The HayWired Earthquake Scenario—Engineering Implications Scientific Investigations Report 2017–5013–I–Q U.S. Department of the Interior U.S. Geological Survey Cover.  This oblique aerial photograph captures a moment during which San Francisco, California, firefighters are extinguishing a fire that occurred shortly after an apartment building collapsed and burned to the ground as a result of the moment-magnitude-6.9 Loma Prieta earthquake of 1989. Another apartment building across the street has fallen into the intersection of Beach and Divisadero Streets, bursting out the walls of the weak first story as the structure buckled and collapsed. The damage (building collapses, damage to gas pipelines and other utilities, and fire) in the city’s Marina District shown here was caused by amplified ground shaking and liquefaction (soils becoming liquid-like during shaking). Since 1989, these and five other collapsed buildings in San Francisco’s Marina District have been replaced or rebuilt, other buildings’ soft first stories have been braced and strengthened, and flexible-conduit gas lines have replaced old, brittle rigid gas lines throughout the neighborhood. Such risk-reduction measures, intended to prevent building collapse and to curtail fire following earthquake, have not yet been taken in many other areas surrounding San Francisco Bay that have been identified as being susceptible to liquefaction. (Photograph copyright Deanne Fitzmaurice/San Francisco Chronicle/Polaris, used with permission.) The HayWired Earthquake Scenario— Engineering Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q U.S. Department of the Interior U.S. Geological Survey U.S. Department of the Interior RYAN K. ZINKE, Secretary U.S. Geological Survey William H. Werkheiser, Deputy Director exercising the authority of the Director U.S. Geological Survey, Reston, Virginia: 2018 For more information on the USGS—the Federal source for science about the Earth, its natural and living resources, natural hazards, and the environment—visit https://www.usgs.gov or call 1–888–ASK–USGS (1–888–275–8747). For an overview of USGS information products, including maps, imagery, and publications, visit https://store.usgs.gov. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. Although this information product, for the most part, is in the public domain, it also may contain copyrighted materials as noted in the text. Permission to reproduce copyrighted items must be secured from the copyright owner. Suggested citation: Detweiler, S.T., and Wein, A.M., eds., 2018, The HayWired earthquake scenario—Engineering implications: U.S. Geological Survey Scientific Investigations Report 2017–5013–I–Q, 429 p., https://doi.org/10.3133/sir20175013v2. ISSN 2328-0328 (online) iii Foreword The 1906 Great San Francisco earthquake (magnitude 7.8) and the 1989 Loma Prieta earthquake (magnitude 6.9) each motivated residents of the San Francisco Bay region to build countermeasures to earthquakes into the fabric of the region. Since Loma Prieta, bay-region communities, governments, and utilities have invested tens of billions of dollars in seismic upgrades and retrofits and replacements of older buildings and infrastructure. Innovation and state-of-the-art engineering, informed by science, including novel seismic-hazard assessments, have been applied to the challenge of increasing seismic resilience throughout the bay region. However, as long as people live and work in seismically vulnerable buildings or rely on seismically vulnerable transportation and utilities, more work remains to be done. With that in mind, the U.S. Geological Survey (USGS) and its partners developed the HayWired scenario as a tool to enable further actions that can change the outcome when the next major earthquake strikes. By illuminating the likely impacts to the present-day built environment, well-constructed scenarios can and have spurred officials and citizens to take steps that change the outcomes the scenario describes, whether used to guide more realistic response and recovery exercises or to launch mitigation measures that will reduce future risk. The HayWired scenario is the latest in a series of like-minded efforts to bring a special focus onto potential impacts when the Hayward Fault again ruptures through the east side of the San Francisco Bay region as it last did in 1868. Cities in the east bay along the Richmond, Oakland, and Fremont corridor would be hit hardest by earthquake ground shaking, surface fault rupture, aftershocks, and fault afterslip, but the impacts would reach throughout the bay region and far beyond. The HayWired scenario name reflects our increased reliance on the Internet and telecommunications and also alludes to the interconnectedness of infrastructure, society, and our economy. How would this earthquake scenario, striking close to Silicon Valley, impact our interconnected world in ways and at a scale we have not experienced in any previous domestic earthquake? The area of present-day Contra Costa, Alameda, and Santa Clara Counties contended with a magnitude-6.8 earthquake in 1868 on the Hayward Fault. Although sparsely populated then, about 30 people were killed and extensive property damage resulted. The question of what an earthquake like that would do today has been examined before and is now revisited in the HayWired scenario. Scientists have documented a series of prehistoric earthquakes on the Hayward Fault and are confident that the threat of a future earthquake, like that modeled in the HayWired scenario, is real and could happen at any time. The team assembled to build this scenario has brought innovative new approaches to examining the natural hazards, impacts, and consequences of such an event. Such an earthquake would also be accompanied by widespread liquefaction and landslides, which are treated in greater detail than ever before. The team also considers how the now prototype ShakeAlert earthquake early warning system could provide useful public alerts and automatic actions. Scientific Investigations Report 2017–5013 and accompanying data releases are the products of an effort led by the USGS, but this body of work was created through the combined efforts of a large team including partners who have come together to form the HayWired Coalition (see chapter A). Use of the HayWired scenario has already begun. More than a full year of intensive partner engagement, beginning in April 2017, is being directed toward producing the most in-depth look ever at the impacts and consequences of a large earthquake on the Hayward Fault. With the HayWired scenario, our hope is to encourage and support the active ongoing engagement of the entire community of the San Francisco Bay region by providing the scientific, engineering, and economic and social science inputs for use in exercises and planning well into the future. David Applegate Associate Director for Natural Hazards, exercising the authority of the Deputy Director U.S. Geological Survey v HayWired Review Panel The HayWired Review Panel, a group whose expertise spans the scope of the HayWired scenario, assessed the overarching goals of the project along with the scientific approach and oversaw the reviews of each individual chapter in this volume. The panel consisted of Jack Boatwright (U.S. Geological Survey, USGS), Arrietta Chakos (Urban Resilience Strategies), Mary Comerio (University of California, Berkeley), Douglas Dreger (University of California, Berkeley), Erol Kalkan (USGS), Roberts McMullin (East Bay Municipal Utility District), Andrew Michael (chair, USGS), David Schwartz (USGS), and Mary Lou Zoback (Build Change, Stanford University). HayWired Coalition Partners Alameda County Mayors’ Conference Alameda County Sheriff’s Office, Office of Emergency Services American Red Cross Art Center College of Design ARUP—Design and Engineering Consultants Association of Bay Area Governments—Metropolitan Transportation Commission Aurecon Bay Area Center for Regional Disaster Resilience Bay Area Council Bay Area Rapid Transit Authority Bay Area Urban Area Security Initiative Bay Planning Coalition Boston University Business Recovery Managers Association California Business, Consumer Services, and Housing Agency California Department of Public Health California Department of Transportation California Earthquake Authority California Earthquake Clearinghouse California Geological Survey California Governor’s Office of Business and Economic Development California Governor’s Office of Emergency Services California Independent Oil Marketers Association California ISO California Public Utilities Commission California Resiliency Alliance California Seismic Safety Commission Carnegie Melon University Silicon Valley City and County of San Francisco City of Berkeley City of Fremont City of Hayward City of Oakland City of Oakland, Fire Department City of San Francisco, Department of Emergency Management City of Walnut Creek Contra Costa County Mayors’ Conference Earthquake Country Alliance Earthquake Engineering Research Institute East Bay Municipal Utility District Federal Emergency Management Agency Joint Venture Silicon Valley Laurie Johnson Consulting Research March Studios Marin Economic Consulting MMI Engineering Office of the Mayor, City and County of San Francisco Pacific Earthquake Engineering Research Center Pacific Gas and Electric Palo Alto University Price School of Public Policy and Center for Risk and Economic Analysis of Terrorism Events, University of Southern California Rockefeller Foundation—100 Resilient Cities San Jose Water Company Southern California Earthquake Center SPA Risk LLC SPUR Strategic Economics Structural Engineers Association of Northern California The Brashear Group LLC University of California Berkeley Seismological Laboratory University of Colorado Boulder University of Southern California U.S. Department of Homeland Security U.S. Geological Survey Wells Fargo vii Contents Foreword.........................................................................................................................................................iii By David Applegate HayWired Review Panel................................................................................................................................v HayWired Coalition Partners........................................................................................................................v Chapters I. Overview of the HayWired Scenario Engineering-Implications Volume........................................1 By Keith A. Porter Introduction............................................................................................................................................1 A Recap of Earthquake Hazards.........................................................................................................2 Engineering Implications......................................................................................................................2 Hazus-MH Analysis......................................................................................................................2 Societal Consequences of Current Code Performance Objectives.....................................3 A Survey of Public Preferences for the Seismic Performance of New Buildings.............5 Earthquake Urban Search and Rescue.....................................................................................5 Water-Network Resilience..........................................................................................................6 Repair Costs and Downtime of High-Rise Buildings...............................................................7 Fire Following Earthquake...........................................................................................................8 Benefit of Combining Earthquake Early Warning with Drop, Cover, and Hold On.............9 Future Research..........................................................................................................................10 Conclusion............................................................................................................................................10 References Cited.................................................................................................................................11 J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks................................13 By Hope A. Seligson, Anne M. Wein, and Jamie L. Jones Abstract.................................................................................................................................................13 Introduction..........................................................................................................................................14 Building Inventory Data......................................................................................................................14 Hazus Results—HayWired Mainshock............................................................................................18 HayWired Liquefaction Implementation in Hazus.................................................................18 HayWired Landslide implementation in Hazus......................................................................19 HayWired Mainshock Hazus Results......................................................................................22 Liquefaction Modeling in Selected HayWired Aftershocks.........................................................27 Ground Shaking Results for the HayWired Earthquake Sequence.............................................27 Unreinforced Masonry Construction................................................................................................30 Population Impacts—Casualties......................................................................................................38 Population Impacts—Displacement and Shelter Requirements.................................................38 Combining Losses in the Mainshock and Aftershocks.................................................................41 Repeat Liquefaction............................................................................................................................43 Model Limitations and Data Gaps.....................................................................................................46 Conclusions..........................................................................................................................................49 viii Mainshock Assessment............................................................................................................49 Aftershock Assessment.............................................................................................................49 Acknowledgments...............................................................................................................................50 References Cited.................................................................................................................................51 Appendix. Calculation of Shelter Parameters for the 2008 Southern California ShakeOut Scenario ..................................................................................................................................52 By Kimberley Shoaf Displacement ..............................................................................................................................52 Shelter Needs..............................................................................................................................53 Other Parameters to Consider .................................................................................................53 References Cited........................................................................................................................54 K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes.....57 By Keith A. Porter Abstract.................................................................................................................................................57 Introduction..........................................................................................................................................57 Background..........................................................................................................................................58 Implications of Seismic Performance Objectives for the HayWired Scenario.........................60 Impairment Rate at MCER Shaking....................................................................................................61 Validation Using Data from the August 2014 Napa Earthquake...................................................63 Impairment Rate at Other Levels of Shaking...................................................................................63 Fraction and Number of Impaired Buildings at the Societal Level..............................................66 Options for Improving Building-Stock Performance......................................................................66 Building Impairment Resulting from Ground Failure......................................................................69 Public Expectations for the Seismic Performance of New Buildings ........................................70 Limitations of this Study......................................................................................................................72 Summary...............................................................................................................................................72 Conclusion............................................................................................................................................73 References Cited.................................................................................................................................73 L. Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings in California and the New Madrid Seismic Zone.....................................................79 By Keith A. Porter Abstract.................................................................................................................................................79 Introduction..........................................................................................................................................80 Objectives....................................................................................................................................81 Survey Approach........................................................................................................................82 Respondent Population and Sampling Procedure................................................................83 Survey Questions and Responses....................................................................................................83 Response Rate............................................................................................................................83 Role or Relation to Building Codes..........................................................................................84 Current and Preferred Code Objectives..................................................................................84 Preferred Performance Measure ...........................................................................................86 Acceptable Cost for Better Performance...............................................................................87 How Important are These Issues?...........................................................................................88 Respondent Demographics.......................................................................................................88 Are Respondents Representative of the Population?....................................................................91 Are Some Groups Willing to Pay More for Better Seismic Performance?................................93 ix Conclusions..........................................................................................................................................94 Study Implications......................................................................................................................94 Limitations and Research Needs.............................................................................................95 Acknowledgments...............................................................................................................................95 References Cited.................................................................................................................................95 M. An Earthquake Urban Search and Rescue Model and its Application to the HayWired Scenario... 99 By Keith A. Porter Abstract.................................................................................................................................................99 Introduction..........................................................................................................................................99 Objective.............................................................................................................................................100 Literature Review...............................................................................................................................100 Literature About People Trapped by Building Collapse......................................................100 Literature About People Trapped in Elevators.....................................................................103 Methodology.......................................................................................................................................103 Methodology for Estimating the Number of People Trapped by Collapse......................103 Methodology for Estimating the Number of People Trapped in Elevators......................111 Application to HayWired Scenario.................................................................................................111 People Trapped in Collapsed Buildings, Based on the Building-Code Objectives........111 People Trapped in Collapsed Buildings, Based on Hazus-MH..........................................112 Scenario Estimate of People Trapped in Collapsed Buildings..........................................116 Number of People Trapped in Stalled Elevators..................................................................116 Conclusions........................................................................................................................................117 USAR Demands Under Current Conditions..........................................................................117 USAR Demands Under Ideal-World Conditions...................................................................118 Limitations...........................................................................................................................................118 Acknowledgments.............................................................................................................................118 References Cited...............................................................................................................................118 Appendixes 1 through 12—National Information Service for Earthquake Engineering e-Library Images of Building Collapse in California, 1965–2014...................................120 Appendix 1. Santa Rosa (1969) Collapse Images.................................................................120 Appendix 2. San Fernando (1971) Collapse Images............................................................122 Appendix 3. Imperial Valley (1979) Collapse Images...........................................................141 Appendix 4. Westmorland (1981) Collapse Images.............................................................143 Appendix 5. Coalinga (1983) Collapse Images.....................................................................144 Appendix 6. Morgan Hill (1984) Collapse Images................................................................152 Appendix 7. Whittier Narrows (1987) Collapse Images......................................................154 Appendix 8. Loma Prieta (1989) Collapse Images................................................................159 Appendix 9. Northridge (1994) Collapse Images..................................................................173 Appendix 10. San Simeon (2003) Collapse Images..............................................................189 Appendix 11. South Napa (2014) Collapse Images..............................................................191 Appendix 12. Earthquakes With No Available Collapse Images.......................................192 N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario....195 By Keith A. Porter Abstract...............................................................................................................................................195 Introduction........................................................................................................................................195 How Water Supply is Important in an Earthquake..............................................................195 x Study Objectives.......................................................................................................................196 Literature Review...............................................................................................................................197 A Panel Approach to Estimating Water-Supply Impacts...................................................197 Analytical Approaches to Estimating Water Supply Impacts...........................................197 Damageability of Buried Pipe.................................................................................................198 Tasks and Methods to Repair Leaks and Breaks................................................................202 Time to Repair Pipe Leaks and Breaks..................................................................................202 Serviceability of Water Supply...............................................................................................205 Lifeline Interaction....................................................................................................................206 Pipeline Damage in Afterslip..................................................................................................210 Measuring Loss of Resilience................................................................................................211 Methodology.......................................................................................................................................211 Overview of the Methodology................................................................................................211 Vulnerability Model...................................................................................................................212 Damage Analysis (Number of Repairs Required)................................................................214 Restoration Model....................................................................................................................215 Measuring Water-Supply Resilience....................................................................................223 Optional Stochastic Simulation Methodology.....................................................................223 Accounting for Afterslip and Aftershocks............................................................................225 Adjusting Hazus-MH’s Lifeline Restoration Model.............................................................226 Mitigation Options....................................................................................................................227 Summary of the Methodology................................................................................................228 Case Study 1—San Jose Water Company....................................................................................228 San Jose Water Company Asset Definition.........................................................................229 San Jose Water Company Hazard Analysis.........................................................................229 San Jose Water Company Damage Analysis.......................................................................230 San Jose Water Company Restoration Analysis.................................................................236 Validation of San Jose Water Company Restoration Analysis.........................................240 Effect of Lifeline Interaction and Consumable Limits.........................................................242 Case Study 2—East Bay Municipal Utility District.......................................................................242 East Bay Municipal Utility District Asset Definition............................................................242 EBMUD Hazard Analysis.........................................................................................................244 EBMUD Damage Analysis.......................................................................................................245 EBMUD Restoration Analysis.................................................................................................251 Validation of EBMUD Damage and Recovery Estimates....................................................254 Effect of Lifeline Interaction and Consumable Limits on EBMUD....................................255 Performance of Other Water Utilities Based on Hazus-MH.......................................................255 Conclusions........................................................................................................................................258 Summary....................................................................................................................................258 Innovations Introduced Here..................................................................................................258 Research Needs.......................................................................................................................260 Acknowledgments.............................................................................................................................260 References Cited...............................................................................................................................260 xi O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Earthquake Scenario Mainshock.....................................................................267 By Ibrahim M. Almuti, Carlos Molina-Hutt, Michael W. Mieler, Nicole A. Paul, and Chad R. Fusco Abstract...............................................................................................................................................267 Introduction........................................................................................................................................268 Objectives..................................................................................................................................268 Report Structure.......................................................................................................................268 HayWired Ground Motions..............................................................................................................269 Description of Archetype Buildings, Design, and Analysis Assumptions................................272 Steel Office Tower....................................................................................................................272 Reinforced-Concrete Residential Tower...............................................................................276 Loss-Assessment Methodology......................................................................................................278 REDi Downtime Methodology.................................................................................................278 Input Parameters......................................................................................................................280 Replacement Value...................................................................................................................280 Utility Disruption........................................................................................................................281 Summary of Loss-Assessment Results..........................................................................................281 Conclusion..........................................................................................................................................283 Acknowledgments.............................................................................................................................283 References Cited...............................................................................................................................284 Appendix 1. Building Structural and Nonstructural Components............................................285 Structural Components............................................................................................................286 Nonstructural Components.....................................................................................................291 Appendix 2. S-SF-B-43—40-Story Steel-Frame Building in San Francisco (Baseline Orientation)............................................................................................................................312 S-SF-B-43 Description.............................................................................................................312 Engineering-Demand Parameters.........................................................................................312 Loss-Assessment Results.......................................................................................................312 Appendix 3. S-SF-R-43—40-Story Steel-Frame Building in San Francisco (Rotated Orientation)............................................................................................................................316 S-SF-R-43 Description..............................................................................................................316 Engineering-Demand Parameters.........................................................................................316 Loss-Assessment Results.......................................................................................................316 Appendix 4. S-SF-B-20—20-Story Steel-Frame Building in San Francisco (Baseline Orientation)............................................................................................................................320 S-SF-B-20 Description.............................................................................................................320 Engineering-Demand Parameters.........................................................................................320 Loss-Assessment Results.......................................................................................................320 Appendix 5. S-SF-R-20—20-Story Steel-Frame Building in San Francisco (Rotated Orientation)............................................................................................................................324 S-SF-R-20 Description..............................................................................................................324 Engineering-Demand Parameters.........................................................................................324 Loss-Assessment Results.......................................................................................................324 xii Appendix 6. S-OK-B-20—20-Story Steel-Frame Building in Oakland (Baseline Orientation)..........................................................................................................328 S-OK-B-20 Description.............................................................................................................328 Engineering-Demand Parameters.........................................................................................328 Loss Assessment Results........................................................................................................328 Appendix 7. S-OK-R-20—20-Story Steel-Frame Building in Oakland (Rotated Orientation)...........................................................................................................332 S-OK-R-20 Description.............................................................................................................332 Engineering-Demand Parameters.........................................................................................332 Loss-Assessment Results.......................................................................................................332 Appendix 8. C-SF-B-46—42-Story Reinforced-Concrete Building in San Francisco (Baseline Orientation)............................................................................................................................336 C-SF-B-46 Description.............................................................................................................336 Engineering-Demand Parameters.........................................................................................336 Loss-Assessment Results.......................................................................................................339 Appendix 9. C-SF-R-46—42-Story Reinforced-Concrete Building in San Francisco (Rotated Orientation)............................................................................................................................342 C-SF-R-46 Description..............................................................................................................342 Engineering-Demand Parameters.........................................................................................342 Loss-Assessment Results.......................................................................................................345 Appendix 10. C-OK-B-46—42-Story Reinforced-Concrete Building in Oakland (Baseline Orientation)............................................................................................................................348 C-OK-B-46 Description.............................................................................................................348 Engineering-Demand Parameters.........................................................................................348 Loss-Assessment Results.......................................................................................................351 Appendix 11. C-OK-R-46—42-Story Reinforced-Concrete Building in Oakland (Rotated Orientation)............................................................................................................................354 C-OK-R-46 Description.............................................................................................................354 Engineering-Demand Parameters.........................................................................................354 Loss-Assessment Results.......................................................................................................357 Appendix 12. Inventory of Existing Tall-Building Stock in San Francisco...............................360 P. Fire Following the HayWired Scenario Mainshock.........................................................................367 By Charles Scawthorn Abstract...............................................................................................................................................367 Introduction........................................................................................................................................367 Purpose......................................................................................................................................368 Background...............................................................................................................................368 Scenario Earthquake and Prevailing Conditions..........................................................................368 Rupture Segment, Magnitude, and Intensity.......................................................................369 Affected Region........................................................................................................................369 Fire Following Earthquake Aspects of the Scenario....................................................................379 Modeling of Fire Following Earthquake................................................................................379 Ignitions......................................................................................................................................380 Initial Response.........................................................................................................................381 Fire Spread.................................................................................................................................384 Lifelines......................................................................................................................................384 xiii Regional and State Response.................................................................................................390 Final Burned Area.....................................................................................................................391 Uncertainty, Verification, and Validation..............................................................................393 Impacts of Fire Following Earthquake............................................................................................394 Human Impacts.........................................................................................................................395 Economic and Insurance Impacts.........................................................................................395 Mitigation of Fire Following Earthquake........................................................................................395 Fire-Service Opportunities......................................................................................................395 Water-Service Opportunities..................................................................................................396 Building-Standards Opportunities.........................................................................................396 Energy-Industry Opportunities...............................................................................................396 Conclusion..........................................................................................................................................397 References Cited...............................................................................................................................397 Q. How Many Injuries Can Be Avoided in the HayWired Scenario Through Earthquake Early Warning and Drop, Cover, and Hold On?...................................................................................401 By Keith A. Porter and Jamie L. Jones Abstract...............................................................................................................................................401 Introduction........................................................................................................................................401 Background...............................................................................................................................401 Objectives..................................................................................................................................402 Literature Review...............................................................................................................................402 Methodology.......................................................................................................................................403 Estimating Avoided Injuries.....................................................................................................403 Estimating the Acceptable Cost to Avoid Injuries...............................................................404 Survey Design...........................................................................................................................404 Findings...............................................................................................................................................404 Distribution of DCHO Reaction Time......................................................................................404 Calculation of Benefits.............................................................................................................406 Conclusions........................................................................................................................................406 Limitations and Research Needs....................................................................................................410 References Cited...............................................................................................................................410 Appendix 1. Survey of DCHO Reaction Times..............................................................................412 Appendix 2. HayWired Mainshock Earthquake Early Warning Time Calculation..................427 By Elizabeth S. Cochrane, Anne M. Wein, Erin R. Burkett, Douglas D. Given, and Keith Porter References Cited......................................................................................................................428 xiv Conversion Factors U.S. customary units to International System of Units Multiply By To obtain Length inch (in.) 2.54 inch (in.) centimeter (cm) 25.4 foot (ft) mile (mi) millimeter (mm) 0.3048 meter (m) 1.609 kilometer (km) Area acre 4,047 acre square meter (m2) 0.4047 hectare (ha) 0.004047 square kilometer (km2) square foot (ft ) 0.09290 square meter (m2) square mile (mi2) 2.590 square kilometer (km2) acre 2 Volume gallon (gal) 3.785 liter (L) Pressure pound per square inch (lb/in2) 6.895 kilopascal (kPa) kilopounds per square inch (lb/in2) 6.895 megapascal (MPa) Cohesion pound per square foot (lb/ft2) 0.04788 kilopascal (kPa) Velocity mile per hr (mi/hr) 1.60934 kilometer per hour (km/hr) Angle degree (°) 0.0174533 radian (rad) International System of Units to U.S. customary units Multiply By To obtain Length centimeter (cm) 0.3937 inch (in.) millimeter (mm) 0.03937 inch (in.) meter (m) 3.281 foot (ft) kilometer (km) 0.6214 mile (mi) square meter (m2) 0.0002471 acre 2.471 acre Area hectare (ha) square kilometer (km2) square meter (m ) 247.1 acre 10.76 2 square foot (ft2) square kilometer (km ) 0.3861 liter (L) 0.2642 kilopascal (kPa) 0.1450377 pound per square inch (lb/in2) megapascal (MPa) 0.1450377 kilopounds per square inch (lb/in2) 2 square mile (mi2) Volume gallon (gal) Pressure xv Multiply By To obtain Velocity centimeter per second (cm/s) centimeter per second (cm/s) 0.3937 0.0223694 inch per second (in./s) mile per hour (mi/hr) meter per second (m/s) 3.281 foot per second (ft/s) meter per second (m/s) 2.23694 mile per hour (mi/hr) kilometer per hour (km/hr) 0.621371 mile per hour (mi/hr) Angle radian (rad) 57.2958 degree (°) Datum Vertical coordinate information is referenced to the North American Vertical Datum of 1988 (NAVD 88). Horizontal coordinate information is referenced to the North American Datum of 1983 (NAD 83). Abbreviations and Acronyms 1D one dimensional 2D two dimensional 3D three dimensional ABAG Association of Bay Area Governments AIS abbreviated injury scale ASCE American Society of Civil Engineers ATC Applied Technology Council BAREPP Bay Area Regional Earthquake Preparedness Project BART Bay Area Rapid Transit BORP Building Occupancy Resumption Program BSSC Building Seismic Safety Council Cal OES California Governor’s Office of Emergency Services Caltrans California Department of Transportation CalWARN California Water/Wastewater Agency Response Network CAPSS Citizens Advisory Panel on Seismic Safety or Community Action Plan for Seismic Safety CDC Centers for Disease Control and Prevention CERT community emergency response team CGS California Geological Survey CPT cone penetration test CUREE Consortium of Universities for Research in Earthquake Engineering xvi DBE design-basis earthquake DCHO drop, cover, and hold on DDR demand-to-design ratio DDR1 1-second DDR DDRS short-period DDR DEM digital elevation model EBMUD East Bay Municipal Utility District EDP engineering-demand parameter EERI Earthquake Engineering Research Institute EEW Earthquake early warning EQE EQE International ESIP Earthquake Safety Improvements Program ETAS epidemic type aftershock sequence FA amplification factor FEMA Federal Emergency Management Agency Fv site coefficient g acceleration due to gravity GMPE ground-motion prediction equation GMPGV geometric mean of the peak ground velocity hr hour IBC International Building Code ICC International Code Council IDR interstory drift Ie seismic importance factor IRB institution review board LA BOMA Los Angeles Building Owners and Managers Association LADWP Los Angeles Department of Water and Power LLEQE Life Line Earthquake Engineering software LPI liquefaction potential index LRFD load- and resistance-factor design M magnitude Ma mega-annum or millions of years ago MCE maximum considered earthquake MCER risk-adjusted maximum considered earthquake MEP mechanical, electrical, and plumbing xvii MMI Modified Mercalli Intensity MRF moment-resisting frame MSA metropolitan statistical area Mw moment magnitude NAD83 North American Datum of 1983 NED National Elevation Dataset NEHRP National Earthquake Hazards Reduction Program NGA-West2 Next Generation Attenuation Relationships for Western United States NIBS National Institute of Building Sciences NISEE National Information Service for Earthquake Engineering NIST National Institute of Standards and Technology NISTIR National Institute of Standards and Technology Interagency Reports NLRHA nonlinear response-history analysis NMSZ New Madrid Seismic Zone NRC National Research Council P probability PACT Performance Assessment Calculation Tool PBEE performance-based earthquake engineering PBEE-2 second generation Performance-based earthquake engineering PDT Pacific Daylight Time PEER Pacific Earthquake Engineering Research Center PG&E Pacific Gas and Electric Company PGA peak ground acceleration PGD permanent ground displacement PGV peak ground velocity PHS U.S. Public Health Service PSA or pSa pseudo-spectral acceleration PSA03 short-period (0.3-second) pseudo-spectral-acceleration response PSA10 long-period (1-second) pseudo-spectral-acceleration response PST Pacific Standard Time PVC polyvinyl chloride PWSS portable water-supply system R coefficient of determination REDi™ Resilience-based Earthquake Design Initiative for the Next Generation of Buildings RIDR residual interstory drift 2 xviii SA spectral acceleration Sa spectral-acceleration response SAFRR USGS Science Application for Risk Reduction project SCEC Southern California Earthquake Center SDC seismic-design categories SEAOC Structural Engineers Association of California SEAONC Structural Engineers Association of Northern California SEI Structural Engineering Institute SHZ seismic hazard zone SIP seismic improvement programs SJWC San Jose Water Company SLE serviceability-level earthquakes SM1 1-second spectral response acceleration parameter SMS short-period spectral acceleration response parameter SPUR San Francisco Bay Area Planning and Urban Research Association SS short-period spectral acceleration response at MCER shaking T period UCERF3 Uniform California Earthquake Rupture Forecast, version 3 UPS uninterruptible power supply URM unreinforced masonry USAR urban search and rescue team USD U.S. Dollars USGS U.S. Geological Survey VS30 time-averaged shear-wave velocity to a depth of 30 meters WGS84 World Geodetic Survey 1984 ρ correlation coefficient Φ standard normal cumulative distribution function The HayWired Earthquake Scenario—Engineering Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q [Also see https://doi.org/10.3133/sir20175013] Chapter I Overview of the HayWired Scenario EngineeringImplications Volume By Keith A. Porter1 Introduction The HayWired scenario is a hypothetical yet scientifically realistic depiction of an earthquake sequence that begins with a moment magnitude (Mw) 7.0 earthquake (mainshock) occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of the San Francisco Bay area, California. The hypothetical mainshock has its epicenter in Oakland, and strong ground shaking from the scenario causes a wide range of severe impacts throughout the greater bay region. In the mainshock, the Hayward Fault is ruptured along its length for 83 kilometers (about 52 miles), an event significant enough to touch the lives of everyone who lives or works in the region. The HayWired Earthquake Scenario—Engineering Implications is the second volume of U.S. Geological Survey (USGS) Scientific Investigations Report (SIR) 2017–5013, which is planned to be published as three volumes. The previous volume of this work (SIR 2017–5013–A–H; Detweiler and Wein, 2017) described a Mw 7.0 mainshock on the Hayward Fault, along with an aftershock sequence and other geologic hazards. This volume (SIR 2017–5013–I–Q) presents engineering impacts that could result from the effects of the HayWired scenario. Together with environmental, social, and economic impacts (including impacts to telecommunications and the internet) that are planned to be described in a third volume, these works describe the HayWired scenario, which was developed by USGS and its partners. The engineering-related implications of the Mw 7.0 HayWired scenario mainshock and aftershocks for the San Francisco Bay region discussed in this volume are based on: 1. An analysis using the Federal Emergency Management Agency’s (FEMA) Hazus-MH computer program (Federal Emergency Management Agency, 2012a) that suggests 800 deaths and 16,000 nonfatal injuries (from the shaking hazard alone) could occur, as well as property and direct business interruption losses of more than $82 billion (for shaking, liquefaction, and landslide hazards). University of Colorado Boulder. 1 2. A study of the societal consequences of the International Building Code’s seismic-performance objectives for new buildings. The scenario indicates that, even for new buildings, the code protects life well but is not robust enough to ensure that hundreds of thousands of code-compliant buildings are not red-tagged (rendered unsafe to enter or occupy) or yellow-tagged (safe only for limited use). As a consequence, even if they were all new, code-compliant buildings, a significant fraction of the San Francisco Bay region’s buildings—perhaps 1 in 4—could have no or restricted occupancy. A moreresilient building stock is achievable for an additional 1 to 3 percent of construction cost and could allow 95 percent of homes and workplaces to be occupied following a powerful earthquake. 3. The first survey of public preferences for the tradeoff between cost and building resilience, showing that most people expect, prefer, and would be willing to pay for greater resilience of the building stock. 4. A new model of urban search and rescue, indicating that (1) more than 22,000 people in the region could require fire departments to free them from stalled elevators and (2) more than 2,400 people could require rescue from collapsed buildings. 5. A new, nonproprietary model of water-network resilience that accounts for the entire earthquake sequence, lifeline (for example, transportation infrastructure and buried utilities) interaction, resource limitations, and service restoration over time. This water-network resilience model shows that the average east bay resident could lose water service for 6 weeks (some for as long as 6 months). Two options to help improve these outcomes are evaluated. 6. A state-of-the-art performance-based earthquakeengineering study showing that an earthquake like the HayWired mainshock could cause damage sufficient to render older regular steel-frame high-rise office buildings and new regular reinforced-concrete 2   The HayWired Earthquake Scenario—Engineering Implications residential buildings in downtown Oakland and San Francisco unusable for as long as 10 months. 7. A study of fire following earthquake showing that the HayWired mainshock could cause about 450 large fires in counties nearest the fault rupture, burning buildingfloor area equivalent to that of more than 52,000 singlefamily dwellings. Such fires would kill hundreds of people and cause property (building and content) losses approaching $30 billion. A first joint exercise of portable firefighting water-supply systems was also conducted by fire agencies in the San Francisco Bay region. 8. An analysis of the benefits of combining earthquake early warning (EEW) and drop, cover, and hold on (DCHO) actions, including the first study of the time it takes people to complete DCHO. Combining EEW and DCHO could prevent as many as 1,500 nonfatal injuries out of 18,000 estimated nonfatal injuries (from shaking and liquefaction hazards combined) in the HayWired scenario, a benefit valued at about $300 million. A Recap of Earthquake Hazards The HayWired Earthquake Scenario—Earthquake Hazards volume (Detweiler and Wein, 2017) sets the stage for this volume on the engineering implications of the scenario. The Hayward Fault is arguably the most urbanized active fault in the United States. Therefore, it offers an informative case study of the effects of a large urban earthquake on a modern U.S. metropolitan region. The earthquake-hazards volume described the hypothetical Mw 7.0 HayWired scenario mainshock, with additional descriptions of the cascading hazards of fault rupture, aftershocks (subsequent earthquakes), afterslip (subsequent movement on a fault), landslides, and liquefaction (soils becoming liquid-like during shaking). The earthquake hazards volume also describes a largely physics-based model of ground motion for the HayWired mainshock. The model uses physical modeling of wave propagation but with a kinematic (motion-based) rupture model. This model of ground motion shows that damaging shaking (Modified Mercalli Intensity VI or higher) in the scenario occurs over a region of approximately 50,000 square kilometers (about 19,000 square miles)—170 kilometers (km) (about 105 miles) west to east (from the Pacific Coast to the Sierra Nevada) and 300 km (about 185 miles) north to south, including almost all the urbanized area of the nine counties bordering San Francisco Bay, as well as Santa Cruz County to the south. The differences between this largely physicsbased model and conventional ground-motion prediction equations were examined, and an explanation was provided of why the physics-based model was used to avoid a systematic underestimate of damage and loss for the HayWired scenario. The modeled scenario earthquake sequence causes as much as 2 meters (about 6.5 feet) of fault offset either in the form of coseismic slip (fault slip during the mainshock) or afterslip. The aftershock sequence includes 16 aftershocks of Mw 5.0 or larger that occur over 2 years and as far as 50 km (about 30 miles) from the Hayward Fault; several of the aftershocks cause local damaging ground shaking that is stronger than in the mainshock. Earthquake-induced liquefaction and landslides, caused by the mainshock, further threaten people, property, and lifeline infrastructure in every county in the San Francisco Bay region. Engineering Implications The HayWired Earthquake Scenario—Engineering Implications volume (this volume) examines how the HayWired scenario earthquake sequence would affect buildings, watersupply pipelines, and other infrastructure. Many researchers have estimated the engineering impacts of similar earthquakes on the Hayward Fault, and several previous studies of San Francisco Bay region earthquakes were discussed in chapter A (Hudnut and others, 2017). The engineering implications volume is not meant to be an exhaustive study of a Hayward Fault earthquake like the ShakeOut scenario (Jones and others, 2008), which examined a hypothetical Mw 7.8 earthquake on the San Andreas Fault in southern California and presented a Hazus-MH analysis along with an examination of 18 special engineering topics. This volume ignores some important, but already well-studied, topics such as soft-story buildings and nonductile-concrete buildings. It sacrifices breadth for innovation to explore some engineering and infrastructure aspects of a San Francisco Bay region earthquake that others have not yet examined, either for a Hayward Fault earthquake or indeed for any earthquake. Hazus-MH Analysis Chapter J (Seligson and others, this volume) estimates the spectrum of damage and loss the San Francisco Bay region would experience in the HayWired scenario mainshock using FEMA’s public risk-analysis software, Hazus-MH (Federal Emergency Management Agency, 2012a). The Hazus-MH analysis uses detailed estimates of liquefaction and landslide probability that are customized using estimates of shaking from the HayWired mainshock. It uses a map of developed area to better constrain the effects of the liquefaction and landslide hazards on the built environment than would typical Hazus analyses. For example, landslide damage is estimated at several probability levels to quantify a range of possible impacts to pockets of developed areas in large census tracts in the San Francisco Bay region. Hazus-MH analysis is done for each of the 16 modeled aftershocks of magnitude 5.0 or greater. The analyses estimate cumulative damage from aftershocks, explore the vulnerability of unreinforced masonry and tilt-up buildings to aftershocks, highlight geographic areas where aftershocks produce greater damage than does the HayWired mainshock, and show repeat Chapter I. Overview of the HayWired Scenario Engineering-Implications Volume  3 damage from liquefaction. The HayWired earthquake sequence could cause 800 deaths and about 16,000 nonfatal injuries from the shaking hazard alone. If there were even a single collapse of a high-rise building, it could greatly increase the casualty figures. Direct losses related to damaged building stock are more than $82 billion, not including losses from fire following earthquake, over the entire scenario earthquake sequence, with very little of it insured. These losses, estimated using Hazus-MH, include about: to earthquake ground shaking, about 80 percent is attributable to the Mw 7.0 mainshock, 12 percent to three aftershocks of Mw 6.0 to 6.4, and 8 percent to 13 aftershocks of Mw 5.0 to 5.9. Figure 1 shows that building damage from the mainshock and aftershocks, as a percentage of replacement value, is highest (greater than 10 percent) and most prevalent in Alameda County, followed by Contra Costa and Santa Clara Counties. It also shows widespread damage throughout the bay region of at least 0.5 percent of building replacement costs. • $53 billion in building repair costs, • $17 billion in damage to contents and inventory, and Societal Consequences of Current Code Performance Objectives • $12 billion in direct business interruption (that is, business interruption losses and additional expenses suffered by building occupants because of damage to their buildings). The Hazus-MH analysis for the HayWired scenario examines what happens to the existing building stock in the San Francisco Bay region in a powerful earthquake. Many previous studies have examined the potential effects of earthquakes on buildings (for Most of these estimated losses in the HayWired scenario are attributable to shaking damage (about 86 percent), the rest are attributable to liquefaction and landslides. Of the losses due 123° 122° 121° EXPLANATION YOLO SONOMA County boundary NAPA Building damage ratio—damage divided SACRAMENTO SOLANO by replacement value, expressed as a percent) 0 <0.1% 0.1–0.5% 0.5–2.5% 2.5–5% 5–10% >10% MARIN 38° SAN JOAQUIN CONTRA COSTA SAN FRANCISCO ALAMEDA STANISLAUS SAN MATEO SANTA CLARA IC CIF PA 37° MERCED SANTA CRUZ OC EAN Area CALIF of map SAN BENITO MONTEREY 0 Coordinate System: GCS North American 1983 Datum: North American 1983 0 1 10 2 20 3 30 4 40 5 MILES 50 MILES 10 KILOMETERS 1 20 2 30 3 40 4 50 5 KILOMETERS Figure 1.  Map of the San Francisco Bay region, California, showing estimated building-damage ratios (repair cost as a percent of building replacement cost) for the hypothetical magnitude-7.0 mainshock and aftershocks of the HayWired earthquake scenario on the Hayward Fault. The building-damage ratio was calculated using the public risk-analysis software Hazus-MH (Federal Emergency Management Agency, 2012a) and considered shaking, liquefaction (soils becoming liquid-like during shaking), and landslide hazards. %, percent. (From Seligson and others, this volume.) 4   The HayWired Earthquake Scenario—Engineering Implications example, Jones and others, 2008). The sometimes shockingly large estimates of building damage tend to draw attention to certain existing building types that contribute disproportionately to the overall damaged building stock, such as older unreinforcedmasonry buildings, nonductile concrete buildings, or older weldedsteel moment-frame buildings. In chapter K (Porter, Societal Consequences, this volume), I use this disaster-planning scenario for a new purpose—as a lens through which to view the intended performance of new buildings. The International Building Code (IBC), on which the California Building Code is based, aims to protect life safety by ensuring that fewer than 1 percent of buildings collapse because of earthquake shaking during their 50-year design life. However, earthquakes do not randomly affect individual buildings in the way traffic accidents affect people, one or a few at a time. Large urban earthquakes strike millions of people and hundreds of thousands of buildings simultaneously, so when a powerful urban earthquake strikes a region, 1 percent of buildings throughout the region could collapse simultaneously. The remaining buildings do not survive 123° 39° 122° 121° Area of map CALIFORNIA 38° Oakland San Francisco PA CI FI C O CE San Jose AN Figure 2.  Map of the San Francisco Bay region, California, showing impairment of existing buildings, if they all complied with current building codes, for the magnitude-7.0 mainshock of the HayWired earthquake scenario on the Hayward Fault. Impaired buildings include those that collapsed, are unsafe to occupy, or have restricted use. Warmer colors show areas with the greatest number of impaired buildings. Even if all buildings in the bay region complied with current building codes, 0.4 percent could collapse, 5 percent could be unsafe to occupy, and 19 percent could have restricted use. For only a small percentage cost increase, more resilient buildings constructed to more stringent building codes could allow 95 percent of the bay region’s population to remain in their homes and workplaces following such an earthquake. (From Porter, Societal Consequences, this volume.) unscathed. Many are damaged to the point that they cannot be used, or even economically repaired, after the earthquake. Chapter K uses survey data from the Mw 6.9 1989 Loma Prieta and Mw 6.7 1994 Northridge, California, earthquakes to show that, for every collapsed building, approximately 60 are damaged to the point that they are unsafe to enter or occupy (red tag) or have their use restricted to a part of the building or to limited duration (yellow tag). Many more buildings experience damage costing tens of thousands of dollars, often exceeding the owners’ financial resources to repair. Again, these results follow indirectly from the IBC’s explicit performance objectives for new buildings, which means that, with the code’s current performance objectives, the State’s existing building stock in 50 years (the year 2067) or even 100 years (the year 2117), for example, will still pose a serious threat to its economic well-being and the lives, livelihoods, and financial stability of the individual people who own, live, or work in those buildings. Figure 2 is a map showing the impairment of existing buildings in the San Francisco Bay region after the HayWired mainshock, even if they all complied 37° 0 0 0.0 20 20 40 40 60 MILES 60 KILOMETERS 0.2 0.4 0.6 Fraction of impaired buildings 0.8 1.0 Chapter I. Overview of the HayWired Scenario Engineering-Implications Volume  5 with current building codes. Chapter K also shows how stronger buildings, costing about 1 to 3 percent more to build (less than the cost to reroof most buildings) would suffer about 75 percent less damage in terms of collapse, red-tagging, and yellow-tagging in an earthquake like that modeled in the HayWired scenario. A A Survey of Public Preferences for the Seismic Performance of New Buildings The IBC strikes a balance for new buildings between better performance and less expense, but what does the public prefer? Chapter L (Porter, Not Safe Enough, this volume) reports on the first survey of the public’s expectations and preferences for the seismic performance of new buildings. It turns out that the current building code appears to aim for something different than the public prefers. A survey of 800 people (400 Californians and 400 people from the Memphis, Tennessee, and St. Louis, Missouri, metropolitan statistical areas near the New Madrid Seismic Zone in the Central United States) shows that the majority expect new buildings generally to be habitable or functional after large earthquakes, not merely safe. Survey respondents expressed a willingness to pay at least an additional 1 percent in construction costs to achieve their preferred degree of postearthquake performance (fig. 3). The willingness to pay more for better performance crosses geographic, wealth, and educational boundaries—people from the Central United States express almost the same preferences as do Californians, people from lower income households express almost the same preferences as those from higher income households, and level of education does not seem to affect these preferences. Other 2% Functional 18% Do not know 17% Life safe 22% Occupiable 41% B Do not know 17% $0/ft2 12% $1/ft2 20% Earthquake Urban Search and Rescue How does building collapse in an earthquake affect demands on urban search and rescue (USAR) teams? Collapse here is defined as the loss of vertical load-carrying capacity in at least part of a building’s structural system, which may or may not trap building occupants or passersby. Chapter M (Porter, Earthquake Urban Search and Rescue Model, this volume) estimates the number of people trapped in collapsed buildings in earthquakes, using a new model that draws on an examination of photographs of building collapses in California earthquakes of the past 5 decades. This examination suggests that, when a California building experiences at least some collapse in an earthquake, an average of 25 percent of its occupiable area collapses in such a way that occupants could be trapped under debris. It seems realistic that an earthquake like the HayWired scenario mainshock could trap about 2,500 people in 5,000 collapsed buildings. (There are more collapses than occupants, because not every building collapse traps people.) Chapter M also examines the question of how many people would require rescue from stalled elevators in a San Francisco Bay region earthquake like the HayWired scenario mainshock. The ShakeOut scenario hinted that many people could be trapped in stalled elevators when electric power is lost throughout a metropolitan region (Schiff, 2008), but it did not quantify the problem. AAXXXX_fig 01 $10/ft2 20% $3/ft2 31% Figure 3.  Pie charts showing public preferences for the seismic performance of new buildings from a survey of 800 people (400 Californians and 400 people from the Memphis, Tennessee, and St. Louis, Missouri, metropolitan statistical areas near the New Madrid Seismic Zone in the Central United States). Most people surveyed thought that (A) new buildings should remain occupiable or functional after a large earthquake and (B) building buyers would be willing to pay as much as an additional $3.00 or more per square foot (ft2) to achieve this preferred outcome. %, percentage of respondents. (Modified from Porter, Not Safe Enough, this volume.) 6   The HayWired Earthquake Scenario—Engineering Implications Chapter M uses data about the age distribution of bay-region buildings along with the history of elevator emergency-power requirements to show that about 25,000 elevators in the bay region lack emergency power to operate briefly after an earthquake, even to travel to the nearest floor and open doors to let passengers escape. It is reasonable to assume that an earthquake like the HayWired mainshock will cause electric power to go out across the region as utility operators act to protect generators and transmission stability, causing many buildings to lose power before shaking reaches them. Elevator earthquake-safety devices (seismic switches or so-called ring-and-string devices) will therefore not be triggered. As a result, it seems possible that more than 22,000 people could be trapped in approximately 4,600 stalled elevators, requiring USAR personnel—generally firefighters—to free them (fig. 4). The job of extricating more than 22,000 people from 4,600 stalled elevators, along with about 2,400 people from 5,000 collapsed buildings, would fall to the approximately 19,000 firefighters who work in the San Francisco Bay region, at the same time as those firefighters are fighting fires. Water-Network Resilience Many authors have studied the effects of earthquakes on water-supply systems. Chapter N (Porter, Water-Network Resilience, this volume) offers a new water-network resilience model with a unique combination of features. It deals with lifeline interaction by directly modeling how individual repairs are Figure 4.  Photograph of firefighters practicing rescue techniques to free people trapped in stalled elevators. Firefighters tend to provide the vast majority of urban search and rescue expertise. After an earthquake like the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario on the Hayward Fault in California’s San Francisco Bay region, they would be called on to extricate thousands of people trapped in stalled elevators and collapsed buildings at a time when they are also called on to fight fires. (U.S. Air Force photograph by Senior Airman Preston Webb; see Porter, Earthquake Urban Search and Rescue Model, this volume.) slowed by limitations in other lifelines and by human and other resource limitations. It quantifies damage and restoration over an entire earthquake sequence—that is, considering damage in the mainshock, aftershocks, and afterslip. It offers an empirical model of water-service restoration as a function of the number of pipeline repairs performed (as opposed to more rigorous, but computationally demanding, hydraulic analysis). It can be used by water-agency staff with a spreadsheet and geographic information system (GIS), rather than requiring a consultant with proprietary software. It can be implemented either deterministically or stochastically, meaning that a sophisticated user can quantify uncertainty but is not required to do so. It does not require hydraulic analysis of the damaged water-supply system or the system as repairs proceed, although that simplification necessarily limits insight into how water pressure would vary throughout the system. The model offers a procedure to adjust Hazus-MH estimates of restoration to account for an earthquake sequence and lifeline interaction and corrects for Hazus’ default assumptions about the number of available repair crews during a disaster. The new water-network resilience model is used to estimate damage and restoration to counties served by two water networks in the San Francisco Bay region for the HayWired scenario earthquake sequence (fig. 5)—those of the San Jose Water Company (SJWC) and of the East Bay Municipal Utility District (EBMUD). The more seriously damaged of the two networks would likely be EBMUD because of its proximity to the Hayward Fault. In the scenario, EBMUD’s 4,162 miles (6,698 km) of pipe suffer about 1,800 breaks and 3,900 leaks during the earthquake sequence, equivalent to 1.4 repairs per mile of pipe (about 0.85 repairs per kilometer of pipe). More than half of water-pipeline damage results directly from ground shaking (60 percent); the remaining damage occurs from liquefaction (29 percent), landslides (3 percent), and coseismic slip (4 percent) and afterslip on the fault (4 percent). In the HayWired scenario, the average EBMUD customer would be without water for 6 weeks, some for as many as 6 months. EBMUD customers suffer a total of 19 million lost service days (each day of lost water supply to a service connection). That loss can be reduced by half if current efforts to replace old, brittle pipe are completed before the next large bay-region earthquake occurs, because such pipe is more susceptible to earthquake damage, and replacing it would reduce damage and therefore restoration time. Also, about 200,000 lost service days could be saved by decreasing or eliminating EBMUD’s dependence on commercial fuel supplies. In the scenario, SJWC suffers less damage—1,000 pipe repairs, of which about 70 percent are water leaks, the rest breaks—and customers would suffer approximately 1 million lost service-days. A utility can reduce its reliance on commercial fuel supplies in a disaster by installing fuel-storage tanks in its service centers or by otherwise ensuring that repair crews have access to fuel. Implementing such a fuel plan would reduce SJWC’s losses by about 25 percent. If SJWC completes replacement of all brittle cast-iron and asbestos-cement pipe (about 25 years at current replacement rates) before an earthquake occurs, losses would be reduced by about half. Chapter I. Overview of the HayWired Scenario Engineering-Implications Volume  7 Repair Costs and Downtime of High-Rise Buildings New earthquake-engineering procedures have emerged since the ShakeOut scenario (Jones and others, 2008) that allow one to estimate repair costs and duration of loss of function for individual, particular buildings. High-rise buildings like the one shown in figure 6 are a particular concern for the engineering community, because a single building can house more than 1,000 people and dozens of businesses, making such buildings potential threats to lives and livelihoods in the San Francisco Bay region should they be damaged in an earthquake. In chapter O (Almufti and others, this volume), second-generation, performance-based earthquake-engineering procedures are applied to 10 example ent Curritions cond Faster restoration of water service Prolonged outages buildings. These examples show that an earthquake like the Mw 7.0 mainshock of the HayWired scenario could damage a pre-1994 (Northridge earthquake) welded-steel moment-frame office tower in downtown Oakland or San Francisco such that it would require 6 to 13 months to reoccupy and cost 7 to 21 percent of its replacement cost to repair, mostly because of nonstructural damage. Even a modern reinforced-concrete high-rise residential building could require 4 to 7 months of repairs before it could be reoccupied and cost 3 to 6 percent of its replacement cost to repair damaged nonstructural elements. Collapse of such high-rise buildings appears to be unlikely in downtown Oakland and San Francisco in an earthquake like the HayWired mainshock. However, this is not to say that other kinds of high-rise buildings could not collapse in these or other locations in the bay region. pipe Withcementn a repl fuel pla and Marin County 95 percent of service returned in 7 days Full service returned in 30 days Marin County Full service returned in 7 days Santa Clara County 73 percent of service returned in 7 days Full service returned in 30 days Santa Clara County 85 percent of service returned in 7 days Full service returned in 30 days San Francisco County 67 percent of service returned in 7 days Full service returned in 30 days San Francisco County Full service returned in 7 days San Mateo County 35 percent of service returned in 7 days Full service returned in 90 days San Mateo County 41 percent of service returned in 7 days Full service returned in 90 days Alameda and Contra Costa Counties 30 percent of service returned in 7 days Full service returned in 210 days Alameda and Contra Costa Counties 47 percent of service returned in 7 days Full service returned in 210 days Figure 5.  