 Resilience of Transportation Systems 3 1.2 Other Dimensions of Asset Management in Transportation Systems 6 CHAPTER 2 A FRAMEWORK FOR COMPREHENSIVE ASSESSMENT OF RESILIENCE IN METROPOLIS-SCALE TRANSPORT SYSTEMS 9 2.1 Background 9 2.1.1 Topology-Based and System-Based Vulnerability 10 2.1.2 System-Based Vulnerability Studies 12 2.1.3 Economic Impact Analysis of Transportation Disruptions 16 2.1.4 Research Gaps in System-Based Vulnerability Analysis: 3 Perspectives 21 2.1.5 Shortcomings in Transportation System Analysis 21 2.1.6 Shortcomings in Hazard Characterization and Damage Assessment 23 2.1.7 Shortcomings in Economic Impact Analysis of Transportation Disruptions 25 2.2 Recent Advances in Modeling 28 2.3 Towards Comprehensive Resilience Assessment 30 2.4 Research Objectives 32 2.5 CRAFT: Comprehensive Resilience Assessment for Transportation Systems in Metropolitan Areas 34 2.5.1 Resilience at System Component Level: Hazard Characterization and Damage Assessment 36 2.5.2 Resilience at System Level: Transportation System Analysis 42 2.5.3 SCAG Regional Travel Model 45 2.5.4 Resilience at Regional Economic Level 49 2.6 Case Study: Magnitude 7.3 Earthquake on Palos Verdes Fault 53 2.6.1 Hazard Characterization 53 2.6.2 Damage Assessment for Bridges 54 2.6.3 Transportation Systems Analysis 57 iv 2.6.4 Damage Assessment for Other Critical Infrastructures and General Building Stock 66 2.6.5 Damage Assessment for Ports of Los Angeles and Long Beach 66 2.6.6 Damage Assessment for General Building Stock 66 2.6.7 Functionality Loss and Recovery at the Ports of Los Angeles and Long Beach 68 2.6.8 General Building Stock Damages in Greater Los Angeles 69 2.6.9 Economic Resilience Tactics: Case of Ports and Hinterland Transportation 74 2.6.10 Construction of the Multi-Sector Income Distribution Matrix of California 76 2.6.11 Aggregate Impacts of Port Disruptions 77 2.6.12 Aggregate Impacts of Increasing Truck Transportation Costs and Building Stock Damages 79 2.6.13 Economic Impacts of Combined Disruptions/Damages 81 2.6.14 Summary on Socioeconomic Impacts and Resilience 81 2.6.15 Income Distribution Impacts 82 2.6.16 Understanding Disruption and Recovery of Regional Transportation under Dynamic Resilience Tactics 84 CHAPTER 3 OTHER DIMENSIONS OF TRANSPORTATION ASSET MANAGEMENT 88 3.1 Bridge Asset Management in the United States 90 3.1.1 Deterioration/Condition Rating Assessment 90 3.1.2 Options in Bridge Asset Management to Maintain Bridge Serviceability 92 3.1.3 How do owners and operators of bridges take maintenance decisions? 94 3.1.4 Federal Guidance 95 3.2 Pathways for Data-Driven Advancements in Bridge Asset Management 96 3.3 Review of Accessibility Concepts in Asset Management and Equity Contexts 97 3.3.1 Quantifying and Mitigating Impacts on Vulnerable Populations 100 3.3.2 Measurements of Accessibility 101 3.4 Framework for Transportation Systems, Accessibility and Equity Analyses for Bridge Asset Management 102 3.4.1 Transportation System Analysis 104 3.4.2 Accessibility Analysis 105 3.4.3 Equity Analysis 106 3.1 Case Study: Investigating Engineering Alternatives 106 3.2 Economic Impacts of MRR-related Closures 115 CHAPTER 4 DISCUSSION ON LIMITATIONS AND FUTURE WORK 119 4.1 Discussion on CRAFT 119 4.2 Discussion on Other ‘Data’ Dimensions of Asset Management 124 CHAPTER 5 CONCLUSION 127 v REFERENCES 129 APPENDIX A. CONSTRUCTION OF INCOME DISTRIBUTION MATRIX 146 A1. Employee Compensation 146 APPENDIX B. METHOD OF MODELING RESILIENCE TACTICS IN TERM 154 APPENDIX C. DETAILED INCOME DISTRIBUTION IMPACT RESULTS 156 vi List of Figures Figure 1.1 Economic damage by disaster type (EM-DAT 2020), graphic adopted from Our World in Data. 1 Figure 1.2 Citations to works on urban sustainability and resilience. .................................................................. 2 Figure 2.1 Example illustrating critical links using V/C rsatios (adopted from Scott et al. 2006). .................. 14 Figure 2.2 Disaster measures, their boundaries and interactions. Adopted from Faturechi and Miller-Hooks (2014). ................................................................................................................................................................ 31 Figure 2.3 Measure of seismic resilience—conceptual definition. Adopted from Bruneau et al. (2013). .... 32 Figure 2.4 A conceptual illustration of CRAFT: Comprehensive Resilience Assessment Framework for Transportation Systems. .................................................................................................................................. 35 Figure 2.5 Several Google Street View images (a), fragility functions (b), and geometric model (c) obtained for a bridge considered in this study. ............................................................................................ 41 Figure 2.6 (a) High resolution, multi-modal network model underlying the SCAG RTDM. ............................. 46 Figure 2.7 (a) PSHA results for the Ports of Los Angeles and Long Beach, and (b) their deaggregation. .. 55 Figure 2.8 Deterministic hazard analysis results for 1.0 sec spectral accelerations at the general study region level (a), the location of damaged bridges within the general study region (b), 1.0 sec spectral acceleration results around rupture length (c), and the corresponding bridge damage in terms of downtimes (d). ................................................................................................................................................... 56 Figure 2.9 Modeled bridge closures on Day 1. Region-of-Interest contains bridges modeled with the image-to-model methodology. ........................................................................................................................ 57 Figure 2.10 Network topology manipulation on links corresponding to bridges deemed closed to traffic (e.g., Bridge 53 1867 on the I210 freeway) implemented with TransCAD’s map editing functionalities. Colored, thicker lines indicate public transit routes such as buses, metro, etc. ..................................... 58 Figure 2.11 System functionality Q(t) based on Vehicle Hours Traveled (VHT). ............................................. 61 Figure 2.12 System functionality Q(t) based on Vehicle Hours Delayed. ......................................................... 61 Figure 2.13 Changes in Vehicle Flow from Day 0 to Day 1 after the scenario event during the AM peak. Available online at: https://arcg.is/HjDO8 ...................................................................................................... 62 Figure 2.14 Changes in Vehicle Flow from Day 0 to Day 7 after the scenario event during the AM peak. Available online at: https://arcg.is/1PeDOa ................................................................................................... 63 Figure 2.15 Spatial distribution of changes in VMT from Day 0 to Day 1, aggregated up to Tier 2 TAZ level. Available online at: https://arcg.is/1mSC9T .................................................................................................. 64 Figure 2.16 Spatial distribution of changes in VMT from Day 0 to Day 7, aggregated up to Tier 2 TAZ level. Available online at: https://arcg.is/1PeDOa ................................................................................................... 65 Figure 2.17 Bridges Deemed Open under the Dynamic Resilience Tactic between Days 1-7, and Bridges that Remain Closed during the Same Time Period. .................................................................................... 86 Figure 2.18 Regional System Functionality Q(t) before and after resilience tactic. ......................................... 87 Figure 3.1 Bridge Action Categories ....................................................................................................................... 93 Figure 3.2 Bridge Conditions over Time (adopted from U.S. Department of Transportation, 2018) ............ 94 vii Figure 3.3 A framework to investigate system-level impacts of MRR decision making on transportation system functionality as well as accessibility and equity. .......................................................................... 103 Figure 3.4 Changes in Vehicle Hours Traveled (VHT) under Scenario 1: Half Closure. .............................. 109 Figure 3.5 Changes in Vehicle Hours Traveled (VHT) under Scenario 2: Full Closure. .............................. 110 Figure 3.6 TAZs most affected by changes in hospital accessibility under Scenario 1: Half Closure. ...... 111 Figure 3.7 TAZs most affected by changes in hospital accessibility under Scenario 2: Full Closure. ....... 112 Figure 4.1 Network resilience curve adapted from Frangopol and Bocchini (2011) and revised. Dashed red line indicates faster recovery path achieved through implementation of resilience tactics. ....... 121 viii List of Tables Table 2.1 Reviewed studies not accommodating explicit network modeling and analysis. ........................... 19 Table 2.2 Reviewed studies accommodating explicit network modeling and analysis. .................................. 19 Table 2.3 Shortcomings in system-based analyses of transportation disruptions: 3 perspectives. .............. 27 Table 2.4 Specific Occupancy Classes for the General Building Stock in HAZUS ......................................... 68 Table 2.5 Percentage of Port Functionality and Recovery Estimates for Different Cargo-Handling Terminals ............................................................................................................................................................ 69 Table 2.6 Direct Loss Estimates (Building and Content Losses Only) for the LA Metro Region (Los Angeles, Orange and Riverside Counties) .................................................................................................... 71 Table 2.7 General Building Damage for LA Metro Region .................................................................................. 74 Table 2.8 Summary of Resilience Tactics Relating to Port and Highway Transportation Disruptions .......... 76 Table 2.9 Real GDP Impact of Port Disruptions – Base Case and Resilience Cases (in millions 2019 dollars and percent reduction from pre-disaster levels). ......................................................................................... 79 Table 2.10 Real GDP Impact of the Combined Disruptions/Damages in Base Case and Resilience Cases (in millions 2019 $ and percent reduction from pre-disaster levels). ........................................................ 82 Table 2.11 Gini Coefficient Impacts ........................................................................................................................ 84 Table 3.1 Common actions based on Bridge Element Condition State (adopted from U.S. Department of Transportation, 2018) ....................................................................................................................................... 91 Table 3.2 Engineering alternatives for Bridge No. 53-1867 on the San Gabriel River. ................................ 108 Table 3.3 System functionality indicators for the regional network (SCAG region) under Scenarios 1 and 2. ....................................................................................................................................................................... 113 Table 3.4 Equity results for both scenarios across different racial and ethnic groups. ................................ 114 Table 3.5 Real GDP Impacts of closure scenarios. ............................................................................................ 117 ix Abstract Disasters exert profound impacts on human societies. Direct costs of disasters have exceeded 2.5 trillion US dollars in the 21st century affecting more than 3 billion people and causing more than 1.2 million casualties around the globe (UNISDR 2018) 1 . Stemming from the inability to grow sustainably, as well as to build resilience at the rate of urban growth, many cities are facing increasingly complex resilience challenges. In the definition of urban resilience (Meerow et al. 2016a), the urban system is characterized by its components such as its governance networks, networked material and energy flows, urban infrastructure and form, and socio-economic dynamics. Among the components of the urban system, infrastructure systems or lifelines are key facilitators that support the lives, interactions, and dynamics of urban dwellers. Lifelines or Critical Infrastructures (CIs) (transportation, water, power, telecommunications, etc.) are essential to the well- being of the society not only under business as usual conditions but also during times of disaster for the entire response and recovery timeline. In this setting, it is argued that the transportation system is one of the most significant lifelines, because disturbance to transportation imposes extra burdens on other lifelines (Hopkins et al. 1991) (e.g., handling of a power substation failure due to earthquake damage requires a connected road network or other functioning modes of transportation for effective response). Despite this, analytical tools and approaches advising policy making to improve resilience are scarce (Ganin et al. 2017). In this thesis, the author contends that there is also a lack of synthetic approaches that address the diversity of challenges associated with civil system disruptions with a special focus on transportation. Most investigations practically exclude one or more dimensions of the problem that stem innately from exposure to hazards, the vulnerability of the physical infrastructure, and the direct and indirect losses that result from this coupling. Having this in mind, the author intends to achieve a balance between 1 These statistics do not account for the ongoing COVID-19 pandemic with devastating impacts on global economy. COVID-19 to slash global economic output by $8.5 trillion over next two years. (2021). Retrieved May 21, 2021, from https://www.un.org/en/desa/covid-19-slash-global-economic-output-85-trillion-over-next-two-years) x the two overlapping views—analytical and synthetic—with a framework that is designed to generate holistic and actionable resilience insights related to transportation network disruptions in metropolitan areas. This convergent framework is called CRAFT for Comprehensive Resilience Assessment Framework for Transportation Systems and consists of: (1) a hazard characterization and damage assessment module that simulates the governing event causing the disruption and estimates the physical damages to network components leveraging a novel image-based modeling methodology, (2) a transportation analysis module (implemented with a high-resolution metropolis-scale travel demand model) investigating the disruption and (3) a socioeconomic impact analysis and economic resilience module based on CGE (computable general equilibrium) analysis, further supplemented by a multi-sector income distribution matrix that calculates the business interruption losses and income distribution impacts. CRAFT is deployed in a Greater Los Angeles Area case study to investigate a magnitude 7.3 earthquake scenario on Palos Verdes Fault. The author additionally asserts that the shortcomings in transportation systems research in targeting holistic and actionable insights (that are strengthened by an enhanced use of traditional and novel data sources, and state-of-the-art modeling) are not limited to the disaster context but also exist in broader asset management of transportation systems. In the case of bridges, the most critical links in multi-modal transportation networks due to the low redundancy associated with their closure, the asset management decisions on alternative maintenance, rehabilitation, and replacement (MRR) strategies are often not as informed as they could be since the data, tools, and methods employed are not up to speed with the ongoing ‘data revolution’ in the civil systems domain. Thus, this thesis also explores the currently underutilized dimensions of data by exploring improvements in the maintenance, repair, and reconstruction decision- making, particularly in the context of its impacts on system-level performance indicators (e.g., changes in total VHT, VMT, Delay, etc. in a region), on accessibility to jobs and services as well as on local/regional/national economies fueled by the supply chains utilizing the transportation infrastructure. This is achieved through another system-based xi framework including data, tools, and methods leveraged to investigate transportation and socioeconomics in urban areas. The mentioned explorations are demonstrated with a Los Angeles case study developed in collaboration with the asset management team of California Department of Transportation (Caltrans) District 7. 1 Introduction Disasters exert profound impacts on human societies. Direct costs of disasters have exceeded 2.5 trillion US dollars in the 21 st century affecting more than 3 billion people and causing more than 1.2 million casualties around the globe (UNISDR 2018) 1 . Figure 1.1 illustrates the scale of economic damage by natural disaster type since 1900. Figure 1.1 Economic damage by disaster type (EM-DAT 2020), graphic adopted from Our World in Data 2 . Stemming from the inability to grow sustainably, as well as to build resilience at the rate of urban growth, many cities are facing increasingly complex resilience challenges. This 1 These statistics do not account for the ongoing COVID-19 pandemic with devastating impacts on global economy. COVID-19 to slash global economic output by $8.5 trillion over next two years. (2021). Retrieved May 21, 2021, from https://www.un.org/en/desa/covid-19-slash-global-economic-output-85-trillion-over-next-two-years) 2 Our World in Data. Retrieved April 20, 2021 from https://ourworldindata.org/. 2 chronic trend is expected to continue as urban settlements grow in number and size. Approximately 70% of the human population is projected to live in cities by 2050 3 . This is foreshadowing even more exposure with higher concentrations of people, capital, and infrastructure in urban areas. Events of the early 21 st century around the globe, including Hurricane Katrina and catastrophic seismic events in Haiti, Chile, and Japan have increased the awareness and the importance of resilience (Freckleton et al. 2012). This Figure 1.2 Citations to works on urban sustainability and resilience 4 . has been strongly highlighted during the ongoing COVID-19 pandemic as well. Consequently, the research area of ‘urban sustainability and (community) resilience’ is 3 68% of the world population projected to live in urban areas by 2050, says UN | UN DESA Department of economic and social affairs. (2018). Retrieved April 20, 2021, from https://www.un.org/development/desa/en/news/population/2018-revision-of-world-urbanization-prospects.html 4 Dimensions.ai database. Retrieved April 20, 2021 from https://www.dimensions.ai/. 3 attracting increasing interest over the last decade from researchers in many related disciplines. Figure 1.2 shows citations to works on urban sustainability and resilience since 2000. The root of its definition originating from ecological resilience, urban resilience refers to “the ability of an urban system—and all its constituent socio-ecological and socio- technical networks across temporal and spatial scales—to maintain or rapidly return to desired functions after a disturbance, to adapt to change and to quickly transform systems that limit current or future adaptive capacity” (Meerow et al. 2016a). In this definition, the urban system is characterized by its components such as its governance networks, networked material and energy flows, urban infrastructure and form, and socio-economic dynamics. Among the components of the urban system, infrastructure systems or lifelines are key facilitators that support the lives, interactions, and dynamics of urban dwellers. Lifelines or Critical Infrastructures (CIs) (transportation, water, power, telecommunications, etc.) are essential to the well-being of the society not only under business as usual conditions but also during times of disaster for the entire response and recovery timeline. Moreover, CIs have become increasingly interdependent which makes them vulnerable to domino effects that distribute failures through time and space. In this setting, it is argued that the transportation system is one of the most significant lifelines, because disturbance to transportation imposes extra burdens on other lifelines (Hopkins et al. 1991) (e.g. handling of a power substation failure due to earthquake damage requires a connected road network or other functioning modes of transportation for dispatch teams). Resilience of Transportation Systems Transportation networks play a central role in providing the mobility of people and goods, which is an immediate functional need in the immediate aftermath of a disaster (for emergency response activities such as search and rescue efforts, delivery of aid, etc.) and longer-term recovery. Some of the recent events of significance that involved major disruptions to transportation include Hurricane Sandy (2012), Hurricane Irene (2011), the 4 Japanese Tsunami (2011) and subsequent nuclear meltdown, the Sichuan Earthquake in China (2008), the Christchurch earthquake in New Zealand (2011), the Minneapolis I-35W bridge collapse (2007), and Hurricane Katrina (2005) (Faturechi and Miller-Hooks 2014). The contribution of transportation systems to the full economic recovery from a disaster is often omitted in studies investigating the engineering aspects of resilience 5 . For example, a detailed survey of business closures after the Northridge Earthquake in 1994 shows that 56.4% of the businesses reported Employees Unable to Get to Work and 24.0% reported Could Not Deliver Products or Services as reasons for closure (Tierney 1995). Thus, the extent of transportation damage and the speed of its restoration are critical determinants of how quickly a disaster-stricken urban area can recover (Chang 2003). Despite their multifaceted importance, transportation systems in the United States have been in poor condition, as they are aged and need a major retrofit or replacement effort. The American Society of Civil Engineers continuously gives the road and bridge inventories poor grades in its periodic infrastructure report card (ASCE 2017). Transportation systems in large metropolitan areas that are exposed to natural hazards and possess characteristic vulnerabilities (e.g., earthquake exposure in the Greater Los Angeles Area along with the vulnerability of bridges, storm surge risks in cities of Southeast US with vulnerable infrastructure in low-lying urban areas, etc.) are especially at risk. In addition, transportation systems have been commonly targeted in antagonistic acts of terrorism, and resilience to such attacks has become a strategic policy objective for many developed countries including the US (Reggiani 2013). It is also worth noting that many metropolitan areas operate their transportation systems at or near capacity during peak hours which projects significant functionality losses in case of a disturbance, despite a relatively higher redundancy in transportation systems compared to other lifelines. 5 This is also the case for many other civil systems, however, a discussion across all civil systems is beyond the scope here. 5 Given the setting above, transportation disruptions have increased interest in academic research. Despite this, analytical tools and approaches advising policy making to improve resilience are scarce (Ganin et al. 2017). In this thesis, the author contends that there is also a lack of synthetic approaches that handle the diversity of challenges associated with transportation system disruptions. Most investigations practically exclude one or more dimensions of the problem that stem innately from exposure to hazards, the vulnerability of the physical infrastructure, and the direct and indirect losses that result from this coupling. Having this in mind, the author intends to achieve a balance between the two overlapping views—analytical and synthetic as highlighted by Goldberg (1975) as an alternate view on complex problems and decisions— with a framework that is designed to generate holistic and actionable resilience insights related to transportation network disruptions in metropolitan areas. The framework is called CRAFT for Comprehensive Resilience Assessment Framework for Transportation Systems and consists of: (1) a hazard characterization and damage assessment module that simulates the governing event causing the disruption (e.g. seismic hazard, tsunami, extreme wind, etc.) and estimates the physical damages to network components leveraging a novel image-based modeling methodology, (2) a transportation analysis module (implemented with a high- resolution travel demand model) investigating the disruption and (3) a socioeconomic impact analysis module based on CGE (computable general equilibrium) analysis supplemented by a multi-sector income distribution matrix that calculates the business interruption losses and income distribution impacts, and quantifies economic resilience. Standing on these three modules, CRAFT assesses 3 levels of resilience in a study region: (1) system component level (e.g. bridges, tunnels), (2) system level (e.g. multi-modal transportation system of a metropolitan area), (3) regional economic level. This thesis introduces the framework and demonstrates the workings of its couplings between the hazard, transportation, and economics modules through a case study in Greater Los Angeles Area. The framework is designed to achieve holistic analyses of transportation disruptions by addressing the many shortcomings and research gaps in this domain. It couples a novel structure-specific modeling methodology 6 with a high-fidelity metropolis-scale travel demand model based on real socioeconomic data and produces results, which, in turn, serve as input for a state-of-the-art socioeconomic impact analysis methodology that is based on computable general equilibrium (CGE) analysis. By the virtues of its data-intensive, model-based, and cross- disciplinary nature, CRAFT aims to capture and incorporate many details that are usually neglected in traditional approaches. The author will elaborate in detail on these aspects in the following sections of the thesis. Other Dimensions of Asset Management in Transportation Systems As argued previously, among the components of the urban system, transportation networks are among the most critical facilitators that support the lives, interactions, and dynamics of urban dwellers. They are essential to the well-being of the society not only during times of disaster but also during business-as-usual times for the mobility of people and goods, giving life to supply chains across the entire economy. Over the years of cooperation in research with transportation authorities in Los Angeles, the author and his collaborators realized that the shortcomings mentioned above, centered around the lack of approaches targeting holistic and actionable insights that are strengthened by an enhanced use of traditional and novel data sources, are not limited to the disaster context but also exist in broader asset management of transportation systems. This, in pairing with the mentioned poor conditions of the transportation infrastructure in the US prevent decision-makers to achieve a full view of the first and higher-order effects on network users (e.g., increased travel time, distance; higher fuel and maintenance costs; loss of accessibility to jobs, goods, and services, etc.) and on businesses and regional economies by increasing the costs of production as well as distribution of goods and services. ASCE estimates the resulting economic losses as $2 trillion business losses per year, $1 trillion in GDP losses per year, 2.5 million jobs lost per year, and $3,400 disposable income losses per household per year. Against this backdrop, the lack of funding for maintenance, rehabilitation, and replacement activities result in a rapid deterioration of transportation 7 infrastructure which exacerbates the mentioned effects, a dynamic that has been captured by strategy documents at the macro level such as the U.S. National Academy of Engineering’s ‘grand challenges’ and internationally by the U.N. Sustainable Development Goals. In this climate, transportation asset management functions across a diverse body of decision-makers (e.g., city/state/federal level owners and agencies) possess a critical role in ensuring public safety, cost-effectiveness, desirable socioeconomic outcomes as well as satisfying long-term sustainability goals. AASHTO defines transportation asset management as ‘a strategic and systematic process of operating, maintaining, upgrading, and expanding physical assets effectively throughout their lifecycle’ (USDOT, FHWA 2007). In the case of bridges, the most critical links in multi-modal transportation networks due to the low redundancy associated with their failure, the asset management decisions on alternative maintenance, rehabilitation, and replacement (MRR) strategies are often not as informed as they could be since the data, tools, and methods employed are not up to speed with the ongoing ‘data revolution’ in the civil systems domain. There are a wide spectrum of needs for advancements starting from the assessment of element-level (e.g., bridge deck) or component-level (e.g, a bridge serving a road network) conditions to utilizing such evaluations for bridge-by-bridge and network-level decisions. In this thesis, the authors particularly focus on the latter ‘decision-making’ aspect with the broader objective of designing methodologies that empower decision-makers to produce data- driven insights. This thesis explores the currently underutilized dimensions of data (and its analytics) by investigating maintenance, repair, and reconstruction decision-making in the context of its impacts on system-level performance indicators (e.g., changes in total VHT, VMT, Delay, etc. in a region), on accessibility to jobs and services as well as on local/regional/national economies fueled by the supply chains utilizing the transportation infrastructure. These objectives motivate Chapter 3 on other dimensions of asset management in transportation systems. To achieve them, first, a review of the current practices in bridge asset management in the United States is presented. The review motivates explorations for data-driven and cross-disciplinary improvements through another system-based, analytical framework including data, tools, and methods from 8 transportation systems analysis, economic impact analysis, and social impact analysis areas. The mentioned explorations are demonstrated with a Los Angeles case study developed in collaboration with the asset management team of California Department of Transportation (Caltrans) District 7. It is essential to note that the resilience context presented in Chapter 2: and the broader decision-making context for asset management presented in Chapter 3: are handled as two separate ‘problems’ (1) to highlight the resilience context where most of the author’s doctoral work focused on and (2) to avoid a premature synthesis between the two analytical frameworks, although their synthesis could be a potentially fruitful direction in future research, e.g, for a multi-criteria decision analysis problem. Thus, Chapters 2 and 3 are formatted accordingly, with sections on research gaps, research objectives, methodology of their own. The thesis is then concluded with a global discussion and conclusion with Chapter 4: and Chapter 5:. 9 A Framework for Comprehensive Assessment of Resilience in Metropolis-Scale Transport Systems Background Research investigating transportation systems in the context of disasters originates from concepts and tools of traditional risk analysis. Therefore, understanding system vulnerability has long been the focus that underlines the capability of anticipation (knowing what to expect in Hollnagel’s four cornerstones of resilience (Hollnagel 2011) as opposed to knowing what to look for, knowing what to do, and knowing what has happened). To that end, there is a substantial amount of literature on network vulnerability. Vulnerability for transportation networks, as defined by Berdica (2002), is the susceptibility to incidents that can result in considerable reductions in network serviceability, where serviceability is defined as the possibility of using a link/route/network during a given period. Resilience, on the other hand, offers a broader perspective that highlights how the anticipating ability of vulnerability analysis must interact with the monitoring (knowing what to look for), responding (knowing what to do), and learning (knowing what has happened) abilities to contribute to a more resilient system (Mattsson and Jenelius 2015). Berdica’s definition of vulnerability goes hand in hand with traditional definitions of risk consisting of the probability and consequence of an incident. However, as argued openly by Berdica, it is often impossible to fully capture disruptive events in terms of probabilities in an accurate manner—e.g. terrorist actions, extreme weather events, earthquakes. This makes it necessary to investigate what-if scenarios—"scenario-specific vulnerability” in the methodological categorization proposed by Murray et al. (2008)—and more importantly, the associated post-disturbance phenomena such as maximum disturbance from which the system can recover and speed of recovery. For this reason, i.e. by covering the disruption timeline holistically, resilience has recently been the 10 preferred concept and the way of thinking in the transportation area as well as in neighboring fields such as urban sciences (Meerow et al. 2016b) as opposed to being only a component of vulnerability analysis in the early prescription by Berdica (2002). The concept can be thought of as spanning both pre-event measures that seek to prevent hazard-related damage and losses and post-event strategies designed to cope with and minimize disaster impacts (Bruneau et al. 2003). Still, research on transportation network disruptions is founded on an abundant body of vulnerability work and many of the shortcomings are common with studies of resilience. Moreover, vulnerability analysis is essential within the treatment of resilience in the author’s research. Therefore, this section begins with an account of the transportation network vulnerability analysis research. In terms of approaches to transportation network vulnerability, the studies are generally categorized into two main categories. Topology-Based and System-Based Vulnerability Studies on the vulnerability of transportation systems are generally grouped into two main methodological categories: topological (graph theory-based) vulnerability and system- based vulnerability approaches (Mattsson and Jenelius 2015). Topological approaches are founded on the pillars of graph theory and model the road network as a set of connected nodes and edges to leverage the elegant graph-theoretical representation. Vulnerability is then investigated by the change in network efficiency metrics such as the sum of the distance of shortest paths between all node pairs in the network, size of the largest connected component, etc. (Pitilakis et al. 2016). Attacks on the network are mimicked with the removal of nodes or links—randomly or strategically—and with link/node criticality quantified in terms of topological metrics built on centrality measures such as degree, betweenness, etc. (Duan and Lu 2015) or connectivity based measures. The mentioned efficiency metrics may be quantified at the component (local) or network (global) levels depending on the purpose of the analysis. Topological approaches are not data-hungry; besides the network topology representing the studied transportation infrastructure, there is virtually no data or modeling effort required to investigate 11 vulnerability 6 . However, the said practicality in topological approaches is gained while sacrificing insights regarding network supply and demand, consequently leaving out a considerable portion of transportation system analysis carried out today (Cascetta, 2009). For example, Duan and Lu (2015) investigate the relationship between topological structure and vulnerability of city road networks concerning three types of network representations (segment, stroke, and community). The authors acknowledge that “some roads are more frequently used than other roads” and assert that previous studies did not discuss whether road network robustness analysis should consider this inherent demand attribute of roads. Still, they use betweenness centrality and shortest-path type performance measures to quantify vulnerability. Furthermore, they identify the need to study different types of disruptions (e.g., punctiform traffic disruption due to a small accident, linear closures during marathons, zonal paralysis during crowd movements of a large event, etc.), however, they limit their empirical analyses to three types of attacks that are frequently used in topology-based approaches: random, betweenness-based, and degree-based attacks. Such limitations virtually exist in every topology-based study of disruptions in transportation networks since the capacity-flow fundamentals of a traffic network are excluded from the beginning and the disruption-causing events are not linked to the inventories through formal hazard analysis and damage assessment methodologies. On the other hand, system-based vulnerability methodologies have a more holistic approach to the transportation system at hand and specifically focus on the interaction of network supply and demand—mostly utilizing models built around the traditional 4-step framework or new generation activity-based modeling approaches—allowing formal treatments of disruption related phenomena such as reduced link capacities, increasing levels of capacity utilization and decreasing redundancy. This way, resulting losses in network functionality that manifest in the form of surging travel times and distances are 6 For transit networks in particular, vulnerability in public transportation networks have largely been studied topologically due to the lack of sophisticated modeling tools available to carry out system-based studies. 12 quantified. System-based approaches also present the best opportunity for the modeling of post-disaster traveler behavior (despite the lack of data and calibrated models thereof) that could enable user-centric insights such as demand loss due to disruption, changes in mode, and destination choice behaviors, etc. However, these approaches are data- hungry and require calibrated demand and supply models as well as sophisticated simulation platforms operating on traffic assignment algorithms to simulate mobility 7 . Moreover, if formal damage assessment is to be carried out to determine the vulnerability of network components (e.g., bridges, tunnels, etc.), hazard simulation models and detailed infrastructure inventories are required. This means the availability of data and models usually dominates the approach and a unified methodology may not always be possible in contrast with topology-based methods (Koc et al. 2019b). All in all, system-based approaches allow the measurement of direct and indirect impacts of transportation disruptions in a more realistic fashion, at different levels of complexity (user-centric or system-level travel costs, economic or financial losses, environmental impacts, etc.) and in a wide range of scenarios and for different purposes. Therefore, the focus of this research is on the development of a system-based framework that leverages the integration of a metropolis scale travel demand model, a novel model- based damage assessment methodology, and a computable general equilibrium model determining economic impacts. Before elaborating on the framework itself, the author presents a perspective on the literature in system-based vulnerability and the identified gaps guiding this section of the thesis. System-Based Vulnerability Studies Following a small number of studies in earlier decades on finding hot spots in the transportation networks, studies in system-based vulnerability have increasingly been conducted after Berdica’s (2002) review who defined vulnerability as a problem of reduced accessibility different from the earlier treatment of the concept from a safety point 7 This is especially the case for metropolitan areas such as Los Angeles, New York, etc. that require the simulation of millions of travelers. 