This illustration shows water-service restoration times for counties in California’s San Francisco Bay region following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The image at left shows current conditions, and the image at right shows how water-service restoration times could be substantially improved if water utilities replaced all brittle pipe in their systems and had a fuel-management plan and emergency generators with fuel at all pumping stations. (From Hudnut and others, 2018, https://doi.org/10.3133/fs20183016, using data in Porter, Water-Network Resilience, this volume.) Figure 6.  Photograph of a San Francisco Bay area, California, high-rise building. High-rise buildings account for roughly 3 percent of the total building square footage in the bay area, and each of these buildings can house more than 1,000 people and dozens of businesses. However, total collapse of such a building appears to be unlikely in downtown Oakland or San Francisco in an earthquake like the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario on the Hayward Fault. (Photograph by Ken Lund, Creative Commons 3.0, https://www.flickr.com/photos/ kenlund/10753946294; see Almufti and others, this volume.) 8   The HayWired Earthquake Scenario—Engineering Implications Fire Following Earthquake In chapter P (Scawthorn, this volume), a standard model of fire following earthquake is applied to the HayWired scenario. The HayWired mainshock could produce about 670 ignitions requiring the response of a fire engine, 90 percent of which would occur in Alameda, Contra Costa, and Santa Clara Counties. Approximately 450 of these fires would not be immediately contained such that large fires would be likely to merge into numerous conflagrations destroying tens of city blocks. The fires would burn approximately 79 million square feet (about 7.3 million square meters) of building floor area, equivalent to more than 52,000 single-family dwellings. These fires would kill hundreds of people and cause property (building and content) losses approaching $30 billion. The property losses are almost fully insured. Fire following earthquake in the HayWired scenario would produce one of the largest single-loss events in the history of the insurance industry. Other potential economic impacts from fire following earthquake include the loss of perhaps $1 billion in local tax revenues. A number of opportunities exist for mitigating this problem, including greatly enhancing the postearthquake supply of water for firefighting; the mandatory use of automated gas shut-off valves, or seismic shut-off meters, in densely built areas; and the use of portable water-supply systems. Toward that end, the first joint exercise of fire agencies in the San Francisco Bay area to practice the use of portable water-supply systems was conducted as part of HayWired (fig. 7). 123° 122° San Pablo Bay 38° C PA IF C I OC EA N San co cis an Fr EXPLANATION 37.5° y Ba Fire losses, in millions of dollars 0 1–10 10–50 50–100 100–500 500+ HayWired rupture Area of map CALIF Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. 0 0 10 20 10 30 40 KILOMETERS 20 MILES Figure 7.  This map of California’s San Francisco Bay region shows areas burned as a result of fires caused by the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario on the Hayward Fault. Warmer colors show areas with greater losses in millions of dollars. Areas (polygons) shown are based on distance to the closest fire station. The black line shows the length of the fault rupture in the scenario. (From Scawthorn, this volume.) Chapter I. Overview of the HayWired Scenario Engineering-Implications Volume  9 Benefit of Combining Earthquake Early Warning with Drop, Cover, and Hold On Since the advent of ShakeOut exercises following the publication of The ShakeOut Scenario (Jones and others, 2008), every year millions of people worldwide practice drop, cover, and hold on (DCHO) earthquake self-protective actions (https:// www.shakeout.org/). During the same period, earthquake early warning (EEW) systems have become available (for example, the Android phone and iPhone app Yurekuru Call in Japan) to warn people about earthquakes in the seconds before strong motion arrives. Chapter Q (Porter and Jones, this volume) offers a new study of the potential benefits of combining EEW and DCHO. It expresses those benefits in terms of avoided injuries and the acceptable cost to avoid those injuries. A survey was made of 500 people who first took DCHO training and then reported how long it took them to complete the actions. Using these new data and estimates of the advanced warning (as much as 25 seconds) that EEW would provide San Francisco Bay region residents in the event of an earthquake like the HayWired mainshock, it is shown that combining EEW and DCHO could prevent as many as 1,500 nonfatal injuries out of 18,000 estimated nonfatal injuries (from shaking and liquefaction hazards) in the HayWired scenario (fig. 8). Using U.S. Government standard figures for the acceptable cost to avoid future statistical injuries, the combination of EEW and DCHO in a powerful bay region earthquake would be a benefit valued at approximately $300 million. 123° 122° 121° A PACIFIC OCEAN 38° Oakland SAN San Francisco FR AN CI SC O BA Y San Jose 37° Google Earth Data LDEO-Columbia, NSF, NOAA Image Landsat/Copernicus Data SIO, NOAA, U.S. Navy, NGA, GEBCO Data MBARI 0 Area of map CALIF 0 10 20 10 30 40 20 50 KILOMETERS 30 MILES Figure 8.  In the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario on the Hayward Fault in California’s San Francisco Bay region, combining earthquake early warning (EEW) and drop, cover, and hold on (DCHO) practices could result in as many as 1,500 nonfatal injuries being prevented out of 18,000 estimated nonfatal injuries, a benefit valued at about $300 million. A, Satellite image showing EEW times in seconds (s) for the HayWired mainshock. The red line, shows the length of the fault rupture in the scenario, and the epicenter (star) is beneath the City of Oakland. (Satellite image from Google Earth.) B, Illustration showing how to DCHO (courtesy of ShakeOut.org). 10   The HayWired Earthquake Scenario—Engineering Implications B Figure 8.—Continued Future Research The developers of the HayWired earthquake scenario did not intend or expect to examine every engineering-related issue (or other important topic) arising from a large Hayward Fault earthquake, partly because of all the previous work on the subject. However, several outstanding engineering issues remain unexplored and would be worth examining. A few of these include: 1. The cost of performing Applied Technology Council (ATC) ATC-20 post-earthquake safety evaluations.—The 1994 Northridge earthquake resulted in ATC-20 earthquake safety-inspection (Applied Technology Council, 2005) of 100,000 buildings. The HayWired mainshock is a larger earthquake than the one in Northridge and takes place in a more densely populated urban area, suggesting that many more ATC-20 evaluations would be required. California has approximately 6,000 certified evaluators, another 4,000 live elsewhere in the United States (Jim Barnes, California Governor’s Office of Emergency Services, oral commun., September 29, 2017). Most have other day jobs. They would only be briefly available to volunteer to perform ATC-20 evaluations. It might take weeks or more to complete the required ATC-20 evaluations. During that time, many occupants would be displaced from their buildings while awaiting evaluations. What would be the cost of delayed ATC-20 evaluations, and how much time and economic value might be saved through automation, such as using FEMA’s Rapid Observation of Vulnerability and Estimation of Risk (ROVER) software (Federal Emergency Management Agency, 2013)? 2. Effectiveness of DCHO to avoid injuries during shaking.—People believe DCHO to be effective, but how effective, and how can we be sure? Nobody has ever tested the effectiveness of DCHO self-protective actions. The question matters for several reasons, some of which include judging the costs versus benefits of further reducing nonstructural hazards, and estimating needs for emergency medical care. There is very little informa- tion to answer any of these questions, partly because no research program seems to currently focus on understanding human injuries in earthquakes. 3. Cost effectiveness of seismic gas-shutoff valves.—It seems intuitive that seismic gas-shutoff valves would reduce the risk of fires, but they introduce costs—upfront construction costs and costs of reopening the valves after an earthquake. Under what conditions are the costs justified? The answer is not self-evident. 4. Modeling electric-utility-service damage and restoration.—Electric-service impairment and subsequent restoration depends on electric network stability as much as it does on physical damage to generation facilities, transmission and distribution lines, substations, and neighborhood and pole-mounted transformers. It would be desirable to construct a model akin to the water-network resilience model introduced here, which does not require proprietary software. Such a model of electricutility resilience would need to be able to estimate damage and restoration time, while accounting for lifeline interaction, human and other resource limitations, and the engineering characteristics of the various components just listed, without requiring sensitive data from the electric-service provider. Conclusion This chapter and the volume it summarizes describe realistic engineering implications of a large earthquake on what may be the most urbanized active fault in the United States—the Hayward Fault. It is not intended to represent a best case, a worst case, or an average case, merely one that is worth planning for. It shows how such an earthquake could cause more than $82 billion in property loss and direct business interruption losses over the modeled earthquake sequence, plus another $30 billion in property (building and content) losses resulting from fires after the mainshock. Chapter I. Overview of the HayWired Scenario Engineering-Implications Volume  11 Because it aims at depth rather than breadth, this volume mostly omits a number of important topics that have been addressed well elsewhere; for example: • What to do about high-risk existing buildings is not addressed. Interested readers can refer to the San Francisco Community Action Plan for Seismic Safety (CAPSS) and Earthquake Safety Improvement Program (ESIP) (see http://sfgov.org/esip/program/) or to the City of Santa Monica’s comprehensive seismic retrofit program (see https://www.smgov.net/Departments/PCD/Programs/Seismic-Retrofit/) for valuable guidance on that topic. • Very little is discussed about dealing with nonstructural building components. The interested reader can see FEMA’s E-74 document (Federal Emergency Management Agency, 2012b). • There is little discussion about electricity and gas. The interested reader can refer to Pacific Gas and Electric Company’s web pages on residential and business earthquake preparedness (https://www.pge.com/en_ US/safety/emergency-preparedness/natural-disaster/ earthquakes/earthquakes.page). The HayWired scenario’s developers hope that the information here and in other HayWired volumes will inform the reader’s decisions about how to prepare for a large earthquake, whether by strengthening infrastructure to better resist earthquakes or through improved planning to recover more quickly despite damage. Ideally, the HayWired scenario volumes will help readers collectively improve their own and their community’s resilience in future disasters. Work that straddles the boundary between engineering and social and economic consequences is planned to be described in a third HayWired volume. Beyond that, a group of government and other organizations called the HayWired Coalition (see chapter A, Hudnut and others, 2017; and Hudnut and others, 2018) intends to build on the work presented here through a series of planning and preparedness activities. As HayWired volumes are published, they will be made available at https://doi.org/10.3133/sir20175013. Detweiler, S.T., and Wein, A.M., eds., 2017, The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., accessed September 13, 2017, at https://doi.org/10.3133/ sir20175013v1. Federal Emergency Management Agency, 2012a, Hazus multihazard loss estimation methodology, earthquake model, Hazus®-MH 2.1 technical manual: Federal Emergency Management Agency, Mitigation Division, 718 p., accessed September 13, 2017, at https://www.fema.gov/media-librarydata/20130726-1820-25045-6286/hzmh2_1_eq_tm.pdf. Federal Emergency Management Agency, 2012b, Reducing the risks of nonstructural earthquake damage—A practical guide: Federal Emergency Management Agency, FEMA E-74, 885 p., accessed September 13, 2017, at https://www.fema.gov/fema-e74-reducing-risks-nonstructural-earthquake-damage. Federal Emergency Management Agency, 2013, ROVER, end-to-end mobile software for managing seismic risk: Federal Emergency Management Agency flyer, 1 p., accessed September 13, 2017, at https://www.fema.gov/media-library/ assets/documents/21350. Hudnut, K.W., Wein, A.M., Cox, D.A., Perry, S.C., Porter, K.A, Johnson, L.A., and Strauss, J.A., 2017, The HayWired scenario—How can the San Francisco Bay region bounce back from or avert an earthquake disaster in an interconnected world?, chap. A of Detweiler, S.T., and Wein, A.M., eds., The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017– 5013–A–H, 126 p., accessed September 13, 2017, at https://doi. org/10.3133/sir20175013v1. Hudnut, K.W., Wein, A.M., Cox, D.A., Porter, K.A., Johnson, L.A., Perry, S.C., Bruce, J.L., and LaPointe, D., 2018, The HayWired earthquake scenario—We can outsmart disaster: U.S. Geological Survey Fact Sheet 2018–3016, 6 p., https://doi. org/10.3133/fs20183016. References Cited Jones, L.M., Bernknopf, R., Cox, D., Goltz, J., Hudnut, K., Mileti, D., Perry, S., Ponti, D., Porter, K., Reichle, M., Seligson, H., Shoaf, K., Treiman, J., and Wein, A., 2008, The ShakeOut scenario: U.S. Geological Survey Open-File Report 2008–1150 and California Geological Survey Preliminary Report 25, 312 p. and appendixes, accessed September 13, 2017, at https://pubs. usgs.gov/of/2008/1150/. Applied Technology Council, 2005, ATC-20-1—Field handbook—Procedures for postearthquake safety evaluation of buildings: Redwood City, Calif., Applied Technology Council, 144 p. Schiff, A., 2008, The ShakeOut scenario supplemental study—Elevators: Denver, Colo., SPA Risk LLC, accessed September 13, 2017, at http://www.sparisk.com/pubs/ ShakeOutScenarioElevatorsSchiff.pdf. The HayWired Earthquake Scenario—Earthquake Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q [Also see https://doi.org/10.3133/sir20175013] Chapter J HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks By Hope A. Seligson,1 Anne M. Wein,2 and Jamie L. Jones2 Abstract The HayWired scenario examines a hypothetical earthquake (mainshock) with a moment magnitude (Mw) of 7.0 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of California’s San Francisco Bay area. Analyses of building damages and direct economic losses from the HayWired mainshock and its aftershocks have been conducted using Hazus, the Federal Emergency Management Agency’s (FEMA) multi-hazard geographic information system-based loss estimation methodology and software. The initial Hazus analysis of the HayWired scenario mainshock (Aagaard and others, 2017)—the baseline default run—was conducted by FEMA using Hazus-MH 2.1 with improved inventory data originally developed for the 100th Anniversary Earthquake Conference commemorating the 1906 San Francisco earthquake. In this study, the mainshock was further analyzed using custom HayWired landslide probability and displacement estimates and custom HayWired liquefaction probability estimates. The baseline default Hazus shaking and liquefaction run was also revised to use the same groundwater depth data as used in the U.S. Geological Survey (USGS) analysis of liquefaction probability. The aftershock sequence (a catalog of earthquakes each with a date/time, magnitude, location, and depth) was simulated by the USGS using aftershock statistics. Hazus analyses of 16 aftershock events of magnitude 5 or greater were performed using the same improved inventory databases as the initial analysis of the mainshock. This chapter (1) describes the results of the initial and revised Hazus analyses of the mainshock (considering shaking, liquefaction, and landslide hazards) and aftershocks (considering shaking and Hazus default liquefaction in selected events), (2) assesses the potential impact of liquefaction and repeated liquefaction, Seligson Consulting, initial efforts conducted while with MMI Engineering. 1 U.S. Geological Survey. 2 (3) reviews expected performance of unreinforced masonry construction in the various events, (4) compares population displacement and shelter estimates using default parameters and custom parameters developed for the Southern California ShakeOut scenario, (5) compares various approaches for combining losses in the mainshock and aftershocks, and (6) identifies knowledge gaps and study limitations. The results of the mainshock assessment yield building losses of $35.2 billion (in 2005 dollars, estimated as $43.3 billion in 2016 dollars): $30.3 billion (in 2005 dollars, $37.3 billion in 2016 dollars) in damage from shaking, $4.6 billion (in 2005 dollars, $5.7 billion in 2016 dollars) in damage from liquefaction, and $300 million (in 2005 dollars, $360 million in 2016 dollars) in damage from landslide. Over the entire earthquake sequence, the total direct economic loss can be approximated as $67.0 billion (in 2005 dollars, $82.6 billion in 2016 dollars). This includes the following: • $43.3 billion (in 2005 dollars, $53.3 billion in 2016 dollars) in building damage for the mainshock and all aftershocks, using USGS-modeled liquefaction and landslide hazard and probability data for the mainshock, where available, and the Hazus default liquefaction modeling approach for other areas in the mainshock as well as for three aftershocks (M5.98 Mountain View, M6.4 Cupertino, and M5.42 Oakland). The remainder of aftershocks were modeled for shaking only. • $13.8 billion (in 2005 dollars, $17.0 billion in 2016 dollars) in damage to contents and commercial inventories, estimated using the Hazus default liquefaction modeling approach for the mainshock and the three aftershocks identified above, and considering only ground-shaking hazards for the remainder of events. • $10.0 billion (in 2005 dollars, $12.3 billion in 2016 dollars) in building damage-related income losses (for example, relocation costs, lost rent, and so on), modeled in the same manner as contents and inventory damage. 14   The HayWired Earthquake Scenario—Engineering Implications About 80 percent of the losses from the earthquake sequence are from the Mw 7.0 mainshock, 12 percent from the three largest aftershocks of Mw 6.0 to 6.4, and 8 percent from the 13 aftershocks of Mw 5.0 to 5.9. Displaced household assessments range from tens of households for the smallest aftershocks, to hundreds of households for the largest aftershocks, to tens of thousands of households (77,000– 153,000) for the mainshock (based on the 2000 U.S. Census). Introduction The HayWired scenario examines a hypothetical earthquake (mainshock) with a moment magnitude (Mw) of 7.0 occurring on April 18, 2018, at 4:18 p.m. on the in the east bay part of California’s San Francisco Bay area. The rupture of the Hayward Fault starts under the city of Oakland, with fault slip progressing north into San Pablo Bay and south to the city of Fremont (Aagaard and others, 2017). The mainshock shaking data (U.S. Geological Survey [USGS], 2014), simulated using a three-dimensional model (Aagaard and others, 2017), were used directly in the Federal Emergency Management Agency’s (FEMA) damage and loss modeling tool, Hazus (FEMA, 2012). The shaking data for the HayWired mainshock were also used to derive data for (1) liquefaction probability (Jones and others, 2017) and (2) landslide probability and displacement (McCrink and Perez, 2017). The HayWired aftershock sequence has thousands of aftershocks and is described in Wein and others (2017). The larger aftershocks (magnitude [M]>5) in the sequence simulated for the Mw 7.0 HayWired mainshock are identified in table 1. The sequence includes two M>6 events (one in Palo Alto and one in Cupertino) and spans 2 years. The locations of the 16 aftershocks are shown in figure 1. The regional ground-shaking data for these aftershocks (USGS, 2015) were also used directly in Hazus. Building Inventory Data The initial Hazus analysis conducted for the HayWired scenario mainshock was executed using Hazus-MH 2.1 by FEMA personnel (Doug Bausch of FEMA Region VIII) using the custom ground motion data for the HayWired scenario mainshock (USGS, 2014) and enhanced Hazus building inventory data originally developed in 2005–2006 as part of a modeling effort for the 100th Anniversary Earthquake Conference Commemorating the 1906 San Francisco Earthquake (Kircher and others, 2006). The Hazus database improvements include retrofitted building models for unreinforced masonry (URM), non-ductile concrete, and softstory wood frame1 structures; replacement cost escalation factors to better reflect actual building inventory values; and enhanced mapping schemes (relationships by building occupancy class that describe the percentage distribution of square footage among various structural types or model building types (MBTs), as referred to within Hazus). A total of 22 custom mapping schemes were applied at the census-tract level across the bay area, reflecting differing building age and height patterns, as well as building density; a separate set of schemes were applied to San Francisco and Alameda Counties to reflect urban core 1 Within Hazus, non-ductile concrete is generally represented by “pre-code” design level concrete moment frame (model building type C1) and concrete frame with URM infill wall (C3) buildings. Pre-code precast concrete frame buildings with concrete shear walls (PC2) may also be considered non-ductile. Soft-story wood frame is similarly represented by pre-code wood frame (W1 and W2) buildings. Table 1.  Sequence of aftershocks of magnitude 5 or greater of the HayWired earthquake scenario, San Francisco Bay region, California. [Data from U.S. Geological Survey (2014). Date format is month/day/year. Day is relative to the day the HayWired mainshock event occurred, with April 18, 2018, counted as day 1. Latitude is in decimal degrees north; longitude is in decimal degrees west. Depth is how far below the Earth’s surface the aftershock hypocenter is located. No., event number; PDT, Pacific Daylight Time; km, kilometers; Mag., magnitude; Short name, short event name] No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Date 4/18/2018 4/19/2018 4/29/2018 5/02/2018 5/20/2018 5/28/2018 5/28/2018 5/28/2018 5/28/2018 6/23/2018 7/01/2018 9/30/2018 10/01/2018 10/01/2018 10/01/2018 8/22/2019 Day 1 2 12 15 33 41 41 41 41 67 75 166 167 167 167 492 Time (PDT) 4:49 p.m. 4:16 a.m. 11:13 p.m. 8:44 p.m. 8:37 a.m. 4:47 a.m. 8:11 a.m. 6:22 p.m. 11:53 p.m. 8:27 p.m. 11:19 a.m. 8:16 p.m. 12:33 a.m. 2:24 a.m. 6:10 a.m. 10:45 p.m. Latitude 37.6008 37.9630 38.1916 37.4829 37.7561 37.3867 37.4528 37.4604 37.4099 37.4391 37.4435 37.4386 37.3068 37.3835 37.3334 37.4145 Longitude 122.0172 122.3473 122.1483 121.9146 122.1508 122.1780 122.1671 122.1753 122.1184 122.1511 122.1561 122.0770 122.0592 122.0153 121.9541 122.1235 Location Union City San Pablo Fairfield Fremont Oakland Palo Alto Menlo Park Atherton Palo Alto Palo Alto Palo Alto Mountain View Cupertino Sunnyvale Santa Clara Palo Alto Depth (km) 2.60 2.60 11.05 7.15 8.45 15.97 7.26 7.91 8.36 2.85 8.69 11.29 14.45 18.89 7.00 11.98 Mag. 5.23 5.04 5.58 5.10 5.42 6.20 5.52 5.11 5.69 5.22 5.26 5.98 6.40 5.35 5.09 5.01 Short name UC523 SP504 FF558 FR510 OK542 PA62 MP552 AT511 PA569 PA522 PA526 MV598 CU64 SV535 SC509 PA501 Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  15 122.5° 122° NAPA SOLANO SONOMA SOLANO SAN PABLO BAY MARIN 38° CONTRA COSTA SAN FRANCISCO N SA O SC CI AN FR ALAMEDA Y BA EXPLANATION County boundary Hayward Fault Fault 37.5° SAN MATEO Mainshock and aftershock epicenters Magnitude 5.0–5.5 5.5–6.0 SANTA CLARA 6.0–6.5 6.5–7.0 Coordinate System: GCS North American 1983 Datum: North American 1983 0 Area of map CALIF 0 5 5 10 10 15 15 20 MILES 20 KILOMETERS Figure 1.  Map showing epicenters of the mainshock (largest red star) and aftershocks of magnitude 5 or greater of the hypothetical HayWired earthquake scenario in the San Francisco Bay region, California. areas with concentrations of mid-rise and high-rise construction. These schemes are considered a substantial improvement over the Hazus default mapping schemes, wherein all construction is assumed to be low-rise. Hazus analyses using the custom HayWired liquefaction and landslide data, as well as the aftershock shaking data, were done using the same improved building inventory as that for the initial mainshock Hazus run. Summaries of the improved building inventory data are provided in tables 2, 3, and 4, by county name, by occupancy, and by structural type, respectively. Note that totals may not match exactly as a result of rounding. Exposure values (and resulting estimated losses) for the Hazus custom bay area inventory data are in 2005 dollars. These values can be approximately escalated to 2016 dollars using Consumer Price Index (CPI) ratios (see https://www.bls.gov/cpi/ data.htm); the U.S. CPI 2016:2005 ratio is approximately 1.23. 16   The HayWired Earthquake Scenario—Engineering Implications Table 2.  Improved building inventory data by county used for the Hazus analyses of the HayWired earthquake scenario mainshock and aftershock sequence, San Francisco Bay region, California. [Exposure value reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23] County Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Sonoma Stanislaus Yolo  Total Building count 413,505 321,281 93,195 56,678 109,838 45,053 378,791 16,279 172,931 159,215 219,815 495,282 90,140 120,823 169,235 130,688 46,049 3,038,798 Building square footage (thousands of square feet) 1,134,537 717,509 237,269 117,942 271,611 111,729 890,201 32,842 671,672 358,055 557,525 1,263,479 208,512 265,812 385,085 287,260 120,019 7,631,059 Total exposure value (thousands of dollars) 155,699,818 102,806,780 36,050,257 12,901,176 33,772,799 14,579,197 110,561,701 4,135,710 100,178,548 42,755,589 84,301,336 183,312,185 28,382,925 34,820,221 50,857,518 33,827,997 14,478,828 1,043,422,585 Table 3.  Improved building inventory data by occupancy class used for the Hazus analyses of the HayWired earthquake scenario mainshock and aftershock sequence. [Exposure value reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23] Hazus occupancy class RES1 (single-family homes) RES2 (manufactured housing) RES3A (multifamily residential: duplex) RES3B (multifamily residential: triplex/quad) RES3C (multifamily residential: 5–9 units) RES3D (multifamily residential: 10–19 units) RES3E (multifamily residential: 20–49 units) RES3F (multifamily residential: 50+ units) RES4 (hotel/motel) RES5 (institutional dormitories) RES6 (nursing homes) COM1 (retail trade) COM2 (wholesale trade) COM3 (personal/repair services) COM4 (offices) COM5 (banks) COM6 (hospitals) COM7 (medical offices/clinics) COM8 (entertainment/recreation) COM9 (theaters) COM10 (parking garages) Building count 2,702,528 108,696 60,464 52,642 16,195 7,588 1,447 1,781 57 4,157 184 1,569 6,379 15,791 4,125 5,524 420 9,710 23,845 231 0 Building square footage (thousands of square feet) 4,324,156 117,362 193,083 171,161 163,087 127,115 118,537 189,909 22,169 116,399 5,184 396,411 227,375 174,448 473,123 22,853 25,336 72,953 120,160 3,860 0 Total exposure value (thousands of dollars) 610,850,583 4,151,673 18,451,037 17,701,795 29,271,495 20,470,095 18,794,663 29,180,378 3,395,429 20,146,620 773,093 38,858,298 20,214,758 21,226,826 70,944,986 4,975,356 6,678,754 13,915,026 24,748,686 559,818 0 Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  17 Table 3.—Continued Hazus occupancy class IND1 (heavy industrial) IND2 (light industrial) IND3 (food/drugs/chemical) IND4 (metals/minerals processing) IND5 (high technology) IND6 (construction) AGR1 (agriculture) REL1 (churches) GOV1 (government/general services) GOV2 (government/emergency response) EDU1 (education/grade schools) EDU2 (education/colleges and universities)  Total Building count 3,154 2,992 903 99 628 1,478 611 2,654 2,326 231 44 345 3,038,798 Building square footage (thousands of square feet) 104,632 107,895 49,298 6,115 30,336 82,533 34,953 64,015 26,109 2,826 38,997 18,668 7,631,058 Total exposure value (thousands of dollars) 10,793,540 9,631,660 8,382,266 1,035,700 5,166,822 7,333,744 3,068,518 10,457,728 3,279,910 547,978 5,277,262 3,138,088 1,043,422,585 Table 4.  Improved building inventory data by model building type used for the Hazus analyses of the HayWired earthquake scenario mainshock and aftershock sequence. [Exposure value reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23. ft2, square feet; CIP, cast in place; PC, precast; RM, reinforced masonry; URM, unreinforced masonry; w/, with; conc., concrete] Hazus model building type W1 (wood, light frame, ≤5,000 ft2) W2 (wood, commercial and industrial, >5,000 ft2) S1L (steel moment frame, low-rise) S1M (steel moment frame, mid-rise) S1H (steel moment frame, high-rise) S2L (steel braced frame, low-rise) S2M (steel braced frame, mid-rise) S2H (steel braced frame, high-rise) S3 (steel light frame) S4L (steel frame w/ CIP concrete shear walls, low-rise) S4M (steel frame w/ CIP Concrete shear walls, mid-rise) S4H (steel frame w/ CIP concrete shear walls, high-rise) S5L (steel frame w/ URM infill walls, low-rise) C1L (concrete moment frame, low-rise) C1M (concrete moment frame, mid-rise) C1H (concrete moment frame, high-rise) C2L (concrete shear wall, low-rise) C2M (concrete shear wall, mid-rise) C2H (concrete shear wall, high-rise) C3L (concrete frame w/ URM infill walls, low-rise) C3M (concrete frame w/ URM infill walls, mid-rise) C3H (concrete frame w/ URM infill walls, high-rise) PC1 (PC concrete tilt-up walls) PC2L (PC concrete frames w/ conc. shear walls, low-rise) PC2M (PC concrete frames w/ conc. shear walls, mid-rise) PC2H (PC concrete frames w/ conc. shear walls, high-rise) Building count 2,734,372 22,888 7,491 4,747 9,109 4,900 3,842 2,565 3,779 2,529 2,936 1,352 4,796 755 3,031 2,865 13,796 9,991 5,087 3,424 3,466 1,890 5,370 1,060 622 757 Building square footage (thousands of square feet) 4,540,623 602,193 123,223 68,584 109,743 90,922 52,920 33,002 76,168 46,719 33,876 22,107 89,765 25,034 33,257 30,933 270,262 127,918 49,514 46,946 53,021 18,651 182,812 30,568 7,900 6,612 Total exposure value (thousands of dollars) 634,173,432 87,505,195 16,882,881 9,587,560 16,598,964 11,632,242 7,014,666 5,042,406 8,909,731 6,328,477 5,096,293 3,442,221 11,334,447 3,225,732 5,193,239 4,649,490 36,497,177 17,573,591 7,368,727 6,464,956 6,354,659 2,799,477 20,193,977 3,620,206 1,022,077 963,776 18   The HayWired Earthquake Scenario—Engineering Implications Table 4.—Continued Hazus model building type Building count RM1L (RM bearing walls w/ wood or metal deck diaphragms, low-rise) RM1M (RM bearing walls w/ wood or metal deck diaphragms, mid-rise) RM2L (RM bearing walls w/ PC conc. diaphragms, low-rise) RM2M (RM bearing walls w/ PC conc. diaphragms, mid-rise) RM2H (RM bearing walls w/ PC conc. diaphragms, high-rise) URML (URM bearing walls, low-rise) URMM (URM bearing walls, mid-rise) MH (mobile homes)  Total 46,314 Building square footage (thousands of square feet) 396,562 Total exposure value (thousands of dollars) 54,177,952 5,094 72,129 9,448,147 1,747 1,255 382 10,965 3,919 111,702 3,038,798 39,094 19,765 3,772 148,408 31,261 146,792 7,631,056 5,205,624 2,592,305 559,241 19,899,424 4,262,510 7,801,366 1,043,422,168 Hazus Results—HayWired Mainshock The initial HayWired scenario mainshock was analyzed using regional liquefaction susceptibility data developed by the USGS and others (Knudsen and others, 2000; as shown in fig. 2), in conjunction with the Hazus default approach for computing liquefaction displacements and probabilities of occurrence and the default assumption of uniform shallow groundwater conditions, to produce a baseline “default” Hazus run. As part of the HayWired earthquake scenario hazard studies, more detailed liquefaction probability calculations were performed for the mainshock in Alameda and Santa Clara Counties (Jones and others, 2017). Likewise, more detailed landslide probability and displacement calculations were performed for the mainshock for the bay area (McCrink and Perez, 2017). As a result of the focused liquefaction study, the baseline Hazus default run was modified to reflect a dual groundwater depth assumption (16 feet in western Santa Clara County, 5 feet elsewhere) consistent with the more detailed study. Implementing the HayWired mainshock liquefaction and landslide data in Hazus required unconventional implementations of Hazus. Descriptions of these implementations precede the discussion of HayWired mainshock loss results in the following sections. HayWired Liquefaction Implementation in Hazus Liquefaction probabilities were derived for most of Santa Clara and Alameda counties for the HayWired mainshock based on a method by Holzer and others (2008, 2010, 2011) (see Jones and others, 2017). The liquefaction probabilities were provided for 50-meter (m) pixels throughout the coverage area. The implementation of these data in Hazus required creating census tract-based liquefaction probability data. National Land Cover Data (NLCD; Homer and others, 2015) were used to identify pixels located in developed areas (identified as low-, medium- or high-intensity development), as shown in figure 3. Census tract-level liquefaction probabilities were then calculated as the average across pixels with nonzero liquefaction probability in developed areas such that (1) liquefaction probabilities in undeveloped areas would not influence results, and (2) resulting loss estimates for the census tract were assumed uniform in the census tract (consistent with the Hazus methodology) and could then be scaled down in proportion to how much of the developed area had a non-zero probability of liquefaction. Further, because the Holzer and others method does not provide liquefaction displacement values, Hazus-estimated displacements were required. Within Hazus, liquefactioninduced spread and settlement displacement estimates, as well as liquefaction probability estimates, require the input of maps of ground motion, liquefaction susceptibility and groundwater depth. For consistency with the liquefaction probability calculations, the liquefaction susceptibility data input into Hazus was modified from the basic “point-in-polygon” centroidal value for each census tract to a weighted (by probability) average of the liquefaction susceptibility class in non-zero liquefaction probability areas for each census tract. The liquefaction susceptibility map (used in the Hazus displacement calculations) and the Holzer and others method are both based on the same Quaternary geologic map. In addition to the change in liquefaction susceptibility input, the assumed uniform shallow depth to groundwater (5 feet) was modified to better reflect groundwater contour data provided by the California Geological Survey and used in the liquefaction probability assessment (Jones and others, 2017); in western Santa Clara County, groundwater depths of 16 feet have been assumed. Because the custom probability data were only available for part of the study area, Hazus-estimated liquefaction probabilities were required for the remainder of the study region. This necessitated (1) a Hazus run that incorporated the custom (weighted) liquefaction susceptibility to produce appropriate displacements in the custom liquefaction study area, and (2) a second Hazus run incorporating the custom liquefaction probability data (overwriting the Hazus-estimated probabilities directly in Hazus’ SQL tables) and interrupting Hazus’ computational flow to avoid overwriting the various Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  19 124° 123° 122° NAPA IC PACIF SACRAMENTO SOLANO OC MARIN EA N SAN FRANCISCO EXPLANATION SAN JOAQUIN CONTRA COSTA ALAMEDA SAN MATEO NT AC County boundary STANISLAUS SANTA CLARA SA Hazus study region boundary 37° 120° YOLO SONOMA 38° 121° MERCED RU Z Liquefaction susceptibility None/not mapped SAN BENITO Very low Low Moderate MONTEREY High 36° Very high Coordinate System: GCS North American 1983 Datum: North American 1983 0 0 Area of map 20 20 40 40 60 60 80 MILES 80 KILOMETERS CALIF Figure 2.  Map showing liquefaction susceptibility in the San Francisco Bay region, California. Data developed by the U.S. Geological Survey (based on Knudsen and others, 2000). liquefaction hazard data tables. The building damage results produced by this second run were then manually scaled down by the proportion of development exposed to nonzero liquefaction probabilities to arrive at final losses for the developed areas. It should be noted that as a result of conducting part of the analysis outside of Hazus, the availability of Hazus results normally derived from the modified outputs became limited. That is, although we have estimated building damage and associated economic loss using the custom liquefaction probability, dependent impact estimates, such as casualties and shelter, are not available for the customized liquefaction assessment. HayWired Landslide implementation in Hazus Landslide displacement and probability data were provided for the HayWired mainshock for 10-m pixels in nine bay area Counties (McCrink and Perez, 2017): Alameda, Contra Costa, Marin, Napa, San Francisco, San Mateo, Santa Clara, Solano, and Sonoma. The landslide probabilities are derived from the displacements, and are categorized as given in table 5. Because the landslide hazard is so localized, census tractlevel representations did not adequately capture the highly variable and concentrated nature of the hazard. Accordingly, an alternate approach was taken whereby multiple Hazus runs were conducted, assuming various displacement/probability combinations, and the census tract results were weighted according to the pixel level data, outside of Hazus, as described below. Six Hazus runs were conducted, each assuming uniform landslide displacement and probability values for a given category, as shown in table 5. These runs yielded loss results for each midpoint displacement and probability value, by census tract. The Hazus inventory was assumed to be uniformly distributed across each census tract (and contributing pixel) within the developed area, as defined by NLCD (Homer and others, 2015), discussed above. (Developed pixels with landslide displacements in Alameda, Marin and Santa Clara counties are shown in fig. 4.) The number of developed pixels in each displacement/probability range were counted for each census tract. The losses for the pixels in each bin were then obtained by multiplying the census 20   The HayWired Earthquake Scenario—Engineering Implications 122°15' 122° 121°45' EXPLANATION CONTRA COSTA Alameda Oakland County boundary Area where Holzer and others' (2008, 2010) methods are used to map liquefaction probability Liquefaction probability <5 percent 30–40 percent 5–10 percent 40–50 percent 10–20 percent >50 percent 20–30 percent Non-developed land 37°45' N SA Livermore O SC CI AN FR Hayward ALAMEDA Y BA Fremont 37°30' SAN MATEO San Jose 37°15' Area of map CALIF Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. SANTA CLARA 0 0 4 4 8 MILES 8 KILOMETERS Figure 3.  Map showing liquefaction probabilities for the HayWired earthquake scenario mainshock computed using the Holzer and others (2008, 2010) method in developed areas of Alameda and Santa Clara Counties, California (data modified from Jones and others, 2017). Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  21 Table 5.  Landslide displacement categories and associated probability ranges for the HayWired earthquake scenario mainshock. [Data modified from McCrink and Perez, 2017. cm, centimeters; L, low, M, medium, H, high, VH, very high] Displacement range (cm) 0–1 1–5 5–15 15–30 30–100 100+ Displacement range (inches) 0.0–0.4 0.4–2.0 2.0–6.0 6.0–12.0 12.0–39.0 39.0–196 Assumed landslide displacement (inches) 0.2 1.2 4.0 9.0 25.5 117.5 122°15' Landslide probability category L M H VH VH VH Probability range 0.0–0.016 0.016–0.15 0.15–0.323 0.323–0.335 0.335 0.335 122°10' A. Alameda County Assumed probability 0.008 0.08 0.237 0.329 0.335 0.335 122°30' 37°55' B. Marin County Mill Valley 37°50' Belvedere Sausalito 121°50' Oakland 37°25' C. Santa Clara County Alameda San Jose 37°45' Non-developed land from Multi-Resolution Land Characteristics Consortium National Land Cover Dataset, 2015. Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. B CONTRA COSTA Oakland A San Francisco SAN MATEO 0 1 2 MILES ALAMEDA C San Jose 0 1 2 KILOMETERS EXPLANATION Landslide probability on developed land (corresponding displacement, in centimeters [cm]) 0–2 percent (0–1 cm) 2–15 percent (1–5 cm) 15–32 percent (5–15 cm) Major highways >32 percent Secondary roads (>15 cm) Non-developed land Figure 4.  Map showing custom landslide probabilities for the HayWired earthquake scenario mainshock in developed areas of parts of Alameda (A), Marin (B), and Santa Clara Counties (C), California (data modified from McCrink and Perez, 2017). 22   The HayWired Earthquake Scenario—Engineering Implications tract-level Hazus result for the bin by the number of pixels in the bin, divided by the total number of pixels in the developed area of the census tract. The landslide losses were then summed for each census tract. This method removed the influence of higher displacement estimates in undeveloped areas, assuming uniform distribution of building construction (consistent with the Hazus methodology). It should be noted, however, that the landslide data only pertain to landslide initiation, and do not capture damages downstream of landslides in other pixels. As with the custom liquefaction data implementation, because we are conducting part of the landslide analysis outside of Hazus, the availability of Hazus results normally derived from the modified outputs become limited. HayWired Mainshock Hazus Results The final mainshock Hazus analyses may be summarized as follows: • Shaking only • Shaking and default liquefaction, with modified groundwater depth assumption (modified baseline default run) • Shaking, custom liquefaction, and custom landslide (custom runs, combined) A summary of the liquefaction and landslide data available by county is provided in table 6. As shown, no liquefaction or landslide data were available for the seven outlying counties (Merced, Monterey, Sacramento, San Benito, San Joaquin, Stanislaus and Yolo). Figure 5 displays the geographic distribution of hazards ultimately contributing to building damage for each census tract. It should be noted that the custom liquefaction data did not cover all census tracts in Alameda and Santa Clara Counties (see figs. 3, 5); default liquefaction results were used in census tracts not covered by the custom data. Building damage estimates for the three versions of the HayWired scenario mainshock, as described above, are summarized in tables 7, 8, and 9 respectively, by county and Hazus building damage state. It should be noted that the data in tables 7 and 8 are taken directly from Hazus, whereas the data in table 9 are approximately derived from resulting distributions of building square footage by damage state, multiplied by Hazus’ damage state mean percentage loss assumptions and building exposure values. Best estimates of loss from shaking and default liquefaction have been taken directly from Hazus (tables 7, 8); best estimates of loss from the custom liquefaction and landslide analyses are taken from the approximate results derived from the square footage distributions (table 9). (Estimates of shaking and default liquefaction contributing to the values in table 9 may not exactly match those in table 8; differences between estimates taken directly from Hazus and those derived using the approximate square footage damage state distribution average 4 percent at the county level.) Table 10 provides the resultant county-level best estimates of building damage by hazard. Table 6.  Availability of liquefaction and landslide data for counties in the San Francisco Bay region, California. [Y, yes. Shading indicates counties with no liquefaction or landslide data available] County Liquefaction Liquefaction Landslide (default approach) (custom approach) Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Y1 Y Y Sonoma Stanislaus Yolo Y Y Y Y Y Y Y Y Y Y1 Y Y Y Y Y Y Y Y 1 Default liquefaction results were used for census tracts in Alameda and Santa Clara Counties not covered by the custom liquefaction data. As noted above, exposure values and losses estimated using the custom bay area Hazus inventory data are in 2005 dollars. These values may be approximately escalated to 2016 dollars using the net CPI ratio of 1.23. Reported and tabulated dollar values have been stated in the native 2005 dollars to facilitate review and ease comparison. As shown in the tables, the combined best estimate for liquefaction adds more than $4.6 billion to the building damage in this event (see table 10). Most of the losses caused by liquefaction occur in Alameda County ($3.1 billion), San Mateo County ($510 million), and Contra Costa County ($460 million). Figure 6A shows the geographical distribution of Hazus building damage ratio estimates caused by shaking, while figure 6B shows the distribution of building damage ratios for liquefaction, estimated using the Hazus default liquefaction modeling approach and adjusted groundwater depths. The difference between default and custom liquefaction damage in Alameda and Santa Clara Counties reflect the differences in the liquefaction probability and estimated displacements at the census tract level. The default and custom probabilities are compared in Jones and others, 2017. The use of custom liquefaction probability data modestly increases liquefaction building damage estimates in Alameda County, impacting a broader area than the default approach, but substantially decreases damage estimates in Santa Clara County. The lower custom liquefaction probability estimates for Santa Clara County and southwest Alameda County are reflected in the lower damage ratios in the same areas in figure 6B and C. Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  23 123° 122° 121° YOLO NAPA SONOMA SACRAMENTO SOLANO SOLANO MARIN 38° SAN JOAQUIN CONTRA COSTA SAN FRANCISCO IC CIF PA ALAMEDA OC STANISLAUS N EA SAN MATEO SANTA CLARA SANTA CRUZ MERCED 37° EXPLANATION County boundary Contributing Hazards SAN BENITO None Shaking Shaking + default liquefaction MONTEREY Shaking + custom liquefaction Shaking + landslide Shaking + default liquefaction + landslide Shaking + custom liquefaction + landslide 00 Coordinate System: GCS North American 1983 Datum: North American 1983 Area CALIF of maps 00 1 10 10 1 20 2 2 20 30 3 3 30 40 4 4 40 5 MILES 50 MILES 50 KILOMETERS 5 KILOMETERS Figure 5.  Map of hazards caused by the HayWired earthquake scenario mainshock contributing to building damage estimated for census tracts in the San Francisco Bay region, California. 24   The HayWired Earthquake Scenario—Engineering Implications Table 7.  Building damage resulting from ground shaking caused by the HayWired earthquake scenario mainshock in counties of the San Francisco Bay region, California, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). Damage reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23] County Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Sonoma Stanislaus Yolo  Total Slight Moderate 1,192,789 742,250 133,432 4,987 16,421 18,538 12,409 8,435 585,608 96,226 527,695 1,272,221 53,518 86,575 20,810 35,564 5,069 4,812,547 4,739,455 1,867,978 126,505 1,800 6,486 10,327 3,450 6,852 842,303 61,923 778,106 2,314,206 31,126 67,398 7,291 16,351 1,514 10,883,071 Extensive Complete (thousands of dollars) 4,254,356 6,063,511 1,200,315 1,560,538 20,149 1,843 118 2 443 2 1,027 33 176 5 1,985 1,387 183,617 25,099 6,610 339 193,827 29,036 810,956 245,904 3,179 137 9,408 782 475 6 1,387 24 84 0 6,688,112 7,928,648 Total 16,250,111 5,371,081 281,929 6,907 23,351 29,926 16,040 18,658 1,636,626 165,099 1,528,665 4,643,287 87,960 164,163 28,582 53,327 6,667 30,312,379 Table 8.  Building damage resulting from ground shaking and default liquefaction caused by the HayWired earthquake scenario mainshock in counties of the San Francisco Bay region, California, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). Damage reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23] County Alameda Contra Costa Marin Merced1 Monterey1 Napa Sacramento1 San Benito1 San Francisco San Joaquin1 San Mateo Santa Clara2 Santa Cruz1 Solano Sonoma Stanislaus1 Yolo*  Total 1 Slight Moderate 1,134,931 733,337 132,990 4,987 16,421 18,538 12,409 8,435 582,164 96,226 518,759 1,253,336 53,518 86,393 20,810 35,564 5,069 4,713,887 4,623,506 1,857,355 127,445 1,800 6,486 10,327 3,450 6,852 846,023 61,923 780,983 2,307,609 31,126 67,688 7,291 16,351 1,514 10,757,729 Extensive Complete (thousands of dollars) 6,094,483 7,248,794 1,496,618 1,741,030 44,173 16,077 118 2 443 2 1,027 33 176 5 1,985 1,387 332,753 116,177 6,610 339 520,077 222,057 1,440,357 636,429 3,179 137 18,389 6,104 475 6 1,387 24 84 0 9,962,334 9,988,603 Total 19,101,714 5,828,340 320,686 6,907 23,351 29,926 16,040 18,658 1,877,116 165,099 2,041,877 5,637,731 87,960 178,575 28,582 53,327 6,667 35,422,556 Counties outside the areas of mapped liquefaction susceptibility. Estimates reflect the dual groundwater depth assumption (16 feet in western Santa Clara County, 5 feet elsewhere) ​made for ​consistency with the custom liquefaction study 2 Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  25 Table 9.  Approximate building damage resulting from ground shaking, custom liquefaction, and landslide caused by the HayWired earthquake scenario mainshock in counties of the San Francisco Bay region, California. [Damage reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23] Slight County Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Sonoma Stanislaus Yolo  Total Moderate Extensive Complete (thousands of dollars) 4,567,748 6,805,683 7,654,758 1,808,999 1,661,298 1,787,929 117,380 102,445 50,314 1,696 140 2 6,078 486 2 9,258 1,313 168 3,138 193 4 6,791 2,485 1,427 896,174 373,868 118,965 58,211 7,898 318 767,512 545,032 225,213 2,318,778 1,163,761 370,790 29,281 4,590 622 63,439 21,460 6,955 6,414 3,461 1,846 15,045 1,561 20 1,326 85 0 10,677,268 10,695,759 10,219,333 1,134,930 746,639 130,835 4,930 15,883 17,625 11,651 8,516 596,005 95,074 525,836 1,283,936 52,599 85,753 19,652 34,412 4,695 4,768,971 Total 20,163,118 6,004,865 400,973 6,768 22,449 28,364 14,986 19,219 1,985,012 161,500 2,063,593 5,137,265 87,093 177,607 31,372 51,037 6,106 36,361,327 Table 10.  Total building damage from the HayWired earthquake scenario mainshock in counties of the San Francisco Bay region, California. [Damage reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23. cm/s, centimeters per second; PGV, peak ground velocity. Shading indicates no estimated damage from liquefaction or landslide] County Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Sonoma Stanislaus Yolo  Total Shaking 16,250 5,371 282 7 23 30 16 19 1,637 165 1,529 4,643 88 164 29 53 7 30,313 Default liquefaction1 Custom liquefaction2 2,852 457 39 (millions of dollars) 3,057 3,081 457 39 Landslide Combined (within 20-cm/s best estimate PGV limit) building damage 147 14 84 0 0 0.4 240 240 9 513 291 9 21 1 2 5 513 994 264 14 0 5,109 With adjusted groundwater depth. 1 Limited to Holzer and others (2008, 2010) method study area. 2 Combined best estimate liquefaction 14 3,321 4,635 292 19,478 5,842 405 7 23 30 16 19 1,886 165 2,051 4,955 89 180 34 53 7 35,240 26   The HayWired Earthquake Scenario—Engineering Implications A. Shaking B. Default liquefaction 123° 122° 121° 123° 122° YOLO SONOMA YOLO NAPA SACRAMENTO SONOMA NAPA SACRAMENTO SOLANO SOLANO MARIN 38° MARIN SAN JOAQUIN CONTRA COSTA SAN FRANCISCO SAN JOAQUIN CONTRA COSTA SAN FRANCISCO ALAMEDA ALAMEDA STANISLAUS SAN MATEO STANISLAUS SAN MATEO SANTA CLARA SANTA CLARA MERCED SANTA CRUZ IC CIF PA IC CIF PA 37° 121° MERCED SANTA CRUZ OC EAN EAN OC SAN BENITO SAN BENITO EXPLANATION MONTEREY County boundary Building damage ratio MONTEREY (damage divided by replacement value, expressed as a percent) 0 <0.1% 0.1–0.5% 0.5–2.5% 2.5–5.0% 5.0–10.0% >10% C. Custom liquefaction YOLO SONOMA NAPA SACRAMENTO D. Landslide Area CALIF of maps YOLO SONOMA NAPA SACRAMENTO SOLANO SOLANO MARIN 38° MARIN SAN JOAQUIN CONTRA COSTA SAN FRANCISCO SAN FRANCISCO ALAMEDA ALAMEDA STANISLAUS SAN MATEO STANISLAUS SAN MATEO SANTA CLARA MERCED SANTA CRUZ IC CIF PA IC CIF PA 37° SANTA CLARA MERCED SANTA CRUZ OC EAN EAN OC SAN BENITO SAN BENITO MONTEREY Coordinate System: GCS North American 1983 Datum: North American 1983 SAN JOAQUIN CONTRA COSTA MONTEREY 0 0 1 10 2 20 3 30 4 40 5 MILES 50 MILES 10 KILOMETERS 1 20 2 30 3 40 4 50 5 KILOMETERS Figure 6.  Maps showing building damage ratio by causative hazard for the HayWired earthquake scenario mainshock, San Francisco Bay region, California, estimated using Hazus (FEMA, 2012). A, Shaking only. B, Default liquefaction (applied where liquefaction computed using the Holzer and others [2008, 2010] method was not available). C, Liquefaction computed using the Holzer and others (2008, 2010) method. D, Landslide. Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  27 Landslide building damage, as estimated using the custom landslide hazard data, totals $291 million in the nine counties in which it was modeled, and is most substantial in Alameda County ($147 million) and Marin County ($84 million). Census tract damage ratios resulting from landslide (fig. 6D) are generally modest, peaking at just 3.3 percent versus 17.1 percent for custom liquefaction and 45.6 percent for shaking. Figure 7 provides the best estimate census tract damage ratio map, combining damage from shaking, liquefaction (using custom data where available, default data elsewhere), and landslide. McCrink and Perez (2017) report slope failure displacements and probabilities in areas of peak ground velocity of 20 centimeters per second (cm/s) and larger; their approach has more uncertainty in areas below that threshold. Results reported above are those within the 20 cm/s contour. Beyond the 20 cm/s contour, an additional $145 million in damage could be expected, including $41 million in Marin, $27 million in each of San Francisco and Santa Clara counties, $20 million in San Mateo and $17 million in Sonoma counties. As a reference point for comparison, the losses from the Loma Prieta earthquake were tallied by Holzer (1994) as $5.8 billion from shaking, $97 million from liquefaction and $30 million from landslides. Loma Prieta earthquake landslide losses were 0.5 percent of shaking losses compared to 1 percent (or 1.5 percent for the whole area) estimated for the HayWired earthquake scenario. Loma Prieta liquefaction losses were 1.5 percent of shaking losses, compared to 15 percent estimated for the HayWired earthquake scenario. The higher liquefaction losses for the HayWired earthquake scenario are attributed to the developed areas affected by liquefaction in close proximity to the fault in Alameda County. In addition to the summary data provided here, detailed results were generated for use by other HayWired researchers, including the following: • The distribution of square footage across the various Hazus damage states by Hazus occupancy class and census tract, for the combined liquefaction and landslide mainshock assessment • Estimated economic loss associated with building damage, by occupancy and county, for the combined liquefaction and landslide mainshock assessment Liquefaction Modeling in Selected HayWired Aftershocks Because the custom liquefaction data were only available in Alameda and Santa Clara Counties, and not available for the other counties potentially subject to liquefaction (Marin, Sonoma, Napa, Solano, Contra Costa, San Mateo, and San Francisco) and because custom liquefaction and landslide hazard data were only developed for the mainshock, it was decided that the focus of the comparative studies of building damage in the various aftershocks would primarily be damage resulting from ground shaking alone. However, to allow for selected comparisons of loss resulting from liquefaction and repeated liquefaction, several of the aftershocks were analyzed both for shaking only and for shaking and default liquefaction, including the M5.98 Mountain View, M6.4 Cupertino, and M5.42 Oakland aftershocks. A comparison of damage estimates for the mainshock and three selected aftershocks, with and without liquefaction assessed using the Hazus default approach and the original uniform depth to groundwater assumption, is provided in table 11. As noted above, in the mainshock, liquefaction is estimated to add approximately $5.1 billion in building damage (an additional 17 percent), whereas in the selected aftershocks, liquefaction adds between $78 million (an additional 15 percent) in the M5.42 Oakland aftershock and $218 million (an additional 9 percent) in the M6.4 Cupertino aftershock. Ground Shaking Results for the HayWired Earthquake Sequence Economic loss resulting from building damage caused by ground shaking has been estimated for each aftershock, as was done for the mainshock. Because Hazus is not able to estimate additional damage to damaged buildings, the mainshock and aftershocks have been modeled as independent events. That is, for each event, the building inventory is assumed to be in an undamaged state at the time of the earthquake; previous damage resulting from the mainshock or preceding aftershocks is not considered. The results of the Hazus analyses are summarized in table 12. As shown in table 12, damage in each of the aftershocks is at least an order of magnitude smaller than that of the mainshock. The aftershocks that result in the largest economic losses from regional building damage are the M6.4 Cupertino ($2.48 billion in building damage), M6.2 Palo Alto ($1.37 billion in building damage), M5.98 Mountain View ($890 million in building damage), and M5.42 Oakland ($510 million in building damage) events. The total direct economic losses resulting from these aftershocks are $3.81 billion (M6.4 Cupertino), $2.11 billion (M6.2 Palo Alto), $1.38 billion (M5.98 Mountain View), and $780 million (M5.42 Oakland). For comparison, FEMA’s initial estimate of expected damage and loss in the 2014 M6.0 South Napa earthquake (including liquefaction), generated using the same inventory data as that used to model the HayWired earthquake scenario and its aftershocks, totaled $347 million in building damage and $575 million in total direct economic losses related to building damage (Doug Bausch, written commun., Federal Emergency Management Agency, 2014). The expected damage in the South Napa event is smaller than the four largest HayWired aftershocks, but larger than the remaining 12 aftershocks. In terms of building damage ratio (defined in Hazus as the ratio of repair to replacement cost), the South Napa earthquake simulation had an overall building damage ratio of 28   The HayWired Earthquake Scenario—Engineering Implications 123° 122° 121° YOLO NAPA SONOMA SACRAMENTO SOLANO MARIN 38° CONTRA SAN JOAQUIN COSTA SAN FRANCISCO IC CIF PA ALAMEDA OC STANISLAUS SAN N EA MATEO SANTA CLARA EXPLANATION 37° SANTA CRUZ MERCED County boundary Best estimate building damage ratio (damage divided by replacement value, expressed as a percent) 0 SAN BENITO <0.1% 0.1–0.5% 0.5–2.5% MONTEREY 2.5–5.0% 5.0–10.0% >10% 00 Coordinate System: GCS North American 1983 Datum: North American 1983 Area CALIF of maps 00 1 10 10 1 20 2 2 20 30 3 3 30 40 4 4 40 5 MILES 50 MILES 50 KILOMETERS 5 KILOMETERS Figure 7.  Map showing best estimate building damage ratio for the HayWired earthquake scenario mainshock, San Francisco Bay region, California, considering liquefaction computed using the Holzer and others (2008, 2010) method and landslide data, where available (Jones and others, 2017; McCrink and Perez, 2017). Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  29 Table 11.  Direct economic losses from the HayWired earthquake scenario mainshock and selected aftershocks, with and without default liquefaction, San Francisco Bay region, California, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). See table 1 for explanation of aftershock short names and magnitudes. Damage reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23. liq., liquefaction. Shading indicates net loss from the liquefaction hazard] HayWired scenario event Mainshock shaking only Mainshock shaking + liq. Mainshock liq. loss CU64 shaking only CU64 shaking + liq. CU64 liq. loss MV598 shaking MV598 shaking + liq. MV598 liq. loss OK542 shaking OK542 shaking + liq. OK542 liq. loss Nonstructural damage Total building damage1 Content damage (millions of dollars) Building damage ratio2 (percent) 24,495.2 28,419.5 3,924.3 2,145.2 2,313.5 168.3 813.2 892.7 79.5 476.8 536.3 59.5 2.91 3.40 0.49 0.24 0.26 0.02 0.09 0.10 0.01 0.05 0.06 0.01 8,003.0 9,232.8 1,229.8 888.3 948.0 59.7 382.4 410.0 27.6 237.3 256.3 19.0 Structural damage 5,817.2 7,037.5 1,220.3 333.3 382.9 49.6 81.1 104.5 23.4 32.6 51.1 18.5 30,312.4 35,457.0 5,144.6 2,478.5 2,696.4 217.9 894.3 997.2 102.9 509.4 587.3 77.9 Inventory loss Building Total direct damage-related economic income loss loss (millions of dollars) 305.6 376.8 71.2 35.9 40.0 4.1 22.3 24.1 1.8 9.3 10.2 0.9 8,012.7 9,059.4 1,046.7 405.2 455.6 50.4 82.4 105.6 23.2 24.0 42.4 18.4 46,633.7 54,126.0 7,492.3 3,807.9 4,139.9 332.0 1,381.4 1,536.9 155.5 780.0 896.3 116.3 Total of structural and nonstructural damage. 1 Ratio of repair to replacement cost, calculated as the total estimated building damage divided by the total building replacement value (as defined in Hazus). 2 Table 12.  Direct economic losses from ground shaking caused by the HayWired earthquake scenario mainshock and aftershock sequence, San Francisco Bay region, California, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). See table 1 for explanation of aftershock short names and magnitudes. Damage reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23] HayWired scenario event Mainshock UC523 SP504 FF558 FR510 OK542 PA62 MP552 AT511 PA569 PA522 PA526 MV598 CU64 SV535 SC509 PA501 NonTotal structural building damage damage1 (millions of dollars) 5,817.2 24,495.2 30,312.4 14.9 333.5 348.4 5.6 100.7 106.3 1.1 21.5 22.6 5.9 100.4 106.3 32.6 476.8 509.4 159.9 1,207.6 1,367.5 8.4 142.9 151.3 9.2 129.5 138.8 27.3 326.6 353.8 13.3 338.4 351.7 16.7 239.0 255.7 81.1 813.2 894.3 333.3 2,145.2 2,478.5 32.7 255.8 288.5 14.6 210.3 224.8 6.1 65.6 71.7 Structural damage Building damage ratio2 (percent) 2.91 0.03 0.01 0.00 0.01 0.05 0.13 0.01 0.01 0.03 0.03 0.02 0.09 0.24 0.03 0.02 0.01 Content damage 8,003.0 176.2 50.4 10.5 48.1 237.3 525.4 71.2 59.1 157.1 182.7 113.4 382.4 888.3 102.2 101.6 27.1 Building Total direct damage-related economic income loss loss (millions of dollars) 305.6 8,012.7 46,633.7 12.4 8.2 545.2 1.3 3.7 161.7 0.5 0.6 34.1 3.7 3.1 161.2 9.3 24.0 780.0 23.6 193.7 2,110.2 2.6 6.9 232.1 2.5 6.0 206.4 7.1 25.7 543.9 7.0 9.1 550.4 5.2 12.1 386.4 22.3 82.4 1,381.4 35.9 405.2 3,807.9 6.1 23.9 420.7 5.1 10.2 341.8 1.3 3.9 104.0 Inventory loss Total of structural and nonstructural damage. 1 Ratio of repair to replacement cost, calculated as the total estimated building damage divided by the total building replacement value (as defined in Hazus). 2 30   The HayWired Earthquake Scenario—Engineering Implications 0.16 percent, similar to the M6.2 Palo Alto aftershock’s damage ratio of 0.13 percent. Regional building damage patterns, in terms of census tract level building damage ratios (total estimated building damage for each census tract divided by total building replacement value), are mapped in figures 8–11. As shown, the aftershocks with the largest losses also show the broadest damage ratio patterns and the highest maximum damage ratios. Building damage is summarized by structural system or Hazus’ MBTs (see table 4 for definitions), grouped across building height classes, in tables 13 and 14; table 13 provides building damage in terms of economic losses (in millions of dollars), and table 14 provides the results in terms of building damage ratio. Although wood frame construction (MBT W1 and W2) makes up the bulk of the exposure (69 percent of exposure, see also table 4), damage to this type of construction makes up a smaller percentage of the losses (47 percent in the mainshock, and between 45 and 56 percent in the aftershocks). Conversely, the more vulnerable building types represent larger proportions of loss than exposure, including URM (2 percent of exposure, 6 percent of loss in the mainshock, and as much as 8 percent in the M5.42 Oakland aftershock), concrete frame with URM infill walls (MBT C3—1 percent of exposure, 5 percent of loss in the mainshock, and 2–3 percent of loss in the aftershocks), and steel frame with URM infill walls (MBT S5—1 percent of exposure, 5 percent of loss in the mainshock, and 2–4 percent of loss in the aftershocks). This vulnerability is further demonstrated by the damage ratios tabulated in table 14; S5, C3, and URM are among the MBTs with the largest net damage ratios in the mainshock (14, 10, and 7 percent, respectively), along with mobile homes (MH—9 percent) and light steel frame (S3—9 percent). Damage ratios in the aftershocks are generally two orders of magnitude smaller than those of the mainshock, except for the larger aftershocks (M6.2 Palo Alto, M5.98 Mountain View, and M6.4 Cupertino), where the ratios tend to be just one order of magnitude smaller than the mainshock. Concentrations of vulnerable building types may also result in higher damage ratios from a nearby aftershock, such as (1) precast concrete tilt-up wall structures (PC1) with a higher damage ratio in the M5.23 Union City aftershock than the other M5 aftershocks and (2) S3, S5, and URM damage in the M5.42 Oakland aftershock. Building exposure (in thousands of dollars) for tilt-up structures (PC1) and low-rise URM is mapped in figure 12, highlighting the concentrations of these vulnerable buildings near the epicenters of the M5.23 Union City and M5.42 Oakland aftershocks. Unreinforced Masonry Construction It should be noted that the MBT damage ratios given in table 14 represent net damage across the entire 17-county study area and include damage for all MBT subclasses (that is, subclasses by height and seismic design level); localized damage and damage ratios for MBT subclasses may be substantially different. For example, for all URM construction (URML and URMM), the net damage ratio in the mainshock is 7 percent, but census tract damage ratios for pre-code (unretrofitted), low-rise URM (URML) range from 0 to 94 percent. Damage to this subclass of URM building represents 60 percent of the URM damage but just 24 percent of the building square footage. Census tract damage ratio maps for pre-code URML construction for the mainshock and three of the more damaging aftershocks (M6.4 Cupertino, M6.2 Palo Alto, and M5.42 Oakland) are provided in figure 13. In the mainshock, pre-code URML damage ratios in Alameda and Contra Costa Counties are substantial, with 51 percent of census tracts in these two counties having damage ratios exceeding 50 percent. Total loss for pre-code URML in the mainshock is $885 million, whereas losses in the selected aftershocks are $20.5 million (M6.4 Cupertino), $15.7 million (M6.2 Palo Alto), and $17.4 million (M5.42 Oakland)—2.3, 1.8, and 2.0 percent of the mainshock loss, respectively. Damage ratios in the selected aftershocks are lower than the mainshock (all are less than 25 percent), and damage is more localized. A small pocket of census tracts has damage ratios between 15 and 25 percent in Santa Clara County in the M6.4 Cupertino aftershock, and a smaller pocket of tracts with damage ratios between 5 and 10 percent are on the border of Santa Clara and San Mateo Counties in the M6.2 Palo Alto aftershock, and census tract damage ratios do not exceed 5 percent in the M5.42 Oakland aftershock. As noted previously, Hazus is not able to estimate additional damage to damaged buildings; each analysis conducted for this study is independent and assumes that the inventory is in an undamaged condition before the earthquake. Given that the aftershocks may, in reality, cause damage in areas already substantially damaged, it is instructive to examine the URM damage results more closely. To assess whether adding the aftershock damage to the mainshock damage would result in overestimating damage to substantially damaged structures, individual census tract level results in the mainshock and the aftershocks were reviewed. At the most basic level, the sum of losses in multiple events should not exceed the value of the exposed inventory, assuming no repair between events. For pre-code URML, building damage in the mainshock represents 18.5 percent of exposure, with a maximum census tract damage ratio of 93.8 percent. After adding the damage from all aftershocks, the total building damage ratio is 20.4 percent, with a maximum census tract damage of 96.4 percent. Fortunately, for the simulated aftershock sequence, census tracts with the largest damage to pre-code URML buildings in the aftershocks are not the same as those tracts with substantial damage in the mainshock. At the census tract level, the sum of economic losses from building damage to pre-code URML in the HayWired mainshock and all of its aftershocks does not exceed the building’s exposure value. Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  31 A. M 5.23 Union City B. M 5.04 San Pablo 123° 122° 121° 123° 122° YOLO SONOMA YOLO NAPA SACRAMENTO SONOMA NAPA SACRAMENTO SOLANO SOLANO MARIN 38° MARIN SAN JOAQUIN CONTRA COSTA SAN FRANCISCO SAN JOAQUIN CONTRA COSTA SAN FRANCISCO ALAMEDA ALAMEDA STANISLAUS SAN MATEO STANISLAUS SAN MATEO SANTA CLARA SANTA CLARA MERCED SANTA CRUZ IC CIF PA IC CIF PA 37° 121° MERCED SANTA CRUZ OC EAN EAN OC SAN BENITO SAN BENITO EXPLANATION MONTEREY Epicenter County boundary Building damage ratio MONTEREY (damage divided by replacement value, expressed as a percent) 0 <0.1% 0.1–0.5% 0.5–2.5% 2.5–5.0% 5.0–10.0% C. M 5.58 Fairfield YOLO SONOMA NAPA D. M 5.10 Fremont Area CALIF of maps SACRAMENTO YOLO SONOMA NAPA SACRAMENTO SOLANO SOLANO MARIN 38° MARIN SAN JOAQUIN CONTRA COSTA SAN FRANCISCO SAN FRANCISCO ALAMEDA ALAMEDA STANISLAUS SAN MATEO SANTA CLARA IC CIF PA IC CIF PA MERCED SANTA CRUZ STANISLAUS SAN MATEO SANTA CLARA 37° SAN JOAQUIN CONTRA COSTA MERCED SANTA CRUZ OC EAN EAN OC SAN BENITO SAN BENITO MONTEREY Coordinate System: GCS North American 1983 Datum: North American 1983 MONTEREY 00 00 10 1 20 2 30 3 40 4 50 MILES 5 MILES 1 20 2 30 3 40 4 50 5 KILOMETERS 10 KILOMETERS Figure 8.  Maps showing building damage ratio for HayWired earthquake scenario aftershocks 1–4 (table 1), San Francisco Bay region, California, estimated using Hazus (FEMA, 2012): A, magnitude (M ) 5.23 Union City; B, M   5.04 San Pablo; C, M   5.58 Fairfield; and D, M    5.10 Fremont. 32   The HayWired Earthquake Scenario—Engineering Implications B. M 6.2 Palo Alto A. M 5.42 Oakland 123° 122° 121° 123° 122° YOLO YOLO SONOMA SONOMA NAPA NAPA SACRAMENTO SACRAMENTO SOLANO SOLANO MARIN MARIN 38° SAN FRANCISCO SAN FRANCISCO SAN JOAQUIN CONTRA COSTA SAN JOAQUIN CONTRA COSTA ALAMEDA STANISLAUS SAN MATEO SANTA CLARA SANTA CLARA IC CIF PA IC CIF PA MERCED SANTA CRUZ MERCED SANTA CRUZ 37° OC EAN EAN OC SAN BENITO SAN BENITO EXPLANATION Epicenter County boundary Building damage ratio MONTEREY MONTEREY (damage divided by replacement value, expressed as a percent) 36° 0 <0.1% 0.1–0.5% 0.5–2.5% 2.5–5.0% 5.0–10.0% C. M 5.52 Menlo Park YOLO SONOMA NAPA SACRAMENTO D. M 5.11 Atherton Area CALIF of maps YOLO SONOMA NAPA MARIN MARIN 38° SACRAMENTO SOLANO SOLANO SAN JOAQUIN CONTRA COSTA SAN JOAQUIN CONTRA COSTA 38° SAN FRANCISCO SAN FRANCISCO ALAMEDA ALAMEDA STANISLAUS SAN MATEO SANTA CLARA IC CIF PA IC CIF PA MERCED SANTA CRUZ STANISLAUS SAN MATEO SANTA CLARA 37° 38° ALAMEDA STANISLAUS SAN MATEO 37° 121° MERCED SANTA CRUZ OC EAN EAN OC SAN BENITO SAN BENITO MONTEREY MONTEREY 36° Coordinate System: GCS North American 1983 Datum: North American 1983 37° 00 10 1 20 2 30 3 40 4 50 MILES 5 MILES 1 20 2 30 3 40 4 50 5 KILOMETERS 00 10 KILOMETERS Figure 9.  Maps showing building damage ratio for HayWired earthquake scenario aftershocks 5–8 (table 1), San Francisco Bay region, California, estimated using Hazus (FEMA, 2012): A, magnitude (M )  5.42 Oakland; B, M  6.2 Palo Alto; C, M  5.52 Menlo Park; and D, M  5.11 Atherton. 36° Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  33 B. M 5.22 Palo Alto A. M 5.69 Palo Alto 123° 122° 121° 123° 122° YOLO YOLO SONOMA NAPA SONOMA SACRAMENTO NAPA MARIN MARIN SAN JOAQUIN CONTRA COSTA SAN JOAQUIN CONTRA COSTA ALAMEDA ALAMEDA STANISLAUS SAN MATEO STANISLAUS SAN MATEO SANTA CLARA SANTA CLARA IC CIF PA IC CIF PA MERCED SANTA CRUZ MERCED SANTA CRUZ OC EAN EAN OC EXPLANATION MONTEREY Epicenter County boundary Building damage ratio MONTEREY (damage divided by replacement value, expressed as a percent) 36° 0 <0.1% 0.1–0.5% 0.5–2.5% 2.5–5.0% 5.0–10.0% C. M 5.26 Palo Alto YOLO SONOMA NAPA SACRAMENTO 36° D. M 5.98 Mountain View Area CALIF of maps YOLO SONOMA NAPA SACRAMENTO SOLANO SOLANO MARIN MARIN SAN JOAQUIN CONTRA COSTA SAN JOAQUIN CONTRA COSTA 38° SAN FRANCISCO SAN FRANCISCO ALAMEDA ALAMEDA STANISLAUS SAN MATEO SANTA CLARA IC CIF PA IC CIF PA MERCED SANTA CRUZ STANISLAUS SAN MATEO SANTA CLARA 37° 37° SAN BENITO SAN BENITO 38° 38° SAN FRANCISCO SAN FRANCISCO 37° SACRAMENTO SOLANO SOLANO 38° 121° MERCED SANTA CRUZ OC EAN EAN OC SAN BENITO SAN BENITO MONTEREY MONTEREY 36° Coordinate System: GCS North American 1983 Datum: North American 1983 37° 00 10 1 20 2 30 3 40 4 50 MILES 5 MILES 1 20 2 30 3 40 4 50 5 KILOMETERS 00 10 KILOMETERS Figure 10.  Maps showing building damage ratio for HayWired earthquake scenario aftershocks 9–12 (table 1), San Francisco Bay region, California, estimated using Hazus (FEMA, 2012): A, magnitude (M )  5.69 Palo Alto; B, M  5.22 Palo Alto; C, M  5.26 Palo Alto; and D, M  5.98 Mountain View. 36° 34   The HayWired Earthquake Scenario—Engineering Implications A. M 6.4 Cupertino B. M 5.35 Sunnyvale 123° 122° 121° 123° 122° YOLO YOLO NAPA SONOMA SACRAMENTO SONOMA NAPA SOLANO MARIN SAN JOAQUIN CONTRA COSTA SAN FRANCISCO SAN JOAQUIN CONTRA COSTA ALAMEDA STANISLAUS SAN MATEO STANISLAUS SAN MATEO SANTA CLARA SANTA CLARA IC CIF PA IC CIF PA MERCED SANTA CRUZ MERCED SANTA CRUZ EAN EAN SAN BENITO Epicenter County boundary Building damage ratio MONTEREY MONTEREY (damage divided by replacement value, expressed as a percent) 0 <0.1% 0.1–0.5% 0.5–2.5% 2.5–5.0% 5.0–10.0% C. M 5.09 Santa Clara YOLO SONOMA NAPA SACRAMENTO 36° D. M 5.01 Palo Alto Area CALIF of maps YOLO SONOMA NAPA SOLANO SACRAMENTO SOLANO MARIN MARIN SAN JOAQUIN CONTRA COSTA SAN FRANCISCO SAN JOAQUIN CONTRA COSTA 38° SAN FRANCISCO ALAMEDA ALAMEDA STANISLAUS SAN MATEO SANTA CLARA IC CIF PA IC CIF PA MERCED SANTA CRUZ STANISLAUS SAN MATEO SANTA CLARA 37° 37° OC OC SAN BENITO EXPLANATION 38° 38° SAN FRANCISCO ALAMEDA 37° SACRAMENTO SOLANO MARIN 38° 121° MERCED SANTA CRUZ 37° EAN EAN OC OC SAN BENITO SAN BENITO MONTEREY MONTEREY 36° 36° Coordinate System: GCS North American 1983 Datum: North American 1983 00 10 1 20 2 30 3 40 4 50 MILES 5 MILES 1 20 2 30 3 40 4 50 5 KILOMETERS 00 10 KILOMETERS Figure 11.  Maps showing building damage ratio for HayWired earthquake scenario aftershocks 13–16 (table 1), San Francisco Bay region, California, estimated using Hazus (FEMA, 2012): A, magnitude (M ) 6.4 Cupertino; B, M  5.35 Sunnyvale; C, M  5.09 Santa Clara; and D, M  5.01 Palo Alto. Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  35 Table 13.  Building damage by model building type from ground shaking caused by the HayWired earthquake scenario mainshock and aftershock sequence, San Francisco Bay region, California, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). See table 4 for definitions of Hazus model building types (MBT). See table 1 for explanation of aftershock short names and magnitudes. Damage in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23] SP504 FF558 FR510 OK542 PA62 MP552 AT511 PA569 PA522 PA526 MV598 CU64 SV535 SC509 PA501 C1 636 C2 2,095 C3 1,580 MH 667 PC1 788 PC2 158 RM1 1,814 RM2 271 S1 2,257 S2 973 S3 843 S4 514 S5 1,561 URM 1,793 W1 11,329 W2 3,034  Total 30,312 UC523 MBT Mainshock Building damage (millions of dollars) 3.6 29.1 5.3 5.8 25.8 3.2 30.1 4.7 16.3 17.1 9.9 4.9 6.3 10.0 135.7 40.6 348.4 0.9 9.0 3.0 1.7 2.4 0.5 7.2 1.1 3.3 2.8 2.2 1.4 2.9 4.8 51.2 11.7 106.3 0.2 2.2 0.6 0.5 1.2 0.2 1.8 0.3 1.0 0.7 0.6 0.3 0.6 0.6 9.5 2.1 22.6 1.1 10.8 1.8 2.5 9.8 1.3 9.6 1.5 5.2 6.1 3.2 1.7 2.0 2.3 36.6 10.7 106.3 3.7 35.1 13.9 5.6 16.0 2.0 30.8 5.1 17.9 14.4 12.9 6.1 20.2 41.1 229.1 55.5 509.4 26.7 127.7 41.7 18.1 57.1 9.8 103.6 17.4 105.3 57.6 33.9 22.5 32.5 35.6 534.6 143.3 1,367.5 1.8 15.5 4.2 2.0 5.5 0.8 12.9 1.9 7.3 5.3 3.6 2.3 4.2 4.2 61.6 18.1 151.3 1.3 14.0 3.6 2.1 6.0 0.9 11.6 1.8 5.3 4.6 3.6 2.0 3.6 4.4 58.6 15.5 138.8 5.7 35.8 8.9 5.6 17.4 2.7 30.0 5.0 21.9 15.5 8.6 5.7 7.7 8.3 135.0 40.1 353.8 3.5 34.6 7.9 4.1 15.4 2.3 31.8 4.6 14.2 13.0 7.7 5.3 8.6 9.3 145.4 43.9 351.7 2.9 25.6 6.2 3.5 12.4 1.8 22.1 3.5 10.8 10.1 6.3 3.9 5.9 7.3 104.2 29.1 255.7 15.8 89.5 22.3 14.8 54.2 8.2 76.9 12.9 63.3 43.3 24.2 14.8 19.9 21.7 311.7 100.5 894.3 53.9 207.7 67.5 34.1 97.4 18.1 175.9 29.6 185.0 98.6 51.8 39.6 42.2 47.2 1,102.0 227.8 2,478.5 4.6 28.9 6.2 5.5 21.0 2.9 24.3 4.5 14.4 13.2 7.9 4.6 4.1 6.5 112.4 27.6 288.5 2.7 25.5 5.7 4.1 14.6 2.1 20.9 3.5 9.7 9.6 5.9 3.8 3.9 5.1 83.7 23.9 224.8 0.7 7.7 1.7 1.5 4.0 0.5 6.1 1.0 2.6 2.7 2.1 1.0 1.6 2.1 29.0 7.4 71.7 Table 14.  Building damage ratio by model building type from ground shaking caused by the HayWired earthquake scenario mainshock and aftershock sequence, San Francisco Bay region, California, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). See table 4 for definitions of Hazus model building types (MBT). See table 1 for explanation of aftershock short names and magnitudes] FF558 FR510 OK542 PA62 MP552 AT511 PA569 PA522 PA526 MV598 CU64 SV535 SC509 PA501 5 3 10 9 4 3 3 3 5 4 9 3 14 7 2 3 SP504 C1 C2 C3 MH PC1 PC2 RM1 RM2 S1 S2 S3 S4 S5 URM W1 W2 UC523 MBT Mainshock Building damage ratio1 (percent) 0.03 0.05 0.03 0.07 0.13 0.06 0.05 0.06 0.04 0.07 0.11 0.03 0.06 0.04 0.02 0.05 0.01 0.01 0.02 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.01 0.03 0.02 0.01 0.01 0.00 0.00 0.00 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.01 0.00 0.00 0.00 0.01 0.02 0.01 0.03 0.05 0.02 0.02 0.02 0.01 0.03 0.04 0.01 0.02 0.01 0.01 0.01 0.03 0.06 0.09 0.07 0.08 0.04 0.05 0.06 0.04 0.06 0.14 0.04 0.18 0.17 0.04 0.06 0.20 0.21 0.27 0.23 0.28 0.17 0.16 0.21 0.24 0.24 0.38 0.15 0.29 0.15 0.08 0.16 0.01 0.03 0.03 0.03 0.03 0.01 0.02 0.02 0.02 0.02 0.04 0.02 0.04 0.02 0.01 0.02 0.01 0.02 0.02 0.03 0.03 0.02 0.02 0.02 0.01 0.02 0.04 0.01 0.03 0.02 0.01 0.02 0.04 0.06 0.06 0.07 0.09 0.05 0.05 0.06 0.05 0.07 0.10 0.04 0.07 0.03 0.02 0.05 0.03 0.06 0.05 0.05 0.08 0.04 0.05 0.05 0.03 0.05 0.09 0.04 0.08 0.04 0.02 0.05 0.02 0.04 0.04 0.05 0.06 0.03 0.03 0.04 0.03 0.04 0.07 0.03 0.05 0.03 0.02 0.03 0.12 0.15 0.14 0.19 0.27 0.15 0.12 0.15 0.15 0.18 0.27 0.10 0.18 0.09 0.05 0.11 0.41 0.34 0.43 0.44 0.48 0.32 0.28 0.35 0.43 0.42 0.58 0.27 0.37 0.20 0.17 0.26 0.03 0.05 0.04 0.07 0.10 0.05 0.04 0.05 0.03 0.06 0.09 0.03 0.04 0.03 0.02 0.03 0.02 0.04 0.04 0.05 0.07 0.04 0.03 0.04 0.02 0.04 0.07 0.03 0.03 0.02 0.01 0.03 0.01 0.01 0.01 0.02 0.02 0.01 0.01 0.01 0.01 0.01 0.02 0.01 0.01 0.01 0.00 0.01 Ratio of repair to replacement cost, calculated as the total estimated building damage divided by the total building replacement value (as defined in Hazus). 1 36   The HayWired Earthquake Scenario—Engineering Implications 123° 122° 38° A. PC1 FIC CI PA N SA C AN FR N EA OC CO IS 37.5° Y BA EXPLANATION M 5.23 Union City aftershock epicenter M 5.42 Oakland aftershock epicenter County boundary Exposure ($1,000) <20,000 20,000–50,000 50,000–75,000 75,000–100,000 >100,000 38° B. URML N SA C AN FR CO IS FIC CI PA Y BA N EA OC 37.5° 0 Coordinate System: GCS North American 1983 Datum: North American 1983 0 Area of map 15 15 2 10 3 15 4 20 5 MILES 25 MILES 10 KILOMETERS 2 15 3 20 4 25 5 KILOMETERS CALIF Figure 12.  Maps showing building exposure value in census tracts in the vicinity of two HayWired earthquake scenario aftershocks, San Francisco Bay region, California. A, Precast concrete tilt-up wall buildings (PC1). B, Low-rise unreinforced masonry buildings (URML). Values are in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23. M, magnitude. Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  37 A. Mainshock 122.5° 123° 122° NAPA SONOMA SOLANO SACRAMENTO 121° YOLO SONOMA 38° B. M 6.4 Cupertino 122° NAPA SACRAMENTO SOLANO MARIN CONTRA COSTA MARIN SAN FRANCISCO SAN FRANCISCO STANISLAUS SAN MATEO OC EAN SANTA CLARA EAN OC SAN MATEO 38° ALAMEDA IC CIF PA IC CIF PA ALAMEDA SAN JOAQUIN CONTRA COSTA MERCED SANTA CRUZ SANTA CLARA 37° SAN BENITO CALIF Area of map 37° 0 4 Area of map SANTA CRUZ 8 0 12 16 MILES 0 4 8 12 16 KILOMETERS 0 MONTEREY 123° 15 30 30 45 45 60 MILES 60 KILOMETERS SOLANO MARIN SAN JOAQUIN CONTRA COSTA ALAMEDA NAPA SOLANO SACRAMENTO 38° MARIN CONTRA COSTA SAN FRANCISCO ALAMEDA IC CIF PA IC CIF PA OC OC EAN EAN SANTA CLARA MERCED SANTA CRUZ 122° SONOMA <1% 1–5% 5–10% 10–15% 15–25% 25–50% 50–100% STANISLAUS SAN MATEO 37° 122.5° (damage divided by replacement value, expressed as a percent) SACRAMENTO NAPA SAN FRANCISCO D. M 5.42 Oakland Epicenter County boundary URML-PC building damage ratio 122° YOLO 38° MONTEREY 15 EXPLANATION C. M 6.2 Palo Alto SONOMA CALIF SAN MATEO SANTA CLARA SAN BENITO Area CALIF of map 0 0 15 15 30 30 45 Area CALIF of map MONTEREY 45 60 MILES 60 KILOMETERS SANTA CRUZ 0 4 8 37° 12 16 MILES 0 4 8 12 16 KILOMETERS MONTEREY Coordinate System: GCS North American 1983 Datum: North American 1983 Figure 13.  Maps showing building damage ratio for pre-code low-rise unreinforced masonry (URML-PC) for the HayWired earthquake scenario mainshock and selected aftershocks, San Francisco Bay region, California, estimated using Hazus (FEMA, 2012): A, Mainshock; B, magnitude (M ) 6.4 Cupertino aftershock; C, M  6.2 Palo Alto aftershock; and D, M  5.42 Oakland aftershock. 38   The HayWired Earthquake Scenario—Engineering Implications Population Impacts—Casualties • Severity level 4—“Instantaneously killed or mortally injured.” The breakdown of Hazus-estimated casualties by severity level for the mainshock and each aftershock, for each time of day modeled (day, 2 p.m.; night, 2 a.m.; and commute, 5 p.m.), is provided in table 15. The HayWired scenario mainshock has been modeled as a daytime event occurring at 4:18 p.m.; occurrence times for the aftershocks were given in table 1. Results in table 15 have been shaded for the time of day closest to the modeled occurrence times for each event. Casualties in the aftershocks are as much as three orders of magnitude smaller than those in the mainshock. Expected casualties in most aftershocks are minor (primarily severity level 1 with some severity level 2), with the exception of the M6.2 Palo Alto and M6.4 Cupertino aftershocks, which also result in a few serious injuries (Severity Level 3) or deaths (Severity Level 4). The Hazus methodology estimates indoor and outdoor casualties by MBT as a function of each building’s damage state. Casualties are estimated at four severity levels as follows (FEMA, 2012): • Severity level 1—“Injuries requiring basic medical aid that could be administered by paraprofessionals. These types of injuries would require bandages or observation. Some examples are: a sprain, a severe cut requiring stitches, a minor burn (first degree or second degree on a small part of the body), or a bump on the head without loss of consciousness. Injuries of lesser severity that could be self treated are not estimated by Hazus.” • Severity level 2—“Injuries requiring a greater degree of medical care and use of medical technology such as x-rays or surgery, but not expected to progress to a life threatening status. Some examples are third degree burns or second degree burns over large parts of the body, a bump on the head that causes loss of consciousness, fractured bone, dehydration or exposure.” Population Impacts—Displacement and Shelter Requirements The Hazus shelter model allows users to estimate the number of displaced households caused by residential building damage and resulting habitability, and the number of people seeking public short-term shelter. The number of displaced households is derived from the estimated distribution of singlefamily homes and multifamily residential buildings across the • Severity level 3—“Injuries that pose an immediate life threatening condition if not treated adequately and expeditiously. Some examples are: uncontrolled bleeding, punctured organ, other internal injuries, spinal cord injuries, or crush syndrome.” Table 15.  Casualties from ground shaking caused by the HayWired earthquake scenario mainshock and aftershock sequence, San Francisco Bay region, California, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). See “Population Impacts: Casualties” section for definitions of casualty severity levels. Shading indicates simulated time of occurrence for each event. See table 1 for explanation of aftershock short names and magnitudes] FR510 OK542 PA62 MP552 AT511 PA569 PA522 PA526 MV598 CU64 SV535 SC509 PA501 Commute (5 p.m.) FF558 Night (2 a.m.) SP504 Day (2 p.m.) Casualty severity U523 Time of day Mainshock Number of casualties Level 1 Level 2 Level 3 Level 4  Total Level 1 Level 2 Level 3 Level 4  Total Level 1 Level 2 Level 3 Level 4  Total 12,263 3,007 461 837 16,568 7,827 1,512 179 340 9,858 10,600 2,966 1,300 834 15,700 13 0 0 0 13 14 0 0 0 14 12 0 0 0 12 6 0 0 0 6 7 0 0 0 7 5 0 0 0 5 1 0 0 0 1 1 0 0 0 1 1 0 0 0 1 6 0 0 0 6 5 0 0 0 5 5 0 0 0 5 31 1 0 0 32 36 1 0 0 37 29 1 0 0 30 168 13 1 1 183 146 9 0 0 155 151 17 10 3 181 9 0 0 0 9 8 0 0 0 8 7 0 0 0 7 9 0 0 0 9 8 0 0 0 8 8 0 0 0 8 28 1 0 0 29 28 1 0 0 29 25 2 1 0 28 13 1 0 0 14 12 0 0 0 12 12 0 0 0 12 17 1 0 0 18 15 0 0 0 15 14 1 0 0 15 89 6 0 0 95 76 3 0 0 79 79 8 6 1 94 323 28 2 2 355 327 22 1 1 351 309 37 22 6 374 32 1 0 0 33 33 1 0 0 34 30 2 3 0 35 15 0 0 0 15 14 0 0 0 14 13 0 0 0 13 6 0 0 0 6 5 0 0 0 5 5 0 0 0 5 Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  39 Hazus building damage states combined with weighting factors (or displacement probabilities) for households residing in buildings in each of the various damage states, given in table 16. The Hazus model for the number of people seeking publiclyprovided, short-term shelter recognizes that only a portion of displaced households will actually use public shelter resources; the model uses population data and weighting factors, which reflect the income, ethnicity, ownership, and age of the residents in the displaced households to estimate the fraction expected to seek public shelter (for additional details, refer to chapter 14 of the Hazus Technical Manual; FEMA, 2012.). It should be noted, however, that the default values for the age and ownership weighting factors are set to zero, effectively removing these factors from consideration, as shown in table 17. Table 18 provides the Hazus estimates of displaced households and people seeking shelter for the HayWired scenario mainshock and each of the aftershocks. These results reflect the shelter impacts related to ground shaking only (liquefaction is not included) and were estimated using the Hazus default shelter model parameters. Expected household displacement in the aftershocks is generally minimal, except in the four larger events. Custom shelter parameters were developed for the ShakeOut scenario (Jones and others, 2008), based on available population survey data from the 1994 Northridge and other California earthquakes (Seligson, 2008), as described in the appendix. The default Hazus parameters and the modified parameters used in the ShakeOut analysis are provided in table 19. Adjustments included modifying the parameters for the percentage of households seeking shelter based on building damage state to include some shelter-seeking behavior for occupants of moderately damaged residential structures (normally set to zero in the Hazus default) and modifications Table 16.  Default damage state factors for the Hazus population displacement model. Model parameter Description Default value WSFM Displacement weight for single-family homes in the moderate damage state 0.0 WSFE Displacement weight for single-family homes in the extensive damage state 0.0 WSFC Displacement weight for single-family homes in the complete damage state 1.0 WMFM Displacement weight for multifamily residences in the moderate damage state 0.0 WMFE Displacement weight for multifamily residences in the extensive damage state 0.9 WMFC Displacement weight for multifamily residences in the complete damage state 1.0 Table 17.  Default weighting and modification factors for the Hazus short-term shelter model. Model parameter Description Default value AW Age weighting factor 0 EW Ethnicity weighting factor 0.27 IW Income weighting factor 0.73 OW Ownership weighting factor 0 AM1 Modification factor for percentage of population under 16 years old 0.4 AM2 Modification factor for percentage of population between 16 and 65 years old 0.4 AM3 Modification factor for percentage of population over 65 years old 0.4 EM1 Modification factor for ethnicity: White households 0.24 EM2 Modification factor for ethnicity: Black households 0.48 EM3 Modification factor for ethnicity: Hispanic households 0.47 EM4 Modification factor for ethnicity: Asian households 0.26 EM5 Modification factor for ethnicity: Native American households 0.26 IM1 Modification factor for household income <$10,000 0.62 IM2 Modification factor for household income $10,000–$15,000 0.42 IM3 Modification factor for household income $15,000–$25,000 0.29 IM4 Modification factor for household income $25,000–$35,000 0.22 IM5 Modification factor for household income >$35,000 0.13 OM1 Modification factor for percentage of households that are owner occupied 0.4 OM2 Modification factor for percentage of households that are renter occupied 0.4 40   The HayWired Earthquake Scenario—Engineering Implications Table 18.  Displaced households and shelter demands from ground shaking caused by the HayWired earthquake scenario mainshock and aftershock sequence, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). See table 1 for explanation of aftershock short names and magnitudes] HayWired scenario event Displaced households Mainshock UC523 SP504 FF558 FR510 OK542 PA62 MP552 AT511 PA569 PA522 PA526 MV598 CU64 SV535 SC509 PA501 64,410 2 2 0 0 19 741 3 2 46 3 5 152 1,880 18 3 1 People seeking shortterm shelter 47,009 1 2 0 0 17 408 2 1 24 2 3 83 1,080 11 2 0 Table 19.  Hazus shelter model default and custom parameter values developed for the ShakeOut scenario (see appendix). [Hazus, Federal Emergency Management Agency (2012)] Model parameter Description Hazus default value Custom ShakeOut value WSFM Displacement weight for single-family homes in the Moderate damage state 0.0 0.2 WSFE Displacement weight for single-family homes in the Extensive damage state 0.0 0.4 WSFC Displacement weight for single-family homes in the Complete damage state 1.0 1.0 WMFM Displacement weight for multifamily residences in the Moderate damage state 0.0 0.4 WMFE Displacement weight for multifamily residences in the Extensive damage state 0.9 0.65 WMFC Displacement weight for multifamily residences in the Complete damage state 1.0 AW Age weighting factor 0 No change EW Ethnicity weighting factor 0.27 No change IW Income weighting factor 0.73 No change OW Ownership weighting factor 0 No change EM1 Modification factor for ethnicity: White households 0.24 0.1 EM2 Modification factor for ethnicity: Black households 0.48 0.2 EM3 Modification factor for ethnicity: Hispanic households 0.47 0.2 EM4 Modification factor for ethnicity: Asian households 0.26 0.1 EM5 Modification factor for ethnicity: Native American households 0.26 0.1 IM1 Modification factor for household income <$10,000 0.62 0.3 IM2 Modification factor for household income $10,000–$15,000 0.42 0.3 IM3 Modification factor for household income $15,000–$25,000 0.29 0.14 IM4 Modification factor for household income $25,000–$35,000 0.22 0.08 IM5 Modification factor for household income >$35,000 0.13 0.05 1.0 Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  41 Table 20.  Displaced households and short-term shelter requirements for the HayWired earthquake scenario mainshock, including liquefaction, in counties of the San Francisco Bay region, California, estimated using Hazus with default and custom ShakeOut shelter parameters. [Data from Hazus (Federal Emergency Management Agency, 2012)] Hazus default shelter parameters Displaced People seeking households short-term shelter 51,975 38,430 12,483 8,641 128 121 0 0 1 1 1 1 0 0 32 29 2,251 1,265 11 9 1,908 1,104 7,649 5,641 9 13 52 39 0 0 1 1 0 0 76,501 55,295 County Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Sonoma Stanislaus Yolo  Total to the income and ethnicity factors considered by the model. It should be noted that, based on the default shelter category weights (table 17), income is a more important indicator of shelter-seeking behavior than ethnicity (that is, IW is larger than EW), and those with lower incomes are more likely to seek shelter than those with higher incomes (IM1>IM2>IM3). The modifications made for the ShakeOut analysis were based on data from recent California earthquakes and resulted in reductions in the proportions of the population expected to seek shelter at all income levels. A comparison of the displacement and shelter estimates for the HayWired scenario mainshock and those estimated using the custom ShakeOut shelter parameters is provided in table 20; both of these Custom ShakeOut shelter parameters Displaced People seeking shorthouseholds term shelter 87,629 28,922 21,856 6,623 513 155 5 2 25 12 29 9 9 3 54 23 11,741 2,986 218 87 6,167 1,640 24,179 7,408 89 48 310 110 14 4 39 16 4 2 152,881 48,050 analyses include the impacts of liquefaction. The net result of applying the custom ShakeOut shelter parameters was an increase in the estimated number of displaced households from 76,500 to more than 150,000 and a slight decrease in the number of people seeking shelter from 55,000 to 48,000. Combining Losses in the Mainshock and Aftershocks Two approaches have been used to bound the estimate of potential additional losses suffered as a result of ground Table 21.  Combined losses from ground shaking caused by the HayWired earthquake scenario mainshock and aftershocks, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). Loss reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23] Combined event Building damage Total direct economic loss (millions of dollars) Mainshock Maximum aftershock All aftershocks Net building damage ratio (percent) 30,312.4 46,633.7 2.9 3,478.7 5,380.8 0.3 7,669.4 11,767.1 0.7 Lower Bound 1: mainshock + maximum aftershock 33,791.0 52,014.5 3.2 Lower Bound 2: maximum individual event 30,551.1 47,015.4 2.9 Upper Bound: mainshock + all aftershocks 37,981.8 58,400.8 3.6 42   The HayWired Earthquake Scenario—Engineering Implications 123° 122° 121° YOLO NAPA SONOMA SACRAMENTO SOLANO SOLANO MARIN 38° SAN JOAQUIN CONTRA COSTA SAN FRANCISCO ALAMEDA EXPLANATION County boundary STANISLAUS SAN MATEO SANTA CLARA HayWired aftershocks 37° MERCED SANTA CRUZ IC CIF PA SAN BENITO OC N EA Event causing max DR None CU64 FF558 FR510 MV598 OK542 PA522 PA62 SP504 UC523 MONTEREY 00 Coordinate System: GCS North American 1983 Datum: North American 1983 Area CALIF of maps 00 1 10 10 1 20 2 2 20 30 3 3 30 40 4 4 40 5 MILES 50 MILES 50 KILOMETERS 5 KILOMETERS Figure 14.  Map showing the HayWired earthquake scenario aftershock causing the maximum building damage ratio (max DR) by census tract, the San Francisco Bay region, California. See table 1 for explanation of aftershock short names and magnitudes. Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  43 shaking in the aftershock sequence; a lower bound estimate for each census tract has been developed as the sum of the mainshock shaking losses and shaking losses from the aftershock causing the maximum damage for that census tract, and an upper bound estimate has been developed as the sum of shaking losses in the mainshock and all of the aftershocks. The aftershocks generating the maximum building damage in each census tract are identified in figure 14. For most census tracts, the largest aftershock (the M6.4 Cupertino) causes the largest building damage ratio. Other events causing maximum damage ratios for selected census tracts include the M5.58 Fairfield, M5.10 Freemont, M5.98 Mountain View, M5.42 Oakland, M5.22 and M6.2 Palo Alto, M5.04 San Pablo, and M5.23 Union City aftershocks. In some cases, a more distant aftershock causes larger damage ratios than a nearby aftershock, as can be seen in parts of Sonoma County, where the M6.2 Palo Alto aftershock generates higher damage ratios than closer aftershocks. However, the damage ratios in Sonoma County are less than 0.1 percent for all aftershocks (figs. 8–11). The upper and lower bound estimates of total loss in HayWired scenario events are summarized in table 21. Shaking damage in the modeled aftershock sequence adds between $3.5 billion (maximum aftershock) and $7.7 billion (all aftershocks) to the building damage total—an additional 11 to 25 percent, respectively. As an alternate to the lower bound estimate calculated above, one could use the maximum loss in all events (that is, the maximum shaking loss occurring in either the mainshock or any of the aftershocks). For most census tracts, this maximum shaking loss occurs in the mainshock, but 49 census tracts (out of 2,122 in the 17-county study area) suffer larger shaking losses in one of the aftershocks, including 12 census tracts where the M6.2 Palo Alto aftershock dominates the losses, 35 tracts where the M6.4 Cupertino event generates the largest building damage, and 2 census tracts where the M5.58 Fairfield event generates the largest building damage ratio; these tracts are mapped in figure 15, and the resulting losses are provided in table 21 as Lower Bound 2. This alternate lower bound assumes that no other earthquake adds any damage to the maximum damage from one earthquake in each census tract and is lower than the initial version; total building damage is $30.6 billion versus $33.8 billion in the initial lower bound estimate. The various resulting census tract level building damage ratio maps are shown in figures 16 and 17, in the same order as presented in table 21. Figure 16 presents the census tract shaking damage ratio map for the HayWired mainshock, the maximum aftershock, the sum of all aftershocks, and the sum of the mainshock and the maximum aftershock (Lower Bound 1). Figure 17 provides the final two shaking damage ratio maps for the maximum of mainshock and aftershock (Lower Bound 2) and the mainshock plus all aftershocks (Upper Bound). Repeat Liquefaction To examine losses resulting from repeated liquefaction in subsequent events, liquefaction impacts within Alameda County (evaluated using the Hazus default liquefaction approach and uniform depth to groundwater assumption) were reviewed at the census tract level for the mainshock and the M5.42 Oakland aftershock. Alameda County was selected for its proximity to the epicenter of the mainshock and the epicenter of the Oakland aftershock, as well as the presence of highly liquefiable soils (see fig. 2). Maps of expected liquefaction/ lateral spread displacements as estimated by Hazus using its default methodology, considering liquefaction susceptibility, depth to groundwater (assumed shallow), and probability of liquefaction, are provided in figure 18. As shown, many of the bay-front census tracts are estimated to have substantial expected lateral spread in the mainshock, with some additional smaller movements in the M5.42 Oakland aftershock. As has been noted, Hazus is not able to estimate additional damage to damaged buildings; each of the current analyses is independent and assumes that the inventory is in an undamaged condition before the earthquake. Although this may underestimate cumulative damage, it is still instructive to examine the potential impact of repeated liquefaction in liquefaction prone areas. In Alameda County, 264 tracts are subject to liquefactioninduced lateral spread displacements in the mainshock; 131 of these tracts are subject to additional lateral spread displacements in the M5.42 Oakland aftershock. Shaking and liquefaction losses in these 131 tracts are summarized in table 22. Even though liquefaction is less widespread in the aftershock, and expected liquefaction damage is minimal compared to the mainshock ($77 million versus $1.45 billion), liquefaction damage is predicted by Hazus even in events with magnitudes less than 6. And in these smaller events, liquefaction has the potential to represent a larger proportion of the overall damage; in these 131 tracts, building damage caused by liquefaction represents 21 percent of the total building damage caused by shaking and liquefaction in the mainshock but increases to 26 percent in the aftershock. 44   The HayWired Earthquake Scenario—Engineering Implications 123° 122° YOLO SONOMA NAPA SACRAMENTO SOLANO MARIN 38° SAN JOAQUIN CONTRA COSTA FIC CI PA SAN FRANCISCO N EA OC ALAMEDA SAN MATEO SANTA STANISLAUS CLARA EXPLANATION SANTA CRUZ County boundary 37° MERCED HayWired mainshock and aftershocks SAN Event causing max DR BENITO HayWired mainshock FF558 CU64 MONTEREY PA62 00 00 1 10 10 1 20 2 2 20 30 3 3 30 40 4 4 40 5 MILES 50 MILES 50 KILOMETERS 5 KILOMETERS Coordinate System: GCS North American 1983 Datum: North American 1983 Area CALIF of maps Figure 15.  Map showing the HayWired earthquake scenario mainshock or aftershock causing the maximum building damage ratio (max DR) by census tract, San Francisco Bay region, California. See table 1 for explanation of aftershock short names and magnitudes. Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  45 B. Maximum aftershock A. Mainshock 123° 122° 121° 123° 122° YOLO SONOMA YOLO NAPA SACRAMENTO SONOMA NAPA SACRAMENTO SOLANO SOLANO MARIN MARIN 38° CONTRA COSTA SAN JOAQUIN SAN FRANCISCO SAN JOAQUIN CONTRA COSTA SAN FRANCISCO ALAMEDA HayWired M 7.05 mainshock ALAMEDA STANISLAUS SAN MATEO IC CIF PA SAN MATEO IC CIF PA SANTA CLARA SANTA CRUZ MERCED STANISLAUS SANTA CLARA SANTA CRUZ EAN OC EAN OC 37° 121° MERCED EXPLANATION SAN BENITO SAN BENITO County boundary MONTEREY Building damage ratio MONTEREY (damage divided by replacement value, expressed as a percent) 0 <0.1% 0.1–0.5% 0.5–2.5% 2.5–5.0% 5.0–10.0% >10.0% C. Cumulative aftershock YOLO SONOMA NAPA SACRAMENTO D. Mainshock + maximum aftershock Area CALIF of maps SONOMA YOLO NAPA SACRAMENTO SOLANO SOLANO MARIN MARIN 38° CONTRA COSTA SAN JOAQUIN SAN FRANCISCO SAN FRANCISCO ALAMEDA SAN MATEO ALAMEDA STANISLAUS IC CIF PA IC CIF PA SANTA CLARA SANTA CRUZ MERCED EAN OC EAN OC 37° SAN JOAQUIN CONTRA COSTA SAN MATEO STANISLAUS SANTA CLARA SANTA CRUZ MERCED SAN BENITO SAN BENITO MONTEREY Coordinate System: GCS North American 1983 Datum: North American 1983 MONTEREY 00 10 1 20 2 30 3 40 4 50 MILES 5 MILES 1 20 2 30 3 40 4 50 5 KILOMETERS 00 10 KILOMETERS Figure 16.  Maps showing building damage ratio from ground shaking caused by the HayWired earthquake scenario mainshock and aftershocks, San Francisco Bay region, California. A, Mainshock building damage ratio; B, Maximum aftershock building damage ratio; C, Cumulative aftershock building damage ratio; and D, Mainshock plus maximum aftershock building damage ratio (lower bound 1). 46   The HayWired Earthquake Scenario—Engineering Implications A. Maximum event B. Cumulative 123° 122° 121° 123° 122° YOLO YOLO SONOMA NAPA SACRAMENTO SONOMA NAPA SACRAMENTO SOLANO SOLANO MARIN 38° MARIN SAN JOAQUIN CONTRA COSTA SAN FRANCISCO SAN JOAQUIN CONTRA COSTA SAN FRANCISCO ALAMEDA ALAMEDA STANISLAUS SAN MATEO SANTA CLARA MERCED SANTA CRUZ STANISLAUS SAN MATEO SANTA CLARA IC CIF PA IC CIF PA MERCED SANTA CRUZ OC EAN EAN OC 37° 121° SAN BENITO SAN BENITO MONTEREY MONTEREY EXPLANATION County boundary Building damage ratio (damage divided by replacement value, expressed as a percent) Coordinate System: GCS North American 1983 Datum: North American 1983 Area CALIF of maps 0 <0.1% 0.1–0.5% 0.5–2.5% 2.5–5% 5–10% >10% 00 00 10 1 20 2 30 3 40 4 50 MILES 5 MILES 1 20 2 30 3 40 4 50 5 KILOMETERS 10 KILOMETERS Figure 17.  Maps of building damage ratio from ground shaking caused by the HayWired earthquake scenario mainshock and aftershocks, San Francisco Bay region, California. A, Maximum individual event building damage ratio (lower bound 2). B, Cumulative (mainshock plus all aftershocks) building damage ratio (upper bound). Model Limitations and Data Gaps Hazus was developed by FEMA to “produce loss estimates for use by Federal, State, regional and local governments in planning for earthquake risk mitigation, emergency preparedness, response and recovery” (FEMA, 2012). At the time of its development in the early 1990s, its methods were considered state of the art; little has changed in the underlying earthquake methodology since its original development. With time, other more comprehensive analytical methodologies have been developed by the earthquake engineering community, such as Performance Based Earthquake Engineering (see, for example, Applied Technology Council, 2006). Nevertheless, the Hazus framework still provides the only publicly available software and methodology that allows for regional earthquake analysis anywhere in the United States using pre-packaged (default) building inventory data. Studies directly comparing the results of the various approaches have yet to be conducted. It should be noted that although the inventory data used in the HayWired analyses were enhanced with improved mapping schemes and escalated replacement values, other exposure data (building square footage and building counts) were Hazus defaults based on 2000 census data for residential occupancies and 2006 Dun & Bradstreet data for non-residential occupancies. With the release of Hazus 2.2 in January of 2015, updated Hazus inventory data reflecting 2010 census data for residential occupancies are now available. Because these new data are based on revised census geographies, they are, unfortunately, currently incompatible with the improved mapping schemes developed for the San Francisco Bay region based on the 2000 census geometries. In the future, to avoid obsolescence of the improved San Francisco Bay region mapping schemes, an evaluation and conversion to the 2010 census geographies will need to be performed. Chapter J. HayWired Scenario—Hazus Analyses of the Mainshock and Aftershocks  47 122.5° 122° A. Mainshock MARIN HayWired M 7.0 mainshock epicenter CONTRA COSTA N SA SAN FRANCISCO O SC CI AN FR Y BA EXPLANATION County boundary Alameda County census tracts ALAMEDA Expected lateral spread 37.5° None <5 inches 5–10 inches 10–15 inches 15–20 inches SAN MATEO SANTA CLARA B. M 5.42 Oakland Oakland M 5.42 epicenter SAN FRANCISCO CONTRA COSTA N SA O SC CI AN FR Y BA PACIFIC OCEAN MARIN ALAMEDA SAN MATEO 37.5° SANTA CLARA 0 Coordinate System: GCS North American 1983 Datum: North American 1983 0 Area of map 5 5 10 MILES 10 KILOMETERS CALIF Figure 18.  Maps showing expected liquefaction-induced lateral spread displacements by census tract in Alameda County, California, estimated using the Hazus (FEMA, 2012) default approach. A, Spread from mainshock. B, Spread from magnitude (M) 5.42 Oakland aftershock. 48   The HayWired Earthquake Scenario—Engineering Implications Table 22.  Liquefaction-induced building damage for 131 census tracts in Alameda County, California, suffering lateral spread displacements from the HayWired earthquake scenario mainshock and Oakland M5.42 aftershock, estimated using Hazus. [Data from Hazus (Federal Emergency Management Agency, 2012). See table 1 for explanation of aftershock short names and magnitudes. Damage reported in 2005 dollars; U.S. Consumer Price Index 2016:2005 ratio is approximately 1.23] HayWired scenario event Building damage ratio (percent) Mainshock, shaking only 6,802 12.7 Mainshock, shaking and liquefaction 8,254 15.4 Mainshock, liquefaction 1,452 2.7 OK542, shaking only 293 0.55 OK542, shaking and liquefaction 369 0.69 77 0.14 OK542, liquefaction The current study uses Hazus to assess damage and loss in 17 earthquake events ranging from M5.0 to M7.0. Hazus’ known limitations include increased uncertainty for small events, wherein empirical data from past events may be lacking. As stated in the Hazus Earthquake Technical Manual, “…the losses from small magnitude earthquakes ($35,000 Old parameter 0.27 0.73 0.24 0.48 0.47 0.26 0.26 0.62 0.42 0.29 0.22 0.13 Revised parameter No change No change 0.1 0.2 0.2 0.1 0.1 0.3 0.3 0.14 0.08 0.05 future modeling. Given that utility outage is highly correlated with building damage, utilizing the lower bound parameter of 20 percent would be appropriate. A final estimate of displacement would be 120 percent of the Hazus estimate. References Cited Jones, L.M., Bernknopf, R., Cox, D., Goltz, J., Hudnut, K., Mileti, D., Perry, S., Ponti, D., Porter, K., Reichle, M., Seligson, H., Shoaf, K., Treiman, J., and Wein, A., 2008, The ShakeOut Scenario: U.S. Geological Survey Open-File Report 2008–1150 and California Geological Survey Preliminary Report 25, 312 p. and appendixes, accessed April 12, 2017, at https://pubs.usgs.gov/of/2008/1150/. Bourque, L.B., Siegel, J.M., and Shoaf, K.I., 2002, Psychological distress following urban earthquakes in California: Prehospital and Disaster Medicine, v. 17, no. 2, p. 81–90, accessed October 2, 2012, at http://journals.cambridge.org/download.php?file=/ PDM/PDM17_02/S1049023X00000224a.pdf&code=2bc1e69b 4f70ad90559d97c60a9e146d. Shoaf, K.I, and Bourque, L.B., 1999, Correlates of damage to residences following the Northridge earthquake, as reported in a population-based survey of Los Angeles County residents: Earthquake Spectra, v. 15, no. 1, p. 145–172, accessed at http://cat.inist.fr/?aModele=afficheN&cps idt=1768150. The HayWired Earthquake Scenario—Engineering Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q [Also see https://doi.org/10.3133/sir20175013] Chapter K Societal Consequences of Current Building Code Performance Objectives for Earthquakes By Keith A. Porter Abstract When discussing how to reduce future building impairment during earthquakes, structural engineers tend to focus on the existing building stock. However, today’s new buildings are tomorrow’s existing buildings, and tomorrow’s building stock will not be risk free. The leading model building code in the United States, the International Building Code (IBC), aims to protect the safety of life by providing that a newly engineered building will have no more than a 1-percent probability of collapse in the coming 50 years. However, what about damage short of collapse? Red-tagged (rendered unsafe to enter or occupy) and yellow-tagged (safe only for limited use) buildings can significantly impact the performance of the building stock as a whole in the months or years after a large, but not exceedingly rare, earthquake, such as that simulated in the HayWired earthquake scenario. The HayWired scenario examines a hypothetical earthquake with a moment magnitude (Mw) of 7.0 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of California’s San Francisco Bay area. I explore the implications of the current building-code objectives for future large earthquakes by asking this question: If the existing building stock were replaced with one that entirely complies with current code requirements, what would be the consequences in the Big One? If recent studies for the Building Seismic Safety Council, Federal Emergency Management Agency, and the National Institute of Standards and Technology are correct about the collapse fragility of engineered buildings, a magnitude (M) 7 earthquake on the Hayward Fault could result in the collapse of 8,000 buildings and nearly 500,000 buildings could be impaired—red- or yellow-tagged—leaving businesses and households with no or restricted use of the buildings they occupy. Low vacancy rates for space in the San Francisco Bay area would likely lead to the loss of a large number of residents and businesses for the region. However, there are simple and relatively inexpensive alternatives. If cities required all new buildings to be constructed with an importance factor of 1.5 (a relatively simple local modification to the IBC), construction costs would be about 1 to 3 percent higher overall and could reduce building impairment by 75 percent, keeping several thousand residents in their homes and thousands of businesses in operation in the event of a M7 earthquake. Additionally, preliminary surveys suggest that the public’s expectations for building performance may be higher than current code objectives—that is, the public wants buildings to be habitable after a disaster, beyond having a low probability of collapse. The surveys also indicate that the public would be willing to pay 1–3 percent more in construction cost (about 0.5–1.5 percent more in purchase price) for buildings that meet this objective. Introduction When discussing how to reduce future earthquake losses, structural engineers tend to focus on the existing building stock, especially buildings that are fragile and numerous. Among these are unreinforced masonry (URM) buildings (see, for example, Hess [2008] on URM buildings in the 2008 ShakeOut scenario—a hypothetical moment magnitude, Mw, 7.8 earthquake on the San Andreas Fault in southern California), nonductile concrete moment-resisting frame buildings (see, for example, Taciroglu and Khalili-Tehrani [2008] in the ShakeOut scenario), and large soft-story wood-frame buildings (see, for example, Porter and Cobeen [2012] on San Francisco’s recent effort to manage risk from this building type). However, today’s new buildings are the existing buildings of tomorrow, and the building stock of tomorrow will not be risk free. It is not intended to be. This chapter explores the implications of the current leading building code in the United States for future natural disasters—the International Building Code (IBC). It asks the questions: Where will the current code lead, if its current performance objectives are all met? If the existing building stock were replaced today with one that complies with current seismic design requirements, what would be the consequences? What are some options for cities that adopt building codes and engineers who write them, if those consequences are unsatisfactory? Seismic performance of buildings can be expressed in terms of building collapse, post-earthquake safety, functionality, economic loss, and other measures. This chapter focuses on collapse and other safety impairment in the sense of the 58   The HayWired Earthquake Scenario—Engineering Implications Applied Technology Council’s (ATC) ATC-20 post-earthquake safety inspections. Buildings are red-tagged if they are deemed unsafe to enter or occupy and yellow-tagged if they are deemed safe for restricted use (ATC, 2005). The ATC-20 tagging process is an imperfect and sometimes inconsistent indicator of building safety, but nonetheless dominates cities’ efforts to determine whether buildings are safe to enter and occupy after earthquakes and to control the use of damaged buildings. This chapter is part of the HayWired earthquake scenario, which examines a hypothetical earthquake of Mw 7.0 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of California’s San Francisco Bay area. The simulated earthquake causes Modified Mercalli Intensities of VI–X in the region, with very strong shaking along the fault in the densely populated east bay. In a later volume, the HayWired scenario will address the economic consequences of building damage. The reader may be aware that some of the content in this chapter appears in Porter (2016a). This chapter builds on the findings of Porter (2016a) and introduces new material, including a discussion on the number of impaired buildings in the HayWired scenario, cities’ options for reducing damage in a code-compliant building stock, what-if loss and cost estimates if bay area cities were to require all ordinary buildings to be designed with an importance factor of 1.5, and new research to elicit public understanding and preferences for the performance of a code-compliant building stock. Background First, I review some of the important steps that led to the seismic design requirements in current building codes. Since approximately 1980, engineers have designed California’s buildings to achieve a quantified safety objective—that is, to ensure that the probability of life-threatening damage to any particular structural element or connection is less than a specified maximum value, given the occurrence of a fairly rare and severe level of shaking. The procedure for doing so is called load- and resistance-factor design (LRFD). LRFD was developed in the 1970s and early 1980s. Ellingwood and others (1980) offer one of the seminal works on LRFD, raising LRFD to the status of building-code requirements. In that work, the authors provide design parameters that would ensure a probability of life-threatening damage of no greater than approximately 4 percent given design-level shaking. At that time, design-level shaking was defined with a 10-percent exceedance probability in 50 years, which equates to a mean recurrence interval of 475 years. That 4 percent figure applies individually to each component in a building, including the beam, column, shearwall, brace, and connection, rather than to the building as a whole. In the 1990s, a procedure called performance-based earthquake engineering (PBEE) was developed to manage the seismic performance of existing individual buildings as a whole system, on the basis of four whole-building performance levels: operational, immediate occupancy, life safety, and collapse prevention. See, for example, ATC (1997) for more information. From the early 1970s through the 2000s, researchers working with the Massachusetts Institute of Technology, URS/ John A. Blume and Associates, the Consortium of Universities for Research in Earthquake Engineering (CUREE), the Pacific Earthquake Engineering Research Center (PEER), and the ATC developed another approach to PBEE that quantifies the seismic performance of individual buildings in terms of repair costs, life-safety impacts, and loss of functionality—in other words, dollars, deaths, and downtime. In this study, I refer to the procedure as second-generation performance-based earthquake engineering (PBEE-2). See, for example, Czarnecki (1973), Kustu and others (1982), Beck and others (1999), Porter and Kiremidjian (2001), Porter (2003), Aslani and Miranda (2006), and ATC (2012) for more information. In the late 2000s, researchers and practitioners working with the Federal Emergency Management Agency (FEMA) and the National Institute of Building Sciences quantified the likely seismic performance of code-compliant buildings in terms of the probability that a building would collapse when subjected to very rare shaking (the maximum considered earthquake, MCE). See ATC (2009) and National Earthquake Hazards Reduction Program (NEHRP) Consultants Joint Venture (2012) for more information. These authors recommend using the estimated collapse probability as an acceptable probability of collapse for new buildings, suggesting that “the probability of collapse due to MCE ground motions . . . be limited to 10 percent . . . A limit of twice that value, or 20 percent, is suggested . . . for evaluating the acceptability of potential ‘outliers’ . . .” (ATC, 2009). The skeptical reader might take issue with the ATC (2009) recommendation because it suggests formulating a performance objective only after performance data have already been gathered and examined. However, the ATC authors’ objective was to ensure that new structural systems would achieve a level of reliability consistent with older ones, using collapse probability as the consistent reliability objective. The process is similar to the way that developers aimed to ensure that new design under LRFD would be consistent with the safety of designs performed using allowable stress design, the precursor to LFRD. Also in the late 2000s, some of the same researchers were developing a new basis for building design in which the performance objective was to achieve a uniform probability of collapse in a given time period. In particular, Luco and others (2007) offer a procedure for defining a so-called risk-adjusted maximum considered earthquake (MCER) motion such that, if a new building satisfied LRFD design requirements at shaking based on MCER motion, then the collapse probability in MCER motion is limited to 10 percent and, more to the point, the long-term collapse risk is limited to 1 percent in 50 years (considering all ground motion levels that could occur in those 50 Chapter K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes   59 120° 110° 100° 45° 90° 80° 70° EXPLANATION Spectral response acceleration for 5 hertz, expressed as a fraction of gravitational acceleration (g) 0.8 0.4 0.3 0.2 0.14 0.1 0.06 0.04 0.02 0 40° 35° Areas where suspected nontectonic earthquakes have been deleted 30° 25° 0 0 500 1,000 KILOMETERS 500 1,000 MILES Figure 1.  Map of risk-adjusted maximum considered earthquake (MCER) shaking for the contiguous United States (modified from Petersen and others, 2014). years and their estimated recurrence frequency). To illustrate this concept, a recent map showing MCER ground motion for short-period construction is shown in figure 1. What is meant by “based on MCER motion” is that design acceleration is the product of the mapped MCER motion multiplied by a factor of 2/3, a factor to account for site amplification, a factor to take advantage of ductility, and, in some cases, a factor to reflect greater building importance. Risk-adjusted design has been incorporated into U.S. seismic design guidelines (American Society of Civil Engineers [ASCE], 2010) and the IBC (International Code Council [ICC], 2012). Though they are not both codes, they will hereafter collectively be referred to as “the code,” because ASCE-7 is adopted by reference in the 2012 IBC. To be clear, the primary performance objective of the most recent U.S. seismic design code requirements is to ensure that new engineered buildings have no more than a 1-percent collapse probability over 50 years. There are secondary objectives as well. The 2015 NEHRP Recommended Seismic Provisions state that ordinary buildings in earthquakes will “Avoid serious injury and life loss due to structural collapse, failure of nonstructural components or systems, and release of hazardous materials . . . and reduce structural and nonstructural repair costs where practicable to do so.” The mention of reducing nonstructural repair costs refers to construction requirements that are believed to cost-effectively limit damage, such as bracing large suspended ceilings, as shown in figure 2. However, only the collapse probability is quantified. Note that the code does not aim to ensure buildings are usable or even repairable after an earthquake, but rather that occupants can safely exit a building after designlevel shaking. Engineers sometimes refer to this objective as “life-safety performance,” acknowledging that some buildings are designed stronger (for example, those housing hazardous or essential facilities). The code appears to achieve its life-safety performance objective. Table 1 compares some of the leading threats to life in the United States with the modeled earthquake lifesafety risk of continuous 24 hours a day, 7 days a week (24/7) occupancy of a building designed to meet the current building code. The number of deaths in new buildings during earthquakes assumes that the average new building somewhat exceeds the minimum safety requirement, exhibiting a 0.6 percent probability of collapse in 50 years instead of 1 percent. It also assumes that collapse affects 25 percent of the building area, as described in Porter (Urban Search and Rescue, this volume), and that 10 percent of occupants located in collapsed building areas die as a result of the collapse (FEMA, 2012a). Deaths in uncollapsed parts of the building are omitted from this calculation. Thus, the mortality rate can be estimated as 0.006/50 years×0.25×0.10=3×10-6/year, or 0.3 per 100,000 people per year. The figure would apply to 24/7 occupancy of buildings that comply with the 2012 IBC—that is, in buildings with earthquake loads (as opposed to wind loads) that govern the strength of the lateral force resisting system. 60   The HayWired Earthquake Scenario—Engineering Implications Table 1.  Mortality rates due to building collapse during earthquakes compared with other causes of death in the United States. Peril Deaths per 100,000 people per year Where and when Heart disease 194 United States 2010 (Heron, 2013) All accidents 39 United States 2010 (Heron, 2013) Occupational fatality, roofers 32 United States 2011 (Bureau of Labor Statistics, 2013) Auto accidents 11 United States 2009 (U.S. Census Bureau, 2012) Firearms 10 United States 2010 (Heron, 2013) New building collapse during earthquakes 0.3 24/7 occupancy, wherever earthquake loads govern the strength of the lateral force resisting system Earthquakes past ~50 years 0.007 California 1964–2014 (Wikipedia [2017], drawing on a variety of sources) Implications of Seismic Performance Objectives for the HayWired Scenario The foregoing brief history discussion of seismic performance objectives raises some interesting questions that can be explored with the HayWired scenario. 1. Quantitative building-code objectives are expressed in terms of collapse risk during a 50-year building design life, considering all possible levels of shaking and their likelihood of occurrence. What about other modes of building impairment, namely red- and yellow-tagging? As used here, a red tag refers to a placard attached to an earthquake-damaged building that declares the building unsafe to enter or occupy. A yellow tag declares the building to be damaged such that its use is restricted. Restrictions might prevent the use of part of the building or they might allow only brief occupancy (for example, to remove possessions). See figure 3. 2. Quantitative code objectives are also expressed in terms of collapse probability in MCER shaking. However, in a large earthquake, like that in the HayWired scenario, only a small geographic area might experience MCER shaking or greater. What happens at lower levels of shaking? 3. Quantitative code objectives are expressed in terms of collapse probability for a single building. What happens at the societal level, for example, in terms of the fraction or number of impaired buildings in the epicentral region? 4. Suppose a city or a state, such as California, adopted the IBC but decided to modify it to require stronger buildings. What would be the result in terms of the fraction or number of impaired buildings in the epicentral region? Stiffness also matters for building performance, especially repair cost and duration, but I focus here on strength, which more strongly affects collapse and red tagging. See Multihazard Mitigation Council (2017) for a discussion on the effects of greater stiffness as well as strength. The answers to these questions can inform the public’s understanding and expectations for code-compliant buildings and the understanding of their elected officials who adopt model building codes like the IBC. The remainder of this report seeks to answer questions 1–4. After an earthquake, red- and yellow-tagged buildings can be impaired for months. Comerio (2006), who studied building restoration after the 1994 Northridge earthquake, wrote that: Figure 2.  Photograph showing examples of building-code requirements to limit nonstructural repair costs. Note the diagonal wires and vertical post (called a compression post) that support the suspended ceiling. (Photograph from Federal Emergency Management Agency, 2012a.) Initial inspections of more than 440,000 units listed 7,000 single-family homes, 2,000 mobile homes, and 49,000 multifamily units [as] red- and yellowtagged. Three years after the event, when insurance claims were tallied, it became clear that moderate damage to single-family homes was under-counted in the post-event inspections, as more than 195,000 homeowners made insurance claims, for an average of $30,000 to $40,000 . . . About 40 percent of homeowners began repairs within one year. . .For Chapter K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes   61 the remainder, it took two to three years to resolve the insurance claim, and repairs were likely to be delayed until the insurance funding was available. The time needed for repairs of large apartment and condominium buildings was often longer. Comerio’s (2006) findings imply that even a yellow-tagged building may be unavailable for months or more after the mainshock earthquake. In light of Comerio’s observation that 58,000 housing units were red- or yellow-tagged, but 195,000 homeowners made insurance claims averaging $30,000 to $40,000, it seems that more than 3 times the number of collapsed, red-tagged, and yellow-tagged buildings may be at least moderately damaged—that is, with repair costs around $30,000 to $40,000. Whereas this study focuses on impaired— that is, collapsed, red-tagged, and yellow-tagged—buildings, it is important to keep in mind that many more buildings may be adversely affected by earthquake shaking. Impairment Rate at MCER Shaking To quantify impairment rates of buildings at MCER shaking, I begin with some observations. Figure 4 summarizes the results of incremental dynamic analyses of a large number of sample buildings (NEHRP Consultants Joint Venture, 2012). The figure shows the fraction of nonlinear dynamic analyses at MCER-level spectral acceleration response in which the building is estimated to have collapsed, as a function of the building’s estimated fundamental period of vibration (“design period” in the figure). In private conversation, the lead author expressed the opinion that many of the sample buildings ought to be excluded from this analysis, especially the circled data points and all of the data for buildings with design period less than 0.5 seconds (C. Kircher, oral commun., May 8, 2014). The former should be excluded because they would not A comply with the current building code, the latter because he believes the structural analyses incorrectly associated a threshold level of peak transient drift ratio with collapse. Excluding these and taking the average of the remaining data suggests the average probability of collapse is 6 percent at MCER shaking. In addition to collapsed buildings, how many buildings are red- or yellow-tagged in MCER shaking? Unfortunately, there are no analytical studies of red and yellow tagging comparable to the studies of collapse by ATC (2009) and NEHRP Consultants Joint Venture (2012). Table 2 lists evidence for the ratio of red tags to collapses, where both are known—an admittedly limited dataset. The data suggest approximately 13 buildings are red-tagged for every collapsed building. Table 3 shows evidence for 3.8 yellow tags per red tag. Let r denote the ratio of impaired buildings to collapsed buildings. With 13 red tags per collapse and 3.8 yellow tags per red tag, r=63 impaired buildings for each collapsed building (1+13+13×3.8=63.4), including the collapsed building. The data in tables 2 and 3 do not reflect a code-compliant stock and they do not account for variability with shaking intensity. However, using this empirical evidence is preferable to using Hazus-MH software (FEMA, 2012b) or another forward analytical model, because engineers generally prefer to rely on empirical evidence over analytical results. Remember, I am only using evidence from past earthquakes to estimate ratios of red tags to collapse and yellow tags to red tags, not rate of collapse. Table 4 summarizes the implications of these findings for the building stock in a small area that experienced MCER shaking. The values in table 4 rely solely on the building code’s explicit objective (10 percent collapse probability conditioned on MCER shaking), reduced to provide a best estimate rather than an upper limit (6 percent rather than 10 percent), and that the ratio of red and yellow tags to collapses of new buildings would be the same as has been observed for the 1989 Loma Prieta and 1994 Northridge earthquakes. B UNSAFE DO NOT ENTER OR OCCUPY (THIS PLACARD IS NOT A DEMOLITION ORDER) This structure has been inspected, found to be seriously damaged and is unsafe to occupy, as described below: Date Date Time (Caution: Aftershocks since inspection may increase damage and risk.) Time This facility was inspected under emergency conditions for: (Jurisdiction) Do not enter, except as specifically authorized in writing by jurisdiction. Entry may result in death or injury. RESTRICTED USE Caution: This structure has been inspected and found to be damaged as described below: Entry, occupancy, and lawful use are restricted as indicated below: (Jurisdiction) Inspector ID / Agency Inspector ID / Agency Facility Name and Address: Do Not Remove, Alter, or Cover this Placard until Authorized by Governing Authority This facility was inspected under emergency conditions for: Facility Name and Address: Do Not Remove, Alter, or Cover this Placard until Authorized by Governing Authority Figure 3.  Images of (A) red and (B) yellow tags used to indicate the level of impairment to an earthquake-damaged building (images from Applied Technology Council, 2005). 62   The HayWired Earthquake Scenario—Engineering Implications A 35 EXPLANATION Bearing-wall systems Building-frame systems Moment-frame systems Collapse probability, in percent 30 25 20 15 10 5 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Design period (T ), in seconds B 35 EXPLANATION Bearing-wall systems Building-frame systems Moment-frame systems Collapse probability, in percent 30 25 20 15 10 5 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Design period (T ), in seconds Figure 4.  Graphs showing collapse probability conditioned on risk-adjusted maximum considered earthquake (MCER) shaking versus design period for buildings examined by NEHRP Consultants Joint Venture (2012). The buildings were subjected to multiple nonlinear dynamic structural analyses at a level of motion associated with MCER shaking. A, Graph showing all buildings examined by NEHRP Consultants Joint Venture (2012). B, Graph excluding the buildings that the lead author of NEHRP Consultants Joint Venture (2012) thought should be excluded from the analysis, because they would not comply with current code or did not correctly associate peak transient drift ratio with collapse (C. Kircher, oral commun., May 8, 2014). The remaining data define an average collapse probability of 6 percent for all design periods considered (dashed red line). (Images modified from NEHRP Consultants Joint Venture, 2012.) Chapter K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes   63 Table 2.  Red tags and building collapses in select California earthquakes. [NIST, National Institute of Standards and Technology; SEAONC, Structural Engineers Association of Northern California; EQE, EQE International; Cal OES, California Governor’s Office of Emergency Services, Geographic Information Systems Group] Earthquake Red tags Collapse Reference 1989 Loma Prieta, San Francisco Marina District 110 7 NIST (1990), Harris and others (1990) 1989 Loma Prieta, City of Santa Cruz1 100 40 SEAONC (1990), Fradkin (1999) 1994 Northridge2 2,157 133 EQE and Cal OES (1995) Total 2,367 180 Ratio=13:1 One hundred red tags is an estimate; 300 red tags county wide, factored by number of structures in city versus county, and reduced by number of collapses to avoid double counting. 1 2 One hundred thirty-three collapsed buildings is taken from the ATC-20 form data in an unpublished database described by EQE and Cal OES (1995); red tags reduced by number of collapses to avoid double counting. Table 3.  Yellow and red tags in select California earthquakes. [SEAONC, Structural Engineers Association of Northern California; EQE, EQE International; Cal OES, California Governor’s Office of Emergency Services, Geographic Information Systems Group] Earthquake Yellow tags Red tags 1989 Loma Prieta 3,330 1,114 SEAONC (1990) 1994 Northridge 9,445 2,290 EQE and Cal OES (1995) 12,775 3,404 Ratio=3.8:1 Total Reference Table 4.  Performance of new buildings in a small area subjected to risk-adjusted maximum considered earthquake (MCER) shaking. Building condition Percent/ratio Fraction of stock Collapsed 6 percent of stock 6 percent Red-tagged and not collapsed 13 red tags per collapse 78 percent Yellow-tagged 3.8 yellow tags per red tag Most of the remaining stock Validation Using Data from the August 2014 Napa Earthquake Data acquired from the City of Napa, California, after the completion of this analysis can be compared with the data in tables 2 and 3. As of October 21, 2014, the City of Napa recorded 1,767 yellow tags and 175 red tags (F. Turner, California Seismic Safety Commission, written commun., March 10, 2015). Damage descriptions acquired from the City of Napa in a separate database (K. Wallis, GIS Coordinator for City of Napa, written commun., March 12, 2015) suggest 34 structures appear to meet the definition of collapse. According to FEMA P–154 (ATC, 2015), building collapse means that any part of the gravity system experienced dynamic instability leading to the loss of load-bearing capacity. The dynamic instability leads to severe structural deformation of a potentially life-threatening nature, especially falling of all or portions of a structure. Note that, as used here, partial building collapse means that the dynamic instability occurs only in a portion of the building. The 34 collapses are included among the yellow and red tags, such that the ratio of impaired buildings to collapses is (1,767+175)/34=57:1. The ratio of yellow to red is higher in the 2014 Napa earthquake than in the Loma Prieta and Northridge earthquakes and the ratio of red tags to collapse is lower. These two ratios counteract each other to result in a ratio of impaired buildings to collapses (57:1) that closely approximates those from the Loma Prieta and Northridge earthquakes (63:1). Impairment Rate at Other Levels of Shaking Until now, I have considered the impairment rates of a code-compliant building stock that is shaken to exactly MCER shaking. Impairment can be estimated at other levels of shaking as well. It is probably higher at higher levels of shaking, and lower at lower levels of shaking. Because the strength of new buildings varies from place to place depending largely on MCER, I define new measures of shaking called demandto-design ratio (DDR), with two varieties: DDRS and DDR1 defined as follows. • DDRS is the 5-percent damped spectral acceleration response at a period of 0.2 seconds in a particular earthquake at a particular location divided by SMS— the MCER 5-percent damped spectral acceleration response parameter at short periods adjusted for site class effects, as defined in American Society of Civil Engineers/Structural Engineering Institute (ASCE/ SEI) 7-10 section 11.4.3. Note that the subscript in DDRS indicates short period, similar to that in SMS. 64   The HayWired Earthquake Scenario—Engineering Implications • DDR1 is the 5-percent damped spectral acceleration response at a period of 1.0 second in a particular earthquake at a particular location divided by SM1— the MCER 5-percent damped spectral acceleration response parameter at a period of 1 second adjusted for site class effects as defined in ASCE/SEI 7-10 section 11.4.3. The subscript in DDR1 and SM1 refers to the period length. One can map DDRS for a given earthquake scenario by normalizing the ground-motion map with respect to ASCE/ SEI 7’s MCER ground motion parameter map for a 5-percent damped, 0.2-second spectral acceleration response, adjusted for site class SMS. The result for the HayWired scenario mainshock—at least for the study area modeled by Aagaard and others (2010)—is shown in figure 5. I further idealize the collapse capacity of all code-compliant buildings as a lognormally distributed random variable measured in terms of DDRS or DDR1. I will use DDRS when 123° 39° considering the performance of buildings with period less than the corner period, which corresponds to the intersection of the constant-acceleration and constant-velocity parts of the ASCE/SEI 7 design spectrum and is equal to SM1/SMS. I will use DDR1 when the period is at or above the corner period. More qualitatively, DDRS provides a better measure for lowrise buildings (less than 3 or 4 stories, or so), whereas DDR1 is the better measure for taller buildings. Taller buildings are actually designed based on SM1/T, where T is the building’s estimated small-amplitude fundamental period of vibration. If I were to define DDR for these taller buildings as the spectral acceleration (Sa) response at their fundamental period of vibration, that is, Sa(T), I could then estimate Sa(T)=Sa(1 second)/T, normalize by SM1/T, and the period T would cancel out of the numerator and denominator, leaving DDR1 as already defined. At the upper limit, I ignore those buildings that are so tall that their fundamental period of vibration lies on the constant-displacement portion of the response spectrum. They tend to have lateral strength governed by wind loads rather than earthquake 122° 121° 38° PA CI San Francisco Oakland FI C O CE AN San Jose 37° Area CALIF of map 0 0 0.0 20 20 40 40 60 MILES 60 KILOMETERS 0.2 0.4 0.6 Demand-to-design ratio 0.8 1.0 Figure 5.  Map of the San Francisco Bay region, California, showing demand-todesign ratio for the moment-magnitude-7.0 mainshock of the HayWired earthquake scenario, calculated for a 5-percent damped, 0.2-second spectral acceleration. (Map created using OpenSHA Generic Mapping Tools Map Plotter; Field and others, 2003.) Chapter K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes   65 loads, so the formulation developed here would grossly overestimate collapse of these buildings. By “lognormally distributed” it is meant that this work idealizes the collapse capacity of a building (the level of shaking that the building can tolerate without collapsing) as having a cumulative distribution function of the form: Now, the collapse probability of an arbitrary, code-compliant, engineered building can be modeled using equation 1 with q=3.47 and b=0.8. The collapse probability therefore is: ⎛ ) ⎞⎟ . Pc ( x ) = Φ ⎜⎝ ln( x/3.47 ⎠ 0.8 (3) Collapse probability, in percent As shown in figure 6, one can use this equation to estimate that at DDR=0.5, the collapse probability is 0.008, meaning ⎛ ln( x/q ) ⎞ , (1) that 0.8 percent of buildings are estimated to collapse at half ⎜ ⎟ P [ X ≤ x ] = Pc ( x ) = Φ ⎝ b ⎠ of MCER shaking. When one uses a lognormal fragility function for probawhere bilistic seismic risk analysis and fixes a point on the curve P[A] denotes the probability that statement A is near x=0.1 or 0.2 with a credible y-value, the long-term failure true, rate tends to be fairly insensitive to the value of b (Kennedy X denotes the uncertain collapse capacity of a and Short, 1994; Porter, 2017). That means that selecting a building measured in terms of DDRS or different value of b than the one suggested by Luco and othDDR1, ers (2007) should make little difference, within a reasonable x is a particular value of X, range. A lower value of b will tend to decrease the estimated Pc(x) denotes the probability that the building will number of failures at low values of x and increase the number collapse when subjected to shaking of x, of high values of x, with little change to the total. A similar Φ(z) denotes the standard normal cumulative statement can be made about selecting a higher value of b. distribution function evaluated at z, Applying the same ratios of red tags to collapses and q is the median value of X, and yellow to red tags determined in the previous sections, one b is the standard deviation of ln(X). can further estimate that under the model used for collapses Equation 1 represents a fragility function, which here by ASCE/SEI 7-10, approximately half of buildings shaken at means a relation between a measure of environmental excitahalf the MCER will be impaired (fig. 7). tion on the x-axis (here, DDR) and the occurrence probability of an undesirable outcome on the y-axis (here, collapse). Since the undesirable outcome is building collapse, equation 1 represents a collapse fragility function. Equation 1 is in a form that is commonly used by many PBEE-2, loss estimation, and 25 probabilistic seismic risk analysis studies to estimate damage to structures and components, including collapse. It is not that collapse capacity of buildings and building components are actually lognormally distributed, but rather that the lognor20 mal often fits observed performance data reasonably well, is convenient, has the advantage of decades of tradition, and has support in information theory. The mention of information 15 theory refers to the fact that the lognormal distribution is the maximum-entropy distribution—the one that assumes the least information—for a positively valued random variable with fixed median and logarithmic standard deviation. 10 To return to the problem of impairment rate at levels of shaking other than at MCER, equation 1 requires two parameters, q and b. Luco and others (2007) examine a range of 5 values of b, from 0.6 to 1.0, and settled on b=0.8 as their best estimate. This value of b was subsequently used in developing the ASCE/SEI 7-10 maps of MCER and is also applied here. With the knowledge that the collapse probability, Pc, is 0 0.5 1 1.5 Pc(1.0)=0.06 (as calculated in fig. 4), one can calculate q as: q = 1.0 × exp (−b × Φ −1 ( 0.06 )) , Demand-to design ratio (2) which leads to q=3.47. That is, the median collapse DDR is 3.47, which means the median collapse capacity is 3.47 times MCER shaking, whether measured in terms of SMS or SM1. Figure 6.  Graph showing the cumulative distribution function of building collapse probability for a range of demand-to-design ratio (DDR) values. When DDR is 0.5—that is, half of the risk-adjusted maximum considered earthquake (MCER) shaking—the probability of collapse is about 0.8 percent. 66   The HayWired Earthquake Scenario—Engineering Implications ⎛ ) ⎞⎟ ≤ 1.0 , Pi ( x ) = 63.4 × Φ ⎜⎝ ln( x/3.47 ⎠ 0.8 (4) Using the map of DDR (fig. 5), one can map impairment rate spatially, as shown in figure 8. The figure only shows impairment results from the shaking model; it does not reflect impairment owing to ground failure, including liquefaction, landslide, and surface faulting, which would collectively tend to increase impairment rates by some unknown, but probably smaller, degree. One can then overlay the impairment rate on a map of the estimated building stock to estimate the number of impaired buildings in a scenario like HayWired. To overlay this information, scale the impairment rate by the estimated building stock: n N i = ∑ Pi ( x j ) × B j j=1 , (5) where Ni j denotes the number of impaired buildings, is an index to locations in the inventory of building stock (such as census blocks), n is the number of locations in the inventory, xj denotes the shaking in location j expressed in terms of DDR, and Bj denotes the number of buildings in location j. Evaluating equation 5 at the level of census tracts and using FEMA’s enhanced building inventory, I find that shaking in the HayWired scenario would impair approximately 495,000 buildings. Table 5 lists the estimated number of shakinginduced impairment to buildings. In this scenario, using an average of three people per building (homes, workplaces, or other buildings), up to 1.5 million people could be displaced from those buildings for months or more while the buildings were repaired or replaced. Collapse, 0.8 0 afe, 1 More generally, the impairment rate (Pi) can be estimated as a function of DDR—denoted by Pi(x)—using equation 4. To do so, simply multiply equation 3 for collapse probability by the ratio of impaired buildings to collapsed buildings: Oklahoma, for example, recovering from a catastrophic tornado, recently required that new buildings be designed to resist windspeeds 50 percent higher than that required by the IBC—that is, 135 miles per hour (mi/hr) rather than 90 mi/hr (City of Moore, 2014). Because wind pressure is proportional to the square of velocity, the 50-percent increase in design windspeed equates to a 125-percent increase in design strength (because 1.52=2.25), which means that the City of Moore was effectively adopting a wind importance factor of 2.25 for ordinary buildings. For reference, buildings along much of the U.S. Gulf and Atlantic Coasts are required to resist windspeeds in excess of 140 mi/hr, so the City of Moore was not requiring unprecedented or even particularly uncommon design strength. Suppose all San Francisco Bay area buildings were suddenly replaced with new ones built with Ie=1.5—that is, 50 percent stronger than the current code minimum. (Structural engineering readers should not infer that this hypothetical situation suggests that all buildings are designed as Risk Category IV under ASCE/SEI 7-10, with the attendant additional requirements. The situation considered here is that the State of California or every jurisdiction adopts the IBC with the exception that ASCE/SEI 7-10 table 1.5-2 is altered so that Ie=1.5 for all risk categories.) Similar to the case in the City of Moore, adopting a seismic importance factor of Ie=1.5 would not impose extraordinary or particularly expensive requirements on the design of new buildings, at least for much of California. Consider the map of short-period MCER shaking in figure 9. The contours show the mapped short-period (0.2 second) MCER shaking Uns Fraction and Number of Impaired Buildings at the Societal Level Usable after earthquake, 49 Restricted use, 40 Options for Improving Building-Stock Performance Cities and states tend to have limited options for modifying the provisions of model building codes like the IBC. They tend to lack the necessary resources to do anything other than accept or reject the code. However, there is one code parameter called the seismic importance factor (denoted by Ie in ASCE/SEI 7-10) that one can think of as a volume dial to uniformly increase the design strength of ordinary buildings in a community by a constant factor. The City of Moore, Figure 7.  Pie diagram showing the impairment rate, in percent, of buildings at one half the risk-adjusted maximum considered earthquake (MCER) shaking. Buildings that are considered impaired are either collapsed, red-tagged, or yellow-tagged. Chapter K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes   67 on rock, denoted by SS in ASCE/SEI 7-10 and measured in percent of gravitational acceleration (g). For mapped values of 100 or greater (1.0 g or greater), the amplification factor to account for site conditions (denoted by Fa in ASCE/SEI 7-10) is approximately 1.0, so the map also shows SMS, the MCER shaking accounting for site conditions other than rock. After accounting for common site conditions, SMS in Sacramento, California, is approximately 0.8 g, whereas in San Francisco SMS varies between 1.5 and 2.3 g. This means that if a lowrise building in western San Francisco were built with an importance factor of Ie=1.0, picked up, and moved bodily to Sacramento, the building would satisfy seismic design requirements for a building with Ie of almost 3.0 in Sacramento. If the same building were instead picked up and moved bodily only 5 miles east inside of San Francisco, it would satisfy seismic design requirements for a building with Ie=1.5. Similarly, an ordinary building constructed in Concord, California, with Ie=1.0 (SMS=2.25 g) and moved 35 miles south near the San Jose International Airport (SMS=1.5 g) would satisfy design requirements for Ie=1.5. Buildings are built in both locations 123° 39° Table 5.  Shaking-related building impairment estimates for a hypothetical moment-magnitude-7.0 earthquake occurring on the Hayward Fault in the east bay part of the San Francisco Bay area, like that modeled in the HayWired earthquake scenario. Building condition Collapse 122° Estimated number of buildings 8,000 Red-tagged 101,000 Yellow-tagged 386,000 Total 495,000 121° Area of map CALIFORNIA 38° Oakland San Francisco PA CI FI C O CE AN San Jose 37° 0 0 0.0 20 20 40 40 60 MILES 60 KILOMETERS 0.2 0.4 0.6 Fraction of impaired buildings 0.8 1.0 Figure 8.  Map of the San Francisco Bay region, California, showing impairment of existing buildings, if they all complied with current building codes, for the momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Building impairment includes collapsed, red-tagged, and yellowtagged buildings. (Map created using OpenSHA Generic Mapping Tools Map Plotter; Field and others, 2003.) 68   The HayWired Earthquake Scenario—Engineering Implications 24° 23° 22° 1.5 PA 39° CI 1.5 FIC 38° OC EA 0.5 N 0.4 Concord Oakland 1.0 San Francisco Figure 9.  Map of risk-adjusted maximum considered earthquake (MCER) shaking in the San Francisco Bay region, California. Map shows contours of maximum-direction, short-period (0.2 second) spectral response acceleration with 5-percent critical damping on sites with 760 meters per second average shear wave velocity in the top 30 meters of soil. Contours on the map are expressed as a fraction of gravitational acceleration (g). Contours are shown in increments of 0.1 below 1.0 g and in increments of 0.25 above 1.0 g. (Map modified from Building Seismic Safety Council, 2015.) 1.25 38° 1.5 37° San Jose 1.5 2.0 EXPLANATION Area of map City Contour lines—Short-period (0.2 second) MCER shaking, expressed as a fraction of gravitational acceleration (g) CALIFORNIA 0 0 20 20 40 MILES Areas with a constant MCER shaking of 1.50 g 40 KILOMETERS 23° 22° 37° 21° 1 all the time, practically and cost effectively, which means that imposing a design requirement that ordinary buildings be designed for Ie=1.5 would be practical and not particularly expensive throughout much of the San Francisco Bay area, indeed in much, if not all, of California. Informal discussion between the author and four California engineers suggests that designing to Ie=1.5 would increase construction costs on the order of 1–3 percent (D. Bonneville, Degenkolb Engineers, oral commun., January 2015; E. Reis, U.S. Resiliency Council, oral commun., April 2014; J. Harris, J. R. Harris and Company, oral commun., August 2015; and R. Mayes, Simpson, Gumpertz and Heger, Inc., oral commun., January 2015). The estimate is further supported by NEHRP Consultants Joint Venture (2013), whose authors find that to redesign six particular buildings in Memphis, Tennessee, so that they comply with the 2012 IBC rather than the 1999 Southern Building Code, their strength would increase on average by 60 percent and their construction cost would increase between 0 and 1.0 percent. Olshansky and others (1998) found a similar estimated marginal cost to increase from no seismic design to code minimum. Furthermore, the estimated cost to achieve an immediate occupancy performance level rather than life safety for one of the index buildings of the CUREE-Caltech Woodframe Project is similarly marginal (Porter and others, 2006). If the previous arguments do not convince the skeptical reader, I offer one more. A square-foot cost manual, such as RSMeans (2011), shows that approximately 67 percent of construction cost of a typical new building is spent on the architectural, mechanical, electrical, and plumbing assemblies, approximately 17 percent on overhead and profit, and of the remaining 16 percent that represents structural cost, approximately half is spent on labor. Most of the remaining 8 percent (mostly structural material cost) is spent on the gravity-resisting system: the foundation, floor slabs, columns, and beams. Of the small remaining portion that is spent on materials for the earthquake load resisting system (perhaps 2 percent), it could be increased by 50 percent to achieve a 1-percent overall construction cost increase (the purchase price increase would be less if land value is considered). However, strength does not increase linearly with quantity of material. Quantity need not double or even increase by half to achieve a 50 percent stronger building. Strength can increase with the square or a higher power of material. For example, a W44×230 wide-flange steel shape is about 63 percent stronger than a W30×191 shape, but weighs (and therefore costs) only about 20 percent more without requiring any additional labor to install. In this particular case, strength increases with more than the square of cost (1.202.6=1.63). More extreme cases can be cited. In California, a marginal construction cost increase of 1–3 percent would translate to a smaller marginal development cost increase because land can constitute more than half the value of a building and its value is unaffected by Ie. Suppose the entire 1–3 percent cost increase was passed on to buyers Chapter K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes   69 and not absorbed by a reduction in the value of land and that land accounted for half the value of the property (it is often much more). The increase in purchase price might range between 0.5 and 1.5 percent—that is, on the order of 1 percent. In Porter (Not Safe Enough, this volume), I will examine whether the public is willing to pay this increased purchase price. What would be the consequences in terms of building impairment in the HayWired scenario mainshock if the State of California or every San Francisco Bay area city adopted the IBC but required that all new ordinary buildings be designed with an importance factor Ie=1.5? One can calculate the effect on collapse probability by retaining the assumption that collapse probability is reasonably approximated with a lognormal cumulative distribution function, retaining the assumption that the standard deviation of the natural logarithm of collapse capacity is approximately 0.8, and multiplying the median collapse capacity by a factor of 1.5 (fig. 10). Assuming the ratios of red tags to collapses and yellow tags to red tags (3.8:1 and 13:1, respectively) still hold, then the map of impairment rate would be that shown in figure 11. Again, results are faily insensitive to parameter b. Evaluating equation 5 at the level of census tracts and assuming Ie=1.5 results in the estimate that shaking in the HayWired scenario mainshock would impair about 75 percent fewer buildings than when Ie=1.0—impairing 130,000 buildings total, rather than 495,000. Table 6 shows a side-by-side comparison of building impairment under Ie=1.0 and Ie=1.5 conditions. Recall that the cost of this reduction in impairment might realistically result in a marginal increase in purchase price of 0.5–1.5 percent. To put it another way, it may increase the monthly mortgage for a new $500,000 house in Concord, 25 Collapse probability, in percent 20 15 I e = 1.0 10 I e = 1.5 5 0 0.5 1.0 1.5 Demand-to-design ratio Figure 10.  