13 of view. This understanding focusing on the functionality of road networks was adopted widely by transportation researchers and led to numerous studies published in transportation journals as opposed to avenues in mathematics and physics which has been the case for many topological vulnerability studies. This section provides a broad review of system-based vulnerability studies investigating disruptions resulting from events with severe consequences (earthquakes, terrorist attacks, etc.) affecting network supply drastically. Thus, works on travel time reliability dealing with relatively minor fluctuations in network supply or demand are excluded. One of the main objectives of vulnerability research is to find critical links or nodes in the network. As discussed above, topological approaches quantify criticality with metrics detached from the supply/demand relationships in transportation. On the system- based side, vulnerability is often quantified based on the marginal travel time/cost induced on the users in the degraded network. Nicholson and Du (1997) conducted one of the early works with this understanding. Identifying the lack of attention to the increases in user costs in the traditional lifeline engineering domain, they present a mathematical modeling approach based on a user equilibrium model to identify the mobility-related impacts of degradation to system components. They only use a toy network model for their numerical example, however, consideration of variable demand based on capacity fluctuations and the use of travel costs and system surplus as performance measures are valuable system-based details in their work. Murray-Tuite and Mahmassani (2004), in their two-player game-theoretic approach (non-zero-sum game between an evil entity and a traffic management agency) to identifying critical components, define a vulnerability index accounting for the availability of alternate paths, excess capacity, and travel time. To set the stage for the network robustness index they define, Scott et al. (2006) criticize conventional infrastructure management practices based on local Level-of-Service (LOS) measures (e.g, Volume/Capacity ratio) calculated at the link level. The authors argue local measures are misleading in determining areas of improvement in the network and illustrate the problem with a simple example (See Figure 2). In Figure 2.1 Example illustrating critical links using V/C rsatios (adopted from Scott et al. 2006)., Link 2 appears to be 14 the more critical link-based only on the V/C ratios, however, it is obvious that Link 1 is more critical since Link 2 cannot accommodate the rerouting of 3 units of volume in the case of a Link 1 closure. Based on this insight, Scott et al. define a Network Robustness Index (NRI) for evaluating the critical importance of a given highway segment (i.e., network link) to the overall system as the change in travel-time cost associated with rerouting all traffic in the system should that segment become unusable. NRI takes into account the spatial relationships and rerouting possibilities associated with the network topology, the OD demand and the capacity of individual highway segments. Again, the authors use a hypothetical network for demonstration. It is essential to note that calculating NRI for every link (or node) can be computationally infeasible in the case of a real network of a metropolitan area as it would require running a traffic assignment model for every link or node removal. Still, the demonstrated systems-level thinking by the authors is valuable and the ideas were adopted in the research that followed. Figure 2.1 Example illustrating critical links using V/C rsatios (adopted from Scott et al. 2006). Sullivan et. al. (2010) advance the travel time/cost understanding and improves the NRI to be used in partial link closures as opposed to binary treatment of failures. Using the reciprocals of the travel costs, Nagurney and Qiang (2007) define a measure of importance that is feasible in the case of non-connected networks. This may be an advantage in the sparse, abstract regional networks, however, networks in urbanized areas are highly redundant and non-connectedness is rarely manifested. Rupi et al. (2015) 15 assert that link importance should be based on two components, average daily traffic on the link in business-as-usual conditions with no disruptions (local importance) and the increase in total travel cost in the study region if the link is closed (global importance). If the link is a cut link, i.e., its closure is disconnecting segments of the network, another cost is added to the link proportional to the unsatisfied demand. For brevity, similar works on vulnerability (or robustness) indices based on travel time/cost and focusing on the identification of critical links are excluded. Another direction of research focuses on changes in accessibility to investigate vulnerability. Accessibility is defined as the ease by which individuals from specific locations in a region may participate in activities (e.g., employment, education, shopping, trade, and commerce) that take place in other physical locations in and around the region and by using a transport system to gain access to those locations (Taylor et al. 2006). Based on this definition, Taylor et al. characterize vulnerability as follows. A network node is vulnerable if loss (or substantial degradation) of a small number of links significantly diminishes the accessibility of the node, and a network link is critical if loss (or substantial degradation) of the link significantly diminishes the accessibility of the network or particular nodes; both measured by a standard index of accessibility. Stemming from this conceptualization, the authors quantify generalized travel cost, as well as two different accessibility measures to identify the vulnerability of nodes and criticality of links in the strategic level (regional and national) networks of Australia. The accessibility perspective reveals itself in Chen et al. (2007) as well where authors use a combined travel demand model (4-step) to investigate long-term consequences—quantified in terms of utility- based accessibility measures—of link closures. Approaches discussed so far on identifying critical nodes or links apply to disruptions resulting from the degradation of a single or a pre-defined number of components in a network. In reality, natural hazards (e.g., earthquakes, floods, etc.) have spatially distributed impacts and different combinations of damaged components (i.e., closures) expectedly create different ‘disrupted’ mobility patterns. Furthermore, computing reduced accessibility—based on traffic assignment—for every node and link removal may be feasible for sparse 16 regional/national level networks similar to the Australian highway network used in Taylor et al. (2006), however, this is not the case for dense metropolitan areas modeled with high-resolution networks. This brings the line of thinking to yet another direction of research focusing on evaluating the vulnerability for specific regions and hazards. Based on a GIS methodology leveraging open-source network data, Bono and Gutiérrez (2011) analyzed the accessibility impacts of the Haiti earthquake. Also in this niche, considering their relevance to the thesis, studies on economic impact analysis of transportation disruptions are especially important. In what follows, a review of such works is presented in a separate section before research gaps are discussed in detail. Economic Impact Analysis of Transportation Disruptions Hazards disturb transportation systems at an increasing rate and severity to cause physical damage leading to losses in the functionality of transportation infrastructures. These physical damages result in direct economic losses that diffuse and expand continually through economic activities between different regions and industries, exacerbating the totality of losses. For instance, it is estimated that a hypothetical disruption of the Seikan Tunnel in Japan can cause 1.33 billion dollars losses to China, Korea, and other regions based on a transnational and interregional input-output model (Irimoto et al. 2017). Therefore, investigating the economic impacts of hazards beyond the immediate (direct) losses and studying the diffusion of the impact among industries and regions is critical. A comprehensive accounting of total losses (including direct, indirect, and induced costs) requires the incorporation of interindustry economics in the form of economic impact analysis models. Economic impact analysis is widely used to estimate economic losses due to natural and man-made hazards. In the current state of economic impact analysis research, input-output models (IO) and the computable general equilibrium (CGE) models are the most common approaches. In this domain of economics research, Cochrane (1974) pioneered the use of interindustry economics in disasters context focusing on 17 earthquakes. Hallegatte (2008) proposed the Adaptive Regional Input-Output model and applied the adaptive IO measures to the impact assessment of Hurricane Katrina. Park et al. (2005) and Park (2008) constructed coupled demand-driven and supply-driven regional input-output models based on IMPLAN and CFS data and applied them to the evaluation of the hypothetical terrorist attacks on civil infrastructure. In CGE modeling for disaster economics, Rose (2004) and Rose and Liao (2005) estimated the regional economic impacts of water supply disruptions using a CGE model and considered resilience measures. Other researchers incorporated non-economic methodologies such as the Inoperability Input-Output model (Crowther et al. 2007). To take a snapshot of the status quo of this domain, a literature review was carried out by the author and his collaborators (Wei et al. 2018). Prior research that focuses on estimating the economic impacts of transportation disturbances was retrieved from Web of Science. An initial list of papers was derived by searching various sets of keywords and keyword groups such as “economic losses”, “hazard”, “disaster”, “disruption”, “transportation”, “economic impact analysis”, and “supply chain disruption”. A threshold date of publication constraint was not set. However, some articles were excluded from the review based on the following criteria. First, articles written in other languages than English were filtered out. In addition, articles with explicit transportation network modeling but missing economic analysis were not included because these works only study the impacts of transportation disturbances from an engineering point of view, which are discussed in detail in earlier sections of the thesis. Lastly, studies investigating the economic impacts of the disruptions to other kinds of infrastructure systems (e.g. power plants or water supply system disruptions) or articles without a clear analysis of the losses resulting from the disturbance of the transportation sector were removed from the review inventory (Aloughareh et al. 2016; Koks et al. 2016; Koks and Thissen 2016). This way, 25 articles were identified, and another 17 articles were found by surveying the references of the original 25 or by going through their listed publications of the authors. Out of the 42 publications identified in the end, 37 were peer-reviewed journals, 5 were conference papers, and 1 was a technical report. 18 The publications in the review inventory were categorized based on the following three dimensions. (1) Scope of Network Modeling and Analysis, to identify whether the article accommodates an explicit transportation network modeling approach; (2) Scope of Hazard Impact Information, to distinguish the articles based on the damage data used to determine component degradation in the network. For this scope, the categorization is (i) simple assumptions for hazard impacts, (ii) reported or reviewed impacts, (iii) impacts found from realistic hazard simulations, or (iv) no hazard impact information. The last dimension is (3) Scope of Economic Modeling and Analysis, to identify the methodologies of economic impact analysis used in the articles. It should be noted that all articles were placed into the six categories that are possible combinations of the first two scopes. The third scope was used in each of the six categories to determine the scope of economic modeling. The results of this scheme are presented in Table 2.1 and Table 2.1 Reviewed studies not accommodating explicit network modeling and analysis. Scope of Hazard Impact Information Simple Assumptions Reported/Reviewed Impacts Realistic (Simulated) Hazard Impacts Scope of Economic Modeling Simple Math (Gueler et al. 2012) (Jaiswal et al. 2010) (Zhang and Lam 2015) (Tan et al. 2015) (Kajitani et al. 2013) (Zhang and Lam 2016) (Oztanriseven and Nachtmann 2017) IO (Park et al. 2005) (MacKenzie et al. 2012) (Park 2008) (Tokui et al. 2017) (Park et al. 2008) (Li et al. 2013) (Rose and Wei 2013) (Irimoto et al. 2017) IIM (Santos and Haimes 2004) (Yu et al. 2013) (Lian and Halmes 2006) (Wei et al. 2010) (Pant et al. 2011) (Thekdi and Santos 2016) CGE (Xie et al. 2014) SCGE (Ueda et al. 2001) 19 (Tatano and Tsuchiya 2008) CGE; IO (Rose et al. 2016) Others (Thissen 2004) Table 2.1 Reviewed studies not accommodating explicit network modeling and analysis. Scope of Hazard Impact Information Simple Assumptions Reported/Reviewed Impacts Realistic (Simulated) Hazard Impacts Scope of Economic Modeling Simple Math (Xie and Levinson 2011) (Mesa-Arango et al. 2016) (Postance et al. 2017) (Ashrafi et al. 2017) (Zhou et al. 2010) (Omer et al. 2013) (Vadali et al. 2015) IO (Park et al. 2011) (Cho et al. 2001) (Cho et al. 2015) (Gordon et al. 2004) (Sohn et al. 2003) CGE (Tirasirichai and Enke 2007) SCGE (Tsuchiya et al. 2007); (Kim and Kwon 2016); Econometric (Greenberg et al. 2013); Table 2.2 Reviewed studies accommodating explicit network modeling and analysis. Considering the economic modeling approaches used in the reviewed studies, most articles only present an estimation of the “direct” impacts by simple mathematics. These articles do not take inter-industry diffusion effects or inter-regional economic activities into consideration. Among the articles with formal economic impact estimation methodologies, IO modeling and IIM modeling are widely used approaches. In addition, there are several 20 examples of CGE and Spatial CGE models (SCGE) as well. However, most of these works leverage hypothetical hazard scenarios as the basis of their economic impacts analysis. With respect to the hazard impact information that is incorporated into the studies, most of the articles are based on simplified assumptions such as the shutting down of a port over a week due to a hypothetical hazard. This type of approach does not utilize a sophisticated understanding of the hazard. Also, only a small subset of the articles in the review inventory carry out their economic analyses based on reviewed or reported hazard information, i.e. hazards that have occurred in the past with documented and reported impacts. Out of the 42 publications reviewed, only 7 incorporated hazard impact information derived from realistic hazard simulations. Among them, 5 articles incorporated explicit network modeling. Zhou et al. (2010) calculated the social costs of drive delay and loss of opportunity caused by the degraded network under a set of earthquake scenarios to evaluate the social-economic effect of seismic retrofit of bridges. Sohn et al. (2003) evaluated the significance of several bridges by quantifying the economic losses due to the 1812 New Madrid earthquake. Postance et al. (2017) conducted economic estimation for scenarios of road segment disruptions simply by multiplying increasing travel time with national user generalized cost without considering any ripple effects of transportation disturbances. Cho et al. (2001) and Gordon et al. (2004) estimated direct, indirect, and induced economic losses of Elysian Park earthquake scenarios. Cho et al. (2001) modified the original Southern California Planning Model (SCPM1) to develop an integrated, operational model that measures losses due to earthquake impacts on transportation and industrial capacity, and how these losses affect the metropolitan economy. This is one of the rare truly interdisciplinary efforts with an overarching approach despite the limitations of the economic and transportation models used. On the other hand, it is noticed that articles without transportation network modeling outnumber articles with explicit network modeling. Within the subset of articles that do not accommodate explicit network modeling, only Gueler et al. (2012) and Park (2008) investigate multimodal issues (e.g. waterway, rail, and truck). Among articles with 21 explicit network modeling, most focus on the calculation of the direct transportation- related costs such as increased travel or warehouse costs. Few studies estimated the indirect economic losses based on the direct losses (i.e. decreased proportion of initial production or demand), which were hypothetical or simply set according to historical records. Next, research gaps in system-based vulnerability studies are discussed from 3 perspectives. Research Gaps in System-Based Vulnerability Analysis: 3 Perspectives It is asserted—despite the extensive data and modeling requirements—that system- based approaches provide the opportunity to capture the realities of transportation disruptions holistically while keeping desired granularity in analyses intact and allow for collaborations across disciplines and stakeholders to translate the outcomes of disaster research in different silos to actionable insights for decision-makers. However, not utilizing recent advances in transportation demand modeling and increased availability of network and inventory data as well as simulation platforms, studies in system-based vulnerability have key shortcomings that prevent researchers from fully capturing transportation disruptions. Moreover, frequently used economic models for impact analysis are outdated, and economic impact analysis studies suffer from several shortcomings. In what follows, shortcomings in all three research silos are discussed as they apply to transportation disruptions. These gaps motivated the collaboration behind the proposed research. Shortcomings in Transportation System Analysis In the transportation systems analysis context, previous efforts have particularly fallen short in two major dimensions. First, post-disaster travel behavior based on the degraded network is largely treated as a mystery due to a lack of open, reliable, and high-resolution mobility data for post-disaster situations. In the case of catastrophic earthquakes, waiting for the disaster to happen to collect mobility data is not an option. However, research should not refrain from utilizing existing demand models to predict post-disaster traveler behavior, e.g. (Chen et al. 2007) even if the predictions depend on what-if type 22 assumptions that pave the way for sensitivity analyses. This is especially feasible for developed countries where earthquakes do not change travel patterns profiles as drastically as developing countries that are less prepared (Khademi et al. 2015). For example, if it is not viable to develop a calibrated model for the stay-at-home behavior in a setting of disaster-related network functionality loss, can researchers devise sensitivity scenarios based on behavioral assumptions to understand the potential improvements in resilience? Addressing this shortcoming has not been part of the author’s dissertation work and was left out as an item of future work. Second, there is a lack of holistic and granular network modeling representing the actual infrastructure inventory present in large metropolitan areas. This shortcoming is the result of an over-simplified physical abstraction of the transportation networks when they could be modeled explicitly (e.g., modeling freeways only and neglecting arterials or surface streets) (Miller and Baker 2015). Abstract network models do not allow the incorporation of realistic and locally relevant hazard simulations into the analyses. Abstraction or simplification may be acceptable for regional or national networks (e.g., US Interstate System) that are sparse, however, dense networks in metropolitan areas need to be modeled completely not to blind assessments to the inherent redundancy (key enabler of resilience) of transportation systems. Broader human mobility research has this limitation as well. An abundance of studies and models on human mobility assumes that human movements happen in a continuous space without considering constraints and/or gaps (Wang and Taylor 2016). Such a limitation causes not only an underestimation of the role of networked transportation infrastructure but also the neglect of mobility constraints resulting from damaged infrastructure. In addition, most of the studies focus on single link failures and/or a single mode of transportation. Nagae et al. (2012) point out Asakura’s (2007) suggestions on network models to be utilized in vulnerability research. According to Asakura, a network model developed for an ordinary network state should be modified and applied to the recovery state of a network, and the network flow model should have the characteristics of explicit link capacity constraints, decreasing demand due to traffic congestion and the uncertainty of a traveler’s choice behavior. As discussed, 23 such models have been rarely used in the area, and most studies use fixed demand assumptions and simplistic networks. Lastly, there is a lack of attention towards equity issues of vulnerability. Vulnerability and resilience assessments usually focus on the travel cost-related consequences of transportation disturbances and quantify network functionality indicators such as increasing travel times and distances. The worsening in terms of such indicators results in environmental impacts (e.g., surging emissions due to increased use of vehicles) that are less focused on. This secondary level of impacts calls for research, especially in the context of environmental justice and transportation equity. Equity can also be discussed in terms of reduced accessibility or financial losses. Shortcomings in Hazard Characterization and Damage Assessment Before the discussions on hazard analysis shortcomings in the literature, it is critical to highlight that formal consideration of the hazard itself and detailed inventories of the infrastructure systems are rarely included in vulnerability studies. This is partially due to the research objectives. If identification of critical links and nodes is the sole objective independent of the type of hazard (Scott et al. 2006; Taylor et al. 2006), then hazard analysis is practically omitted. However, if the objective is to evaluate the vulnerability of a network to earthquakes, floods, etc., formal hazard analyses and damage assessment procedures need to be incorporated. The shortcoming here is that researchers traditionally resorted to what-if assumptions to determine physical damages and component failures. Khademi et al. (2015) find in their review that many studies look at failures without their causes and focus on the failure of a single, hypothetical link. This treatment of network degradation is not founded on the ability of network components to meet demands from hazards, thus they are generally limited to more academic applications. It is important to acknowledge that what-if analyses might be the only viable option for certain events, e.g., terrorist attacks. However, in the case of earthquakes, tsunamis, or floods, there is a large body of research to be benefited from. In this section, the shortcomings of hazard analyses carried out to inform transportation vulnerability 24 assessments are discussed to shine a light on the contributions of the model-based methodology presented later which plugs into the proposed resilience framework discussed in the following sections. In the context of damage assessment, significant shortcomings exist in engineering representations of the infrastructure serving the transportation networks. Due to the inherent uncertainty in the number, sizes, and locations of future earthquakes (or other natural hazards), traditionally, the seismic hazard at a site is defined probabilistically, in terms of return periods (Cornell, 1968). A direct consequence of this approach is the use of probabilistic functions, namely fragility functions, to connect the seismic hazard defined as an intensity measure (IM) to the probability of damage to a network component as outlined in Baker et al. (2015). In analyzing complex spatially distributed systems such as transportation networks, the typical assumption made for the bridge and tunnel fragility functions is that they can be grouped into archetype structures (Kiremidjian et al., 2006; Fabozzi et al., 2018). This simplification may be warranted for tunnels, given they constitute a diminutive portion of network components, hence in developing their fragility functions, there is a better chance of capturing the actual engineering properties by averaging. However, this is proved to be inapplicable to bridges. As reported by Jeon et al. (2016) and Soleimani et al. (2017), geometric properties such as the column height and shape, horizontal curvature, and abutment skew significantly impact the fragility functions of bridges. Nonetheless, widely used HAZUS (1999) fragility functions do not consider any of these geometric features in formulating a relationship between IM and damage probabilities. As a result, studies based on HAZUS relationships are expected to result in inaccurate — most likely, as evidenced in Kircher et al. (2006) for buildings, over- conservative — estimation of system performance. Accurate representation of site conditions is another area requiring improvements. The standard approach in defining site properties such as average shear-wave velocity for the upper 30-m, V s30, liquefaction susceptibility, etc. for spatially distributed systems is to resort to proxy data sources such as topographic or geologic information without consideration to open access in-situ measurements. Such renditions may introduce notable degradation in the precision of 25 ground-motion simulation or analysis results (Thompson et al., 2014), hence reduce the overall quality of network performance evaluations (Koc et al. 2020). Shortcomings in Economic Impact Analysis of Transportation Disruptions In socioeconomic impact assessments of transportation disruptions, economic modeling techniques are key to complementing the ‘engineered resilience’ with the quantification of service/business disruption perspectives, i.e., investigating infrastructure systems not only as civil systems and physical components of the built environment but also as assets delivering essential goods and services to maintain economic output up (export producers) and down (import providers) supply chains. Based on the review presented in 2.1.3, economic impact analysis literature related to transportation disruptions focuses exclusively on individual components of infrastructure systems (e.g., a bridge instead of the road network, a port instead of the coupled seaway-inland network). This leads to the inability to incorporate the spatially distributed and networked nature of civil infrastructures into the impact assessments. This is a major shortcoming of the works in this domain leading to the omission of the fact that components of infrastructure systems are fundamentally dependent on the status of the network to carry out the desired functions. Thus, e.g., if one studies the impact of an earthquake on a single freeway bridge or even a small group of bridges that do not represent the bridge network in an urban area realistically, the analysis cannot produce actionable insights related to the overall economic impacts. This is because when an earthquake (or any other natural hazard with spatially distributed impacts) hits an urban area, it affects a wide area and exposes the entire urban transportation network due to the possible propagation of failed individual components. Moreover, economists focusing on resilience in the transportation systems disruptions context typically resort to premature assumptions about ‘resilience tactics’ 8 and their impacts on system 8 These are tactics that are considered significant due to their potential in reducing economic losses, e.g, ship rerouting, effective road infrastructure asset management, cargo prioritization etc (Wei et al. 2020b). 26 performance, mainly due to a lack of a detailed and explicit model of the system of concern. In terms of the interindustry economic models used for impact analysis, Input- Output analysis (IO) has conventionally been used as the go-to modeling approach despite well-known limitations. IO models are characterized by a linear and rigid response, lacking behavioral content (Rose and Liao 2005) that is pivotal in the course of disruption response. Resilience tactics, such as input substitution, conservation of resources, importing critical inputs, making use of inventories, etc. are not modeled in IO. In summary (see Table 2.3), only a few researchers conducted comprehensive economic analyses based on integrated transportation network modeling and realistic hazard simulations, especially focusing on the economic impacts of urban mobility disruptions. Simplified assumptions were widely used, and most of the research in the area did not leverage explicit transportation network modeling during this process, which compromised the reliability of the research outcomes. Naturally, this led to a lack of attention towards the fine-resolution analysis of urban transportation disturbances coupled with economic impact analysis. Lastly, more advanced economic modeling techniques used for regional science are not utilized. 27 I. Transportation System Analysis § Lack of holistic and granular network modeling representing actual transportation infrastructure § An abundance of work in single link failures and/or a single mode of transportation. Limited for spatially distributed impacts and multiple modes § Post-disaster travel behavior treated as a mystery § Lack of attention to transportation equity-related consequences and environmental impacts II. Seismic Hazard Characterization and Damage Assessment § Omission of formal hazard considerations, a common tendency is to look at failures without causes § Omission of structure-specific details as fragility analyses predominantly use archetypes § Misrepresentation of site-specific details such as shear-wave velocity, liquefaction susceptibility due to lack of in-situ measurements III. Economic Impact Analysis § Predominantly focus on individual component failures instead of a systems-level analysis § Commonly used IO models have well-known limitations: they are linear, have a rigid response, and lack behavioral content § Rarely use explicit network modeling and formal hazard considerations, both in impact assessments and in investigations of resilience tactics § Lack of attention to disparities across income groups or impacts on racial/ethnic minorities Table 2.3 Shortcomings in system-based analyses of transportation disruptions: 3 perspectives. 28 Recent Advances in Modeling Recent advances related to structure-specific and site-specific details in component modeling for regional damage assessment, and increased availability and standardization of spatial (network) data, as well as the development of metropolis-scale travel demand models for transportation systems analyses, are paving the way for improvements in the investigation of transportation disruptions. From a component modeling viewpoint, significant improvements can be made to bridge fragility functions if the geometric and structural traits for individual structures are considered in detail (Cetiner et al. 2019). With the recent advancements in automated model generation based on LIDAR data, photogrammetric reconstructions (Morgenthal et al. 2019), Computer Vision methods capable of extracting bridge models from street-level photographs (Cetiner 2020; Koc et al. 2020), and statistical studies on the moments, i.e. mean, variance, etc., of bridge structural properties (Mangalathu 2017), it is possible to attain substantial enhancements in the bridge fragility functions. In many cases, these automated procedures are capable of the capturing structural response of bridges within a negligible margin of error. In addition, as illustrated by Thompson et al. (2014), the site conditions at bridge locations can be defined at greater detail by constraining the proxy-based estimations of site characteristics with the public domain geologic, geophysical and geotechnical data. Thus, numerous site measurements that are otherwise disregarded in evaluating network damage can be effectively incorporated into the analyses. As a consequence, the damage induced by the principal damage mechanisms, e.g., ground shaking, liquefaction and surface rupture, etc. can be better estimated. From a network modeling viewpoint, recent advances in GIS technology and increased availability of public domain sources for network data (e.g., OpenStreetMaps) together with standardized data formats (e.g. the General Transit Feed Specification) enable holistic modeling of multi-modal transportation networks of many urban areas around the world. Therefore, research in the area needs to advance towards utilizing real 29 scale networks that can incorporate spatially distributed impacts of catastrophic events rendering realistic insights possible for metropolitan transportation planning. In terms of travel demand modeling, large scale models are developed and maintained for planning purposes in many metropolitan areas in the US. The author believes that collaborations are necessary to bring governmental agencies (e.g, metropolitan planning organizations, DOTs, etc.) and researchers together. These agencies, develop, calibrate and validate state-of-the-art models, having gathered large amounts of data to model travel demand in their respective locales. Such models are rarely utilized for vulnerability and resilience studies. In terms of economic impact analysis, CGE analysis used for regional economic modeling—particularly for policymaking and impact analysis—is promising to improve upon the shortcomings of the IO analyses. The approach is based on a multi-market simulation model based on simultaneous optimizing behavior of individual consumers and firms, subject to economic account balances and resource constraints (Rose 2004). CGE analysis is a competing approach to Input-Output analysis (IO) which has the ability to model the mentioned resilient actions such as input substitution, conservation of resources, importing critical inputs, making use of inventories, etc. Its capabilities make CGE analysis a more promising alternative in policy planning and disaster impact analysis. CGE analyses are applied extensively on infrastructure systems. Literature in the area provides findings related to transportation systems (Lofgren and Robinson 2002), energy policy (Schumacher and Sands 2007), water infrastructure (Rose and Liao 2005) and so on. There has been recent efforts towards similar objectives, i.e. applying CGE analysis to interdependent systems (Wittwer 2012a; Zhang and Peeta 2011, 2014), and although there are concepts in common, this section of the thesis focuses solely on hazard impact analysis and entertains resilience as a key concept. Hence, to the author’s knowledge, the economic facet of the framework proposed here is a pioneering effort in studying resilience of networked infrastructure systems resilience from both engineering and economic perspectives. 30 Towards Comprehensive Resilience Assessment As mentioned, research on disrupted transportation stems from and has been dominated by investigation of consequences, i.e. vulnerability approaches. Closely related disaster related measures such as robustness and survivability have also been used to quantify consequences of hazards either from a weakness (vulnerability) or a strength (robustness, survivability) point of view. It is significant to put these concepts into perspective. Faturechi and Miller-Hooks (2014) argue that selection of an appropriate disaster measure for a particular application is an important first step in system analysis, and they list the disaster measures used for transportation systems as risk, vulnerability, reliability, robustness, flexibility (also known as agility and adaptability), survivability, and resilience). Figure 2.2 illustrates the boundaries and interactions between them. One insight obtained from their delineation of the concepts is the far reach of resilience to all of the system capabilities relating to hazards. For an engineered system, resilience is defined as the ability of the system to reduce the chances of a shock, to absorb a shock if it occurs (abrupt reduction of performance) and to recover quickly after a shock (re-establish normal performance) (Bruneau et al. 2003), i.e., resilience builds on system’s strengths and weaknesses measured by risk, vulnerability, reliability, robustness, and survivability (i.e., resistance) and adaptability measures, while also encapsulating the benefits the system’s ability to adapt to post-disaster circumstances as in flexibility measures (Faturechi and Miller-Hooks 2014). 31 Figure 2.2 Disaster measures, their boundaries and interactions. Adopted from Faturechi and Miller-Hooks (2014). From multiple reviews of the literature (Faturechi and Miller-Hooks 2014; Khademi et al. 2015; Mattsson and Jenelius 2015), the conclusion drawn is that the literature on transport system resilience is less extensive in comparison to the works on vulnerability. Review articles also identify the limited signs of adoption of vulnerability related work by practitioners, planners and decision makers despite the abundance of studies. In this context, resilience is a promising disaster measure by encapsulating other system capabilities and by covering the entire disaster timeline from disaster mitigation to responae and recovery. This is a key motivation of the proposed research. In their seminal paper, Bruneau et al. (2003) define resilience for an engineered system as “the ability of the system to reduce the chances of a shock, to absorb a shock if it occurs (abrupt reduction of performance) and to recover quickly after a shock (re- establish normal performance)” and use it as a measure that quantifies system performance over time given the abrupt changes in performance, followed by a gradual restoration to normal performance levels. This idea has been adopted by other researchers and the curve illustrating the concept on Figure 2.3 is universally applied in different forms. In subsequent sections, the application of the concept to CRAFT is discussed in more elaborate detail. 32 Figure 2.3 Measure of seismic resilience—conceptual definition. Adopted from Bruneau et al. (2013). Research Objectives The mentioned shortcomings of network vulnerability analyses (topological and system- based), and the general lack of attention to resilience are obstacles in front of research, especially in regards to developing results at a level of refinement expected by planners, practitioners, and operators. To bridge the gap, the following research objectives and questions are identified. These are partially achieved and answered, still there is a significant amount of work to be done. In terms of hazard characterization and damage assessment, the research objective was to integrate a model-based methodology that would offer improvements in the representation of infrastructure inventories. This objective is achieved thanks to a collaboration between the author and researchers at University of California, Los Angeles. The novel model-based methodology is currently used for bridge modeling and it is discussed briefly in the following section. In transportation network analysis, the research objective was to use a metropolis- scale demand model for resilience assessment. The key advantage of the model is the high resolution network modeling its founded on without compromising the metropolitan scale and it has many features that are drastic improvements over the more simplistic 33 models used in the literature with more abstract networks 9 . The regional travel demand model has never been used for a resilience assessment. Research questions in this truss include: “How to model for the degradation in the network model given damage assessment results?”, “How to consistently quantify system resilience across different network versions resulting from a specific hazard scenario and across different hazard scenarios?”, “What are the dynamic resilience tactics that can enhance system resilience and how to model identified tactics?”, “What are the effects of fixed demand assumptions versus supply sensitive demand conditions on measured resilience?”, “How to calibrate the demand models to mimic post disaster travel behavior?”, “How to analyze environmental justice and transportation equity issues given a hazard scenario?”, “What are the ways to integrate transportation results with economic impact analysis methods?”. The current version of CRAFT introduced in the next section is able to answer some of the mentioned questions and is receptive to improvements that can answer the remaining ones. The latter are elaborated on as items of future work at the end of Section 2. In terms of economic impact analysis and the treatment of economic resilience within the scope of this research, the research objective was to integrate advanced computable general equilibrium (CGE) approaches that are used for regional policy analysis into the framework to achieve formal considerations of socioeconomic impacts. CGE models have well established advancements not present in the commonly applied, conventional IO analysis. The author and his dissertation advisor initiated a collaboration within University of Southern California to achieve this research objective and received funding support from Caltrans to establish the linkages between the research silos of ‘engineered resilience’ of transportation systems and ‘economic resilience’. The high- level research question associated with this objective is stated above and the progress on this facet of the work is presented broadly in section 4.4. and a vision is described in Section 6. 9 Other features of the Southern California Assoc. of Governments (SCAG) Regional Travel Demand Model are introduced in 4.3. 