Graph showing collapse probability as a function of demand-to-design ratio for an importance factor (Ie) of 1.0 and 1.5. Note the significant drop in collapse probability for buildings with a higher value of Ie. California, for example, from $3,400 to $3,430. The values in table 6, however, do not include buildings that are moderately damaged. Recall that Comerio (2006) found that moderately damaged housing units numbered three times as many as collapsed, red-tagged, or yellow-tagged buildings after the 1994 Northridge earthquake. I consider the Ie=1.0 case that the HayWired mainshock would impair 500,000 buildings if all buildings behaved like new ones. The San Francisco Bay area contains approximately 2 million buildings, so 500,000 represents about 25 percent of buildings in the area. What happens to the people who live or work in those buildings? Vacancy rates in recent years are estimated to be approximately 0.5 percent of single-family dwellings, 3 percent of residential rental units (U.S. Census Bureau, [n.d.]), and 12–15 percent of San Francisco office space (Wells Fargo Securities, 2014). These rates are far too low to accommodate the displaced homes and businesses, and may force people to move out of the area. With a building stock designed to Ie=1.5, an estimated 130,000 buildings, or approximately 6 percent of the building stock, would be impaired in the HayWired mainshock. Though that number still exceeds the vacant stock, it is approximately within the bounds called for by the San Francisco Planning and Urban Research (SPUR) Shelter-in-Place Task Force (2012), an urban-planning policy document that recommends housing in San Francisco be resilient enough that 95 percent of the population could shelter in place. With doubling up and other emergency accommodations, this scenario would be more bearable than the alternative Ie=1.0 case. Building Impairment Resulting from Ground Failure The building impairment estimates thus far discussed do not account for ground failure—liquefaction, landslide, or surface faulting. The effect of surface faulting can be estimated by assuming that every building whose footprint crosses a fault rupture where there is offset at the surface would be impaired. In the HayWired scenario, surface rupture would result in damage to approximately 500 existing buildings, adding about 0.1 percent to the total impairment quantity if all buildings were designed to Ie=1.0, or 0.4 percent if all buildings were designed to Ie=1.5. However, the damage to the code-compliant building stock would probably be much lower, for reasons explained next. The 1972 Alquist-Priolo Fault Zoning Act (California Department of Conservation, [n.d.]) requires that local agencies regulate most development projects within about 50 feet of an active fault, such as the Hayward Fault. Projects include all land divisions and most buildings. The act generally serves to prevent new construction on top of a known active fault. Single-family wood-frame and steel-frame dwellings as high as two stories that are not part of a development of four or more units are exempt. Cities are free to impose greater 70   The HayWired Earthquake Scenario—Engineering Implications restrictions, such as including the exempt buildings. If every city were to extend the restrictions of the 1972 Alquist-Priolo Fault Zoning Act to all buildings, and all buildings in the San Francisco Bay area were compliant, then there would be no buildings damaged by fault rupture. More recently, ASCE/SEI 7-10 section 11.8 requires a geotechnical investigation and report for most new ordinary California buildings. The report includes an evaluation of the potential for slope instability, liquefaction, total and differential settlement, and surface displacement owing to faulting or seismically induced lateral spreading or flow. The report must contain recommendations for foundation designs or other measures to mitigate the effects of these hazards. Although section 11.8 does not preclude damage as a result of ground failure, it does tend to prevent it. One can conclude that earthquake damage owing to ground failure in code-compliant buildings will be minor, probably overshadowed by uncertainty in damage resulting from shaking. The 2015 NEHRP Provisions Update Committee Chair points out that “Our seismic codes don’t do much to address 123° 39° Area of map liquefaction damage. . . We are doing more in the current cycle.” (D. Bonneville, written commun., March 2015). Because this chapter addresses the hypothetical situation where all buildings are replaced with new ones, I also imagine that liquefaction damage is largely prevented by current developments. Public Expectations for the Seismic Performance of New Buildings It appears likely that the public is unaware that a codecompliant building stock could perform as described in table 5, largely because the public has no involvement in the codewriting process. I now review that process. Seismic design provisions have been present in model U.S. building codes since 1927 and have almost continuously evolved with new editions of the building codes. (Again, recommended provisions, standards, and model building codes 122° 121° CALIF 38° Oakland San Francisco PA CI FI C O CE AN San Jose 37° 0 0 0.0 20 20 40 40 0.2 60 MILES 60 KILOMETERS 0.4 0.6 Fraction of impaired buildings 0.8 1.0 Figure 11.  Map of the San Francisco Bay region, California, showing estimated building impairment for the momentmagnitude-7.0 mainshock of the HayWired earthquake scenario if buildings were designed with a seismic importance factor (Ie) of 1.5. Estimated building impairment includes collapsed, red-tagged, and yellow-tagged buildings. (Map created using OpenSHA Generic Mapping Tools Map Plotter; Field and others, 2003.) Chapter K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes   71 Table 6.  Shaking-related building impairment for the momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. [Ie, seismic importance factor; numbers are rounded to the nearest 1,000 for simplicity] Building condition Collapse Buildings with Ie=1.0 Buildings with Ie=1.5 8,000 2,000 Red-tagged 101,000 27,000 Yellow-tagged 386,000 101,000 Total 495,000 130,000 are collectively and loosely referred to here as codes.) Code writers are required by engineering codes of ethics—such as that of the ASCE (2006)—to “hold paramount the health, safety and welfare of the public,” and there is little doubt that that mandate constantly informs the code-writing process. However, code writers have expended little effort to elicit from the public its preferences for the seismic performance of new buildings or its preferred tradeoffs of safety and construction cost. Discussions of acceptable seismic performance have been largely carried out within the structural engineering community. A philosopher of engineering ethics, Davis (2015), has recently concluded that the ASCE Code of Ethics implicitly requires that those discussions include a reasonable effort to involve members of the public and that civil engineers among the code writers are obligated to consider the public’s expressed preferences in the code. Several engineering ethicists strongly agree with Davis’ (2015) findings (see Porter, 2016b). As previously mentioned, U.S. building codes aim to protect life safety and limit property damage. The explicit intent of the 2009 IBC, for example, is to “establish minimum requirements to safeguard the public health, safety, and general welfare . . .” and includes “safety to life and property” among its goals (ICC, 2009). The NEHRP Provisions (Building Seismic Safety Council [BSSC], 2009) aim “to avoid structural collapse in very rare, extreme ground shaking” and “to provide reasonable control of damage to structural and nonstructural systems that could lead to injury and economic or functionality losses for more moderate and frequent ground shaking.” Note the inclusion of protecting the general welfare and avoiding property loss along with protecting life safety, suggesting that the authors did not feel legally constrained to only protect life safety. The NEHRP Provisions could, if its authors choose, change the requirements to control economic loss as well as to protect life. For earthquake loads in particular, authors of modern codes have assumed it is impractical or uneconomical to achieve seismic resistance much greater than what is implicit in prior codes. (Here, seismic resistance refers generally to the capacity of buildings to resist damage or loss of functionality in earthquakes.) This implies that the authors believe the public would be unwilling to pay for safer buildings, such as those that would remain functional after very strong shaking. ASCE 7 (ASCE, 2010) and, by adoption, the IBC (ICC, 2012) measure seismic acceptability on a per-building basis, which implies that per-building or per-person risk is the best (or at least the most practical) measure of risk. To say that it is uneconomical to provide greater seismic resistance is roughly equivalent to saying that people would be unwilling to pay for it (“economical” being a subjective judgment, not necessarily measured in terms of, say, benefitcost ratio). Is that true? In some sources (ASCE, 2010; ICC, 2012) and in others that underlie current code requirements for new buildings, there is no examination of the public’s willingness to pay. Building owners and tenants have generally not been asked to express their preferences and are typically absent from the committees that establish codes. The authors of National Bureau of Standards Special Publication 577 (Ellingwood and others, 1980) expressed the idea that the A58 standards committee (precursor to ASCE 7) represented “those substantially concerned with its [the standard’s] scope and provisions.” Although committee members included “a broadspectrum group of professionals from the research community, building code groups, industry, professional organizations and trade associations,” it did not include representatives of building owners or tenants. The authors of a FEMA-sponsored workshop on communicating earthquake risk interviewed a small group of primarily commercial real-estate professionals about their preferred measures of seismic performance (ATC, 2002). The authors expressed the belief that “this workshop represents one of the first significant attempts to obtain input on issues of acceptable levels of seismic risk used as a basis for design, from other than the technical community . . . ,” although they acknowledged that “several important stakeholder groups, notably residential and institutional building owners, and retailers were not represented at all.” Only very recently did the National Institute of Building Sciences (NIBS, 2012)—NIBS is the parent body that led the development of BSSC (2009), a foundational document for the IBC—carry out its first survey to “get a baseline read about what Americans know or don’t know about building codes.” The survey was at a fairly high level and did not inquire about the public’s understanding and preferences for the seismic performance of new buildings. The point is that the public is generally not consulted about their preferences for the seismic performance of the buildings on which their lives and livelihoods rely. It would be easier, perhaps necessary, for engineers to support judgments about what is economical if the public were asked what they would be willing to pay for greater seismic resistance. It is necessary to define “the public” for purposes of this discussion. Here, I adopt the definition described by Davis and Porter (2016), “the public should be understood as including all those anywhere whose lack of information, technical knowledge, ability, or conditions for adequate deliberation renders them more or less vulnerable to the power that engineers wield on behalf of client or employer. The public is a collection or aggregate rather than an organized body. Unlike an electorate or corporation, it has interests but no decision procedure—no will of its own. The public is an abstraction 72   The HayWired Earthquake Scenario—Engineering Implications (much like a set in mathematics). In the present context, however, the public does not include engineers, building officials, or members of the building industry, because these groups largely control the code.” Davis (2015) recently showed that the ASCE Code of Ethics requires that (1) the public has a role in establishing preferred tradeoffs among health, safety, and welfare; (2) engineers have an ethical obligation to ask about those preferences (within limits of reasonable effort); (3) engineers have an ethical obligation to respond to that information; and (4) the public, as used here, does not include builders, developers, or structural engineers. To study public expectations as part of the HayWired scenario, I surveyed the public to ask a series of questions (Porter, Not Safe Enough, this volume): • What do you believe to be the performance objectives most new buildings are intended to meet in a large earthquake? • What do you think the building code should provide? • What do you believe are the building performance measures of greatest interest to your community? • What do you consider to be a reasonable tradeoff between safer buildings and higher initial construction cost? • How important do you believe the issue is? • What would be the best way to explore the issue further with local government or community? The survey seeks information about the public’s and local officials’ understanding of and preferences for the seismic performance of new buildings that comply with the most modern building codes. In Porter (Not Safe Enough, this volume), I carried out a population-based survey of 400 members of the public within each of two regions: statewide throughout California and the two largest metropolitan statistical areas closest to the New Madrid Seismic Zone, namely Saint Louis, Missouri, and Memphis, Tennessee. The survey shows that the majority of respondents prefer better performance from new buildings than the results presented here for code-compliant buildings. Most prefer that buildings should be habitable or functional after a large earthquake. Survey respondents believe the public would be willing to pay 1 to 3 percent more in construction costs to achieve a higher level of seismic performance. Limitations of this Study There are many uncertainties in studies such as this and it may be that some parameters used here are unintentionally overly conservative. Perhaps the estimated 6-percent collapse rate at MCER shaking drawn from the FEMA and NIST analyses (ATC, 2009, and NEHRP Consultants Joint Venture, 2012) is unrepresentative of real buildings and the true average collapse rate at MCER shaking among new, engineered buildings will be much lower. For example, those studies only considered hypothetical frame buildings and, on the recommendation of the lead author of the two studies, the 6-percent figure reflects a subset of those hypothetical buildings—those with period greater than 0.5 seconds. Guidelines developed for the Global Earthquake Model present procedures for selecting samples of the general building stock so that the sample is representative of the building stock as a whole (Porter and others, 2014). A study similar to ATC (2009) and NEHRP Consultants Joint Venture (2012), but using a representative sample following the Global Earthquake Model procedures, could test the validity of the 6-percent figure for the general building population. Perhaps the logarithmic standard deviation of 0.8 used by Luco and others (2007) to calculate MCER is too high. The figure could be tested using the representative sample method described above (although long-term collapse rates and numbers of collapsed buildings in a single earthquake should be fairly insensitive to the exact value of the logarithmic standard deviation, as explained above). Perhaps the 63:1 ratio of redand yellow-tagged to collapsed buildings in the Loma Prieta and Northridge earthquakes and the 57:1 ratio in the Napa earthquake overestimate what would happen in future California earthquakes among code-compliant buildings. Procedures presented in ATC (2012) could be used to better estimate the ratio of red tags to collapses. ATC-20 tagging data from the 2014 Napa earthquake, 1994 Northridge earthquake, and 1989 Loma Prieta earthquakes could be analyzed further to develop analytical models for yellow tagging. Future studies could test the possibility that one or more of the parameters derived here do not represent reality and result in impairment estimates that are much greater than would happen in a real earthquake. However, until better estimates are available, the estimates presented here provide insight into the unintended consequences of current building code performance objectives on large-scale building impairment during a large, but not exceedingly rare, earthquake. Summary The IBC attempts to ensure the safety of life in new, engineered buildings that are subject to earthquake shaking. Current building-code goals include no more than a 1-percent collapse probability in 50 years and no more than 10-percent collapse probability during MCER shaking. However, the code does not control lesser degrees of impairment, such as red and yellow tagging, nor does it control the overall number of impaired buildings in a large, but not exceedingly rare, earthquake, such as the mainshock modeled in the HayWired scenario. This study shows that in three California earthquakes, approximately 60 times as many buildings have been red- or yellow-tagged as have collapsed. In one earthquake, three Chapter K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes   73 times as many again experienced at least moderate damage—evidenced by insurance claims averaging $30,000 to $40,000—without collapsing or being red- or yellow-tagged. This chapter also shows that FEMA and NIST research supporting the upper-bound goal of no more than 10-percent collapse probability results in an average 6-percent collapse probability during MCER shaking. This work proposes a collapse fragility function for new, engineered buildings whose input parameter is the ratio of shaking at the building site to the siteclass-adjusted MCER; the ratio is referred to as the demandto-design ratio (DDR). The impairment fragility function takes the form of a lognormal cumulative distribution function multiplied by the ratio of total impaired buildings to collapsed buildings (63) with an upper bound of 1.0. The logarithmic standard deviation of the fragility function can be applied from the work of Luco and others (2007) that proposed establishing MCER. The median of the collapse fragility function follows directly from the observation that 6 percent of new, engineered buildings are expected to collapse in MCER shaking, with a logarithmic standard deviation. Using the map of ground motion in the HayWired scenario mainshock (fig. 5) and ASCE/SEI 7-10’s map of MCER shaking, adjusted to account for site amplification, this study calculates shaking in the HayWired scenario mainshock in terms of DDR. The DDR is evaluated for each location in FEMA’s 2010 enhanced building inventory for California to estimate the number of buildings exposed to ground shaking in the HayWired mainshock for various levels of DDR. Using the impairment fragility function derived here, one can estimate the number of buildings impaired in the HayWired scenario, assuming that all of the buildings behaved as if they were new, code-compliant buildings. The result is an estimate that approximately 500,000 buildings would collapse, be redtagged, or be yellow-tagged. As many as 1.5 million people could be displaced from those buildings for months or more while the buildings are repaired or replaced. There are too few vacant buildings in the San Francisco Bay area to accommodate this many displaced people, suggesting that many would be forced to leave the San Francisco Bay area for at least several months, and possibly permanently. One possible option for states or cities to reduce future building impairment is to adopt the IBC with the modification that all risk category I, II, or III buildings must be designed with a seismic importance factor (Ie) greater than 1.0. In the hypothetical case that all San Francisco Bay area cities require design to Ie=1.5, the total number of impaired buildings in the HayWired mainshock would be reduced by approximately 75 percent, to 130,000 impaired buildings. The estimated 500,000 impaired buildings under current code objectives fails the 95-percent shelter-in-place performance objective of a San Francisco Bay area urban planning organization, SPUR, but the reduced impairment with Ie=1.5 (130,000 buildings) meets SPUR’s objective. The estimates presented here—500,000 impaired buildings under current code and 130,000 impaired buildings under the what-if scenario—follow solely from current code objectives, the research that established MCER, the evidence of collapse, red tags, and yellow tags in California’s recent earthquake history, and FEMA’s 2010 enhanced building inventory. The construction cost to build new buildings with Ie=1.5 has been approximately estimated here to be 1–3 percent greater than under the current code, or about a 1 percent greater purchase price considering land value. Such a marginal cost seems entirely practical. Seismic design procedures in ASCE/SEI 7-10 already require buildings on one side of San Francisco to be designed with 50 percent greater strength than buildings on the other side. As another example, the City of Moore, Oklahoma, now requires buildings to be designed with 125 percent greater strength than before the devastating 2013 Moore tornado occurred. A survey to elicit the public’s understanding, expectations, and preferences for the code’s seismic performance objectives was undertaken as part of the HayWired earthquake scenario (Porter, Not Safe Enough, this volume). It shows that the majority of the public prefers better performance— that buildings should be habitable or functional after a large earthquake. Survey respondents believe the public would be willing to pay 1 to 3 percent more in construction costs to achieve a higher level of seismic performance. Some argue that civil engineers have an ethical obligation to elicit the public’s preferences in drafting seismic design requirements (Davis, 2015), suggesting that surveys such as those undertaken as a part of the HayWired scenario may help inform the next generation of building-code objectives in seismically active areas. Conclusion Current code objectives seem to unintentionally produce a high degree of building impairment in a large, but not exceedingly rare, earthquake, such as the Mw 7.0 mainshock of the HayWired scenario. The impairment, in terms of collapsed, red-tagged, or yellow-tagged buildings, would displace a large number of residents from the San Francisco Bay region. The building-code objectives appear, based on a large survey of Californians and people living in the Saint Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas, to be substantially below what the public may prefer and may be willing to pay for. References Cited Aagaard, B.T., Graves, R.W., Rodgers, A., Brocher, T.M., Simpson, R.W., Dreger, D., Petersson, N.A., Larsen, S.C., Ma, S., and Jachens, R.C., 2010, Ground-motion modeling of Hayward Fault scenario earthquakes, part II—Simulation of long-period and broadband ground motions: Bulletin of the Seismological Society of America, v. 100, no. 6, p. 2945–2977. 74   The HayWired Earthquake Scenario—Engineering Implications American Society of Civil Engineers [ASCE], 2006, Code of ethics: Reston, Va., American Society of Civil Engineers web page, accessed January 28, 2015, at https://www.asce. org/codeofethics/. American Society of Civil Engineers [ASCE], 2010, Minimum design loads for buildings and other structures: Reston, Va., Structural Engineering Institute, ASCE/SEI 7-10, 608 p. Applied Technology Council [ATC], preparer, 1997, NEHRP guidelines for the seismic rehabilitation of buildings: Washington D.C., Federal Emergency Management Agency publication 273, prepared for the Building Seismic Safety Council, 386 p., accessed May 21, 2015, at https://www. wbdg.org/ccb/DHS/ARCHIVES/fema273.pdf. Applied Technology Council [ATC], 2002, ATC-58-1—Proceedings of FEMA-sponsored workshop on communicating earthquake risk: Redwood City, Calif., Applied Technology Council, 74 p. Applied Technology Council [ATC], 2005, ATC-20-1—Field handbook—Procedures for postearthquake safety evaluation of buildings: Redwood City, Calif., Applied Technology Council, 144 p. Applied Technology Council [ATC], preparer, 2009, Quantification of building seismic performance factors: Washington D.C., Federal Emergency Management Agency publication P-695, 421 p., accessed May 21, 2015, at https://www. fema.gov/media-library-data/20130726-1716-25045-9655/ femap695.pdf. Applied Technology Council [ATC], preparer, 2012, Seismic performance assessment of buildings—Methodology: Washington D.C., Federal Emergency Management Agency publication P-58-1, v. 1, 319 p., accessed May 21, 2015, at https://www.fema.gov/media-library-data/13964950198480c9252aac91dd1854dc378feb9e69216/FEMAP-58Volume1508.pdf. Applied Technology Council [ATC], preparer, 2015, Rapid visual screening of buildings for potential seismic hazards—A handbook (3d ed.): Washington D.C., Federal Emergency Management Agency publication P-154, 388 p., accessed May 21, 2015, at https://www.fema.gov/medialibrary-data/1426210695633-d9a280e72b32872161efab26a602283b/FEMAP-154508.pdf. Aslani, H., and Miranda, E., 2006, Delivering improved information on seismic performance through loss disaggregation, in Proceedings of the 8th U.S. National Conference on Earthquake Engineering: Earthquake Engineering Research Institute, 10 p. Beck, J.L., Kiremidjian, A., Wilkie, S., Mason, A., Salmon, T., Goltz, J., Olson, R., Workman, J., Irfanoglu, A., and Porter, K., 1999, Decision support tools for earthquake recovery of businesses—Final report: Richmond, Calif., Consortium of Universities for Research in Earthquake EngineeringKajima Joint Research Program, Phase III, 146 p. Building Seismic Safety Council [BSSC], preparer, 2009, NEHRP recommended seismic provisions for new buildings and other structures: Washington D.C., Federal Emergency Management Agency publication P-750, 388 p., accessed May 21, 2015, at https://www.fema.gov/media-librarydata/20130726-1730-25045-1580/femap750.pdf. Building Seismic Safety Council [BSSC], preparer, 2015, NEHRP recommended seismic provisions for new buildings and other structures: Washington D.C., Federal Emergency Management Agency publication P-1050, 515 p. Bureau of Labor Statistics, 2013, National census of fatal occupational injuries in 2012 (preliminary results): U.S. Department of Labor publication USDL-13-1699, 14 p. California Department of Conservation, [n.d.], California public resources code, division 2, chapter 7.5, Earthquake fault zoning: California Department of Conservation web page, accessed February 3, 2015, at https://www.conservation. ca.gov/cgs/codes/prc/Pages/chap-7-5.aspx. City of Moore, 2014, City adopts new building codes, first in the Nation: City of Moore web page, accessed November 25, 2014, at https://www.cityofmoore.com/node/2111. Comerio, M.C., 2006, Estimating downtime in loss modeling: Earthquake Spectra, v. 22, no. 2, p. 349–365. Czarnecki, R.M., 1973, Earthquake damage to tall buildings: Cambridge, Mass., Massachusetts Institute of Technology, Structures Publication 359, 125 p. Davis, M., 2015, What part should the public have in writing engineering standards?, in Security and Disaster Preparedness Symposium—Codes and Governance in the Built Environment, National Institute of Building Sciences Third Annual Conference and Expo: National Institute of Building Sciences, accessed March 1, 2015, at https://www.nibs.org/ store/ViewProduct.aspx?id=4108575. Davis, M., and Porter, K., 2016, The public’s role in seismic design provisions, Earthquake Spectra, v. 32, no. 3, p. 1345–1361, https://doi.org/10.1193/081715EQS127M. Ellingwood, B., Galambos, T.V., MacGregor, J.G., and Cornell, C.A., 1980, Development of a probability-based load criterion for American National Standard A58: Washington D.C., National Bureau of Standards Special Publication 577, 222 p. Chapter K. Societal Consequences of Current Building Code Performance Objectives for Earthquakes   75 EQE International and the Geographic Information Systems Group of the California Governor’s Office of Emergency Services [EQE and Cal OES], 1995, The Northridge earthquake of January 17, 1994—Report of data collection and analysis; Part A, Damage and inventory data: Irvine, Calif., EQE International, project number 36386.02, 322 p. Luco, N., Ellingwood, B.R., Hamburger, R.O., Hooper, J.D., Kimball, J.K., and Kircher, C.A., 2007, Risk-targeted versus current seismic design maps for the conterminous United States, in Proceedings of the 2007 Convention of the Structural Engineers Association of California: Sacramento, Structural Engineers Association of California, p. 163–175. Federal Emergency Management Agency [FEMA], 2012a, Reducing the risks of nonstructural earthquake damage—A practical guide: Federal Emergency Management Agency, FEMA E-74, accessed June 1, 2014, at https://www.fema. gov/earthquake-publications/fema-e-74-reducing-risksnonstructural-earthquake-damage. Multihazard Mitigation Council, 2017, Natural hazard mitigation saves—2017 interim report: Washington, D.C., National Institute of Building Sciences, 344 p., accessed April 3, 2018, at https://www.nibs.org/page/mitigationsaves/. Federal Emergency Management Agency [FEMA], 2012b, Hazus multi-hazard loss estimation methodology, earthquake model, Hazus®-MH 2.1 technical manual: Federal Emergency Management Agency, Mitigation Division, 718 p. Field, E.H., Jordan, T.H., and Cornell, C.A., 2003, OpenSHA, a developing community-modeling environment for seismic hazard analysis: Seismological Research Letters, v. 74, no. 4, p. 406–419. Fradkin, P.L., 1999, Magnitude 8—Earthquakes and life along the San Andreas Fault: University of California Press, 348 p. Harris, S.K., Scawthorn, C., and Egan, J.A., 1990, Damage in the Marina District of San Francisco in the October 17, 1989 Loma Prieta Earthquake, in Proceedings of the 8th Japan Earthquake Engineering Symposium: Tokyo, Japan Earthquake Engineering Symposium. Heron, M., 2013, Deaths—Leading causes for 2010: National Vital Statistics Reports, v. 62, no. 6, 97 p. Hess, R.L., 2008, The ShakeOut scenario supplemental study—Unreinforced masonry (URM) Buildings: Denver, Colo., SPA Risk LLC, 15 p., accessed October 26, 2010, at https://www.sparisk.com/pubs/ShakeOutScenarioURMHess. pdf. International Code Council [ICC], 2009, International Building Code: Country Club Hills, Ill., International Code Council, 716 p. International Code Council [ICC], 2012, International Building Code: Country Club Hills, Ill., International Code Council, 722 p. Kennedy R.P., and Short, S.A., 1994, Basis for seismic provisions of DOE-STD-1020: Lawrence Livermore National Laboratory and Brookhaven National Laboratory Technical Report UCRL–CR–111478 and BNL–52418, 65 p., https:// doi.org/10.2172/10146835. Kustu, O., Miller, D.D., and Brokken, S.T., 1982, Development of damage functions for highrise building components for the U.S. Department of Energy: San Francisco, Calif., URS/John A. Blume & Associates, 46 p. National Institute of Building Sciences [NIBS], 2012, Results from a survey of Americans on building codes and building safety: Washington D.C., National Institute of Building Sciences, 13 p. National Institute of Standards and Technology [NIST], 1990, Performance of structures during the Loma Prieta Earthquake of October 17, 1989: National Institute of Standards and Technology, NIST SP-778, 212 p. NEHRP Consultants Joint Venture, 2012, Tentative framework for development of advanced seismic design criteria for new buildings: National Institute of Standards and Technology, NIST GCR 12-917-20, 302 p., accessed May 21, 2015, at https://www.nehrp.gov/pdf/nistgcr12-917-20.pdf. NEHRP Consultants Joint Venture, 2013, Cost analyses and benefit studies for earthquake-resistant construction in Memphis, Tennessee: National Institute of Standards and Technology, NIST GCR 14-917-26, 249 p., accessed May 21, 2015, at https://www.nist.gov/customcf/getpdf. cfm?pubid=915569. Olshansky, R.B., Bancroft, R., and Glick, C., 1998, Promoting the adoption and enforcement of seismic building codes—A guidebook for State earthquake and mitigation managers: Washington D.C., Federal Emergency Management Agency publication 313, 211 p. 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, 255 p., https://dx.doi. org/10.3133/ofr20141091. Porter, K.A., 2003, An overview of PEER’s performancebased earthquake engineering methodology, in Proceedings of the Ninth International Conference on Applications of Statistics and Probability in Civil Engineering: Civil Engineering Risk and Reliability Association, p. 973–980, accessed June 1, 2014, at https://spot.colorado. edu/~porterka/Porter-2003-PEER-Overview.pdf. 76   The HayWired Earthquake Scenario—Engineering Implications Porter, K.A., 2016a, Safe enough?—A building code to protect our cities as well as our lives: Earthquake Spectra, v. 32, no. 2, p. 677–695, https://dx.doi.org/10.1193/112213EQS286M. Porter, K.A., 2016b, Not safe enough—The case for resilient seismic design: Structural Engineers Association of California 2016 Convention, Maui, Hawaii, October 12–15, 2016, 10 p., accessed April 3, 2018, at http://www.sparisk.com/ pubs/Porter-2016-SEAOC-Resilience.pdf. Porter, K.A., 2017, When addressing epistemic uncertainty in a lognormal fragility function, how should one adjust the median?, in World Conference on Earthquake Engineering, 16th, Santiago, Chile, 2017, Proceedings: International Association for Earthquake Engineering, no. 2617. Porter, K.A., and Cobeen, K., 2012, Informing a retrofit ordinance—A soft-story case study, in Proceedings of the 2012 Structures Congress: Reston, Va., American Society of Civil Engineers. Porter, K., Farokhnia, K., Vamvatksikos, D., and Cho, I., 2014, Analytical derivation of seismic vulnerability functions for highrise buildings: Global Vulnerability Consortium, 136 p., accessed January 1, 2015, at https://www.nexus.globalquakemodel.org/gem-vulnerability/posts/. Porter, K.A., and Kiremidjian, A.S., 2001, Assembly-based vulnerability of buildings and its uses in seismic performance evaluation and risk-management decision-making: Stanford, Calif., Stanford University, John A. Blume Earthquake Engineering and Research Center report 139, Ph.D. dissertation, 196 p., available at https://www.sparisk.com/ pubs/Porter-2001-ABV-thesis.pdf. Porter, K.A., Scawthorn, C.R., and Beck, J.L., 2006, Costeffectiveness of stronger woodframe buildings: Earthquake Spectra, v. 22, no. 1, p. 239–266, https://dx.doi. org/10.1193/1.2162567. RSMeans Co., Inc., 2011, Square Foot Costs 2012 (33d ed.): Kingston, Mass., RSMeans, Inc., 505 p. Structural Engineers Association of Northern California [SEAONC], 1990, Posting of buildings after the Loma Prieta Earthquake: San Francisco, Calif., Structural Engineers Association of Northern California, 32 p. SPUR Shelter-in-Place Task Force, 2012, Safe enough to stay: San Francisco, Calif., San Francisco Planning and Urban Research, 44 p. Taciroglu, E., and Khalili-Tehrani, P., 2008, M7.8 Southern San Andreas Fault Earthquake Scenario—Non-ductile Reinforced Concrete Building Stock: Denver, Colo., SPA Risk LLC, accessed October 26, 2010, at https://arazprojects.com/wp-content/uploads/concrete_buildings. pdf. U.S. Census Bureau, 2012, Table 1103, motor vehicle accidents—Number and deaths, 1990–2009: Washington D.C., U.S. Census Bureau, accessed May 21, 2015, at https://www.census.gov/compendia/statab/2012/ tables/12s1103.pdf. U.S. Census Bureau, [n.d.], Quarterly vacancy and homeownership rates by State and MSA: Washington D.C., U.S. Census Bureau, accessed May 21, 2015, at https:// www.census.gov/housing/hvs/data/rates.html. Wells Fargo Securities, 2014, San Francisco Bay area real estate outlook: Wells Fargo Securities Economic Group Special Commentary, May 6, 2014, accessed May 21, 2015, at https://www.realclearmarkets.com/ docs/2014/05/San%20Francisco%20Real%20Estate%20 Outlook_05062014.pdf. Wikipedia, 2017, List of earthquakes in California, accessed December 5, 2017, at https://en.wikipedia.org/ wiki/List_of_earthquakes_in_California. The HayWired Earthquake Scenario—Engineering Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q [Also see https://doi.org/10.3133/sir20175013] Chapter L Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings in California and the New Madrid Seismic Zone By Keith A. Porter1 Abstract 4. Expect better seismic performance than ASCE/SEI 7 intends to provide; Earlier disaster planning scenarios developed by the U.S. Geological Survey (USGS), especially the ShakeOut earthquake scenario, suggest the public is unaware of the lifesafety performance objective in U.S. seismic-design standards and may want better building performance. As part of the USGS HayWired earthquake scenario, I (through the University of Colorado Boulder) undertook an effort to explore this hypothesis by performing a public survey in California and near the New Madrid Seismic Zone in the Central United States. The HayWired scenario examines a hypothetical earthquake with a moment magnitude (Mw) of 7.0 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of the San Francisco Bay area. The survey was designed to determine (1) whether the public understands the current life-safety objective of the building code’s seismic-design requirements, (2) what the public prefers in terms of the performance of the building stock in a large earthquake, (3) whether the public would be willing to pay the costs for stronger buildings, and (4) how important the public finds the issue of the seismic performance of buildings. The survey found that, without major regional differences, respondents: 5. By a large majority, are willing to pay for greater seismic safety, with the modal response (the most common response) being $3.00 per square foot additional construction cost to achieve such a higher level of performance; 6. Believe the degree of seismic performance of buildings is important or very important— the response of approximately 80 percent of respondents—even in the Central United States where earthquakes happen much less frequently than in California; and 7. Tend to be somewhat more commonly of European descent, wealthier, and more educated than the general public, but regression analyses found no strong trends in either region relating education to acceptable cost for better performance or relating household income to acceptable cost for better performance; 1. Are largely unaware of the life-safety seismic performance objective of American Society of Civil Engineers ASCE/SEI 7 and the International Building Code; 2. Are more interested in controlling total number of deaths and injuries in a large earthquake than in controlling per-building collapse probability; 3. Are also interested in more than the total number of casualties, specifically that buildings should remain functional or habitable after a large earthquake (the “Big One” in the language of the survey), and prefer better performance than the code is intended to deliver for new buildings; University of Colorado Boulder. 1 Key implications of the survey indicate that: 1. There is a potential need for writers of seismicdesign criteria in ASCE/SEI 7 to revisit the seismicperformance objectives for new buildings, considering the public’s apparent preferences for better performance; 2. There is a need for better communication with the public about the building code’s performance objectives for new buildings; 3. Practical options for stronger buildings are needed that an elected official can select in case they, like respondents here, want more from new buildings than the life-safety performance objective delivers; 80   The HayWired Earthquake Scenario—Engineering Implications 4. More narrowly, the study has some implications for the HayWired scenario. First, the HayWired study of the potential outcomes of designing stronger buildings addresses a real, measured public preference among Californians; 5. Elected officials in the San Francisco Bay area might be interested in hearing about public preferences for the seismic performance of new buildings; and 6. Those same elected officials might be interested in hearing about the costs and benefits of higher design requirements, in light of this study and the damage estimated by the HayWired scenario in general. Introduction This work deals with public preferences for the seismic performance of new buildings. Before addressing what the public might want, I first briefly discuss what the building code presently provides. ASCE/SEI 7–10 (American Society of Civil Engineers, 2010) recommends minimum design loads for buildings and other structures, including seismicdesign requirements. The 2012 International Building Code (IBC; International Code Council, 2012) adopts ASCE/SEI 7–10 by reference, and the majority of communities adopt the IBC, sometimes with minor modifications. Thus, ASCE/ SEI 7 tends to control how buildings perform in earthquakes. Oversimplifying somewhat, ASCE/SEI 7 helps to ensure new buildings are designed so less than 1 percent will collapse in earthquakes during the first 50 years of their existence, called their design life. As discussed in Porter (2016), studies by the Federal Emergency Management Agency (FEMA) and National Institute of Standards (NIST) imply that “fewer than 1 percent” means, on an expected-value basis, about 0.6 percent of buildings will collapse during their design life. ASCE/SEI 7 provides various other requirements to control repair costs, but essentially the 0.6-percent goal ensures a reasonable degree of life safety and is commonly referred to as the building code life-safety performance objective. In Porter (Societal Consequences, this volume), I showed that the current building code’s life-safety seismic performance objective has a serious unintended consequence—that a magnitude 7.0 earthquake on the Hayward Fault with an Oakland epicenter can impair hundreds of thousands of buildings, potentially displacing a million people or more. This earthquake is the mainshock of the U.S. Geological Survey’s (USGS) HayWired scenario, which examines a hypothetical earthquake with a moment magnitude (Mw ) of 7.0 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of the San Francisco Bay area. In the opinion of a USGS scientist who deals regularly with local governments, and based largely on her experience during the development of the ShakeOut earthquake scenario (Jones and others, 2008), city councils and mayors “absolutely do not know” how a code-compliant building stock designed to meet the life-safety objective will perform in a large earthquake, and are unsatisfied when they do learn of it (Lucile Jones, USGS, oral commun., November 19, 2013). The seismic-design requirements adopted by the building code are written with little input from the public. The 1927 Uniform Building Code (International Conference of Building Officials, 1927) provided the earliest seismic-design provisions in a model building code and was based on the experience and judgment of 60 building officials. In the subsequent 90 years, professional engineers and structural engineers (for the most part) have driven the development of seismic-design provisions. Developers of load- and resistance-factor design once called for a professionwide debate among structural engineers on the proper seismic reliability of new buildings (Ellingwood and others, 1980), but there is no record that such a debate ever took place, let alone a discussion with the public. Of course, state legislatures and city councils adopt or adapt model buildings codes such as the IBC on behalf of their community. By adopting these codes, there is a measure of consent by community officials. Suppose State and local officials’ ignorance of the life-safety objective only points to a failure to communicate on the part of engineers, code officials, or other building professionals. Better communication could conceivably address that problem. Imagine then an effort by structural engineers and others to explain what the building code does and does not provide. If such an effort were undertaken, then when a city or State adopts a model building code, one could say that well-informed city and State governments that adopt the IBC would be giving informed consent on behalf of the public to the risk imposed by the code? Davis (1991), a philosopher of engineering ethics, argues that if one cannot practically formulate and choose an alternative to a risk, one cannot give informed consent. Whether or not policymakers understand the code’s life-safety objectives, most cities and many States lack the resources to formulate an alternative to a model code and, therefore, cannot give informed consent. Those cities and States that cannot formulate and choose an alternative—the elected leaders and their constituents—represent “the public” in the sense meant by the American Society of Civil Engineers’ (2006) Canon 1 of its Code of Ethics, which holds that “Engineers shall hold paramount the safety, health and welfare of the public . . .” Being a member of the public in that sense has consequences— Davis (2015) has recently shown that the American Society of Civil Engineers’ Code of Ethics implicitly requires engineers to consult the public’s preferences when writing building standards for earthquakes. He excludes from “the public” the people who write the building code, such as engineers, building officials, and people who work in the building trades. In this work, I detail what may be the first effort to elicit the public’s preferences for the seismic performance of new buildings through a large, rigorous population survey. I conducted the survey (through the University of Colorado Boulder) as part of the HayWired scenario. See Davis and Porter (2016) for Chapter L. Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings 81 a summary of the survey, along with a discussion of who comprises the public in the present context and the ethics of consulting the public when establishing performance objectives for model building codes. Objectives This report addresses the question, what does the public expect and prefer from new code-compliant buildings in a big earthquake? Would the public be willing to pay to achieve their desired performance objective? It addresses these questions through population surveys in two high-hazard earthquakeprone regions. “High hazard” here means at least seismic-design category D as defined by ASCE/SEI 7-10 (American Society of Civil Engineers, 2010; fig. 1). The study employs a randomsample survey of adults 18 years or older living in two highseismic-hazard geographic regions. There are many ways to perform a survey; this study uses a web-based survey for efficiency. Web-based surveys have advantages and pitfalls, as discussed by Dillman and others (1998). Potential pitfalls include coverage error (the chance that some groups have no chance of selection), sampling error (differences between responses of a sample and those of the entire population), measurement error (resulting from poorly worded questions), and nonresponse error (the difference between responses received and ones that would have been received from people who declined to respond). Dillman and others (1998) offer a number of recommendations to overcome these pitfalls, as discussed later. The survey covers two geographic areas to probe for regional differences in public preferences. Those regional differences might reflect different degrees of recent experience with earthquakes, different wealth, or other regional issues. The two selected geographic regions for the web-based survey discussed in this report are California, virtually all of which is in seismic-design category D or above, and a part of the New Madrid Seismic Zone in the Central United States. In particular, respondents to represent people in the New Madrid Seismic Zone were recruited from residents of the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas (MSAs), both of which qualify as high-seismic-hazard areas as defined here. An earlier preliminary survey of 66 Californians suggested the public was largely unaware of the building EXPLANATION Seismic-design category A B C 0–10 10–20 20–30 E 30–45 45–60 60–112.5 >112.5 D N Figure 1.  Map of the contiguous United States showing risk-adjusted maximum considered earthquake (MCER) motion 1-second spectral response acceleration parameter (SM1) in percent of acceleration due to gravity (g) for seismic-design categories (SDC) A–E. Note that most of California and the New Madrid Seismic Zone in the Central United States—the areas examined in the web-based survey discussed in this report—are in seismic-design category D. (Map from Building Seismic Safety Council, 2009.) 82   The HayWired Earthquake Scenario—Engineering Implications code’s seismic performance objectives for new, codecompliant buildings; the public generally wanted more than life-safety performance from ordinary buildings, and that they would be willing to pay $3.00 to $10.00 per square foot more for a code-compliant building stock, in which the population could shelter in place after a large earthquake. The preliminary survey constituted a so-called convenience sample because the respondents could be conveniently accessed rather than because they represented a random sample. The preliminary survey also tested the survey instrument; meaning that I personally administered the survey several times and respondents were encouraged to ask questions. None of the questions they asked suggested the survey was unclear. Furthermore, Dr. Liesel Ritchie, a sociologist with expertise in surveys, reviewed the questions and expressed the opinion that the language was clear. Data from the preliminary survey are not included here, to ensure the convenience sample is not mixed with the random sample. See Porter and Davis (2015) for details of the preliminary survey. At least one other survey has attempted to ascertain the public’s confidence in the safety of existing buildings (International Association of Plumbing and Mechanical Officials and others, 2012). Those authors found, “The majority of Americans understands that building codes exist to keep us safe. However, beyond that basic fact their knowledge and appreciation of building codes appears weak. . . . Americans are generally confident that the structures where they live and work have adequate safeguards given the types of natural hazards in their areas.” The Urban Institute’s (1991) study of the public’s willingness to pay for life safety informed later decisions about acceptable cost to avoid statistical injuries and fatalities, such as can be found in Federal Highway Administration (1994). The present report may represent the first population survey to elicit the public’s understanding of how the code measures seismic performance, its quantitative objective (that is, numerically, how safe are new buildings), the public’s preferences for seismic performance, and its willingness to pay for performance in excess of current code requirements. This study is intended primarily to address willingness to pay for seismic safety of new buildings, a new risk context domain to which one can compare findings in other domains. The survey can also be compared with other methods to assess willingness to pay. Because a public university, the University of Colorado Boulder, performed the survey it satisfies regulations designed to implement ethical principles and preserve the public trust. The 1966 U.S. Public Health Service (PHS) policy, “Clinical research and investigation involving human beings,” requires an institution review board (IRB) to independently review research (Office of the Surgeon General, 1966). The 1974 PHS policy 45 CFR 46 specifies requirements for institutional assurances, IRB review, informed consent and ethical conduct. In 1991, 17 Federal agencies issued uniform regulations under the title “Common Rule.” The survey discussed here was approved by the University of Colorado’s IRB in light of these requirements. Survey Approach The survey was administered using an online survey company, SurveyMonkey.com, because of the ease, speed, and low cost of web surveys compared with alternatives such as randomdigit dialing. The survey aimed to elicit at least 400 responses from adults 18 years and older throughout the State of California and 400 adults in the Memphis and St. Louis MSAs. As shown later, it had 413 individual responses in California and 401 individual responses in Memphis and St Louis. All mentions of those two cities refer to the MSA, rather than the incorporated city. Because building professions have a significant role in establishing seismicdesign requirements, they do not qualify as “the public” in the present context, and so are excluded for the survey. The excluded professions include construction, structural design, architecture, building trades, building officials, and building inspectors. The sample size of 400 was chosen for each region because, for a population in excess of 10,000,000 people, a sample size of 400 provides a ±5-percent margin of error with 95-percent confidence. That is, responses to survey questions in each region are expected to be within 5 percent of what the population as a whole would say, with 95-percent confidence. Considering both regions together, the margin is approximately ±3.4 percent. (The accuracy of the estimates depends on how representative the sample is of the population; this question is addressed later.) The large sample size helps to overcome the potential for coverage error. As shown later, responses were received from all major demographic groups (considering gender, income, education, age, race, and ethnicity). Sampling error is inevitable when one cannot survey the entire population, but the large sample size limits the potential error. Measurement error was minimized through the preliminary survey, as previously discussed. Nonresponse error was addressed by keeping the survey brief. As will be shown later, the high response rate limits nonresponse error. The survey was administered by a survey company according to a human-subjects research protocol approved by the University of Colorado Boulder’s IRB on July 2, 2015. Survey responses were collected in July 2015. The survey asked the following general questions: 1. Do any of the following apply to you? (Asking if respondent belongs to any of the excluded building professions that have a relation to existing building codes) 2. Respondent’s role or relation to building codes? (If disqualified from further answering the survey by answering yes to question 1) 3. Performance objectives that most new buildings are intended to meet in a large earthquake? 4. What should the building code provide? 5. Preferred measure of seismic performance? 6. Acceptable cost to increase seismic performance? 7. How important are the issues raised by the survey? Chapter L. Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings 83 8. Age? Sampling Procedure 9. Gender? We email invitations to respondents who match your targeting criteria. Our system selects a random group from the SurveyMonkey Contribute member base who match the demographic targeting criteria you requested. 10. Race or ethnicity? 11. Education? 12. Household income? Individual survey responses were aggregated in the form of pie diagrams. Respondents took on average 6 minutes to complete the survey. We use a standard template email notification to notify respondents that they have a new survey to take. It’s not possible to customize the invitation email sent by SurveyMonkey Contribute. Targeting Criteria Respondent Population and Sampling Procedure We target members based on the information they provide to us in their profile. The survey aimed to sample at least 400 adults in California and 400 adults in the combination of the Memphis and St. Louis metropolitan statistical areas, excluding adults involved in a buildings profession. SurveyMonkey.com offers this explanation of its sampling procedure: The more variables or criteria you target, the more it constricts the population we can use to build your sample. A more constricted sample may slow down the pace at which your survey can complete—or even make it impossible for us to run your survey at all. We take great care to ensure that we have a diverse group of members who are interested in sharing their opinions with you. When a panelist joins our community of respondents and becomes a SurveyMonkey Contribute member, they fill out a profile. This profile contains demographic questions (gender, age, region, etc.) as well as some other targeting characteristics you might care about (cell phone usage, job type, and more). Incentive Structure Each time a SurveyMonkey Contribute member completes an eligible survey, SurveyMonkey makes a contribution to a charity of the member’s choice, and the member can choose to enter a sweepstakes. Recruitment We recruit Contribute members from a diverse population of 45+ million people who take SurveyMonkey surveys every month. For example, after completing a survey, respondents are redirected to a page that may feature an advertisement for SurveyMonkey Contribute. SurveyMonkey Contribute panelists come from the United States, the United Kingdom, and Australia. You choose the country you’d like your respondents to come from. If you need respondents from other countries, please contact us. Although we recruit panelists ages 13 and up, we have the ability to target respondents by age and can target 18 and older. Balancing If you send your survey to a general audience, your results are generally representative of the population you’re surveying. We automatically balance results according to census data for age and gender, whereas location tends to balance out naturally. Balancing precision and granularity improves as the number of responses increases. When you choose specific targeting criteria, your results are no longer representative of the general population because you’re purposefully focusing on a particular subset of the population. (SurveyMonkey, 2015) Survey Questions and Responses The survey asked 12 questions to determine the public’s preferences for the seismic performance of new buildings. The initial survey question asks whether the respondent is employed in the building industry. Employment in that industry disqualifies the respondent from answering further questions. Incomplete responses are all the people who were invited to take the survey but who declined to begin it (N. Teckman, written commun., July 14, 2015). Response Rate The survey was sent to a total of 1,506 potential participants. The response rate was quite high—60 percent of those eligible in California and 56 percent of those eligible in St. Louis and 84   The HayWired Earthquake Scenario—Engineering Implications Memphis. Here, response rate refers to the ratio of “completed” to the sum of “completed” and “incomplete” responses for example, 413/(413 + 278)=60 percent. The pie diagrams in figure 2 show the number of people disqualified, number of completed responses, and number of incomplete responses to the following question: A 1. Do any of the following apply to you? (Check all that apply.) • Employed in the construction industry? Completed 413 (55%) • Employed in the structural design industry? Incomplete 278 (37%) • Employed in the architecture industry? • Employed in the building-trades industry? • Employed as a building official or building inspector? Disqualified 62 (8%) • None of these apply? Role or Relation to Building Codes B Figure 3 shows the number and percentage of responses from people disqualified to the following question: 2. What is your role or relation to building codes? Please mark all your roles that apply? • Local elected official? • Local government staff who advise local officials on building codes? Completed 401 (53%) Incomplete 309 (41%) • Building owner? • Building tenant (renter)? • Other (please specify)? (Detailed information about respondent demographics is discussed below in Respondent Demographics.) Figure 4 shows the relation of respondents to building codes. (Responses to this and all subsequent questions reflect only completed responses, for example, from respondents who were qualified for and completed the survey.) Current and Preferred Code Objectives Figure 5 shows the percentage of responses to the following question: 3. Although local codes vary, which of these performance objectives do you think most new buildings are intended to meet in a large earthquake? That is, what do you think the current code actually says, not what should it say. Please mark only one answer. • New buildings will generally be functional after an earthquake, and will require minimal repairs. Disqualified 43 (6%) Figure 2.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing response rate in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Completed, those completing the survey; Incomplete, those who were invited to take the survey but who declined to begin it; disqualified, those disqualified from the survey. Number and percentage (%) of respondents are shown. • New buildings will generally be occupiable after an earthquake. Although they might require some repairs to be fully functional, the occupants will be able to remain in the building during the repairs. • New buildings are safe enough that occupants won’t be killed, but are not generally intended to be occupiable after the earthquake. That is, a person will be able to exit a building safely, but not necessarily be able to go back in. • I don’t know. Chapter L. Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings 85 A B Building official or building inspector 4 (13%) Building official or building inspector 4 (18%) Structural design 6 (19%) Architecture 6 (19%) Building trades 15 (49%) Elected official 0.25% B Owner 27.3% Other 28.6% Architecture 7 (32%) Building trades 7 (32%) Elected official Staff 2.5% 2.5% A Structural design 4 (18%) Tenant 39.0% Staff 0.25% Owner 35.9% Other 41.1% Tenant 22.5% Other 3% Figure 3.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing relation to building codes for people disqualified from participating in the survey in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Number and percentage (%) of respondents are shown. Figure 4.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing relation to building codes in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Percentage (%) of respondents is shown. Other 3% A B Functional 18% Functional 9% Do not know 31% Do not know 43% Occupiable 22% Occupiable 22% Life safe 26% Life safe 23% Figure 5.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “Which of these performance objectives do you think most new buildings are intended to meet in a large earthquake?” in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Percentage (%) of respondents is shown. 86   The HayWired Earthquake Scenario—Engineering Implications Figure 6 shows the percentage of responses to the following question: That is, a person should be able to exit a building safely, but not necessarily be able to go back in. 4.   What should the building code provide? That is, if someone builds a new building in your community and it meets buildingcode requirements for seismic safety, which one of these would you most prefer the code to ensure? In some of the responses below I use the term “the Big One,” by which I mean an earthquake that might be considered a once-in-a-lifetime event. Please mark only one answer. • New buildings should generally be functional after the Big One, possibly requiring minimal repairs. • New buildings should generally be occupiable after the Big One. Although they might require some repairs to be fully functional, the occupants should be able to remain in the building during the repairs. • I don’t know. Preferred Performance Measure Figure 7 shows the percentage of responses to the following question: 5. Which of these building performance measures do you think is of greatest interest to your community? That is, if the building code controlled only one of these measures, which one should it control? Again, “the Big One” here means an earthquake that might be considered a once-in-alifetime event. Please mark only one answer. • The chance that any given building will collapse in the Big One. (“Per-building collapse probability” in fig. 7.) • New buildings should be safe enough that occupants won’t be killed, but need not be occupiable after the Big One. Other 2% A Figure 6.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “If someone builds a new building in your community and it meets building-code requirements for seismic safety, which one of these would you most prefer the code to ensure?” for people in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Percentage (%) of respondents is shown. Figure 7.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “Which of these building performance measures do you think is of greatest interest to your community?” in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Percentage (%) of respondents is shown. Other 2% B Functional 19% Do not know 16% Functional 16% Do not know 18% Life safe 21% Life safe 24% Occupiable 40% Occupiable 42% A B Other 7% Per-building collapse probability 23% Repair cost 9% Per-building collapse probability 14% Unoccupiable buildings AAXXXX_fig 01 Number of collapses 12% Repair cost 10% Unoccupiable buildings 9% 11% Community casualties 38% Other 9% Community casualties 48% Number of collapses 10% Chapter L. Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings 87 • The total number of people killed or injured by building damage in your community in the Big One. (“Community casualties” in fig. 7.) • The total number of buildings in your community that might collapse in the Big One. (“Number of collapses” in fig. 7.) • The number of buildings that would not be occupiable after the Big One. (“Unoccupiable buildings” in fig. 7.) • The total cost to repair damaged buildings in your community in the Big One. • Something else or some combination of these (please specify). Responses to this question beg for some interpretation. The responses suggest respondents are more interested in controlling total number of deaths and injuries in a large earthquake than in controlling per-building collapse probability. Note well that the two measures are not the same. Although building collapse drives deaths and injuries in earthquakes, the per-building probability is blind to the number of simultaneous collapses. A large remote earthquake can subject a small town to very strong motion and produce high collapse probability among that small number of buildings. A large earthquake in a metropolitan area can subject a large number of buildings to the same level of motion and the same collapse probability. ASCE/SEI 7 does not distinguish between the two cases, but the public does. The public cares very much about the total simultaneous numbers. In one of the most cited studies on public risk perception, Slovic and others (1981) show that the leading factor affecting the public’s perception of risk is “associated with lack of control, fatal consequences, high catastrophic potential, reactions of dread, inequitable distribution of risks and benefits (including transfer of risks to future generations), and the belief that the risks are increasing and not easily reducible.” They refer to factor 1 as dread risk. This phenomenon partially explains why Americans tolerate more than 32,000 deaths per A year as a result of automobile accidents and more than 11,000 annual firearm homicides (both of which cause at most only a few deaths at a time), but found the 2,996 (nearly simultaneous) deaths in the September 11, 2001, attacks on the United States traumatic enough to launch two wars that ultimately cost more than $1 trillion (Daggett, 2010). Acceptable Cost for Better Performance Figure 8 shows the percentage of responses to the following question: 6.  This question is intended to obtain information about the tradeoffs between safer buildings and higher initial construction cost (not retrofit cost). Suppose that in the Big One (a oncein-a-lifetime earthquake), up to 1 out of every 5 buildings in your community would collapse or require major repairs, taking a year or more to repair before they could be reoccupied. Also suppose that you could change the building code so that it would reduce that fraction to 1 in 100 buildings or less, but at the cost of higher initial construction costs. What additional cost do you think building buyers should be willing to pay to achieve that end? Please mark only one answer. • The current risk is already tolerable. No additional cost seems justified. • Maybe $1 per square foot, which would increase the monthly mortgage for the purchase of a new, typical California house from about $2,000 per month to about $2,010 per month. [In St Louis and Memphis, instead of “...California... $2,000... $2,010....” the question says “...St Louis... $750... $758....”] • Maybe $3 per square foot, which would increase the monthly mortgage for the purchase of a new, typical California house from about $2,000 per month to about $2,030 per month. [In St Louis and Memphis, instead of “...California... $2,000... $2,030....” the question says “...St Louis... $750... $770....”] B $0/ft 14% 2 Do not know 16% Do not know 19% $1/ft2 19% $10/ft2 23% $3/ft2 28% $0/ft 10% 2 $1/ft2 21% $10/ft2 16% $3/ft2 34% Figure 8.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “What additional cost do you think building buyers should be willing to pay to achieve that end [acceptable tradeoff between safer buildings and higher initial construction cost?]” for people in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Percentage (%) of respondents is shown. 88   The HayWired Earthquake Scenario—Engineering Implications • Maybe $10 per square foot, which would increase the monthly mortgage for the purchase of a new, typical California house from about $2,000 per month to about $2,100 per month. [In St Louis and Memphis, instead of “...California... $2,000... $2,100....” the question says “...St Louis... $750... $824....”] • White/Caucasian • I don’t know, or you would have to measure the cost some other way (please specify). • Native American How Important are These Issues? Figure 9 shows the percentage of responses to the following question: 7. How important are these issues? Please mark only one answer. • Very important • Important • Not very important • Unimportant Respondent Demographics Figure 10 shows the percentage of responses to the following question: 8. Age? • <18 • 18–29 • African American • Hispanic • Asian • Pacific Islander • Other Figure 13 shows the percentage of responses to the following question: 11. What is the highest level of education you have received? • Less than high school • High school/GED • Some college • 2-year college degree • 4-year college degree • Masters degree • Doctoral degree • Professional degree (JD, MD) Figure 14 shows the percentage of responses to the following question: 12. How much total combined money did all members of your HOUSEHOLD earn last year? • 30–44 • $0 to $9,999 • 45–59 • $10,000 to $24,999 • 60+ • $25,000 to $49,999 Figure 11 shows the percentage of responses to the following question: • $50,000 to $74,999 9. Gender? • $100,000 to $124,999 • Female • Male Figure 12 shows the percentage of responses to the following question: 10. Which of the following best describes your race or ethnicity? (Check all that apply.) • $75,000 to $99,999 • $125,000 to $149,999 • $150,000 to $174,999 • $175,000 to $199,999 • $200,000 and up • Prefer not to answer Chapter L. Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings 89 Unimportant Unimportant 5% 6% A B Not very important 12% Not very important 15% Very important 38% Important 45% Very important 27% Important 52% <18 0% <18 0% A B 60+ 25% 18–29 25% 18–29 25% 60+ 25% 45–59 25% Figure 9.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “How important are these issues?” for people in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Percentage (%) of respondents is shown. 30–44 28% 30–44 28% 45–59 25% Figure 10.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “[What is your] Age?” for people in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Percentage (%) of respondents is shown. AAXXXX_fig 01 A B Male 31% Female 53% AAXXXX_fig 01 Male 47% Female 69% Figure 11.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “[What is your] Gender?” for people in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Percentage (%) of respondents is shown. 90   The HayWired Earthquake Scenario—Engineering Implications A B 3 11 8 2 14 2 9 2 68 1 86 3 EXPLANATION African American Native American Hispanic Pacific Islander Asian Other Caucasian Figure 12.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “Which of the following best describes your race or ethnicity?” for people in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Bold numbers show percentage of respondents. A 3 5 2 B 8 18 4 3 1 9 24 18 21 11 10 33 30 EXPLANATION AAXXXX_fig 01 Less than high school 2-year degree PhD degree High school/GED 4-year degree JD, MD degree Some college Masters degree Figure 13.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “What is the highest level of education you have received?” for people in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Bold numbers show percentage of respondents. HS/GED, high school degree or General Educational Development certificate; JD, doctor of jurisprudence; MD, doctor of medicine. Chapter L. Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings 91 A B 5 15 4 18 9 9 17 2 3 7 4 17 2 4 4 7 8 12 18 9 14 12 EXPLANATION Household income, in thousands of dollars $0–9,999 $50,000–74,999 $125,000–149,999 $175,000–199,999 $10,000–24,999 $75,000–99,999 $150,000–174,999 >$200,000 $25,000–49,999 $100,000–124,999 $150,000–174,999 Figure 14.  Pie diagrams for survey of the public’s preferences for the seismic performance of new buildings showing responses to “How much total combined money did all members of your HOUSEHOLD earn last year?” for people in (A) California and (B) in the St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas. Bold numbers show percentage of respondents. Are Respondents Representative of the Population? This section addresses differences between the demographics of survey respondents and the population of their state or community. If one is to draw any conclusions about what American adults think based on what these respondents said, one must ask how representative respondents are of the larger group. According to SurveyMonkey (2015), respondents “represent a diverse group of people and are reflective of the general population. However, as with most online sampling, respondents have Internet access and voluntarily joined a program to take surveys.” Most of the U.S. population regularly uses the internet—87 percent according to the Pew Research Center (2014). However, not all volunteer to take SurveyMonkey surveys. Respondents receive an incentive to take surveys. “Each AAXXXX_fig 01 time a SurveyMonkey Contribute member completes an eligible survey, SurveyMonkey makes a [$0.50] contribution to a charity of the member’s choice, and the member can choose to enter a sweepstakes.” I quantify how representative the survey seems to be. The demographics of the survey do deviate from those of the population as a whole. More women than men responded to the survey in both regions. Responses included 496 women and 317 men, or a 3:2 ratio rather than the approximately 1:1 ratio in the general public (fig. 15A). Respondents were also generally wealthier than the population (fig. 15B). California income data are taken from U.S. Census Bureau’s (2015) American Community Survey for California households in 2013. Median household income data for Memphis and St. Louis MSAs in 2010 were taken from U.S. Conference of Mayors (2012). Respondents in both regions report more education than the population as a whole (fig. 16). Educational attainment of Californians is as of 2009 from U.S. Census Bureau (2012). Memphis MSA data are for 2014 from Memphis Chamber of Commerce (2015). Data for St. Louis are for 2012 from University of Missouri St. Louis, Public Policy Research Center (2014). More respondents are of European (white/Caucasian) descent than the population as a whole (fig. 17). Race and ethnicity data are taken from U.S. Census Bureau (2011). Survey respondents seem to be more likely to be of European descent than the general population in either region. They are somewhat wealthier and have more education than the general population. These differences may produce a difference between survey participants’ responses and those of the general population. The next section tests that hypothesis. 92   The HayWired Earthquake Scenario—Engineering Implications 100 New Madrid Seismic Zone California A $100,000 B EXPLANATION 75 50 $75,000 St. Louis $50,000 Memphis $25,000 25 0 New Madrid Seismic Zone California Median household income, in dollars Gender, as percentage of population Female Male Sample Population Sample Population $0 Sample Population Sample Population Figure 15.  Histograms of respondent (A) gender and (B) median household income for survey of the public’s preferences for the seismic performance of new buildings in the California and St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas (MSA). EXPLANATION California sample California population NMSZ sample High school or less Figure 16.  Histogram of respondent education attainment for survey of the public’s preferences for the seismic performance of new buildings in the California and St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas (MSA) as compared to relevant populations 25 years and older. %, percent; HS/GED, high school degree or General Educational Development certificate; BS, bachelor’s degree; MS, master’s degree; CA, California; NMSZ, New Madrid Seismic Zone; <, less than. Memphis population St. Louis population At least high school or GED At least Bachelors degree Masters and professional degree or higher 0 25 50 75 Education attained, as percentage of population 100 Chapter L. Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings 93 Caucasian Figure 17.  Histogram of respondent race or ethnicity for survey of the public’s preferences for the seismic performance of new buildings in the California and St. Louis, Missouri, and Memphis, Tennessee, metropolitan statistical areas (MSA) as compared to relevant populations. %, percent; NMSZ, New Madrid Seismic Zone. African American Hispanic EXPLANATION California sample California population Asian NMSZ sample Memphis population St. Louis population Other 0 25 50 75 100 Race, as percentage of population Are Some Groups Willing to Pay More for Better Seismic Performance? This section examines whether some groups are more willing to pay for better seismic performance than others. Better seismic performance is measured here in terms of building impairment. Suppose that people also equate better seismic performance with more life safety. In such a case, one could compare these survey results with expectations from other methods to estimate willingness to pay for safety. There is a rich literature addressing people’s willingness to pay for safety; I consider just a few sources and compare what one may predict from them with observations from the present survey. Needleman (1982) examines several methods to value life safety, including lifetime earning potential (the human capital approach), questionnaires to ascertain people’s willingness to pay to reduce their own risk, observed willingness to pay to reduce risk, and observed willingness to take on additional risk for extra pay. He found that the last method produced the most reliable estimates of people’s valuation of small changes in their own risk, and the upper bound, the value of avoiding a fatality, is equivalent to 20 times average annual income. In Porter (2002), I briefly review various methods to estimate willingness to pay for more safety in earthquakes, adding to Needleman’s list the application of Stanford-style decision analysis, such as proposed by Howard (1980, 1989). In that work I show that, if people behaved as predicted by decision analysis, they should be more willing to pay for life safety in direct proportion to their annual consumption (roughly equivalent to household income), which is consistent with Needleman’s finding. I also show that under Howard’s decision-analysis framework, older people should be less willing to pay for life safety than younger people, and men should be less willing to pay for life safety than women. Based on these works, one might suspect that the survey results would exhibit a relation between household income and willingness to pay for better performance or between age and willingness to pay. If either were true, the survey would exhibit a sampling error—a difference between survey responses and the opinions of the broader population. However, a regression analysis of the data from the present survey found no strong trend relating household income and willingness to pay for better building performance. Nor do the survey responses show a relation between education, age, or gender and willingness to pay. 94   The HayWired Earthquake Scenario—Engineering Implications The coefficient of determination (R2), between individual respondents’ household income and acceptable cost for better performance was 0.010 among 350 Californians who provided both quantities. It was 0.0001 among 303 people from St. Louis and Memphis who provided both quantities. Regression analysis to relate acceptable cost and age in years similarly exhibited low coefficients of determination—0.001 in St. Louis and Memphis, 0.002 in California. Men and women were approximately equally willing to pay for better seismic performance—in St. Louis and Memphis, women were willing to pay 13 percent more on average than men, but in California, men were willing to pay 6 percent more than women. One might also suppose that years of secondary and postsecondary education would correlate with acceptable cost for better performance. Among 413 Californians who answered both questions, R2=0.020, and among 170 people from St Louis and Memphis who provided both quantities, R2=0.022. Nor does one derive higher coefficients of determination from nonlinear regression, fitting polynomials of second, third, or fourth order. (Actually the R2 values do rise slightly, but the rise has more to do with overfitting than with more information.) All these coefficients of determination are too low to reject the null hypothesis at the 5-percent significance level, where the null hypothesis is that the correlation coefficient (ρ) is in fact zero. Put another way, there appears to be no strong relation between education and acceptable cost for better performance, between household income and acceptable cost for better performance, or between respondent age and acceptable cost for better performance. Conclusions As part of the HayWired scenario’s study of the unintended consequences of the building code’s seismic performance objective, I undertook a public survey in two highly seismically active regions of the United States—one of California adults and another of adults in two metropolitan areas near the New Madrid Seismic Zone in the Central United States, namely the Memphis and St. Louis MSAs. Sample sizes in both regions (413 adults in California and 401 near the New Madrid Seismic Zone) ensure that the results reflect the opinions of the public with ±5-percent margin of error with 95-percent confidence, at least insofar as respondents resemble the public. The purpose of the survey was to determine whether the public understands the current life-safety objective of the building code’s seismic-design requirements; what the public prefers in terms of the performance of the building stock in a large earthquake; whether the public would be willing to pay the cost of stronger buildings; and how important the public finds the issue of the seismic performance of buildings. The survey found that respondents in both regions: 1. Are largely unaware of the life-safety seismic performance objective of ASCE/SEI 7 and the International Building Code; 2. Are more interested in controlling total number of deaths and injuries in a large earthquake than in controlling perbuilding collapse probability; 3. Are also interested in more than the total number of casualties, specifically that buildings should remain functional or habitable after a large earthquake (the “Big One” in the language of the survey), and prefer better performance than the code is intended to deliver for new buildings; 4. Expect better seismic performance than ASCE/SEI 7 intends to provide; 5. By a large majority, are willing to pay for greater seismic safety, with the modal response (the most common response) being $3 per square foot additional construction cost to achieve such a higher level of performance; 6. Believe that the degree of seismic performance of buildings is important or very important—the response of approximately 80 percent of respondents, even in the Central United States where earthquakes happen much less frequently than in California; 7. Are more commonly of European descent, wealthier, and more educated than the general public, but regression analyses found no strong trends in either region relating education to acceptable cost for better performance or relating household income to acceptable cost for better performance. Study Implications The survey implies several needs: 1. The Building Seismic Safety Council, which originates many innovations of ASCE/SEI 7, could revisit the performance objectives that underlie the seismic-design criteria of ASCE/SEI 7, considering (among other issues) the public’s preferences as elicited here. 2. Structural engineers could communicate better with elected officials who adopt building codes if they want to reduce the public’s apparent misunderstanding of the code’s performance objectives for new buildings. Engineers might, for example, create brief (1 page or so) documents about the community-level outcomes of the “Big One,” written in plain English, and targeted to elected officials, building owners, urban planning organizations, and other stakeholders among the public (as meant here). Such documents could perhaps be included in an appendix to ASCE/SEI 7, or distributed by other means such as by the National Institute of Building Sciences to the United States Conference of Mayors, and the Building Owners and Managers Association International. Chapter L. Not Safe Enough—A Survey of Public Preferences for the Seismic Performance of New Buildings 95 3. Practical options are needed that elected officials can select in case they find the code’s outcomes unacceptable. For example, an appendix of ASCE/SEI 7 or of the International Building Code could explain the costs and benefits of higher design requirements and offer optional adoption language to increase required design strength. The optional language might locally modify the code to require all ordinary buildings (called “risk category II” in ASCE/SEI 7-10) to be designed with a seismic importance factor of 1.5. More narrowly, the study has some implications for the HayWired scenario. It suggests that: Juliette Hayes, Kevin Mickey, Evan Reis, and Phil Schneider. Dr. Liesel Ritchie reviewed survey questions and provided valuable advice about wording, demographic questions, and survey size. Sharyl Rabinovici (private consultant), Phil Schneider (National Institute of Building Sciences), and Anne Wein (USGS) provided peer reviews and valuable constructive criticism. The Association of Bay Area Governments and the Association of Contingency Planners provided opportunities to test and administer the preliminary survey. Anne Wein performed some preliminary surveys at meetings of the Association of Bay Area Governments. The author thanks all these people and organizations. 1. The HayWired scenario study of the potential outcomes of designing stronger buildings (Porter, Societal Consequences, this volume) addresses a real, measured public preference for more resilient buildings among Californians. References Cited 2. Elected officials in the San Francisco Bay area might be interested in hearing about public preferences for the seismic performance of new buildings. American Society of Civil Engineers, 2006, Code of ethics: American Society of Civil Engineers web page, accessed January 28, 2015, at http://www.asce.org/code_of_ethics/. 3. Those same elected officials might be interested in hearing about the costs and benefits of higher design requirements in light of this study and the damage estimated by HayWired in general. American Society of Civil Engineers, 2010, Minimum design loads for buildings and other structures: Reston, Va., Structural Engineering Institute, ASCE/SEI 7–10, 608 p. Limitations and Research Needs Respondents spanned the domain of age, gender, race and ethnicity, educational attainment, and income of both regions, so there are no significant unrepresented groups. However, this survey only examined two regions. It may be that people in other high-hazard areas have different preferences. Further regression analysis might detect a trend relating other parameters to preferred performance level or to acceptable cost for better performance. Respondents seemed to understand the questions, especially regarding acceptable cost for better performance, but it may be that they would act differently when actually confronted with a real purchasing decision. The present survey only begins to study the public’s preferences for the seismic performance of new, code-compliant buildings. More research might better measure any differences between what the code provides and what the public wants. It may be interesting to survey engineers and others involved in the building trades to explore whether and how their preferences differ from those of the public. It would also be interesting to explore the question of preferred performance measure more deeply, such as degree of preference for each option or alternative ways of dealing with probability. Acknowledgments Survey questions were drafted with the assistance of the National Institute of Building Sciences’ Multihazard Mitigation Council Public Expectations Subcommittee—Gary Ehrlich, Building Seismic Safety Council [BSSC], preparer, 2009, NEHRP recommended seismic provisions for new buildings and other structures, 2009 ed.: Washington, D.C., Federal Emergency Management Agency publication P-750, 388 p., accessed May 21, 2015, at http://www.fema.gov/medialibrary-data/20130726-1730-25045-1580/femap750.pdf. Daggett, S., 2010, Costs of Major U.S. Wars: Washington, D.C., Congressional Research Service, 7-5700, RS22926, 5 p., accessed December 29, 2015, at https://www.fas.org/sgp/crs/ natsec/RS22926.pdf. Davis, M., 1991, Thinking like an engineer—The place of a code of ethics in the practice of a profession: Philosophy and Public Affairs, v. 20, no. 2, p. 150–167. Davis, M., 2015, What part should the public have in writing engineering standards?, in Security and Disaster Preparedness Symposium—Codes and Governance in the Built Environment: National Institute of Building Sciences Third Annual Conference and Expo, Building Innovation, 2015— Creating High-Performing Resilient Communities, January 6–9, 2015, Washington D.C., presentations available at https:// www.nibs.org/store/ViewProduct.aspx?id=4108575. Davis, M., and Porter K., 2016, The public’s role in seismic design provisions: Earthquake Spectra, v. 32, no. 3, p. 1345– 1361, https://doi.org/10.1193/081715EQS127M. Dillman, D.A., Tortora, R.D., and Bowker, D., 1998, Principles for constructing web surveys: American Statistical Association, Proceedings of the Joint Meetings of the American Statistical Association, August 9–13, 1998, Dallas, Texas. 96   The HayWired Earthquake Scenario—Engineering Implications Ellingwood, B., Galambos, T.V., MacGregor, J.G., and Cornell, C.A., 1980, Development of a probability-based load criterion for American National Standard A58: Washington D.C., National Bureau of Standards, Special Publication 577, 222 p. Federal Highway Administration, 1994, Technical Advisory— Motor Vehicle Accident Costs: Washington, D.C., U.S. Department of Transportation Technical Advisory #7570.2, 5 p. Howard, R.A., 1980, On making life and death decisions, in Shwing, R.C., and Albers, W.A., eds., Societal Risk Assessment: New York, Plenum Press, p. 89–106. Howard, R.A., 1989, Microrisks for medical decision analysis: International Journal of Technology Assessment in Health Care, v. 5, p. 357–370. International Association of Plumbing and Mechanical Officials, International Code Council, National Association of Mutual Insurance Companies, National Institute of Building Sciences, National Fire Protection Association, and Insurance Institute for Business and Home Safety, 2012, Public Survey on Building Codes and Building Safety: Washington, D.C., National Institute of Building Sciences, 3 p. International Conference of Building Officials, 1927, Uniform Building Code: Whittier, Calif., International Conference of Building Officials, 265 p. International Code Council, 2012, International Building Code 2012: Country Club Hills, Ill., 690 p. Jones, L.M., Bernknopf, R., Cox, D., Goltz, J., Hudnut, K., Mileti, D., Perry, S., Ponti, D., Porter, K., Reichle, M., Seligson, H., Shoaf, K., Treiman, J., and Wein, A., 2008. The ShakeOut Scenario. U.S. Geological Survey OpenFile Report 2008–1150 and California Geological Survey Preliminary Report 25, http://pubs.usgs.gov/of/2008/1150/. Memphis Chamber of Commerce, 2015, Detailed demographics: Greater Memphis Chamber of Commerce web page, accessed September 1, 2015, at http://www. memphischamber.com/Articles/DoBusiness/pdfMemphis_ MSA_Demographics.aspx. Needleman, L., 1982, Methods of valuing life, in Lind, N.C., ed., Technological Risk: Waterloo, Ontario, University of Waterloo Press, p. 89–99. Office of the Surgeon General, 1966, Surgeon General’s directives on human experimentation—Index clinical research human subjects, investigations involving individuals, rights and welfare of: Office of the Surgeon General, p. 350–355, accessed September 8, 2015, at https://history.nih.gov/research/downloads/ Surgeongeneraldirective1966.pdf. Pew Research Center, 2014, The web at 25: Pew Research Center web page, accessed August 17, 2015, at http://www. pewinternet.org/2014/02/25/the-web-at-25-in-the-u-s. Porter, K.A., 2002, Life-safety risk criteria in seismic decisions, in Taylor, C.E., and VanMarcke, E., eds., Acceptable Risk Processes—Lifelines and Natural Hazards: Reston, Va., American Society of Civil Engineers, Technical Council for Lifeline Earthquake Engineering, Monograph 21, accessed September, 27, 2015, at http://spot.colorado. edu/~porterka/Porter-2002-Life-Safety-Risk-Criteria.pdf. Porter, K., and Davis, M., 2015, Not safe enough—The public’s expectations for seismic performance: Journal of the National Institute of Building Sciences, no. 3, p. 22–25. Slovic, P., Fischhoff, B., and Lichtenstein, S., 1981, Perceived risk—Psychological factors and social implications, in The Assessment and Perception of Risk: London, The Royal Society, p. 17–34 SurveyMonkey, 2015, SurveyMonkey Audience: SurveyMonkey web page, accessed July 14, 2015, at http:// help.surveymonkey.com/articles/en_US/kb/SurveyMonkeyAudience. University of Missouri St. Louis, Public Policy Research Center, 2014, Metropolitan mirror—Facts and Trends Reflecting the St. Louis Region—Changes in educational attainment for the St. Louis Region, 2009 to 2012: University of Missouri St. Louis, Public Policy Research Center, accessed July 19, 2015, at https://pprc.umsl.edu/ pprc.umsl.edu/data/Metro-PDFS/MM-EdAtt-Nov14.pdf. Urban Institute, 1991, The costs of highway crashes, final report: Washington, D.C., The Urban Institute, Federal Highway Administration contract DTFH61-85-C-00107, 144 p. U.S. Conference of Mayors, 2012, U.S. metro economies—2012 employment forecast and the impact of exports: U.S. Conference of Mayors, accessed July 19, 2015, at http://usmayors.org/pressreleases/uploads/2012/ MetroEconomiesReport_011812.pdf. U.S. Census Bureau, 2011, 2010 Census interactive population search: U.S. Census Bureau web page, accessed July 21, 2015, at http://www.census.gov/2010census/popmap/ ipmtext.php?fl=06. U.S. Census Bureau, 2012, The 2012 statistical abstract— Education—Educational attainment: U.S. Census Bureau web page, accessed December 29, 2015, at https:// www.census.gov/library/publications/2011/compendia/ statab/131ed/education.html. U.S. Census Bureau, 2015, State and county quickfacts: U.S. Census Bureau web page, accessed July 17, 2015, at http:// quickfacts.census.gov/qfd/states/06000.html. The HayWired Earthquake Scenario—Engineering Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q [Also see https://doi.org/10.3133/sir20175013] Chapter M An Earthquake Urban Search and Rescue Model for Earthquake Response and its Application to the HayWired Scenario By Keith A. Porter1 Abstract Few researchers have examined the potential for earthquake-induced building collapses and electrical failures to trap building occupants. The HayWired earthquake scenario examines a hypothetical moment magnitude (Mw) 7.0 earthquake (mainshock) occurring on April 18, 2018, at 4:18 p.m., on the Hayward Fault in the east bay part of the San Francisco Bay area. To estimate demand for urban search and rescue (USAR) related to building collapse in the HayWired scenario, I compiled a database of photographic evidence of 73 building collapses in California earthquakes between 1965 and 2014. The database includes all images in the University of California, Berkeley, National Information Service for Earthquake Engineering (NISEE) e-library whose descriptions use any of the words “collapse,” “fail,” “fell,” or “parapet,” along with data taken from other sources about 14 additional buildings. I interpreted each image to estimate the fraction of building area that collapsed and the fraction of occupants in the collapsed area who would realistically be trapped and require extrication by others. The proportions vary by structural material, but on average, collapse involves 23 percent of the area of the buildings and traps 66 percent of the occupants in the collapsed area. Using this new knowledge and other information about the number of collapsed buildings, I can estimate the number of people requiring extrication by USAR personnel. In the case of the HayWired scenario, two alternative methods suggest approximately 5,000 buildings could collapse. The two methods are described elsewhere, but briefly, they are Hazus-MH (Seligson and others, this volume) and the approach described in Porter (2015). The latter essentially relies on the model of building collapse underlying the maps of risk-adjusted maximum considered earthquake ground motions in ASCE Standard ASCE/SEI 7-10 Minimum Design Loads for Buildings and Other Structures (American Society 1 University of Colorado Boulder. of Civil Engineers, 2010). It seems realistic that 2,500 people would be trapped in 5,000 collapsed buildings. (Not every building collapse traps people). If all buildings were designed to be 50 percent stronger than currently required under the International Building Code, both figures could be reduced by approximately a factor of four. Using statistics about how many elevators there are in the United States, how many have emergency power, and what fraction of them are occupied and traveling between floors, I estimated that loss of power in the HayWired mainshock could trap 22,000 people in 4,500 stalled elevators, placing additional demands on USAR personnel. If newer elevators were provided with emergency power, the number trapped in elevators could be reduced to 14,000 people in 3,000 elevators. Introduction The HayWired earthquake scenario examines a hypothetical moment magnitude (Mw) 7.0 earthquake (mainshock) occurring on April 18, 2018, at 4:18 p.m., on the Hayward Fault in the east bay part of the San Francisco Bay area. This chapter estimates demand for urban search and rescue (USAR) related to building collapse in the HayWired scenario. What is meant when one says a building has collapsed in an earthquake? When it collapses, what does the damage look like? The answer matters for at least two reasons. (1) Engineers would like to create second generation, performancebased earthquake engineering (PBEE-2) models of the effects of collapse on safety (see, for example, an early effort by Yeo and Cornell, 2002). So and Pomonis (2012) recently proposed a process for estimating fatalities in collapsed buildings during earthquake ground shaking using their engineering judgment of fatality rate by building type, informed by fatality data from various recent earthquakes. (2) Building collapse affects the demand for urban search and rescue. Elevators stalled without 100   The HayWired Earthquake Scenario—Engineering Implications power may also trap substantial numbers of people who must be rescued by USAR personnel (see, for example, Schiff, 2008). The present HayWired study seeks to (1) clarify what is meant by “collapse,” (2) advance demand for mathematical modeling of USAR, and (3) illustrate a new model by applying it to the HayWired mainshock. Objective This report describes the use of USAR modeling and addresses the following questions: 1. When engineers use the word “collapse” to describe the seismic performance of a building, what fraction of the occupiable floor area deforms severely enough to threaten lives in that area? (I offer an empirical answer by examining a database of photographs of building collapses. Such an empirical study complements numerous analytical assessments of collapse.) 2. What fraction of occupants in the collapsed areas require extrication, and by whom? (I answer this question by interpreting the image database in light of Federal Emergency Management Agency [FEMA] Urban Search and Rescue guidelines; for example, PerformTech, Inc., 2011.) 3. How many elevators are in the affected metropolitan area, how many of them are carrying passengers (and how many passengers) between floors at the time of the earthquake, and what fraction of those elevators have emergency power to bring the elevator to a floor and open the doors? To keep the level of effort commensurate with the value of the information, I consider only one extensive, although not exhaustive, data source—the Earthquake Engineering Online Archive provided by the National Information Service for Earthquake Engineering (NISEE), University of California, Berkeley. NISEE refers to the archive as the NISEE e-library (see http://nisee.berkeley.edu/elibrary/). NISEE describes the e-library as “a database of significant, publicly funded research and development literature, photographs, data and software in earthquake, structural, and geotechnical engineering.” I exclude manufactured housing, fences, equipment, and bridges from the objective. I also acknowledge that the NISEE e-library is not exhaustive. It is treated here as a sample, not as documentation of the population of collapsed buildings, with the expectation that it is a diverse and representative sample. One could conceivably address the building collapse questions with structural analysis, either instead of, or in addition to, the empirical approach of examining photographic evidence. But it seems doubtful that structural analysis would reliably reveal the extent of collapse, because structural analysis is not yet capable of reliably predicting the onset of collapse, its dynamics, and the eventual shape of a collapsed building. The authors of FEMA P–695 (Applied Technology Council, 2009), for example, identified collapse as the condition that lateral dynamic instability appeared during incremental dynamic analysis, meaning that collapse occurs when structural analysis of a mathematical representation of the building fails to converge. Failure of a mathematical model to converge following the loss of vertical load-carrying capacity provides little information about how much of or how far a floor or roof diaphragm falls. The authors of FEMA P-695 further cite examples of possible nonsimulated collapse modes, meaning collapse modes that a structural analysis might not reveal. These include “shear failure and subsequent axial failure in reinforced-concrete columns, fracture in the connections or hinge regions of steel moment frame components, or failure of tie-downs in light-frame wood shear walls. Component failures such as these may be difficult to simulate directly.” Another reason to favor an empirical study over an analytical study is that empirical models are more credible than analytical ones, at least among the loss-estimation community, where an empirical model is always preferred to an analytical one, all else being equal. Analytical models often serve to validate an empirical one or to provide insight where empirical data are lacking. None of this is to say that an analytical study will never have anything to offer to the question of area affected by collapse, but rather an empirical study seems more likely to provide defensible results in the near term for much less effort. Literature Review Literature About People Trapped by Building Collapse It is believed that building collapse dominates earthquake casualty risk and contributes substantially to USAR demands. The 2009 National Earthquake Hazards Reduction Program (NEHRP) provisions (Building Seismic Safety Council, 2009) assert that “Most earthquake injuries and deaths are caused by structural collapse.” Although that statement could be true in general, it probably exaggerates the importance of structural collapse for nonfatal injuries, at least in California earthquakes such as those of the late 1980s and early 1990s, judging by the analysis Shoaf and others (1998). The National Fire Protection Association (2014) offers descriptive patterns of earthquakeinduced building collapses in earthquakes and explains the causes and nature of voids where occupants can escape injury in collapsed buildings (fig. 1). The authors of the Hazus-MH technical manual (Federal Emergency Management Agency, 2012) offer estimates of the fraction of occupants in collapsed area who are killed. Their estimates draw on the judgment-based ATC-13 (Applied Technology Council, 1985), which they “revised based on comparison with a limited amount of historical data,” and validated against “several recent events, including the Northridge, Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   101 Loma Prieta and Nisqually earthquakes . . .” They estimate that 10 percent of occupants in collapsed areas of buildings are killed and 65 percent are injured to some degree. The two leading public models of earthquake risk, Hazus-MH (Federal Emergency Management Agency, 2012) and ATC-13 (Applied Technology Council, 1985), do not address search and rescue demands. Collapse fragility functions, which estimate the probability that a building will collapse under various levels of excitation, are available or can be derived (see, for example, Applied Technology Council [2009] or Federal Emergency Management Agency [2012]). However, I could find no prior work that quantifies the fraction of the building area that collapses when a building experiences some collapse. When buildings in California collapse, they rarely pancake. That is, they rarely collapse such that the floor or roof over every square foot of occupiable floor area drops because of the loss of vertical load carrying capacity of the portion of the gravity system that supports it. One could conceivably use structural analysis to model the collapse behavior of sample Offset collapse pattern—light-frame construction buildings, but the state of the practice enables structural engineers only to estimate the excitation associated with the onset of collapse, as the authors of FEMA P–695 (Applied Technology Council, 2009) did quite extensively. Another approach, explored here, is to review postearthquake observations of building collapse. The present work focuses on California buildings. The International Building Code (International Code Council, 2009) does not use the word “collapse” at all. The authors of ASCE/SEI 7-10 (American Society of Civil Engineers, 2010) use the word “collapse” in defining the probabilistic (MCER) ground motion and in describing the anticipated maximum probability of failure for earthquake loading. It does not define collapse, but it does define progressive collapse as “the spread of an initial local failure from element to element, resulting eventually in the collapse of an entire structure or a disproportionately large part of it.” It also defines the term “limited local collapse” with an example: “the containment of damage to adjacent bays and stories following the destruction of one or two neighboring columns in a multibay structure.” Overturn collapse pattern—heavy-floor or heavy-steel construction Wall-fall collapse pattern—heavy wall— unreinforced-masonry construction Soft first-story collapse pattern—heavy-floor construction Wall-fall collapse pattern—heavy wall—tilt-up construction Wall-fall collapse pattern—heavy wall—tilt-up construction Random-fall collapse pattern—precast-concrete construction Figure 1.  Illustrations of building collapse patterns in earthquakes (modified from National Fire Protection Association, 2014). 102   The HayWired Earthquake Scenario—Engineering Implications The 2009 NEHRP provisions (Building Seismic Safety Council, 2009) mention structural collapse, collapse of small structural systems (such as a hospital canopy), and collapse of nonstructural components (such as light fixtures, ductwork, and piping systems), but they do not define the word. FEMA P–695 defines collapse as “including both partial and global instability of the seismic-force-resisting system,” excluding “local failure of components not governed by global seismic performance factors, such as localized out-of-plane failure of wall anchorage and potential life-threatening failure of nonstructural systems” (Applied Technology Council, 2009). It does not include in its consideration of collapse damage to, or failure of, “components that are not designated as part of the seismic-force-resisting system” because those components “are not controlled by seismic-force-resisting system design requirements,” and they are therefore not within the scope of the project. The authors of FEMA P–695 include among the possible definitions of collapse the occurrence of a sidesway mechanism, and more generally the “state of lateral dynamic instability.” In more recent work, I and colleagues developing the third edition of FEMA P–154 and FEMA P–155 (Applied Technology Council, 2015a,b) and proposed the following definition. We generally define building collapse as the condition in which “any part of the gravity system experiences dynamic instability leading to the loss of load-bearing capacity. The dynamic instability leads to severe structural deformation of a potentially life-threatening nature, especially falling of all, or portions of, a structure. Partial building collapse means that the dynamic instability occurs only in a portion of the building . . . In the case of manufactured housing and wood frame buildings, building collapse also includes the condition that the manufactured home falls off one or more of its supports, or the cripple walls of a wood frame building experience a sidesway mechanism and lose their vertical load-carrying capacity” (Applied Technology Council, 2015b). Building collapse does not include wood frame buildings sliding relative to their foundations if there is no vertical drop in any part of the floor or roof. Nor is the falling of a parapet from an unreinforced masonry (URM) building or brick veneer or chimney from any FEMA building type considered to constitute building collapse. The United States Federal Emergency Management Agency National Urban Search and Rescue Response System (2009) estimates that, of people injured in buildings in earthquakes, 50 percent are injured but not trapped, and can be aided by emergent, untrained volunteers—civilians—who happen to be nearby at the time of the earthquake (fig. 2). Another 30 percent are injured and trapped, but not by structural components, for example, by overturning of furniture, and are extracted by trained local community emergency response teams (CERTs). CERTs are trained to perform search and rescue in buildings that have damage to decorative work and to interior contents but are not collapsed or fallen from their foundations; that would presumably include chimney and parapet damage (PerformTech, Inc., 2011). An additional 15 percent of people injured are rescued from the collapse of light structures, such as wood frame construction and manufactured housing, by emergency services rescue forces—generally firefighters—without the need for heavy excavation equipment. The remaining 5 percent must be extricated by trained urban search and rescue forces aided by equipment to penetrate heavy structures—masonry, concrete, and structural steel. There do not appear to be any published statistics on the frequency of each collapse pattern or what fraction of occupants require extrication by search and rescue personnel, although there is limited anecdotal evidence about individual buildings, such as Krimgold’s (1988) statistics from the 12-story Juarez Hospital that collapsed in the 1985 Mexico City earthquake. 5 to 10 percent Urban search and rescue task force Lightly trapped 15 percent Light search and rescue Nonstructure entrapment 30 percent Community response teams Injured and not trapped 50 percent Spontaneous response by civilians nearby Trapped Figure 2.  Pyramid charts showing the distribution of assistance in a large earthquake (after National Urban Search and Rescue Response System, 2009). Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   103 Literature About People Trapped in Elevators The vast majority of San Francisco Bay area buildings do not have uninterruptible power supplies or emergency generators to power elevators in the absence of commercial power. According to National Elevator Industry, Inc. (2014), there are 900,000 elevator units in the United States, or approximately one elevator per 344 people. Each elevator makes an average rise of 4 to 5 floors, or 40 feet, and each carries an average of 5 people per trip. Each passenger averages 4 trips per day, 250 days per year. According to the Emporis Corporation (2007) database of high-rise buildings, there are approximately 600 high-rise buildings with approximately 3,700 elevators in the San Francisco Bay area. Sample calculations in Strakosch and Caporale (2010) suggest that an elevator is in motion with the doors closed approximately 30 percent of the time that it is in use with passengers inside. Some elevators have battery power to operate briefly to move the cab to a floor and open doors. According to San Francisco Bay area elevator consultant von Klan (written commun., 2015), elevators installed in highrise buildings in the past 40 years or so have been required to have emergency power for elevators, and he estimates that perhaps 60 percent of high-rise buildings in the bay area were constructed since this time. He also estimates that less than 5 percent of elevators in mid- and low-rise buildings have emergency power. Even if there were emergency power available, seismic safety devices installed in newer elevators may stop the elevator between floors until a technician inspects the elevator. Methodology Methodology for Estimating the Number of People Trapped by Collapse The illustrations in figure 1 do not represent an exhaustive typology of collapses that could trap occupants or passersby. If a portion of a parapet falls, it does not constitute building collapse, but engineers do speak of parapets collapsing, and parapet collapse does not appear in figure 1. I therefore include in collapse (1) the falling of a floor or roof such that the clear height is reduced to less than 2 meters (m), and (2) the falling of parapets, chimneys, and other elements, but I exclude the falling of other contents and movable furnishings, such as cubicles. For purposes of estimating the probability of being injured or trapped by collapse, I define collapse as follows: Collapse constitutes the condition where, in a portion of the building or in the entire building, the gravity load-carrying system (for example, its beams, columns, floors, and shear walls) loses the ability to carry its own weight and the weight of whatever else it supports. That failure leads to severe building deformation of a potentially life-threatening nature, especially if all or portions of a building fall. The nonstructural portions of a building are included in our definition of collapse, along with the structural portions, such as parapets, chimneys, and porches. Thus, some nonstructural collapses are included (parapets, chimneys, and porches), but some structural failures are not (permanent lateral displacement of the building relative to the foundation where no vertical drop occurs). I estimate fatality rate and USAR needs in future earthquakes as follows. I estimate fatality rate as the product of the collapse probability conditioned on ground motion, the fraction of the building floor area that actually collapses when there is at least some collapse, and the fraction of occupants in that collapsed area that are killed, as in equation 1: F ( h ) = P ( h ) × A × R (1) In the equation, F(h) represents the fatality rate in a building (fraction of occupants killed) that is shaken with severity h. P(h) denotes collapse probability given shaking h. A denotes affected area, that is, the fraction of the building area that collapses, given that at least some collapse occurs. R denotes the fatality rate in the collapsed area. I model search and rescue needs by an analogous equation where S(h) and E denote, respectively, the fraction of building occupants requiring extrication, and the fraction of occupants in the collapsed area who need extrication, as in equation 2: S ( h ) = P ( h ) × A × E (2) Implicit in equation 2 is the assumption that people are uniformly distributed throughout the building: an occupant is as likely to be in one place as another. This assumption might be conservative; for example, buildings with soft-story conditions are likely to collapse onto the soft story, which tends to be less densely occupied garage space rather than more densely occupied living space. To account for that fact requires a model of the number of buildings that collapse onto soft garage levels. I assume for the remainder of this work that one lacks a damage model that detailed. If one already has an estimate of the number of collapsed buildings (I denote this number by Nb), then the estimated number of people, Nc, who are trapped in collapsed buildings and require extrication by USAR personnel can be estimated as: N c (t ) = N b × O (t ) × A × E , (3) where O(t) denotes the average number of occupants per building at time t, and A and E again denote the fraction of the building area that collapses and the average fraction of occupants in the collapsed area who need extrication by USAR personnel, respectively. One might condition each term in equations 1, 2, and 3 on building type, era of construction, or other parameters. The analyst must estimate the quantity O(t), for example, using estimates of average building area per occupant from Hazus-MH 104   The HayWired Earthquake Scenario—Engineering Implications (Federal Emergency Management Agency, 2012) or ATC-13 (Applied Technology Council, 1985). To estimate A, I examined every photograph of a building in the NISEE e-library images database from every California earthquake in the past 50 years in which the photo description uses the word “collapse,” “fail,” “fell,” or “parapet.” I supplemented these images with photos of buildings where I knew collapse had occurred. I also added data on tilt-up roof collapses in the 1971 San Fernando earthquake extracted from a 1973 National Oceanic and Atmospheric Administration (NOAA) report that showed building plan area and area of roof collapse. I labeled each building with a building category: wood (consisting of FEMA model building types W1, W1A, and W2), steel (types S1 through S5), concrete (C1 through C3), precast concrete (PC1 and PC2), reinforced masonry (RM1 and RM2), and unreinforced masonry (URM). To make the assignments I used the procedures recommended in FEMA P-154 and FEMA P-155 (Applied Technology Council, 2015a,b). Finally, I aggregated these to four simpler categories: tiltup concrete (PC1), all concrete and precast concrete except tiltup, wood, and unreinforced masonry. No other building types appeared in the photo database. Of course, making these assignments based mostly on photographic evidence can be problematic, because architectural finishes often conceal the true structural system. I do not claim complete accuracy in making the assignments. Still, it is hard to mistake wood, URM, and tiltup construction, especially when one is familiar with the historic architectural styles of the region and given that collapse tends to peel away architectural finishes. I estimated E, the fraction of occupants in the collapsed area requiring extrication, as the fraction of the collapsed area in which heavy debris or structural elements fell to the floor or ground. For example, in the case of bricks littering a sidewalk from collapsed parapets or chimneys, it seems reasonable to assume that anyone in that debris field would be injured or killed and would require extrication by others. In the case of collapsed porch roofs resting entirely on the ground or porch, anyone beneath the porch would require extrication. In the case of houses off their foundations, but where the roof or upper floors do not fall, I assume that residents can generally escape through a window or a door that is not blocked. It seems realistic that there will be cases of injured or physically disabled people who cannot escape through a window unaided, but I assigned E=0 based on the assumption of the more likely case, that the occupant is not physically disabled or seriously injured. Social scientists speak of such an approach to sampling as a convenience sample, a nonprobability sampling technique where subjects are selected because of their convenient accessibility and proximity to the researcher. The main problem with convenience sampling is the potential for sampling bias, in which one does not know that the sample is representative of the entire population. If a database existed of all collapsed buildings in a particular earthquake or particular geographic region, one could perform a randomized sample or an exhaustive survey and avoid worries about sampling bias, but such a database does not exist, so for present purposes I fall back on this convenience sample and advocate for a better database in the future. In the present convenience sample, the first California earthquake in the 50-year period studied here is the Mw 6.5 1968 Borrego Mountain earthquake; the last is the Mw 6.0 2014 South Napa earthquake. In each case, I estimated the fraction of the building area affected by the collapse. In many cases, particularly ones where only a small portion of a large building was affected, the photograph shows the affected area but not the overall size of the building, and the building no longer exists. In many cases, I found additional evidence of the building location and other photographs that show more of the building, and in several cases, I estimated building area from the area of building shown in Google Earth Pro, which includes parcel outlines and recent and historical satellite imagery, and has a tool for measuring area. The results of my analysis are summarized in table 1. The columns list the earthquake associated with the collapse, NISEE’s image identifier number, NISEE’s photo description, the building type (using FEMA’s building typology), the estimated fraction of the building’s occupiable floor area that was affected by the collapse (A), the fraction of occupants in the affected area that would require extrication by others (E), and the technical qualifications of the people most likely to perform the extrication (T). The quantities A and E are bounded by 0 and 1. Options for T are labeled by the order in which USAR personnel would arrive: 1=emergent civilian volunteers (neighbors); 2=CERT; 3=firefighters; and 4=FEMA USAR Task Force. Details of each estimate of A are provided in the appendixes. I binned the fraction of affected area on a quarter orderof-magnitude basis, that is, approximately 10-2, 10-1.75, 10-1.5, ... 100, which is to say 1 percent, 2 percent, 3 percent, 6 percent, 10 percent, 18 percent, 32 percent, 56 percent, and 100 percent. From these data, one can create histograms of the data as a whole and subdivide by the structural material (wood, unreinforced masonry, or concrete). I estimated T, the technical qualifications of the USAR personnel, as 1 (untrained emergent civilian volunteer) if the extrication could be done by a single person without tools, as in picking up bricks. I assigned T=2 (CERT) if the extrication requires two or more people but no heavy equipment and would not violate the CERT training guidelines (PerformTech, Inc., 2011). I assigned T=3 if the extrication requires equipment but not heavy lifting or cutting of reinforced concrete, such as in the case of a collapsed wood frame building where a roof or an upper floor falls onto the floor or furnishings below. For example, firefighters extracted Sherra Cox from a collapsed building in the San Francisco Marina District after the 1989 Loma Prieta earthquake (Scawthorn and others, 1992). I assigned T=4 if the extrication requires heavy lifting or cutting of reinforced concrete. I made no assignment (T=blank) if E=0, that is, no extrication is required. The database of photos of collapse that I compiled from NISEE and the other sources contains 73 California buildings that experienced at least some collapse in earthquakes between Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   105 Table 1. Summary of parameters used in the urban search and rescue (USAR) model for the HayWired earthquake scenario. [ID, image identifier from Earthquake Engineering Online Archive; type, model building type according to the Applied Technology Council (2015a); A, affected area; E, fraction of occupants trapped; T, technical qualifications of USAR personnel; Mw, moment magnitude; %, percent; in., inch; ft, feet; St., Street, Rd., Road; Ave., Avenue] Earthquake ID Damage description Type A E T Santa Rosa 1969 S3715 (Mw 5.6 and 5.7) Two-story wood frame building off foundations. Foundations were rotted and poorly braced. Gas lines ruptured when house fell. 718 Beaver St., Santa Rosa, California. W1 0% 0 S3726 Miramar Building. Collapsed portion of a wall fell on a car. 203 Old Courthouse Square, Santa Rosa, California. URM 1% 1.0 1 Damage to porches (probable cripple wall failure?); chimney fell away from house. In the vicinity of Knox and Orange Grove Streets, in the fault zone. W1 8% 0.5 3 S4533 Chimney fell towards otherwise undamaged wood frame house. W1 0% 0 S4581 Furniture store. Unreinforced masonry parapet collapsed, dumping bricks into the street and on to the sidewalk. Large plate-glass windows are gone, presumably shattered by the earthquake. URM 19% 1.0 1 S4597-S4602 Apartments over retail space. Note that the failure of the unreinforced bearing walls did not result in collapse. Unit masonry construction, built prior to 1933. Downtown San Fernando commercial area. URM 3% 1.0 1 S4489 Partial collapse on older wood frame house, probable cripple wall failure of house. Between Glen Oaks and Hubbard Streets. W1 0% 0 S4491, S4492 Pink structure at the rear was a residence over a garage. The first story collapsed; note remains of automobile under the building. W1 50% 1.0 3 S4624 Roof to the wall failed first. Ground cracks in the vicinity. Rear wall bulged out, and rear roof fell. See S4625-4633. Light industrial buildings. Bradley Tract. 12884 Bradley Ave. TU 11% 0.1 3 Benfe and Coffman (1973, p. 123) 12840 Bradley Ave. TU 44% 0.1 3 12874 Bradley Ave. TU 12% 0.1 3 12950 Bradley Ave. TU 10% 0.1 3 12881 Bradley Ave. TU 10% 0.1 3 12975 Bradley Ave. TU 23% 0.1 3 13001 Bradley Ave. TU 8% 0.1 3 13069 Bradley Ave. TU 16% 0.1 3 15200 Bledsoe St. TU 19% 0.1 3 15151 Bledsoe St. TU 8% 0.1 3 12860 San Fernando Rd. TU 16% 0.1 3 12806 San Fernando Rd. TU 18% 0.1 3 12744 San Fernando Rd. TU 26% 0.1 3 San Fernando S4473 1971 (Mw 6.7) 12814 Bradley Ave. TU 15% 0.1 3 GoddenJ53 Collapse of a split-level wooden home. Large numbers of these split-level homes suffered significant damage because of a lack of adequate ties between the two levels. The upper level ripped away and crushed the lower garage walls, which did not have adequate lateral bracing. W1 33% 1.0 3 S4195 Collapsed Semi-Ambulant Building at Veterans Authority Hospital, built in 1925, masonry construction. URM 50% 1.0 3 S4529 Damage to older house caused by cripple wall collapse. W1 0% 0 S4065 Collapsed tower at southeast corner. Olive View Hospital. Rear [east] elevation of Medical Treatment Building. C2 3.3% 1.0 3 106   The HayWired Earthquake Scenario—Engineering Implications Earthquake ID Damage description S4070 Ambulance garage collapsed. Olive View Hospital. Southern elevation of Medical Treatment Building. See also S4139-44. S4115, S4117 Type C1 A E T 100% 0.5 3 Soft-story collapse, most evident at upper right of photo. Original- C1 ly a one- and two-story building, irregular in plan, the first story collapsed in the earthquake. 67% 1.0 4 S4519 Collapsed wood frame house under construction on Tucker Street near Pacoima Dam. W1 67% 0.5 3 S4501 Two-story section over garage of this wood frame house on Almetz Street has collapsed in the first story. In a new housing tract in Sylmar at base of hills and between Olive View and Veterans Administration Hospitals. W1 33% 1.0 3 R0070 Old masonry building in upper center of photo has completely collapsed. Constructed in 1925–26, with major additions in 1938 and 1949, the entire complex was demolished after the 1971 earthquake and the entire 97 acres were dedicated in 1977 as Veterans Memorial Park. URM 100% 1.0 4 Imperial Valley S5584 1979 (Mw 6.4) Cripple wall collapse—wood frame house on G Street in Brawley, W1 California. 0% 0 S5585 Cripple wall collapse—wood frame house on G Street in Brawley, W1 California. 0% 0 Westmorland N/A 1981 (Mw 6.0) Collapsed two-story building on West Main Street in Westmorland, California URM 100% 1.0 3 Coalinga 1983 (Mw 6.2) GoddenJ52 Chimney collapse of a modern house, 1983 Coalinga earthquake. Most of the chimneys were thrown down because of the lack of proper connections (straps) to the buildings. W1 9% 1.0 1 GoddenJ19 This two-story wood frame dwelling underwent a lateral displacement of more than half a meter as illustrated by the slant in the porch columns and also fell more than half a meter from its foundation, owing to lack of adequate anchorage and support W1 0% 0 GoddenJ23 Collapse of a wooden porch (owing to lack of proper anchorage to the wooden frame of the house and of a proper later-resistant supporting system) was owing to vibratory response. W1 15% 1.0 3 GoddenJ29 The second-story, 8-in. unreinforced solid brick masonry walls of this commercial building collapsed because of inadequate tying at the floor, roof, and transverse walls. URM 30% 0.60 1 R0323 Porch running the full width of the church simply pulled away URM from the rest of the building. Built in 1946, the stabilized adobe building was heavily damaged but did not collapse. On the corner of Jefferson St. 7% 1.0 3 Most severely damaged dwelling. Sheathing between first floor W1 and foundation was fiberboard with little strength. Morgan Hill, California Anderson Lake area. 0% 0 S5839 Dwelling on the left moved, owing to landsliding from the earthquake. Morgan Hill, California, Anderson Lake area. W1 20% 1.0 S6014 Damage to roof from chimney collapsing. Whittier, California. W1 0% 0 S6023 Chimney collapsed away from the house. Whittier, California. W1 3% 1.0 1 S6020 Chimney fell through porch roof. See S6021 and s6040. Whittier, California. W1 2% 1.0 2 S6022 One chimney collapsed, but not the other. Whittier, California. W1 3% 1.0 1 Wall collapse in unreinforced masonry building. Santa Cruz, California. URM 1% 1.0 1 Morgan Hill S5840 1984 (Mw 6.2) Whittier Narrows 1987 (Mw 5.9) Loma Prieta LP0042 1989 (Mw 6.9) 3 Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   107 Earthquake ID Damage description Type A E T LP0070 Older building with failed parapets on Main Street. 307 Main Street, Watsonville, California. URM 18% 1.0 1 LP0072 Older building with failed parapets on Main Street. 311 Main Street, Watsonville, California. URM 9.4% 1.0 1 LP0462, LP0460 Collapse of unreinforced brick wall. 6th and Bluxome Streets, South of Market District, San Francisco, California. URM 5.3% 1.0 1 LP0375 Collapse of two four-story apartment buildings (soft ground floors). Marina District, San Francisco, California. W1A 25% 1.0 3 LP0375, S6120 Collapse of two four-story apartment buildings (soft ground floors); there were two buildings in the image. W1A 25% 1.0 3 LP0499 Collapsed apartment building at 2090 Beach Street, after the fire was much advanced. Note the firefighter directing water onto exposed side of building. Marina District, San Francisco, California. W1A 75% 1.0 3 S6144 Soft-story collapse of apartment building in the Marina District, San Francisco, California. W1A 33% 1.0 3 LP0459 Collapse of unreinforced masonry wall from third floor of building. 235 Front St. at Davis St., Embarcadero/Financial District, San Francisco. URM 2.9% 1.0 1 LP0041 Interior structural failures at Ford’s Department Store. Santa Cruz, URM California. 33% 1.0 3 LP0081-LP0085 Front view of damaged St. Patrick’s church. Watsonville, California. URM 4.5% 1.0 1 LP0087 Damaged bike store with failed parapet. Watsonville, California. URM 25% 1.0 1 LP0090 Pink frame house with failed foundation. Watsonville, California. W1 0% 0 Collapsed apartment building, three-story wood frame. Northridge, California. According to Todd and others (1994, p. 23), four buildings experienced collapse. This is the first. W1A 33% 1.0 3 Collapsed apartment building, three-story wood frame. Northridge, California. Second building. W1A 33% 1.0 3 Collapsed apartment building, three-story wood frame. Northridge, California. Third building. W1A 17% 1.0 3 Collapsed apartment building, three-story wood frame. Northridge, California. Fourth building. W1A 4% 1.0 3 NR408-409 1004 West Channel Road at Pacific Coast Highway (near Pacific Palisades). Damage to two-story masonry building. Heavy shear cracking on side walls. Out of plane failure of the second story. State Beach Cafe, Santa Monica, California. URM 13% 1.0 1 NR412-414 Four-story masonry building, 827 Fourth Street, Santa Monica, California. Damage to the fourth and third floor of the building. The masonry facade fell out of plane and took with it the fourth-floor terrace. This building had been scheduled for a retrofit to begin on Monday, January 17, 1994. Three-layers-thick unreinforced masonry. Damage in the top story and balcony. Little damage on the sides and below the third story. See also NR412–414. URM 2.1% 1.0 1 20101224 This residential chimney of unreinforced blocks collapsed. W1 2.7% 1.0 1 NR559 Parking structure on Zelzah Ave., California State University, C1 Northridge, campus. This is a three-story precast concrete parking structure. Overall view showing collapse at east end of the structure. 35% 1.0 4 Northridge 1994 NR327, NR353, (Mw 6.7) NR357, NR358 108   The HayWired Earthquake Scenario—Engineering Implications Earthquake ID Damage description Type A E T NR579 Collapse of parking garage floors. See NR459–461 for damage to Broadway department store. Fashion Center, Northridge, California. PC1 35% 1.0 4 NR221 Northridge Fashion Island Center. Interior reinforced-concrete columns remain standing following collapse of second- and third-floor concrete waffle slabs. Intact portion of waffle slab roof shows typical slab construction. C1 78% 1.0 4 NR303 View of partial roof collapse. South elevation, east of front entry. View from east. Taken at 3 p.m. California State University, Northridge. C1? C2? 1% 1.0 4 NR542, NR543 Complete collapse of parking structure. Los Angeles, California. C1 100% 1.0 4 NR328 Soft-story collapse of apartment building, at Hazeltine Ave. and Milbank St. Sherman Oaks, California. W1A 33% 1.0 3 NR160, NR162 Overall view of Kaiser Permanente office building looking toward C1 the northeast. The brick facades at either end of the structure have separated from the concrete frame, and the second floor of the structure has completely collapsed. The bays at the north and south ends of the building are also partly collapsed from the second to the fifth floor. Granada Hills, California. 30% 1.0 4 San Simeon NM0001-NM0012 House of Bread, was located in the Mastagni/Acorn Building, URM 2003 (Mw 6.7) which collapsed. By the time these pictures were taken, emergency personnel had removed the front wall of the building and a great deal of debris. Built in 1892, the clock tower of this unreinforced masonry building had become a symbol of the town of Paso Robles. The second story of the building collapsed during the earthquake, killing two employees of Ann’s Dress Shop. The roof of the building collapsed directly westward onto Park Street and landed on a row of parked cars. Debris from the north wall went through the roof of an adjacent shop at 1220 Park Street. Paso Robles, California. 78% 1.0 3 South Napa P9050177, 2012 (Mw 6.6) P9080152 8.3% 1.0 1 Don Perico’s Restaurant in Napa. At the time of the earthquake, W2 the restaurant was located at 1025 1st St., Napa, California, in the west end of the building at lat 38.299029 N., long 122.285868 W. That address seems to occupy approximately 60 ft×60 ft. The collapsed wall appears to fill 25 ft by 12 ft, suggesting a collapsed portion of 8.3%. 1965 and 2014. The database contains wood, concrete, and unreinforced masonry buildings. Areas affected range from zero (for example, cripple wall collapse that did not cause height reduction of an occupiable area) to 100 percent (for example, complete collapse of a parking structure). Among the sample of collapsed California buildings of the last 50 years, the average had 23 percent of its occupiable floor affected area. Therefore, on average 23 percent of occupants or passersby—people walking within a few feet of the building—could have been trapped or injured by a portion of building falling on them. On average, I estimate that 66 percent of occupants in the collapsed area would need extrication by USAR personnel, even if only by emergent civilian volunteers. Statistics by structural material are shown in table 2. In California, the 1934 Field Act outlawed the use of unreinforced masonry in most buildings. Consequently, URM buildings have become rarer in California than elsewhere in the Western United States, and many have been retrofitted, thus including the data of their past performance could conceivably bias estimates of future performance. Nonetheless, removing unreinforced masonry buildings and chimneys from the data does not substantially change the average affected area. The weighted average considering only tilt-up, other reinforced concrete, and wood is 22 percent. If one removes the cases where the collapse was limited to or caused by chimney collapse (that is, also removing the case where a chimney penetrated a roof), the average increases to 25 percent. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   109 Table 2.  Average affected area (A) and average fraction of occupants in collapsed areas requiring extrication (E) in the urban search and rescue (USAR) model for the HayWired earthquake scenario. there is a constant probability f that the affected area is zero. If the affected area is nonzero, it is exponentially distributed: [%, percent] where f and L are constants. The affected area data and equation 4 are plotted together in figure 3 for comparison. I refer to equation 4 as a frequency-and-exponential-severity model. Of all the forms examined, only the one shown in equation 4 passed the Lilliefors (1967) goodness-of-fit test at the 5 percent significance level. The Lilliefors test is intended to check whether a sample is drawn from a normally distributed population with parameters of the distribution estimated from the sample. The test is not intended for this frequency-andexponential-severity model. I am aware of no comparable test for this frequency-and-exponential-severity model, so let the passing of the Lilliefors test merely indicate reasonableness in a qualitative manner. A parametric expression similar to a power law is also shown in figure 3. It does not fit quite as well as the frequency-and-exponential distribution, but it is simpler. Given a building that is modeled as having collapsed, one could simulate affected area by inverting equation 4 at the value of a sample of a random variable uniformly distributed between 0 and 1; that is, if I draw a sample u~U(0,1), then the sample of affected area is the following: Material Count Average A Average E All 73 23% 0.66 Tilt-up concrete 14 17% 0.10 Other concrete 9 50% 0.94 Unreinforced masonry 18 28% 0.98 Wood 32 17% 0.66 All except unreinforced masonry 54 22% 0.56 All except chimneys 66 25% 0.65 The estimated distribution of minimum USAR technical qualifications is shown in table 3. It suggests that most search and rescue would have to be done by firefighters, rather than by untrained emergent civilian volunteers. This estimate is not necessarily inconsistent with figure 2, whose bottom two strata are people who are not trapped by collapse and are not represented in the collapse photos examined here. Considering all buildings, the distribution of affected area resembles an exponential distribution (though it does not pass a Lilliefors, 1967, goodness-of-fit test at the 5-percent significance level). An exponential distribution would mean that a building is approximately equally likely to collapse on 1 percent (10-2) of its occupiable area, 2 percent (10-1.75), 3 percent (10-1.5), and so on, through 100 percent (100). Among the wood buildings, the affected area tends to be lower; among the nine concrete buildings, the affected area tends to be higher, but nearly the full range is exhibited among each building type (reinforced concrete, tilt-up unreinforced masonry, and wood), as discussed later. Suppose one wanted to perform a Monte Carlo simulation of USAR needs using a simple parametric model, for example, a mathematic idealization of the data presented here. To inform such simulations, I evaluated a few common parametric cumulative distribution functions for affected area: uniform, exponential, lognormal, power-law, and the distribution shown in equation 4. The equation reflects a model in which P [ X ≤ x ] = 1− f × exp (−Lx ) ; X ≥ 0 , x=0 u≥ f = u< f −1 (1− u ) ln f L (4) . (5) The mean number of people trapped in the collapsed area can be estimated as n in equation 6 where the symbols ⎣⎦ mean “floor,” that is, the largest integer less than or equal to the value inside: n = ⎢⎣ x × N × E ⎥⎦ , (6) where N denotes the number of occupants in the building and E=0.66. Alternatively, to account for building type and to treat uncertainty at least to a limited degree, use the cumulative distribution function of area affected from figure 3B. Invert the expression for P[X≤x] shown in the figure at a random sample of U(0,1) to simulate the affected area x. Then calculate n according to equation 6 using the value of E from table 2, and invert the Table 3.  Distribution of minimum technical qualifications for urban search and rescue (USAR) personnel, in percent. [CERT, community emergency response team; URM, unreinforced masonry] Technical qualifications 1 Civilian 2 CERT All URM Not URM 27 67 11 Tilt-up 0 Other concrete Wood Chimney 0 23 80 Not chimney 22 2 0 2 0 0 5 20 0 3 Firefighter 59 28 71 100 22 73 0 64 4 USAR Task Force 13 6 16 0 78 0 0 14 110   The HayWired Earthquake Scenario—Engineering Implications • Figure 5 shows that collapse of buildings with bearing walls composed of wood or unreinforced masonry generally affected the smallest total area in these buildings, followed by tilt-up concrete, then other reinforced concrete. • Most collapses involving wood frame buildings affect less than 10 percent of the building area, that is, the median affected area is less than 10 percent. Furthermore, 95 percent of collapses affect less than half the building area. More than 30 percent do not collapse into occupied space at all. As shown in figure 5, the modal affected area (the tallest bar on the ¼-logincrement bar charts) for wood frame buildings was between 0 and 1 percent. A common example of a building with such an affected area is one in which the unbraced cripple wall collapsed, without the loss of load-bearing capacity supporting a ceiling or roof above an occupied space (fig. 6A). The median affected 1.00 Cumulative probability A 0.75 P[X≤x] = 1 – f × e -Lx f = 0.922 L = 4.52 0.50 0.25 0.00 0.25 0.50 0.75 Affected area 1.00 1.00 B Cumulative probability binomial cumulative distribution function with parameters n and p, where p is another sample of U(0,1). One could go farther, treating N as random and using a separate cumulative distribution function for affected area that varies by building type, but such a treatment is omitted here for brevity. If one wanted to use the data presented here for modeling future performance of buildings, one must assume that the past is indicative of the future. Is it? There does not appear to be a correlation between affected area and earthquakes occurring in later years, as shown in figure 4. The trend line has almost no slope, and the coefficient of determination (R2) is so low (0.0006) that one can be fairly confident that no trend actually exists. Because each earthquake affects an existing building stock that was built up over decades, the relation would be a lagging indicator, meaning a measurable factor that changes only after the process it measures has begun to follow a particular pattern or trend. But because approximately half the building stock was replaced over the five decades examined here, if newer buildings tended to experience lower collapse areas, one would expect to see a stronger downward trend. The implication is that, while collapse probability of an arbitrary building in the building stock may or may not change over time, if a building does collapse, its collapse area is not related to the year of collapse. To be clear, figure 4 does not say anything about the collapse probability of older versus newer buildings. It says only that, in that subset of buildings where at least some collapse occurs, the affected area does not vary with the year in which the earthquake occurred. One can reasonably assume that buildings in near-future earthquakes (the next several decades) will have approximately the same distribution of affected area as in the previous five decades. Note that the catalog does not indicate the age of the building that collapsed. Newer buildings presumably have a lower collapse probability than older buildings, all else being held equal, but that issue is separate from the one examined here. A few additional observations of the nature and extent of collapse. 0.75 P[X≤x] = x0.32 0.50 0.25 0.00 0.25 0.50 0.75 Affected area 1.00 Figure 3.  Graphs showing approximate parametric forms of the cumulative distribution function for affected area of all building types—A, frequency-and-exponential-severity; B, a simpler expression similar to a power law. The axis of affected area spans from 0.00 (no area affected) to 1.00 (100 percent of area affected). area (the value with 50 percent probability of being exceeded) was between 6 percent and 10 percent of building area, commonly the collapse of a chimney or porch roof (for example, fig. 6B). The distribution of affected area in wood frame collapses is likely biased high. The reason for this is that the collapse of brick chimneys was likely too widespread and too uninteresting for NISEE e-library contributors to photograph instances in proportion to their actual occurrence within the population of wood frame buildings with collapse. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   111 Methodology for Estimating the Number of People Trapped in Elevators Affected area, in percent 100 75 50 y = 0.06x −102 R² = 0.0006 25 0 1960 1970 1980 1990 2000 2010 2020 Earthquake year Figure 4.  Graph showing affected area of all types of buildings versus year of earthquake, 1965–2014. • Although the database includes instances of complete collapses of URM buildings, most URM collapses affect less than 18 percent of floor area. The modal affected area is between 18 and 32 percent of the building area, such as the collapse of brick parapets on the sidewalk, parking areas, and lower buildings adjacent to the URM building. That is, the URM collapses sampled here are commonly more dangerous to neighbors and passersby than to occupants. See figure 7 for representative examples. • In the case of pre-1971 tilt-up construction examined here, most collapses affected less than 18 percent of the building area. The modal affected area was between 10 and 18 percent of the building area, almost always just inside the building perimeter where roofto-wall connection fractures occurred. The interior gravity system kept supporting interior subdiaphragms (away from the edge) even after perimeter subdiaphragms collapsed. See figure 8 for an example. • Complete collapses of concrete buildings in California have occurred, but they are the exception rather than the rule. In most cases, less than 50 percent of the floor area is affected. The modal affected area on this ¼-log-increment scale was between 32 and 56 percent of building area. An example of such a modal collapse was that of a partial collapse of a parking structure, shown in figure 9. No obvious spatial pattern of collapse was observed in these images. It is reasonable to assume that electric power will go out across the bay area as soon as substation equipment and perhaps buildings in the area near the earthquake’s epicenter are damaged. Hence, the vast majority of elevators in the bay area will lose power before P-waves trigger seismic switches or ring-on-a-string devices. How many people will be in elevators with doors closed and traveling between floors when power goes out? I take the number of elevators in a metropolitan area Vm as: P Vm = m , p (7) where Pm is the population of the metropolitan area, and p is the average number of people per elevator, which as noted earlier is approximately 344 in the United States. Let Vo of (t) denote the number of elevators in motion with people inside and no emergency power at time t, and I estimate it as shown in equation 8: Vo (t ) = Vm × fo (t ) × fc × (1− fb ) , (8) where fb denotes the fraction of elevators with emergency power, fo(t) is the estimated fraction of all elevators that are in use at time t, and fc is the fraction of the time that an elevator in use with passengers in it is traveling between floors with the doors closed, which as noted earlier is on the order of 30 percent of the time. If the average elevator with passengers has d passengers (as previously noted, d≈5), then the number of people that will be trapped in elevators Ne can be estimated as shown in equation 9: N e = Vo (t ) × d = Pm × fo (t ) × fc × (1− fb ) × d p . (9) Application to HayWired Scenario People Trapped in Collapsed Buildings, Based on the Building-Code Objectives I now turn to the question of urban search and rescue needs in the HayWired earthquake scenario. The HayWired scenario uses two approaches to estimate building damage—(1) based on building-code objectives (the Safe Enough approach documented in Porter [2015] and Porter [Not Safe Enough, this volume]) and (2) based on a combination of empirical observations, structural analysis, and engineering judgment, as encoded in the Hazus-MH model. FEMA performed a Hazus-MH analysis for the 112   The HayWired Earthquake Scenario—Engineering Implications A 0.50 Reinforced-concrete buildings Probability 0.40 0.30 0.20 0.10 0.00 0 1 2 3 6 10 18 32 56 100 Affected area, in percent 0.50 B Tilt-up buildings Probability 0.40 0.30 0.20 0.10 0.00 people = 7,800 buildings × 2.8 building × 0.25 × 0.66 . (10) 0 1 2 3 0.50 = 3,600 people 6 10 18 32 56 100 Affected area, in percent C mainshock (Doug Bausch, written commun., Federal Emergency Management Agency, 2014), and Seligson and others (this volume) performed the Hazus-MH analysis for selected aftershocks. Using the Safe Enough approach, I estimated the number of collapsed buildings in the HayWired mainshock to be Nb=7,800—if all buildings were to perform as well as modern (code-compliant) buildings—as estimated by a recent FEMA study (Applied Technology Council, 2009). California is home to 38 million people and approximately 11 million buildings, or approximately 3.5 people per building. If 80 percent of people were indoors at the time of the earthquake (which seems realistic at 4:18 p.m. on a workday and consistent with Hazus-MH on an overall average basis), then there would be an average of about O(t)=2.8 occupants in each collapsed building at 4:18 p.m. As previously observed, the overall average fraction of building area that experiences collapse can be taken as A≈0.25. The overall average fraction of occupants in the collapsed area requiring USAR extrication can be taken as E≈0.66. Thus, if all buildings in the bay area just met current code requirements, equation 3 can estimate the number of people trapped in collapsed buildings: N c (t ) = N b × O (t ) × A × E Unreinforced-masonry buildings That is, by the approach that uses building-code objectives, I estimate 3,600 people trapped in 7,800 collapsed buildings; however, many buildings with collapse would not have people trapped in them requiring USAR assistance. Probability 0.40 People Trapped in Collapsed Buildings, Based on Hazus-MH 0.30 0.20 0.10 0.00 0 1 2 3 6 10 18 32 56 100 Affected area, in percent D 0.50 Wood buildings Probability 0.40 N c (t ) = O (t ) × ∑ (( N b,i × Ai ) × Ei ) , i (11) where i is an index for the structural materials, Nb,i×Ai is taken as the product of Hazus-MH’s estimated number of buildings in the complete damage state and its estimate of the fraction of that building area that collapses, and Ei is the fraction of 0.30 0.20 0.10 0.00 Hazus-MH does not estimate the number of people trapped in collapsed buildings, but it does estimate the number of buildings in the complete structural damage state and the fraction of their area that experiences collapse, the product of which I can take as Nb×A. Applying the values of E, estimated here by structural material, and applying a uniform occupant load of 2.8 occupants per collapsed building, I can estimate: 0 1 2 3 6 10 18 32 56 100 Affected area, in percent Figure 5.  Graphs showing distribution of affected area by structural material: A, reinforced-concrete buildings except tilt-up; B, tilt-up concrete buildings; C, unreinforced-masonry buildings; and D, wood buildings. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   113 A B Figure 6.  Photographs of homes damaged in earthquakes. A, An example of the modal affected area (0 percent) of a collapsed wood frame building—in the moment-magnitude-6.4 1979 Imperial Valley, California, earthquake. B, An example of the median affected area (6–10 percent)—collapse of a porch roof in the moment-magnitude-6.6 1971 San Fernando Valley, California, earthquake. (Photographs by M. Hopper, and V. Bertero, respectively, courtesy of the National Information Service for Earthquake Engineering, PEER-NISEE, University of California, Berkeley.) 114   The HayWired Earthquake Scenario—Engineering Implications A B Figure 7.  Photographs of examples of modal (A) and median (B) affected areas in unreinforced masonry (URM) buildings. A, Brick building damage in the moment-magnitude 6.9 1989 Loma Prieta, California, earthquake; B, store front collapsed in the momentmagnitude-6.6 1971 San Fernando Valley, California, earthquake. (Photographs by J. Blacklock and E. Schader, respectively, courtesy of the National Information Service for Earthquake Engineering, PEER-NISEE, University of California, Berkeley). Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   115 Figure 8.  Photograph of an example of both modal and median collapse of tilt-up construction. damage was caused in the moment-magnitude-6.6 1971 San Fernando Valley, California, earthquake (Photograph by V. Bertero, courtesy of the National Information Service for Earthquake Engineering, PEER-NISEE, University of California, Berkeley.) Figure 9.  Photograph of an example of a partly collapsed reinforced-concrete structure, a parking garage at California State University, Northridge, damaged in the moment-magnitude-6.7 Northridge earthquake. (Photograph by P. Weigand, courtesy of the National Information Service for Earthquake Engineering, PEER-NISEE, University of California, Berkeley.) 