34 CRAFT is founded on a cross-disciplinary collaboration to address the shortcomings in the literature and the author has enjoyed fantastic contributions from his collaborators. In the next section, the author introduces the proposed framework. CRAFT: Comprehensive Resilience Assessment for Transportation Systems in Metropolitan Areas In the light of the discussions on the background of transportation disruptions, perspectives presented on the shortcomings of research in this area and leveraging the mentioned advancements, a comprehensive framework, CRAFT, is designed for the assessment of resilience in metropolis-scale transportation networks. A modular approach was pursued in which—coming back full circle on Goldberg (1975)—modules analyze corresponding portions of the problem with rigor and precision while the convergence captures a view of holism/operationality/generality for resilience at 3 levels (See Figure 2.4). The framework couples a novel structure-specific modeling methodology with a metropolis-scale travel demand model, and uses results thereof to inform an advanced economic impact analysis methodology. By virtue of its data-intensive and model-based nature, the proposed approach is capable of capturing and incorporating many details that are usually neglected in traditional analyses, enabling an improved thoroughness in the estimation of resilience and sustainability metrics of transportation networks. As the test-bed for the proposed framework, a study region aligning with the Southern California Association of Governments (SCAG) area of responsibility including 6 counties in Southern California (Imperial, Los Angeles, Orange, Riverside, San Bernardino and Ventura) is used. 35 Figure 2.4 A conceptual illustration of CRAFT: Comprehensive Resilience Assessment Framework for Transportation Systems. 36 Resilience at System Component Level: Hazard Characterization and Damage Assessment 10 CRAFT is not hazard-specific. The methodology behind the framework is, by design, capable of accommodating resilience assessment for various hazards in a coupled or decoupled manner. Given a comprehensive description of the steps required for representing each type of hazard cannot be realized within the space limitations of this work, in the following, only the seismic characterization approach used for the case study discussed in the next section is described. Characterizing seismic hazards for a region require translating the knowledge of potential seismic sources into simulations of realistic and locally relevant deterministic hazard scenarios. The seismic demands resulting from these hazard scenarios are then used to predict damage to—and ultimately, the recovery of—network components. The procedure for converting seismic hazards to direct inventory damage consists of three principal components: (i) quantifying deterministic seismic hazard governing an urban transportation network, (ii) coupling the intensity measures (IMs) resulting from the seismic hazard with component fragility functions to estimate the damage state probabilities, and (iii) combining the damage state probabilities with restoration functions to calculate the downtime for each network component (Koc et al. 2019a). The first step in quantifying the deterministic seismic hazard is identifying the scenario earthquake(s) controlling the overall seismic hazard for the network—also known as probabilistic seismic hazard analysis (PSHA) (McGuire 1995). The regular practice in determining the fault line, location, and magnitude associated with a scenario earthquake is ”deaggregating” the probabilistic seismic hazard (PSH) results so that the relative contributions of all seismic sources to that hazard are displayed for all possible magnitude and distance measures (Bazzurro and Cornell 1999). Once the scenario earthquake is 10 This portion of the joint effort on CRAFT (image-based modeling of bridges) has been led by collaborators at UCLA’s Taciroglu Research Group, Cetiner and Taciroglu, co-authors on Koc et al. (2020). 37 identified, it is possible to calculate the rupture length of the fault segment using the Wells- Coppersmith relationship 𝑙𝑜𝑔10 𝐿 = 𝑎 + 𝑏 · 𝑀 (1) where L is the rupture length in km, M is the moment magnitude of the earthquake, and a and b are the regression coefficients (Wells and Coppersmith 1994). The final step in computing the deterministic hazard is to calculate the IMs resulting from the defined earthquake event. This is typically performed by passing the source information to ground motion prediction equations (GMPEs), such as the ones developed through the NGA- West 2, and NGA East projects (PEER 2015; Stewart et al. 2016). In a GMPE, the natural logarithm of a ground motion IM (ln Y) is coupled with a source function (F E), path function, (FP ), site function (F S), and a residual term ε nσ(.) through the relationship 𝑙𝑛𝑌 = 𝐹 1 (𝑴,𝑚𝑒𝑐ℎ) + 𝐹𝑃 (𝑅,𝑴,𝑟𝑒𝑔𝑖𝑜𝑛) + 𝐹𝑆(𝑉 @AB ,𝑅,𝑴,𝑟𝑒𝑔𝑖𝑜𝑛,𝑧 D ) + 𝜀𝑛𝜎(𝑴,𝑅,𝑉 @AB ) (2) where M is the earthquake moment magnitude, R is the closest distance to the rupture plane in km, V S30 is the time-averaged shear-wave velocity in the top 30 m of the site in m/sec, z1 is the basin depth in km, mech is the fault mechanism parameter, and region is the regional correction parameter. The main outputs of GMPEs are the IMs peak ground acceleration (PGA), peak ground velocity (PGV), and 5%–damped elastic pseudo-absolute spectral acceleration (SA). Combining PGA with information such as liquefaction susceptibility, the IM peak ground displacement (PGD) at a site can also be calculated. Each of these IMs correlate with a particular type of seismic demand and the corresponding damage to individual network components. During an earthquake, the primary factors that contribute to bridge losses are ground and structural failures, which are typically well-correlated with PGA (Mackie 38 and Stojadinović 2001) and SA (Padgett et al. 2008), respectively. Damages to tunnels, on the other hand, are caused by ground shaking, ground failure due to liquefaction, fault displacement, or slope instabilities. As such, tunnel seismic damage levels are well correlated to (and thus can be described as a function of) PGA and PGD (Argyroudis and Pitilakis 2012). The relationship between network components and IMs are defined using fragility functions. Fragility functions are log normally-distributed functions that give the probability of reaching or exceeding different damage states for a given IM. In performance-based earthquake engineering (PBEE) practice, damage to a network component is categorized into five damage states, namely no damage (ds 1), and slight (ds 2), moderate (ds 3), extensive (ds 4), and complete (ds 5) damage states (Kiremidjian et al. 2006). Each fragility function corresponds to one of these damage states and is characterized by a median value IM (M), and a log-normal standard deviation value (β). The generic form of a fragility function is given by 𝑃𝑟G𝐷 I ≥ ds K L = 1 −φ P I (ln (𝑥 I /𝑀) 𝛽 (3) where k is the index for IMs, j is the index for PBEE damage states, D k is the damage state of network component due to IM k, Φ is the normal cumulative distribution function, and x k is the IM k at the site of the network component. Note that the probability of a system being in or exceeding the no damage, ds 1, state is always 1 (Pr(D k ≥ ds 1) = 1). An essential input for resilience assessment is the downtime estimates. Thus, for resilience studies, damage probabilities computed using fragility functions are converted into this metric. Translating bridge fragilities to downtime requires calculating the probability of network components being in one of the five damage states, and aggregating these probability measures to restoration functions that correspond to individual damage states. Open literature on restoration functions is particularly limited, and the restoration functions published by FEMA (FEMA 2003) are the main tool used for 39 tying component damage information to downtime estimates. For a set of IMs, the probability of a network component being in a damage state (P k j ) is calculated as 𝑃 P I ={ 𝑃𝑟G𝐷 I ≥ 𝑑𝑠 P L−𝑃𝑟G𝐷 I ≥ 𝑑𝑠 PXD L 𝑗 = 1,2,3,4 𝑃𝑟(𝐷 I ≥ 𝑑𝑠 ] ) 𝑗 = 5 (4) For a set of IMs, expected downtime (E[D k ]) is defined with respect to P k j as in 𝐸[𝐷 I ] = b 𝑃 P I · 𝑅𝐶 P ] PdD (5) where RCj is the recovery function corresponding to the damage state denoted by index j. As outlined by NIBS (NIBS 2002) and FEMA (FEMA 2004), HAZUS-type fragility functions based on archetype classifications have a tendency to overestimate physical damages for earthquakes with M < 6 and underestimate the damages for events with M ≥ 6. In other words, traditional fragility functions offer limited accuracy in predicting component-level disruptions. One of the novel contributions of CRAFT is its use of detailed bridge fragility functions generated using the image-based modeling approach proposed by Cetiner (Cetiner 2020). The method establishes structural models of bridges via a fusion of geotagged street-level and satellite imagery, OpenStreetMaps centerline curves (OpenStreetMap contributors 2019), 2018 version of National Bridge Inventory (NBI) metadata (FHWA 1995), Shuttle Radar Topography Mission (SRTM) 1 Arc-Second Global data (USGS 2015), and class statistics for bridge structural properties available in the literature (Mangalathu 2017). The central premise of the method is that through auto- 40 calibration from multiple uncalibrated street-level images, the camera matrix can be determined. Also, by semantic segmentation, individual components of a bridge can be identified in images. Subsequently, by measuring the distance between the back- projection of image origin and the image locations of components, world coordinates of the components can be computed and superimposed on the bridge centerline. Furthermore, object dimensions and deck height can be extracted, and a geometric model of the bridge can be computed. The geometric model can then be laid over the SRTM digital elevation model and populated with class statics to attain a comprehensive structural model of the bridge. Here, the described process is implemented using an in- house code developed specifically for this purpose (Cetiner 2020). Figure 2.5(a) shows a snapshot of Google StreetView images selected for one of the bridges modeled for this study. Figure 2.5(c) displays the 3D model generated for this bridge using the procedure summarize above. The process of obtaining bridge-specific fragility functions for ground shaking involves performing incremental dynamic analysis (IDE) (Vamvatsikos and Cornell 2002) of individual bridge models. By calculating the seismic response of a bridge for multiple ground motions at a range of SA 1.0 levels and comparing the demands determined for each realization against the corresponding damage thresholds for each damage, IDE computes the Pr(D k ≥ ds j). The choice of ground motion records and the damage thresholds for each damage state depends on a variety of factors. In selecting ground motions, using a dataset, which consists of waveforms from earthquakes with magnitude and distance measures compatible with the scenario earthquake, that covers a broad band of frequencies is crucial. The key aspect of determining damage thresholds, on the other hand, is to ascertain that they are applicable to the structural systems that comprise the bridge. In this study the ground motion dataset suggested by Baker et al. (2011) and the threshold values employed by Ramanathan et al. (2015) are utilized. Figure 2.5(b) shows the fragility functions obtained for the bridge model displayed in Figure 2.5(c) for the damage states ds2 to ds 5. 41 (a) Google Street View images (b) Bridge fragility functions (c) 3-D geometric model Figure 2.5 Several Google Street View images (a), fragility functions (b), and geometric model (c) obtained for a bridge considered in this study. 42 Resilience at System Level: Transportation System Analysis In the light of the presented perspective on shortcomings of system-based analysis of transportation disruptions, the novel image-based damage assessment methodology discussed in 2.10. is coupled with a transportation system analysis methodology to achieve resilience insights at the system level (See Figure 2.4.). Based on the results of the damage assessment procedure in the previous section, the analysis of the transportation network disruption is realized with a metropolis scale 4- step travel demand model. Specifically, functionalities of damaged system components are evaluated with respect to a link closure policy with a threshold parameter. This is done to replace the decisions based on post-disaster manual inspections in a consistent manner. In the case of bridges, if a bridge is damaged beyond the threshold, the link corresponding to that bridge in the network model underlying the travel demand model is—partially or fully—closed to operation. The restoration functions embedded into the damage assessment provide the information on the duration of closures. With this information, a number of network topologies (pre-disaster baseline and post-disaster degraded versions) are modeled to capture the network supply conditions throughout the disruption timeline (Initialization). Initial skim matrices are computed to find the OD costs for TAZ (Traffic Analysis Zone) pairs (Network Skimming). These costs inform Trip Generation where trip production and trip attraction models estimate the number of trips generated for all trip purposes (from and to all TAZs) which are then balanced and distributed throughout the region via different modes (Trip Distribution and Mode Choice). The calculated travel demand is then segmented into finer time periods (Time-of-Day Choice) and are used to assign the loads into the network to solve for the traffic assignment problem (Assignment). With the new ‘congested’ link costs, a new iteration begins with Network Skimming and this loop runs until convergence to user-equilibrium. CRAFT implements this methodology, for every network topology (pre-disaster baseline and post-disaster degraded versions). The methodology, however, may differ based on the modeling the travel demand after the initial disruption. Often in system-based analyses of transportation disruptions, 43 researchers assume fixed demand conditions in which the same trip matrices are fed into traffic assignment for different topologies. This way, an understanding of overall system functionality is gathered in settings where various levels of lesser network supply attempts to serve the same travel demand. In this case, the analysis employs trip generation, trip distribution, mode choice and time-of-day choice models only for the pre-disaster baseline network. Another option is to run trip generation, trip distribution and mode choice models based on the degraded topologies and try to capture the interaction between varying travel demand and network supply. However, since data on post-disaster travel demand and calibrated models thereof are largely incomplete (relative to the data and models for pre-disaster baseline settings), such analyses fundamentally depend on existing models of travel demand that are commonly generated from and calibrated to data from a typical weekday in the study region. Traffic assignment results for all network topologies allow for the assessment of network functionality until full recovery with respect to a business-as-usual (pre-disaster) baseline. This way, the system resilience curve can be drawn for the regional transportation system, shining light on 3 of the 4 resilience properties proposed by Bruneau et al. (2003): robustness, redundancy and rapidity. Investigating resourcefulness is traditionally more challenging due to the unknowns associated with the resources of transportation agencies and other stakeholders for response and recovery. However, CRAFT allows for sensitivity analyses focusing on optimal response and recovery interventions as tactics of enhancing resilience (Wei et al. 2020b). In the investigation of system resilience as an emergent capability for the transportation system, the following analytical resilience definition of Frangopol and Bocchini (2011a) is adopted here: 𝑅 = 1 ℎ ∫ f fXg 𝑄(𝑡)𝑑𝑡 (6) 44 where t is the instant in which the disruption occurs and h is the investigated time horizon and Q(t) is an indicator of system functionality. In this definition, resilience is quantified as the area under the functionality curve with respect to 100% functionality throughout the investigated time horizon. Integrating functionality over time in this manner gives the resilience of the networked system, R, to the specific hazard scenario. Revising the definition by Frangopol and Bocchini (2011), the authors define Q(t) relative to a baseline, Γ i(0), indicating system functionality with respect to indicator I on a typical day in pre- disaster settings, i.e. day 0. This is done to quantify the functionality of the disrupted versions of the system relative to a business-as-usual baseline. This way, the extreme case considered by Frangopol and Bocchini where all the bridges are out-of-service is also left out as it is not a realistic one for metropolis-scale systems. 𝑄(𝑡) = 1 − |𝛤(𝑡) – 𝛤(0)| 𝛤(0) ; 𝑄(𝑡) ∈ [0,1] ⊂ 𝑅 (7) A number of functionality indicators, Γ, are proposed in literature that are commonly quantified based on the total travel time spent or total travel distance covered in the system by all users. Other indicators such as average speed or emission levels can be used. Being centered around a detailed, high-resolution model of the transportation system, CRAFT allows for the quantification of virtually all such indicators. Γ i(t) is typically calculated as a sum over all links in the network at time t for indicator I for indicators such as VMT (Vehicle-Miles-Traveled), VHT (Vehicle-Hours-Traveled), Delay (Vehicle-Hours- Delayed). Other less visited indicators such average speed can be calculated as a mean over all the links. 𝛤𝑖(𝑡) = b 𝐹𝑢𝑛𝑐𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑡𝑦𝐼𝑛𝑑𝑖𝑐𝑎𝑡𝑜𝑟 s tsuIv (𝑡) ; 𝛤𝑖(𝑡) ∈ 𝑅 X (8) 45 SCAG Regional Travel Model The transportation system analysis component of CRAFT illustrated with the workflow in Figure 2.4 is currently implemented with the regional travel demand model (RTDM) developed by the Southern California Association of Governments (SCAG), as part of their regional transportation plan (Southern California Association of Governments 2019). The peer-reviewed model is developed and operated on TransCAD. The SCAG RTDM uses a number of datasets for modeling and validation purposes. For modeling, zonal data at the level of Traffic Analysis Zones (TAZs) including population and household characteristics (household size, annual income, education, employment etc.), travel survey, land use data are essential for estimating the travel demand in the region. Most of these data come from Public Use Microsample Data (PUMS), U.S. Census, California Department of Finance, California Employment Development Department (EDD), Land Use and County Assessor’s Parcel Database among other local data sources. Consequently, the model’s travel demand estimation components utilize 65 socio- economic variables and 8 joint distributions of two or more variables. For calibration and validation, California Household Travel Survey, Vehicle Miles of Travel (VMT) from Highway Performance Monitoring System (HPMS) and speed data from Freeway Performance Measurement System (PeMS) as well as transit on board survey and ridership data from L.A. Metro are used. The underlying network in the RTDM includes over 21,000 centerline miles of freeways, arterials and major urban collectors modeled with over 115,000 links (See Figure 2.6(a) for model resolution). Some of the features of the SCAG RTDM include an auto ownership model, advanced mode and destination choice models, a highly granular 2-tier TAZ (over 11,000 Traffic Analysis Zones in the 6-county SCAG Study Region) system for higher spatial resolution, trip market strata defined by car sufficiency and household income groups used throughout the entire demand models (10 trip purposes), an HDT (Heavy Duty Truck) model, a high-occupancy-vehicle (HOV) diversion model splitting carpool trips from vehicles on the general purpose lanes, and refined—with respect to the earlier model—congestion pricing components. To the author’s knowledge, it has not been 46 utilized for a study investigating resilience in a comprehensive manner. For detailed discussions regarding the data sources and the modeling efforts related to SCAG RTDM, readers are referred to the model validation report (Southern California Association of Governments 2019). Figure 2.6 (a) High resolution, multi-modal network model underlying the SCAG RTDM. Most recent version of the model is used by the authors with socioeconomic data from 2016. The model is highly granular and accommodates a holistic transportation network enabling a wide range of analyses including investigations of expansion projects, highway pricing strategies, introduction of new types of transportation services, etc. Figure 2.6(a) and Figure 2.6(b) show the level of detail in the network model and zoning, respectively. 47 Zoning relates to the development of geographically defined areas known as Traffic Analysis Zones (TAZs) containing socioeconomic data and other information for the model. In terms of topology, TAZs are connected to the network with centroid connectors that allow travelers (trips) to access the transportation system. In other words, TAZs provide the spatial units within which travel behavior is estimated in a data-driven fashion. In summary, the transportation system analysis discussed above is carried out with the SCAG RTDM. Investigating resilience with a metropolis scale model allows the framework to be applicable by policy makers, practitioners and operators. The right-hand-side transportation network analysis column in Figure 2.4 makes another interface with economic impact analysis and becomes the central piece of the comprehensive framework. This way resilience is quantified at component, system (network) and economic levels. In the next section, this second coupling that forms the overall framework is discussed. 48 Figure 2.6 (b) Zoning level of detail in the SCAG RTDM at Tier 2 level (11,267 traffic analysis zones). In summary, the transportation system analysis module in CRAFT is carried out with the SCAG RTDM which allows the framework to generate resilience insights at a scale and granularity directly translatable to the use of policy makers, practitioners and operators. As mentioned in earlier sections, this underlines the ability of CRAFT to bridge a major gap between transportation resilience research and policy-making. For example, current asset management practices do not account for system resilience while prioritizing maintenance, repair and restoration activities (e.g., bridge maintenance). Deployment of CRAFT could enable decision-makers to integrate data and model-driven resilience assessment into their planning activities. 49 Resilience at Regional Economic Level 11 Shortcomings in economic impact and resilience analysis were discussed in detail to highlight the advantages of CGE analysis in capturing resilience. In this section, a characterization of economic resilience is presented and discussions on the last integrated component of the framework are made. Rose (2004) borrows ideas from early definitions of the resilience concept for ecological systems and characterizes economic resilience in two main branches: Static Resilience refers to the ability of the system to maintain a high level of functioning when shocked (Holling 1973). Static Economic Resilience is the efficient use of remaining resources at a given point in time. It refers to the core economic concept of coping with resource scarcity, which is exacerbated under disaster conditions. Dynamic Resilience refers to the ability and speed of the system to recover (Pimm 1984). Dynamic Economic Resilience is the efficient use of resources over time for investment in repair and reconstruction. Investment is a time-related phenomenon—the act of setting aside resources that could potentially be used for current consumption in order to re-establish productivity in the future. Static Economic Resilience does not completely restore damaged capacity and is therefore not likely to lead to complete recovery. Rose (2004) also makes a distinction between inherent and adaptive resilience where inherent resilience refers to resilience capacity that is either already built into the system or that can be incorporated in advance of the disruption by enhancing resilience capacity through ‘pre-positioning’. On the other hand, adaptive resilience is exemplified by undertaking conservation that was not previously thought possible, changing technology, or devising new government post-disaster assistance programs. 11 This facet of the joint effort on CRAFT has been mostly led by Dr. Dan Wei and Dr. Adam Rose of USC’s Price School of Public Policy. 50 Various types of economic impact models—as discussed previously—have been used for economic resilience analysis. For spatially distributed infrastructure systems such as transportation, multi-regional CGE models are particularly advantageous due to the diffusion of economic losses regions other than immediate impact areas. In addition, for critical links of the global supply chains such as seaports, disasters could mean the disruption of imports and exports. Multi-regional CGE models are capable of analyzing the spatial allocation of the direct imports and exports and the spatial reallocation of the economic activity of direct users of the commodities up and down the supply chain. CGE models are the state-of-the-art models among all economy-wide modeling approaches used to study the economic consequences of a disaster. The CGE formulation incorporates many of the best features of other popular model forms, but without many of their limitations. For example, CGE models retain the major strengths of input-output models (full accounting of all inputs, multi-sector detail, and ability to capture interdependencies), but overcome the limitations of linearity, lack of behavioral content, lack of input and import substitution possibilities, and difficulty of incorporating resource constraints (Rose 1995). This modeling approach has been shown to represent an excellent framework for analyzing natural and man-made hazard impacts and policy responses, including disruptions of transportation infrastructure (Chen and Rose 2018; Rose 2017). In CRAFT, to investigate resilience at regional economic level, a multi-regional computable general equilibrium (CGE) model – TERM – is used to analyze the total economic impacts of and the cost effectiveness of resilience to seaport and highway transportation network disruptions. A CGE model is a multi-market simulation model based on the simultaneous optimizing behavior of individual consumers and firms, subject to economic account balances and resource constraints (see, e.g., Burfisher (2017)). TERM is a “bottom-up” multi-regional CGE model, meaning that it treats each region as a separate economy and then links regions through commodity imports and exports. The modeling structure of TERM is similar to that of other CGE models. Producers in each region are assumed to minimize production costs subject to a combination of intermediate 51 and primary factor inputs, whose relationship is structured by a series of nested Constant Elasticity of Substitution (CES) functions. A representative household in each region maximizes utility through purchases of optimal bundles of goods in accordance with its preferences and budget constraint. The model was custom built by the research team at the Centre of Policy Studies at Monash University and Victoria University in Australia (V. U. Center of Policy Studies (CoPS) 2019; Wittwer 2012b). It was designed specifically for the U.S. on the basis of IMPLAN regional input-output data (IMPLAN Group 2012), supplemented by various elasticities gleaned from the latest literature. The TERM database used for this study consists of 97 economic sectors and divides the U.S. economy into three interconnected regions (three-county Southern California Region, Rest of California, and Rest of the U.S.) to examine the spatial spread of disaster impacts. An important contribution of the CRAFT effort is to develop a methodological linkage between hazard characterization and damage assessment as well as transportation systems analysis modules, and the TERM CGE model (See Figure 2.4.). The following steps are used to enable this modeling linkage: 1) Reduced seaport capacity and damage to general building stocks resulting from the earthquake are determined by FEMA HAZUS 12 ; 2) Reductions in commodity supply and demand estimated based on the property damages in Step 1 are fed into the TERM CGE model to determine the general equilibrium impacts (broad multiplier effects including the implications of both quantity and price changes) throughout the region; 3) The transportation systems analysis module specifies the likely re-configuration of freight traffic given physical damage to the network and changes in commodity supply and demand; 4) The increased freight delivery times and distances are converted into increased transportation costs, which are in turn fed into the TERM CGE model to determine the effect on commodity demand and modal choice; 5) An iterative process is used such that the results of the CGE model are fed into the Transportation Network model to re-estimate its results and then fed back to the CGE 12 These inputs are required for the estimation of business interruptions and loss of production and employment. The computer-vision enabled approach applied in the case of bridges is not capable of modeling seaports and buildings. HAZUS is leveraged for this purpose. 52 model, and so on; 6) Various economic resilience tactics (e.g. input conservation, use of inventories, relocation) are fed into the CGE model to estimate the improvements that these tactics can bring about. In order to analyze the income distribution impacts of the port and transportation network disruptions, the TERM Model is also supplemented with a Multi-Sector Income Distribution Matrix (MSIDM) that provides a basic accounting of income payments across nine income brackets (i.e., what proportion of total personal income paid out by each sector accrues to each income bracket). Following the method developed by Rose et al. (Rose et al. 2012, 1988), the MSIDM is constructed using data from the U.S. Bureau of Labor Statistics Occupational Employment Statistics (OES) Division and the U.S. Census. The 2018 California MSIDM was developed using a combination of state and federal tax agency and IMPLAN data. Linkage of the two aspects of our socioeconomic impact analysis module (the TERM Model and the MSIDM) involves two steps. The first step simply involves multiplying the changes in income in each sector by the income distribution matrix to determine the profile of income changes by bracket associated with the initial earthquake damages, initial changes in transportation patterns, and/or optimal transportation route re- configurations reflecting resilience. The results are summed across sectors to obtain an overall change in income distribution of the economy as a result of the shock. The second step is more complicated, in that it adjusts the MSIDM for changes in two other major factors influencing the income distribution: the changes in capital-labor ratio and changes in factor returns, both of which can be calculated by the TERM Model. This yields a new MSIDM to be applied to the vector of sectoral income changes in a manner similar to that described above. Hence, the bottom-line income distribution impacts are determined by this second set of calculations. The initial distribution and the changed distribution (for both the base case and various resilience cases) can be compared by a number of metrics, such as the Gini coefficient, to determine whether the income distribution has been worsened or improved. 53 Case Study: Magnitude 7.3 Earthquake on Palos Verdes Fault Hazard Characterization Without losing generality, a scenario earthquake that poses a significant seismic risk to the Ports of Los Angeles and Long Beach (also known as San Pedro Bay port complex) is considered. The port complex forms one of the world’s busiest seaport handling more containers per ship call than any other port complex in the world. The complex comprises 32% of the total national market; hence, potential interruptions to its operations bear significance to more than just the City of Los Angeles. In order to quantify the deterministic seismic hazard controlling the area where the port facilities are located, the PSH results from the 2014 version of U.S. Conterminous Seismic Hazard Maps (Petersen et al. 2014) are deaggregated. Based on the deaggregation of the PSH for a 975-year return period as shown in Figure 2.7b, the Palos Verdes fault was found to govern the seismic hazard for the complex. Palos Verdes fault is a predominantly right-lateral strike-slip fault system extending in northwest-southeast direction for more than 100 km (Brothers et al. 2015). According to the deaggregated hazard results, the moment magnitude with the highest contribution to the overall hazard is determined as 7.3. The Wells-Coopersmith relationship (Wells and Coppersmith 1994) gives a subsurface rupture length of 90.37 km, and a surface rupture length of 71.12 km for a 7.3-moment magnitude reverse-slip fault event. As such, throughout the deterministic hazard calculations conducted for this study, a magnitude 7.3 earthquake generated by rupturing 90.37 km segment of the Palos Verdes fault line is used. Figure 6a displays the PSHA results for the epicentral location of the defined deterministic event. In order to simplify that calculation of damages resulting from the scenario event, the effect of earthquakes are limited to ground shaking only. Thus, the IMs required for direct damage analyses consisted of 0.3 sec and 1.0 sec spectral accelerations (SA 0.3 and SA 1.0 respectively) for HAZUS predictions and SA 1.0 and for image-based model predictions for the physical damage. The weighted average of the median SA values computed from 2013 GMPEs by Abrahamson et al., Boore et al., Campbell and Bozorgnia, Chiou and 54 Youngs, and Idriss with weights for the first four equations set to 0.22 and the last one set 0.12. Site effects are taken into account using the slope-based V S30 proxy method suggested by Wald, whereas the basin effects were neglected for all site locations. Given the V S30 map computed by Thompson et al. (2014), and the position of the study region on the Los Angeles basin, both of these assumptions are warranted. Figures 2.8.a. and 2.8.c show maps of the SA 1.0 resulting from the scenario event. Damage Assessment for Bridges For the assessment of physical damage, 98 bridges in the immediate periphery of the port complex, i.e., bridges that fall within the region of interest (ROI) in Figure 2.9, were modeled using the image-based modeling procedure mentioned above. For the remaining bridges in the area, HAZUS fragility functions (FEMA 2003), created using the 2018 NBI data (FHWA 1995), are utilized. The purpose of this decision is to reduce the considerable computational load required for bridge-specific fragility calculations. The traffic mobility in and out the ports are at large controlled by the functionality of Interstate 110 and 710 and their connections to Interstate 405. Hence, it is assumed that simulating only the physical damages for the bridges within the ROI at great detail through 55 Figure 2.7 (a) PSHA results for the Ports of Los Angeles and Long Beach, and (b) their deaggregation. ( a ) ( b ) 56 Figure 2.8 Deterministic hazard analysis results for 1.0 sec spectral accelerations at the general study region level (a), the location of damaged bridges within the general study region (b), 1.0 sec spectral acceleration results around rupture length (c), and the corresponding bridge damage in terms of downtimes (d). the image-based method will yield resiliency metrics with satisfactory accuracy. It is also realized that, in certain study regions, particular segments of the transportation network may be equipped with extra redundancies, and these redundancies may render defining a representative ROI difficult. When that is the case, the entire bridge inventory within the study region shall be modeled using the image-based approach. See Figure 2.9 for the defined ROI in this case study as well as a map of closed bridges on Day 1 after the earthquake. Damage state probabilities for each bridge in the study region are computed using Eq. (4) for ds1 through ds 5. Subsequently, HAZUS bridge restoration functions (FEMA 2003) and Eq. (5) are utilized to calculate the downtime associated with the defined scenario 57 event. Figure 2.8 (b and d) show maps of the computed downtimes. It is observed that the bridge downtimes are highest around the rupture port facilities due to the proximity of the scenario event to the complex. However, it is possible to observe bridge closures as far as the Marina Del Rey area. The reason for such damage occurrences is because of the coupled effect of increased ground shaking levels due to the presence of softer soil deposits and pre-1971 construction of the structures. Figure 2.9 Modeled bridge closures on Day 1. Region-of-Interest contains bridges modeled with the image-to-model methodology. Transportation Systems Analysis With the damage assessment information for the scenario event, 6 representative network versions (for days 0, 1, 7, 30, 90 and 104) are modeled according to the methodology described in 2.11. to achieve resilience insights at the system level. A 75% 58 functionality threshold is determined for the bridge closure policy, i.e. a damaged bridge under recovery is deemed closed to service for the period of time its expected functionality is below 75%. This threshold was adopted from literature (Gordon et al. 2004) due to the lack of standardized data on the relationship between damage or functionality information and closure decisions made by inspectors. This way, 137 bridges are closed to service on day 1, 62 on day 7, 58 on day 30, 45 on day 90 and 19 on day 104. On day 105, full recovery is achieved which brings network functionality back to its pre-disaster baseline level. Figure 2.10 Network topology manipulation on links corresponding to bridges deemed closed to traffic (e.g., Bridge 53 1867 on the I210 freeway) implemented with TransCAD’s map editing functionalities. Colored, thicker lines indicate public transit routes such as buses, metro, etc. 59 Six network topologies are modeled for the mentioned days 13 and bridge closures by manipulating the multi-modal network underlying the SCAG RTDM. This manipulation includes: (1) identifying the links corresponding to the bridges to be closed, (2) modifying and discontinuing the public transit routes crossing those bridges so that the routes carry on serving the remaining portions of the routes, (3) removing the identified links from the network altogether 14 . Consequently, six complete traffic distribution and assignment problems are solved by implementing the transportation systems analysis methodology shown in Figure 2.4 to quantify system functionality in pre-disaster and post-disaster settings representing the entire disruption and recovery timeline. In the case study, the authors primarily focus on total travel time indicators V HT (Vehicle Hours Traveled) and Delay (Vehicle Hours Delayed) since travel time is typically regarded as the primary indicator of utility which makes it the main decision variable of travel behavior. Moreover, total travel distance covered indicators (e.g. VMT) may not indicate significant impacts at the system level for dense urban networks even for severe disruptions due to the short detours associated with high redundancy networks. Nevertheless, the authors visualize the impacts on vehicle flow and total travel distance (VMT) with maps to demonstrate the reconfiguration of traffic flow due to the disruption (Figures 2.13. to 2.16.). User-equilibrium results for the 6 network versions allow the quantification of Γ i(t) where I = VHT, Delay and t = 0, 1, 7, 30, 90 and 104 days. Consequently, Q(t) is quantified for all Γ i(t). Figures 10 and 11 show system functionality, Q(t), based on ΓV HT (t) and ΓDELAY 13 The selected days align with the hybrid—image-to-model and HAZUS—damage assessment information and represent the timeline until full recovery in sufficient detail. 14 Partial closures can be modeled with CRAFT’s current capabilities thanks to the detailed model, SCAG RTDM. In this initial work, the authors assume full closure until functionality reaches levels above 75% during recovery. 60 (t), respectively 15 . Data are presented for the entire study region as well as the counties of Los Angeles, Orange and Riverside 16 . Significant disruptions in regional mobility is observed, particularly in the Day 1 network in which 137 bridges are deemed closed to service. During the first week, CRAFT estimates a total of approximately 850,000 hours/day additional travel time spent in traffic which corresponds to a 6.52% decrease in the study region in terms of ΓV HT . As Figure 2.11 demonstrates, Los Angeles County burdens most of this functionality loss with a 11.81% loss in functionality. A visualization of the redistributed vehicle flow—due to the modeled bridge closures— reveals interesting insights. Figure 2.13 and Figure 2.14 illustrate the reconfiguration of vehicle flow during the AM Peak on Days 1 and 7 in terms of percent change relative to the Day 0. In other words, Figure 2.11 and Figure 2.12 illustrate the reconfiguration of vehicle flow after the dramatic loss of functionality on Day 1, and the vehicle flow after a significant portion of system functionality is recovered on Day 7 17 . It is observed that, detouring due to the closed bridges on high volume corridors (e.g., 110, 710, 405 Interstate Highway corridors around the port) shifts the flow to the surface streets, i.e. forces high volume traffic to pass through the neighborhoods nearby 12 . Same effect is seen during the PM peak which is not included for brevity. 15 In Figures 2.11. and 2.12., data points indicate functionality levels on representative days for which a simulation run is carried out. Data points are labeled for lines corresponding to Study Region and Los Angeles only. The linearity in these graphs do not imply a constant rate of system recovery, it is only for visualization purposes to indicate system recovery is still happening even if a full simulation run is not carried out to quantify the exact functionality level on the days between the representative days. 16 Being further away from the disruption, and due to their relative size and the marginal impacts they experience, Ventura, Imperial and San Bernardino. 17 ArcGIS suite is used to visualize and share the transportation simulation results obtained with the SCAG RTDM on TransCAD. "Join" functions are employed within Arc Pro to match the tabular data from TransCAD with the spatial network data. This enables rich visualizations of network functionality and the publication of results online (Figures 12, 13, 14 and 15). Visualizations and online maps provide vehicle flow and vehicle miles traveled (VMT) results for Days 1 and 7 only where most of the disruption happens. Shared online maps are hosted on a University of Southern California server and are open to public access for viewing purposes. 61 Figure 2.11 System functionality Q(t) based on Vehicle Hours Traveled (VHT). Figure 2.12 System functionality Q(t) based on Vehicle Hours Delayed. 62 Figure 2.13 Changes in Vehicle Flow from Day 0 to Day 1 after the scenario event during the AM peak. Available online at: https://arcg.is/HjDO8 63 Figure 2.14 Changes in Vehicle Flow from Day 0 to Day 7 after the scenario event during the AM peak. Available online at: https://arcg.is/1PeDOa 64 Figure 2.15 Spatial distribution of changes in VMT from Day 0 to Day 1, aggregated up to Tier 2 TAZ level. Available online at: https://arcg.is/1mSC9T 65 Figure 2.16 Spatial distribution of changes in VMT from Day 0 to Day 7, aggregated up to Tier 2 TAZ level. Available online at: https://arcg.is/1PeDOa 66 Damage Assessment for Other Critical Infrastructures and General Building Stock As illustrated in Figure 2.4, the economic impact analysis module in CRAFT requires (on top of bridge damages) an assessment of damages to other critical infrastructures and the general building stock. This requirement needs to be satisfied to achieve the model linkages presented previously for a full deployment of the framework. Damage Assessment for Ports of Los Angeles and Long Beach To simulate the Mw 7.3 earthquake scenario, the author utilized HAZUS 4.2 (FEMA 2010), FEMA’s hazard loss estimation software with the default building inventory data on ports and harbors to estimate damage and functionality at the Ports of Los Angeles and Long Beach. Probabilities for each damage state as well as the expected functionality levels throughout the recovery timeline are estimated for 171 facilities in total for both ports (berths, terminals, etc.). Official facility maps published by the port authorities are used to manually classify the berths in the HAZUS inventory into main cargo categories handled by the facilities such as automobile, containerized cargo, dry bulk, liquid bulk, etc., to link the facility downtimes to disruptions of imported and exported commodity flows. Damage Assessment for General Building Stock HAZUS has a detailed loss estimation methodology for the damages and the direct losses resulting from the vulnerability of the general building stock (GBS) to the simulated event. GBS includes residential, commercial, industrial, agricultural, religious, government, and educational buildings, and the damage state probability of the general building stock is computed at the centroid of the census tract. The damage states for each building type are linked to the default economic data supplied within HAZUS (e.g., structural repair costs for each of the damage states, model building types and occupancies, contents damage as a function of damage state, etc.) to estimate the direct economic losses. Detailed discussions on the building damage and loss estimation methodology can be found in the corresponding technical manual published by FEMA (FEMA, 2012). 67 The database behind the approach stores building information at the census tract level for the specific occupancy classes listed in Table 2.4. For each census tract and per each occupancy class, structural and nonstructural cost of repair or replacement, loss of contents, business inventory loss, relocation costs, business income loss, employee wage loss and loss of rental income are estimated. Definitions of each loss category are provided in the cited HAZUS technical manual. To inform the TERM Model, all such results are first aggregated up to the county level of detail. Then the percentage of building/content damages are calculated for the 3-county LA Metro Region. RES1 Single Family Dwelling IND1 Heavy RES2 Mobile Home IND2 Light RES3 Multi Family Dwelling IND3 Food/Drugs/Chemicals RES4 Temporary Lodging IND4 Metals/Minerals Processing RES5 Institutional Dormitory IND5 High Technology RES6 Nursing Home IND6 Construction AGR1 Agriculture COM1 Retail Trade REL1 Church/Non-Profit COM2 Wholesale Trade GOV1 General Services COM3 Personal and Repair Services GOV2 Emergency Response COM4 Professional/Technical Services EDU1 Grade Schools COM5 Banks EDU2 Colleges/Universities COM6 Hospital COM7 Medical Office/Clinic COM8 Entertainment and Recreation COM9 Theaters COM10 Parking 68 Table 2.4 Specific Occupancy Classes for the General Building Stock in HAZUS Functionality Loss and Recovery at the Ports of Los Angeles and Long Beach Due to the special focus on the port complex in this project, instead of making simple assumptions on the remaining functionality at the ports after the scenario event, FEMA HAZUS methodology is employed to find estimates of port remaining functionality following the strike of the earthquake as well as its recovery path. HAZUS accommodates a database that provides detailed information (owner, address, coordinates, etc.) about the berths at the ports. However, an industry classification scheme is not given. Therefore, as presented previously, official facility maps published by the port authorities are used to manually classify the berths in the HAZUS inventory into main categories of cargos handled by the damaged facilities. This way, functionality loss and recovery information was fed into the TERM Model for 5 different cargo categories (containerized, breakbulk, dry bulk, liquid bulk, automobiles). Table 2.5 shows these estimates separately for Ports of Los Angeles and Long Beach. 69 Port of Los Angeles Facilities Day 1 Day 3 Day 7 Day 14 Day 30 Day 90 Day 150 Container Terminals 51.57 66.47 71.96 72.96 75.74 87.19 100.00 Breakbulk 51.40 66.30 71.80 72.80 75.60 87.10 100.00 Dry Bulk 52.00 66.90 72.40 73.40 76.10 87.40 100.00 Liquid Bulk 51.48 66.38 71.86 72.88 75.64 87.14 100.00 Automobiles 53.40 68.25 73.70 74.70 77.30 88.20 100.00 Port of Long Beach Facilities Day 1 Day 3 Day 7 Day 14 Day 30 Day 90 Day 150 Container Terminals 59.73 74.03 79.22 80.05 82.16 91.04 100.00 Breakbulk 60.00 74.26 79.41 80.21 82.32 91.15 100.00 Dry Bulk 58.91 73.34 78.54 79.39 81.56 90.72 100.00 Liquid Bulk 58.38 72.82 78.06 78.88 81.14 90.48 100.00 Automobiles 60.70 74.90 80.00 80.80 82.80 91.40 100.00 Table 2.5 Percentage of Port Functionality and Recovery Estimates for Different Cargo- Handling Terminals General Building Stock Damages in Greater Los Angeles Table 2.6 presents the general building damage data calculated based on the HAZUS earthquake simulation results. The percent building and content losses (last column) is calculated by dividing the sum of dollar losses in building and contents across the three counties of the LA Metro Region by the total exposure values of building stocks. On average, various business sectors in the LA Metro Region experience property damages ranging from 1.4% to 6.21%. The sectors that experience the highest property damages include High Technology Industry, Food/Drugs/Chemicals Industry, and Wholesale Trade. In 70 Table 2.7, the percentage property damages from HAZUS occupancy classes are first mapped to TERM economic sectors. Next, the weighted average recovery period by sector is calculated in Column 4 using the information on building damage states (None, Slight, Moderate, Extensive, and Complete) and the associated recovery time (measured in days) both obtained from HAZUS. In the last column, the percent destruction of capital input is calculated on an annual basis by multiplying the percent building and content losses by the percent of time over one year it takes to recover for each sector. Occupancy Class Definition Building Loss (M $) Content Loss (M $) Total Exposure Value (M $) % Building & Content Loss AGR1 Agriculture 55,931 21,286 2,692,677 2.87% COM1 Retail Trade 1,136,455 407,194 36,584,328 4.22% COM2 Wholesale Trade 1,550,634 547,609 44,862,051 4.68% COM3 Personal/Repair Services 790,633 297,464 26,745,228 4.07% COM4 Prof./Technical Services 2,890,156 1,096,050 97,526,179 4.09% COM5 Banks 133,441 51,242 4,919,510 3.75% COM6 Hospital 248,196 130,784 12,262,975 3.09% COM7 Medical office/Clinic 566,333 312,068 18,581,869 4.73% COM8 Entertainment & Rec. 958,030 342,208 28,844,919 4.51% COM9 Theaters 33,098 11,575 957,793 4.66% EDU1 Schools 324,234 129,665 14,528,751 3.12% EDU2 Colleges/Universities 67,382 42,778 3,914,696 2.81% GOV1 General Services 145,763 51,576 4,348,859 4.54% GOV2 Emergency Response 31,361 17,461 1,350,226 3.62% IND1 Heavy 710,297 394,763 22,782,954 4.85% IND2 Light 703,994 399,674 22,937,561 4.81% IND3 Food/Drugs/Chemicals 343,729 194,114 10,176,421 5.29% IND4 Metals/Minerals Processing 94,854 51,968 2,893,606 5.07% IND5 High Technology 73,124 41,350 1,843,089 6.21% IND6 Construction 272,069 102,633 11,127,149 3.37% REL1 Church/N.P. Offices 489,390 185,631 18,533,206 3.64% 71 RES1 Single Family Dwelling 11,577,632 3,264,336 1,036,315,753 1.43% RES2 Mobile Home 390,076 44,305 10,450,711 4.16% RES3A Multi Family Dwelling – Duplex 312,416 73,417 14,271,921 2.70% RES3B Multi Family Dwelling – 3-4 Units 686,953 161,152 33,351,463 2.54% RES3C Multi Family Dwelling – 5-9 Units 1,345,659 315,986 67,098,742 2.48% RES3D Multi Family Dwelling – 10-19 Units 1,184,211 278,017 59,984,901 2.44% RES3E Multi Family Dwelling – 20-49 Units 1,217,032 285,070 61,599,244 2.44% RES3F Multi Family Dwelling – 50+ Units 1,235,564 289,291 63,712,570 2.39% RES4 Temporary Lodging 264,024 54,375 9,184,425 3.47% RES5 Institutional Dormitory 492,615 107,345 26,341,777 2.28% RES6 Nursing Home 43,986 9,603 3,588,936 1.49% Total 30,369,272 9,711,990 1,774,314,490 2.26% Note: Building losses include both structural and non-structural losses. Table 2.6 Direct Loss Estimates (Building and Content Losses Only) for the LA Metro Region (Los Angeles, Orange and Riverside Counties) TERM Sector # Short names % Building & Content Loss Average Recovery Time (Days) % Capital Input Destruction on an Annual Basis 1 Crops 2.87% 60 0.47% 2 PoultryEggs 2.87% 60 0.47% 3 Livestock 2.87% 60 0.47% 4 OthLivestock 2.87% 60 0.47% 5 ForestFrsHnt 2.87% 60 0.47% 6 OilGas 5.07% 211 2.93% 7 Coal 5.07% 211 2.93% 8 OtherMining 5.07% 211 2.93% 9 BiomassGen 4.09% 251 2.81% 10 CoalsGen 4.09% 251 2.81% 11 GasGen 4.09% 251 2.81% 12 HydroGen 4.09% 251 2.81% 13 NuclearGen 4.09% 251 2.81% 14 RenewGen 4.09% 251 2.81% 72 15 ElecDist 4.09% 251 2.81% 16 NatGasDist 4.09% 251 2.81% 17 WaterSewage 4.09% 251 2.81% 18 ResidConstrt 3.37% 164 1.51% 19 OthConstruct 3.37% 164 1.51% 20 HwyBrdgCons 3.37% 164 1.51% 21 OthMaintain 3.37% 164 1.51% 22 Mrstreets 3.37% 164 1.51% 23 FoodProc 5.29% 211 3.05% 24 BevTobManu 5.29% 211 3.05% 25 Textiles 4.85% 211 2.80% 26 Apparels 4.81% 211 2.78% 27 LeathFtwr 4.81% 211 2.78% 28 WoodProds 4.85% 211 2.80% 29 PulpPaperPbd 4.85% 211 2.80% 30 Printing 4.81% 211 2.78% 31 PetrolRefine 5.29% 211 3.05% 32 OthPetrolCl 5.29% 211 3.05% 33 Chemicals 5.29% 211 3.05% 34 RubPlastic 4.81% 211 2.78% 35 NonMetMinPrd 5.07% 211 2.93% 36 PrimMetals 5.07% 211 2.93% 37 FabriMetals 4.85% 211 2.80% 38 AgriMachinry 4.85% 211 2.80% 39 IndustrMach 4.85% 211 2.80% 40 CommrcMach 4.85% 211 2.80% 41 AirConHeat 4.85% 211 2.80% 42 MetalWkMach 4.85% 211 2.80% 43 TurbnEngine 4.85% 211 2.80% 44 OtherMach 4.85% 211 2.80% 45 Computers 6.21% 266 4.53% 46 CmptrStorage 6.21% 266 4.53% 47 ComptrTrmEtc 6.21% 266 4.53% 48 CommunicEqp 4.81% 211 2.78% 73 49 MscElctEqp 4.81% 211 2.78% 50 Semicondctr 4.81% 211 2.78% 51 ElecInstrmnt 4.81% 211 2.78% 52 HholdEqp 4.81% 211 2.78% 53 MVPManu 4.85% 211 2.80% 54 AerospaceMan 4.85% 211 2.80% 55 RlrdCars 4.85% 211 2.80% 56 ShipsBoats 4.85% 211 2.80% 57 OthTrnEqp 4.85% 211 2.80% 58 Furniture 4.81% 211 2.78% 59 MiscManuf 4.81% 211 2.78% 60 WholesaleTr 4.68% 217 2.78% 61 AirTrans 4.09% 251 2.81% 62 RailTrans 4.09% 251 2.81% 63 WaterTrans 4.09% 251 2.81% 64 TruckTrans 4.68% 217 2.78% 65 GrdPassTrans 4.09% 251 2.81% 66 Pipeline 4.09% 251 2.81% 67 OthTransprt 4.85% 211 2.80% 68 Warehousing 4.68% 217 2.78% 69 RetailTr 4.22% 217 2.51% 70 Publishing 4.81% 211 2.78% 71 MovieSound 4.09% 251 2.81% 72 BroadcastSrv 4.51% 186 2.30% 73 Telecomm 4.51% 186 2.30% 74 InfoSvce 4.09% 251 2.81% 75 DataProcScv 4.09% 251 2.81% 76 FinancBank 3.92% 219 2.36% 77 RealEstate 4.09% 251 2.81% 78 RentLease 4.09% 251 2.81% 79 AssetLessors 4.09% 251 2.81% 80 PrfSciTchSrv 4.09% 251 2.81% 81 WasteMgmt 4.54% 259 3.22% 82 Education 3.12% 223 1.91% 74 83 HealthSocAs 3.91% 249 2.63% 84 ArtsRecreat 4.51% 186 2.30% 85 Accommodatn 3.47% 247 2.35% 86 EatDrinkPlce 4.51% 186 2.30% 87 OthService 4.07% 217 2.42% 88 GovEnterprs 4.07% 217 2.42% 89 StaLocGov 4.54% 259 3.22% 90 OwnOccDwell 4.08% 238 2.69% 91 FedGovt 4.08% 238 2.69% 92 Holiday 4.51% 186 2.30% 93 FgnHol 4.51% 186 2.30% 94 ExpTour 4.51% 186 2.30% 95 ExpEdu 3.12% 223 1.91% 96 WT_EXP 4.09% 251 2.81% 97 AT_EXP 4.09% 251 2.81% Table 2.7 General Building Damage for LA Metro Region Economic Resilience Tactics: Case of Ports and Hinterland Transportation The earthquake scenario investigated in this case study has a special focus on ports and their hinterland connectivity. Serving as critical portals of a nation’s supply-chain, seaports and their associated inland transportation infrastructure are especially vulnerable to major disruptions from a variety of causes. The economic impacts of these disasters can extend well beyond the on-site operations at the port complex, through supply-chain curtailments and/or delays of delivering imports and exports to their destinations. Port resilience is a special case of economic resilience (Rose and Wei 2013; Wei et al. 2020a). In the context of a port shutdown or disruption, static economic resilience refers to how ports and businesses can utilize remaining resources effectively to maintain functioning to the extent that they can. Supplier-side resilience refers to various tactics to maintain functionality at the port. On the customer-side, businesses affected by the 75 import or export disruptions could initiate a broad range of coping activities, including both businesses directly and indirectly affected by the port disruptions throughout the economy-wide supply chain. Expanding on our prior research, various supplier-side and customer-side resilience options are defined relating to ports and their hinterland transportation system disruptions in Table 2.8. Supplier-Side Resilience Options Customer-Side Resilience Options Excess capacity. Utilization of unused capacity at undamaged terminals. Use of inventories. Stockpiling critical inputs for the production of goods and services. Cargo prioritization. Altering schedules for unloading or loading based on the characteristics or value of the cargo. Conservation. Finding ways to utilize less disrupted imported goods and curtailed critical inputs in production processes. Ship re-routing. Sending ships to other ports. Input substitution. Utilizing similar goods in the production process to those whose production has been disrupted. Export diversion for import use. Sequestering goods that were intended for export to substitute for lack of availability of imports or domestically-produced goods. Import substitution. Bringing in goods and services in short supply from outside the region. Effective management. Improvements in decision-making and expertise that enhance functionality 18 . Production relocation. Shifting production to branch plants Production recapture (Rescheduling). Working extra shifts or over-time to clear up a backlog of vessels after the resumption of port operation. Production recapture (Rescheduling). Making up lost production by working extra shifts or over time after the port re-opens 18 Such effective management measures can be potentially built on existing data-driven commodity throughput portal and port community communication systems. See Appendix A for a summary of the Port Optimizer that is being deployed in POLA/POLB. 76 Effective road infrastructure asset management. Improvements in decision- making and expertise that enhance functionality and recovery. Effective travel demand management. Establishing measures to decrease travel demand during recovery. Table 2.8 Summary of Resilience Tactics Relating to Port and Highway Transportation Disruptions Construction of the Multi-Sector Income Distribution Matrix of California In Section 2.13. on the economic resilience module of CRAFT, discussions on income distribution impacts were made. Within the context of this case study, to evaluate the impacts of port and transportation network disruptions and the effectiveness of resilience tactics across socioeconomic groups (specifically income groups), a Multi-Sector Income Distribution Matrix (MSIDM) for the state of California is constructed. The matrix provides the earnings profile according to nine income brackets for each producing sector in the economy, i.e., what proportion of the personal income (including both labor income and capital income) paid out by each sector accrues to each income bracket (Li et al. 1999; Rose et al. 2012, 1988) In 2018, the total personal income in California was more than $2.4 trillion (Bureau of Economic Analysis (BEA) 2019). The first major component is Wages and Salaries, which include the total remuneration of employees. The total Employee Compensations, which were $1.35 trillion in 2018, are the sum of Wages and Salaries and Employer Contributions for Employee Pension and Insurance Funds. The total Proprietors’ Income in 2018 was $249.6 billion. The next major component of the personal income accounts is capital income, which amounted to $538.3 billion in California in 2018. These include dividends, interest payments, and rental income. The final major component of the personal income accounts is Personal Current Transfer Receipts, with were $341.2 billion 77 in California in 2018. These mainly include payments from government welfare and benefit programs. The BEA Personal Income accounts for California were used as control totals when the individual income matrices are constructed in the following sections. In this case study, nine household income brackets that are used in IMPLAN are adopted (the largest provider of regional input-output and social accounting data in the U.S.). The income distribution matrix for wages and salaries is constructed using the Occupation-Industry Employment matrix and the Occupation-Industry Wage matrix obtained from the BLS Occupational Employment Statistics (OES) (Bureau of Labor Statistics (BLS) 2019). IMPLAN data are used as the main data source to distribute proprietors’ income, dividends, other property income, and transfers across sectors and income brackets (see Appendix A for more details). The total personal income matrix is constructed by combining each individual matrix of various personal income components. Then, the total income distribution coefficient (structural) matrix is calculated for California by dividing the income value for each bracket in a given sector by the total income for that sector. Aggregate Impacts of Port Disruptions To simulate the macroeconomic impacts of port disruptions, reductions in port functionality are translated into disruptions of import and export flows. The percentage reductions in import uses and export production by sector in each of the TERM regions are then calculated. Similar to other CGE models, the TERM Model automatically takes into consideration three types of inherent economic resilience that work through the price system. These include input substitution, import substitution, and regional production shifts (IIR). Following the methodology developed in Wei et al. (Wei et al. 2020a), the loss reduction potential of the IIR is estimated by comparing the simulation results using the input-output (I-O) analysis and the TERM simulation results of the port disruptions. The I- O analysis is based on the assumption of fixed production coefficients (or a linear relationship between the changes in production inputs and changes in the output), and thus can be used as the Base Case with no resilience tactics incorporated. Table 2.6 first 78 presents the GDP impacts of the Base Case (no resilience) and the GDP impacts obtained from the TERM simulations that take into consideration IIR. For the LA Metro Region, the port disruptions are estimated to result in $1.76 billion GDP losses (0.219% reduction) after the three major types of inherent resilience tactics are considered. The other regions in California, as well as the Rest of U.S., also experience GDP losses, but in smaller magnitudes in percentage terms since the LA Region is the direct recipient and user of nearly 50% of the import shipments through the ports. The total GDP losses for the U.S. as a whole is about $11.0 billion, though this is only less than a one-tenth of one percent decline. The last two columns in Table 2.9 present the loss reduction potential for the resilience tactics in percentage terms. A comparison of the results from the TERM Model (second row) and the I-O analysis (first row) indicates that the inherent economic resilience estimated by the TERM CGE Model reduces the potential GDP losses by 87.2% for the LA Metro Region and 86.1% at the national level. Next a simulation run is conducted of other types of inherent and adaptive resilience tactics pertaining to port disruptions (See the last row of Table 2.6). The combined resilience can further reduce GDP losses to $0.28 billion for LA and $0.76 billion for the U.S., or a reduction of GDP losses by about 97.9% and over 99%, respectively, compared to the Base Case. Note that the effects of the various resilience tactics are not additive because of overlaps in their application. 79 LA Metro SF Metro Rest of CA Rest of US US Total Loss Reduction Potential (for LA) Loss Reduction Potential (for US) Base Case (no resilience) -13,754.92 -5,912.57 -6,342.29 -52,900.52 -78,910.42 -1.71% -1.24% -1.12% -0.41% -0.54% With Inherent Resilience (IIR) -1,763.81 -936.19 -739.53 -7,535.55 -10,975.09 87.18% 86.09% -0.22% -0.20% -0.13% -0.06% -0.08% With Combined Resilience (IIR, Other Inherent, Adaptive Resilience) -284.92 -64.06 -77.21 -331.96 -758.15 97.93% 99.04% -0.04% -0.01% -0.01% 0.00% -0.01% Table 2.9 Real GDP Impact of Port Disruptions – Base Case and Resilience Cases (in millions 2019 dollars and percent reduction from pre-disaster levels). Aggregate Impacts of Increasing Truck Transportation Costs and Building Stock Damages The transportation cost increases over the 104-day recovery period, translate to a 0.53% increase in truck transportation cost within the LA Metro Region and a 0.26% increase between the LA Metro Region and the Rest of CA on an annual basis. The GDP losses in the LA Metro Region due to increasing truck transportation costs are estimated to be $17.0 million. The Rest of California will only experience very slight GDP losses. The Rest of U.S. is estimated to have a small increase in GDP of $4.21 million, which can be explained by the effect of regional production shifts automatically captured by the TERM Model that is caused by a truck transportation cost increase (and thus an implicit production cost increase) in the LA Metro Region and the Rest of CA in 80 comparison with the Rest of US. The resilience tactic of a more rapid opening of critical highway corridors during the first week after the seismic event is estimated to have a loss reduction potential of about 10% over the entire recovery period. The simulated earthquake also results in destructions and damages to general building stock in the LA metro region, which in turn results in interruptions to the flow of goods and services emanating from the productive capital stock. The total GDP losses stemming from property damages are estimated to be nearly $23.4 billion (or a 2.8% reduction from the annual baseline level) in the LA Metro Region. The Rest of CA (excluding Northern California) is estimated to experience very slight GDP losses due to the scenario earthquake. The Rest of the U.S. is estimated to have an increase in GDP of $3.3 billion, which again can be explained by the effect of regional production shifts or locational substitution of economic activities caused by the simulated disaster. After two major types of resilience tactics to cope with general building stock damages are taken into consideration—the use of undamaged spare/excess capacity and production recapture—GDP losses decrease from $22.4 billion to $13.8 billion for the LA Metro Region and from $19.3 billion to $11.8 billion for the U.S. as a whole, or a reduction of about 38% compared to the Base Case. When we compare the economic impacts of the three types of disruptions/damages of the earthquake scenario – port disruption, hinterland transportation cost increase, and general building damages – the impacts from general building damages account for over 92% of the total impacts in the LA Metro Region before resilience, other than IIR, is taken into account. The hinterland transportation system disruption results in the smallest impacts because of the high redundancy of the transportation network in the study region. For the U.S. as a whole, impacts from general building damages account for about 63% of the total impacts (without resilience adjustments), while the port disruptions account for another one third of the total impacts. After the adjustment of the various resilience tactics, impacts from general building 81 damages account for nearly 98% of the total impacts in the LA Metro Region and 94% for the U.S. This is because there are more effective resilience tactics businesses can implement to deal with port disruptions and supply chain shortages than are available in the case of physical damages to buildings and facilities. Economic Impacts of Combined Disruptions/Damages Another simulation run is conducted in which all three types of disruptions/damages are combined. This simulation is run for both the Base Case and the Combined Resilience Case (See Table 2.7). The total GDP losses are estimated to be $24.2 billion (or a 3% reduction) in the LA Metro Region, and $30.2 billion in the U.S. as a whole (or a 0.21% reduction). Various resilience tactics can help reduce the total impacts to $14.2 billion in the LA Metro Region and $12.8 billion for the U.S. as a whole, or a loss reduction of 41.3% and 57.6%, respectively. The reason that total impacts are less on the U.S. as a whole than on the LA Metro Region in the Combined Resilience Case is that the Rest of the U.S. is expected to experience an overall increase in economic activities due to regional production shifts after the earthquake hits the LA area. Summary on Socioeconomic Impacts and Resilience Our analysis indicates that it takes 150 days for the ports to fully recover from the simulated seismic event. The total GDP impacts stemming from both import and export disruptions are estimated to be $11.8 billion in the LA Metro Region and $67.5 billion for the U.S. before we consider any resilience. These impacts are reduced to $1.5 billion and $9.4 billion, respectively, after we take into consideration three major types of inherent economic resilience (input substitution, import substitution, and regional production shifts) that are automatically captured by the TERM CGE Model. After we consider the other types of inherent resilience tactics and adaptive resilience tactics, the total impacts are further reduced to $0.24 billion in the LA Metro Region and $0.65 billion in the U.S. In addition, damages to the highway transportation system cause a 0.53% increase in truck transportation cost within the LA Metro Region and a 0.26% increase between the 82 LA Metro Region and the Rest of CA (on an annual basis). The associated GDP losses are estimated to be only $15 million in the LA Metro Region because of the general high redundancy of the transportation network. The total GDP losses from damages to the building stock are estimated to be $19.2 billion in the LA Metro Region, which is reduced to $11.8 billion after the adjustment for resilience. The GDP losses for the U.S. are $16.5 billion with no resilience, and $10.1 billion after the resilience adjustments. The lower impacts at the national level are due to the offsetting effect stemming from regional production shifts from the earthquake-impacted region to elsewhere in the country. The combined simulation of all three types of disruptions/damages yields GDP losses of $12.1 billion for the LA Metro Region and $10.9 billion for the U.S. after we consider all the relevant resilience tactics. The loss reduction potential of resilience is 41.3% at the LA regional level and 57.6% at the national level. LA Metro SF Metro Rest of CA Rest of US US Total Loss Reduction Potential (for LA) Loss Reduction Potential (for US) Base Case (no resilience) -24,207.64 -827.70 -855.49 -4,295.82 -30,186.64 -3.00% -0.17% -0.15% -0.03% -0.22% Combined Resilience Case -14,200.35 -11.71 -166.77 1,570.97 -12,807.86 41.34% 57.57% -1.76% 0.00% -0.03% 0.01% -0.09% Table 2.10 Real GDP Impact of the Combined Disruptions/Damages in Base Case and Resilience Cases (in millions 2019 $ and percent reduction from pre-disaster levels). Income Distribution Impacts Based on the simulation results obtained from the TERM Model, income distribution analyses are performed for the LA Metro Region (the most affected region by the 83 simulated earthquake scenario). Detailed income distribution impacts for port disruption, transportation cost increase, general building damage, and all three combined, are presented in Appendix C. The tables first present the distribution of personal income across income brackets in the baseline, followed by the income distribution impacts for both post-disruption simulation cases (Base Case and Combined Resilience Case). For port disruptions, the percentage reductions in income are relatively higher for the lower- to middle-income groups in the Base Case, but are relatively higher for the middle- to high-income groups in the Combined Resilience Case. For the transportation cost increase simulation, the percentage changes are relatively higher for some middle- income and upper-income groups. For the building damage simulations, the middle- income and higher-income groups in general experience relatively higher income losses in both the Base Case and Resilience Case. This can be explained by the fact that a higher proportion of capital-related income is earned by higher-income groups. Table 2.11 first presents the Gini coefficients for the income distribution in the baseline and various simulation cases. The changes in the Gini coefficient relative to the baseline level are next presented. Finally, the Gini coefficient is computed for the income loss alone. The Gini coefficient increases in the port disruption Base Case, which indicates that the disruption is borne slightly disproportionately by lower- and middle- income groups. The Resilience Case results in a Gini coefficient slightly lower than the baseline level, which is explained by the fact that the various resilience tactics are more effective in reducing impacts in the sectors that employ more people from the lower- income groups. For example, manufacturing sectors have a higher potential of inventory uses and production recapture compared with service sectors such as healthcare, finance and insurance, professional and business services. The Gini coefficients of the other cases decrease compared to the baseline level, which indicates that the income losses stemming from transportation cost increase and general building damages are borne disproportionately by middle- and higher-income groups. This is because these groups earn a higher proportion of capital-related income and thus are expected to experience a higher proportion of income losses from capital stock damages. Since the impacts of 84 general building damages account for over 90% of the total impacts in the LA Metro Region, the combined simulation of all three types of disruptions/damages also leads to lower Gini coefficients. Disruption Type Baseline Scenario Gini Coefficient Change in Gini Coefficient Gini Coefficient of the Income Loss Port Disruption_Base Case 0.465478 0.465614 0.000136 0.413109 Transportation Cost Increase_Base Case 0.465478 0.465478 0.000000 0.490154 Building Damage_Base Case 0.465478 0.463904 -0.001574 0.508171 Combined Disruptions_Base Case 0.465478 0.464041 -0.001438 0.501768 Port Disruption_Resilience Case 0.465478 0.465473 -0.000006 0.471813 Transportation Cost Increase_Resilience Case 0.465478 0.465478 0.000000 0.490157 Building Damage_Resilience Case 0.465478 0.464243 -0.001235 0.508481 Combined Disruptions_Resilience Case 0.465478 0.464238 -0.001240 0.507328 Table 2.11 Gini Coefficient Impacts Understanding Disruption and Recovery of Regional Transportation under Dynamic Resilience Tactics So far in this thesis, the resilience of the transportation system—in a ‘static resilience’ manner—has been automatically integrated in the estimates of system recovery in terms of the transportation system functionality metrics (i.e., VHT, VMT, VHD), by assuming that the trips are loaded on to the degraded transportation network to find new UE (user 85 equilibrium) based solutions to the traffic assignment problem 19 . In contrast, the recovery of the damaged bridges is estimated through the HAZUS restoration functions. In other words, there has been no modeling for the resilience tactics potentially available at the system level that can accelerate the speed of the recovery of the degraded system, i.e., dynamic resilience. In this section, results are presented from a testing of such a tactic, specifically a hypothetical intervention by the decision makers to keep 8 bridges open to service on the I405 corridor between the I405/I110 and I405/I10 intersections. For instance, such an intervention could entail the rapid installment of temporary support structures such as shoring systems that are often used in bridge construction (WashDOT 2001). On Day 1 after the scenario earthquake, these 8 bridges on the mentioned corridor were observed to have estimated functionality levels close 75% which is the closure threshold mentioned previously (See the green features in the map presented with Figure 2.17 for these bridges). Due to the lack of data and literature on the dynamic resilience capacities of owners and agencies, as an effort to test a potential tactic, the author assumes that these 8 bridges are kept open on Day 1, hence contributing to a faster recovery for the entire system. To quantify the contribution of the more rapid opening of the 8 bridges to system functionality during Days 1-7 after the scenario event, the authors reiterated the transportation systems analysis component of the analytical framework mentioned in earlier sections. From the results (See Figure 2.18, we find a significant improvement to system functionality for this time period in comparison with the transportation system base case where the 8 bridges are closed to service during Week 1. Specifically, there is over a 3% reduction in functionality loss in L.A. County and a reduction of 220,000 hours/day spent in traffic in the study region with respect to the base case. This is all due to the rapid recovery of 8 bridges (out of 147 closures on Day 1) identified in the tactic, which emphasizes the capability to achieve large improvements in resilience strategically. 19 In parallel, the assumption that travelers have perfect information about the state of the network is made. Consequently, UE outcomes of the degraded network versions are used to measure system disruption and recovery. 86 Figure 2.17 Bridges Deemed Open under the Dynamic Resilience Tactic between Days 1-7, and Bridges that Remain Closed during the Same Time Period. 87 Note: Based on Vehicle Hours Traveled (VHT) and shown for counties of Los Angeles, Orange and Riverside and entire SCAG region (Study Region) Figure 2.18 Regional System Functionality Q(t) before and after resilience tactic. 88 Other Dimensions of Transportation Asset Management In many developed countries around the world, aging civil infrastructure is deteriorating at an alarming rate, with overwhelming transportation, power, water, and pipeline system failures warning of a coming crisis. In the U.S., currently, ‘over 240,000 water main breaks occur each year in the U.S.’, ‘over 188 million trips are taken daily across deficient bridges’, and ‘21% of the nation’s highways are in poor condition’ (ASCE 2017). Infrastructure in the U.S. received an overall grade of D+ (poor) in the American Society of Civil Engineers (ASCE)’s report card (ASCE 2017). The “deteriorating infrastructure, long known to be a public safety issue, has a cascading impact on our nation’s economy, impacting business productivity, gross domestic product (GDP), employment, personal income, and international competitiveness” (ASCE 2017). According to ASCE, the poor conditions of the transportation infrastructure, including roads and bridges, result in increased travel times, which “translate into higher costs for businesses to manufacture and distribute goods and provide services. These higher costs in turn get passed along to workers and families”. ASCE estimates the resulting economic losses as $2 trillion business losses per year, $1 trillion in GDP losses per year, 2.5 million jobs lost per year, and $3,400 disposable income losses per household per year (ASCE 2016, 2017). Improving the civil infrastructure is a grand societal challenge that can no longer be ignored. “Restoring and improving urban infrastructure” is one of the 14 grand challenges identified by the U.S. National Academy of Engineering (ASCE 2016). Addressing this challenge requires answering many questions such as: How to better assess the current conditions and performance of our infrastructure systems? How to better predict future deterioration, conditions, and impacts? How to better maintain our infrastructure for public safety, cost-effectiveness, and improved socioeconomic outcomes? How to measure the true costs and benefits of infrastructure decisions and disruptions? How to empower decision making and maximize benefits from infrastructure investments? 