116   The HayWired Earthquake Scenario—Engineering Implications occupants requiring extrication for structural material i, from table 2. See table 4 for results. How many buildings would Hazus-MH estimate had collapsed? Hazus-MH does not provide that estimate, but I can infer: Mc = ∑ i M compl,i × fcoll compl,i , Ai (12) where Mcompl,i denotes Hazus-MH’s estimate of the number of buildings of structural material i in the complete structural damage state (column 2 in table 4); fcoll compl,i denotes the fraction of area collapsed, given that it is in the complete damage state (column 3 in table 4); Ai is the fraction of building area that collapses (from table 2); and i is an index for structural material. See table 5 for results. Thus, one can infer from the combination of Hazus-MH’s damage estimates and the observations of collapsed buildings made here that the HayWired mainshock would trap approximately 1,100 people in 2,100 collapsed buildings. Scenario Estimate of People Trapped in Collapsed Buildings Using Hazus-MH damage estimates for the HayWired scenario mainshock, 1,100 people will be trapped in 2,100 collapsed buildings, whereas by the Safe Enough approach, 3,600 people will be trapped in 7,800 collapsed buildings. That the two approaches differ by a factor of 3 essentially means that they agree within a half order of magnitude, which in the present state of loss modeling represents reasonable agreement. However, the agreement is actually poorer than that, because the Safe Enough figures represent the expected behavior of post-1980 construction, and the Hazus-MH estimates are of the existing building stock, of which 60 to 70 percent predates 1980. One would expect the Safe Enough estimates to be less than those of Hazus-MH, if both were correct; they use the same inventory of buildings. However, I use their estimates as benchmarks, their range representing two approaches to a realistic answer, and their medians, 2,500 people trapped in 5,000 collapsed buildings (in round numbers), as realistic estimates for the HayWired scenario. Number of People Trapped in Stalled Elevators I turn now to the question of people trapped in elevators. In a large bay area earthquake, power would be lost immediately throughout the bay area and return slowly as power plants are inspected, load is carefully restored, and damage is repaired. When power is lost, most elevators in the bay area (those that do not have emergency power) would stop, even before P-waves reached the elevators and triggered their ringand-string safety devices. What would be the USAR impacts of that loss of power to elevators? How many people would be trapped in elevators with their doors closed, traveling between floors? Considering a San Francisco Bay area population of 10 million and using the previously observed average of one elevator per 344 people, one can use equation 7 to estimate the number of elevators in the San Francisco Bay area (Vm): Pm p 10,000,000 people = people 344 elevators Vm = . (13) = 29,000 elevators Subtracting 60 percent of the estimated 3,700 elevators in bay area high-rise buildings that have emergency power, and 2.5 percent of the remaining elevators and low- and mid-rise buildings with emergency power, an estimated 25,300 elevators in the bay area lack emergency power—I estimate 25,000 in round numbers. Recall the fraction of the time that an elevator that is in use with passengers in it is traveling between floors with the doors closed is fc≈0.3. I assume that at peak Table 4.  Number of people trapped in collapsed buildings, using Hazus-MH (Federal Emergency Management Agency, 2012) building damage estimates by Seligson and others (this volume) for the moment-magnitude-7 mainshock of the HayWired earthquake scenario. [E, fraction of occupants in collapsed portion of buildings requiring extrication; O(t), number of building occupants per collapsed building on a Thursday at 4:18 p.m.; Nc, number of people in collapsed buildings requiring extrication] Material Number of buildings in complete structural damage state Fraction of building area collapsed, given complete damage E O(t) Nc Wood 4,946 0.03 0.66 2.8 274 Steel 1,595 0.05 0.66 2.8 147 Concrete 1,241 0.10 0.94 2.8 327 Precast 71 0.15 0.10 2.8 3 725 0.10 0.66 2.8 134 Unreinforced masonry 639 0.15 0.98 2.8 263 Manufactured housing 4,340 0.03 0 2.8 0 Reinforced masonry Total 1,148 Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   117 Table 5.  Number of collapsed buildings, using Hazus-MH (Federal Emergency Management Agency, 2012) building damage estimates by Seligson and others (this volume) for the moment-magnitude-7 mainshock of the HayWired earthquake scenario. Number in complete structural damage state Fraction of area collapsed, given complete damage Fraction of area collapsed in collapsed buildings Collapsed buildings Wood 4,946 0.03 0.17 873 Steel 1,595 0.05 0.23 347 Concrete 1,241 0.10 0.50 248 71 0.15 0.17 63 725 0.10 0.28 259 Unreinforced masonry 639 0.15 0.28 342 Manufactured housing 4,340 0.03 0.00 Material Precast Reinforced masonry Total 2,132 hours (and 4:18 p.m. on a weekday is likely a peak hour), most elevators are in use and most are carrying passengers primarily in one direction, so I assume fo(t)≈0.6. Then by equation 8, the number of elevators stalled with people inside at the time of the HayWired mainshock can be estimated as: V0 (t ) = Vm × fo (t ) × fc × (1− fb ) . (14) = 25,000 × 0.6 × 0.3 = 4,500 elevators And as previously noted, the average elevator carries d=5 people when occupied, so one can use equation 9 to estimate Ne, the number of occupants trapped in elevators by the HayWired mainshock: N e = Vo (t ) × d = 4,500 elevators × 5 occupants elevator . (15) = 22,500 occupants So it seems realistic that on the order of 22,500 people could be trapped in 4,500 elevators by the sudden loss of electric power after the HayWired mainshock, requiring fire department assistance to escape. Untrained first responders will be unable to assist the people trapped in elevators because technical skills and equipment are required to extricate people from elevators. It is possible to retrofit some existing elevators with emergency power to reduce the demand for elevator rescue. Kornfield (San Francisco Department of Building Inspection, retired, written commun., 2015) estimates the cost of retrofitting elevators to be on the order of $20,000 per elevator, and only 30 to 40 percent of elevators in the bay area could be retrofitted, so retrofit could reduce elevator entrapment to 14,000 people in 3,000 stalled elevators. Conclusions USAR Demands Under Current Conditions There are currently no public models of urban search and rescue demands after earthquakes. Although engineers can estimate the number of buildings that collapse in an earthquake, we do not know what fraction of building area experiences collapse when at least some collapse occurs, nor do we know what fraction of occupants in those collapsed areas require extrication by urban search and rescue personnel. To estimate the search-and-rescue demands after the HayWired earthquake scenario mainshock, I compiled a photographic database of 72 buildings known to have experienced at least some collapse (structural or nonstructural) in 10 California earthquakes in the last 50 years. These include all buildings with images in the NISEE e-library whose description includes the word “collapse,” “fail,” “fell,” or “parapet,” plus 12 tilt-up buildings with roof collapse documented in a NOAA report on the 1971 San Fernando earthquake and one collapse from the 2014 South Napa earthquake. Slightly more than half of these were wood frame buildings, 13 were unreinforced masonry, and 9 were of reinforced concrete. I found that on average, about 25 percent of the total square footage collapses, given that at least some collapse occurs. The fraction varies by structural material, from about 17 percent (tilt-up concrete and wood) to about 50 percent (cast-in-place reinforced concrete). I also estimated the fraction of occupants in the collapsed area who would require USAR assistance by various levels of technical expertise, based on CERT training guidelines. Applying the observations from these historical California building collapses, I estimated that on the order of 2,400 people could realistically require extrication from approximately 5,000 collapsed buildings. Older buildings are generally more likely 118   The HayWired Earthquake Scenario—Engineering Implications to collapse, so the trapped population will tend to be in older buildings. There is no public model of USAR demands resulting from power loss to elevators. However, using relevant estimates of the total number of elevators nationwide and local experts’ observations that few San Francisco Bay area elevators have emergency power, I estimated that on the order of 22,500 people would be trapped in 4,500 stalled elevators. USAR Demands Under Ideal-World Conditions In Porter (Societal Consequences, this volume), I estimated that the number of collapsed buildings in the HayWired mainshock could be reduced by a factor of four if all buildings were designed with an earthquake importance factor of I=1.5 (as defined in American Society of Civil Engineers, 2010). Doing so would reduce the number of people trapped in collapsed buildings proportionately, from 2,500 people trapped in 5,000 collapsed buildings to perhaps 600 people trapped in 1,200 collapsed buildings. Retrofit of newer elevators with emergency power could reduce elevator entrapment to 14,000 people in 3,000 stalled elevators. Limitations Other buildings have collapsed in California earthquakes over the last 50 years that do not appear in the NISEE e-library or the other sources examined here. The distribution of affected area in these images may be biased relative to the distribution of affected area in the population of collapsed buildings, for example, if photographers who contributed to the NISEE e-library preferred to photograph buildings with more or less affected area than they would have done if they selected collapsed buildings at random to photograph. Absent a big California earthquake in which one can deliberately select collapsed buildings to examine in an unbiased way, I do not know how to test whether the photographers introduced bias in this way. However, the presence of numerous buildings with affected areas across the entire possible range of 0 to 100 percent shows that the observations are at least diverse, even if their representativeness cannot be known without more data. I find the database sufficiently useful for estimating the distribution of affected area, at least until better data—more definitely representative—come along. Some readers may object that the buildings shown here do not represent an exhaustive list of collapsed California buildings, but few surveys are exhaustive. Samples commonly provide useful statistical information. Acknowledgments Sarah Durphy (Estructure), Craig Stevenson (Aurecon), Lawrence Kornfield (San Francisco Department of Building Inspection, retired), John Osteraas (Exponent Failure Analysis Associates), Marko Schotanus (Rutherford and Chekene), George von Klan (GVK-ECS, Inc.), and Anne Wein (U.S. Geological Survey) reviewed the draft report and offered valuable comments and recommendations. The author thanks them for their contributions. References Cited American Society of Civil Engineers, 2010, Minimum design loads for buildings and other structures, ASCE/SEI 7–10: Reston, Va., American Society of Civil Engineers, 608 p. Applied Technology Council, 1985, ATC-13, Earthquake damage evaluation data for California: Redwood City, Calif., Applied Technology Council, 492 p. Applied Technology Council, 2009, Quantification of building seismic performance factors: Prepared for the Federal Emergency Management Agency, FEMA P–695, 421 p. Applied Technology Council, 2015a, Rapid visual screening of buildings for potential seismic hazards—A handbook (3d ed.): Prepared for the Federal Emergency Management Agency, FEMA P–154, 388 p. Applied Technology Council, 2015b, Rapid visual screening of buildings for potential seismic hazards—Supporting documentation (3d ed.): Prepared for the Federal Emergency Management Agency, FEMA P–155, 206 p. [Also available online at https://www.fema.gov/medialibrary-data/1426210695613-d9a280e72b32872161efab26a602283b/FEMAP-155_508.pdf .] Benfer, N.A., and Coffman, J.L., eds.,1973, San Fernando, California, earthquake of February 9, 1971—Effects on building structures; volume 1: U.S. Department of Commerce, National Oceanic and Atmospheric Administration, 448 p. Bibliop, 2010, Bullock’s Northridge, Northridge – 3rd Floor: Flickr online image taken on September 12, 2010, accessed December 11, 2017, at https://www.flickr.com/photos/53409445@N04/sets/72157625752688863/. Building Seismic Safety Council, 2009, NEHRP recommended seismic provisions for new buildings and other structures: Prepared for the Federal Emergency Management Agency, FEMA P–750, 406 p. California State University Northridge, 2017, Building information: California State University Northridge Oviatt Library web page, accessed June 2015, at https://library. csun.edu/About/BuildingInformation. Celebi, M., and Page, R., 2005, Monitoring earthquake shaking in federal buildings: U.S. Geological Survey Fact Sheet 2005–3052, 2 p. [Also available at https://pubs.usgs.gov/ fs/2005/3052/. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   119 Christensen, A., 2009, Remembering the Loma Prieta earthquake: Loma Prieta Stories blog, November 5, 2009, accessed June 2015, at https://lomaprietastories.wordpress. com/tag/santa-cruz-fires-and-earthquakes/. National Oceanic and Atmospheric Administration, n.d., Natural hazard images database: National Oceanic and Atmospheric Administration, National Centers for Environmental Information, database, doi:10.7289/V5154F01. Earth Science World Image Bank, 2017, Photo ID h32fh3: Earth Science World Image Bank web page, accessed November 2017, at http://www.earthscienceworld.org/ images/search/results.html?begin=10&num=2&numBe gin=1&Category=&Continent=&Country=&Keyword= San%20Andreas%20Fault#null. National Urban Search and Rescue Response System, 2009, Structural collapse technician course student manual: Washington, D.C., Federal Emergency Management Agency, 501 p. Emporis Corporation, 2007, Emporis research: Emporis Web page, accessed June 19, 2007, at www.emporis.com. Federal Emergency Management Agency, 2002, Rapid visual screening of buildings for potential seismic hazards— Training slide set: Federal Emergency Management Agency, Washington, DC, 144 p. Federal Emergency Management Agency, 2012, Hazus multi-hazard loss estimation methodology, earthquake model, Hazus®-MH 2.1 technical manual: Federal Emergency Management Agency, Mitigation Division, 718 p., accessed July 18, 2017, at https:// www.fema. gov/media-library-data/20130726-1820-25045-6286/ hzmh2_1_eq_tm.pdf. International Code Council, 2009, International building code 2009: Country Club Hills, Ill., International Code Council, 716 p. JPG Magazine LLC, 2017, Photograph eq8: JPG Magazine web page, accessed November 2017, at http://jpgmag.com/ photos/1989999. Krimgold, F., 1988, Search and rescue in collapsed reinforced concrete buildings: Proceedings of the Ninth World Conference on Earthquake Engineering, Tokyo and Kyoto, Japan, August 2–9, 1988, v. VII, p. 693–696. Lilliefors, H., 1967, On the Kolmogorov-Smirnov test for normality with mean and variance unknown: Journal of the American Statistical Association, v. 62, no. 318, p. 399–402 Moore, D., 2014, Loma Prieta’s legacy, 25 years later: Press Democrat, October 16, 2014, accessed June 2015, at http:// www.pressdemocrat.com/news/2983451-181/loma-prietaslegacy-25-years. PerformTech, Inc., 2011, Community emergency response team basic training participant manual: Developed for National CERT Program, Federal Emergency Management Agency, accessed November 17, 2015, at http://www.fema.gov/medialibrary/assets/documents/27403. Porter, K.A., 2015, Safe enough? A building code to protect our cities as well as our lives: Earthquake Spectra, v. 32, no. 2, p. 677–695, doi: http://dx.doi.org/10.1193/112213EQS286M. Scawthorn, C.R., Porter, K.A., and Blackburn, F.T., 1992, Performance of emergency-response services after the earthquake, in O'Rourke, T.D., ed., The Loma Prieta, California, earthquake of October 17, 1989—Marina District: U.S. Geological Survey Professional Paper 1551-F, p. F195– F215. Schiff, A., 2008, The ShakeOut scenario supplemental study— Elevators: Denver, Colo., SPA Risk LLC, accessed November 17, 2015, at https://goo.gl/jYJJfZ. Shoaf, K.I., Nguyen, L.H., Sareen, H.R., and Bourque, L.B., 1998, Injuries as a result of California earthquakes in the past decade: Disasters, v. 22, no. 3, p. 218–235. So, E.M.K., and Pomonis, A., 2012, Derivation of globally applicable casualty rates for use in earthquake loss estimation models, in World Conference on Earthquake Engineering, 15th, Lisbon, Portugal, September 24–28, 2012. Strakosch, G.R., and Caporale, R.S., 2010, The vertical transportation handbook: Hoboken, NJ, John Wiley & Sons, 624 p. Taylor, A., 2014, The Northridge earthquake—20 years ago today: The Atlantic, January 17, 2014, accessed June 2015, at https://www.theatlantic.com/photo/2014/01/the-northridgeearthquake-20-years-ago-today/100664/. National Elevator Industry, Inc., 2014, Elevator and escalator fun fact: Salem NY, National Elevator Industry, 1 p. Todd, D., Carino, N., Chung, R.M., Lew, H.S., Taylor, A.W., Walton, W.D., Cooper, J.D., and Nimis, R., 1994, 1994 Northridge earthquake—Performance of structures, lifelines, and fire protection systems: National Institute of Standards and Technology Special Publication 862, 187 p. National Fire Protection Association, 2014, NFPA 1670— Standard on operations and training for technical search and rescue incidents: National Fire Protection Association, NFPA 1670, 116 p., accessed November 15, 2015, at http://www.nfpa.org/codes-and-standards/documentinformation-pages?mode=code&code=1670. Yeo, G.L., and Cornell, C.A., 2002, Building-specific seismic fatality estimation methodology: Proceedings of the Fourth U.S.-Japan Workshop on Performance-Based Earthquake Engineering Methodology for Reinforced Concrete Building Structures, October 22–24, 2002, Toba, Japan, p. 59–74. 120   The HayWired Earthquake Scenario—Engineering Implications Appendixes 1 through 12—National Information Service for Earthquake Engineering e-Library Images of Building Collapse in California, 1965–2014 These appendixes present images of building collapse caused by earthquakes in California in the past 50 years. The appendixes are organized by earthquake in chronological order, beginning with the Borrego Mountain earthquake of 1968 (a list of earthquakes with no available images, such as Borrego Mountain, is in appendix 12). Within each section, collapses are documented with their descriptions and other metadata, followed by the author’s estimate of the affected area, and then images of the collapse. Unless noted otherwise, metadata and images are from collections in the National Information Service for Earthquake Engineering (NISEE) e-library. Permission for their use is granted at http://nisee.berkeley.edu/elibrary/about.html. Note that earthquake magnitudes may not exactly match final moment magnitudes determined by the U.S. Geological Survey. Because of the large number of images, figures are numbered by appendix. [Abbreviations used in the appendixes—ft, feet; ft2, square feet; in., inch; %, percent, Calif., California; St., Street, Ave., Avenue; Rd., Road. --, no data] Appendix 1. Santa Rosa (1969) Collapse Images Image Metadata and Description for Figure 1–1 Karl V. Steinbrugge Collection: S3715 Earthquake date and magnitude (M) October 1, 1969; M5.59 Title Creator Damage to wood Steinbrugge, frame house in Karl V. the fault zone Date Location October 6, 1969 North America/ Sonoma County/ United States/ Santa Rosa/ California Description Two-story wood frame building off foundations. Foundations were rotted and poorly braced. Gas lines ruptured when house fell. 718 Beaver St., Santa Rosa, California. Author’s Estimate of Affected Area 0% Figure 1–1.  Photograph showing two-story wood frame house that collapsed in the 1969 Santa Rosa, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   121 Image Metadata and Description for Figure 1–2 Karl V. Steinbrugge Collection: S3726 Earthquake date and magnitude (M) October 1, 1969; M5.59 Title Damage to wood frame house in the fault zone Creator Date Steinbrugge, Karl V. October 6, 1969 Location Description North America/ Two-story wood frame Sonoma building off foundations. County/ United Foundations were rotted States/ Santa and poorly braced. Gas lines Rosa/ California ruptured when house fell. 718 Beaver St., Santa Rosa, California. Author’s Estimate of Area Plan area≈13,000 ft2×3 stories. Area littered by bricks≈25 ft×15 ft=1% of 39,000 ft2. Figure 1–2.  Photograph showing part of a wall that collapsed onto a car in the 1969 Santa Rosa, California, earthquake. 122   The HayWired Earthquake Scenario—Engineering Implications Appendix 2. San Fernando (1971) Collapse Images Image Metadata and Description for Figure 2–1 Karl V. Steinbrugge Collection: S4473 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Creator Date Location Damage to wood Steinbrugge, February 16, 1971 North America/ Los frame house in Karl V. Angeles County/ the fault zone United States/ San Fernando/ California Description Damage to porches (probable cripple wall failure); chimney fell away from house. In the vicinity of Knox and Orange Grove Streets, in the fault zone. Author’s Estimate of Affected Area Approximately (120 ft2 porch)/(1,500 ft2 house)=8.0%. Figure 2–1.  Photograph showing damage to a wood frame house after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–2 Karl V. Steinbrugge Collection: S4533 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Chimney damage Schader, Eugene E. Author’s Estimate of Affected Area 0% Creator Date -- Location Description North America/ Los Chimney fell towards otherwise unAngeles County/ damaged wood frame house. United States/California Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   123 Figure 2–2.  Photograph showing chimney damage after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–3 Karl V. Steinbrugge Collection: S4581 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Creator Furniture store Schader, Eugene E. Date Location Description February 16, 1971 United States/ Furniture store. Unreinforced masonry San Fernando/ parapet has collapsed, dumping bricks California/North into the street and on to the sidewalk. America/ Los Large plate-glass windows are gone, Angeles County presumably shattered by the earthquake. San Fernando, California. Author’s Estimate of Affected Area Plan area≈40 ft×60 ft (?); area littered by bricks≈30 ft×15 ft=19%. Figure 2–3.  Photograph showing furniture store damage after the 1971 San Fernando, California, earthquake. 124   The HayWired Earthquake Scenario—Engineering Implications Image Metadata and Description for Figure 2–4 Karl V. Steinbrugge Collection: S4597, S4598, S4599, S4600, S4601, S4602. Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Apartments over retail space Creator Date Steinbrugge, Karl V. -- Location Description United States/ Apartments over retail space. Note that the San Fernando/ failure of the nonreinforced bearing walls California/North did not result in collapse. Unit masonry America/ Los construction, built prior to 1933. DownAngeles County town San Fernando commercial area. Author’s Estimate of Affected Area Plan area: 50 ft×75 ft×3 stories; masonry littering 250 ft (?)×15 ft (?)=3%. A B C D E F Figure 2–4.  Photographs (A–F) showing damage to apartments over retail space after the 1971 San Fernando, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   125 Image Metadata and Description for Figure 2–5 Karl V. Steinbrugge Collection: S4624, S4625, S4626, S4628, S4629, S4630, S4631, and S4633 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Creator Roof to the wall failed first Date Location Steinbrugge, February 18, 1971 North America/ Los Karl V. Angeles County/ United States/ Los Angeles/California Description Roof to the wall failed first. Ground cracks in the vicinity. Rear wall bulged out and rear roof fell. See S4625–4633. Light industrial buildings. Bradley Tract. Author’s Estimate of Affected Area No long shots show the length of any wall, address, or way to estimate overall size of the building. Benfer and Coffman (1973, p. 123) show 14 tilt-up buildings in the Bradley Tract with this kind of failure, including the one pictured in S4624. Steinbrugge’s photos seem to show a building on the north side of an east-west street, with failure on along the entire north wall and on the southwest bay. That only matches one building: 12884 Bradley Avenue, 131.5 ft wide (east-west) and 276 ft north-south, for a total floor area of 36,294 ft2. Collapses appear to cover 26 ft×26 ft on the southwest corner and 26 ft×131 ft on the north wall. I estimate 26-ft bays because the panels look approximately square and 131 ft equals 5 bays plus two 6-inch panel thicknesses. Affected area: (6×26 ft×26 ft)/(36,294 ft)=11%. Other tilt-ups in the Bradley Tract: I extracted the map of tilt-up damage from Benfer and Coffman (1973, p. 123) and overlaid it in Google Earth Pro, measuring the collapsed area with Google Earth Pro’s ruler tool. Results are shown in table 2–1. Table 2–1.  Collapsed tilt-up roofs in Bradley Tract, Los Angeles, after the 1971 San Fernando, California, earthquake. Collapsed area, in square feet Plan area, in square feet Affected area, in percent 12840 Bradley Avenue 21,461 48,400 44 12874 Bradley Avenue 2,460 21,000 12 12884 Bradley Avenue 4,056 36,294 11 12950 Bradley Avenue 3,060 30,240 10 12881 Bradley Avenue 5,678 58,500 10 12975 Bradley Avenue 18,180 77,600 23 13001 Bradley Avenue 6,400 85,050 8 13069 Bradley Avenue 7,030 45,000 16 15200 Bledsoe Street 3,700 19,800 19 15151 Bledsoe Street 4,050 51,800 8 12860 San Fernando Road 4,650 29,340 16 12806 San Fernando Road 11,260 63,400 18 12744 San Fernando Road 26,600 101,400 26 2,400 15,600 15 Address 12814 Bradley Avenue 126   An Earthquake Urban Search and Rescue Model for Earthquake Response and Application to the HayWired Scenario A C G B D E F H Figure 2–5.  Photographs (A–H) showing damage to industrial buildings in the Bradley Tract, Los Angeles, after the 1971 San Fernando, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   127 Image Metadata and Description for Figure 2–6 Karl V. Steinbrugge Collection: S4489 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Creator Date Location Description Older wood frame house Steinbrugge, Karl V. 1971 United States/ San Fernando/California/North America/ Los Angeles County Porch partial collapse on older wood frame house, probable cripple wall failure of house. Between Glen Oaks and Hubbard Streets. Author’s Estimate of Affected Area Plan area≈1,500 ft2 (?); collapsed area where people could be trapped=0%. Figure 2–6.  Photograph showing the partial collapse of a porch on older wood frame house, with probable cripple wall failure. Image Metadata and Description for Figure 2–7 Karl V. Steinbrugge Collection: S4491, S4492 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Soft-story failure Creator Date Steinbrugge, Karl V. -- Location Description United States/ San FerPink structure at the rear was a residence nando/ California/North over a garage. The first story collapsed, America/ Los Angeles note remains of automobile under the County building. Author’s Estimate of Affected Area Building area=30 ft×20 ft (?)×2; collapsed area=30 ft×20 ft (?)×1=50%. 128   The HayWired Earthquake Scenario—Engineering Implications A B Figure 2–7.  Photographs (A, B) showing soft-story failure after the 1971 San Fernando, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   129 Image Metadata and Description for Figure 2–8 William G. Godden (v. 4) Collection: GoddenJ53 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Creator Date Split-level house, San Fernando Valley Bertero, Vitelmo V. -- Location United States/California/ North America Description Collapse of a split-level wooden home. Large numbers of these split-level homes suffered significant damage because of a lack of adequate ties between the two levels. The upper level ripped away and crushed the lower garage walls, which did not have adequate lateral bracing.1 Additional discussion of this image is available in Godden Set J. 1 Author’s Estimate of Affected Area Building area≈15 ft×30 ft×3; collapsed area≈15 ft×30 ft×1=33%. Figure 2–8.  Photograph showing damage to a split-level house after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–9 Karl V. Steinbrugge Collection: S4195 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Creator Severe damage Bertero, Vito masonry telmo V. building Date Location Description February 1971 United States/ Los Angeles/ California/ North America/ Los Angeles County Collapsed semi-ambulent building, built in 1925, masonry construction. Structure: Veterans Administration Hospital (Sylmar). 130   The HayWired Earthquake Scenario—Engineering Implications Author’s Estimate of Affected Area Collapsed area: from this photo, it looks as if the lower story collapsed, so 50%. Figure 2–9.  Photograph showing severe damage to masonry building at the Veterans Administration Hospital (Sylmar) after the 1971 San Fernando, California, earthquake. Image Data and Description for Figure 2–10 The 1971 San Fernando earthquake (magnitude 6.7) collapsed four buildings at the San Fernando Veterans Administration Hospital complex, killing 47 people. The buildings had been built in 1925, before modern building codes were in effect. Image and description are from Celebi and Page (2005). Author’s Estimate of Affected Area The view is from the west. The semi-ambulent building was a long building oriented east to west, the second building from the south (that is, second from right), in the middle of the photo. Portions of the building are leaning at various angles to the north. The wing is a complete loss, but it appears as if it did not pancake. The estimate of 50% from NISEE S4195 seems reasonable. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   131 Figure 2–10.  Photograph showing four collapsed buildings at the San Fernando Veterans Administration Hospital complex in the 1971 San Fernando, California, earthquake (from Celebi and Page, 2005). Image Metadata and Description for Figure 2–11 Karl V. Steinbrugge Collection: S4529 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Creator Date Location Damage to older dwellings Olson, Robert A. -- United States/ California/ North America/ Los Angeles County Description Damage to older house caused by cripple wall collapse. Author’s Estimate of Affected Area Although the cripple wall collapsed, the living space does not appear to have experienced any drop in a roof or ceiling relative to the floor, so 0%. 132   The HayWired Earthquake Scenario—Engineering Implications Figure 2–11.  Photograph showing damage to older dwelling after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–12 Karl V. Steinbrugge collection: S4065 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Collapsed tower at southeast corner Creator Date Steinbrugge, Karl V. -- Location United States/ Sylmar/ California/North America/ Los Angeles County Description Collapsed tower at southeast corner. Olive View Hospital. Rear (east) elevation of Medical Treatment Building. Structure: Olive View Medical Treatment Building. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   133 Figure 2–12.  Photograph showing collapsed tower at Olive View Hospital, Sylmar, after the 1971 San Fernando, California, earthquake. Image Data and Description for Figure 2–13 San Fernando earthquake, February 1971, California. Fallen, structurally separated stair tower and leaning north stair tower (left) at Olive View Hospital. Emergency vehicles are visible in the foreground. View is from the west. Image taken by Reuben Kachadoorian, U.S. Geological Survey. Author’s Estimate of Affected Area Each wing appears to be approximately 240 ft×50 ft×5 stories×4 wings=240,000 ft2. The collapsed stair towers appear to be approximately 20 ft×40 ft×5 stories×2 towers=8,000 ft2, or 3.3%. 134   The HayWired Earthquake Scenario—Engineering Implications Figure 2–13.  Photograph showing collapsed and leaning stair towers at Olive View Hospital, Sylmar, after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–14 Karl V. Steinbrugge Collection: S4070 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Ambulance garage collapsed Creator Date Steinbrugge, Karl V. -- Location United States/ Sylmar/ California/ North America/ Los Angeles County Author’s Estimate of Affected Area By inspection (an engineering term meaning “just by looking at it”), 100%. Description Ambulance garage collapsed. Olive View Hospital. Southern elevation of Medical Treatment Building. See also S4139–44. Structure: Olive View ambulance garage. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   135 Figure 2–14.  Photograph showing a collapsed ambulance garage at Olive View Hospital, Sylmar, after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–15 Karl V. Steinbrugge Collection: S4115 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Olive View psychiatric building Creator Location Steinbrugge, United States/ Karl V. Sylmar/ California/ North America/ Los Angeles County Description Description Soft-story collapse, most evident at upper right of photo. Originally a one- and two-story building, irregular in plan, the first story collapsed in the earthquake. Structure: Olive View Medical Center, Calif. Ambulance garage collapsed. Olive View Hospital. Southern elevation of Medical Treatment Building. See also S4139–44. Structure: Olive View ambulance garage. Author’s Estimate of Affected Area Collapsed area: it appears as if the first story was about twice the area of the second, and all of the area of the first story has collapsed, so 67%. 136   The HayWired Earthquake Scenario—Engineering Implications Figure 2–15.  Photograph showing soft-story collapse of psychiatric building at Olive View Hospital, Sylmar, after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–16 Karl V. Steinbrugge collection: S4117 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Creator Psychiatric building collapsed Olson, Robert A. Location Description United States/ West elevation, psychiatric buildSylmar/ Caliing. This was a two-story buildfornia/ North ing—the first story collapsed. America/ Los Olive View. Structure: Olive Angeles County View Medical Center, Calif. Author’s Estimate of Affected Area This is another view of the previous building. Description Ambulance garage collapsed. Olive View Hospital. Southern elevation of Medical Treatment Building. See also S4139–44. Structure: Olive View ambulance garage. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   137 Figure 2–16.  Photograph showing first story collapse of psychiatric building at Olive View Hospital, Sylmar, after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–17 Karl V. Steinbrugge Collection: S4519 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Collapsed wood frame house Creator Date Steinbrugge, February 16, 1971 Karl V. Location Description United States/ Sylmar/ California/ North America/ Los Angeles County Collapsed wood frame house under construction on Tucker St. near Pacoima Dam. Author’s Estimate of Affected Area There is no other view of this house. It looks as if the garage (front left) and perhaps half of the living space (in the rear) at least partly collapsed, so approximately 67%. 138   The HayWired Earthquake Scenario—Engineering Implications Figure 2–17.  Photograph showing a collapsed wood frame house after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–18 Karl V. Steinbrugge Collection: S4501 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Soft-story failure Creator Date Steinbrugge, Karl V. 1971 Location Description United States/ Sylmar/ Cali- Two-story section over garage of this wood fornia/North America/ Los frame house on Almetz St. has collapsed in Angeles County the first story. In a new housing tract in Sylmar at base of hills and between Olive View and Veterans Administration Hospitals. Author’s Estimate of Affected Area There are no other views of this house. Judging by the description, this building resembled S4514 in layout, so approximately 33%. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   139 Figure 2–18.  Photograph showing soft-story failure in a wood frame house after the 1971 San Fernando, California, earthquake. Image Metadata and Description for Figure 2–19 Robert A. Olson Collection: R0070 Earthquake date and magnitude (M) February 9, 1971; M6.6 Title Creator Date Location Description VA Hospital -- 1971 -- Veterans Administration Hospital (Sylmar). Old masonry building in upper center of photo has completely collapsed. Constructed in 1925–26, with major additions in 1938 and 1949, the entire complex was demolished after the 1971 earthquake, and the entire 97 acres were dedicated in 1977 as Veterans Memorial Park. Structure: Veterans Administration Hospital (Sylmar). Author’s Estimate of Affected Area The collapsed building is the gray-roofed one, which appears to have been a one-story building whose entire area collapsed. 100%. 140   The HayWired Earthquake Scenario—Engineering Implications Figure 2–19.  Photograph showing damage to the Veterans Administration Hospital, Sylmar, after the 1971 San Fernando, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   141 Appendix 3. Imperial Valley (1979) Collapse Images Image Metadata and Description for Figure 3–2 Karl V. Steinbrugge Collection: S5584 Earthquake date and magnitude (M) October 15, 1979; M7.0 Title Cripple wall collapse Creator Hopper, Margaret G. Date Location October 1979 United States/ California/ North America/ Imperial County Description Cripple wall collapse—wood frame house on G St. Author’s Estimate of Affected Area By inspection, 0%. Figure 3–2.  Photograph showing cripple wall collapse on a wood frame house after the 1979 Imperial Valley, California, earthquake. 142   The HayWired Earthquake Scenario—Engineering Implications Image Metadata and Description for Figure 3–3 Karl V. Steinbrugge Collection: S5585 Earthquake date and magnitude (M) October 15, 1979; M7.0 Title Cripple wall collapse Creator Date Hopper, Margaret G. October 1979 Location Description Brawley, Imperial County, Cripple wall collapse—wood frame Calif. house on G St. Author’s Estimate of Affected Area By inspection, 0%. Figure 3–3.  Photograph showing cripple wall collapse on a wood frame house after the 1979 Imperial Valley, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   143 Appendix 4. Westmorland (1981) Collapse Images Image Metadata and Description for Figure 4–1 National Oceanic and Atmospheric Administration (n.d.) Earthquake date and magnitude (M) April 26, 1981; M5.6 Title Westmorland 1981 Creator Olsen, Robert O. Date -- Location Description North America/ United View of a two-story building which States/ California partly collapsed in the earthquake. Note the undamaged one story building on the left. Photo credit: California Governor’s Office of Emergency Services Earthquake Program. Author’s Estimate of Affected Area 100% Figure 4–1.  View of a two-story building on West Main Street, Westmorland, after the April 26, 1981, Westmorland, California, earthquake. 144   The HayWired Earthquake Scenario—Engineering Implications Appendix 5. Coalinga (1983) Collapse Images Image Metadata and Description for Figure 5–1 William G. Godden (v. 4) Collection: GoddenJ19 Earthquake date and magnitude (M) Title May 2, 1983; M6.5 2-story building, Coalinga Creator Date Bertero, Vitelmo V. -- Location Description North America/ United This two-story wood frame dwelling States/ California underwent a lateral displacement of more than half a meter, as illustrated by the slant in the porch columns, and also fell more than half a meter from its foundation, owing to lack of adequate anchorage and support. Author’s Estimate of Affected Area 0% Figure 5–1.  Photograph showing lateral displacement of two-story wood frame dwelling after the 1983 Coalinga, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   145 Image Metadata and Description for Figure 5–2 William G. Godden (v. 4) collection: GoddenJ52 Earthquake date and magnitude (M) Title May 2, 1983; M6.5 Chimney collapse, Coalinga Creator Date Bertero, Vitelmo V. -- Location United States/ Coalinga/ California/ North America/ Fresno County Description Chimney collapse of a modern house, 1983 Coalinga earthquake. Most of the chimneys were thrown down because of the lack of proper connections (straps) to the building.1 Additional discussion of this image is available in Godden Set J. 1 Author’s Estimate of Affected Area There are no other views of this building. Typical single-family dwelling is approximately 1,500 ft2, but this one looks a little larger, say 50% larger or 2,250 ft2. Bricks litter an area approximately 20 ft×10 ft=200 ft2, or 9%. Figure 5–2.  Photograph showing chimney collapse of a modern house after the 1983 Coalinga, California, earthquake. Image Metadata and Description for Figure 5–3 William G. Godden (v. 4) Collection: GoddenJ23 Earthquake date and magnitude (M) Title Creator Bertero, Vitelmo V. May 2, 1983; M6.5 Collapse of wooden porch, Coalinga Additional discussion of this image is available in Godden Set J. 1 Date -- Location Description United States/ Collapse of a wooden porch (owing to lack of California/ North proper anchorage to the wooden frame of the America house and of a proper lateral-resistant supporting system) owing to vibratory response.1 146   The HayWired Earthquake Scenario—Engineering Implications Author’s Estimate of Affected Area There are no other views of this building. Typical single-family dwelling is approximately 1,500 ft2. This porch appears to have measured 12 ft×20 ft, so 200 ft2 / 1,500 ft2≈15%. Figure 5–3.  Photograph showing collapse of a wooden porch after the 1983 Coalinga, California, earthquake. Image Metadata and Description for Figure 5–4 William G. Godden (v. 4) collection: GoddenJ29 Earthquake date and magnitude (M) Title Creator Bertero, Vitelmo V. May 2, 1983; M6.5 Unreinforced brick building, Coalinga Date -- Location Description United States/ The second story, 8-in., unreinforced solid brick California/ North masonry walls of this commercial building America in Coalinga collapsed, owing to inadequate tying at the floor, roof, and transverse walls.1 Additional discussion of this image is available in Godden Set J. 1 Author’s Estimate of Affected Area There are no other views of this building. It looks as if about half of the upper story of a two-story building collapsed (25%), plus bricks litter the perimeter, so say 30%. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   147 Figure 5–4.  Photograph showing collapse of an unreinforced brick building after the 1983 Coalinga, California, earthquake. Image Metadata and Description for Figure 5–5 Robert A. Olson Collection: R0321 Earthquake date and magnitude (M) Title May 2, 1983; M6.5 Heavy wooden overhang fell on sidewalk Creator Date Location Description -- -- -- Heavy wooden overhang fell from storefront on to the sidewalk. Damaged concrete block wall at the right. Author’s Estimate of Affected Area No long shot to show how long the building is. No address. No estimate of affected area. 148   The HayWired Earthquake Scenario—Engineering Implications Figure 5–5.  Photograph showing a heavy wooden overhang fallen onto sidewalk in the 1983 Coalinga, California, earthquake. Image Metadata and Description for Figure 5–6 Robert A. Olson Collection: R0323 Earthquake date and magnitude (M) May 2, 1983; 6.5 Title Creator Date Location Porch pulled away from church building -- -- -- Description Porch running the full width of the church simply pulled away from the rest of the building. Built in 1946, the stabilized adobe building was heavily damaged, but did not collapse. On the corner of Jefferson St. Author’s Estimate of Affected Area There are no other views of this building. Guess building area≈30 ft×90 ft=2,700 ft2, guess porch measured 20 ft×10 ft=7%. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   149 Figure 5–6.  Photograph showing porch pulled away from church building after the 1983 Coalinga, California, earthquake. Image Metadata and Description for Figure 5–7 Karl V. Steinbrugge Collection: S5765 Earthquake date and magnitude (M) Title May 2, 1983; M6.5 Veneer also fell into the first story Creator Steinbrugge, Karl V. Date Location Description May 3, 1983 North America/ Fresno Veneer also fell into the first story. County/ United States/ All reinforced brick buildings in Coalinga/ California the downtown Coalinga area were demolished. Author’s Estimate of Affected Area No long shots, no address, no estimate of affected area. 150   The HayWired Earthquake Scenario—Engineering Implications Figure 5–7.  Photograph showing veneer fallen into first story of downtown building after the 1983 Coalinga, California, earthquake. Image Metadata and Description for Figure 5–8 Karl V. Steinbrugge Collection: S5773 Earthquake date and magnitude (M) Title Creator May 2, 1983; M6.5 Parapet damaged Steinbrugge, Karl V. Date Location Description May 3, 1983 North America/ Fresno Parapet damage. All reinforced brick County/ United States/ buildings in the downtown Coalinga Coalinga/ California area were demolished. See S5828– 5830 for "after" views. Author’s Estimate of Affected Area Building was at E. Durian Avenue and Coalinga Plaza, Coalinga, Calif. (https://www.masonryinstitute.org/pdf/909.pdf), possibly 286 Coalinga Plaza. No old satellite imagery. No estimate of plan area. No estimate of effected area. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   151 Figure 5–8.  Photograph showing parapet damage to a building in downtown Coalinga after the 1983 Coalinga, California, earthquake. 152   The HayWired Earthquake Scenario—Engineering Implications Appendix 6. Morgan Hill (1984) Collapse Images Image Metadata and Description for Figure 6–1 Karl V. Steinbrugge Collection: S5840 Earthquake date and magnitude (M) April 24, 1984; M6.19 Title Most severely damaged dwelling Creator Steinbrugge, Karl V. Date Location April 28, 1984 United States/ Morgan Hill/ California/ North America/ Santa Clara County Description Most severely damaged dwelling. Sheathing between first floor and foundation was fiberboard with little strength. Morgan Hill, California, Anderson Lake area. Author’s Estimate of Affected Area By inspection, 0%. Figure 6–1.  Photograph showing the most severely damaged dwelling in the 1984 Morgan Hill, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   153 Image Metadata and Description for Figure 6–2 Karl V. Steinbrugge Collection: S5839 and Federal Emergency Management Agency (2002). Earthquake date and magnitude (M) April 24, 1984; M6.19 Title Creator Dwelling on the left moved owing to landslide Steinbrugge, Karl V. Date Location Description April 28, 1984 Morgan Hill/ California/ Dwelling on the left moved, North America/ Santa Clara owing to landsliding from the County/ United States earthquake. Morgan Hill, California. Anderson Lake area. Author’s Estimate of Affected Area The right-hand image is from the FEMA National Earthquake Technical Assistance Training Program training slideset, entitled “Postearthquake Safety Evaluation of Buildings” (Federal Emergency Management Agency, 2002). Plan area from top to bottom floors appear to be 2:2:1. The bottom floor experienced some collapse, so say 20%. A B Figure 6–2.  Photographs (A, B) showing dwellings that have moved, owing to landslide in the 1984 Morgan Hill, California, earthquake. 154   The HayWired Earthquake Scenario—Engineering Implications Appendix 7. Whittier Narrows (1987) Collapse Images Image Metadata and Description for Figure 7–1 Karl V. Steinbrugge Collection: S6014 Earthquake date and magnitude (M) October 1, 1987; M6.0 Title Chimney collapsed Creator Steinbrugge, Karl V. Date Location October 3, 1987 United States/ Whittier/ California/North America/ Los Angeles County Description Damage to roof from chimney collapsing. Whittier, California. Author’s Estimate of Affected Area By inspection, 0%. Figure 7–1.  Photograph showing damage to roof from collapsed chimney after the 1987 Whittier Narrows, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   155 Image Metadata and Description for Figure 7–2 Karl V. Steinbrugge Collection: S6023 Earthquake date and magnitude (M) October 1, 1987; M6.0 Title Chimney collapsed Creator Steinbrugge, Karl V. Date Location October 3, 1987 United States/ Whittier/ California/ North America/ Los Angeles County Description Chimney collapsed away from the house. Whittier, California. Author’s Estimate of Affected Area There are no other views of this house in adjacent records, so assume typical area 1,500 ft2 and that bricks litter an area 5 ft×10 ft=3%. Figure 7–2.  Photograph showing collapsed chimney in a house in Whittier after the 1987 Whittier Narrows, California, earthquake. 156   The HayWired Earthquake Scenario—Engineering Implications Image Metadata and Description for Figure 7–3 Karl V. Steinbrugge Collection: S6020 Earthquake date and magnitude (M) October 1, 1987; M6.0 Title Creator Chimney damage Steinbrugge, Karl V. Date Location October 3, 1987 United States/ Whittier/ California/ North America/ Los Angeles County Author’s Estimate of Affected Area House looks larger than typical: assume 3,000 ft2. Bricks litter an area 8 ft×8 ft=2%. A B Figure 7–3.  Image showing collapsed chimney in the 1987 Whittier Narrows, California, earthquake. Description One chimney collapsed, but not the other. Whittier, California. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   157 Image Metadata and Description for Figure 7–4 Karl V. Steinbrugge Collection: S6022 Earthquake date and magnitude (M) October 1, 1987; M6.0 Title Creator Chimney damage Steinbrugge, Karl V. Date Location Description October 3, 1987 United States/ Whittier/ California/ North America/ Los Angeles County One chimney collapsed, but not the other. Whittier, California. Author’s Estimate of Affected Area Assume typical plan area for single-family dwelling of 1,500 ft2. Bricks litter an area approximately 5 ft×10 ft=3%. Figure 7–4.  Photograph showing chimney damage from the 1987 Whittier Narrows, California, earthquake. 158   The HayWired Earthquake Scenario—Engineering Implications Image Metadata and Description for Figure 7–5 Karl V. Steinbrugge Collection: S6024 Earthquake date and magnitude (M) October 1, 1987; M6.0 Title May Company parking Creator Steinbrugge, Karl V. Date Location Description October 3, 1987 North America/ Los Angeles County/ United States/ Whittier/ California May Company parking structure. Roof failed; damage shown is from demolition. Whittier, California. Author’s Estimate of Affected Area No long shots. Google Earth imagery does not date back to 1987, so there is no way to estimate total area of lot. No estimate of affected area. Figure 7–5.  Photograph showing failure of parking structure roof after the 1987 Whittier Narrows, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   159 Appendix 8. Loma Prieta (1989) Collapse Images Image Metadata and Description for Figure 8–1 Loma Prieta Blacklock Collection: LP0042 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Wall collapse in unreinforced masonry Creator Blacklock, James R. Date Location Description 1989 United States/ Santa Cruz/ California/ North America/ Santa Cruz County Wall collapse in unreinforced masonry building. Santa Cruz, California. Author’s Estimate of Affected Area This is the historic Hihn Building, 1205 Pacific Avenue, Santa Cruz, California 95060. The parcel (APN 00507517000) covers 8,180 ft2 according to Google Earth. Another photograph with a wider field of vision shows that the building stood two stories tall in 1989 (Moore, 2014) Total building area=16,360 ft2. Bricks litter an area about 16 ft×12 ft, or 1%. A B Figure 8–1.  Photographs (A, B) showing wall collapse in an unreinforced masonry building after the 1989 Loma Prieta, California, earthquake. 160   The HayWired Earthquake Scenario—Engineering Implications Image Metadata and Description for Figure 8–2 Loma Prieta Blacklock Collection: LP0066 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Creator Parapet and wall Blacklock, failures in bakJames R. ery building Date 1989 Location North America/ Santa Cruz County/ United States/ Watsonville/ California Description Parapet and wall failures in bakery building. Watsonville, California. Author’s Estimate of Affected Area Location: 15 E. Beach Street (at Union Street), Watsonville, California. No long shot. No 1989 satellite imagery exists, so there is no estimate of shape or size of the damaged building. No estimate of affected area. Figure 8–2.  Photograph showing parapet and wall failures in Watsonville after the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–3 Loma Prieta Blacklock Collection: LP0070, LP0072, LP0073, and LP0074 Earthquake date and magnitude (M) Title October 17, 1989; M7.09 Failed parapets on Main Street Creator Blacklock, James R. Date 1989 Location North America/ Santa Cruz County/ United States/ Watsonville/ California Description Older building with failed parapets on Main St. Watsonville, California. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   161 Author’s Estimate of Affected Area Two buildings are addressed here. The tall building labeled “Canada” on front and back appears to be 307 Main Street, Watsonville (see https://www.ngdc.noaa.gov/hazardimages/picture/show/259). According to Google Earth Pro, the lot at 307 Main Street measures 30 ft×125 ft. The building (now removed) appears to fill the parcel, with a total building area of 7,500 ft2. Collapsed parapet and second story wall appears to litter an area about 90 ft long (counting collapsed portions of both long walls, on the north and south sides) and perhaps 15 ft wide, for total affected area=(90 ft×15 ft)/(7,500 ft2)=18%. The building with the collapsed parapet on its front facade appears to be located at what is now 311 Main Street, Watsonville, the middle one of three buildings on what is now one parcel. The center building appears to be about 65 ft wide, with the front 35 ft or so, occupying two stories and the back 90 ft a single story. Bricks litter the 65 ft length by 15 ft, for an affected area of (65 ft×15 ft)/(65 ft×125 ft + 65 ft×35 ft)=9.4%. A B C D Figure 8–3.  Photographs showing failed parapets in Watsonville after the 1989 Loma Prieta, California, earthquake. A, The building at the far left is 307 Main Street; the building in the foreground is 311 Main Street. B–D, Three views of the sides and rear of 307 Main Street. Image Metadata and Description for Figure 8–4 Loma Prieta Blacklock Collection: LP0080 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Failed brick parapet fell on sidewalk Creator Blacklock, James R. Date 1989 Location North America/ Santa Cruz County/ United States/ Watsonville/California Description Damaged building near Main St. Failed brick parapet fell on sidewalk 162   The HayWired Earthquake Scenario—Engineering Implications Author’s Estimate of Affected Area No address, no long shots. There is no way to tell how long this wall is or how deep the building is perpendicular to this wall. No estimate of affected area. Figure 8–4.  Photograph showing failed brick parapet fallen onto sidewalk in Watsonville after the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–5 Loma Prieta Blacklock Collection: LP0081–LP0085 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title St. Patrick's Church Creator Blacklock, James R. Date 1989 Location North America/ Santa Cruz County/ United States/ Watsonville/ California Description Front view of damaged street. Patrick's Church. Watsonville, California. Author’s Estimate of Affected Area Littered area≈200 ft2 at front (east) entrance, about 200 ft2 at south transept, and 50 ft2 at east end of north facade. Plan area≈9,070 ft2, and assume 1,000 ft2 of additional galleries. Affected area≈(450 ft2)/(10,000 ft2)=4.5%. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   163 A B D C E Figure 8–5.  Photographs (A–E) showing damage to St. Patrick’s Church, Watsonville, after the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–6 Loma Prieta Blacklock Collection: LP0087 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Damaged bike store with failed parapet Creator Blacklock, James R. Date Late 1989 Location North America/ Santa Cruz County/ United States/ Watsonville/ California Description Damaged bike store with failed parapet. Watsonville, California. Author’s Estimate of Affected Area No other shots. No street name. No indication as to the former location of Watsonville Cyclery. 202 Main Street does not look like this. Littered area≈50 ft×12 ft. Plan area≈40 ft×60 ft. Affected area≈25%. 164   The HayWired Earthquake Scenario—Engineering Implications Figure 8–6.  Photograph showing damaged bike store in Watsonville with failed parapet after the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–7 Loma Prieta Blacklock Collection: LP0090 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Wood frame house with failed foundation Author’s Estimate of Affected Area By inspection, 0%. Creator Blacklock, James R. Date Late 1989 Location North America/ Santa Cruz County/ United States/ Watsonville/ California Description Pink frame house with failed foundation. Watsonville, California. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   165 Figure 8–7.  Photograph showing house with failed foundation in Watsonville after the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–8 Loma Prieta Collection: LP0462 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Creator 6th and Bluxome St. Dickenson, Stephen E. Date 1989 Location United States/ California/ North America/San Francisco Description Collapse of fourth story wall from unreinforced brick building at 6th and Bluxome Streets, San Francisco, South of Market. Author’s Estimate of Affected Area Also see LP0460 (below). The location is sometimes reported as near 5th and Townsend Streets, sometimes on Bluxome Street near 6th and Townsend Streets. If the latter, the building appears to be 178 Bluxome Street, at the south end of Bluxome on the north side of the street (assessor’s parcel number [APN] 3785135), with parcel area 15,300 ft2 according to Google Earth Pro. With four stories, the total building area would be 61,200 ft2. The debris runs the length of the facade (135 ft) and twice as wide as the sidewalk, perhaps 24 ft. Five people were killed by the wall collapse. Affected area=(135×24)/(61,200)=5.3%. 166   The HayWired Earthquake Scenario—Engineering Implications Figure 8–8.  Photograph showing collapse of fourth story wall from unreinforced brick building after the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–9 Loma Prieta Collection: LP0460 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Unreinforced brick building Author’s Estimate of Affected Area Same as LP0462 (above). Creator Date Kayen, Robert E. Late 1989 Location United States/ California/ North America/ San Francisco Description 6th and Bluxome Streets, south of Market. Collapse of unreinforced brick wall. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   167 Figure 8–9.  Photograph showing collapse of unreinforced brick wall from the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–10 Loma Prieta Collection: LP0375 Earthquake date and magnitude (M) Title Creator Date October 17, 1989; M7.09 Collapse of apartment buildings Seed, Raymond B. Late 1989 Author’s Estimate of Affected Area By inspection, two buildings, each with 25% collapse. Location United States/ California/ North America/ San Francisco Description Collapse of two four-story apartment buildings (soft ground floors). Marina District, San Francisco, California. 168   The HayWired Earthquake Scenario—Engineering Implications Figure 8–10.  Photograph showing collapse of apartment buildings with soft ground floors in the Marina District of San Francisco after the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–11 Loma Prieta Collection: LP0499 Earthquake date and magnitude (M) Title October 17, 1989; M7.09 Collapsed building in Marina District Creator Harris, S. P. Date October 17, 1989 Location United States/ California/ North America/ San Francisco Description Collapsed apartment building at 2090 Beach St, after the fire was much advanced. Note firefighter directing water onto exposed side of building. Marina District, San Francisco, California. Author’s Estimate of Affected Area This had been a four-story building, now with only one story remaining somewhat intact, so 75% collapse. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   169 Figure 8–11.  Photograph showing collapsed four-story building in the Marina District, San Francisco, after the 1989 Loma Prieta, California, earthquake (photograph from Scawthorn and others, 1992, p. 204, fig. 11). Image Metadata and Description for Figure 8–12 Karl V. Steinbrugge Collection: S6144 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Soft-story collapse Creator unknown Date -- Location United States/ California/ North America/ San Francisco Description Soft-story collapse of apartment building in the Marina District, San Francisco, California. Author’s Estimate of Affected Area To the author’s personal knowledge, this had been a three-story building, so affected area=67%. What is remarkable about this building is that it appears in many photos of the Marina District, almost entirely without identifying information other than the neighborhood. One photo caption says the building was at Beach Street and Divisadero Street. The view of the Golden Gate Bridge tower in the background tells us that it was at the northwest corner, apparently 3700 Divisadero Street, San Francisco, California 94123-1000 (APN 0913037). 170   The HayWired Earthquake Scenario—Engineering Implications Figure 8–12.  Photograph showing soft-story collapse of an apartment building in the Marina District, San Francisco, in the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–13 Loma Prieta Collection: LP0459 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Front and Davis St. Creator Dickenson, Stephen E. Date Late 1989 Location United States/ California/ North America/ San Francisco Description Front and Davis Streets. Collapse of unreinforced masonry wall from third floor of building. Embarcadero/ Financial District, San Francisco. Author’s Estimate of Affected Area Front Street is parallel to Davis Street, so the recorded location makes no sense. Matching the background buildings, the address seems to be 235 Front Street, San Francisco, California, on the northwest corner of Front Street and Halleck Street. The view is toward the northwest. The building appears to be on APN 0237047, whose area is 4,960 ft2. Aerial photography dating from 1938 and available in Google Earth Pro shows a building of uniform height covering the entire parcel, suggesting a total building area of 14,880 ft2. The collapsed wall faces Front Street. The facade length is 72 ft, so the affected area appears to be 36 ft. I can find no images of the masonry on the sidewalk. I assume it litters an area 36 ft×16 ft wide, for an affected area of (36 ft×12 ft)/(14,880 ft2)=2.9%. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   171 Figure 8–13.  Photograph showing collapse of unreinforced masonry wall in the Embarcadero/Financial District, San Francisco, after the 1989 Loma Prieta, California, earthquake. Image Metadata and Description for Figure 8–14 Loma Prieta Blacklock Collection: LP0041 Earthquake date and magnitude (M) October 17, 1989; M7.09 Title Interior structural failures at department store Creator Blacklock, James R. Date Late 1989 Location North America/ Santa Cruz County/United States/ Santa Cruz/ California Description Interior structural failures at department store. Santa Cruz, California. 172   The HayWired Earthquake Scenario—Engineering Implications Author’s Estimate of Affected Area This may be Ford’s Department Store, the only department store mentioned in connection with collapse in Santa Cruz after the Loma Prieta earthquake. The building was located at the corner of Pacific Avenue and Cathcart Street, Santa Cruz, California. The address is 1101 Pacific Avenue, Santa Cruz, California (APN 00514120000), on the northwest corner of Pacific Avenue and Cathcart Street. The parcel measures 20,900 ft2, according to Google Earth Pro. One can see an exhaust vent above the truss in the background, so Ford’s Department Store must have been one story tall in this portion of the building. The affected area here appears to be perhaps 1,000 ft2. A personal-injury law firm’s website (http://csfwlaw.com/successful_personal_injury_lawsuits) says that the “back of the Ford’s Department Store collapsed,” indicating that the entire interior did not collapse. More images from a local blog (Christensen, 2009) and JPG Magazine LLC (2017) suggest that something like the back one-third of the store collapsed, or roughly 33%. Figure 8–14.  Photograph showing interior structural failure in a department store in Santa Cruz after the 1989 Loma Prieta, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   173 Appendix 9. Northridge (1994) Collapse Images Image Metadata and Description for Figure 9–1 Northridge Collection: NR327 Earthquake date and magnitude (M) Title Creator Date Location Description January 17, 1994; M6.69 Collapsed apartment building Unknown 1994 Northridge/ California/ North America/ Los Angeles County/ United States Collapsed apartment building, three-story wood frame. Northridge, California. Author’s Estimate of Affected Area According to Todd and others (1994, p. 23; fig. 9–2), there were four collapsed three-story buildings. The ground story of two of the buildings completely collapsed, the ground story of about half of a third three-story building collapsed, and approximately one-eighth of a fourth. Thus, the affected areas are 33%, 33%, 17%, and 4%. Figure 9–1.  Photograph showing collapsed apartment building after the 1994 Northridge, California, earthquake. 174   The HayWired Earthquake Scenario—Engineering Implications Figure 9–2.  Diagram showing parking areas, collapsed areas, and locations of deaths on the first level of Northridge Meadows Apartments after the 1994 Northridge, California, earthquake (Todd and others, 1994, p. 23). Image Metadata and Description for Figure 9–3 Northridge Collection: NR335 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title Building components fell onto off-ramp Creator Aschheim, Mark A. Date January 19, 1994 Location Description North America/ Los Building at eastbound off-ramp of Angeles County/ United Route 101 south at Van Nuys States/ Los Angeles/ exit. View to south. Failed buildCalifornia ing components fell onto offramp. Los Angeles, California. Author’s Estimate of Affected Area This building was repaired. It is located at 4717 Van Nuys Boulevard, Sherman Oaks, California 91403. According to Google Earth Pro, the building area is 16,094 ft2. There are no long shots or aerial shots to show the extent of the roof collapse. No estimate of affected area. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   175 Figure 9–3.  Photograph showing building components fallen onto off-ramp after the 1994 Northridge, California, earthquake. Image Metadata and Description for Figure 9–4 Northridge Collection: NR353 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title Creator Northridge Mead- Reitherman, ows Apartments Robert K. Author’s Estimate of Affected Area Same as figure 9–1. Date Location February 12, 1994 Northridge/ California/ North America/ Los Angeles County/United States Description Collapse of ground story in Northridge, California. Structure: Northridge Meadows Apartments. 176   The HayWired Earthquake Scenario—Engineering Implications Figure 9–4.  Photograph showing collapse of ground story at Northridge Meadows Apartments after the 1994 Northridge, California, earthquake. Image Metadata and Description for Figure 9–5 Northridge Collection: NR357 Earthquake date and magnitude (M) Title Creator Date January 17, 1994; M6.69 Northridge Meadows Apartment Reitherman, Robert K. February 12,1994 Author’s Estimate of Affected Area Same as figure 9–1. Location Description Northridge/ California/ Northridge Meadows ApartNorth America/ Los ments. Collapse of ground Angeles County/ United story. Northridge, California. States Structure: Northridge Meadows Apartments. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   177 Figure 9–5.  Photograph showing collapse of ground story at Northridge Meadows Apartments after the 1994 Northridge, California, earthquake. Image Metadata and Description for Figure 9–6 Northridge Collection: NR358 Earthquake date and magnitude (M) Title Creator January 17, 1994; M6.69 Northridge Meadows Apartment Reitherman, Robert K. Author’s Estimate of Affected Area Same as figure 9–1. Date Location February 12, 1994 United States/ Northridge/ California/ North America/ Los Angeles County Description Northridge Meadows Apartments. Collapse of ground story. Northridge, California. Structure: Northridge Meadows Apartments. 178   The HayWired Earthquake Scenario—Engineering Implications Figure 9–6.  Photograph showing collapse of ground story at Northridge Meadows Apartments after the 1994 Northridge, California, earthquake. Image Metadata and Description for Figure 9–7 Northridge Collection: NR408–NR409 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title 2-story masonry building Creator Date Stojadinovic, Bozidar January 19, 1994 Location Description North America/ Los An- 1004 West Channel Road at geles County/ United Pacific Coast Highway (near States/ Santa Monica/ Pacific Palisades). Damage to California two-story masonry building. Heavy shear cracking on side walls. Out of plane failure of the second story. State Beach Cafe, Santa Monica, California. Author’s Estimate of Affected Area The address appears to be 108 West Channel Road, Santa Monica, which is adjacent to 112 (it is not 1004). From size of replacement building, which fills the lot, the damaged building appears to be 1,500 ft2 in plan, or 3,000 ft2 total. Bricks litter 40 ft of facade×10 ft across sidewalk. Affected area is therefore approximately 400 ft2/3,000 ft2=13%. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   179 Figure 9–7.  Photograph showing damage to twostory masonry building in Santa Monica after the 1994 Northridge, California, earthquake. Image Metadata and Description for Figure 9–8 Northridge Collection: NR413, NR414 Earthquake date and magnitude (M) Title January 17, 1994; M6.69 Four-story masonry building Creator Date Stojadinovic, January 19, 1994 Bozidar Location Description North America/ Los Four-story masonry building, 827 Angeles County/ Fourth St. Damage to the fourth United States/ Santa and third floor of the building. The Monica/ California masonry facade fell out of plane and took with it the fourth floor terrace. This building had been scheduled for a retrofit to begin on Monday, January 17, 1994. Three layers of thick unreinforced masonry. Damage to the top story and balcony. Little damage on the sides and below the third story. Santa Monica, California. 180   The HayWired Earthquake Scenario—Engineering Implications Author’s Estimate of Affected Area Building still exists and has been repaired. Google Earth Pro says building area=31,314 ft2. Affected area looks like (55 ft×12 ft)/(31,314 ft2)=2.1%. A B Figure 9–8.  Photograph showing damage to four-story masonry building in Santa Monica after the 1994 Northridge, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   181 Image Metadata and Description for Figure 9–9 Northridge Collection: 201012024 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title Collapsed unreinforced chimney Creator Date Location Description Reitherman, Robert K. 2010 Unknown This residential chimney of unreinforced blocks collapsed during the 1994 Northridge earthquake. Author’s Estimate of Affected Area Masonry litters an area about 10 ft×4 ft, or 40 ft2. Assuming a typical 1,500 ft2 home, the affected area is 2.7%. Figure 9–9.  Photograph showing collapsed unreinforced chimney after the 1994 Northridge, California, earthquake. Image Metadata and Description for Figure 9–10 Northridge Collection: NR559 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title Creator Date Location Description Parking structure on Cal State Northridge campus Unknown 1994 Northridge/ California /North America/ Los Angeles County/ United States Parking structure on Zelzah Ave., California State University, Northridge, campus. This is a three-story precast concrete parking structure. Overall view showing collapse at east end of the structure. Structure: Cal State Northridge Parking Author’s Estimate of Affected Area From figure 9–11, looks like about 35%. 182   The HayWired Earthquake Scenario—Engineering Implications Figure 9–10.  Photograph showing collapse of a parking structure on the California State University, Northridge, campus after the 1994 Northridge, California, earthquake. Figure 9–11.  Photograph from the 1994 Northridge, California, earthquake. Image is from Earth Science World Image Bank (2017), which describes it as follows: California State University, Northridge parking structure that partly collapsed during the 1994 earthquake. Scientists believe it was the lack of shear walls, being precast, and lack of extra steel reinforcements in vertical columns that led to the damage seen here. This is 5 kilometers northeast of the epicenter. (Photograph by P.W. Weigand. Copyright California State University, Northridge, Geology Department, permission granted per earthscienceworld.org.) Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   183 Image Metadata and Description for Figure 9–12 Northridge Collection: NR579 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title Creator Date Location Fashion Center Reitherman, February 12, 1994 Northridge/ Califorparking garage Robert K. nia/North America/ Los Angeles County/ United States Description Collapse of parking garage floors. See NR459–461 for damage to Broadway department store. Fashion Center, Northridge, California. Structure: Northridge Fashion Center Parking. Author’s Estimate of Affected Area From an Atlantic Magazine image (Taylor, 2014), the collapsed area looks like about 35%. Figure 9–12.  Photograph showing collapse of floors in the Northridge Fashion Center parking garage from the 1994 Northridge, California, earthquake. 184   The HayWired Earthquake Scenario—Engineering Implications Image Metadata and Description for Figure 9–13 Northridge Collection: NR221 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title Bullock's retail store Creator Date Location Description Unknown 1994 Northridge/ California/ North America/ Los Angeles County/ United States Northridge Fashion Island Center. Interior reinforced-concrete columns remain standing following collapse of secondand third-floor concrete waffle slabs. Intact portion of waffle slab roof shows typical slab construction. Structure: Bullock’s Department Store. Author’s Estimate of Affected Area A plan of Bullock’s shows that the building has eight bays in each direction (Bibliop, 2017). It appears that the second floor collapsed onto the first floor in all but about 14 square bays: the one on the left and the one in the rear as viewed from the photographer’s viewpoint, so 150 out of 192 floor-bays collapsed, or 78%. Figure 9–13.  Photograph showing collapse of second- and third-floor concrete waffle slabs at Bullock’s retail store after the 1994 Northridge, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   185 Image Metadata and Description for Figure 9–14 Northridge Collection: NR303 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title Oviatt Library, Cal State campus Creator McMullin, Kurt M. Date Location January 20, 1994 Northridge/ California/ North America/ Los Angeles County/ United States Description View of partial roof collapse. South elevation, east of front entry. View from east. Taken at 3 p.m. California State University, Northridge. Structure: Oviatt Library. Author’s Estimate of Affected Area Other photographs, omitted for brevity (Northridge Collection numbers NR299, NR300, and NR302), show about 4×1 bays of roof collapse. One bay generally refers to the space between two columns, so 4×1 bay means a rectangular space between a sequence of five columns in one direction and between two columns in the perpendicular direction. The building has 14 bays east to west and 6 bays north to south and 5 floors, according to California State University Northridge’s web page (California State University Northridge, 2017). Thus, 4×1 bays on 1 story collapsed out of 5 stories, each with 14×6 bays, suggesting a collapsed area of (4×1)/(5×14×6)=1.0% Figure 9–14.  Photograph showing partial roof collapse of Oviatt Library, California State University, after the 1994 Northridge, California, earthquake. 186   The HayWired Earthquake Scenario—Engineering Implications Image Metadata and Description for Figure 9–15 Northridge Collection: NR543 Earthquake date and magnitude (M) Title January 17, 1994; M6.69 Kaiser parking structure Creator Date Location Description Reitherman, January 19, 1994 Los Angeles/ CaliforComplete collapse of parking structure. Robert K. nia/ North America/ Los Angeles, California. Structure: Los Angeles County/ Kaiser Hospital parking. United States Author’s Estimate of Affected Area See also the NISEE e-libraries Northridge Collection, photographs NR519, NR528, NR530, NR539, NR540, NR542, NR544, NR545, NR546, NR549, NR551, NR552, NR543, and NR544. All the photo descriptions ostensibly describe the Kaiser Hospital parking structure, but it appears there were two parking structures. Some descriptions say “complete collapse” and other photos such as NR519, NR528, and NR530 show a parking structure that has not collapsed. Reitherman, in NR549, names the location “Kaiser West Los Angeles Medical Center,” which Google says is located at 6041 Cadillac Avenue, Los Angeles, California, 90034, which Google Earth locates at lat 34.0384 N., long −118.3757 E. Three satellite images from August 1989, April 1994, and March 2002, and shown in Google Earth, reveal two parking structures near here; one with a center near lat 34.0391 N., long −118.3759 E. appears to be the one that did not collapse. Another with a center at lat 34.0389 N., long −118.3733 E. appears in 1989, but is absent in April 1994 (after the earthquake), and it reappears (a replacement) in 2002. I can find no aerial images of the latter collapsed structure or long shots to show the extent of the collapse, so I take the affected area as 100%. Figure 9–15.  Photograph showing complete collapse of the Kaiser parking structure, Los Angeles, after the 1994 Northridge, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   187 Image Metadata and Description for Figure 9–16 Northridge Collection: NR328 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title Soft-story collapse of apartment building Creator Date Unknown 1994 Location Sherman Oaks/ California/ North America/ Los Angeles County/ United States Description Soft-story collapse of apartment building, at Hazeltine Ave. and Milbank St., Sherman Oaks, California. Author’s Estimate of Affected Area By inspection, 33%. Figure 9–16.    Photograph showing soft-story collapse of apartment building in Sherman Oaks after the 1994 Northridge, California, earthquake. 188   The HayWired Earthquake Scenario—Engineering Implications Image Metadata and Description for Figure 9–17 Northridge Collection: NR160 Earthquake date and magnitude (M) January 17, 1994; M6.69 Title Creator Date Soft-story collapse of apartment building Unknown 1994 Location Sherman Oaks/ California/ North America/ Los Angeles County/ United States Description Soft-story collapse of apartment building, at Hazeltine Ave. and Milbank St., Sherman Oaks, California. Author’s Estimate of Affected Area See also the NISEE e-library’s Northridge Collection, photograph NR162. The collapsed second floor amounts to 20% of the building area. The partly collapsed north and south end bays from floors three to five add another 10%, for a total of 30%. Figure 9–17.  