89 Recent advances in data analytics and artificial intelligence have created a unique opportunity to solve this grand societal challenge by learning from past and current conditions to better assess and predict the current and future conditions of civil infrastructure systems, and derive insights about construction, maintenance, and investment decision making – by linking, integrating, and analyzing the wealth of data that exist. For example, the national bridge inventory (NBI) contains inspection data for 604,995 bridges, from 1992 to 2015. In addition, unstructured textual data are generated by bridge owners in the form of inspection, maintenance, and accident reports. The reports include critical data beyond the NBI data, such as the specific types of defects (e.g., crack, decay, scour, and settlement) and the detailed descriptions of the defects (e.g., crack direction, length, and depth; spatial location of crack in element; type of crack whether it is structural or non-structural; observed onset time of crack). With the increasing use of emerging technologies such as sensors, laser scanners, and unmanned aerial vehicles (UAVs), bridge owners are also continuously creating large amounts of data related to the construction, operation, and conditions of the infrastructure. This is in addition to many other relevant data that are continuously generated, including traffic, environmental, and socioeconomic data. The main objective of this chapter of the thesis is to summarize explorations by the author and his collaborators on underutilized dimensions of data (and its analytics) in the management of civil infrastructure systems with a focus on transportation asset management. These explorations are highlighted results of a joint effort with many stakeholders as part of an NSF grant on a Civil Infrastructure Systems Open Knowledge Network (CIS-OKN) (NSF 2020). The CIS-OKN is designed to harness the ongoing data revolution in the civil systems domain, and transform the existing isolated data sources into integrated knowledge and actionable decisions. The team asserted that the CIS domain needs a shared knowledge infrastructure that can (1) extract information from both structured and unstructured sources, (2) connect the scattered data, information, and knowledge from disparate sources, which is challenging because of the data heterogeneity (structured and unstructured) and complexity (technically complex, with 90 different levels of technical detail and quality), and (3) enable deeper interdisciplinary analysis of connected data/information/knowledge. The author assumed an active role in envisioning analytical frameworks and potential use cases for these new modes of data- driven discovery, innovation, and decision making in the CIS domain for improved safety, accessibility, and economic opportunity. These frameworks and use cases were focused on maintenance, repair and reconstruction decision-making in the context of its impacts on system-level performance indicators (e.g., changes in total VHT, VMT, Delay, etc. in a region), on accessibility to jobs and services as well as on local/regional/national economies fueled by the supply chains utilizing the transportation infrastructure. The discussion presented in this chapter is focused on bridges similar to Chapter 2, however, one can extend the motivations behind this research to other assets as well to pursue similar directions in utilizing data. First, a brief review of the current practices in bridge asset management in the United States is presented. The review motivates explorations for data-driven and cross disciplinary improvements in understanding the impacts of construction, maintenance, and investment decision making through a preliminary framework including data, tools and methods from transportation systems analysis, economic impact analysis and social impact analysis areas. The mentioned explorations are demonstrated with a Los Angeles case study developed in collaboration with the asset management team of California Department of Transportation (Caltrans) District 7. Bridge Asset Management in the United States Deterioration/Condition Rating Assessment In the existing Bridge Management Systems (BMS), deterioration is often assessed based on visual inspections with condition states attributed to bridge elements (Ghodoosipoor et al. 2013). The National Bridge Inventory (NBI) general condition ratings (GCRs) are used to rate the bridge compared to its as-built status. There are many pieces of information considered in the rating such as the bridge materials, deck, superstructure, 91 and substructure components physical conditions. As per the FHWA Coding Guide (USDOT FHWA 1995), the GCRs are then assigned to the bridge on a numerical scale ranging from 0 (failed) to 9 (excellent). This assignment is based on the GCRs of the decks (NBI Item 58), bridge superstructures (NBI Item 59), and bridge substructures (NBI Item 60), and the conditions are categorized as 1- Good Condition (when the minimum CR of the three NBI items for a bridge is 7, 8, or 9), 2- Fair Condition (when the minimum CR of the three NBI items for a bridge is either 5 or 6), and 3- Poor Condition (when the minimum condition rating of any of the three NBI items for a bridge is 4 or below). Although GCRs help in providing general categorization of a bridge’s preservation, rehabilitation, and/or replacement needs, they remain to be too broad to allow the determination of specific work activities and the estimated cost. Thus, the National Bridge Elements (NBE) data are used (data on primary structural components of bridges) to assess the condition of the primary load carrying members, such as the various material and construction decks types, railings, superstructures, and substructures. The bridge elements each also have four condition states (1– good, 2–fair, 3–poor, and 4–severe), denoted by CS1, CS2, CS3, and CS4. Total quantities are assigned to these four condition states depending on the element (USDOT 2018). Table 3.1 shows common actions taken based on the Bridge Element Condition State. Table 3.1 Common actions based on Bridge Element Condition State (adopted from U.S. Department of Transportation, 2018) Condition State Description Common Actions 1 Varies depending on element – Good Preservation/Cyclic Maintenance 2 Varies depending on element – Fair Cyclic Maintenance or Condition-Based Maintenance when cost effective 92 3 Varies depending on element – Poor Condition-Based Maintenance, or Rehabilitation (if quantity of poor more than limits making Condition-Based not cost effective), or Replacement (if Rehabilitation is not cost effective) 4 Varies depending on element – Severe Rehabilitation or Replacement Options in Bridge Asset Management to Maintain Bridge Serviceability Before delving into the different available options for agencies to maintain and preserve their infrastructure, it is necessary to differentiate between the different levels of actions available (as shown in Figure 3.1. USDOT’s report (USDOT 2018) classifies it into 3 major levels starting from preservation/preventive maintenance and moving to rehabilitation, and then replacement. Maintenance is described as the “work that is performed to maintain the condition of the transportation system or respond to specific conditions or events that restore the highway system to a functional state of operations.” It is described as a key aspect of the entire asset management plan of an agency, and further includes: ● Routine maintenance which includes work done with no long term preservation value, conducted merely as a reaction to seasons, events, or activities, such as snow removal, fixing concrete decks as a result of an accident, or graffiti removal. ● Bridge preservation entails strategies aimed at preventing, delaying, or reducing bridges deterioration to keep them in good or fair condition, which extends their service life and thus delay the necessity for costly rehabilitation or replacement before they seriously deteriorate. It could be cyclical (set frequencies such as deck sealers or bridge cleaning/washing) or condition-based (in response to defects to 93 improve the defective component/element condition thus may not end up increasing the overall bridge condition such as exposed reinforcement). Figure 3.1 Bridge Action Categories (adopted from U.S. Department of Transportation, 2018) USDOT (2018) spells out the best practices for bridge preservation to include identifying the need in a consistent manner, which could be based on the NBI GCR component condition ratings, inspections, or element condition ratings, a bridge management system that aims at preservation strategies, a process or some form of prioritization that aligns with the agency’s objectives, and a feedback loop to verify the robustness of the process. The next level of intervention is rehabilitation which would entail major work and restoration of one or multiple elements of a structure that is needed to restore for the bridge’s structural integrity and to fix any safety issues, such as deck replacement. Because of the major work entailed, it requires significant resources whether such resources are in terms of manpower, time, and/or financial cost. The final and highest level of intervention is the total replacement of an existing bridge, also requiring a significant amount of resources. 94 How do owners and operators of bridges take maintenance decisions? The decision to rehabilitate versus replace requires a consideration of life-cycle costs as well as other economic factors. The decisions are also driven by limited funding as well as the prioritization of which assets should be maintained first to support the overall corridor mobility. A successful bridge asset management program will thus need to strategically balance between preservation, rehabilitation, and replacement, given these funding constraints (USDOT 2018). Figure 3.2. shows how these three options relate to the bridge’s condition over time, with preventive maintenance (while the bridge is in fair or good condition) regarded as a measure that extends the bridge’s service life, thus postponing the costly options of rehabilitation or replacement. FHWA highlights that an agency’s asset management process needs be a ‘strategic and systematic’ one where the focus is not only on engineering but also on economic analysis that leverages quality data and information to understand how the mentioned options could result in the improvement or the preservation of the asset (USDOT 2018; Weykamp et al. 2009). More specifically, the bridge management process synthesizes information on the bridge conditions and risks as well as the actions and means of execution that are appropriate and available for efficient maintenance work programming (Weykamp et al. 2009). Figure 3.2 Bridge Conditions over Time (adopted from U.S. Department of Transportation, 2018) 95 Federal Guidance It is widely asserted that the ‘worst first’ approach that was used in bridge asset management for a long time is no longer sustainable and does not allow agencies to maintain fair to good condition ratings for their bridge infrastructure given the backdrop of underfunding. This has led many agencies to initiate a paradigm shift to a sustainable asset management approach capable of a ‘mix of fixes’. Further, the federal Moving Ahead for Progress in the 21 st Century Act (MAP-21) of 2012 has clearly directed state DOTs to use Transportation Asset Management (TAM) practices for tracking the conditions of highways and bridges, and mandated it in federally funded projects (Robjent et al. 2020). With bridge preservation being an instrumental element in sustaining the state of good repair for US highway facilities—as clearly highlighted by MAP-21 and the Fixing America’s Surface Transportation (FAST) Act—, maintenance decision making is addressed and outlined for agencies in many guidebooks and reports at both the federal and the state level (USDOT 2018). At the federal level, there has been a plethora of guidelines, policies, and reports aiming to inform state agencies (USDOT 2018; Weykamp et al. 2009). USDOT’s Preservation Guidebook published in 2018 outlines general strategies used to improve existing bridge preservation programs, and further defines bridge preservation terms and commonly employed maintenance actions (USDOT 2018). The Scan 07-05 Project, requested by AASHTO aimed at scanning bridge management programs across the US. The project team reviewed 24 DOTs bridge maintenance guidelines, manuals, and policies as well as held meetings with 13 DOTs staff. Based on the data collected, they highlighted best practices that were proven to improve the overall bridge conditions. These practices recommended focus on three overarching categories; bridge management process, preventive maintenance, and agency’s support. As for the bridge management process, the team recommended (1) a consistent strategy for identifying maintenance needs, and further storing the needs in a corporate database for all program managers, (2) an alignment of the performance measures used to monitor the network conditions to the agency’s objectives, as well as continuous management support 96 of these measures, (3) a projects prioritization procedure that is based on agency objectives for maintenance and preventive maintenance of deficient and good bridges respectively, as well one that accounts for the overall network performance, risk, and the effects of deferred intervention (automated evaluations using multi-objective approaches are crucial in analyzing these different factors), (4) both main and regional DOT offices engagement in this prioritization procedure that encompasses a network-level and a bridge-by-bridge projects selection, (5) finally, a verification process integrated with the DOT data systems that confirms and updates information on maintenance actions completed and performance measures met. As for preventive maintenance, it needs to be continually assessed and forecasted so maintenance is not sought as a response to a bridge reaching a poor condition, but rather to have a timely plan for bridges when they are in fair or good condition. Finally, the DOT organization at all levels should support and embrace the bridge management program, including the central DOT office that even though operating with quantitative measures should account for the regional staff first- hand knowledge (Weykamp et al. 2009). Pathways for Data-Driven Advancements in Bridge Asset Management In terms of bridge condition assessment, despite the needs for more intelligent condition assessment and prediction tools that could be utilized in a proactive manner, current processes are able to identify the maintenance needs and bridge conditions in a holistic fashion, i.e., there exist some data on every bridge and its condition. Knowing such information, agencies are able to work out the engineering alternatives given the MRR options described in previous sections. However, current processes are failing to consider social, environmental and economic impacts of the engineering decisions and alternative scenarios that arise from them. The author asserts that more attention is needed—for improving the socioeconomic outcomes of asset management decision-making—to the impacts of MRR activities (1) on system-level (network wide) performance indicators (e.g., Total VHT, VMT, Delay), (2) on accessibility to jobs, goods and services, (3) on the local, 97 regional, national economies fueled by the supply chains served by the transportation infrastructure, (4) on environmental and public health outcomes such as emissions and air quality. To account for all such impacts and to implement decision-making procedures accommodating such advancements require a more widespread use of data, tools and methods as well as working across disciplines. While the mentioned types of impacts are quantified, an important implication of their accounting is the need to also consider their spatial distribution with the objective of identifying the disproportionately affected communities (e.g., by loss of accessibility to employment, healthcare and recreation opportunities). Within the scope of this chapter, the author focuses on (1), (2) and (3) only and leave environmental and public health outcomes for consideration in his future work. Given this scope, in the following section, the authors briefly introduce concepts in accessibility (in equity context), in an effort to establish a foundation for the analytical framework and its implementation in a case study that follows 20 . Review of Accessibility Concepts in Asset Management and Equity Contexts Accessibility, or spatial accessibility more precisely, is the concept that describes the level of ease with which people are able to access spatially distributed opportunities and activities. Since Hansen (1959) conceptualized it as a potential opportunity for interaction, this area has been researched extensively by researchers from various domains. Accessibility has played the major explanatory role in spatial econometric theories of the city. It has been considered to be the key variable in the determination of urban rents densities and land uses. Recently, research attention has been centered on the role of accessibility in the evaluation of equity of transportation investments. This is due to the known relationship between accessibility and quality of life, i.e. enhancing access to fundamental living opportunities such as employment, healthcare, green spaces, 20 Transportation systems analysis was extensively discussed in Chapter 2, therefore the author leaves a repetitive background discussion out. 98 education, social networks leads to improvements in the quality of life (Martens et al. 2012). One of the major applications of spatial accessibility is to assess the employment environment. Basically, it is believed that increasing accessibility to businesses helps achieve economic progress. Forming and developing a city comes from the positive effect of urban agglomeration (phenomenon describing the benefits resulting from the concentration of economic activities (Fujita 1989)). For example, different stores operating in the same market utilize opportunities to advertise and sell to customers who are in the area to visit another store, leading to an increase in sales. To meet the increased demand, stores expand their business and hire more people, creating an even more attractive location for that market. Moreover, higher accessibility enables goods to be delivered quickly and efficiently, which results in more efficient supply chains driving costs down. Overall, urban agglomeration stimulates consumer activity. This idea is also supported by previous studies highlighting that employment outcomes are highly dependent on accessibility (Helling 1998; Hu 2017). Understanding spatial barriers to employment in cities is a prime interest for geographers, urban planners, and many other specialists trying to mitigate unemployment and poverty. In the context of equity, the Spatial Mismatch Hypothesis is well-known as a classical example of accessibility analysis of the employment environment. The SMH, initially proposed by Kain (1992), is the theorem that many jobs are difficult to access for socially vulnerable groups such as African-Americans, other minorities or lower income segments in general. Grengs (2012), for example, illustrated that members of minority communities living in central Detroit have less opportunities to find a job due to their slower travel modes (using mainly public transit), and their residential location. Another interface of accessibility and equity is healthcare. It is widely accepted that geographical accessibility (usually assessed in terms of travel time required to reach facilities) and availability (usually assessed in terms of the number of beds available) play the main role in determining healthcare access in a given region (Neutens 2015). Utilizing GIS, many scholars and practitioners have been revealing the spatial disparities in 99 healthcare accessibility and people’s health conditions. For example, Gage and Calixte (2007) proved that closeness to maternal health care is a key factor in decision-making for pregnant women to use the service. The same phenomenon is observed in other healthcare services. Hiscock et al. (2008), for example, provided evidence from a study in New Zealand that there is an inverse relationship between travel time and frequency of visits to general practitioners and pharmacies. To maintain one’s health, securing access to green spaces such as parks is also important, especially for urban dwellers living in densely populated areas. This is due to the known positive effects of daily exposure to green space both mentally and physically. For example, Mitchell and Popham (2008) found that inequalities of the risk of circulatory diseases derived from socio-economic status can be improved by increasing green- accessible areas. In addition, people living in areas having a higher accessibility to green spaces have more ability to cope with stressful life events (Van den Berg et al. 2010). Transportation owners and operators as well as researchers utilize accessibility measures to assess plans for the construction of a new transportation infrastructure (e.g., roads, bridges) because accessibility analysis enables them to underscore the expected regional economic outcomes of a planned investment. For instance, Ribeiro et al. (2010) demonstrated that purchasing power in Portugal, especially in the less-developed regions, increased thanks to better accessibility associated with new transportation infrastructure investments. Although many DOTs publish the results of the accessibility measures especially for job accessibility when installing new routes or facilities (e.g., TPB 2018), little accessibility research has been conducted to show the impact in terms of maintenance. For example, AASHTO (2013) listed some key elements of evaluating asset performance toward effective management but did not include accessibility. One major reason attributed to the relative lack of attention to accessibility is that maintenance activities are temporary events which are determined by the physical conditions of the infrastructure (Burningham and Stankevich 2005). Another potential reason is that traffic-based performance, such as changing traffic speeds, is considered sufficient to assess the 100 impact of transportation planning. In fact, AASHTO (2013) touches on traffic delay as an indirect factor of assessment. However, accessibility is the ultimate goal of mobility and the main driver behind travel demand. People do not travel without a purpose with rare exceptions (Litman 2017) and accessibility is a fundamental aspect of our daily lives. Further, populations with socio-economic disadvantages are likely to be more affected by accessibility changes due to the environments and circumstances they live in. For example, in many urban areas, lower income earners have to travel longer to reach their workplaces due to various reasons including a higher public transit-dependency (e.g., (Antipova et al. 2020)). Because MRR of transportation infrastructure causes negative impacts on accessibility, accessibility needs to be a major determinant in MRR decision making. Quantifying and Mitigating Impacts on Vulnerable Populations In order to promote transportation justice, it is expected that planners investigate which segments of the society will shoulder the burden during the MRR projects. It should be avoided that the burden accumulates on vulnerable groups. Evenness measures of burden exposure are computed by comparing the spatial distributions of different groups among units in a metropolitan area. In order to convey the results of the analysis and to communicate with stakeholders, making web maps such as ArcGIS Dashboard is good way because maps enable the visualization of spatially distributed impacts making their communication much easier. Mapping can be used for conducting counter activities as well, which is called Counter Mapping. Peluso (1995) exemplified the case of countermapping in Indonesian forests resulting in this movement quickly spreading all over the world. Counter mapping empowers socioeconomically vulnerable groups to protest or demand change from governments or large corporations. Besides map-making, quantifying an easy-tu-understand index would be expected to help identify the bias. The most widely used measure of evenness is the Dissimilarity Index (Duncan and Duncan 1955). However, this index has shortcomings at times in demonstrating which way inequity leans. Therefore, some other approaches have been 101 also proposed and widely used. One of them, used in this research, is the Location Quotient (LQ). This index expresses the level of concentration on a particular industry, cluster, occupation, or a demographic group in a sub-region as compared to the whole region. Pruitt et al. (2015) applied LQ to reveal the relationships between residential racial segregation and mortality of breast cancer. Delineating the social inequality with a simple number helps its integration with other criteria such as physical conditions of bridges, toward comprehensive decision-making. Measurements of Accessibility There are several ways to measure accessibility. A common categorization has two types: location (or place)-based and person (or individual)-based. In order to measure location- based accessibility, information on 3 parameters are needed: (1) a reference point of origin and destination (e.g., often this is home location or zonal centroid), (2) a volume of activities at destinations (e.g., number of jobs in a zone) and (3) the cost of moving from the origin to destinations, (e.g., travel time). Then, accessibility is measured by counting the total volume of a given activity at destinations that can be reached within a specific time or distance (i.e., with respect to a predetermined threshold) from each origin. This approach is termed cumulative opportunities. A slightly more complex approach is attained by assigning more weight to values of destinations closer to the origin than ones further which is called the gravity-based approach. Although these approaches only focus on the supply side originally, recent studies have started taking into account the demand side and its interactions to seek methodological advancements. Especially in studies of healthcare accessibility, this relatively new method is called the two-step floating catchment area (2SFCA) method (Luo and Wang 2003). Assessment of person-based accessibility is divided further into two approaches based on the measures they use: utility-based measures and space-time measures. The former is based on the log-sum of discrete choice models applied to destination choice analysis (Ben-Akiva and Lerman 2021) where the latter is based on time geography studies (Hagerstrand 1970). These person-based approaches will bring more precise 102 results for individuals but require the collection of data on personal characteristics to define preference or time-constraints. Therefore, they are considered hard to implement given data availability challenges. To make the outcomes of this research more intuitively understandable for stakeholders and more generalizable, the author and his team prefer on the location- based approach. This is due to more relaxed constraints with regards to data availability, which is a critical determinant for public sector agents facing financial and technical constraints. Location-based approaches can be readily implemented with public data including those from Census Transportation Planning Products (CTPP), Topologically Integrated Geographic Encoding and Referencing System (TIGER), and other open datasets such as cities’ own GIS portals and OpenStreetMap (OSM). Location-based approach also fits the scale of analysis at the regional level, which is the SCAG (Southern California Assoc. of Governments) area for the case study presented in the following sections. Framework for Transportation Systems, Accessibility and Equity Analyses for Bridge Asset Management In the light of mentioned gaps in research and practice, the objective was to design and test a preliminary framework based on transportation systems analysis and accessibility/equity analyses supported by economic impact analysis, all of which were largely or entirely missing from the partner agency’s (Caltrans District 7) approach to prioritization of MRR projects. Figure 3.3 demonstrates the first two components elaborated on in this section. 103 Figure 3.3 A framework to investigate system-level impacts of MRR decision making on transportation system functionality as well as accessibility and equity. 104 Transportation System Analysis The analysis of the transportation network disruption due to engineering alternatives (e.g., closures resulting from MRR projects as well as new investments) can be quantified with a metropolis scale 4-step travel demand model in a similar fashion with CRAFT (See Chapter 2 and Figure 3.3). Specifically, system components are evaluated with to the planned projects in terms of closures and durations of those closures. This is done to simulate the effects of the planned project (and all its alternative scenarios) in a consistent manner. In the case of bridges, if a bridge is planned to be closed for a project, the link corresponding to that bridge in the network model underlying the travel demand model is—partially or fully—closed to operation, where the nature of the closure is determined by project requirements. For example, a full replacement of a bridge will naturally result in a full closure of its corresponding link in the network model throughout the duration of the project. With such information, a number of network topologies representing engineering alternatives can be modeled to capture the network supply conditions throughout the disruption timeline (Initialization). Initial skim matrices are computed to find the OD costs for TAZ (Traffic Analysis Zone) pairs (Network Skimming). These costs inform Trip Generation where trip production and trip attraction models estimate the number of trips generated for all trip purposes (from and to all TAZs) which are then balanced and distributed throughout the region via different modes (Trip Distribution and Mode Choice). The calculated travel demand is then segmented into finer time periods (Time-of-Day Choice) and are used to assign the loads into the network to solve for the traffic assignment problem (Assignment). With the new ‘congested’ link costs, a new iteration begins with Network Skimming and this loop runs until convergence to user- equilibrium. In this manner, equilibrium in a given network version (of a given engineering alternative) allows the quantification of all system functionality indicators (VMT, VHT, Delay, congested links and level of congestion, average speed information, patterns of reconfiguration of flow due to closure, etc.). Along with the information on system functionality, accessibility analysis module (See Figure 3.3) requires spatial data on point-like features representing all facilities of 105 concern, i.e., facilities over which accessibility is to be investigated (e.g., hospitals, green space, banks, grocery shops, etc.). Using the information on the system and the facilities, a 2-step catchment area problem is modeled for every engineering alternative. Accessibility Analysis As mentioned previously, the two-step floating catchment area method (2SFCA) is used to assess service availability in response to the demand. Since Luo and Wang (2003) developed this method, it has been applied widely in accessibility studies. Two genres of datasets are needed for 2SFCA: (1) locational data of service facilities with attributes on capacity; (2) locational data on the demand (the population that is meant to access the facilities). Methodologically, the 2SFCA method is operated as follows: Step 1-1. Set the travel time threshold t from each facility j; this is set as 60 minutes in the case study that follows. Step 1-2. Compute the facility capacity to population ratio Rj for each facility. Step 2-1. Find all facilities within the travel time t from each populational reference point, these are the centroids of each TAZ in the case study that follows. Step 2-2. For each populational point, sum up the Rj values of those facilities, obtaining the accessibility index for the area A i. These steps are mathematically written as Step 1: For each facility j, Rj = "# ∑ & ! " # Step 2: For each TAZ i, , Ai = ∑ R ( ) * where, t is travel time threshold, Cj is the capacity of facility j, P i is total population living in area i, Rj = the facility to population ratio of facility j. 106 Equity Analysis Based on considerations discussed previously, this research employs the Location Quotient as an equity index. This index, being rooted in economic geography, has been used to evaluate the degree of specialization to a specific industry in a target area compared to the entire region. Recently, it is also being used for assessing the level of specialization in other contexts. For instance, Pruitt et al. (2015) used the LQ to assess residential segregation and mortality due to cancer among races in Texas. In this case, the LQ is defined as follows: 𝐿𝑄 = 𝑃 '( 𝑃 ( 𝑃 ' 𝑃 ) ( where Pij is the population belonging to specific group i in sub-area j (this is set as the people living in the top 5% affected TAZs in the case study that follows), Pj is the total population in sub-area j, Pi is the total regional population belonging to a specific group i, Pt is the total population in the region. Hence, LQ greater than 1 indicates that the impact is more concentrated on a given group, and less than 1 indicates that other groups of people are relatively more affected. Case Study: Investigating Engineering Alternatives As part of the joint work on the mentioned NSF project (NSF 2020), the author and his team worked closely with Caltrans District 7 Asset Management Team to understand the ‘data’ gaps in their asset management processes. The shortcomings revealed by literature review was largely confirmed as the agency evaluates engineering alternatives with fewer dimensions of data than currently possible. To summarize, despite a prioritization scheme based on data on bridge conditions, seismic risk, average daily/annual vehicle counts, etc., (1) the agency was not utilizing a travel demand model to simulate the impacts of engineering alternatives on regional travel with holism and at high resolution, (2) the agency was not quantifying expected shifts in accessibility to services and associated consequences in equity, and (3) the agency did not have means to quantify regional 107 economic impacts and the economic facet of their decision only included conventional benefit/cost analyses. Apart from obstacles in obtaining the mentioned insights, these shortcomings create a number of operational challenges, e.g., in communicating investment and MRR decisions with the public. In that junction, the agency often finds itself unprepared to justify investments as the ‘not-in-my-backyard’ type of complaints arise from the public. It is believed that the insights generated through the explorations discussed in chapter and their advancements in the author’s future work will help owners and operators of infrastructure in justifying their decision-making. This could also make some other processes easier such as seeking federal and state funding towards MRR which is critical to enhancing the sustainability and resilience all civil systems. As a an initial deployment of the framework illustrated with Figure 3.3, the asset management team of Caltrans District 7 offered alternative decision making contexts with examples. A bridge on the I-210 freeway in Los Angeles (in Irwindale, on the San Gabriel River, Bridge No. 53-1867) was selected. This bridge was selected for a maintenance project where 2 hinge diaphragms were to be reconstructed, bridge railing was to be upgraded and electroliers were to be reinstalled. The bridge was selected based on the agency’s Average Daily Traffic (ADT) and Waterway criteria. The question the author and his team were interested in is shown in Table 3.2. There are two alternatives on how the project would be conducted in terms of schedule and closures. Alternative 2: Full Closure costs 3 million USD less than Alternative 1: Half Closure in terms of project cost. 108 Scenario 1: Half Closure Scenario 2: Full Closure Cost Schedule Cost Schedule Total ($k) Begin Duration Phasing Total ($k) Begin Duration Phasing $29,000 10/28/2021 Two 126 hours freeway closure Each closure will close all lanes in one direction and divert traffic to lanes in the opposite direction, going from 6 lane to 3 lanes in each direction. $26,000 10/28/2021 134 hours closure One single closure of both directions. Table 3.2 Engineering alternatives for Bridge No. 53-1867 on the San Gabriel River. The author deployed the framework to first quantify the changes in system level functionality indicators for both scenarios (VMT, VHT, Delay, etc.) that show the outlook in traffic conditions. 109 Figure 3.4 Changes in Vehicle Hours Traveled (VHT) under Scenario 1: Half Closure. 110 Figure 3.5 Changes in Vehicle Hours Traveled (VHT) under Scenario 2: Full Closure. 111 Figure 3.6 TAZs most affected by changes in hospital accessibility under Scenario 1: Half Closure. 112 Figure 3.7 TAZs most affected by changes in hospital accessibility under Scenario 2: Full Closure. 113 Scenario 1 - Half Closure Performance Measure Change %Change Average Speed (mph), L_AND_M -0.03 -0.08 Vehicle Miles Traveled ('000), L_AND_M -26.24 -0.01 Vehicle Hours Traveled ('000), L_AND_M 9.36 0.07 Vehicle Hours Delay ('000), L_AND_M 3.93 0.14 Average Speed (mph), HDT -0.10 -0.21 Vehicle Miles Traveled ('000), HDT 1.88 0.01 Vehicle Hours Traveled ('000), HDT 1.49 0.22 Vehicle Hours Delay ('000), HDT 1.37 1.05 Average Speed (mph), ALL -0.03 -0.09 Vehicle Miles Traveled ('000), ALL -24.37 -0.01 Vehicle Hours Traveled ('000), ALL 10.85 0.08 Vehicle Hours Delay ('000), ALL 5.30 0.17 Scenario 2 - Full Closure Performance Measure Change %Change Average Speed (mph), L_AND_M -0.54 -1.56 Vehicle Miles Traveled ('000), L_AND_M 103.35 0.02 Vehicle Hours Traveled ('000), L_AND_M 201.38 1.61 Vehicle Hours Delay ('000), L_AND_M 157.48 5.44 Average Speed (mph), HDT -0.62 -1.34 Vehicle Miles Traveled ('000), HDT 108.21 0.34 Vehicle Hours Traveled ('000), HDT 11.65 1.70 Vehicle Hours Delay ('000), HDT 8.15 6.26 Average Speed (mph), ALL -0.54 -1.54 Vehicle Miles Traveled ('000), ALL 211.57 0.05 Vehicle Hours Traveled ('000), ALL 213.03 1.61 Vehicle Hours Delay ('000), ALL 165.64 5.47 Table 3.3 System functionality indicators for the regional network (SCAG region) under Scenarios 1 and 2. The author—to conduct the transportation system analysis component of the framework—simulated both alternatives with the Southern California Association of Government’s Regional Travel Demand Model (RTDM) (SCAG 2016) to estimate the disruption of regional mobility (such as change in average speed, traffic delays, and vehicle-miles-traveled). The results are given with Table 3.3. Based on new equilibriums 114 in the network for Scenario 1 and Scenario 2 (See Figure 3.4 and Figure 3.5), hospital accessibility was also assessed and considered within equity context (See Figure 3.6 and Figure 3.7). Results of the analysis showed how the full closure (Scenario 2) expectedly creates a more significant disruption in the transportation system, compared to Scenario 1. The reconfiguration of flow in both alternatives were also observed in high spatial resolution enabling the identification of neighborhoods suffering unusually high levels of congestion. Scenario 1: Half Closure Scenario 2: Full Closure Race LQ Race LQ White 0.899430 White 0.869471 Black 0.7448 Black 0.691562 Asian 1.341535 Asian 1.857111 Others 1.140246 Others 0.940899 Non White 1.13943 Non White 1.180832 Ethnicity LQ Ethnicity LQ LQ-His/Lat 1.149593 LQ-His/Lat 0.914471 LQ-Others 0.871324 LQ-Others 1.073624 Table 3.4 Equity results for both scenarios across different racial and ethnic groups. 115 Equity results presented in Table 3.4 show that the Asian group suffer the most from a loss of healthcare accessibility in both scenarios where the full closure scenario has a significantly higher adverse effect. In general, LQ greater than 1 shows more concentrated effect as mentioned previously. Other equity results can be read this way. Economic Impacts of MRR-related Closures As mentioned, asset management teams of agencies including Caltrans District 7’s, do not formally study impacts of MRR decisions on regional economies. Typically on the economic facet of their decisions are project costs and financing considerations. This section elaborates on how the impacts on regional economy (e.g., GDP) are assessed. To complement the framework illustrated on Figure 3.3 with an analysis of economic impacts, The TERM model was employed to conduct the CGE simulations. The TERM model is a multi-regional CGE model with a high degree of regional details, developed by the Centre of Policy Studies at Victoria University (CoPS 2019). As a CGE model, TERM overcomes many limitations of IO models such as linearity, lack of substitution possibilities and resources constraints, and lack of price-response behavioral content. Apart from that, the TERM model has a well-extended sector structure and regional detail as well as a well-constructed database which was developed based on detailed official national and regional statistics. For instance, in the TERM model, well-built trade matrices are constructed based on regional and inter-regional trade data, which contains information about sources (i.e. imported or domestic), origins, and destinations of commodities. In the model, transportation costs are included in the trade margin matrices, which record the value of any trade or transport margins used to deliver commodities to users (Horridge 2012). This brings convenience to the investigation of indirect economic impacts caused by freight flow transportation disturbances resulting from closures. 116 All identified direct impacts caused by transportation infrastructure closures are put into TERM models as shocks through a series of “CGE drivers” (Rose 2017). Changes in freight transportation costs due to increased travel time can be simulated by adjusting trade margin variables, which will affect the delivered prices of commodities and services. The increased costs caused by increased travel times of freight flow are introduced into TERM models as a shock to the technical efficiency parameter of labor factor usage for transportation sectors. In this research, the costs of increased travel time of passenger flows are based on the value of travel time. Further investigation can be done by looking into detailed results such as output losses distribution across industries and regions, employment and price changes, etc. To calculate the scale of ‘shocks’ induced on the model, the increased transportation costs are used and calculated in the following ways. For increased travel distances due to closures, associated costs are calculated according to vehicle-based operating costs and ownership costs, such as fuel costs, repair and maintenance fees, tires, tolls, insurance premiums, permits, and license fees and vehicle lease or purchase payments. For increased travel times, the method depends on the type of flow. For freight flows, costs can be estimated based on transportation sectors' labor costs, such as driver wages and driver benefits. For passenger flows, because CGE models do not take time as one of the endowments like labor, capital, land, etc., a direct link between increased travel times and economic costs is not available. This is handled by calculating the monetary costs of increased travel time of passenger flows based on the value of travel time (or the value of saved travel time) and by using the loss in terms of value of time to an increase in labor costs in the region. Table 3.5 shows the results of economic impact analysis calculations for both scenarios in the case study. As for the economic loss (in GDP terms), while the project cost was estimated at $29 million for Scenario 1 and $26 million for Scenario 2, the GDP loss was estimated at $2.2 million and $23 million for the LA Metro region, respectively. This clearly shows how a less expensive project cost alternative (Alternative 2) from a 117 project cost perspective, is not overall less expensive considering its resulting socioeconomic impacts. These impacts, if accounted (and quantified) for when MRR decisions are being made, would affect the agency’s prioritization of their maintenance projects. It would also aid decision-makers (transportation agencies in this example) in providing tangible data-driven reasoning to various stakeholders and taxpayers of why a project decision, although seemingly more expensive, is selected in lieu of the other. Scenario 1: Half Closure Truck Transportation Disturbances LA Metro SF Metro Rest of CA Rest of US US Total (%) -0.000008 - - - - GDP Losses ($76.35) - - - - Labor Costs (%) -0.000224 0.000005 0.000001 0.000003 -0.00001 GDP Losses ($2,137.84) $38.50 $7.77 $455.64 ($1,768.89) Combined Impacts (%) -0.000232 0.000005 0.000001 0.000003 -0.00001 GDP Losses ($2,214.19) $38.50 $7.77 $455.64 ($1,768.89) Scenario 2: Full Closure Truck Transportation Disturbances LA Metro SF Metro Rest of CA Rest of US US Total (%) -0.000087 - 0.000004 0.000001 -0.000004 GDP Losses ($830.32) - $31.07 $151.88 ($707.56) Labor Costs (%) -0.002337 0.00005 0.000011 0.000028 -0.000102 GDP Losses ($22,304.16) $384.97 $85.43 $4,252.62 ($18,042.67) Combined Impacts (%) -0.002424 0.000051 0.000015 0.000029 -0.000106 GDP Losses ($23,134.48) $392.67 $116.49 $4,404.50 ($18,750.22) Table 3.5 Real GDP Impacts of closure scenarios. 118 (in thousand dollars based on 2012 prices and percent reduction from baseline levels) 119 Discussion on Limitations and Future Work In this chapter, key outcomes from the presented research work as well as limitations and ideas on future work are discussed. According to the format of the thesis so far, this will be done in 2 consecutive sections on Comprehensive Assessment of Resilience (CRAFT) and Other Dimensions of Asset Management. Discussion on CRAFT The objective of the work behind CRAFT was to design and deploy a framework for the comprehensive assessment of resilience in transportation systems. CRAFT advances the research in resilience of transportation systems in 3 directions by: (1) incorporating a novel image-based modeling methodology advancing the regional hazard analysis component beyond the inventories traditionally used in this area, (2) adopting a metropolis scale travel demand model based on real socioeconomic data to achieve high- fidelity analyses of the transportation system at scale, (3) proposing an advanced socioeconomic impact analysis methodology leveraging a state-of-the-art CGE model informed by the hazard and transportation modules. The mentioned advancements are demonstrated in this thesis through the deployment of CRAFT for a scenario earthquake (M 7.3 Earthquake on Palos Verdes Fault in Los Angeles) which motivates the discussions on the case study presented in Chapter 2. The full integration of CRAFT for a seamless collaboration between its 3 modules was shown with the case study. So far in this research, based on the resilience definition of Frangopol and Bocchini (Frangopol and Bocchini 2011b), resilience is quantified—with respect to the continuous black line showing Q(t) on Figure 4.1—as the remaining functionality in the network following the disturbance stemming from direct physical damages to the infrastructure (e.g., bridge closures due to structural damages). This conceptualization of resilience is similar to the static resilience definition by Rose and Dormady (2018) which refers to using remaining resources efficiently to maintain function. In other words, static resilience in this context refers to the system-level indicators quantified by the user-equilibrium traffic 120 assignment results under the new network supply conditions in the degraded network. The author intends to investigate tactics associated with the concept of dynamic resilience defined by Rose and Dormady which is characterized as investing efficiently in repair and reconstruction in order to reestablish capacity as quickly as possible to regain function. An improved recovery curve that may be achieved through dynamic resilience is indicated as Q’(t) on Figure 4.1. Some of the resilience tactics to achieve the faster recovery curve could be allocating resources to more rapidly open critical corridors (e.g., Interstate 405) to service. CRAFT is tested for such a scenario in Section 2.6.9, and results are presented on Figure 2.18. Despite the success of the framework there, more resilience tactics need to be devised and studied with the framework. Other options include shifting the heavy duty truck traffic to less congested time periods during the day to compensate for less desirable congestion levels due to the hazard. One interesting resilience tactic that is currently considered is the change of signalization routines (traffic lights) in the major arterials neighboring the closed segments of the freeways. These arterials are simulated to carry volumes beyond their capacity which results in congestion. Signal control policies that take into account real-time queue length measurements to dynamically change green time allocations among various phases can help facilitate larger volumes through the urban arterials mitigating the rerouting based congestion partially (Hosseini and Savla 2016). Overall, adaptive tactics are expected to increase resilience, however, these need to be modeled for and simulated to measure improvements in network efficiency. Such analyses will bridge the gap between the research presented here and the practical management of the disaster by producing more actionable insights for operators and transportation agencies. Naturally, any analysis of this sort could also be utilized as input for economic impact analyses, paving the way for the economic meanings of increased network efficiency through the implementation of these tactics. In economic resilience, the difference between potential resilience and actual resilience is also noted. The existence of various coping measures that are modeled in this research does not mean they will be optimally realized given the likelihood of restrictive regulations, bounded rationality, and market failures. The idea is to estimate 121 the loss reduction effects of only potential resilience. However, the analysis provides insights to port managers and operators, businesses that rely on operations of the ports and the freight transportation network, and policy makers to identify and implement these powerful resilience tactics and enhance business contingency and continuity planning to cope with seaport and transportation network disruptions as targets for their decisions. Figure 4.1 Network resilience curve adapted from Frangopol and Bocchini (2011) and revised. Dashed red line indicates faster recovery path achieved through implementation of resilience tactics. With regards to modeling the post disaster travel behavior, the author has two options. The first and the simpler option is to make assumptions on behavioral responses such as the stay-at-home behavior due to reduced network functionality. In this case, the author could raise questions and devise sensitivity scenarios such as “What if 20% of the home- based-work trips (a commuting trip in the trip market strata) do not happen for a month after the earthquake in the immediate impact area?”. Stemming from that question, a new model run could explore the potential improvements in network performance and quantify the reduced congestion with respect to the fixed demand conditions. The second and the more complex option is to use the trip distribution (destination choice) and the mode 122 choice modeling components of travel demand models including the SCAG RTDM. Destination choice and mode choice models—apart from the models of home-based- school (college) trips—are based on utility theory with utility functions accommodating terms for travel distance and time parameters. Therefore, these functions can be used to estimate the change in destinations and modes. This can enable insights related to post disaster travel behavior, however, it is essential to note that these results will always be prone to a lack of validation with empirical data, simply because such data does not exist. Khademi et al. (2015) identifies the work of Nagae et al. (2012) who take traffic congestion and travelers' route choice behavior into account and performs a UE-based assignment to predict post disaster situation. Nagae et al. admit that whether or not such a static/equilibrium-based assignment is suitable for representing the actual traffic flows on a malfunctioning network after the earthquake is unknown, and emphasize the importance and necessity of further analyses and modeling of post-disaster traffic flows. In general, the loss of network functionality in the network affects transportation costs for virtually every industry in the region. The TERM model is used to capture the economic consequences of such effects. Income distribution is also investigated as discussed in Section 2.6.10. Therefore, the model linkages between network analysis and economic impact analysis modules are complete. The author has quantified economic impacts of disturbed transportation with simple IO approaches in a previous study (Wei et al. 2018), however, the current version of the framework is founded on more advanced CGE analysis. Some of the dynamic resilience tactics that will be studied in the future will require additional model runs (e.g., mode substitution of disrupted cargo, ship rerouting, etc.). For instance, if the port is assumed to be closed to service for a period of time, the truck trip tables that generate the truck trips in the Port HDT model will have to be updated not to load trips onto the network. On the other hand, if excess capacity 21 is incorporated as a resilience tactic, truck trips can still happen through the undamaged terminals. Such cross-disciplinary links that have not been made before in the analysis of port resilience 21 Excess capacity: utilization of unused capacity at undamaged terminals of the port to unload or load cargo that was originally handled in other terminals that experience facility downtime. 123 by economists will be made for the first time thanks to the complete coupling of a transportation model with an economic model. Yet another item of future work consists of investigating the environmental justice issues as well as advancing the resilience work towards considerations of accessibility and equity. Environmental justice is an important policy objective. It emphasizes reducing disparities in negative impacts of economic activity across racial/ethnic groups. Transportation equity is a broader term that began with concern about disparities across income groups in relation to matters of accessibility and affordability but has now been extended to disparities across racial/ethnic groups as well. The author will make attempts at extending these concepts of environmental justice and transportation equity to the area of natural hazards that affect transportation systems. In general, the economic and environmental impacts of disasters have in general been found to fall disproportionally on lower income groups and racial/ethnic minorities. The intention is to examine the extent to which this is the case for the scenario earthquake (7.3 M w Palos Verdes earthquake). A statistical test shows significant differences across mentioned groups in terms of impacts, however, these calculations are left out of the thesis for future work. A key limitation of CRAFT is that it isolates the transportation system and does not consider the interdependent interactions of transportation with other civil systems that are known to create additional exposure through cascading failures. Working across system- of-systems has not been the objective in author’s dissertation work due to the emerging research objective of designing a holistic framework for transportation systems that did not compromise from granularity and scale. The challenges of extending CRAFT to a system-of-systems setting will be many, but these could motivate exciting research problems in the author’s future work. In terms of generalizing CRAFT to hazard setting other than earthquakes, the author asserts that there is not a significant barrier. The hazard characterization and damage assessment module, with modifications, could be applied to other settings such as hurricanes, floods, etc. relatively quickly. The choice here naturally depends on the 124 study region itself and what types of disaster risk are of concern. If a multi-hazard setting is to be investigated, CRAFT could again be modified accordingly given the participation of subject experts in the domains of interest. The key idea here is that the framework, as per the objectives mentioned in the beginning of the thesis, motivates researchers to work across traditional silos. Today, in civil engineering departments as well as across other engineering or social sciences areas, disasters are being investigated more or less in an isolated fashion which prevents researchers from producing convincing and actionable insights with a potential to alter the way disaster risk is managed. The author plans to push resilience research forward by operating at the interface while keeping the data-intensive system-based investigations intact. Discussion on Other ‘Data’ Dimensions of Asset Management The project that motivated this portion of the presented work in this thesis (NSF 2020) brought a multidisciplinary team of experts together and leveraged government and industry partnerships to focus on the grand challenge of restoring and improving the national urban infrastructure, which cannot be solved without a convergence research approach that provides a framework to tightly integrate efforts across academic disciplines and stakeholders. The mentioned open knowledge network CIS-OKN represents an accelerated effort by a diverse group of researchers and stakeholders from universities, research institutions, federal and state agencies, and industry partners working together to develop effective and sustainable strategies, methods, and tools for data-driven discovery and decision making in the CIS domain. The team expertise includes multiple civil engineering disciplines (infrastructure systems, structural engineering, transportation engineering, water resources, construction engineering), urban and regional planning disciplines, and computer science and data science disciplines (data science and data analytics, AI and machine learning, sensing systems, information retrieval, natural language processing, image and video processing, mobile and cloud computing, energy-efficient computing, security and privacy of computer and 125 networked systems). The author took part in developing use cases for the proposed CIS- OKN on which the discussions in Chapter 3 are based. The larger effort included identifying alternative maintenance, preservation, and repair strategies and decisions for a bridge (or a set of bridges) based on the current conditions and the predicted future deterioration. Discussions on those tasks were left out in this thesis. However, based on improved prediction of future deterioration in bridges and enhanced datasets on infrastructure conditions will strengthen the methodology presented in Chapter 3 where the author will advance the work on identifying direct and life cycle maintenance costs, as well as maintenance scheduling and closure durations for all alternatives; capturing, simulating, and analyzing regional economic and social impacts of these decisions that are often very high but not considered in the decision- making process due to lack of integrated frameworks. These impacts include costs of increased travel times for passengers and freight; reduction in accessibility to employment, healthcare, and recreation opportunities; as well as the spatial distributions of these impacts, particularly when vulnerable populations are disproportionately affected. Following such advancements, based on the aforementioned analytics results and funding constraints, the author and his collaborators can work on optimizing, prioritizing, and visualizing effective maintenance strategies and actions. Presented work in Chapter 3 is perceived as a precursor of such end goals. Some exciting research questions on this front shall focus on how to bridge the data gap between agents in the government and private sector. Most businesses today make spatial decisions with segregated data because data linking and fusion across different stakeholders’ sources are not complete. For example, how does the insights generated at the regional economic level translate to individual businesses across the supply chains? Can we, in the future, inform business decisions whose success are directly related to the performance of networked civil systems that serve the consumers as well as the producers and other facilitators? What are the implications of MRR decisions on business interruptions? What are the implications for the insurance industry as a result 126 of such interruptions? In more catastrophic events where resilience is assessed, how does civil systems resilience relate to the economic recovery at the individual business level? These are highly complex but interesting questions as the presented works advance with time. The challenges are not limited to resolution but also relate to addressing such challenges in a consistent manner which has implications on modeling and simulation both for civil systems and economic/social systems. Another aspect in this vein of limitations and future work relates to how to address human decision making under varying network states. For example, in the estimation of economic impacts of passenger flow disturbances, because CGE models do not take time as one of the endowments like labor, capital, land, etc., a direct link between increased travel times and economic costs is not available. Although the calculations in Section 3.2 aim to calculate the monetary costs of increased travel time of passenger flows based on the value of travel time (or the value of saved travel time), the complexity of the effects caused by people changing their behaviors under transportation disruption circumstances is commonly overlooked. For instance, increasing time spending on the road for users could eventually lead to less participation in other activities. Considering that the utility and necessity vary from activities for different users, this will have a bigger impact on non-essential activities (e.g., leisure) instead of essential activities such as commuting. Consequently, final demands of commodities and services could be affected due to changes in traveler behaviors based on human preferences. Additionally, increased transportation costs can lead to changes in households’ consumption behaviors given budget constraints. The changes in final demand can be estimated based on household consumption patterns through observation or investigation. These can be simulated in terms of changes in households’ consumption preferences in CGE models. In this way, changes in the state of the transportation system can be converted into monetary values and thus can be introduced into CGE models as initial shocks. It needs to be mentioned that, throughout the thesis, it is assumed that people maintain their commuting trips after road closures. 127 Conclusion In summary, the presented convergent frameworks assessing transportation resilience and other dimensions of asset management are opening doors to further research in this domain. The author believes that the current state of this research strikes the analytic- synthetic balance discussed in the beginning of the thesis (Goldberg 1975). The main contribution of CRAFT to the field of resilience engineering, transportation and society-at-large is the multi-disciplinary framework itself. The metrics that are impacted by implementing the framework instead of conventional approaches are at 3 levels; (1) advanced modeling of infrastructure inventory and hazard characterization improvements addressing site-specific factors results in improved (structure-specific) fragility analyses; (2) high granularity transportation system analysis at scale, informed by (1) and based on authoritative network and socioeconomic data, allows the framework to holistically capture the disrupted (degraded) states of the network, (3) advanced economic impact and resilience analyses informed by (1) and (2), further enriched with considerations on income distribution, enable a multi-dimensional view on economic implications of transportation disruptions quantifying the effects on regional economy, impacts on different income brackets and effectiveness of resilience tactics. The author contends that the linkages across different participating disciplines in CRAFT are uncompromising in terms of leveraging data availability and the state-of-the- art in corresponding fields of modeling and simulation while seeking advancements both in terms of individual modules of the framework and the emergent capabilities that arise from their synergy. CRAFT’s proof-of-feasibility example, the case study presented in 2.6, demonstrates these emergent capabilities. Pushing the envelope further could also help achieve two-way linkages, e.g., the economic model feeding back information to transportation analysis, potentially motivating further research with implications on urban land use planning. Accessibility is also an important context of analysis in this vein as more of the newer investments in big 128 urban centers prioritize improved accessibility as it also aligns with other sustainability objectives such as decreasing transportation emissions. Coming back full circle on the ideas presented in the beginning, sustainable and resilient cities served by sustainable and resilient civil systems can only be fully achieved by ensuring the effective use of data. The author’s dissertation work has been pursued with this understanding which also shines light into his future work. The users of this research are professionals that are directly involved in disaster risk management and asset management related decision making for transportation infrastructure including bridges and road networks. Reactions from such professionals whether it is a modeling task force meeting at SCAG or research meetings at Caltrans have been positive and highlighted the need of further research in the investigated areas through more effective use of data, tools and methods. In the longer term, the vision is to establish the information transfer between academic research and policy-making to help engineer more sustainable and resilient transportation systems. Creating this broader impact is the ultimate goal. 129 References AASHTO. (2013). “Transportation Asset Management Guide.” 3(2), 54–67. Aloughareh, I. R., Ashtiany, M. G., and Nasserasadi, K. (2016). “An Integrated Methodology For Regional Macroeconomic Loss Estimation Of Earthquake: A Case Study Of Tehran.” Singapore Economic Review, WORLD SCIENTIFIC PUBL CO PTE LTD, 5 TOH TUCK LINK, SINGAPORE 596224, SINGAPORE, 61(4). Antipova, A., Sultana, S., Hu, Y., and Rhudy, J. P. (2020). “Accessibility and Transportation Equity.” Multidisciplinary Digital Publishing Institute. Argyroudis, S. A., and Pitilakis, K. D. (2012). “Seismic fragility curves of shallow tunnels in alluvial deposits.” Soil Dynamics and Earthquake Engineering, Elsevier, 35, 1–12. Asakura, Y. (2007). “Requirements for transport network flow models used in reliability analysis.” International journal of critical infrastructures, Inderscience Publishers, 3(3–4), 287–300. ASCE. (2016). “Failure to Act: Closing the infrastructure investment gap for America’s Economic Future.” American Society of Civil Engineers, 32. ASCE. (2017). Infrastructure Report Card. Ashrafi, Z., Shahrokhi Shahraki, H., Bachmann, C., Gingerich, K., and Maoh, H. (2017). “Quantifying the Criticality of Highway Infrastructure for Freight Transportation.” Transportation Research Record: Journal of the Transportation Research Board, 2610, 10– 18. Baker, J. W., Lin, T., Shahi, S. K., and Jayaram, N. (2011). New Ground Motion Selection Procedures and Selected Motions for the PEER Transportation Research Program. PEER Report 2011/03, Berkeley, CA. Bazzurro, P., and Cornell, C. A. (1999). “Disaggregation of seismic hazard.” Bulletin of the Seismological Society of America, 89(2), 501–520. Ben-Akiva, M., and Lerman, S. R. (2021). “Disaggregate travel and mobility-choice models and measures of accessibility.” Behavioural travel modelling, Routledge, 654–679. 130 Berdica, K. (2002). “An introduction to road vulnerability: what has been done, is done and should be done.” Transport Policy, 9, 117–127. Van den Berg, A. E., Maas, J., Verheij, R. A., and Groenewegen, P. P. (2010). “Green space as a buffer between stressful life events and health.” Social science & medicine, Elsevier, 70(8), 1203–1210. Bono, F., and Gutiérrez, E. (2011). “A network-based analysis of the impact of structural damage on urban accessibility following a disaster: The case of the seismically damaged Port Au Prince and Carrefour urban road networks.” Journal of Transport Geography, Elsevier Ltd, 19(6), 1443–1455. Brothers, D. S., Conrad, J. E., Maier, K. L., Paull, C. K., McGann, M., and Caress, D. W. (2015). “The Palos Verdes Fault offshore Southern California: Late Pleistocene to present tectonic geomorphology, seascape evolution, and slip rate estimate based on AUV and ROV surveys.” Journal of Geophysical Research: Solid Earth, 120(7), 4734–4758. Bruneau, M., Chang, S. E., Eguchi, R. T., Lee, G. C., O’Rourke, T. D., Reinhorn, A. M., Shinozuka, M., Tierney, K., Wallace, W. A., and Von Winterfeldt, D. (2003). “A Framework to Quantitatively Assess and Enhance the Seismic Resilience of Communities.” Earthquake Spectra, 19(4), 733–752. Bureau of Economic Analysis (BEA). (2019). “Quarterly Personal Income by State: Personal Income by Major Component (SAINC4). .” <https://apps.bea.gov/iTable/iTable.cfm?reqid=99&step=1#reqid=99&step=1> (Jun. 2, 2021). Bureau of Labor Statistics (BLS). (2019). “Occupational Employment Statistics.” <https://www.bls.gov/oes/tables.htm> (Jun. 2, 2021). Burfisher, M. E. (2017). Introduction to computable general equilibrium models. Cambridge University Press. Burningham, S., and Stankevich, N. (2005). “Why road maintenance is important and how to get it done.” World Bank, Washington, DC. Cetiner, B. (2020). “Image-Based Modeling of Bridges and Its Applications to Evaluating 131 Resiliency of Transportation Networks.” University of California, Los Angeles, Los Angeles, CA. Cetiner, B., Koc, E., Soibelman, L., Taciroglu, E., Lee, E. J., and Nutakki, A. (2019). “A Granular Simulation of Bridge Closures Due to a Southern California Scenario Earthquake and Its Effects on the Disruption and Recovery of Freight Traffic to and from Ports of Los Angeles and Long Beach.” International Conference on Sustainable Infrastructure 2019: Leading Resilient Communities through the 21st Century, American Society of Civil Engineers Reston, VA, 575–585. Chang, S. E. (2003). “Transportation planning for disasters: An accessibility approach.” Environment and Planning A, 35(6), 1051–1072. Chen, A., Yang, C., Kongsomsaksakul, S., and Lee, M. (2007). “Network-based acessibility measures for vulnerability analysis of degradable transportation networks.” Networks and Spatial Economics, 7(3), 241–256. Chen, Z., and Rose, A. (2018). “Economic resilience to transportation failure: a computable general equilibrium analysis.” Transportation, Springer, 45(4), 1009–1027. Cho, J. K., Gordon, P., Moore, J. E., Pan, Q., Park, J. Y., and Richardson, H. W. (2015). “TransNIEMO: economic impact analysis using a model of consistent inter-regional economic and network equilibria.” Transportation Planning and Technology, 38(5), 483– 502. Cho, S., Gordon, P., Moore, J. E., Richardson, H. W., Shinozuka, M., and Chang, S. (2001). “Integrating transportation network and regional economic models to estimate the costs of a large urban earthquake.” Journal of Regional Science, 41(1), 39–65. Cochrane, H. C. (1974). Predicting the economic impacts of earthquakes. Natural Hazards Research Paper. Crowther, K. G., Haimes, Y. Y., and Taub, G. (2007). “Systemic valuation of strategic preparedness through application of the inoperability input-output model with lessons learned from hurricane Katrina.” Risk Analysis, 27(5), 1345–1364. Duan, Y., and Lu, F. (2015). “Robustness analysis of city road network at different granularities.” 132 Space-Time Integration in Geography and GIScience: Research Frontiers in the US and China, Elsevier B.V., 411, 127–143. Duncan, O. D., and Duncan, B. (1955). “A methodological analysis of segregation indexes.” American sociological review, JSTOR, 20(2), 210–217. EM-DAT. (2020). “The International Disaster Database.” <https://www.emdat.be/>. Faturechi, R., and Miller-Hooks, E. (2014). “Measuring the Performance of Transportation Infrastructure Systems in Disasters: A Comprehensive Review.” Journal of Infrastructure Systems, 21(1), 04014025. FEMA. (2003). “Hazus-MH 2.1 Technical Manual: Earthquake Model.” Washington, DC. FEMA. (2004). FEMA 433: Using HAZUS-MH for Risk Assessment. Reston, VA. FHWA. (1995). Recording and Coding Guide for the Structure Inventory and Appraisal of the Nation’s Bridges. Washington, D.C. Frangopol, D., and Bocchini, P. (2011a). “Resilience as Optimization Criterion for the Rehabilitation of Bridges Belonging to a Transportation Network Subject to Earthquake.” Structures Congress, 2044–2055. Frangopol, D., and Bocchini, P. (2011b). “Resilience as Optimization Criterion for the Rehabilitation of Bridges Belonging to a Transportation Network Subject to Earthquake.” Structures Congress, 2044–2055. Freckleton, D., Heaslip, K., Louisell, W., and Collura, J. (2012). “Evaluation of Resiliency of Transportation Networks after Disasters.” Transportation Research Record: Journal of the Transportation Research Board, 2284(1), 109–116. Fujita, M. (1989). Urban economic theory: land use and city size. Cambridge university press. Gage, A. J., and Calixte, M. G. (2007). “Population Studies Effects of the physical accessibility of maternal health services on their use in rural Haiti.” Taylor & Francis, Routledge, 60(3), 271–288. Ganin, A. A., Kitsak, M., Marchese, D., Keisler, J. M., Seager, T., and Linkov, I. (2017). “Resilience and efficiency in transportation networks.” Science Advances, 3(12), 1–9. 133 Ghodoosipoor, F., Ph, A., and Dissertation, . D. (2013). DEVELOPMENT OF DETERIORATION MODELS FOR BRIDGE DECKS USING SYSTEM RELIABILITY ANALYSIS. Goldberg, M. A. (1975). “On the Inefficiency of Being Efficient.” Environment and Planning A: Economy and Space, SAGE Publications Ltd, 7(8), 921–939. Gordon, P., Moore II, J. E., Richardson, H. W., Shinozuka, M., An, D., and Cho, S. (2004). “Earthquake Disaster Mitigation for Urban Transportation Systems: An Integrated Methodology that Builds on the Kobe and Northridge Experiences.” Modeling Spatial and Economic Impacts of Disasters, 205–232. Greenberg, M., Lioy, P., Ozbas, B., Mantell, N., Isukapalli, S., Lahr, M., Altiok, T., Bober, J., Lacy, C., Lowrie, K., Mayer, H., and Rovito, J. (2013). “Passenger Rail Security, Planning, and Resilience: Application of Network, Plume, and Economic Simulation Models as Decision Support Tools.” Risk Analysis, 33(11), 1969–1986. Grengs, J. (2012). “Equity and the social distribution of job accessibility in Detroit.” Environment and Planning B: Planning and Design, 39(5), 785–800. Group, M. I. (2012). “Impact Analysis for Planning (IMPLAN) System.” Gueler, C. U., Johnson, A. W., and Cooper, M. (2012). “Case Study: Energy Industry Economic Impacts from Ohio River Transportation Disruption.” ENGINEERING ECONOMIST, TAYLOR & FRANCIS INC, 325 CHESTNUT ST, SUITE 800, PHILADELPHIA, PA 19106 USA, 57(2), 77–100. Hagerstrand, T. (1970). “What about people in regional science?” Ninth European Congress of the Regional Science Association. Hallegatte, S. (2008). “An adaptive regional input-output model and its application to the assessment of the economic cost of Katrina.” Risk Analysis, 28(3), 779–799. Hansen, W. G. (1959). “How accessibility shapes land use.” Journal of the American Institute of planners, Taylor & Francis, 25(2), 73–76. Hazus. (2012). “Hazus–MH 2.1: Technical Manual.” Federal Emergency Management Agency, 718. 134 Helling, A. (1998). “The effect of residential accessibility to employment on men’s and women’s travel.” Women’s Travel Issues Second National ConferenceDrachman Institute of the University of Arizona; Morgan State University; Federal Highway Administration. Hiscock, R., Pearce, J., Blakely, T., and Witten, K. (2008). “Is neighborhood access to health care provision associated with individual-level utilization and satisfaction?” Health services research, Wiley Online Library, 43(6), 2183–2200. Holling, C. S. (1973). “Resilience and stability of ecological systems.” Annual review of ecology and systematics, Annual Reviews 4139 El Camino Way, PO Box 10139, Palo Alto, CA 94303-0139, USA, 4(1), 1–23. Hollnagel, E. (2011). “Prologue: The scope of resilience engineering.” Resilience engineering in practice a guidebook /, Ashgate, Farnham, Surrey, England ; Hopkins, D. C., Lumsden, J. L., and Norton, J. A. (1991). “Wellington’s Lifelines in Earthquake Project: An Outline of a Major Review.” Lifeline Earthquake Engineering, ASCE, 774–804. Horridge, M. (2012). “The TERM model and its database.” Economic modeling of water, Springer, 13–35. Hosseini, P., and Savla, K. (2016). “A comparison study between proportionally fair and max pressure controllers for signalized arterial networks.” TRB 95th Annual Meeting Compendium of Papers. Hu, L. (2017). “Job accessibility and employment outcomes: which income groups benefit the most?” Transportation, Springer, 44(6), 1421–1443. Irimoto, H., Shibusawa, H., and Miyata, Y. (2017). “Evaluating the economic damages of transport disruptions using a transnational and interregional input-output model for Japan, China, and South Korea.” AIP Conference Proceedings, AIP - American Institute of Physics, USA, 110002 (6 pp.). Jaiswal, P., Van Westen, C. J., and Jetten, V. (2010). “Quantitative assessment of direct and indirect landslide risk along transportation lines in southern India.” Natural Hazards and Earth System Science, 10(6), 1253–1267. Kain, J. F. (1992). “The spatial mismatch hypothesis: three decades later.” Housing policy 135 debate, Taylor & Francis, 3(2), 371–460. Kajitani, Y., Chang, S. E., and Tatano, H. (2013). “Economic Impacts of the 2011 Tohoku-oki earthquake and tsunami.” Earthquake Spectra, 29(SUPPL.1), S457–S478. Khademi, N., Balaei, B., Shahri, M., Mirzaei, M., Sarrafi, B., Zahabiun, M., and Mohaymany, A. S. (2015). “Transportation network vulnerability analysis for the case of a catastrophic earthquake.” International Journal of Disaster Risk Reduction, 12, 234–254. Kim, E., and Kwon, Y. J. (2016). “Indirect Impact of Nuclear Power Plant Accidents Using an Integrated Spatial Computable General Equilibrium Model with a Microsimulation Module on the Korean Transportation Network.” Quantitative Regional Economic and Environmental Analysis for Sustainability In Korea, New Frontiers in Regional Science-Asian Perspectives, B. Kim, E and Kim, ed., Springer, Singapore, 141–152. Kiremidjian, A., Moore, J., Fan, Y. Y., Basoz, N., Yazali, O., and Williams, M. (2006). Highway Demonstration Project, Peer Report 2006/02. Koc, E., Cetiner, B., Lee, J., Nutakki, A., Soibelman, L., and Taciroglu, E. (2019a). “System- Based Resilience Assessment of Networked Transportation Systems in Metropolitan Areas: Case of Greater Los Angeles.” Proceedings of the 26th International Workshop on Intelligent Computing in Engineering, 2019. Koc, E., Cetiner, B., Rose, A., Soibelman, L., Taciroglu, E., and Wei, D. (2020). “CRAFT: Comprehensive Resilience Assessment Framework for Transportation Systems in Urban Areas.” Advanced Engineering Informatics, Elsevier, 46(July), 101159. Koc, E., Cetiner, B., Soibelman, L., and Taciroglu, E. (2019b). “System-based vulnerability and resilience assessment in mega-scale transportation systems: Towards data and model- driven methodologies.” Computing in Civil Engineering 2019: Smart Cities, Sustainability, and Resilience, American Society of Civil Engineers Reston, VA, 444–450. Koks, E. E., Carrera, L., Jonkeren, O., Aerts, J. C. J. H., Husby, T. G., Thissen, M., Standardi, G., and Mysiak, J. (2016). “Regional disaster impact analysis: comparing input-output and computable general equilibrium models.” Natural Hazards and Earth System Sciences, COPERNICUS GESELLSCHAFT MBH, BAHNHOFSALLEE 1E, GOTTINGEN, 37081, GERMANY, 16(8), 1911–1924. 136 Koks, E. E., and Thissen, M. (2016). “A Multiregional Impact Assessment Model for disaster analysis.” ECONOMIC SYSTEMS RESEARCH, ROUTLEDGE JOURNALS, TAYLOR & FRANCIS LTD, 2-4 PARK SQUARE, MILTON PARK, ABINGDON OX14 4RN, OXON, ENGLAND, 28(4), 429–449. Li, J., Crawford-Brown, D., Syddall, M., and Guan, D. (2013). “Modeling Imbalanced Economic Recovery Following a Natural Disaster Using Input-Output Analysis.” RISK ANALYSIS, WILEY-BLACKWELL, 111 RIVER ST, HOBOKEN 07030-5774, NJ USA, 33(10), 1908–1923. Li, P.-C., Rose, A., and Eduardo, B. (1999). “Construction of an Input-Output Income Distribution Matrix for the U.S.” 191–213. Lian, C., and Halmes, Y. Y. (2006). “Managing the risk of terrorism to interdependent infrastructure systems through the Dynamic Inoperability Input-Output Model.” Systems Engineering, 9(3), 241–258. Litman, T. (2017). Evaluating accessibility for transport planning. Victoria Transport Policy Institute. Lofgren, H., and Robinson, S. (2002). “Spatial-network, general-equilibrium model with a stylized application.” Regional Science and Urban Economics, 32(5), 651–671. Luo, W., and Wang, F. (2003). “Measures of spatial accessibility to health care in a GIS environment: synthesis and a case study in the Chicago region.” Environment and planning B: planning and design, SAGE Publications Sage UK: London, England, 30(6), 865–884. MacKenzie, C., Santos, J., and Barker, K. (2012). “Measuring changes in international production from a disruption: Case study of the Japanese earthquake and tsunami.” International Journal of Production …, 138(2), 293–302. Mackie, K., and Stojadinović, B. (2001). “Probabilistic seismic demand model for California highway bridges.” Journal of Bridge Engineering, American Society of Civil Engineers, 6(6), 468–481. Mangalathu, S. (2017). “Performance based grouping and fragility analysis of box-girder bridges in California.” Georgia Institute of Technology. Martens, K., Golub, A., and Robinson, G. (2012). “A justice-theoretic approach to the distribution 137 of transportation benefits: Implications for transportation planning practice in the United States.” Transportation Research Part A: Policy and Practice, Elsevier Ltd, 46(4), 684–695. Mattsson, L. G., and Jenelius, E. (2015). “Vulnerability and resilience of transport systems - A discussion of recent research.” Transportation Research Part A: Policy and Practice, 81, 16–34. McGuire, R. K. (1995). “Probabilistic seismic hazard analysis and design earthquakes: closing the loop.” Bulletin of the Seismological Society of America, The Seismological Society of America, 85(5), 1275–1284. Meerow, S., Newell, J. P., and Stults, M. (2016a). “Defining urban resilience: A review.” Landscape and Urban Planning. Meerow, S., Newell, J. P., and Stults, M. (2016b). “Defining urban resilience: A review.” Landscape and Urban Planning, Elsevier B.V., 147, 38–49. Mesa-Arango, R., Zhan, X., Ukkusuri, S. V, and Mitra, A. (2016). “Direct transportation economic impacts of highway networks disruptions using public data from the United States.” JOURNAL OF TRANSPORTATION SAFETY & SECURITY, TAYLOR & FRANCIS INC, 530 WALNUT STREET, STE 850, PHILADELPHIA, PA 19106 USA, 8(1), 36–55. Mitchell, R., and Popham, F. (2008). “Effect of exposure to natural environment on health inequalities: an observational population study.” The lancet, Elsevier, 372(9650), 1655– 1660. Morgenthal, G., Hallermann, N., Kersten, J., Taraben, J., Debus, P., Helmrich, M., and Rodehorst, V. (2019). “Framework for automated UAS-based structural condition assessment of bridges.” Automation in Construction, Elsevier, 97, 77–95. Murray-Tuite, P. M., and Mahmassani, H. S. (2004). “Methodology for Determining Vulnerable Links in a Transportation Network.” Transportation Research Record: Journal of the Transportation Research Board, 1882(1), 88–96. Murray, T. A., Matisziw, T. C., and Grubesic, T. H. (2008). “A Methodological Overview of Network Vulnerability Analysis.” Growth and Change, 39(4), 573–592. Nagae, T., Fujihara, T., and Asakura, Y. (2012). “Anti-seismic reinforcement strategy for an 138 urban road network.” Transportation Research Part A: Policy and Practice, Elsevier Ltd, 46(5), 813–827. Nagurney, A. (2007). “A Network Efficiency Measure for Congested Networks.” Europhysic, Letters, 71(2007), 1–5. Neutens, T. (2015). “Accessibility, equity and health care: Review and research directions for transport geographers.” Journal of Transport Geography, Elsevier Ltd, 43, 14–27. NIBS. (2002). A Guide to Using HAZUS for Mitigation. Reston, VA. Nicholson, A., and Du, Z.