Photograph showing second-floor collapse at Kaiser Permanente office building, Granada Hills, after the 1994 Northridge, California, earthquake. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   189 Appendix 10. San Simeon (2003) Collapse Images Image Metadata and Description for Figure 10–1 NISEE Miscellaneous Collection: NM0008 Earthquake date and magnitude (M) December 22, 2003; M6.6 Title View of collapsed building from intersection of 12th and Park Streets Creator Sakai, Junichi Date Location Description December 23, 2003 Paso Robles/ Cali- This unreinforced masonry building was fornia/ North built in 1892, and its clock tower became America/ San a symbol for the town of Paso Robles. The Luis Obispo second story of the building collapsed durCounty/ United ing the earthquake, killing two employees States of Ann's Dress Shop as they tried to flee onto Park Street. The roof of the building collapsed directly westward onto Park Street and landed on a row of parked cars. Debris from the north wall went through the roof of an adjacent shop at 1220 Park Street, Paso Robles, California. Structure: Mastagni Building. Author’s Estimate of Affected Area Also see the NISEE e-library’s Northridge Collection, photographs NM0009 and NM0012 for this building and its photographs NM0001–NM0004 for 1220 Park Street. The building at the west end of the 800 block of 12th Street (807 12th Street is mentioned in the description of NM0009) and the south end of the 800 block of Park Street appears in September 1994 satellite imagery in Google Earth. It has a plan area of approximately 5,960 ft2, so a total area of approximately 11,920 ft2. The collapse of the second floor constitutes 5,960 ft2. In addition, the roof collapsed onto 12th Street. The building was approximately 120 ft long north to south, and it looks as if the roof covered the sidewalk and half the depth of the diagonal street parking, about 19 ft total, so another 120 ft×19 ft=2,280 ft2. The building at 1220 Park Street, just to the north, was a one-story building that appears from photograph NM0009 to have had its roof completely collapse under debris falling from the Mastagni Building. The floor area of 1220 Park Street looks like 50 ft deep by perhaps 20 ft wide. The total affected area is therefore approximately (5,960 ft2 + 2280 ft2 + 1,000 ft2)/(11,920 ft2)=78%. Figure 10–1.  Photograph showing collapsed building from intersection of 12th and Park Streets in Paso Robles after the 2003 San Simeon, California, earthquake. 190   The HayWired Earthquake Scenario—Engineering Implications Image Metadata and Description for Figure 10–2 NISEE Miscellaneous Collection: NM0012 Earthquake date and magnitude (M) December 22, 2003; M6.6 Title Creator Old Clocktower Unknown Date Location December Paso Robles/ California/ 23, 2003 North America/ San Luis Obispo County/ United States Description Before and after images of the Old Clocktower. This unreinforced masonry building was built in 1892, and its clock tower had become a symbol of Paso Robles. The second story of the building collapsed directly westward onto Park Street. Paso Robles, California. Author’s Estimate of Affected Area The author could not find an image that provides a wide enough perspective to judge the affected area. The image is included for completeness. A B Figure 10–2.  Photographs of the Old Clock Tower, Paso Robles, before (A) and after (B) the 2003 San Simeon, California, earthquake. This unreinforced masonry building was built in 1892, and its clock tower had become a symbol of Paso Robles. The second story of the building collapsed directly westward onto Park Street. Chapter M. An Earthquake Urban Search and Rescue Model for Earthquake Response   191 Appendix 11. South Napa (2014) Collapse Images Image Metadata and Description for Figure 11–1 Sarah Durphy: P9050177 (outside) and P9080152 (inside). Earthquake date and magnitude (M) August 24, 2014, M6.0 Title Don Perico’s Restaurant Creator Sarah Durphy Date Unknown Location Napa/ California/ North America/ Napa County/ United States Description Exterior and interior of Don Perico’s Resturant. Napa, California. Author’s Estimate of Affected Area At the time of the earthquake, the restaurant was located at 1025 1st Street, Napa, California, in the west end of the building at lat 38.299029 N., long -122.285868 E. That address seems to occupy approximately 60 ft×60 ft. The collapsed wall appears to fill 25 ft×12 ft, suggesting an affected area of 8.3%. A B Figure 11–1.  Photographs (A, B) showing damage to Don Perico’s Restaurant in Napa in the 2014 South Napa, California, earthquake. 192   The HayWired Earthquake Scenario—Engineering Implications Appendix 12. Earthquakes With No Available Collapse Images Borrego Mountain (1968) Livermore (1980) Mammoth Lakes (1980) Cape Mendocino (1980) Humboldt County (1980) North Palm Springs (1986) Oceanside (1986) Chalfant Valley (1986) Superstition Hills (1987) Lake Elsman (1989) Sierra Madre (1991) Joshua Tree (1992) Cape Mendocino (1992) Landers (1992) Big Bear (1992) Eureka Valley (1993) Hector Mine (1999) Yountville (2000) Parkfield (2004) Anza (2005) Cape Mendocino (2005) Alum Rock (2007) Chino Hills (2008) Inglewood (2009) Eureka (2010) Pico Rivera (2010) El Mayor-Cucapah (2010) Borrego Springs (2010) Brawley swarm (2012) Avalon (2012) The HayWired Earthquake Scenario—Engineering Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q [Also see https://doi.org/10.3133/sir20175013] Chapter N A New Model of Water-Network Resilience, with Application to the HayWired Scenario By Keith A. Porter1 Abstract Introduction Damage to potable water-supply systems can profoundly affect a society after an earthquake. For at least 25 years, engineers have performed computerized risk analyses of earthquake damage to water-supply systems to estimate earthquake damage and restoration. A new stochastic simulation model is offered here that employs a fairly traditional lossestimation approach but with three notable improvements— (1) it deals with lifeline interaction by directly modeling how individual repairs are slowed by limitations in so-called upstream lifelines and other prerequisites; (2) it quantifies damage and restoration over the entire earthquake sequence, that is, considering damage in the mainshock, aftershocks, and afterslip; and (3) it offers an empirical model of service restoration as a function of the number of pipeline repairs performed (as opposed to more rigorous, but computationally demanding, hydraulic analysis). A fourth improvement is that it offers a procedure to adjust estimates of restoration from the Federal Emergency Management Agency’s Hazus-MH computer program to account for an earthquake-sequence’s interactions with lifelines and corrects for Hazus-MH’s default assumptions about the number of available repair crews. The model is applied for two water-supply systems in California’s San Francisco Bay region subjected to the hypothetical but highly realistic HayWired earthquake sequence—a moment magnitude (Mw) 7.0 mainshock on the Hayward Fault in the east bay part of the San Francisco Bay area, plus 16 aftershocks of magnitude 5 or greater occurring over 17 months after the mainshock. The model quantifies water-system damage and restoration, including delays due to fuel and other lifeline limitations, and setbacks in restoration because of aftershocks. It estimates the benefit of a fuel-management plan and an accelerated pipe-replacement plan in terms of accelerated restoration of service. The model is validated several ways for each of the two case-study water-supply systems and seems reasonable. One San Francisco Bay region water utility anticipates using the model to target vulnerable segments of its system for accelerated pipe replacement. The HayWired scenario examines a hypothetical earthquake (mainshock) with a moment magnitude (Mw) of 7.0 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of California’s San Francisco Bay area. The HayWired scenario evaluates, among other things, the potential for damage to water-supply systems in the San Francisco Bay region during the HayWired mainshock and subsequent aftershocks. University of Colorado Boulder. 1 How Water Supply is Important in an Earthquake People need potable water for daily life. Businesses need water for air conditioning and as a raw material for production. Water is an input to many natural and manufactured products and processes. Damage to a water-supply system can contribute greatly to the life-safety and economic consequences of an earthquake, as illustrated by the economic analyses performed for the 2008 ShakeOut scenario (Rose and others, 2011). In that study, the authors found that water-supply interruption from a hypothetical Mw 7.8 earthquake on the San Andreas Fault in southern California could result in $24 billion in business interruption losses when macroeconomic responses are not considered and only minimal business-resilience actions are taken to reduce the losses. The figure that represents more than one third of the $68 billion in total business interruption losses and 13 percent of the total of property damage plus business interruption. A potable water supply is crucial for residences, businesses, government facilities, and hospitals and other critical-care facilities. Long aware of the importance of water supply and the potential for earthquakes to interrupt water supply, earthquake experts recommend that homes and businesses have enough water to provide for 1 gallon per person per day after a major earthquake to last at least 3 days and ideally for 2 weeks. Loss of water supply in the hypothetical ShakeOut earthquake would also contribute substantially to the fire damage to property, which itself could realistically account for $65 billion of the $113 billion in property losses (Scawthorn, 2008). The ShakeOut scenario was not a worst-case earthquake; the earthquake fault rupture it dealt with has a mean recurrence interval 196   The HayWired Earthquake Scenario—Engineering Implications of 150 years, and it has been 300 years since the last rupture. Furthermore, the fire simulation assumed mild winds rather than the fast, hot, dry, Santa Ana winds that commonly blow in the fall and notoriously fan wildfires. These earlier estimates, although particular to the ShakeOut, reflect a general truth—earthquake damage to water-supply systems in the United States (and elsewhere) threatens the health, safety, and welfare of the population, possibly more than earthquake damage to any other utility or other element of the built environment, in part because repairs are so costly and time consuming. More narrowly, earthquakes can pose a serious financial threat to any water supply utility in a seismically active region. If a utility cannot deliver water it cannot collect revenues, which can threaten its ability to make payroll. Every water utility in earthquake country may be at risk. The HayWired scenario hypothetical mainshock is a large but not exceedingly rare Mw 7.0 earthquake that damages water-supply systems in the San Francisco Bay region. Earthquakes damage water-supply systems, and the damage causes other problems, such as for firefighting. The Mw 7.8 1906 San Francisco earthquake damaged so much of that city’s potable water-supply system that pressure dropped too low for firefighters to fight the fires that eventually destroyed much of the city. The Mw 6.9 1989 Loma Prieta earthquake caused at least 761 breaks and leaks to water mains and pipelines made of various materials throughout the San Francisco Bay area (Lund and Schiff, 1991). The loss of firefighting water supply in San Francisco’s Marina District contributed to the fire that damaged 7 structures, destroying 4 buildings containing 33 apartments and flats (Scawthorn and others, 1991). Cast iron, steel, ductile iron, plastic, and copper pipes all broke both within and outside areas of liquefaction and other ground failure. The Mw 6.0 2014 South Napa earthquake caused 249 pipeline breaks or leaks in the City of Napa, in the northern San Francisco Bay area (Douglas DeMaster, Engineer, City of Napa, written commun., March 23, 2017). The largest total number of breaks and leaks and the highest repair rate (repairs per mile) in the 1989 earthquake occurred in cast-iron pipe subjected to liquefaction-induced ground failure, but other materials were also damaged, including ductile iron, polyvinyl chloride (PVC), and steel. Pipes were damaged in 1989 in places that were not known to have experienced ground failure, so that damage has been attributed to ground strain associated with wave passage, especially Rayleigh surface waves. There was no observed liquefaction damage to buried pipeline in Napa in the 2014 earthquake, reinforcing the idea that wave passage alone can damage buried pipe. Even the modest Mw 4.0 Piedmont, California, earthquake of August 17, 2015, caused at least 7 instances of damage to buried cast iron water-supply pipe in the east bay (Bay City News, 2015). Repairs to an earthquake-damaged water-supply system can take months or more. Each repair can take as little as 2 hours to repair, but large numbers of repairs and larger pipes can take much longer. A 30-inch water main that broke near the UCLA campus at 3:30 p.m. on Tuesday July 29, 2014, took almost 5 days, until 11:00 a.m. Sunday, August 3 to repair (Los Angeles Department of Water and Power, 2014). During an earthquake sequence, with many simultaneous instances of damage, repairs take longer for many reasons. Some of these are: 1. When a pressure zone loses pressure because of many breaks and leaks, it can be necessary to repair damage closer to the source (that is, nearer the tank, reservoir, or other water source) before one discovers damage farther from the source. 2. Similarly, it may be necessary to repair damage to a pumping plant, reservoir, or regulator before damage in the downstream pipeline network can be addressed. 3. Water districts have an upper limit to their ability to field and manage multiple repair crews operating in parallel, even when the crews are from outside contractors or from water districts that provide mutual aid. 4. Limited supplies of repair resources such as spare pipe, clamps, fuel, and repair crews. 5. Damage to other systems—for example, electrical and gas—can hinder pipeline repair, and in some cases those repairs can cause pipeline damage. Coordination with other agencies can conceivably idle repair crews. 6. Aftershocks can hinder repair efforts because they pose an ongoing safety threat to repair crews. They can also cause new damage or aggravate earlier damage. Study Objectives In this work, I attempt to depict a realistic outcome of the damage and restoration of water supply in the HayWired earthquake sequence. I review available models of earthquakeinduced pipeline damage and repair, propose one for use in the HayWired scenario, and apply it to the water-supply systems of two San Francisco Bay area water utilities—the San Jose Water Company and the East Bay Municipal Utility District (EBMUD). These two systems were chosen because they are strongly shaken in the scenario, are affected by the mainshock and by aftershocks, and were willing to share their system maps. The maps were shared under strict requirements of confidentiality, so map details are not available to the reader. This study supplements conventional loss estimation by examining the detailed activities involved in discovering and repairing water-pipeline damage. It identifies steps in the repair process that rely on other lifelines to inform a new model of the effects of lifeline interaction to delay water service repairs and restoration. This study focuses on damage and repair of buried water pipe, which tends to dominate the effort to restore water supply. It considers damage resulting from wave passage, liquefaction, landsliding, and fault offset. It ignores earthquake damage to other elements in the water-supply system, including raw-water aqueducts, tanks, tunnels, canals, valves, and reservoirs. The decision to focus this study on buried pipelines without including Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 197 other critical facilities, such as tanks, reservoirs, and tunnels, seems reasonable, because a majority of water utilities have implemented seismic improvement programs (SIP) that, for the most part, focused on seismically retrofitting their tanks, reservoirs, and other such facilities but not their old distribution pipelines. As such, old distribution pipelines, as an asset class, present the most significant seismic vulnerability for most water utilities, because for the most part smaller diameter distribution mains were not replaced with seismic-resistant mains because it simply wasn’t economically feasible to replace them all as part of a SIP. Elsewhere, the HayWired scenario quantifies damage to buildings; this chapter does not address restoration of water utilities’ customer base or the change in demand for water as homes and businesses relocate because of building damage or other reasons. Literature Review Before proposing a model to estimate water-supply pipeline restoration considering an earthquake sequence and lifeline interaction, I first consider some key aspects of previous efforts. At least two general approaches exist to estimating damage and restoration of water supply after earthquakes—(1) expert opinion and (2) engineering analysis. The present work will pursue an analytical approach, which requires one to consider some important details—pipe damageability, repair effort, postearthquake serviceability, lifeline interaction, afterslip, and measuring loss of resilience. so panelists from each panel actually reviewed and revised the write-ups. However, the panel process worked reasonably well. Panelists were well qualified and seemed to fairly assess realistic earthquake impacts and restoration. They gained insight into lifeline interaction, mutual-aid needs, communication capabilities, and backup supplies. Figure 1 shows the water-supply restoration timeline that the water-supply panel estimated for strongly shaken (Modified Mercalli Intensity, MMI, VIII+) geographic areas using expert opinion (Jones and others, 2008). See Porter and Sherrill (2011) for electric power restoration curves in ShakeOut and Porter and others (2011) for various restoration curves and modes of lifeline interaction in the ARkStorm scenario. Analytical Approaches to Estimating Water Supply Impacts Analytical approaches to estimating impacts to watersupply from an earthquake typically involve acquiring a map of a water-supply system, identifying component materials and sizes, associating each with one or more vulnerability functions or fragility functions (depending on the desired output), estimating ground motion and ground-failure severity in one or more scenarios, estimating mean damage and sometimes uncertainty in damage with reference to the vulnerability functions, and sometimes estimating repair costs and duration of loss of function. The Federal Emergency Management Agency’s (FEMA) publication FEMA 224 (Applied Technology Council, 1991), 100 The ShakeOut scenario (Jones and others, 2008) assessed Earth-science impacts, physical damage, and socioeconomic impacts of a hypothetical Mw 7.8 southern San Andreas Fault earthquake. Among many detailed studies were special studies of 12 lifelines, 7 of which were performed by panels of employees of the utilities at risk. The panel process is described in detail in Porter and Sherrill (2011). Briefly, panels meet for several hours (generally 4 hours in the case of ShakeOut). Panelists are presented with a scenario’s Earth-science impacts and previously estimated damage to supposedly upstream lifelines—lifelines whose damage would seem to affect the damage or repair to the lifeline in question but not vice versa. They then hypothesize a realistic outcome of the earthquake on damage and service restoration, identifying research needs and mitigation options. Panels’ discussion and initial findings are documented in brief memos, which are then circulated to the panelists. Panelists are asked to review the memos and asked to reconsider lifeline interaction in light of damage to supposedly downstream lifelines, as well as upstream ones. The process iterates until panelists are satisfied with their estimates of damage and restoration. In practice in the ShakeOut, as well as in ARkStorm (see Porter and others, 2011), only one iteration was used and only two or Water customers receiving service, in percent A Panel Approach to Estimating Water-Supply Impacts 75 50 25 0 0 30 60 90 120 150 180 Time after earthquake, in days Figure 1.  Graph showing water-restoration timeline in the ShakeOut scenario—a hypothetical magnitude-7.8 earthquake on the southern San Andreas Fault—for water customers in areas that experienced shaking of Modified Mercalli Intensity VIII or higher (modified from Jones and others, 2008). 198   The HayWired Earthquake Scenario—Engineering Implications Scawthorn and others (1992), Hazus-MH (Federal Emergency Management Agency, 2012), the Mid-America Earthquake Center’s seismic loss estimation system MAEViz (Mid-America Earthquake Center, 2006), and Marconi (Prashar and others, 2012) all use such an approach. The last three implement their methodologies in software, as do many others. In the case of Hazus-MH, the software assumes that a water main exists under each street, 80 percent of pipes are brittle (such as cast iron), and 20 percent of pipes are ductile (such as ductile iron). MAEViz and Marconi allow the user to specify the location and characteristics of each pipe segment. Neither Hazus-MH, MAEViz, nor Marconi performs hydraulic analysis. MAEViz and Marconi estimate damage. Hazus-MH estimates damage and estimates repair costs and system restoration time using methods described later. Khater and Grigoriu (1989) describe an analytical model of water-supply damage and serviceability that does perform hydraulic analysis. Coded in software called GISALLE, it involves three tasks—(1) generate damage states for water-system components consistent with the seismic intensity at the site, (2) perform hydraulic analysis for simulated damage state of the system, and (3) develop statistics on the available water flow for postulated levels of seismic intensity. Some of the available software, such as MAEViz and Urban Infrastructure and Lifelines Interactions of Systems (UILLIS) (Javanbarg and Scawthorn, 2012), have the ability to treat lifeline interaction—how damage or loss of function in one lifeline system affects the functionality or restoration of another. For example, loss of power and limitations in fuel supply can affect the functionality of a water-supply system or delay repairs. These programs use a system-of-systems approach to modeling the lifelines. That is, they model two or more lifelines in the same framework, relating the condition of an element of one lifeline to the condition of an element in another. Damageability of Buried Pipe The first step of a water-network resilience analysis once a ground-motion map has been developed is to estimate damage. Many authors have written extensively about the damageability of buried pipe, only some of this work is discussed here. Vulnerability and Fragility Functions As used here, a vulnerability function relates the degree of damage—in this case, number of breaks or leaks per unit length of pipeline—as a function of the degree of environmental excitation such as peak ground velocity (PGV). A fragility function by contrast measures the probability of reaching or exceeding some undesirable state conditioned on the degree of environmental excitation. The terminology is not universal but will be consistently applied here. In the present context, vulnerability functions are most useful for estimating the number of breaks and leaks in a pipeline network subjected to ground shaking (usually referred to as wave passage in the pipeline literature), landsliding, and liquefaction. However, at a fault crossing, a fragility function is more useful— here, I am interested in the probability that a pipeline requires repair at the point where it crosses the fault, as a function of the fault offset and possibly as a function of the angle at which the pipeline crosses the fault. Both vulnerability functions and fragility functions are commonly conditioned on the pipeline’s engineering attributes, such as material, diameter, connections at joints, and sometimes soil conditions. Hazus-MH, O’Rourke and Ayala (1993), and Honneger and Eguchi (1992) The Hazus-MH computer software (Federal Emergency Management Agency, 2012) currently uses a vulnerability function for pipeline subjected to wave passage by O’Rourke and Ayala (1993) and one for pipe in liquefied soil from Honegger and Eguchi (1992). The median rates of repairs per kilometer of pipeline for these two relations are given by equations 1 and 2, respectively: Rˆ = 0.0001 × K × PGV 2.25 , × PGD 0.56 , R = PL × K   (1) (2) where PL denotes the probability of liquefaction; K=1.0 for asbestos cement, concrete, and cast-iron pipe; K=0.3 for steel, ductile iron, and PVC; PGV denotes peak ground velocity measured in centimeters per second (cm/s); and PGD denotes permanent ground deformation—the absolute distance a point on the ground permanently moves due to landsliding, fault offset, or liquefaction-induced ground failure—measured in inches. Equation 1 draws on a number of observed repairs in asbestos cement, concrete, cast iron, and prestressed concrete pipe, with diameters between 3 and 72 inches, in four U.S. and two Mexican earthquakes experiencing ground motion of as much as 50 cm/s of peak ground velocity. (O’Rourke and Ayala, 1993, do not publish the number of repairs or the lengths of pipe.) O’Rourke and Ayala’s (1993) data for equation 1 imply a coefficient of variation in the ratio of observed to estimated repair rate of 0.76 and a ratio of mean repair rate to median repair rate of 1.22. The work by Honneger and Eguchi (1992) used to develop equation 2 reflects an unknown quantity of pipe and number of breaks and leaks. Their data mostly come from four earthquakes—1923 Kanto (Japan), 1971 San Fernando (California), 1976 Tangshan (China), and 1985 Michoacán (Mexico). Pipe diameters range from 4 inches to 48 inches. Materials included cast iron, concrete, precast concrete, and steel. Eidinger (2001) More recently, Eidinger (2001) proposed two vulnerability functions—one for wave passage (that is, ground shaking absent liquefaction) and one for permanent ground deformation (that is, in the presence of liquefaction or landslide-induced ground Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 199 displacement). Equations 3 and 5 present Eidinger’s recommended vulnerability functions. In the equations, Rw (PGV, p) and Rl(PGD, p) denote repair rate per 1,000 linear feet of pipe associated with nonexceedance probability p, as a result of wave passage and liquefaction, respectively. For example, the median repair rate is estimated using p=0.5. PGV refers to geometric mean horizontal peak ground velocity in inches per second, PGD denotes permanent ground displacement relative to pre-earthquake location, measured in inches, and Φ-1(p) denotes the inverse standard normal cumulative distribution function evaluated at p. For the reader who is unfamiliar with probability distributions, the standard normal distribution is the bell-shaped curve that represents how likely are various possible values of an uncertain quantity. Uncertain or random variables can take on a variety of probability distributions; the standard normal distribution is one of many. It has a peak (the expected or mean value and also the value with 50-percent probability of not being exceeded, called the median) at 0. Its standard deviation (a measure of how wide the bell is, and therefore how uncertain is the random quantity) is 1.0. Its cumulative distribution function is an S-shaped curve that tells the probability that a sample of a quantity with a standard normal distribution takes on a value less than or equal to any given quantity between ˗∞ and ∞. The inverse of the standard normal cumulative distribution function is the value of the uncertain quantity that has a specified probability of not being exceeded. Most statistics textbooks provide more information about probability distributions (see, for example, Ang and Tang, 1975, or Benjamin and Cornell, 1970). The quantities K1 and K2 are factors to account for pipe material, joints, soil corrosivity, and pipe diameter—either small (4 to 12 inches diameter) or large (16-inch diameter or greater). See table 1 for their values. Eidinger (2001) does not provide values for some combinations, so they appear blank in the table. The authors acknowledge that permanent ground displacement produces damage rates two orders of magnitude greater than wave passage and that damage rate in areas with ground failure is fairly insensitive to PGD. The terms exp(β×Φ-1(p)) in equations 3 and 5 reflect that the equations treat the repair rate as uncertain and lognormally distributed conditioned on the value of PGV or PGD. (Lognormal is like normal, except that the natural logarithm of the uncertain quantity in question is normally distributed. A lognormal variable can take on any positive value but not zero or a negative number. The bell shape is skewed to the right.) Setting p to 0.5 sets the exp term to 1.0 and makes R(p) produce the median (not the mean) repair rate. The mean repair rate would be substantially higher than the median. Equations 4 and 6 provide the mean (average) repair rate, given Eidinger’s values of β shown in equations 3 and 5 and Eidinger’s assumption of lognormality. The interested reader who is unfamiliar with lognormally distributed variables can refer to any of several common textbooks (see, for example, Ang and Tang, 1975). The interested reader who is unfamiliar with vulnerability functions can refer to Porter (2017a) for a short primer. Equation 3 gives Eidinger’s (2001) pipe vulnerability function for wave passage, drawn from 81 sources reporting 3,350 repairs recorded in 12 earthquakes. The plurality of data come from the Mw 6.7 1994 Northridge (California) earthquake. The data reflect 38 data points for damage to cast iron, 13 data points for damage to steel, 10 data points for damage to asbestos Table 1.  Eidinger (2001) pipe-vulnerability equation factors K1 and K2, which account for pipe material, joints, soil corrosivity, and pipe diameter. ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Pipe material Cast iron Cast iron Cast iron Cast iron Cast iron Welded steel Welded steel Welded steel Welded steel Welded steel Welded steel Welded steel Asbestos cement Asbestos cement Concrete with steel cylinder Concrete with steel cylinder Concrete with steel cylinder PVC Ductile iron Joint type Cement Cement Cement Rubber gasket Mechanical restrained Lap-arc welded Lap-arc welded Lap-arc welded Lap-arc welded Rubber gasket Screwed Riveted Rubber gasket Cement Lap-arc welded Cement Rubber gasket Rubber gasket Rubber gasket Assumed here because no K-value is offered by the source. 1 Soils Diameter All Corrosive Noncorrosive All All All Corrosive Noncorrosive All All All All All All All All All All All Small Small Small Small Small Small Small Small Large Small Small Small Small Small Large Large Large Small Small K1 K2 1.0 1.4 0.7 0.8 0.71 0.6 0.9 0.3 0.15 0.7 1.3 1.3 0.5 1.0 0.7 1.0 0.8 0.5 0.5 1.0 1.0 1.0 0.8 0.7 0.15 0.15 0.15 0.15 0.7 1.31 1.31 0.8 1.0 0.6 1.0 0.7 0.8 0.5 200   The HayWired Earthquake Scenario—Engineering Implications cement, 9 data points for damage to ductile iron, and 2 data points for damage to concrete. Data reflect PGV values between 2 and 52 cm/s. Rw ( PGV , p) = K1 × 0.00187 × PGV × exp (1.15 × Φ −1 ( p)), (3) Rw ( PGV ) = K1 × 0.003623 × PGV .  (4) cast-iron pipes subjected to wave passage. They draw on data about 2,051 repairs in 3,400 kilometers (km) of pipe in the Mw 6.2 February 22, 2011, Christchurch, New Zealand, earthquake and the Mw 6.0 June 13, 2011, Christchurch earthquake. The majority of pipe length in the database was asbestos cement, but the data also included cast iron, PVC, modified PVC, and unnamed other materials. The data were drawn from locations with PGV between 10 and 80 cm/s. Their vulnerability functions are given by equations 9 and 10: Equations 5 and 6 give Eidinger’s (2001) pipe vulnerability function for permanent ground deformation, drawn from 42 data points from 4 earthquakes between the Mw 7.8 1906 San Francisco log10 ( RAC ) = 2.83 × log10 (GMPGV ) − 5 , (9) earthquake and the Mw 6.7 1989 Loma Prieta earthquake. The plurality of data points come from the Mw 7.8 1983 Nihonkailog10 ( RCI ) = 2.38 × log10 (GMPGV ) − 4.52 ,   (10) Chubu (Japan) earthquake. The plurality of pipe material is asbestos cement (20 data points) followed by cast iron (17 data where RAC denotes the median repairs per kilometer of asbestos points), and a mixture of cast iron and steel—presumably meaning cement pipe, RCI is the analogous value for cast-iron pipe, and that the material was one or the other, but it is not known which according to the authors, “GMPGV is the mean of the natural (5 data points). None of the data appear to reflect ductile iron. logs of the two maximum horizontal peak ground velocity They reflect PGD values between 0 and 110 inches. In these two (PGV) values taken from ground motion recordings available equations, Rl(PGD, p) denotes the liquefaction-induced damage from GNS Science . . . at each station.” Despite that definition rate associated with nonexceedance probability p, and R l ( PGD ) of GMPGV, the authors seem actually to mean the geometric denotes the expected value of the liquefaction-induced damage mean of the peak ground velocity values in centimeters per rate: second of the two horizontal orthogonal components. (The inverse of the natural logarithm of the mean of the natural Rl ( PGD, p ) = K 2 × 1.06 × PGD 0.319 × exp ( 0.74 × Φ −1 ( p )) , (5) logarithms of two quantities equals their geometric mean.) They offer vulnerability functions for pipe subjected to liquefaction, Rl ( PGD ) = K 2 × 1.39 × PGD 0.319 . (6) where the ground deformation is measured in terms of (1) the larger principal component of ground strain in the horizontal Eidinger (2001) also proposed models for damage to pipe plane and (2) the rotation of the axis of the pipe about a that crosses an earthquake fault—one for continuous pipelines horizontal axis normal to the axis of the pipe, which the authors (equation 7) and one for segmented pipe (equation 8). In the call angular distortion—essentially a differential permanent equations, PGD denotes mean offset (in inches, in.) over the vertical displacement of two points on the pipe axis, divided by entire length of the fault, presumably at the fault trace rather the distance between the two points. than averaged over the area of the fault, and presumably considering coseismic slip and afterslip. In the equations, P O’Rourke (2003) denotes the probability that the pipe crossing the fault will break: There does not appear to exist any empirical relation between fault offset and probability of pipeline damage. A few PGD P = 0.70 × authors offer analytical formulations between offset and stress 60 in. , (7) or strain in a pipeline that crosses a fault. O’Rourke (2003) ≤ 0.95 summarizes some of these, considering under two conditions that depend on the geometry of the pipeline at the fault crossing—(1) a combination of bending and axial tension and (2) a combination P=0 PGD < 1 in. of bending and axial compression. For the former, he illustrates = 0.5 1 in. ≤ PGD ≤ 12 in. a relation between tolerable fault offset as a function of distance . (8) between points at which the pipeline is anchored on either side = 0.8 12 in. < PGD ≤ 24 in. of the fault (which he refers to as “unanchor length”; see fig. 2A) = 0.95 24 in. < PGD and the angle β subtended by the fault and the pipeline, in which the offset puts the pipeline in tension. Figure 2A is merely an illustration for a particular pipe material and diameter. He offers O’Rourke and others (2014) a second analytical relation (fig. 2B) for segmented pipe subject O’Rourke and others (2014) offer vulnerability functions to fault offset, again for fault-crossing geometry where offset puts for the median repair rate per kilometer of asbestos cement or the pipe into tension. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 201 Unanchor length, in meters 60 25 120 180 240 300 A β = 60° β = 45° 20 Fault offset, in feet 420 EXPLANATION Hc= 3 ft (0.9 m) εa= 3 ft (0.9 m) β = 30° β = 15° 480 540 γ = 110 pcf (1,760 kg/m3) Φ <34° Φp <20° Hc= 10 ft (3 m) εa= 3.5% 15 6.0 4.5 β = 60° β = 45° 10 0 0 200 400 600 Inches 16 300 X – 60 D = 3.5 ft (107 cm) t = 0.56 in (1.43 cm) δv = 1/4δh 800 1,000 1,400 1,600 1,800 Extra long restrained couplings 12 200 Failure 8 Failure depends on lead joint location Mechanical joints 4 100 0 1,200 1.5 Unanchor length, in feet B Millimeters 400 3.0 β = 30° β = 15° 5 Fault displacement 360 Fault offset, in meters 0 0 Lead joints 10 20 30 40 50 60 70 80 90 Angle of pipe-fault intersection, β, in degrees Figure 2.  Graphs showing (A) tolerable fault offset versus unanchor length in continuous pipe (O’Rourke 2003, citing Kennedy and others, 1977) and (B) tolerable fault offset versus pipe-fault intersection angle in segmented pipe (O’Rourke 2003, citing O’Rourke and Trautmann, 1980). (Images modified from O’Rourke 2003.) mm, millimeters; cm, centimeters; m, meters; in, inches; ft, feet; %, percent; kg/m3, kilograms per cubic meter; X−60, grade of pipe material; εa, maximum axial strain due to the elongation of the pipe induced by the fault offset; β, angle at which the pipeline intersects a right lateral strike-slip fault; γ, effective unit weight of soil; D, pipe diameter; t, pipe wall thickness; δV, vertical component of fault offset; δh, horizontal component of fault offset; Hc, burial depth from ground surface to top of pipe; Φ, soil internal friction angle; Φp, contact friction angle. 202   The HayWired Earthquake Scenario—Engineering Implications • Burial depth—Among 67 records with reported burial depth, the average was 4.0 feet (ft) and the standard deviation was 2.2 ft. Tasks and Methods to Repair Leaks and Breaks The City of Winnipeg (2014) in Canada offers a list of tasks to repair a water main break or leak, written for the general public. The tasks are shown in chronological order in the lefthand column of table 2. The task list is generally consistent with a more detailed checklist created by the American Water Works Association (2009), although it omits lists of tools, equipment, disinfecting chemicals, documentation, and testing materials. Column 2 of table 2 lists my interpretation of rate-limiting factors, that is, prerequisites for each task. The rate-limiting factors are mostly potential impacts from other lifelines, that is, lifeline interactions. If they are unavailable, repairs cannot proceed or they proceed more slowly—that is, their rate is limited. These items include communications, electricity, fuel, site safety (that is, no fire or hazardous material release), roadway access, repair crews, and repair supplies (replacement pipe, replacement fittings, clamps, and paving materials). Regarding crew availability, public and private water agencies plan to provide mutual assistance for emergencies (see, for example, California Water/Wastewater Agency Response Network, CalWARN, 2009). Crews may have to travel from great distances, hundreds of miles or more, so their availability can change over time. Table 2 probably omits tasks that are unnecessary or trivial for day-to-day repairs but become significant in a large earthquake. For example, a water agency may also have to arrange repair contracts with contractors, track and prioritize repairs, and manage an unusually large number of repair crews operating simultaneously. Lund and Schiff (1991) surveyed operators of pipeline utilities, asking them to provide detailed information about each pipeline failure they repaired after the 1989 Loma Prieta earthquake (see fig. 3 for the survey form). The resulting database includes information about 862 pipeline failures among 65 water, sewer, drainage, and gas agencies. The data may be useful for estimating repair times, so I extracted the following statistics from the database. • Break or leak—Among the failures where the respondent indicated whether the failure was a break or a leak, it was more common for the pipeline to break (336 failures) than to leak (140 failures). • Pipe failure modes—Among pipe failures, the most common were circumferential cracks (99), followed by splits (43) and corrosion (33). Only one blowout was reported. • Joint failure modes—Almost as common as pipe failures were joint failures (102 pulled, 29 cracks at joints, 25 gasket failures, and 12 other joint failures). • Fitting failure modes—There were a variety of fitting failures (57 threaded couplings, 9 elbows, 6 offsets, 4 hydrants, 3 tees, and 45 miscellaneous other fitting failures). • Repair methods—The most common repair method was to replace the damaged element (185 replacements), which was more than twice the number of clamps installed (77), followed by mechanical couplings (50), epoxy glue (16), and miscellaneous others such as flex couplings and pressure grout. Time to Repair Pipe Leaks and Breaks To repair damaged water-supply pipe, a repair crew must locate the damage, usually eliminate pressure in the pipe by closing an upstream valve, excavate the damaged element (usually with a backhoe), perform the repair, reopen the upstream valve, backfill the excavation, and repave any driving surface over the Table 2.  Water-pipeline repair tasks (City of Winnipeg, 2014), with an interpretation of rate-limiting factors for repairs after an earthquake. Tasks Rate-limiting factors Receive a notice from our 311 Centre about a water main break. Communications, electricity Dispatch a crew to the location. Fuel, site safety (for example, no fire), roadway access, crew availability Control the leak to reduce the risk to public safety, and private and public property. We do this by finding and closing valves. Contact other utilities to make sure that we can dig without damaging other services or endangering staff or the public. Communications Pinpoint the location of the leak using an electronic leak detector. Dig down to the water main and confirm the cause of the leak. Fuel Repair the water main. Depending on the type of break, we may apply a repair clamp or replace a length of pipe. Pipe, fitting, or repair hardware such as clamps Open valves to turn the water main back on, flush the water main and sample water quality. Backfill to temporarily restore the excavated area. Fuel Permanently restore the sod or pavement in the excavated area. Pavement material Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 203 Figure 3.  Image of Lund and Schiff (1991) survey form given to operators of pipeline utilities, asking them to provide detailed information about each pipeline failure they repaired after the moment-magnitude-6.7 1989 Loma Prieta, California, earthquake. location of the repair. Pipe damage can be repaired by replacing the damaged element, by welding over the crack, or by installing repair hardware—generally either a clamp that is mechanically secured over the damage or a closure ring called a butt strap that is welded to the outside of the pipe over the damage. The time required to perform the repair depends on several issues: • How long it takes people to report the damage to the utility or otherwise for the utility to become aware of and locate the damage, which itself depends on power and communication; • Site accessibility; • Availability of crews and equipment; • Availability of fuel and consumable repair material; • Pipe burial depth; • Groundwater presence and depth; • Diameter, material, and jointing of the pipe; • Impact on water flow (break or leak); • Nature of the damaged element—whether to pipe, joint, or fitting; • If pipe, whether circumferential crack, longitudinal split, corrosion, or other such failure; • If joint, whether a crack, pull-out, compression failure, gasket failure, or other such failure; and • If fitting, the nature of the fitting (such as elbow, tee, cross, offset). Schiff (1988) offers repair times for 21 individual water-pipe repairs after the Mw 5.9 1987 Whittier Narrows earthquake in 204   The HayWired Earthquake Scenario—Engineering Implications southern California, mostly of cracks and breaks in 4- to 8-inch steel and cast-iron mains. Repair times were reported by the City of Whittier water distribution superintendent. Times varied between 1 and 16 hours, as shown in table 3. Schiff reports that water pressure in Whittier dropped to 50 pounds per square inch (psi) from the normal 80 to 100 psi as a result of 40 repairs in 133 miles of pipe (or 0.06 repairs per 1,000 linear feet of pipe). EBMUD reports on its mutual assistance to the City of Napa after the August 24, 2014, magnitude (M) 6.0 South Napa earthquake (East Bay Municipal Utility District, 2014). EBMUD crews performed 56 repairs in approximately 252 crew-hours, for an average duration of 4.5 hours per repair. It should be noted that this average duration for completing repairs does not reflect the time it took for the City of Napa or its contractors to complete the excavation and backfill (EBMUD crews focused on repair work and did not complete excavation/backfill/paving-related work). Tabucchi and others (2010) elicited opinions from personnel at the Los Angeles Department of Water and Power (LADWP) on water-pipe repair productivity. They propose a model with triangular probability distributions for each of several repair operations. Each distribution is characterized by a minimum value (the left end of the triangle), a modal value (the peak of the triangle, which is the most likely value), and a maximum value (the right end of the triangle). Table 4 repeats LADWP’s estimates. Distribution-system leak and break repairs are estimated to require no less than 3 hours and no more than 12 hours with modes of 4 to 6 hours. Hazus-MH (Federal Emergency Management Agency, 2012) employs four restoration times for pipe repair—two each for large and small diameter pipes (20-inch diameter and above is large, 12-inch diameter or less is small) times two to distinguish between breaks and leaks (see table 5). Table 3.  Repair times for water-supply pipeline damage in the moment magnitude (Mw) 5.9 1987 Whittier Narrows Earthquake and Mw 5.2 aftershock in southern California (modified from Schiff, 1988). [CI, concrete insert, RC, reinforced concrete; --, no data] 11 12 13 14 15 Pipe diameter, Estimated repair time, Installation in inches (and Type failure and comments in hours date material) Mainshock La Cuarta St. and Whittier Blvd. 3–4 4 (CI) 1920 Blowout—3–4-foot section Citrus Ave. at Beverly Dr. 16 (down 3–4 days) 6 1932 Circumference crack 11741 S. Circle Dr. 3–4 4 (CI) 1929 Circumference crack Bronte Dr. at Bacon Rd. 4–5 6 (CI) 1956 Blowout—Pressure hole Beverly Blvd. (Citris and Pick) 12 24 (RC) 1930 Beam crack Painter Ave. at Broadway Ave. ---Leaked surface from Painter Ave. and Beverly Blvd. and again after the aftershock Dorland St. at Magnolia Ave. 5 6 (CI) 1938 Circumference crack Painter Ave. at Sunset 1 3/4 (steel) Old Greenleaf Ave. at Orange Dr. 16 --Leak surfaced likely from Orange Dr. and Friends Ave., (see no. 10) Orange Dr. at Friends Ave. 16 16 (RC) 1930 Circumference crack—Leak entered abandoned, uncapped steel pipe 13502 Beverly Blvd. 4 6 (CI) 1927 Joint pullout—Likely a flair joint 8041 Michigan Ave. 3–4 4 (steel) Very old Blowout—Hole developed 12101 Rideout Way 2–3 2 (steel) Old Blowout—2–3-foot section service line South Circle Dr. at North Circle Dr. 4–6 6 CI 1929 Circumference crack Panorama Dr. above Orange Dr. 20 24 1967 Leaks at caulked collars 16 17 18 19 20 21 22 23 24 11630 Whittier Blvd. 8053 Michigan Ave. Near no.17 5630 Omelia Rd. Painter Ave. and Beverly Blvd. 14245 Bronte Greenleaf Booster Station Near 14245 Bronte 11630 Whittier Blvd. 25 12906 Orange Dr. No. 1 2 3 4 5 6 7 8 9 10 Location 8 6 (with no. 18) See no.17 8 6–8 8 5 6 1 Aftershock 6×8 (CI) 4 (steel) 4 (steel) 8 (CI) 6×8 (CI) -16 (CI) 6 (CI) 6×8 (CI) 1 (steel) -Very Old Very Old 1938 1935 1948 1930 1948 See no. 16 -- Shear—T-sheared at flange Blowout—Hole a few feet away Blowout—Same as no. 17 Cracked bell T-sheared at flange Service cork pulled from main Lead caulk forced out of bell; leak Blowout This may be from soil settlement associated with no. 16 Corroded—Service line split Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 205 Table 4.  Los Angeles Department of Water and Power (LADWP) water-pipe repair productivity estimates (modified from Tabucchi and others, 2010). [MWD, Metropolitan Water District; hr, hour; km, kilometer] Event Minimum Mode Maximum 0.5 hr 1 hr 1 hr 1 hr 2 hr 0.5 hr 1 hr 1 hr 1 hr 2 hr 1 hr 2 hr 2 hr 2 hr 3 hr 3–6 hr 3 hr 3–4 hr 4–6 hr 1–2 hr 1 hr 6–12 hr 4 hr 6 hr 6–8 hr 2–3 days 2 hr 8–24 hr 8 hr 8–12 hr 8–12 hr 3–4 days 4 hr 3 hr 4 hr 4 days 6 days D/25 hr 4 hr 6 hr 4 days 8 days D/40 hr 6 hr 12 hr 6 days 10 days D/80 hr Inspect a: Trunk or distribution damage location Pump station Regulator station Tank Small reservoir Rerouting operation on a trunk line by: Truck redundancy (major)1 Trunk redundancy (minor)1 Connecting to MWD1 Connecting to well1 Using a fire truck1 Isolate distribution damage at one demand node Repair a: Distribution leak Distribution break Trunk leak Trunk break Travel a distance D (km) 1 Major trunk lines are the pipelines that are the sources for each of the 13 LADWP subsystems; minor trunk lines are the remaining ones. Task duration for major trunk line rerouting operations vary by specific trunk line, as listed in Tabucchi and Davidson (2008). Table 5.  Hazus-MH estimates of repair time per pipe repair as a result of earthquake (modified from Federal Emergency Management Agency, 2012). Diameter from, in inches Diameter to, in inches 60 36 20 12 8 Unknown diameter or for default data analysis 300 60 36 20 12 Unknown diameter or for default data analysis Number of fixed breaks per day per worker 0.33 0.33 0.33 0.50 0.50 0.50 Seligson and others (1991) offer an empirical relation for time required to restore water service as a function of number of pipeline repairs per square mile, based on evidence from two earthquakes in southern California, the Mw 6.7 1971 San Fernando and Mw 5.9 1987 Whittier earthquakes. In equation 11, B denotes repairs per square mile and d denotes number of days of watersupply outage: d = 2.18 + 2.51 × ln B =0 B > 0.42 . B ≤ 0.42 (11) Serviceability of Water Supply As previously noted, some analytical models are capable of modeling the serviceability of a damaged water-supply system Number of fixed leaks per day per worker 0.66 0.66 0.66 1.0 1.0 1.0 Number of available workers User specified User specified User specified User specified User specified User specified Priority 1 (highest) 2 3 4 5 (lowest) 6 (lowest) using hydraulic or connectivity analysis (see, for example, Khater and Grigoriu, 1989). As in the case of the closely related Life Line Earthquake Engineering (LLEQE) software, the Applied Technology Council (1991) noted that such systems can be data intensive and computationally demanding. What can be done to estimate water-supply serviceability without a hydraulic model? Isoyama and Katayama (1982) proposed to measure a quantity they called serviceability as the probability that the demand at a system node (such as a customer-service connection) is fully satisfied, or in the aggregate, the average fraction of nodes in the entire system whose demand is fully satisfied. Demand seems to mean the pre-earthquake consumption plus postearthquake leakage. Markov and others (1994) propose to measure serviceability using a serviceability index, SS, defined as the ratio of the total available flow to the total required flow, 206   The HayWired Earthquake Scenario—Engineering Implications 100 s (r ) = 1 −Φ ln (( r L ) q ) b Serviceability index, in percent which is similar but not identical to Isoyama and Katayama’s serviceability. If demand at 10 nodes were fully satisfied and demand at 10 other nodes were partially satisfied, the two measures of serviceability would take on different values—0.5 in the case of Isoyama and Katayama (1982) and somewhat higher in the case of Markov and others (1994). The developers of the Hazus-MH water system use data from Isoyama and Katayama (1982) and Markov and others (1994) to propose to estimate the serviceability index, s(r), as a function of break rate (breaks, not leaks, per km of service main pipe) using equation 12. They seem to use the serviceability index to measure the fraction of customers receiving any water service, because the software expresses loss of serviceability in terms of “households without water”: 80 60 40 20 0 . (12) In equation 12, ln denotes natural logarithm, r/L denotes the average break rate (r main breaks per L kilometer of pipe), q and b are model parameters, and Φ is the standard normal cumulative distribution function (the y-value of the S-shaped curve in x–y space that depicts the probability that an uncertain quantity with standard normal distribution will take on a value less than or equal to x). Hazus-MH employs values of q=0.1 and b=0.85, respectively, fitting the curve to Isoyama and Katayama’s (1982) modeling of Tokyo’s water-supply system, Markov and others (1994), modeling of the San Francisco Auxiliary Water-Supply System (a dedicated firefighting system), and analyses of EBMUD’s water-supply system. Hazus-MH’s serviceability model is illustrated in figure 4, in the curve labeled “NIBS.” Thus, the Hazus-MH serviceability index might measure: • The fraction of service connections receiving preearthquake flows, regardless of the degree of postearthquake flow received at other service connections, which would seem to be consistent with Isoyama and Katayama’s (1982) serviceability; • The fraction of pre-earthquake flow being delivered after the earthquake, consistent with Markov and others (1994); or • The fraction of service connections receiving any water, as the Hazus-MH reports indicate. Lund and others (2005), citing Kobe, Japan, Municipal Waterworks Bureau’s M. Matsushita, present a restoration curve for the Kobe water system after the Mw 6.9 1995 Kobe earthquake. Tabucchi and Davidson (2008) offer an analogous plot for the restoration of water service in the San Fernando Valley after the 1994 Northridge earthquake. The two restoration curves are shown in figure 5. Restoration after Northridge appears fairly linear; Kobe less so. 0.01 0.1 1 Average break rate, in breaks per kilometer EXPLANATION NIBS AWSS (simulation results at Cornell) AWSS (average) EBMUD(G&E) (large pipes) EBMUD (G&E) (small pipes) Isoyama and Katayama (1982) (lower range) Isoyama and Katayama (1982) (upper range) Figure 4.  Graph showing Hazus-MH (Federal Emergency Management Agency, 2012) water-pipe serviceability model and other models of water-pipe serviceability. Hazus-MH uses the curve labeled “NIBS.” AWSS, San Francisco Auxiliary Water Supply System (Markov and others,1994); EBMUD, East Bay Municipal Utility District; G&E, G&E Engineering Systems, Inc.; NIBS, National Institute of Building Sciences (Federal Emergency Management Agency, 2012). Lifeline Interaction Many authors have characterized lifeline interaction after natural disasters. A few but not all relevant works are summarized here. For ease of reference, I recall here some evidence previously noted—City of Winnipeg (2014) and American Water Works Association (2008) suggest that prerequisites for the repair of buried pipeline include cellular communications and electricity to learn about and coordinate repairs, fuel and roadway access to travel to and perform the repairs, site safety (especially that no fires, gas leaks, or electrical hazards are present), and consumable repair materials including pipe, fittings, repair hardware, and disinfecting chemicals. AAXXXX_fig 01 Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 207 100 100 A B Water-service recovery, in percent 80 80 60 40 60 20 40 0 1 2 3 4 5 6 7 8 9 10 Time following earthquake, in weeks 20 0 0 0 1 2 3 4 5 6 Figure 5.  Graphs showing (A) restoration of water service to customers experiencing outages after the moment magnitude (Mw) 6.7 1994 Northridge, California, earthquake and (B) after the Mw 6.9 1995 Kobe, Japan, earthquake. A, Modified from Tabucchi and Davidson (2008); B, modified from Lund and others (2005). 7 Time following earthquake, in days Nojima and Kameda (1991) compiled instances of lifeline interaction in the 1989 Loma Prieta earthquake, noting particularly the loss of wastewater treatment because of the loss of electricity, and the degradation of telecommunications resulting from the loss of electricity and difficulty acquiring fuel for central offices’ emergency generators as a result of highway problems. See table 6 for a matrix summarizing lifeline interaction in the Loma Prieta earthquake. It shows that water supply was impaired for 18 hours in Santa Cruz because of loss of electric power for pumping. It also shows that electricity failure impaired EBMUD’s Lafayette filtration plant and its Oakland control center. Repairs in Santa Cruz were also impaired by delays transporting repair equipment over the damaged Oakland-San Francisco Bay Bridge. In San Francisco and Santa Cruz, overloaded telecommunications impaired repair efforts. Scawthorn (1993) reviews literature and then-recent disaster experience on lifeline interaction in several disasters (for example, 1989 Loma Prieta Earthquake and 1991 Oakland Hills firestorm in the east bay) to construct a model and analytical methodology for lifeline interaction. He points out that water supply in the 1991 Oakland Hills fire was impaired in part because of breakage of service connections in buildings that collapsed in the fire and the reliance of water supply on electric power to pumps stations that were required to resupply ridge-top tanks. He suggests characterizing lifeline interactions as (1) cross-impact (impact on one lifeline’s function due to impairment of service to that lifeline by a second lifeline), (2) collocation (direct damage or impact on one lifeline’s function due to failure of another lifeline in a very proximate location), or (3) cascade (increasing impacts on a lifeline due to initial inadequacies, such as water-supply damage as buildings collapse and sever service connections). In Scawthorn’s (1993) quantitative model, one characterizes initial damage to a set of lifelines through a vector, D, of n scalar quantities, each element representing a fraction of customers receiving service for one of n lifelines if there were no interaction, that is, if only damage to that lifeline mattered. Lifeline interaction is quantified by an n×n matrix denoted by L, where element Li,j (row i, column j) denotes the fraction of service of lifeline i that is contributed by lifeline j. A higher value of Li,j indicates greater reliance of lifeline i service on lifeline j. A value Li,j=0 indicates no interaction. The final functional state of the n lifelines is represented by vector F, whose value is given by equation 13. Element i of vector F measures the fraction of customers receiving service from lifeline i, where any reduction below Fi=1.0 is a result of initial damage D to all the lifelines and interaction L between them. Equation 13 is as follows: F = L × D . (13) Scawthorn offers the model but does not propose particular values for matrix L. Note that, because 0≤Di ≤1.0, to ensure that 0≤Fi≤1.0, L must be constrained per equation 14: n i ∈ {1, 2,...n} . ∑ Lij = 1.0 j =1   (14) The San Francisco Lifelines Council (2014) adapted the panel process of Porter and Sherrill (2011) to involve San Francisco Bay region lifeline operators in qualitatively characterizing the potential effects of lifeline interaction on the post-earthquake functionality of their systems. The authors sought to identify key assets and restoration schemes to prioritize postdisaster restoration and reconstruction activities for San Francisco and ultimately the entire region. Through panel discussion with 11 lifeline operators, the authors identified lifeline interaction effects in the context of a hypothetical M7.9 earthquake on the northern San Andreas Fault. They propose a qualitative interaction matrix (table 7) that describes modes of interaction similar to Nojima 208   The HayWired Earthquake Scenario—Engineering Implications Table 6.  Lifeline interaction matrix in the moment-magnitude-6.7 1989 Loma Prieta, California, earthquake (modified from Nojima and Kameda, 1991). [SF, San Francisco; BART, Bay Area Rapid Transit; EBMUD, East Bay Municipal Utility District; PBX, private branch exchange; hrs, hours; *, general interaction, meaning among components of the same system] Lifeline type Electricity Gas Water Sewer Road Rail Telephone Electricity Gas * Santa Cruz gas explosion due to electricity comeback (spark ignition). Recovery work arrangement with electric powersupply system. SF and Santa Cruz: gas leak inspection before recovering electricity. * Santa Cruz: recovery work arrangement with water-supply system. Water Sewer Santa Cruz: pump SF and Santa stopped for 18 hrs Cruz: (gravity flow area power survived; no water failure in pump-based at pump supply area) station. SF: power failure due to gas leak inspection, no water in pumpbased supply area and Marina district. No power for repair work. EBMUD: short-term loss of power at Lafayette filtration plant. Oakland Control Center power loss, no service. Santa Cruz: no home treatment. Recovery work arrangement with gas supply system. * SF and Santa Cruz: traffic jam due to malfunction of traffic signal. Rail Telephone SF: BART Capacity diminished omitted stops by use of storage at some cells. PBX stations with no battery, to save malfunction. electricity. Pacific Bell Bush/ Pine Street office (SF) coolant trouble; no service for 3 hrs. Pacific Bell Hollister office generator failure no service for 3 hrs. GTE Corp.: Monte Bello office (Los Gatos) failure of generator fuel tank; malfunction (6–7 hrs). SF: road closed due to propane fire (Route 80 westbound at Central Avenue). Santa Cruz: SF Marina damage District: road detection failure due to by analogy. water leakage. * Santa Cruz: suspicion of underground water contamination due to outflow or crude sewage from pipeline. Santa Cruz: no Santa Cruz: transporting damage machinery due to detection bridge damage. by analogy. SF and Santa Cruz: overload. Road * BART riders increased (October 23, +40 percent) due to Bay Bridge closure. * * Collocation, restoration Functional, collocation, restoration Functional, restoration Restoration Substitute, restoration Restoration Restoration Restoration City streets Electric power Natural gas Telecom Restoration Substitute, restoration Restoration Restoration Restoration Waste-water Transit Port Airport Fuel Restoration Water Auxiliary water Restoration Restoration Functional, restoration Collocation, restoration Functional, substitute, collocation, restoration Collocation, restoration City streets Restoration, substitute General Regional roads Regional roads General Lifeline Functional, restoration Collocation, restoration Restoration Functional, restoration Functional, restoration Restoration Functional, restoration Restoration Substitute General Collocation, restoration Restoration General Collocation, restoration Electric power Natural gas Restoration Restoration Restoration Collocation, restoration Restoration Restoration Restoration Restoration Restoration General Restoration Restoration Collocation, restoration Telecom Restoration Restoration Collocation, restoration Restoration Functional, restoration Functional, restoration Collocation, restoration Collocation, restoration General Collocation, restoration Collocation, restoration Collocation, restoration Water Restoration Collocation Collocation, restoration General Collocation, restoration Collocation, restoration Collocation, restoration Collocation, restoration Auxiliary water Restoration Collocation General Collocation, restoration Collocation, restoration Collocation, restoration Collocation, restoration Collocation, restoration Waste-water Restoration Collocation, restoration Collocation Collocation, general Collocation, substitute, restoration Transit Substitute Restoration General Collocation, restoration Collocation, restoration Collocation, restoration Collocation Collocation Collocation Collocation, restoration Port Restoration General Restoration Restoration Restoration Airport Restoration Functional, restoration General Restoration Functional, restoration Restoration Restoration Restoration Restoration Restoration Restoration Restoration Fuel Restoration [Darker shading indicating greater interaction. For lifeline operator’s dependency on other lifeline systems, read across rows. For the overall interaction and dependency on a particular lifeline system read down columns] Table 7.  The San Francisco Lifelines Council’s lifeline-system interdependencies matrix (modified from San Francisco Lifelines Council, 2014). Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 209 210   The HayWired Earthquake Scenario—Engineering Implications and Kameda (1991) and shows a degree of interaction, with darker shading indicating greater interaction, like a higher value in Scawthorn’s (1993) matrix. The authors found that restoring water supply in San Francisco depends significantly on city streets, telecommunications, and fuel and to a lesser extent on regional roads, electric power, and the Port of San Francisco. The matrix characterizes the mode of each interaction, with five possible modes. The following quotations are taken from San Francisco Lifelines Council (2014) but the interpretations are mine: • “Functional disaster propagation and cascading interactions from one system to another due to interdependence.” This means that a system relies on one or more other systems to operate, each of which can rely on still others. I refer to these other systems as “upstream,” in the sense that failure of an upstream system flows or cascades down to the system in question and causes its failure. For example, consider water service in a pressure zone that is supplied from tanks whose source is water pumped from lower elevation. Water service in that pressure zone is functionally dependent on electricity, which may be functionally dependent on natural gas. Failure of fuel supplies or electric generation, transmission, or distribution propagates or cascades to cause water-supply failure through interdependence. • “Collocation interaction, meaning physical disaster propagation among lifeline systems.” This means that one or more elements of the system in question are located close to one or more elements of another system and that the other system can fail in such a way that an area around the failure can impair the system in question. For example, fiber optic cable that serves the telecommunications network may be installed in a conduit on a roadway bridge. Excessive displacement of the bridge, such as a result of settlement of an abutment, can sever the fiber conduit. • “Restoration interaction, meaning various hindrances in the restoration and recovery stages.” This means that one or more elements of the system in question are located close to one or more elements of another system, and that repairs to the other system can damage or hinder the repair of the system in question. For example, consider a water main (the system in question) that is located above a damaged sewer line. Repair to the sewer line could require the temporary removal of or inadvertently lead to damage to the water main. • “Substitute interaction, meaning one system’s disruption influences dependencies on alternative systems.” This means that the system in question may have substitutes (alternative systems), and that disruption of one of the alternatives can affect the system in question. For example, damage to the San Francisco-Oakland Bay Bridge in the Mw 6.9 1989 Loma Prieta earthquake caused a 32-percent increase in Bay Area Rapid Transit (BART) ridership during October and November 1989 (Bay Area Rapid Transit District, 2015). • “General interaction, meaning between components of the same system.” Nojima and Kameda (1991) use an asterisk (*) to mean the same thing. This means that impairment of elements of the system in question can affect other elements of the same system. For example, overturning of electrical switchgear in a pumping station can cause the pumps to fail. Pipeline Damage in Afterslip Several authors have considered lifeline damage due to afterslip, which is fault slip immediately following an earthquake rupture that involves creep much faster than the interseismic rate. According to Aagaard and others (2012), “Afterslip develops very quickly and can have similar impacts as coseismic slip, with the added complexity that the slip continues for months to years, albeit with a decreasing rate.” They discuss afterslip in various Hayward Fault earthquake scenarios, including the one adopted for use in HayWired— “Afterslip makes a substantial contribution to the long-term geologic slip and may be responsible for up to 0.5–1.5 m [meters] (median plus one standard deviation) of additional slip following an earthquake rupture.” The authors offer a powerlaw expression for afterslip as a function of time t, denoted D(t), as follows in equations 15–19 (s, second): D (t ) = A + B × A= 1 (1 + T t ) C , (15) 1 × ( Dtotal − a × Dcoseismic ) , 1− a (16) −a × ( Dtotal − Dcoseismic ) , 1− a  (17) B= C T , a = 1+ 1s   (18) Cmedian = 0.881 − 0.111 × M w ,  (19) where T is referred to as the afterslip time constant, taken here as 365 days per Aagaard and others (2012). For example, with Mw=7.0, equation 19 leads to Cmedian=0.0984. With T=365 days, equation 18 leads to a=5.47. With Dtotal=1.86 meters (m) and Dcoseismic=0.83 m, equations 15, 16, and 17 produce A=0.608 m, B=1.25 m, and the estimate of slip versus time shown in figure 6. O’Rourke and Palmer (1996) point out that understanding observations of pipeline damage at fault crossings requires estimating fault slip from the time of pipeline installation to the time of its excavation for inspection after an earthquake, including preseismic slip, coseismic slip, and afterslip. Treiman and Ponti (2011) suggested that afterslip could realistically account for 40 percent of the total surface slip in the Coachella Valley, California, resulting from a M7.8 earthquake on the southern San Andreas Fault. The afterslip could aggravate damage to the Coachella Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 211 Canal, railroads, fiber optic cable, electrical lines, gas and oil pipelines, and highways. Hudnut and others (2014) measured deformation in a temporarily decommissioned 26-inch diameter gas pipeline that crosses the fault rupture involved in the 2014 South Napa earthquake. They observed that the pipeline was “subtly warped more than 35 cm [centimeters] by fault offset, most of which accumulated as afterslip that is still continuing as of 3 months after the earthquake.” They argue that “Lifeline performance in future events, with both coseismic slip and afterslip, deserves additional consideration.” Measuring Loss of Resilience Bruneau and others (2003) propose to measure the loss of resilience as the area above the curve Q(t), where Q(t) is defined (somewhat vaguely) as the “quality of the infrastructure of a community” They denote a quantity they call “community earthquake loss of resilience” by R and calculate it as in equation 20: R= t1 ∫ (1 − Q (t )) dt ,  (20) 0 where t=0 and t=t1 denote the times of the initiating event and the time of full restoration, respectively. For brevity, I refer to R more simply as the loss of resilience. Bruneau and others (2003) do not define t=0 and t=t1 precisely. I define t=0 here as the time of the first earthquake in the earthquake sequence under consideration, and I define t1 as the time when Q(t)=1 after the last earthquake in the sequence under consideration. Let Q(t) measure the fraction of water customers receiving at least an adequate degree of service at time t, meaning sufficient water flow and pressure at the tap for drinking (even if it needs to be boiled first), bathing, and using toilets. R has units of time, and as applied here, can be seen here as the expected value of the time that an arbitrary customer receives less water than a useful degree of service. To be clear, a reduction in the loss of resilience indicates a briefer average time that an arbitrary customer lacks adequate service, but I will not equate a reduction in the loss of resilience with an increase in resilience. In Bruneau and others’ (2003) terminology, resilience is not a quantity but rather a quality that means “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 (reestablish normal performance).” Resilience is not the mathematical complement of the loss of resilience. Methodology Using the brief literature review described above as a basis, I now turn to the proposed methodology. The methodology used in this chapter to evaluate water-network resilience for the HayWired scenario is discussed below. Overview of the Methodology 2.0 A lifeline earthquake performance and restoration model typically involves the following analytical elements: 1. Asset definition, in which the system is described in terms of nodes and links. Nodes have a location, flow capacity, sometime a value (for example, replacement cost), and an asset category that associates the component with one or more relations among environmental excitation (for example, severity of shaking) and loss (for example, in terms of dollars, deaths, downtime, or some combination). Links connect nodes. They have a path, sometime a direction, flow capacity, sometimes a value, and an asset category. The assets in question here are defined in the San Jose Water Company case study (described later in this chapter). Slip, in meters 1.5 1.0 0.5 0.0 0.00001 0.001 0.1 10 1,000 Time after mainshock, in days Figure 6.  Graph showing afterslip versus time after a moment-magnitude-7.0 mainshock as derived from equations 15 through 19 (see text). 2. Hazard model, relating geographic location to environmental excitation. In the case of earthquake hazard, the hazard model typically includes a mathematical idealization of seismic sources in the region, their locations, the frequency with which they can produce earthquakes of various magnitudes, and one or more ground motion prediction equations to relate earthquake magnitude and location to shaking and other site effects. In the present study, the hazard model is presented elsewhere. Briefly, it is a physics-based model of the San Francisco Bay Area, depicted in Aagaard and others (2010a, b). 212   The HayWired Earthquake Scenario—Engineering Implications • yi(x), the expected value of the degree of loss experienced by a component of class i when subjected to excitation x. One can refer to yi(x) as the mean vulnerability function for class i. 3. Hazard analysis, in which one evaluates the hazard model for one or more realizations of an earthquake. In the present analysis, I use the realization from Aagaard and others (2010a, b) depicting a Mw 7.0 rupture of the Hayward Fault north and south segments with an epicenter under Oakland. Accompanying the model of shaking from the mainshock are estimates of liquefaction probability, landslide probability, coseismic slip, and afterslip associated with the mainshock, along with shaking in each of a sequence of 16 aftershocks of M5.0 and greater, as described in Wein and others (2017). 4. Vulnerability model (described later in this chapter), which relates environmental excitation at a particular location to the potentially uncertain loss in each of a set of asset classes. • εi(x), the error term for class i. The error term can be constant for class i or it can depend on degree of excitation x. The error term has unit mean and usually has some parametric distribution, such as lognormal with a specified standard deviation of the natural logarithm of the error term, referred to here as the logarithmic standard deviation. The vulnerability model can provide mean vulnerability functions and error terms for one or more modes of damage j, such as damage by wave passage and damage by ground failure due to liquefaction, landslide, or fault offset. 5. Damage analysis (described later in this chapter), in which one evaluates the vulnerability model for each lifeline component at the level of environmental excitation to which the component is subjected. • yi,j(xi,j)=the mean loss to a component i of a specified class in damage mode j when component i is subjected to mode-j excitation xi,j, such as the peak ground velocity to which a particular segment of pipe is subjected. 6. Restoration model (described later in this chapter), which characterizes the time to restore the damaged components to their pre-disaster condition, and calculates the degree of service at each of many points in time. (The restoration model developed for the HayWired scenario is new. It includes a new method for quantifying the effects of lifeline interaction. It includes an initial assessment period suggested by engineers of the East Bay Municipal Utility District, along with a period during which repair crews and other resources ramp up from an initial, in-house quantity to one that includes mutual aid.) • εi,j(xi,j)=the mode-j vulnerability error term for the class to which component i belongs, when the component is subjected to excitation xi,j. 7. Aftershock analysis, in which one inserts one or more aftershocks into the restoration process, which in a sense restarts the hazard, damage, and restoration analyses with a still-damaged lifeline system. Vulnerability Model Here, “vulnerability model” means a mathematical formulation of the relation between loss (usually normalized by quantity; for example, pipeline breaks and leaks per 1,000 feet, ft, of pipe) and environmental excitation (for example, degree of PGV). Basic Elements of a Vulnerability Model These relations often apply to classes of components that share common engineering features (for example, pipe sharing common material, diameter range, or joint type). All specimens of the class are assumed to be interchangeable and indistinguishable for purposes of estimating loss. A vulnerability model can be deterministic, providing for example only a mean estimate of loss conditioned on excitation, or probabilistic, providing both a mean value and error term. Some terms need to be defined: • Yi,j(xi,j)=the uncertain normalized loss is denoted by (for example, uncertain total pipeline breaks and leaks per 1,000 linear feet of pipe), where the index i refers to the component class to which component i belongs, j refers to the damage mode under consideration (for example, pipeline breaks and leaks per 1,000 linear feet of pipe as a result of wave passage), and xi,j is as previously defined. The uncertain normalized loss is calculated as: Yi , j ( x i , j ) = yi , j ( xi , j ) × ε i , j ( xi , j ) . (21) The vulnerability model comprises the set of functions y and ε, the component classes to which they refer, and the domain of excitations for which the functions are valid. Selecting a Vulnerability Model for a Pipeline Network Several authors have created and published pipeline vulnerability functions; a few are discussed above in the Literature Review. There are no commonly accepted rating systems for pipeline vulnerability functions, but it seems reasonable to choose among competing vulnerability functions based on at least the following criteria: • Vulnerability functions reflect diverse conditions—pipe material, diameters, joint systems, age, and corrosivity similar to the conditions where the vulnerability functions will be applied. • Vulnerability functions are drawn from numerous damage data. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 213 10.0 • The articles in which the vulnerability functions are presented are respected, highly cited, and frequently used for similar applications, which here means estimating and depicting realistic outcomes of a hypothetical U.S. earthquake (table 8). Breaks per 1,000 linear feet of pipe • Vulnerability functions are drawn from ground-motion levels reaching as strong as those where they will be applied. Rl (0.95) 7.5 5.0 O’Rourke and others (2014) have been more cited in far fewer years than Eidinger (2001), suggesting somewhat greater credibility. Maximum PGV values are greater in the Rl O’Rourke work, suggesting greater applicability in strong shaking. However, Eidinger (2001) draws on a larger dataset 2.5 and his vulnerability functions cover both wave passage and ground failure. For these reasons, it seems the Eidinger vulnerRl (0.05) ability functions are most suited to the present problem. Thus, for wave passage, one can use equation 3 to 0.0 calculate repair rate with probability p of nonexceedance, or 0 5 10 15 20 alternatively equation 4 for the mean repair rate. However, Permanent ground deformation, in inches there is a problem applying a liquefaction and landslide model Figure 7.  Graph showing Eidinger’s (2001) pipe liquefaction vulnerability that requires permanent ground displacement to HayWired, (Rl) for K2=1.0, mean and 90-percent bounds (see equations 5 and 6). Factor as in Eidinger’s (2001) model equations 5 or 6. The problem K2 accounts for pipe material, joints, soil corrosivity, and pipe diameter. here is that PGDs are unavailable for the HayWired scenario. HayWired instead has liquefaction probability and landslide not to say that it would not be a little better to estimate PGD, it probability. How to apply Eidinger’s ground-failure model just would not be much better. without an estimate of PGD? In light of the very high uncertainty in repair rate and the The solution employed here takes advantage of the fact relatively modest sensitivity of the vulnerability function to PGD, that equation 6 is not very sensitive to PGD. One can see the I assume a reasonable moderate PGD associated with liquefaction, limited sensitivity in the small power to which PGD is raised, say 6 inches, and rewrite equations 5 and 6 using liquefaction 0.319. At the same time, the logarithmic standard deviation probability, as shown in equations 22 and 23: β=0.74 in equation 5, which gives the marginal distribution of repair rate, is very large, suggesting the 90th percentile bounds Rl ( PL , p ) = K 2 × PL × 1.06 ⋅×6 0.319 × exp ( 0.74 × Φ −1 ( p )) differ by more than an order of magnitude. In this case, the ,  (22) 95th and 5th percentiles of repair rate conditioned on PGD = 1.88 × K 2 × PL × exp ( 0.74 ⋅ Φ −1 ( p )) differ by a factor of 11.4. So Eidinger’s (2001) liquefaction equation tells us that an increase in PGD from 1 inch to 10 inches only increases mean Rl ( PL ) = K 2 × 1.39 × 60.319 × PL ,   (23) repair rate by a factor of 2. See figure 7; for K2=1, the repair = 2.46 × K 2 × PL rate for PGD=1 inch and the repair rate for PGD=10 inches where PL denotes probability of ground failure, either through are 1.4 and 2.9 repairs per 1,000 linear feet, respectively. liquefaction, landslide, or fault offset. The equation estimates At either point, PGD=1 inch or 10 inches, the uncertainty mean repair rate per 1,000 linear feet of pipe. in repair rate is much greater, that is, even if I knew PGD, I How does one sum repair rates from wave passage and would still be very uncertain as to repair rate. The apparent ground failure if one uses Eidinger’s (2001) model? He says that improvement in accuracy gained by estimating liquefaction“wave propagation effects are masked within the more destructive induced or landslide-induced PGD would be illusory. That is Table 8.  Comparison of criteria for selecting pipeline vulnerability functions. [PGV, peak ground velocity; PGD, permanent ground displacement; cm/s, centimeter per second; NA, not applicable] Diverse pipe Repairs Max PGV Max PGD Citations O’Rourke and Ayala (1993) Source Yes Unknown 50 cm/s NA 87 Honneger and Eguchi (1992) Yes Unknown Unknown Unknown 21 Eidinger (2001) Yes 3,350 52 cm/s 110 in 18 O’Rourke and others (2014) Yes 2,051 80 cm/s NA 20 214   The HayWired Earthquake Scenario—Engineering Implications effects of [peak ground displacements].” If one knew where ground failure occurs, one would ignore the wave-passage model so as not to double-count it. The problem here is that Eidinger’s (2001) empirical model of damage due to liquefaction probably includes some damage that was caused by wave passage, but he does not know which breaks and leaks were caused by which peril. In zones of liquefaction, he treats all damage as caused by liquefaction. That is, his empirical relation for damage in zones of liquefaction include an unknown (but probably small) fraction of damage caused by wave passage. As a result, one must not double-count wave passage damage by applying both the liquefaction and wave-passage models in zones of liquefaction. To eliminate double-counting, I modify the wave-passage model of equations 3 and 4 by multiplying the repair rate by a factor (1–PL), where PL denotes ground-failure probability. After eliminating double-counting, one can sum the wave-passage and ground-failure models as shown in equations 24 and 25. In both equations, R denotes repair rate in repairs per 1,000 ft of buried pipe, PGV is peak ground velocity in inches per second, and p denotes nonexceedance probability. In equation 24, R gives repair rate with nonexceedance probability p, whereas equation 25 gives mean (average) repair rate. The coefficients are smaller in equation 24 than they are in equation 25 because the median is smaller than the average for a lognormally distributed random variable, and the difference depends on the logarithmic standard deviation: R ( PGV , PL , p ) = (1 − PL ) × Rw ( PGV , p ) + RL ( PL , p ) = (1 − PL ) × K1 × 0.00187 × PGV × exp (1.15 ⋅ Φ −1 ( p )) + 1.88 × K 2 × PL , (24) × exp ( 0.74 × Φ −1 ( p )) R ( PGV , PL ) = (1 − PL ) × Rw ( PGV ) + Rl ( PL ) = (1 − PL ) × K1 × 0.003623 × PGV . (25) + 2.46 × K 2 × PL For damage resulting from fault offset, one could apply Eidinger’s (2001) proposed model. Given the absence of supporting data, the relatively small number of breaks that occur at the fault trace compared with breaks and leaks that occur as a result of wave passage and liquefaction, and the desire to model breaks as a function of offset at the location of the pipe rather than average offset over the entire trace; a simpler model is adopted here. I assume that any pipe segment that crosses a fault is broken if the fault offset exceeds 6 inches and use the same threshold regardless of pipe material, jointing, and angle subtended by the fault and pipeline alignment. The fault trace is treated as a collection of line segments rather than as a zone on the surface of the Earth with a finite width. The offset therefore is lumped at the line rather than distributed over the width of the zone. Mathematically, let Zi denote a binary variable to indicate that pipe segment i is damaged by fault offset (1 if true, 0 if false), d denotes the fault offset distance, df denotes the threshold of fault offset distance that produces damage, and I(⋅) is the indicator function (1.0 if the value in parentheses is positive, 0.0 if negative), then Z i = I ( d − df ) , (26) where df=6 inches, consistent with the fault offset that Eidinger (2001) equates with a 50-percent failure probability for all segmental pipe. One could use a more refined model such as American Society of Civil Engineers (ASCE) (1984), which applies engineering first principles of stress and strain, the engineering characteristics of pipe and backfill, the geometry of how a pipe crosses a fault, and other factors. In the context of an earthquake-planning scenario in which we care about the total number of pipe breaks and leaks over an entire strongly shaken region, such an analysis seems like excessive effort for a relatively small contributor to overall damage. Furthermore, considering the necessary assumptions about unknown backfill characteristics and probably other model parameters, such an analysis would probably provide illusory precision. Damage Analysis (Number of Repairs Required) A damage analysis applies the vulnerability model and the hazard model to the assets under consideration to estimate degree of damage or loss. An example would be the total number of pipeline repairs required when a particular pipeline network is affected by a particular earthquake. Basic Elements of a Damage Analysis I employ a common general formulation for number of repairs required for a system of ni discrete components (for example, segments of pipe) that can each be uniquely identified with a class of components (for example, type of pipe). Each component i has an associated quantity or value Vi (for example, length of pipeline segment), and is assumed to be subject to damage from up to nj modes of damage (for example, wave passage and liquefaction). Each combination of component class and mode of damage is assigned a vulnerability model yi,j(xi,j) and εi,j(xi,j) as previously defined. Let R denote the total uncertain loss (for example, total number of instances of pipeline damage). It is estimated as shown in equation 27: ni nj i j R = ∑ ∑ Vi × yi , j ( xi , j ) × ε i , j ( xi , j ) .  (27) Equation 27 assumes that damage to one component or in one model is independent of damage to other components or in other modes. That is, that the degree of damage to component i in mode j is unaffected by damage to a different component, and that if component i is damaged in one mode, it can also be damaged in another mode and that the losses resulting from the two modes of damage simply sum. In the case of water-supply pipelines, one implication of this assumption of independence is that it assumes that repairs Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 215 are spaced widely enough apart that it makes sense to repair individual breaks or leaks, at least initially, rather than to remove and replace pipe and thus repair two or more instances of damage with a single repair. Applying the Damage Analysis to a Water-Supply System Applying equation 27 to water-supply pipelines using Eidinger’s (2001) vulnerability model, and adding an additional term for fault crossings, one can estimate mean total number of repairs r as shown in equation 28: n n n i =1 i =1 i =1 r = ∑ Rw ( PGVi ) × Li + ∑ Rl ( PL,i ) × Li + ∑ I ( d − df ) n = ∑ (1 − PL,i ) × 0.003623 × K1,i × PGVi × Li i =1 n n i =1 i =1 , (28) + ∑ 2.46 × K 2,i × PL,i × Li + ∑ I ( d − df ) where i is an index to pipe segments, n is the total number of segments in the network, K1,i and K2,i denote the values of K1 and K2 for pipe segment i, PGVi is the peak ground velocity to which segment i is subjected, PL,i denotes the ground-failure probability at pipe segment i, Li is the length of pipe segment i in thousands of linear feet, I(⋅) is an indicator function that takes on the value 1.0 if the expression in parentheses is positive, 0.0 if negative, di is the fault offset to which a segment of pipe i is subjected, and df is the offset at which breakage occurs. Here, I propose to take df as deterministically equal to 6 inches (15 cm). Note that as long as pipe segments i are relatively short, less than a few hundred meters, there should be little difference between shaking at the ends and thus little error introduced by discretizing a pipeline network in this way. Because the present analysis does not require a probabilistic estimate of loss, I ignore the error term e and deal only with the expected value of loss. I use the lower-case r in equation 28 to indicate a deterministic value rather than the upper-case R of equation 27 that stands for an uncertain quantity. To carry out equation 28, one uses a geographic information system (GIS) to create a table of system components; for example, a table of pipe segments. Components are listed in rows. For each component, assign an identifier, determine its quantity (for example, its length), assign it to a class that has one or more vulnerability or fragility functions (for example, Eidinger’s, 2001, classes that group water-supply pipe by material, joint, soil corrosivity, and diameter), and determine its location (for example, the latitude and longitude of a pipe segment midpoint). Then using the GIS, look up the groundmotion parameter values xi,j. Equation 28 can then be calculated for each component (each row) and the losses summed over all rows to calculate the expected value of loss r (for example, the number of instances of pipeline damage requiring repairs). Breaks or Leaks? Lund and Schiff (1991) define leaks and breaks for purposes of compiling damage data. Under their definition, a pipe with a leak continues to function with minimal loss of service, whereas a pipe with a break completely loses function. It seems as if another, equivalent definition is that a pipe break separates a pipe segment into two, and a leak only partially fractures a pipe. Hazus-MH assumes an 80 percent/20 percent break/leak ratio for liquefaction and 20 percent/80 percent breaks/leaks ratio for wave-passage damage. The authors of the technical manual do not cite a source for their choices. Lund and Schiff (1991) found that, among all pipeline failures in the Mw 6.9 1989 Loma Prieta earthquake, where it was known whether the failure was a break or a leak, it was more common for the pipeline to break (336 repairs, 71 percent) than to leak (140 repairs, 29 percent). A study by Ballantyne and others (1990) of pipe damage in the Mw 6.7 1949 Olympia and Mw 6.7 1969 Seattle (Washington), Mw 5.6 and 5.7 1969 Santa Rosa (California), Mw 6.7 1971 San Fernando Valley, Mw 6.2 1983 Coalinga (California), and Mw 5.9 1987 Whittier Narrows earthquakes, found that ground failure resulted in a 50 percent/50 percent break/leak ratio, and absent ground failure, the ratio was 15 percent/85 percent breaks versus leaks. Because the present model allows one to distinguish between repairs associated with ground failure versus wave passage, and because Ballantyne and others (1990) are highly regarded and offer their evidence, I employ their ratios. Degraded Vulnerability? The model presented here applies the same vulnerability functions to the same system map in the aftershocks that it applies to the mainshock. Is it correct to do so? Perhaps I should consider a system that has already been degraded by the mainshock or a large aftershock to be weaker. Perhaps the mainshock causes small undetected leaks or incipient breaks that become large leaks or breaks in an aftershock. However, there does not seem to be sufficient research available to support explicitly modeling system degradation—making the mathematical model of the system more vulnerable in aftershocks than before the mainshock. This is a topic deserving of future research. Restoration Model As used here, a restoration model relates the damage (the output of the damage model) to a system’s functionality over time. It usually depicts a system’s return to its predisaster condition. Basic Elements of the Lifeline Restoration Model Functionality can be measured a variety of ways, but in the case of a utility such as a pipeline network, it is common 216   The HayWired Earthquake Scenario—Engineering Implications to measure functionality in terms of the number of service connections that receive the lifeline service as a function of time. I do not offer a general mathematical formulation of a lifelinerestoration model, but merely list its elements here, and then propose a particular solution for water-supply pipeline systems. A lifeline-restoration model includes the following elements: • A model of the level of functionality immediately after the disaster • A model of the repair resources—crews and supplies— available over time. V (n) = M × 1 −Φ ln r−n L×q β . (30) • A model of the number of services restored by each repair Or one could model services as being restored in proportion to the number of repairs remaining, as shown in equation 31. I refer to equation 31 as the proportional approach. • A model of the elapsed time after each repair • Ideally, a model of lifeline interaction (that is, accounting for how damage or restoration of other lifelines affects or delays damage or restoration of the lifeline in question). A more general approach is suggested by conversations with engineers of EBMUD. They indicated that their repair strategy in an earthquake would be to focus most of EBMUD’s resources to repair water-transmission lines that serve large areas, then smaller diameter distribution lines that serve smaller numbers of customers, and so on. The strategy would depend on how parts of the system, which may have been impacted by damage in large diameter pipes, could first be isolated to continue to maintain services to as many customers as possible by rerouting water using a combination of temporary system such as portable pumps and flexible hoses. If one were to plot a restoration curve with the fraction of customers receiving service on the y-axis and time after the earthquake on the x-axis, then a plot for EBMUD’s strategy would maximize slope as soon as repairs begin. The slope might increase if the number of crews increases, but with constant resources, the slope will decrease as individual repairs restore fewer and fewer services. Equation 32 would have such a form for values of 00.5 CALIF 0 Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2018. 0 1 1 2 2 3 3 4 4 5 MILES 5 KILOMETERS Figure 16.  Satellite image of Santa Clara Valley, California, at the southern end of San Francisco Bay, annotated with a map of the San Jose Water Company’s buried water-pipeline system (red). Image is overlaid with liquefaction probability for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. 122.2° 122° 121.8° 37.4° San Jose 37.3° EXPLANATION Landslide probability, in percent 0 2–15 0–2 15–32 >32 37.2° 0 0 1 1 2 2 3 3 4 4 5 MILES 5 KILOMETERS Area of map CALIF Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2015. Figure 17.  Satellite image of Santa Clara Valley, California, at the southern end of San Francisco Bay, annotated with a map of the San Jose Water Company’s buried water-pipeline system (red). Image is overlaid with landslide probability for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 233 122.2° 122° 121.8° 121.6° 37.4° 37.3° 37.2° Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2015. 2 0 0 Area of map 2 4 4 6 6 8 8 10 MILES 10 KILOMETERS CALIF Figure 18.  Satellite image of Santa Clara Valley, California, at the southern end of San Francisco Bay, annotated with a map of the San Jose Water Company’s buried water-pipeline system (red). Image is overlaid with peak ground velocity contours (white lines; in increments of 8 centimeters per second) for a moment-magnitude-5.98 aftershock that occurs beneath the City of Mountain View 5 months after the mainshock of the HayWired earthquake scenario. of number of pipe repairs by event in the HayWired scenario earthquake sequence and table 15 for subtotals by day. In those tables, “large diameter” means at least 20 inches. Table 16 summarizes the expected value of the number of repairs by material, summing damage over the entire HayWired earthquake sequence. The table shows that the plurality of repairs are in asbestos cement pipe (481 breaks or leaks), and although the next-largest contributor is ductile-iron pipe (470 breaks or leaks), repairs are disproportionately from damage to asbestos-cement pipe, with an expected value of 0.23 repairs per 1,000 linear feet of pipe (0.75 per km) versus 0.11 repairs per 1,000 linear feet (0.37 per km) for ductile-iron pipe. The unsurprising implication is that it is better to have ductile-iron water pipe than asbestoscement water pipe. Table 13.  Damage estimates for San Jose Water Company, California, buried water pipeline in the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. [%, percent] Description Mean number of repairs Repairs per kilometer of pipe Repairs due to wave passage Repairs due to liquefaction Repairs due to landslide Damage to large diameter pipe (≥20 inch diameter) Damage to small diameter pipe (<20 inch diameter) Breaks Leaks Number 1,054 0.27 665 (63%) 345 (33%) 44 (4%) 30 (3%) 1,024 (97%) 294 (28%) 760 (72%) 234   The HayWired Earthquake Scenario—Engineering Implications Table 14.  Estimated number of leaks and breaks in San Jose Water Company, California, buried water pipeline in the HayWired scenario earthquake sequence (see table 12). Table 15.  Total leaks and breaks by day in San Jose Water Company, California, buried water pipeline in the HayWired scenario earthquake sequence (see table 12). [Day 1 corresponds to April 18, 2018. Mw, moment magnitude] [Day 1 corresponds to April 18, 2018] Day 1 1 1 12 15 32 40 40 41 41 67 74 166 166 166 166 492 Total Epicenter Name Mw Oakland Union City San Pablo Fairfield Fremont Oakland Palo Alto Menlo Park Palo Alto Atherton Palo Alto Palo Alto Mountain View Cupertino Sunnyvale Santa Clara Palo Alto Mainshock uc523 sp504 ff558 fr51 ok542 pa62 mp552 pa569 at511 pa522 pa526 mv598 cu64 sv535 sc509 pa501 7.05 5.23 5.04 5.58 5.10 5.42 6.21 5.52 5.69 5.11 5.22 5.26 5.98 6.40 5.35 5.09 5.01 Large Small Leaks + diameter diameter breaks pipe pipe 1,054 30 1,024 34 1 33 6 0 6 2 0 2 47 1 46 30 1 29 102 3 99 30 1 29 58 2 56 30 1 29 47 2 45 48 2 46 93 3 90 172 6 166 73 2 71 102 3 98 29 1 28 1,957 59 1,897 Day Total leaks + breaks 1 12 15 32 40 41 67 74 166 492 Total 1,094 2 47 30 132 88 47 48 440 29 1,957 Large diameter pipe 31 0 1 1 4 3 2 2 14 1 59 Table 16.  Repair rate by material for San Jose Water Company, California, buried water pipeline in the HayWired scenario earthquake sequence (see table 12). Material1 AC BCL CCCL CI CL CU DCCL DCIL DFK DICL DIMCL DS FKCL GALV GG HDPE PB Repairs per 1,000 linear feet 0.23 0.13 0.19 0.19 0.12 0.31 0.14 0.09 0.23 0.11 0.25 0.07 0.13 0.08 0.23 0.05 0.08 Repairs per kilometer 0.75 0.42 0.62 0.62 0.40 1.00 0.45 0.29 0.74 0.37 0.81 0.21 0.41 0.26 0.74 0.16 0.27 Material1 PE PP PVC RCP S SB SG SI SOMCL SS TBD WI WS WSCL ZCCL Total Repairs per 1,000 linear feet 0.06 0.05 0.09 0.08 0.13 0.15 0.12 0.14 0.16 0.19 0.18 0.22 0.12 0.13 0.07 0.15 Repairs per kilometer 0.20 0.17 0.29 0.26 0.42 0.50 0.39 0.45 0.53 0.63 0.60 0.71 0.39 0.43 0.24 0.50 See table 11 for explanation of material codes. 1 Small diameter pipe 1,063 2 46 29 127 85 45 46 426 28 1,897 Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 235 The estimates in tables 14 and 15 ignore the potential for liquefaction outside the area with estimated liquefaction probability (that is, in places where liquefaction could occur but HayWired has no maps). They also ignore damage from ground failure in aftershocks, for which liquefaction and landslide probability were not estimated. However, because liquefaction requires long duration as well as strong shaking, and because aftershocks would tend to have short duration because of their moderate and small magnitude, they would tend to produce relatively few pipeline repairs as a result of liquefaction. Note that after the mainshock, the Mw 6.4 aftershock near the City of Cupertino (table 14) adds the largest number of aftershock-related breaks and leaks in buried pipelines, likely setting SJWC back in restoring service. Figure 19 is a heatmap (warmer color indicates greater concentration of damage) of SJWC pipeline repair rate in the HayWired scenario mainshock. To be clear, if any additional emphasis is needed, this map shows estimated damage rates in one scenario earthquake—the HayWired mainshock—not all possible earthquakes, not even all possible M7.0 earthquakes on the Hayward Fault. Different earthquakes produce different damage patterns. However, the point of a scenario is to understand what might realistically happen, and a heatmap makes a possible outcome more tangible, more useful for planning purposes. By planning for one scenario, one becomes more prepared for what actually happens, which will invariably differ from a scenario. Figure 19 unsurprisingly shows greater damage near the fault and on soil with high liquefaction probability, with maximum values approaching 12 breaks or leaks per square kilometer (km2). Figure 20 shows an analogous map for the Mw 6.4 Cupertino aftershock. Damage rates just exceed 1 break or leak per square kilometer in the aftershock in the neighborhoods 122° 121°45' Milpitas Palo Alto Area of map Mountain View Sunnyvale CALIF Santa Clara Los Altos Cupertino San Jose Campbell Saratoga 37°15' Los Gatos EXPLANATION Mean pipeline repairs per square kilometer <2 6–8 2–4 8–10 4–6 Lexington Reservoir Calero Reservoir 10–12 Major roadways City boundaries Water Hillshade derived from U.S. Geological Survey National Elevation Dataset, 2013. Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. 0 0 2 2 4 MILES 4 KILOMETERS Figure 19.  Map of buried water-pipeline damage in San Jose Water Company’s, California’s, service area for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Colors indicate mean repairs (breaks and leaks) per square kilometer. A warmer color indicates greater concentration of damage. To be informative, the color scale is shifted from those used in figures 20 and 21. 236   The HayWired Earthquake Scenario—Engineering Implications 122° 121°45' Milpitas Palo Alto Area of map Mountain View Sunnyvale CALIF Santa Clara Los Altos Cupertino San Jose Campbell Saratoga 37°15' Los Gatos EXPLANATION Mean pipeline repairs per square kilometer 0.2 0.6–0.8 0.2–0.4 0.8–1.0 0.4–0.6 Lexington Reservoir Calero Reservoir 1.0–1.1 Major roadways City boundaries Water Hillshade derived from U.S. Geological Survey National Elevation Dataset, 2013. Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. 0 0 2 2 4 MILES 4 KILOMETERS Figure 20.  Map of buried water-pipeline damage in San Jose Water Company’s, California’s, service area for the hypothetical moment-magnitude-6.4 Cupertino aftershock in the HayWired scenario earthquake sequence. Colors indicate mean repairs (breaks and leaks) per square kilometer. A warmer color indicates greater concentration of damage. To be informative, the color scale is shifted from those used in figures 19 and 21. along the northern edge of the SJWC service area. Figure 21 shows the heatmap for the entire sequence, with damage rates of approximately 15 repairs/km2 in the northeastern part of the service area. I simulated damage locations by applying equation 49 once for the mainshock and once for each aftershock. Figure 22 shows pipeline segments with at least one simulated repair resulting from the Mw 7.0 HayWired mainshock. Figure 23 shows pipeline segments with at least one simulated repair resulting from the Mw 6.4 Cupertino aftershock. Figure 24 shows pipeline locations with at least one simulated repair resulting from the entire HayWired earthquake sequence. San Jose Water Company Restoration Analysis I take the following g-value (flow factor as discussed earlier) time series for lifelines upstream of water, and iterate later if necessary: Consumables.— SJWC has one of the best stock of repair materials in the Bay Area (J. Wollbrinck, SJWC, oral commun., October 14, 2015). I assume sufficient repair consumable materials (such as pipe and clamps) are on hand or can be acquired as they are needed, that is, g(t)=1.0 for all t. Fuel.—As of this writing, SJWC is in the process of preparing its fuel plan (J. Wollbrinck, SJWC, oral commun., Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 237 122° 121°45' Milpitas Palo Alto Area of map Mountain View Sunnyvale CALIF Santa Clara Los Altos Cupertino San Jose Campbell Saratoga 37°15' Los Gatos EXPLANATION Mean pipeline repairs per square kilometer <2 6–8 2–4 8–10 4–6 Lexington Reservoir Calero Reservoir 10–12 Major roadways City boundaries Water Hillshade derived from U.S. Geological Survey National Elevation Dataset, 2013. Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. 0 0 2 2 4 MILES 4 KILOMETERS Figure 21.  Map of buried water-pipeline damage in San Jose Water Company’s, California’s, service area for the entire hypothetical HayWired scenario earthquake sequence (events larger than moment-magnitude 5). Colors indicate mean repairs (breaks and leaks) per square kilometer. A warmer color indicates greater concentration of damage. To be informative, the color scale is shifted from those used in figures 19 and 20. October 14, 2015). I treat two possible outcomes—(1) the earthquake (Mw 7.0 HayWired scenario mainshock) happens before the plan is implemented and (2) the earthquake happens afterwards. If afterwards, I assume that the fuel plan is sufficient to ensure adequate supplies throughout the repair and restoration process, in which case g(t)=1.0 for all t. Without the fuel plan, I assume that there is sufficient fuel initially, but that shortages would impair restoration for a few days until emergency supplies were secured. Quantitatively, I assume that before implementing fuel plan: g(t)=1.0 for 0≤t<3 days, g(t)=0.25 for 3≤t<7 days, g(t)=1.0 for t>7 days. After implementing a fuel plan, I assume g(t)=1.0. Electricity.—Pacific Gas and Electric Company (PG&E) was unable to offer a public estimate of the time required to restore power throughout the San Francisco Bay area after the HayWired scenario mainshock. On the basis, in part, of a Hazus-MH analysis, the HayWired project team and SJWC’s emergency manager believe it is realistic that 99.9 percent of customers in Santa Clara County will have power restored within 238   The HayWired Earthquake Scenario—Engineering Implications 122.2° 122° 121.8° 37.4° s 37.3° 37.2° Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2015. 2 0 Area of map 0 2 4 4 6 6 8 8 10 MILES 10 KILOMETERS CALIF Figure 22.  Satellite image of Santa Clara Valley, California, at the southern end of San Francisco Bay, annotated with a map of San Jose Water Company’s buried water-pipeline system. Simulated repairs (red circles) resulting from the hypothetical momentmagnitude-7.0 Cupertino aftershock in the HayWired scenario earthquake sequence are shown as red circles. 122.2° 122° 121.8° 37.4° San Jose 37.3° 37.2° Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2015. 2 0 Area of map CALIF 0 2 4 4 6 6 8 8 10 MILES 10 KILOMETERS Figure 23.  Satellite image of Santa Clara Valley, California, at the southern end of San Francisco Bay, annotated with a map of San Jose Water Company’s buried water-pipeline system. Simulated repairs (red circles) resulting from the hypothetical momentmagnitude-6.4 mainshock of the HayWired earthquake scenario are shown as red circles. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 239 122.2° 122° 121.8° 37.4° 37.3° 37.2° Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2015. 2 0 Area of map 0 2 4 4 6 6 8 8 10 MILES 10 KILOMETERS CALIF Figure 24.  Satellite image of Santa Clara Valley, California, at the southern end of San Francisco Bay, annotated with a map of San Jose Water Company’s buried water-pipeline system. Simulated repairs (red circles) resulting from the entire hypothetical HayWired scenario earthquake sequence are shown as red circles. 2 weeks of the mainshock or 2 days after a M6+ aftershock. Quantitatively, I therefore take: g(t)=1–exp(-0.4934×t), t measured in days after the mainshock, g(t)=1–exp(-3.45×t), t in days after the aftershocks on days 40 and 166. Communication.—SJWC has battery powered radios for its repair crews that reach almost its entire service area. I assume that communication to facilitate coordination between utilities will be such a high priority that coordination will not be a constraint. I therefore assume g(t)=1.0. Crews.—SJWC personnel estimated that they could realistically field between 20 and 25 crews on a work basis of 12 hours on and 12 hours off. I therefore take w(t)=0.5 and c(t)=22 for purposes of equation 35. I consider one more remediation measure. Suppose all of the more vulnerable SJWC buried pipe (especially 609 miles of asbestos-cement and cast-iron pipe) could be replaced with less-vulnerable pipe (for example, ductile-iron or plastic pipe). What would be the benefit in terms of damage reduction and accelerated recovery? To explore this question, I assume that all asbestos cement and cast-iron pipe is replaced with ductileiron or plastic pipe before the HayWired earthquake sequence occurs. I refer to this as the “ideal world” assumption. An SJWC engineer (J. Walsh, SJWC, oral commun., October 2, 2015) informed the HayWired project team that SJWC replaces 1 percent, or 24  miles, of existing water mains every year. He did not imagine this study being used to change that percentage of replacement, but they might change their mix of pipes. Their replacement program is based on both consequences of failure and probability of failure. They consider a multitude of factors and then apply genetic algorithm software to predict leaks. They will likely add additional weighting to asbestos-cement and cast-iron pipe in close proximity to earthquake fault lines based on the present work. If SJWC were to focus on asbestos-cement and cast-iron pipe and were to continue to replace 24 miles of pipe per year, all 609 miles of asbestos-cement and cast-iron pipe could be replaced within 25  years. Figure 25 shows the repair timeline for SJWC water-supply pipelines before and after implementing the fuel-management plan and after replacing all cast-iron and asbestos-cement pipe with ductile-iron or plastic pipe (the “ideal world”). Figure 26 illustrates the simulated restoration curve. If the Hazus-MH serviceability index realistically measures the fraction of services receiving any water, as reports produced by the Hazus software suggest, then “services available” in figure 26 measures the fraction of service connections receiving even small flows. If it means the postearthquake flow as a fraction of pre-earthquake flow, then figure 26 underestimates the number of service connections receiving at least some water. 240   The HayWired Earthquake Scenario—Engineering Implications 1,500 250,000 Number of water-service connections available EXPLANATION Ideal world With fuel plan No plan Fuel shortage 1,000 Repairs remaining Aftershocks 500 0 1 10 100 1,000 200,000 Aftershocks 150,000 Fuel shortage EXPLANATION Ideal world With fuel plan No plan 100,000 50,000 0 1 Figure 25.  Graph showing simulated repair timeline of San Jose Water Company, California, water-supply pipelines under three conditions for the HayWired scenario earthquake sequence. Conditions are (1) ideal world (all asbestos cement and cast-iron pipe is replaced with ductile-iron or plastic pipe before the earthquake sequence occurs and with a fuelmanagement plan), (2) with a fuel-management plan but no pipe replacement, and (3) without fuel-management plan or pipe replacement. As discussed earlier, I view the area above the curves in figure 26 as a measure of resilience—less area means less impact, faster recovery, or both. The areas above the three curves are measured in units of service-days. That is, each day of lost water supply to a service connection equates with one service day. The areas above the three curves are shown in table 17—lost service days under as-is conditions, with a fuel-management plan, and under ideal-world conditions, that is, assuming that all brittle pipe is replaced before the HayWired scenario mainshock. The table shows the lost service days as a multiple of number of water service connections, that is, the average number of days that each service connection is without potable water under as-is conditions, with a fuel-management plan, and under ideal-world conditions. The difference between lost service days under as-is and what-if conditions (fuel-management plan or ideal world) measures the resilience benefit of the what-if condition—with a fuel-management plan and if all brittle pipe were replaced before the HayWired scenario mainshock occurs. Validation of San Jose Water Company Restoration Analysis It is possible to compare the foregoing results for SJWC with other analyses as an initial check of the results. This validation is discussed below. 10 100 1,000 Time after mainshock, in days Time after mainshock, in days Figure 26.  Graph showing simulated service availability of San Jose Water Company, California, water-supply system under three conditions for the HayWired scenario earthquake sequence. Conditions are (1) ideal world (all asbestos cement and cast-iron pipe is replaced with ductile-iron or plastic pipe before the earthquake sequence occurs and with a fuelmanagement plan), (2) with a fuel-management plan but no pipe replacement, and (3) without fuel-management plan or pipe replacement. Cross Validation with San Jose Water Company’s Internal Damage Estimate By scaling up the number of water pipeline breaks and leaks in the Mw 6.0 2014 South Napa earthquake, SJWC personnel estimated that the Mw 7.0 HayWired scenario mainshock would cause 1,200 water main breaks or leaks to their company’s system (J. Wollbrinck written commun., SJWC, October 19, 2015). The estimate follows this logic—the City of Napa has 370 miles of Table 17.  San Jose Water Company, California, total and per-customer average lost service days following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [Numbers rounded to reduce the impression of excessive precision. Ideal world—all asbestos cement and cast-iron pipe is replaced with ductile-iron or plastic pipe before the earthquake sequence occurs and with fuel-management plan. R, number of days that service connection is without potable water; M, number of water service connections] Condition As-is With fuelmanagement plan Ideal world Lost service days, R×M 940,000 750,000 Resilience benefit, D(R×M) 0 190,000 Average lost service days (R) 4 3 470,000 470,000 2 Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 241 Validation Against Northridge, Kobe, and South Napa Earthquakes Jeon and O’Rourke (2005) report that the 1994 Northridge earthquake caused 1,095 breaks or leaks to buried pipeline operated by the Los Angeles Department of Water and Power, most of the damage occurring in the San Fernando Valley. Lund and others (2005) report that repairs took about 1 week. The present calculation suggests that SJWC would take 23 days to repair 1,176 repairs (mainshock plus 4 aftershocks through day 15)—that is, about 3 times as long for about the same number of repairs. Lund and others (2005) do not report the number of crews required to perform those repairs. Presumably LADWP fielded more crews than SJWC has at its disposal. Lund and others (2005) report that, according to M. Matsushita of the Kobe Municipal Waterworks Bureau, 1,757 breaks and leaks occurred in buried water-supply distribution pipe in Kobe after the Mw 6.9 1995 Kobe earthquake and that repairs took 10 weeks. The present estimate of 3 weeks to repair 1,176 breaks and leaks suggests one-third the time to repair two-thirds of the breaks and leaks. Thus, in a sense, the water-pipeline repair estimates for the Northridge and Kobe earthquakes bracket the restoration estimates for SJWC presented here. The City of Napa repaired approximately 120 leaks and breaks in 5 days with approximately 10 crews working 12-hour shifts (SPA Risk LLC, 2014), for a repair productivity of approximately 2.4 repairs per crew-day. The present model suggests that, before its fuel plan is implemented, San Jose Water Company would repair 1,176 breaks and leaks in 26 days with 22 crews working 12-hour shifts, or 2.1 repairs per crew day, suggesting fairly good agreement. Cross Validation with Hazus-MH Using Hazus-MH (Federal Emergency Management Agency, 2012), the estimated restoration of water supply in Santa Clara County following the HayWired scenario mainshock (table 18). The estimates are for the mainshock only and do not reflect lifeline interaction. Applying the percentages to the number of SJWC’s customers, one can compare the HayWired scenario model with fuel management plan discussed in this chapter with the Hazus-MH model (fig. 27). The present model and Hazus-MH disagree wildly in terms of initial level of water service and restoration time, with the Hazus-MH model estimating a six-times increase in time to restore service compared with the present model. Considering the validation against the Napa repair timeline, the present restoration model seems more plausible than the Hazus-MH restoration model. Why would Hazus-MH’s restoration model differ so markedly from the present one? The difference can be partly explained by the user-specified number of repair crews. Table 18.  Hazus-MH estimate of Santa Clara County, California, loss of water supply following the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario (Federal Emergency Management Agency, 2012). [%, percent] Total Households without potable water following mainshock households Day 1 Day 3 Day 7 Day 30 Day 90 565,853 504,596 502,302 497,394 458,220 137,185 89.20% 88.80% 87.90% 81.00% 24.20% 250,000 HayWired model—with fuel plan Number of water-service connections available water main and experienced 120 pipeline leaks the first week and more than 170 pipeline leaks in the first 6 months following the 2014 earthquake. Scaling up by system size, SJWC estimates that Napa’s 120 leaks would equate with 850 leaks for SJWC and the 170 leaks would equate with 1,200 leaks for SJWC. The South Napa earthquake was weaker than the mainshock modeled in the HayWired scenario, so the number of leaks could be more. The similarity between SJWC’s estimates of 850 increasing to 1,200 and the ones produced here (1,054 increasing to 1,956) suggests that either or both are reasonable or neither is. That the two set of figures used two different approaches to arrive at a basically similar set of numbers tends to support both being reasonable, rather than neither. In either case, SJWC engineers found the results presented here to be reasonable (J. Walsh, written commun., SJWC, October 19, 2015), with the exception that Wollbrinck (written commun., SJWC, December 4, 2015) expected more damage to wrapped steel pipe because of its age and corrosion susceptibility. 200,000 150,000 Hazus-MH model—without aftershocks or lifeline interaction 100,000 50,000 0 1 10 100 1,000 Time after mainshock, in days Figure 27.  Graph comparing the estimated restoration of water supply in Santa Clara County, California, between the HayWired scenario model with fuel management plan and the Hazus-MH model (Federal Emergency Management Agency, 2012) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. 242   The HayWired Earthquake Scenario—Engineering Implications Effect of Lifeline Interaction and Consumable Limits Suppose one ignored limitations in fuel and other consumables, and ignored impairment of electricity, telecommunications, and so on. How much difference do lifeline interaction and consumable limits make? Without these effects, damage is unaffected, but repairs could potentially proceed faster. Recall that equation 36 gives the time τ(n,t) required to perform the nth repair, which occurs at time t. It can be expressed as a baseline productivity increased by a factor S(t) that accounts for how lifeline interaction and consumable limits slow repairs, that is: d0 τ ( n, t ) = × S (t ) , w (t ) × c (t ) where S (t ) = 1 ∏ (1 − u × (1 − g (t ))) i i (69) .  (70) i The first multiplicand in equation 69 is the baseline productivity, that is, the time required to perform the nth repair, without lifeline interaction and consumable limits. The factor S(t), which is always at least 1.0, increases the repair time to account for lifeline interaction and consumable limits. Repeating the calculations for SJWC under as-is conditions but with S(t)=1 produces an estimated 740,000 lost service days, about 80 percent of the value estimated considering lifeline interaction. Viewed another way, lifeline interaction and consumable limits decrease the calculated water-supply resilience in the area served by San Jose Water Company by 25 percent in the HayWired scenario. The factor would vary in other earthquakes, generally being larger the more dependent the utility is on other lifelines and consumables and the more these lifelines and consumables are impaired. Case Study 2—East Bay Municipal Utility District A second water system subjected to the same hypothetical earthquake sequence can be examined to see whether the model can produce plausible results twice. This section considers the East Bay Municipal Utility District’s (EBMUD) water-supply buried pipeline network. East Bay Municipal Utility District Asset Definition The following description is largely quoted from Contra Costa Local Agency Formation Commission (2008) and from conversations with EBMUD. EBMUD provides water- and sewage- treatment services for an area of approximately 331 square miles in the eastern side of the San Francisco Bay. EBMUD serves approximately 1.3 million people in portions of Alameda County and Contra Costa County in California. EBMUD currently has an average annual growth rate of 0.8 percent and is projected to serve 1.6 million people by 2030. As of 2015 it provides water to approximately 390,000 service connections. Approximately 100,000 service connections are located east of the East Bay Hills, and the other 290,000 service connections are to the west between the hills and San Francisco Bay. EBMUD’s administrative offices are located in The City of Oakland. EBMUD owns and maintains: • 2 water storage reservoirs on the Mokelumne River, • 5 terminal reservoirs, • 91 miles of three separate water-transmission aqueducts, • 4,162 miles of water mains (the only part of the system modeled here), • 6 water-treatment plants, • 29 miles of wastewater interceptor sewer lines, and • A wastewater treatment facility rated at a maximum treatment capacity of 320 million gallons a day. EBMUD provided an ArcGIS map of its water mains. The system is shown in figure 28. The map shows 6,698  km (4,162 miles) of pipe of various types and lengths. Approximately 2,091 km of pipe are located east of the East Bay Hills, and the other 4,607 km of pipe are to the west between the hills and San Francisco Bay. Total quantities of pipe are summarized by material in table 19 and by diameter in table 20. EBMUD’s system is discretized into segments with an average length of 64 m and a standard deviation of 79 m—short enough that earthquake shaking should vary little between ends of segments. Some of EBMUD’s pipe does not map well to an Eidinger type. One of the material codes do not appear in EBMUD’s glossary of pipe types and is probably a data-entry error. I have made a reasonable assumption about the intended meaning, but in any case, the total quantity is small—0.1 miles. Terentieff and others (2015) report that 176,000 of 390,000 water services are in pumped pressure zones. These pressure zones rely on 130 pumping stations, of which 117 (90 percent) have no emergency generators. Therefore, I set the parameter z used in the lifeline interaction matrix for EBMUD to be 0.9×176,000/390,000=0.41, and uelectr=0.41+0.03=0.44. The factor 0.9 reflects the 90 percent of pumping stations that have no generator. Conceivably EBMUD could install emergency generators with large fuel tanks at all its pumping stations, in which case I can take uelectr=0.03. I take the former as the “real-world” scenario and the latter as an “ideal-world” scenario. I also assume that fuel limitations affect EBMUD the same as SJWC, and like SJWC, EBMUD can optionally develop a fuel-management plan and storage to ensure that fuel does not limit its ability to perform pipeline repairs. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 243 122.50° 122° Figure 28. Satellite image of the San Francisco Bay region, California, annotated with a map of the East Bay Municipal Utility District’s buried water-pipeline system (red) with dividing line (yellow) to approximately separate pipe and services east and west of the East Bay Hills. 38° 37.75° SAN FRANCISCO BAY Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2016. 0 Area of map 0 2 2 4 4 6 6 8 8 10 MILES 10 KILOMETERS CALIF Table 19.  East Bay Municipal Utility District, California, pipe construction, associated with Eidinger (2001) vulnerability functions. [Code, code for pipe material; count, number of segments of that material in East Bay Municipal Utility District buried water-pipeline system; miles, total length of pipe material; material description, description of pipe material; Eidinger type and ID, assumed corresponding vulnerability functions of Eidinger (2001) and their associated vulnerability factors K1 and K2 from table 1; PVC, polyvinyl chloride] Code A C D F H K L N P3 R S T W Total Count 24,543 33,747 43 30 167 50 197 8,613 1 2 37,101 127 71 104,692 Miles 1,136.4 1,322.1 2.1 1.2 8.8 0.7 14.3 380.4 0.1 0.0 1282.4 10.4 2.7 4,162 Material, joint description1 Asbestos cement, unrestrained Cast iron, unrestrained Ductile iron, unrestrained Fusible PVC, welded High-density polyethylene, weld Copper, restrained Reinforced concrete cylinder, unrestrained PVC, unrestrained Pretensioned concrete cylinder, restrained Reinforced concrete noncylinder Steel, welded Pretensioned concrete cylinder, restrained Wrought iron Eidinger (2001) type2 Asbestos cement with cement joint Cast iron with cement gasket Ductile iron with rubber gasket Welded steel with lap-arc welded joint PVC with rubber gasket Welded steel lap-arc weld joint small diameter Concrete with steel-cylinder cement joint PVC with rubber gasket Concrete with steel-cylinder cement joint Concrete with steel-cylinder cement joint Welded steel with lap-arc welded joint Concrete with steel-cylinder lap-arc welded joint Cast iron with cement gasket 1 East Bay Municipal Utility District pipe material and description of joint; descriptions in italics are assumptions. 2 Closest equivalent corresponding vulnerability function from table 1. 3 One length of 48-inch-diameter pipe from 1927; P is probably a typo for T. 4 ID 6=small diameter (<20 inches); ID 9=large diameter (≥20 inches). ID2 14 1 19 6 18 6 16 18 16 16 6, 94 15 1 244   The HayWired Earthquake Scenario—Engineering Implications Table 20.  Quantities of pipe by diameter in the East Bay Municipal Utility District, California, buried water-pipeline system. Diameter, in inches 0.00 0.75 1.00 2.00 3.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 24.00 Length, in miles 0.4 0.2 1.0 18.6 0.7 294.6 1,728.2 1,105.7 38.0 475.8 1.4 157.6 1.4 78.1 74.4 Diameter, in inches 25.00 30.00 36.00 42.00 48.00 54.00 60.00 66.00 69.00 72.00 78.00 84.00 90.00 96.00 108.00 Total EBMUD Hazard Analysis Ground motion, liquefaction, landslide, coseismic slip, and afterslip for the HayWired scenario are quantified in Detweiler and Wein (2017). Figure  29 shows EBMUD’s buried waterpipeline system and PGV for the Mw 7.0 HayWired scenario mainshock. Figure 30 shows EBMUD’s buried water-pipeline system with liquefaction probabilities for the HayWired scenario mainshock. Liquefaction probability was not calculated for Contra Costa County, but it is assumed that liquefaction damage occurs in Contra Costa County in approximately the same proportion to shaking-induced damage as in Alameda County. Landslide probability in the EBMUD service area is mapped in figure 31. See figure 32 for a map of EBMUD’s water-supply buried pipeline system with a fence diagram showing coseismic slip. Figure 33 shows PGV in one of the more damaging HayWired scenario aftershocks, a Mw 5.4 earthquake with an epicenter near Oakland. The HayWired scenario does not have a map of afterslip, which progresses with time and varies spatially along the Hayward Fault. For present purposes, I assume that for most of the fault length, total slip (coseismic slip plus afterslip) equals 1.9 m, except where coseismic slip exceeds that amount. Afterslip evolves Length, in miles 0.5 36.5 66.8 18.5 38.3 8.8 2.6 6.8 4.6 0.0 0.2 1.8 0.2 0.1 0.0 4,161.7 122.5° 122° 38° 37.75° 0.0 0.5 1.0 1.5 2.0 PGV, in meters per second 0 Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2016. Area of map 0 2 2 4 4 6 6 8 8 10 MILES 10 KILOMETERS CALIF Figure 29.  Satellite image of the San Francisco Bay region, California, annotated with a map of the East Bay Municipal Utility District buried water-pipeline system (red). Image is overlaid with mainshock peak ground velocities (PGV) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 245 122.5° 122° EXPLANATION Liquefaction probability Not assigned < 0.05 38° 0.05–0.1 0.1–0.2 0.2–0.3 0.3–0.4 0.4–0.5 >0.5 37.75° SAN FRANCISCO BAY Area of map 0 2 6 4 2 0 Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2018. 4 6 8 8 10 MILES 10 KILOMETERS CALIF Figure 30.  Satellite image of the San Francisco Bay region, California, annotated with a map of the East Bay Municipal Utility District buried water-pipeline system (red). Image is overlaid with liquefaction probability for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. over time following equations 6–9 in Aagaard and others (2012), with coseismic slip=0.09845, calculated using their figure 11 and Mw=7.0. EBMUD Damage Analysis Table 21 summarizes the estimated damage to EBMUD buried water pipelines from the HayWired scenario mainshock and aftershocks. Damage in the mainshock includes ground shaking, liquefaction, landsliding, and surface rupture. Because the liquefaction map does not include Contra Costa County, liquefaction damage in Contra Costa County is assumed to be proportional to earthquake shaking damage in Contra Costa County, in the same proportions as in Alameda County. An analysis of fire following earthquake for the HayWired scenario mainshock (Scawthorn, this volume) requires that the assumed liquefaction damage in Contra Costa County be assigned to particular locations; the damage is assigned to the cities of Pinole, Hercules, and Rodeo, where PGV values are very high and time-averaged shear-wave velocity to a depth of 30 m (VS30) values are low. As table 21 illustrates, aftershock damage ignores landslides, liquefaction, and surface rupture. Note that the mainshock damage estimate for afterslip assumes that pipes that are ruptured by fault slip are damaged a second time by afterslip, EBMUD may decide to either install earthquake-resistant pipe (for example, highdensity polyethylene-plastic pipe or steel pipe with flexible joints that can tolerate extension, compression, and lateral deformation) or to temporarily install flexible hose until the damaged water main can be repaired or earthquake-resistant pipe can be installed. The table shows that the mainshock produces the majority of the overall damage but that after the mainshock 36 percent more damage occurs in aftershocks. Half the mainshock damage is associated with wave passage, the other half to liquefaction, landsliding, and fault offset. 246   The HayWired Earthquake Scenario—Engineering Implications 122.5° 122° 121.5° 38° EXPLANATION 37.75° Landslide probability, in percent 0 0–2 2–15 15–32 >32 SAN FRANCISCO BAY 0 Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2016. 2 4 6 8 10 MILES 0 2 4 6 8 10 KILOMETERS Area of map CALIF Figure 31.  Satellite image of the San Francisco Bay region, California, annotated with a map of the East Bay Municipal Utility District (EBMUD) buried water-pipeline system (red). Image is overlaid with landslide probabilities near the EBMUD service area for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. SAN FRANCISCO BAY 2 0 Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2016. Area of map 0 2 4 4 6 6 MILES 8 KILOMETERS CALIF Figure 32.  Oblique satellite image of the east bay part of California’s San Francisco Bay area, California, annotated with a map of the East Bay Municipal Utility District (EBMUD) buried water-pipeline system (red). Image is overlaid with a red “fence” along the northern Hayward Fault. The height of the red fence represents the right-lateral surface slip occurring on the Hayward Fault at the time of the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The tallest point on the fence, near the location where Interstate 80 crosses the fault between Richmond and Pinole, represents 2.1 meters of offset. At California Route 24 near Berkeley, the offset is 0.84 m. At Interstate 238 near Castro Valley, the offset is approximately 1.68 m. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 247 122.5° 122° 38° Oakland 37.75° SAN FRANCISCO BAY 0 Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2016. 0 Area of map 2 2 4 4 6 6 8 8 10 MILES 10 KILOMETERS CALIF Figure 33.  Satellite image of the San Francisco Bay region, California, annotated with a map of the East Bay Municipal Utility District (EBMUD) buried water-pipeline system (red). Image is overlaid with peak ground velocity contours (white lines; in increments of 5 centimeters per second) for a moment-magnitude-5.42 aftershock that occurs beneath the City of Oakland 32 days after the mainshock of the HayWired earthquake scenario. Table 21.  Mean damage estimate for East Bay Municipal Utility District, California, buried water pipeline in the moment-magnitude-7 mainshock and aftershocks HayWired scenario earthquake sequence (see table 12). [%, percent] Mainshock Mean number of repairs Repairs per kilometers of pipe 4,294 Aftershocks 1,395 Total 5,688 0.64 0.21 0.85 Due to wave passage 2,037 (47%) 1,395 (100%) 3,432 (60%) Due to liquefaction 1,642 (38%) Not calculated 1,642 (29%) Due to landslide 185 (4%) Not calculated 185 (3%) Due to coseismic slip 214 (5%) Not calculated 214 (4%) Due to afterslip 214 (5%) Not calculated 214 (4%) Large diameter (≥20 inches diameter) 218 (5%) Small diameter (<20 inches diameter) 4,076 (95%) Breaks Leaks 84 (6%) 302 (5%) 1,311 (94%) 5,386 (95%) 1,582 (37%) 209 (15%) 1,791 (31%) 2,712 (63%) 1,185 (85%) 3,898 (69%) 248   The HayWired Earthquake Scenario—Engineering Implications Table 22 presents the estimated number of leaks plus breaks in EBMUD buried pipeline in each earthquake of the HayWired sequence. After the mainshock, 6 of 16 aftershocks each produces at least 100 breaks or leaks, and one produces more than 300, with significant numbers of breaks and leaks occurring almost 6 months after the mainshock, a point that is made somewhat clearer by table 23, which summarizes damage by day after the mainshock. Table 24 details repair rate over the entire HayWired sequence by pipe material. Unsurprisingly, it shows that most damage is in brittle cast-iron and asbestos-cement pipe, with damage rates approaching 0.3 per 1,000 linear feet. Table 25 presents repair rate over the entire sequence by pipe diameter, with most repairs required in 6-inch- and 8-inch- diameter pipe, which together represent the majority of pipe in the system. Table 22.  Estimated number of leaks plus breaks in East Bay Municipal Utility District, California, buried water pipeline in the HayWired scenario earthquake sequence (see table 12). [Day 1 corresponds to April 18, 2018. Mw, moment magnitude] Day Epicenter Name Mw 1 1 1 12 15 32 40 40 41 41 67 74 166 166 166 166 492 Total Oakland San Pablo Union City Fairfield Fremont Oakland Menlo Park Palo Alto Atherton Palo Alto Palo Alto Palo Alto Cupertino Mountain View Santa Clara Sunnyvale Palo Alto Mainshock sp504 uc523 ff558 fr510 ok542 mp552 pa62_ pa569 at511 pa522 pa526 sc509 cu640 sv535 mv598 pa501 7.0 5.04 5.23 5.58 5.1 5.42 5.52 6.2 5.11 5.69 5.22 5.26 6.4 5.98 5.09 5.35 5.01 Leaks + breaks 4,294 102 101 49 37 323 44 141 61 44 54 59 25 173 52 102 28 5,688 Diameter ≥20 inches 218 6 6 3 2 20 3 8 4 3 3 4 2 10 3 6 2 302 Diameter <20 inches 4,076 96 95 46 35 304 41 133 57 42 51 55 24 162 49 96 26 5,386 Table 23.  Total leaks plus breaks in East Bay Municipal Utility District, California, buried water pipeline by day in the HayWired scenario earthquake sequence (see table 12). [Day 1 corresponds to April 18, 2018] Day 1 12 15 32 40 41 67 74 166 492 Total Leaks + breaks1 4,496 49 37 323 185 105 54 59 352 28 5,688 Diameter ≥20 inches 230 3 2 20 11 6 3 4 21 2 302 Diameter <20 inches 4,266 46 35 304 174 98 51 55 330 26 5,386 1 Slight differences from numbers in table 22 and summation differences are due to rounding. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 249 Table 24.  Repair rate for East Bay Municipal Utility District, California, buried water pipeline in the HayWired scenario earthquake sequence. Material1 Description Length, in feet Repairs2 Repairs per 1,000 linear feet A Asbestos cement, unrestrained 6,000,452 1,874 0.312 C Cast iron, unrestrained 6,980,888 2,206 0.316 D Ductile iron, unrestrained 10,841 5 0.433 F Fusible PVC, welded 6,478 1 0.156 H High-density polyethylene, weld 46,472 10 0.218 K Copper, restrained 3,646 0 0.135 L Reinforced-concrete cylinder, unrestrained N PVC, unrestrained P Reinforced-concrete cylinder R Reinforced-concrete noncylinder S Steel, welded T W 75,622 22 0.285 2,008,486 438 0.218 372 0 0.081 174 0 0.147 6,771,086 1,119 0.165 Pretensioned concrete cylinder, restrained 55,106 10 0.183 Wrought iron 14,319 4 0.284 21,973,942 5,688 0.259 Total See material codes and descriptions in table 19. 1 Excludes damage estimated to occur in zones of liquefaction in Contra Costa County. 2 Table 25.  Repair rate for East Bay Municipal Utility District, California, buried water pipeline by diameter in the HayWired scenario earthquake sequence. Diameter, in inches 0 Length, in feet 2,201 Repairs1 0 Repairs per 1,000 linear feet1 Diameter, in inches Length, in feet Repairs1 Repairs per 1,000 linear feet1 0.000 25 2,581 0 0.000 192,540 30 0.156 352,938 51 0.145 834 0 0.000 30 1 5,342 3 0.562 36 2 98,006 28 0.286 42 97,882 16 0.163 3 3,665 1 0.273 48 202,222 35 0.173 4 1,555,655 466 0.300 54 46,248 11 0.238 6 9,124,837 2,583 0.283 60 13,778 3 0.218 0.75 8 5,838,048 1,554 0.266 66 35,784 3 0.084 10 200,527 69 0.344 69 24,207 2 0.083 12 2,512,027 501 0.199 72 70 0 0.000 978 0 0.000 14 7,502 2 0.267 78 16 831,998 178 0.214 84 9,598 1 0.104 18 7,394 1 0.135 90 994 0 0.000 20 412,261 83 0.201 96 689 0 0.000 24 393,015 68 0.173 108 122 0 0.000 21,973,942 5,688 0.259 Total 1 Where expected number of repairs is less than 0.5, number of repairs and repair rate per 1,000 linear feet are rounded to 0; table excludes damage assumed to occur in zones of liquefaction in Contra Costa County. 250   The HayWired Earthquake Scenario—Engineering Implications Figures 34, 35, and 36 are damage heatmaps for EBMUD’s system. They map mean repairs (estimated per the present methodology) per square kilometer in the HayWired scenario mainshock, most-damaging aftershock (a Mw 5.4 event with an Oakland epicenter), and the entire HayWired earthquake sequence. Damage rates reach 50 repairs/km2 along the Hayward Fault and in zones of high liquefaction probability. The heatmap for the mainshock shows 20–50 repairs/km2 in large areas west of the fault, and no levels so high east of the fault, consistent with older pipe and higher liquefaction probability west of the fault. As a check of validity of the mainshock heatmap (fig. 34), notice that 122°30' it shows on the order of 10 leaks or breaks per square kilometer over an area of about 500 km2, approximately the 5,200 leaks and breaks estimated here. As with the SJWC case study, the heatmaps show estimated repairs in the HayWired sequence. Different earthquakes would produce different damage patterns. The EBMUD heatmaps depict a single realistic damage pattern to make the HayWired scenario more tangible. EBMUD and its customers, by planning for this scenario, could better prepare for what actually happens in a future earthquake, which will invariably differ in total damage quantities, spatial distribution of damage, and over time. 122°15' 122° San Pablo Bay 38° Pinole Antioch Concord Richmond Lafayette Berkeley Danville Oakland San Francisco N EXPLANATION Dublin San Leandro O ISC NC A FR Pleasanton Y BA Mean pipeline repairs per square kilometer 30–40 <5 5–10 40–50 10–20 50–53 20–30 Major roadways City boundaries SA 37°45' Alameda Hillshade derived from U.S. Geological Survey National Elevation Dataset, 2013. Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. Hayward 0 Area of map 0 3 3 6 MILES 6 KILOMETERS CALIF Figure 34.  Map of buried water-pipeline damage in East Bay Municipal Utility District’s, California’s, service area for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Colors indicate mean repairs per square kilometer. A warmer color indicates greater concentration of damage. To be informative, the color scale is shifted from those used in figures 35 and 36. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 251 EBMUD Restoration Analysis See the San Jose Water Company Restoration Analysis section for the HayWired scenario’s assumptions about electricity, fuel, and communication. I further assume that EBMUD can acquire consumables—for example, pipe, clamps, and replacement valves—as quickly as needed to make repairs after the Mw 7.0 HayWired scenario mainshock. Regarding the availability of repair crews, EBMUD agrees with the scenario assumption that it would take as long as a week to assess the extent of damage and locate leaks before 122°30' repairs can be initiated on a larger scale and that repair efforts would likely need to initially focus on larger diameter water mains. EBMUD staff also estimated that they may be able to field 20 of their own repair crews plus 15 crews provided through mutual aid, for a total about 35 repair crews. Of these, one-quarter are deployed east of the East Bay Hills, the other three-quarters are deployed to the west of the hills. I assume that repairs begin 5 days after the mainshock (the mainshock occurs on day 1), and that c(t) ramps up from 20 to 35 crews over the following 14 days, and that crews work 8-hour days until repairs are completed. Figure 37 illustrates 122°15' 122° San Pablo Bay 38° Pinole Antioch Concord Richmond Lafayette Berkeley Danville Oakland San Francisco Dublin San Leandro O ISC NC A FR Pleasanton Y BA Major roadways City boundaries N EXPLANATION Mean pipeline repairs per square mile 4 1 2 5 3 SA 37°45' Alameda Hayward Hillshade derived from U.S. Geological Survey National Elevation Dataset, 2013. Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. 0 Area of map 0 3 3 6 MILES 6 KILOMETERS CALIF Figure 35.  Map of buried water-pipeline damage in East Bay Municipal Utility District’s, California’s, service area for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Colors indicate mean repairs per square kilometer. A warmer color indicates greater concentration of damage. To be informative, the color scale is shifted from those used in figures 34 and 36. 252   The HayWired Earthquake Scenario—Engineering Implications 122°30' 122°15' 122° San Pablo Bay 38° Pinole Antioch Concord Richmond Lafayette Berkeley Danville Oakland San Francisco N EXPLANATION Dublin San Leandro CO CIS AN FR Pleasanton Y BA Mean pipeline repairs per square kilometer 30–40 <5 5–10 40–50 10–20 50–55 20–30 Major roadways City boundaries SA 37°45' Alameda Hillshade derived from U.S. Geological Survey National Elevation Dataset, 2013. Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. Hayward 0 Area of map 0 3 3 6 MILES 6 KILOMETERS CALIF Figure 36.  Map of buried water-pipeline damage in East Bay Municipal Utility District’s, California’s, service area for the entire hypothetical HayWired scenario earthquake sequence (events larger than moment-magnitude 5). Colors indicate mean repairs per square kilometer. A warmer color indicates greater concentration of damage. To be informative, the color scale is shifted from those used in figures 34 and 35. repair crew availability, which is expressed mathematically here: a(t)=0.33 c(t)=2  t<6 .  =[2t – 5]   6 ≤ t<20  =35 20 ≤ t<41  =20 41 ≤ t The notation [x] means the integer part of the quantity x, used here because there are no such things as fractional crews. Figure 38 illustrates the initial level of water service according to equation 29—the loss of system pressure results in approximately 87 percent of service connections east of the East Bay Hills receiving water shortly after the mainshock, 8 percent of those west of the hills. It seems reasonable to assume that PG&E will take 2 weeks to restore electricity to 99.9 percent of customers in Alameda and Contra Costa Counties. EBMUD personnel concur. I therefore use the following electricity restoration curve for these counties: g (t ) = 1 − exp (−0.493t ) . (71) Figure 39 shows the estimated service restoration curve for EBMUD under three conditions—(1) as is, (2) assuming EBMUD develops a fuel-management plan to ensure that repair crews are never slowed or idled from lack of fuel, and (3) under ideal conditions. If the Hazus-MH serviceability index realistically measures the fraction of service connections receiving any water, as reports produced Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 253 40 1.00 EXPLANATION East Total West East 0.75 Serviceability index Crews available 30 20 0.50 0.25 10 West 0 1 10 100 0.00 0.01 1,000 0.1 Time after mainshock, in days Figure 38.  Graph showing initial East Bay Municipal Utility District, California, water service availability (equation 29) east and west of the East Bay Hills, in the east bay part of the San Francisco Bay area, following the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Figure 37.  Graph showing East Bay Municipal Utility District, California, repair-crew availability by day following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. East, areas east of the East Bay Hills; west, areas west of the East Bay Hills between the hills and San Francisco Bay. 400,000 1 Average pipe break rate per kilometer A B Water service connections available EXPLANATION Total as-is West of hills 300,000 East of hills Loss of resilience 200,000 EXPLANATION As-is conditions 100,000 Fuel plan and brittle pipe replaced Inspection Fuel plan only 0 1 10 100 1,000 1 10 100 1,000 Time after mainshock, in days Figure 39.  Graphs showing East Bay Municipal Utility District, California, water-service restoration curves by day following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. A, water-service restoration curves under as-is conditions, including total for the system and service east and west of the East Bay Hills, in the east bay part of the San Francisco Bay area. B, Total-system water-service restoration curves under several conditions—(1) with a fuel-management plan in place and all brittle pipe replaced, (2) with a fuel-management plan only, and (3) as-is conditions. With a fuel-management plan in place and all brittle pipe replaced before the mainshock, about 8 million service days are lost (lost connections×days), with average lost service days per lost connection being 21 days. With only a fuel-management plan in place and brittle pipe not replaced before the mainshock, about 18.9 million service days are lost, with average lost service days per lost connection being 48 days. Under as-is conditions following the mainshock, about 19.91 million service days are lost, with average lost service days per lost connection being 49 days. 254   The HayWired Earthquake Scenario—Engineering Implications by the Hazus software suggest, then “services available” means the fraction of service connections receiving even small flows. If it means the post-earthquake flow as a fraction of pre-earthquake flow, then the charts underestimate the number of service connections receiving at least some water. Note that “ideal conditions” refers to the case in which the HayWired mainshock occurs after all of EBMUD’s brittle pipe is replaced and all pumping stations are supplied with emergency generators and fuel. The curves show that under as-is conditions, full restoration takes 28 weeks (just more than 6 months). Under ideal-world conditions, full service is restored in 14 weeks (just more than 3 months), roughly 3 months earlier than under as-is conditions. Figure 40 shows the progress of EBMUD repairs in the HayWired earthquake sequence, again illustrating the possible effects of fuel limitations, an estimate of the benefit of an in-place fuel-management plan, and the potential benefit of replacing brittle pipe in advance. One can view the area above the curves in figure 40 as a measure of resilience—less area means less negative impact, faster recovery, or both. The areas above the three curves (as-is conditions, with fuel plan, and ideal-world) are measured in units of service days. That is, each day of lost water supply to a service connection equates with one service day. The areas above the three curves are shown in table 26. The table shows lost service-days under as-is conditions, with a fuel plan, and under ideal-world conditions, that is, in which all cast iron and asbestos cement pipe is replaced before the earthquake. The difference between lost 6,000 A service days under as-is and what-if conditions (fuel-management plan or ideal world) measures the resilience benefit of the what-if condition—with a fuel-management plan and if all brittle pipe were replaced before the hypothetical earthquake occurred. Although full restoration takes much longer, the table also shows average number of days that each service connection goes without potable water under as-is conditions, with the fuel-management plan, and under ideal-world conditions. Validation of EBMUD Damage and Recovery Estimates As with SJWC, it is possible to perform some initial checks of the foregoing results, to see at least whether they are reasonable. This validation is discussed below. Cross Validation with EBMUD Internal Damage Estimates EBMUD commissioned a private study that estimated, among other things, the potential for water-supply pipeline damage resulting from a M7.0 earthquake on the Hayward Fault. As described by Terentieff and others (2015), that 1997 study estimated 4,054 pipe breaks and leaks, most of which occur in cast-iron and asbestos-cement pipe. EBMUD performed an internal study in 1997 of large-diameter pipe (at least 16- to B EXPLANATION EXPLANATION Total as-is With fuel plan Ideal world Total as-is West of hills 5,000 East of hills Repairs remaining 4,000 Aftershocks Aftershocks 3,000 2,000 1,000 0 1 10 100 1,000 1 10 100 1,000 Time after mainshock, in days Figure 40.  Graphs showing East Bay Municipal Utility District, California, water-service repair progress following the hypothetical momentmagnitude-7 mainshock and aftershocks HayWired scenario earthquake sequence (see table 12). A, Curves for remaining repairs under as-is conditions, including total for the system and service east and west of the East Bay Hills, in the east bay part of the San Francisco Bay area. B, Curves for remaining repairs in the total system with (1) a fuel-management plan in place only and (2) a fuel-management plan in place and all brittle pipe replaced (the “ideal world”). Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 255 Table 26.  East Bay Municipal Utility District, California, average lost service days following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [Numbers rounded to reduce the impression of excessive precision. Ideal world—fuel-management plan in place and all asbestos cement and cast-iron pipe is replaced with ductile-iron or plastic pipe before the earthquake sequence occurs. R, number of days that service connection is without potable water; M, number of water service connections] Condition Total as-is With fuel-management plan Ideal world Lost service-days R×M 19,100,000 18,900,000 8,100,000 24-inch diameter, depending on pipe material) and estimated 334 breaks and leaks could result from a M7.0 earthquake on the Hayward Fault. Terentieff and others (2015) report that EBMUD has initiated an infrastructure renewal program with a goal of replacing approximately 1 percent of its pipe per year, focusing first on cast-iron and asbestos-cement pipe. The 1997 estimate of 4,054 breaks and leaks is close to the 4,300 breaks and leaks estimated here for the Mw 7.0 HayWired scenario mainshock and somewhat smaller than the present estimate of 5,688 breaks and leaks in the entire HayWired earthquake sequence. The 1997 estimate of 334 breaks and leaks in large diameter pipes is somewhat higher than the present estimate of 218 breaks and leaks in pipe of at least 20-inches diameter in the HayWired mainshock, although similar to the present estimate of 302 large-diameter breaks and leaks in the entire HayWired sequence. It seems realistic that EBMUD will complete much of its replacement of the 61 percent of its pipes that are constructed of cast iron or asbestos cement before a large earthquake occurs on the Hayward Fault. The issue is more complicated than just whether a Mw 7.0 or larger earthquake occurs on the Hayward Fault. According to the Uniform California Earthquake Rupture Forecast, version 3 (UCERF3), fault section data (Field and others, 2013), the chance that such an earthquake will occur in the next 61 years is approximately 16 percent on the northern segment of the Hayward Fault, 12 percent on the southern segment of the fault, and 26 percent on either of these fault segments—significant, but nowhere near certain. Comparison with EBMUD Judgment, Northridge, Kobe, and Napa Restoration As previously noted, Jeon and O’Rourke (2005) report that the 1994 Northridge earthquake caused 1,095 breaks or leaks to buried pipeline operated by the LADWP, whereas Lund and others (2005) report that repairs took about 1 week. The calculation discussed here for the HayWired scenario suggests that EBMUD would take 28 weeks to repair 5,700 breaks and leaks from the entire HayWired earthquake sequence. Although the estimate of just more than 6 months to restore EBMUD buried water pipelines following the HayWired mainshock agrees with EBMUD’s judgment, it is about 28 times as long for about five times the number of repairs, or about one-fifth on average as fast as LADWP’s actual repair rate following the Northridge earthquake. Resilience benefit, D(R×M) 200,000 11,000,000 Average lost days R 49 48 21 Also, as previously noted, the Kobe Municipal Waterworks Bureau experienced 1,757 breaks and leaks after the 1995 Kobe earthquake and that repairs took 10 weeks, that is, 176 repairs per week. The present estimate of 28 weeks to repair 5,700 breaks and leaks (200 repairs per week) is roughly on par with Kobe. The City of Napa repaired approximately 120 leaks and breaks in 5 days (170 repairs per week) after the 2014 South Napa earthquake, approximately equal to the 200 repairs per week estimated here. Cross Validation with Hazus-MH Table 27 shows the Hazus-MH estimate of water-supply restoration in Contra Costa and Alameda Counties, where EBMUD operates. As before, the estimates are for the HayWired mainshock only and do not reflect lifeline interaction. Applying the percentages to the number of EBMUD’s customers, one can compare the two models, as shown in figure 41. The two models substantially disagree in terms of initial service availability and in terms of restoration time. Effect of Lifeline Interaction and Consumable Limits on EBMUD EBMUD does not expect to begin repairs until about day 7 following the HayWired scenario mainshock, after power and telecommunications have been largely (though not completely) restored. Therefore, lifeline interaction will have little effect on EBMUD in the HayWired scenario. Performance of Other Water Utilities Based on Hazus-MH It is time consuming to acquire the necessary data and to perform the analysis of a water-supply system. The San Francisco Bay region has on the order of 30 of them. To estimate the effects of the HayWired earthquake sequence on the metropolitan area, I apply the proposed modification of the Hazus-MH (Federal Emergency Management Agency, 2012) methodology to the analysis of the bay region’s water-supply system (see Seligson and others, this volume). First, I adjust