-P. (1997). “Degradable Transportation Equilibrium Alan Nicholson Systems : an Integrated Model.” Transportation Research Part B, 31(3), 209–223. NSF. (2020). “Convergence Accelerator Phase I (RAISE): Civil Infrastructure Systems Open Knowledge Network (CIS-OKN).” of Policy Studies (CoPS), V. U. (2019). “The TERM Model.” Omer, M., Mostashari, A., and Nilchiani, R. (2013). “Assessing resilience in a regional road- based transportation network.” International Journal of Industrial and Systems Engineering, Inderscience Enterprises, Switzerland, 13(4), 389–408. OpenStreetMap contributors. (2019). “Planet dump retrieved from https://planet.osm.org .” Oztanriseven, F., and Nachtmann, H. (2017). “Economic impact analysis of inland waterway disruption response.” ENGINEERING ECONOMIST, TAYLOR & FRANCIS INC, 530 WALNUT STREET, STE 850, PHILADELPHIA, PA 19106 USA, 62(1), 73–89. Padgett, J. E., Nielson, B. G., and DesRoches, R. (2008). “Selection of optimal intensity measures in probabilistic seismic demand models of highway bridge portfolios.” Earthquake Engineering \& Structural Dynamics, Wiley Online Library, 37(5), 711–725. Pant, R., Barker, K., Grant, F. H., and Landers, T. L. (2011). “Interdependent impacts of inoperability at multi-modal transportation container terminals.” Transportation Research Part E: Logistics and Transportation Review, 47(5), 722–737. Park, J., Cho, J., Gordon, P., Moore II, J. E., Richardson, H. W., and Yoon, S. (2011). “Adding a freight network to a national interstate input-output model: a TransNIEMO application for 139 California.” JOURNAL OF TRANSPORT GEOGRAPHY, ELSEVIER SCI LTD, THE BOULEVARD, LANGFORD LANE, KIDLINGTON, OXFORD OX5 1GB, OXON, ENGLAND, 19(6), 1410–1422. Park, J., Gordon, P., Ii, J. E. M., and Richardson, H. W. (2005). Simulating the State-by-State Effects of Terrorist Attacks on Three Major U . S . Ports : Applying NIEMO ( National Interstate Economic Model ). Park, J. Y. (2008). “The Economic Impacts of Dirty- Bomb Attacks on the Los Angeles and Long Beach Ports: Applying the Supply-driven NIEMO (National Interstate Economic Model).” Journal of Homeland Security and Emergency Management, 5(1), Article 21-Article 21. Park, J. Y., Gordon, P., Moore, J. S. E., and Richardson, H. W. (2008). “The state-by-state economic impacts of the 2002 shutdown of the Los Angeles-Long Beach ports.” Growth and Change, 39(4), 548–572. PEER. (2015). NGA-East: Median Ground-Motion Models for the Central and Eastern North America Region. Berkeley, CA. Peluso, N. L. (1995). “Whose woods are these? Counter-mapping forest territories in Kalimantan, Indonesia.” Antipode, Wiley Online Library, 27(4), 383–406. Petersen, M. D., Moschetti, M. P., Powers, P. M., Mueller, C. S., Haller, K. M., Frankel, A. D., Zeng, Y., Rezaeian, S., Harmsen, S. C., Boyd, O. S., Field, N., Chen, R., Rukstales, K. S., Luco, N., Wheeler, R. L., Williams, R. A., and Olsen, A. H. (2014). Documentation for the 2014 Update of the United States National Seismic Hazard Maps. U.S. Geological Survey Open-File Report 2014–1091, Reston, VA. Pimm, S. L. (1984). “The complexity and stability of ecosystems.” Nature, Nature Publishing Group, 307(5949), 321–326. Pitilakis, K., Argyroudis, S., Kakderi, K., and Selva, J. (2016). “Systemic Vulnerability and Risk Assessment of Transportation Systems under Natural Hazards Towards More Resilient and Robust Infrastructures.” Transportation Research Procedia, Elsevier B.V., 14, 1335–1344. Postance, B., Hillier, J., Dijkstra, T., and Dixon, N. (2017). “Extending natural hazard impacts: an assessment of landslide disruptions on a national road transportation network.” 140 ENVIRONMENTAL RESEARCH LETTERS, IOP PUBLISHING LTD, TEMPLE CIRCUS, TEMPLE WAY, BRISTOL BS1 6BE, ENGLAND, 12(1). Pruitt, L., Mumuni, T., Raikhel, E., Ademola, A., Ogundiran, T., Adenipekun, A., Morhason-Bello, I., Ojengbede, O. A., and Olopade, O. I. (2015). “Social barriers to diagnosis and treatment of breast cancer in patients presenting at a teaching hospital in Ibadan, Nigeria.” Global public health, Taylor & Francis, 10(3), 331–344. Ramanathan, K., Padgett, J. E., and DesRoches, R. (2015). “Temporal evolution of seismic fragility curves for concrete box-girder bridges in California.” Engineering Structures, Elsevier, 97, 29–46. Reggiani, A. (2013). “Network resilience for transport security: Some methodological considerations.” Transport Policy, Elsevier, 28, 63–68. Ribeiro, A., Antunes, A. P., and Páez, A. (2010). “Road accessibility and cohesion in lagging regions: empirical evidence from Portugal based on spatial econometric models.” Journal of Transport Geography, Elsevier, 18(1), 125–132. Robjent, L., County Perry Clark, C., County Michael Marti, C., Johnson, A., and Services, P. (2020). Asset Management Guide for Local Agencies. Rose, A. (1995). “Input-output economics and computable general equilibrium models.” Structural Change and Economic Dynamics, 6(3), 295–304. Rose, A. (2004). “Defining and Measuring Economic Resilience to Earthquakes.” Disaster Prevention and Management, 13, 307–314. Rose, A. (2017). “Benefit-Cost Analysis of Economic Resilience Actions.” Oxford Research Encyclopedia of Natural Hazard Science. Rose, A., and Dormady, N. (2018). “Advances in Analyzing and Measuring Dynamic Economic Resilience.” SSRN Electronic Journal, 2, 1–8. Rose, A., and Liao, S.-Y. (2005). “Modeling Regional Economic Resilience to Disasters: A Computable General Equilibrium Analysis of Water Service Disruptions*.” Journal of Regional Science, 45(1), 75–112. 141 Rose, A., Sue Wing, I., Wei, D., and Wein, A. (2016). “Economic Impacts of a California Tsunami.” Natural Hazards Review, 17(2013), 4016002. Rose, A., and Wei, D. (2013). “Estimating the Economic Consequences of a Port Shutdown: the Special Role of Resilience.” Economic Systems Research, 25(2), 212–232. Rose, A., Wei, D. A. N., and Prager, F. (2012). “Distributional impacts of greenhouse gas emissions trading: alternative allocation and recycling strategies in California.” Contemporary Economic Policy, Wiley Online Library, 30(4), 603–617. Rose, A. Z., Stevens, B., and Davis, G. (1988). Natural resource policy and income distribution. Johns Hopkins Univ Pr. Rupi, F., Bernardi, S., Rossi, G., and Danesi, A. (2015). “The Evaluation of Road Network Vulnerability in Mountainous Areas: A Case Study.” Networks and Spatial Economics, Networks and Spatial Economics, Springer, 15(2), 397–411. Santos, J. R., and Haimes, Y. Y. (2004). “Modeling the demand reduction input-output (I-O) inoperability due to terrorism of interconnected infrastructures.” Risk Analysis, 24(6), 1437– 1451. SCAG. (2016). “SCAG REGIONAL TRAVEL DEMAND MODEL AND 2012 MODEL VALIDATION.” Schumacher, K., and Sands, R. D. (2007). “Where are the industrial technologies in energy- economy models? An innovative CGE approach for steel production in Germany.” Energy Economics, 29(4), 799–825. Scott, D. M., Novak, D. C., Aultman-Hall, L., and Guo, F. (2006). “Network Robustness Index: A new method for identifying critical links and evaluating the performance of transportation networks.” Journal of Transport Geography, 14(3), 215–227. Sohn, J., Kim, T. J., Hewings, G. J. D., Lee, J. S., and Jang, S. G. (2003). “Retrofit priority of transport network links under an earthquake.” Journal of Urban Planning and Development, 129(4), 195–210. Southern California Association of Governments. (2019). “Regional Transportation Plan.” 142 Stewart, J. P., Boore, D. M., Seyhan, E., and Atkinson, G. M. (2016). “NGA-West2 equations for predicting vertical-component PGA, PGV, and 5{\%}-damped PSA from shallow crustal earthquakes.” Earthquake Spectra, 32(2), 1005–1031. Sullivan, J. L., Novak, D. C., Aultman-Hall, L., and Scott, D. M. (2010). “Identifying critical road segments and measuring system-wide robustness in transportation networks with isolating links: A link-based capacity-reduction approach.” Transportation Research Part A: Policy and Practice, Elsevier Ltd, 44(5), 323–336. Tan, X. M., Zhang, Y., and Lam, J. S. L. (2015). “Economic Impact of Port Disruptions on Industry Clusters: a Case Study of Shenzhen.” The 3rd International Conference on Transportation Information and Safety (ICTIS 2015), Z. Yan, XP and Hu, ZZ and Zhong, M and Wu, CZ and Yang, ed., IEEE, 617–622. Tatano, H., and Tsuchiya, S. (2008). “A framework for economic loss estimation due to seismic transportation network disruption: a spatial computable general equilibrium approach.” Nature Hazards, SPRINGER, 233 SPRING STREET, NEW YORK, NY 10013 USA, 44(2), 253–265. Taylor, M. A. P., Sekhar, S. V. C., and D’Este, G. M. (2006). “Application of accessibility based methods for vulnerability analysis of strategic road networks.” Networks and Spatial Economics, 6(3–4), 267–291. Thekdi, S. A., and Santos, J. R. (2016). “Supply Chain Vulnerability Analysis Using Scenario- Based Input-Output Modeling: Application to Port Operations.” RISK ANALYSIS, WILEY, 111 RIVER ST, HOBOKEN 07030-5774, NJ USA, 36(5), 1025–1039. Thissen, M. (2004). “The indirect economic effects of a terrorist attack on transport infrastructure: a proposal for a SAGE.” Disaster Prevention and Management, 13(4), 315– 322. Thompson, E. M., Wald, D. J., and Worden, C. B. (2014). “A VS30Map for California with geologic and topographic constraints.” Bulletin of the Seismological Society of America, 104(5), 2313–2321. Tierney, K. J. (1995). “Impacts of Recent U.S. Disasters on Businesses : the 1993 Midwest Floods and the 1994 Northridge Earthquake.” National Center for Earthquake Engineering 143 Research Conference on the Economic Impacts of Catastrophic Earthquakes, 189–222. Tirasirichai, C., and Enke, D. (2007). “Case study: applying a regional CGE model for estimation of indirect economic losses due to damaged highway bridges.” Engineering Economist, American Society for Engineering Education and Institute of Industrial Engineers, USA, 52(4), 367–401. Tokui, J., Kawasaki, K., and Miyagawa, T. (2017). “The economic impact of supply chain disruptions from the Great East-Japan earthquake.” Japan and the World Economy, 41, 59– 70. Tsuchiya, S., Tatano, H., and Okada, N. (2007). “Economic loss assessment due to railroad and highway disruptions.” Economic Systems Research, 19(2), 147–162. Ueda, T., Koike, A., and Iwakami, K. (2001). “Economic Damage Assessment of Catastrophes in High Speed Rail network.” Proceedings of the 1st Workshop on “Comparative Study of Urban Earthquake Disaster Management,” 13–19. UNISDR. (2018). “Disaster Statistics.” <https://www.unisdr.org/we/inform/disaster-statistics>. United States Federal Emergency Management Agency. (2010). “Multi-hazard Loss Estimation Methodology: Earthquake Model, Hazus-MH MR5: User Manual.” USDOT, FHWA, O. of A. M. (2007). “Asset Management Overview.” 52. USDOT. (2018). Bridge Preservation Guide Maintaining a Resilient Infrastructure to Preserve Mobility. USDOT FHWA. (1995). Recording and Coding Guide for the Structure Inventory and Appraisal. USGS. (2015). “Shuttle Radar Topography Mission (SRTM) 1 Arc-Second Global.” Vadali, S., Chandra, S., Shelton, J., Valdez, A., and Medina, M. (2015). “Economic costs of critical infrastructure failure in bi-national regions and implications for resilience building: Evidence from El Paso-Ciudad Juarez.” Research in Transportation Business and Management, Elsevier Ltd, 16, 15–31. Vamvatsikos, D., and Cornell, C. A. (2002). “Incremental dynamic analysis.” Earthquake Engineering {\&} Structural Dynamics, John Wiley {\&} Sons, Ltd, 31(3), 491–514. 144 Wang, Q., and Taylor, J. E. (2016). “Process Map for Urban-Human Mobility and Civil Infrastructure Data Collection Using Geosocial Networking Platforms.” Journal of Computing in Civil Engineering, 30, 4015004. WashDOT. (2001). Rapid Design of Temporary Support Systems for Bridges Damaged by Earthquakes. Wei, D., Koc, E., Rose, A., Chen, Z., and Soibelman, L. (2020a). Socioeconomic Dimensions of Resilience to Seaport and Highway Transportation Network Disruptions. Wei, D., Koc, E., Rose, A., Chen, Z., Soibelman, L., and Region, P. S. (2020b). Socioeconomic Dimensions of Resilience to Seaport and Highway Transportation. California. Dept. of Transportation. Division of Research and Innovation. Wei, F., Koc, E., Soibelman, L., and Li, N. (2018). “Disturbances to Urban Mobility and Comprehensive Estimation of Economic Losses.” Polytechnica, Polytechnica, 1(1–2), 48– 60. Wei, H., Dong, M., and Sun, S. (2010). “Inoperability input-output modeling (IIM) of disruptions to supply chain networks.” Systems Engineering, 13(4), 324–339. Wells, D. L., and Coppersmith, K. J. (1994). “New Empirical Relationships among Magnitude, Rupture Length, Rupture Width, Rupture Area, and Surface Displacement.” Bulletin of the Seismological Society of America, 84(4), 974–1002. Weykamp, P., Kimball, T., H., G., Johnson, B. V., Ramsey, K., D’Andrea, A., and Becker, S. (2009). “Best Practices in Bridge Management Decision-Making Overview of Scan Results Overview of Scan Results.” Lawrenceville, NJ. Wittwer, G. (2012a). Economic Modeling of Water. Wittwer, G. (2012b). Economic modeling of water: the Australian CGE experience. Springer Science \& Business Media. Xie, F., and Levinson, D. (2011). “Evaluating the effects of the I-35W bridge collapse on road- users in the twin cities metropolitan region.” Transportation Planning & Technology, 34(7), 691–703. 145 Xie, W., Li, N., Li, C., Wu, J. dong, Hu, A., and Hao, X. (2014). “Quantifying cascading effects triggered by disrupted transportation due to the Great 2008 Chinese Ice Storm: Implications for disaster risk management.” Natural Hazards, 70(1), 337–352. Yu, K. D. S., Tan, R. R., and Santos, J. R. (2013). “Impact Estimation of Flooding in Manila: An Inoperability Input-Output Approach.” 2013 IEEE Systems and Information Engineering Design Symposium (SIEDS), IEEE, 345 E 47TH ST, NEW YORK, NY 10017 USA, 47–51. Zhang, P., and Peeta, S. (2011). “A generalized modeling framework to analyze interdependencies among infrastructure systems.” Transportation Research Part B: Methodological, Elsevier Ltd, 45(3), 553–579. Zhang, P., and Peeta, S. (2014). “Dynamic and disequilibrium analysis of interdependent infrastructure systems.” Transportation Research Part B: Methodological, Elsevier Ltd, 67, 357–381. Zhang, Y., and Lam, J. S. L. (2015). “Estimating the economic losses of port disruption due to extreme wind events.” OCEAN & COASTAL MANAGEMENT, ELSEVIER SCI LTD, THE BOULEVARD, LANGFORD LANE, KIDLINGTON, OXFORD OX5 1GB, OXON, ENGLAND, 116, 300–310. Zhang, Y., and Lam, J. S. L. (2016). “Estimating economic losses of industry clusters due to port disruptions.” Transportation Research Part A, PERGAMON-ELSEVIER SCIENCE LTD, THE BOULEVARD, LANGFORD LANE, KIDLINGTON, OXFORD OX5 1GB, ENGLAND, 91, 17– 33. Zhou, Y., Banerjee, S., and Shinozuka, M. (2010). “Socio-economic effect of seismic retrofit of bridges for highway transportation networks: A pilot study.” Structure and Infrastructure Engineering, 6(1–2), 145–157. 146 Appendix A. Construction of Income Distribution Matrix A1. Employee Compensation To construct the income distribution matrix for wages and salaries, we first collected data from the BLS Occupational Employment Statistics (OES) (BLS, 2019a). The two main matrices we use are the Occupation-Industry Employment matrix and the Occupation- Industry Wage matrix. The industries are disaggregated at 4-digit NAICS level, and the occupation categories follow the 6-digit Standard Occupational Classification. For each occupation type of a given industry, BLS OES data report not only the annual average (mean) wages, but also wage rate in percentiles (10, 25, 50, 75, 90). One limitation of this data set was that the minimum and maximum wage percentiles are 10 and 90, respectively, and hence it does not readily provide information on the wage rate for the highest and lowest earners. In order to deal with this limitation, we estimated annual wage rates for an extended set of percentiles (1, 5, 20, 40, 60, 80, 95, 99) using linear interpolations following the methodology developed by Rose et al. (2012) and Prager (2013). After we calculated the annual employee compensation by sector and occupation for each percentile (1, 5, 20, 40, 60, 80, 95, or 99), we multiply it by the number of employees in each percentile interval of this occupation in this sector to obtain the total employee compensations by percentile. Next, the total employee compensations by percentile and sector are allocated to the relevant household income brackets. The OES sectors are also mapped to the TERM CGE model sectors (see the TERM sectoring scheme in Appendix B). Finally, we used the estimate of Total Employee Compensations in California in 2018, which was $1.346 trillion, as the control total to re-balance the entire Employee Compensation matrix we constructed. A2. Proprietors’ Income The distribution of proprietors’ income across sectors and income brackets is calculated based on IMPLAN data. In addition, IMPLAN also provides data on the amount of 147 proprietors’ income generated in each sector. We apply the distribution percentages across the nine income brackets to the total proprietors’ income for each sector to obtain the distribution of proprietors’ income across income brackets for each sector. The underlying assumption is that the proportional distribution of proprietors’ income among the income brackets is the same across all sectors. Finally, we used the BEA estimate of total proprietors’ income in California in 2018, which was $249.7 billion, as the control total to re-balance the entire proprietors’ income matrix we constructed. A3. Capital Income IMPLAN also provides data on the distribution of the total Dividend Payments and Other Property Income (which mainly includes interest payments and rent income) across income brackets. In addition, IMPLAN also provides data on the amount of Other Property Income by sector. We first calculated the percentage distribution of Other Property Income across sectors, and then apply it to the total amounts of Dividend Payments and Other Property Income in each income bracket to obtain the distribution across sectors for each income bracket. The underlying assumption is that the proportional distribution of Dividend Payments and Other Property Income among the sectors is the same across all income brackets. Finally, we used the BEA estimate of total Dividends, Interest, and Rental Income in California in 2018, which was $538.3 billion, as the control total to re- balance the capital income matrix we constructed. A4. Personal Transfer Receipts The final component of the personal income accounts is the personal transfer receipts (including social security benefits, medical benefits, veteran’s benefits, and unemployment insurance benefits). IMPLAN provides data on the distribution of federal, state and local government transfer payments to each household income bracket. Similar as for the other components of the personal income accounts, we used the BEA estimate of total Personal Current Transfer Receipts in California in 2018, which was $341.2 billion, as the control total to re-balance the transfer income matrix we constructed. 148 Appendix Table A1. Multi-sector Income Distribution Matrix for California, 2018 (millions of 2018$) Sector <10k 10-15k 15-25k 25-35k 35-50k 50-75k 75-100k 100- 150k 150k+ Total 01. Crops 17 158 3,284 4,056 1,648 1,497 1,991 1,606 6,194 20,451 02. Poultry & Eggs 0 3 38 48 23 24 36 29 114 315 03. Livestock 4 35 188 237 191 250 397 319 1,296 2,916 04. Other Livestock 1 3 14 17 17 22 35 28 114 251 05. Forestry, Fishing, & Hunting 2 20 26 38 131 198 243 186 756 1,601 06. Oil & Gas 9 45 54 75 207 353 550 441 1,581 3,315 07. Coal 0 0 0 1 2 5 3 2 4 18 08. Other Mining 4 10 15 62 213 465 294 216 311 1,589 09. Biomass electricity generation 1 6 7 10 25 41 67 60 208 426 10-11. Coal-fired and Gas-fired electricity generation 18 50 53 71 207 361 574 680 1,364 3,378 12. Hydroelectric generation 1 5 5 7 21 41 68 87 150 386 13. Nuclear electricity generation 5 14 15 21 62 116 186 242 399 1,061 14. Renewable electricity generation 10 27 29 38 105 161 249 236 703 1,558 15. Electricity distribution 5 15 16 21 61 105 165 192 399 979 16. Natural gas distribution 8 23 25 45 170 516 904 1,634 1,080 4,404 17. Water and sewage services 3 14 16 22 62 118 197 243 467 1,142 18. Residential Construction 64 313 493 1,250 3,343 6,603 6,417 5,301 11,173 34,957 19. Highway Construction 6 30 71 271 741 1,580 1,276 1,081 1,225 6,281 20. Other Non-Residential Construction 53 252 589 2,213 6,037 12,858 10,439 8,837 10,304 51,583 21. Highway Maintenance 6 28 61 214 583 1,229 1,026 866 1,140 5,152 22. Other Maintenance 24 116 254 903 2,456 5,190 4,308 3,638 4,680 21,569 23. Food Processing 33 99 886 2,243 2,667 2,638 1,647 1,383 3,239 14,835 24. Beverage & Tobacco Product Manufacturing 13 49 225 662 1,100 1,341 1,023 769 1,811 6,992 25. Textile & Textile Product Manufacturing 1 4 93 180 180 145 84 83 150 920 26. Apparel 2 10 373 554 294 311 283 313 509 2,649 27. Leather & Allied Products 0 0 12 45 24 9 4 5 11 110 28. Wood Product Manufacturing 3 8 106 346 410 331 164 136 246 1,749 29. Paper Mills 5 15 76 243 429 450 245 238 522 2,222 30. Printing & Related Support Activities 4 15 164 439 662 751 335 322 592 3,284 149 Sector <10k 10-15k 15-25k 25-35k 35-50k 50-75k 75-100k 100- 150k 150k+ Total 31. Petroleum Refineries 61 169 178 238 651 1,003 1,602 1,270 4,216 9,388 32. Other Petroleum & Coal Products 3 8 9 12 33 58 98 83 217 521 33. Chemicals 175 487 604 1,049 2,553 3,979 4,831 4,739 14,727 33,143 34. Rubber & Plastics 7 21 174 549 703 678 415 428 839 3,815 35. Non-Metallics 6 16 117 604 1,224 1,453 662 551 614 5,246 36. Primary Metal Manufacturing 2 2 35 177 279 260 120 100 87 1,063 37. Fabricated Metal Product 19 58 310 1,372 2,419 3,204 1,444 1,292 2,416 12,535 38. Agriculture Machinery 2 6 12 37 74 118 100 110 234 692 39. Industrial Machinery 1 3 37 185 343 568 373 505 643 2,659 40. Commercial Machinery 4 12 28 100 198 316 251 290 554 1,754 41. Ventilation, Heating & Air- Conditioning 1 4 9 31 62 100 81 92 181 563 42. Metalworking Machinery 1 5 15 60 116 189 142 173 293 995 43. Engines & Turbines 2 5 15 55 107 171 131 156 278 920 44. Other General Purpose Machinery Manufacturing 4 12 30 108 213 341 268 314 580 1,869 45. Computers 97 267 471 1,217 2,759 4,942 5,905 10,644 20,764 47,064 46. Computer Storage Devices 7 20 29 62 147 249 310 487 1,073 2,385 47. Computer Terminals & Other Peripheral Equipment 4 11 23 69 153 283 330 643 1,171 2,686 48. Communications Equipment 6 17 24 46 98 124 154 140 470 1,079 49. Miscellaneous Electronic Equipment 27 73 103 200 423 536 666 606 2,028 4,661 50. Semiconductors & Related Devices 32 87 116 212 464 593 759 664 2,339 5,266 51. Electronic Instruments 8 21 28 53 115 146 184 164 565 1,284 52. Household Equipment, Appliances, and Component Manufacturing 4 12 20 42 84 105 123 121 366 878 53. Motor Vehicle and Parts Manufacturing 12 37 59 186 362 616 663 796 1,625 4,354 54. Aerospace Product & Parts Manufacturing 18 50 131 609 1,087 1,955 1,878 2,612 3,978 12,319 55. Railroad Rolling Stock Manufacturing 0 0 1 6 10 18 16 24 32 106 56. Ship & Boat Building 0 1 7 42 71 133 120 183 228 786 57. Other Transportation Equipment Manufacturing 1 2 4 16 29 51 50 67 110 327 58. Furniture & Related Product Manufacturing 2 9 150 472 598 558 294 240 470 2,793 59. Miscellaneous Manufacturing 21 59 206 745 1,279 1,839 1,383 1,820 3,275 10,626 60. Wholesale Trade 168 537 1,748 5,980 10,434 13,799 11,650 11,277 22,705 78,298 61. Air Transport 23 67 98 314 829 2,083 813 784 4,286 9,299 150 Sector <10k 10-15k 15-25k 25-35k 35-50k 50-75k 75-100k 100- 150k 150k+ Total 62. Rail Transport 2 6 7 9 61 319 230 166 175 976 63. Water Transport 5 16 37 41 115 205 214 164 448 1,247 64. Truck Transport 8 115 386 2,608 5,022 8,940 2,091 1,838 5,006 26,014 65. Transit and Ground Passenger Transport 7 36 160 736 1,212 889 452 343 1,259 5,093 66. Pipelines 0 2 2 4 18 41 55 56 64 242 67. Other Transportation 21 88 361 1,373 2,277 3,818 2,519 1,502 3,403 15,363 68. Warehousing 8 26 454 1,817 2,880 2,892 927 511 823 10,339 69. Retail Trade 109 543 14,099 23,812 21,508 14,635 9,874 8,232 20,770 113,580 70. Publishing Industries 89 249 310 543 1,434 2,787 3,740 5,547 12,041 26,741 71. Motion Picture & Sound Recording Industry 193 533 930 1,390 2,957 4,521 5,995 5,889 16,264 38,672 72. Broadcasting 58 369 494 767 2,001 3,113 4,806 4,202 14,064 29,875 73. Telecommunications 146 406 478 805 2,334 5,356 6,795 5,959 11,769 34,047 74. Information Services 4 10 29 97 400 1,141 1,937 4,477 7,487 15,581 75. Data Processing Services 1 5 14 60 170 584 940 2,190 3,658 7,623 76. Finance & Banking 232 672 1,152 3,683 8,878 15,548 15,050 17,851 41,160 104,226 77. Real Estate 807 2,622 3,951 7,169 15,264 19,639 25,482 20,864 72,803 168,600 78. Rental & Leasing Services 30 127 361 709 1,381 1,705 1,607 1,309 4,214 11,443 79. Lessors of Nonfinancial Intangible Assets 67 189 199 258 700 967 1,434 1,060 4,668 9,541 80. Professional, Scientific, Technical, Administrative, & Support Services 286 1,331 6,431 18,112 28,506 44,173 43,096 66,401 114,559 322,896 81. Waste Management Services 8 23 117 359 822 1,105 896 358 741 4,430 82. Education Services 13 74 1,302 8,120 17,375 27,838 26,785 31,787 14,526 127,820 83. Health Care & Social Assistance 86 445 11,087 19,316 30,097 34,633 25,394 43,896 50,932 215,886 84. Arts, Entertainment & Recreation 51 245 2,247 3,488 3,576 4,256 3,990 3,121 9,099 30,074 85. Accommodations 21 70 1,450 3,291 3,051 2,130 1,168 941 1,973 14,095 86. Eating & Drinking Places 71 276 14,388 20,236 8,741 5,122 3,578 2,267 8,540 63,220 87. Other Services -1 317 2,809 5,868 6,821 8,652 7,475 6,237 15,669 53,847 88. Owner-Occupied Dwellings 461 1,265 1,317 1,674 4,570 6,220 9,175 6,561 29,850 61,094 89. Government Enterprises 35 95 142 365 1,202 2,820 3,175 3,939 3,259 15,033 90. State & Local Government 9,229 24,878 15,481 13,834 23,560 34,338 34,865 36,254 29,481 221,919 91. Federal Government 15,277 41,406 25,560 20,684 31,027 34,537 31,426 20,617 50,210 270,743 Total 28,343 79,882 117,735 190,410 277,948 371,097 344,198 375,087 691,027 2,475,727 Appendix Table A2. Total Personal Income Distribution Coefficient Matrix, 2018 151 Sector <10k 10-15k 15-25k 25-35k 35-50k 50-75k 75-100k 100- 150k 150k+ Total 01. Crops 0.001 0.008 0.161 0.198 0.081 0.073 0.097 0.079 0.303 1.000 02. Poultry & Eggs 0.001 0.009 0.122 0.151 0.073 0.077 0.113 0.092 0.361 1.000 03. Livestock 0.001 0.012 0.065 0.081 0.066 0.086 0.136 0.109 0.444 1.000 04. Other Livestock 0.002 0.013 0.054 0.069 0.067 0.089 0.140 0.110 0.456 1.000 05. Forestry, Fishing, & Hunting 0.002 0.013 0.016 0.023 0.082 0.124 0.152 0.116 0.472 1.000 06. Oil & Gas 0.003 0.013 0.016 0.023 0.063 0.107 0.166 0.133 0.477 1.000 07. Coal 0.003 0.008 0.011 0.037 0.123 0.260 0.179 0.133 0.247 1.000 08. Other Mining 0.002 0.006 0.009 0.039 0.134 0.292 0.185 0.136 0.195 1.000 09. Biomass electricity generation 0.003 0.014 0.017 0.023 0.060 0.095 0.158 0.142 0.488 1.000 10-11. Coal-fired and Gas-fired electricity generation 0.005 0.015 0.016 0.021 0.061 0.107 0.170 0.201 0.404 1.000 12. Hydroelectric generation 0.004 0.012 0.014 0.019 0.056 0.106 0.175 0.225 0.390 1.000 13. Nuclear electricity generation 0.005 0.014 0.014 0.020 0.058 0.109 0.175 0.228 0.376 1.000 14. Renewable electricity generation 0.006 0.018 0.019 0.024 0.068 0.103 0.160 0.151 0.452 1.000 15. Electricity distribution 0.006 0.015 0.016 0.022 0.063 0.107 0.169 0.196 0.408 1.000 16. Natural gas distribution 0.002 0.005 0.006 0.010 0.039 0.117 0.205 0.371 0.245 1.000 17. Water and sewage services 0.003 0.012 0.014 0.019 0.054 0.103 0.173 0.213 0.409 1.000 18. Residential Construction 0.002 0.009 0.014 0.036 0.096 0.189 0.184 0.152 0.320 1.000 19. Highway Construction 0.001 0.005 0.011 0.043 0.118 0.252 0.203 0.172 0.195 1.000 20. Other Non-Residential Construction 0.001 0.005 0.011 0.043 0.117 0.249 0.202 0.171 0.200 1.000 21. Highway Maintenance 0.001 0.006 0.012 0.042 0.113 0.238 0.199 0.168 0.221 1.000 22. Other Maintenance 0.001 0.005 0.012 0.042 0.114 0.241 0.200 0.169 0.217 1.000 23. Food Processing 0.002 0.007 0.060 0.151 0.180 0.178 0.111 0.093 0.218 1.000 24. Beverage & Tobacco Product Manufacturing 0.002 0.007 0.032 0.095 0.157 0.192 0.146 0.110 0.259 1.000 25. Textile & Textile Product Manufacturing 0.001 0.004 0.101 0.196 0.196 0.158 0.091 0.090 0.163 1.000 26. Apparel 0.001 0.004 0.141 0.209 0.111 0.118 0.107 0.118 0.192 1.000 27. Leather & Allied Products 0.000 0.000 0.108 0.413 0.218 0.082 0.034 0.045 0.099 1.000 28. Wood Product Manufacturing 0.002 0.004 0.060 0.198 0.234 0.189 0.094 0.078 0.141 1.000 29. Paper Mills 0.002 0.007 0.034 0.109 0.193 0.203 0.110 0.107 0.235 1.000 30. Printing & Related Support Activities 0.001 0.004 0.050 0.134 0.202 0.229 0.102 0.098 0.180 1.000 31. Petroleum Refineries 0.007 0.018 0.019 0.025 0.069 0.107 0.171 0.135 0.449 1.000 32. Other Petroleum & Coal Products 0.005 0.015 0.017 0.023 0.064 0.111 0.188 0.160 0.416 1.000 33. Chemicals 0.005 0.015 0.018 0.032 0.077 0.120 0.146 0.143 0.444 1.000 34. Rubber & Plastics 0.002 0.006 0.046 0.144 0.184 0.178 0.109 0.112 0.220 1.000 35. Non-Metallics 0.001 0.003 0.022 0.115 0.233 0.277 0.126 0.105 0.117 1.000 36. Primary Metal Manufacturing 0.002 0.002 0.033 0.166 0.263 0.244 0.113 0.094 0.082 1.000 152 37. Fabricated Metal Product 0.001 0.005 0.025 0.109 0.193 0.256 0.115 0.103 0.193 1.000 38. Agriculture Machinery 0.003 0.008 0.017 0.053 0.108 0.170 0.144 0.159 0.338 1.000 39. Industrial Machinery 0.000 0.001 0.014 0.070 0.129 0.214 0.140 0.190 0.242 1.000 40. Commercial Machinery 0.002 0.007 0.016 0.057 0.113 0.180 0.143 0.166 0.316 1.000 41. Ventilation, Heating & Air- Conditioning 0.002 0.007 0.016 0.056 0.111 0.177 0.144 0.164 0.322 1.000 42. Metalworking Machinery 0.001 0.005 0.015 0.061 0.117 0.190 0.143 0.174 0.295 1.000 43. Engines & Turbines 0.002 0.006 0.016 0.059 0.116 0.186 0.143 0.170 0.302 1.000 44. Other General Purpose Machinery Manufacturing 0.002 0.006 0.016 0.058 0.114 0.183 0.143 0.168 0.310 1.000 45. Computers 0.002 0.006 0.010 0.026 0.059 0.105 0.125 0.226 0.441 1.000 46. Computer Storage Devices 0.003 0.008 0.012 0.026 0.062 0.104 0.130 0.204 0.450 1.000 47. Computer Terminals & Other Peripheral Equipment 0.001 0.004 0.009 0.026 0.057 0.105 0.123 0.239 0.436 1.000 48. Communications Equipment 0.006 0.016 0.022 0.043 0.091 0.115 0.143 0.130 0.435 1.000 49. Miscellaneous Electronic Equipment 0.006 0.016 0.022 0.043 0.091 0.115 0.143 0.130 0.435 1.000 50. Semiconductors & Related Devices 0.006 0.017 0.022 0.040 0.088 0.113 0.144 0.126 0.444 1.000 51. Electronic Instruments 0.006 0.016 0.022 0.041 0.089 0.114 0.144 0.128 0.440 1.000 52. Household Equipment, Appliances, and Component Manufacturing 0.005 0.014 0.022 0.048 0.096 0.119 0.140 0.138 0.417 1.000 53. Motor Vehicle and Parts Manufacturing 0.003 0.008 0.014 0.043 0.083 0.141 0.152 0.183 0.373 1.000 54. Aerospace Product & Parts Manufacturing 0.001 0.004 0.011 0.049 0.088 0.159 0.152 0.212 0.323 1.000 55. Railroad Rolling Stock Manufacturing 0.001 0.002 0.009 0.052 0.090 0.165 0.153 0.224 0.305 1.000 56. Ship & Boat Building 0.000 0.001 0.008 0.054 0.091 0.170 0.153 0.232 0.291 1.000 57. Other Transportation Equipment Manufacturing 0.002 0.005 0.011 0.048 0.087 0.155 0.152 0.205 0.335 1.000 58. Furniture & Related Product Manufacturing 0.001 0.003 0.054 0.169 0.214 0.200 0.105 0.086 0.168 1.000 59. Miscellaneous Manufacturing 0.002 0.006 0.019 0.070 0.120 0.173 0.130 0.171 0.308 1.000 60. Wholesale Trade 0.002 0.007 0.022 0.076 0.133 0.176 0.149 0.144 0.290 1.000 61. Air Transport 0.002 0.007 0.011 0.034 0.089 0.224 0.087 0.084 0.461 1.000 62. Rail Transport 0.002 0.006 0.007 0.009 0.062 0.327 0.236 0.171 0.179 1.000 63. Water Transport 0.004 0.013 0.030 0.033 0.093 0.164 0.172 0.132 0.360 1.000 64. Truck Transport 0.000 0.004 0.015 0.100 0.193 0.344 0.080 0.071 0.192 1.000 65. Transit and Ground Passenger Transport 0.001 0.007 0.031 0.144 0.238 0.175 0.089 0.067 0.247 1.000 66. Pipelines 0.001 0.006 0.010 0.016 0.076 0.170 0.228 0.229 0.264 1.000 67. Other Transportation 0.001 0.006 0.023 0.089 0.148 0.249 0.164 0.098 0.222 1.000 153 68. Warehousing 0.001 0.002 0.044 0.176 0.279 0.280 0.090 0.049 0.080 1.000 69. Retail Trade 0.001 0.005 0.124 0.210 0.189 0.129 0.087 0.072 0.183 1.000 70. Publishing Industries 0.003 0.009 0.012 0.020 0.054 0.104 0.140 0.207 0.450 1.000 71. Motion Picture & Sound Recording Industry 0.005 0.014 0.024 0.036 0.076 0.117 0.155 0.152 0.421 1.000 72. Broadcasting 0.002 0.012 0.017 0.026 0.067 0.104 0.161 0.141 0.471 1.000 73. Telecommunications 0.004 0.012 0.014 0.024 0.069 0.157 0.200 0.175 0.346 1.000 74. Information Services 0.000 0.001 0.002 0.006 0.026 0.073 0.124 0.287 0.481 1.000 75. Data Processing Services 0.000 0.001 0.002 0.008 0.022 0.077 0.123 0.287 0.480 1.000 76. Finance & Banking 0.002 0.006 0.011 0.035 0.085 0.149 0.144 0.171 0.395 1.000 77. Real Estate 0.005 0.016 0.023 0.043 0.091 0.116 0.151 0.124 0.432 1.000 78. Rental & Leasing Services 0.003 0.011 0.032 0.062 0.121 0.149 0.140 0.114 0.368 1.000 79. Lessors of Nonfinancial Intangible Assets 0.007 0.020 0.021 0.027 0.073 0.101 0.150 0.111 0.489 1.000 80. Professional, Scientific, Technical, Administrative, & Support Services 0.001 0.004 0.020 0.056 0.088 0.137 0.133 0.206 0.355 1.000 81. Waste Management Services 0.002 0.005 0.026 0.081 0.186 0.249 0.202 0.081 0.167 1.000 82. Education Services 0.000 0.001 0.010 0.064 0.136 0.218 0.210 0.249 0.114 1.000 83. Health Care & Social Assistance 0.000 0.002 0.051 0.089 0.139 0.160 0.118 0.203 0.236 1.000 84. Arts, Entertainment & Recreation 0.002 0.008 0.075 0.116 0.119 0.142 0.133 0.104 0.303 1.000 85. Accommodations 0.001 0.005 0.103 0.234 0.216 0.151 0.083 0.067 0.140 1.000 86. Eating & Drinking Places 0.001 0.004 0.228 0.320 0.138 0.081 0.057 0.036 0.135 1.000 87. Other Services 0.000 0.006 0.052 0.109 0.127 0.161 0.139 0.116 0.291 1.000 88. Owner-Occupied Dwellings 0.008 0.021 0.022 0.027 0.075 0.102 0.150 0.107 0.489 1.000 89. Government Enterprises 0.002 0.006 0.009 0.024 0.080 0.188 0.211 0.262 0.217 1.000 90. State & Local Government 0.042 0.112 0.070 0.062 0.106 0.155 0.157 0.163 0.133 1.000 91. Federal Government 0.056 0.153 0.094 0.076 0.115 0.128 0.116 0.076 0.185 1.000 Total 0.011 0.032 0.048 0.077 0.112 0.150 0.139 0.152 0.279 1.000 154 Appendix B. Method of Modeling Resilience Tactics in TERM Appendix Table B1. Modeling Tactics for Economic Resilience in the TERM Model Simulation Method Description Adjustment Level or Range for Base Case Resilience Conservation Adaptive resilience is captured by adjusting import and export shocks in different regions Utilize less of disrupted imported goods in production processes. Adjust import and export shocks by 2% in all regions Inherent Input Substitution N/A Inherent input substitution is captured by the CGE model automatically. N/A Import Substitution N/A Inherent import substitution is captured by the CGE model automatically by the Armington elasticity of substitution. N/A Ship Rerouting Adjust import and export shocks in different regions Steering ships to other nearby ports. Reduce Base Case import and export shocks by 50% across sectors Effective Management Adjust import and export shocks Improvements in decision- making and expertise that enhance functionality (e.g., information sharing, utilizing digital incoming cargo shipment data to increase cargo handling productivity). Reduce Base Case import shocks by 10% across sectors Export Diversion for Import Use Adjust import and export shocks Using goods that were intended for export as substitutions for the lack of availability of imports. Reduce import and export disruptions between 0% and 100% across sectors (depending on availability of similar 10-digit HTS exported 155 commodity that can be sequestered for import use) Inventory Use Adjust import shock Reducing the direct import disruption by the amount of inventory. Reduce import disruptions between 0% and100% across sectors (based on a comparison of BEA inventory data and Base Case import disruptions by sector) Production Recapture Application of “Recapture Factor Parameter” to output changes A side-calculation to adjust total output losses for production rescheduling. Recapture factors range from 0.223 to 0.429 across sectors. 156 Appendix C. Detailed Income Distribution Impact Results Table C1. Baseline Income Distribution and Income Changes in the Port Disruption Simulation for the LA Metro Region (in millions 2010 dollars) Income Bracket Income Distribution Income Changes relative to Baseline (M $) Income Changes relative to Baseline (%) Baseline Port Disruption Base Case Port Disruption Resilience Case Port Disruption Base Case Port Disruption Resilience Case Port Disruption Base Case Port Disruption Resilience Case <10k 3,474.1 3,470.5 3,472.2 -3.64 -1.87 -0.1048% -0.0539% 10-15k 9,993.2 9,981.3 9,987.6 -11.96 -5.58 -0.1196% -0.0558% 15-25k 20,527.7 20,461.1 20,511.8 -66.55 -15.90 -0.3242% -0.0775% 25-35k 37,426.9 37,290.4 37,392.5 -136.49 -34.40 -0.3647% -0.0919% 35-50k 56,675.1 56,495.4 56,620.7 -179.67 -54.38 -0.3170% -0.0960% 50-75k 77,908.2 77,686.9 77,838.1 -221.32 -70.11 -0.2841% -0.0900% 75-100k 74,636.7 74,459.5 74,573.8 -177.17 -62.89 -0.2374% -0.0843% 100- 150k 80,606.5 80,413.0 80,541.8 -193.49 -64.71 -0.2400% -0.0803% 150k+ 164,409.6 164,042.7 164,261.3 -366.88 -148.22 -0.2231% -0.0902% Total 525,658.0 524,300.8 525,199.9 -1,357.16 -458.05 -0.2582% -0.0871% Table C2. Baseline Income Distribution and Income Changes in the Transportation Cost Increase Simulation for the LA Metro Region (in millions 2010 dollars) Income Bracket Income Distribution Income Changes relative to Baseline (M $) Income Changes relative to Baseline (%) Baseline Transportation Cost Increase Base Case Transportation Cost Increase Resilience Case Transportation Cost Increase Base Case Transportation Cost Increase Resilience Case Transportation Cost Increase Base Case Transportation Cost Increase Resilience Case 157 <10k 3,474.1 3,474.1 3,474.1 -0.05 -0.05 -0.0015% -0.0013% 10-15k 9,993.2 9,993.1 9,993.1 -0.14 -0.13 -0.0014% -0.0013% 15-25k 20,527.7 20,527.3 20,527.3 -0.43 -0.38 -0.0021% -0.0019% 25-35k 37,426.9 37,426.1 37,426.2 -0.80 -0.72 -0.0021% -0.0019% 35-50k 56,675.1 56,674.0 56,674.1 -1.14 -1.03 -0.0020% -0.0018% 50-75k 77,908.2 77,907.0 77,907.1 -1.25 -1.13 -0.0016% -0.0014% 75-100k 74,636.7 74,635.1 74,635.3 -1.53 -1.39 -0.0021% -0.0019% 100- 150k 80,606.5 80,604.9 80,605.0 -1.56 -1.41 -0.0019% -0.0017% 150k+ 164,409.6 164,405.9 164,406.3 -3.66 -3.31 -0.0022% -0.0020% Total 525,658.0 525,647.4 525,648.4 -10.55 -9.55 -0.0020% -0.0018% Table C3. Baseline Income Distribution and Income Changes in the General Building Damages Simulation for the LA Metro Region (in millions 2010 dollars) Income Bracket Income Distribution Income Changes relative to Baseline (M $) Income Changes relative to Baseline (%) Baseline Building Damage Base Case Building Damage Resilience Case Building Damage Base Case Building Damage Resilience Case Building Damage Base Case Building Damage Resilience Case <10k 3,474.1 3,377.9 3,398.4 -96.20 -75.67 -2.7691% -2.1780% 10-15k 9,993.2 9,717.5 9,776.4 -275.76 -216.84 -2.7595% -2.1699% 15-25k 20,527.7 19,961.7 20,083.6 -566.02 -444.08 -2.7573% -2.1633% 25-35k 37,426.9 36,265.9 36,516.9 -1,161.03 -910.02 -3.1021% -2.4314% 35-50k 56,675.1 54,770.7 55,181.9 -1,904.45 -1,493.25 -3.3603% -2.6347% 50-75k 77,908.2 75,260.2 75,831.4 -2,648.01 -2,076.80 -3.3989% -2.6657% 75-100k 74,636.7 71,997.2 72,564.3 -2,639.49 -2,072.35 -3.5365% -2.7766% 100- 150k 80,606.5 77,818.8 78,417.5 -2,787.64 -2,188.98 -3.4583% -2.7156% 150k+ 164,409.6 157,795.0 159,211.6 -6,614.55 -5,197.95 -4.0232% -3.1616% Total 525,658.0 506,964.8 510,982.0 -18,693.14 -14,675.92 -3.5561% -2.7919% 158 Table C4. Baseline Income Distribution and Income Changes in the Combined Disruptions/Damages Simulations for the LA Metro Region (in millions 2010 dollars) Income Bracket Income Distribution Income Changes relative to Baseline (M $) Income Changes relative to Baseline (%) Baseline Combined Simulation Base Case Combined Simulation Resilience Case Combined Simulation Base Case Combined Simulation Resilience Case Combined Simulation Base Case Combined Simulation Resilience Case <10k 3,474.1 3,374.3 3,396.6 -99.81 -77.52 -2.8729% -2.2313% 10-15k 9,993.2 9,705.6 9,770.9 -287.58 -222.35 -2.8778% -2.2250% 15-25k 20,527.7 19,895.6 20,067.7 -632.10 -460.04 -3.0792% -2.2411% 25-35k 37,426.9 36,130.8 36,482.5 -1,296.16 -944.39 -3.4632% -2.5233% 35-50k 56,675.1 54,593.2 55,127.9 -2,081.94 -1,547.22 -3.6735% -2.7300% 50-75k 77,908.2 75,043.0 75,762.3 -2,865.25 -2,145.98 -3.6777% -2.7545% 75-100k 74,636.7 71,822.9 72,502.2 -2,813.79 -2,134.48 -3.7700% -2.8598% 100- 150k 80,606.5 77,629.4 78,353.6 -2,977.10 -2,252.84 -3.6934% -2.7949% 150k+ 164,409.6 157,432.8 159,065.7 -6,976.73 -5,343.91 -4.2435% -3.2504% Total 525,658.0 505,627.5 510,529.2 -20,030.47 -15,128.73 -3.8106% -2.8781%

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

In-situ quality assessment of scan data for as-built models using building-specific geometric features

PDF

A radio frequency based indoor localization framework for supporting building emergency response operations

PDF

Behavioral form finding using multi-agent systems: a computational methodology for combining generative design with environmental and structural analysis in architectural design

PDF

Point cloud data fusion of RGB and thermal information for advanced building envelope modeling in support of energy audits for large districts

PDF

User-centric smart sensing for non-intrusive electricity consumption disaggregation in buildings

PDF

Uncertainty management for complex systems of systems: applications to the future smart grid

PDF

Learning personal thermal comfort and integrating personal comfort requirements into HVAC system control loop

PDF

Towards health-conscious spaces: building for human well-being and performance

PDF

Understanding human-building-emergency interactions in the built environment

PDF

Integration of digital twin and generative models in model-based systems upgrade methodology

PDF

Impact of virtual reality (VR)-based training on construction robotics remote-operation

PDF

Immersive computing for coastal engineering

PDF

Semantic modeling of outdoor scenes for the creation of virtual environments and simulations

PDF

Enabling human-building communication to promote pro-environmental behavior in office buildings

PDF

Integrating systems of desalination and potable reuse: reduced energy consumption for increased water supply

PDF

Introspective resilience for exascale high-performance computing systems

PDF

Understanding human-building interactions through perceptual decision-making processes

PDF

Modeling and simulation testbed for unmanned systems

PDF

COSYSMO 3.0: an extended, unified cost estimating model for systems engineering

PDF

Context-adaptive expandable-compact POMDPs for engineering complex systems

Asset Metadata

Creator Koc, Eyuphan (author)

Core Title A framework for comprehensive assessment of resilience and other dimensions of asset management in metropolis-scale transport systems

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Civil Engineering

Degree Conferral Date 2021-12

Publication Date 09/21/2021

Defense Date 06/21/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag disaster economics,mobility,natural hazards,OAI-PMH Harvest,resilience,Transportation Systems,urban policy

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Soibelman, Lucio (committee chair), Becerik-Gerber, Burcin (committee member), Madni, Azad (committee member), Wei, Dan (committee member)

Creator Email ekoc@usc.edu,eyuphan.koc@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15921166

Unique identifier UC15921166

Legacy Identifier etd-KocEyuphan-10084

Document Type Dissertation

Format application/pdf (imt)

Rights Koc, Eyuphan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu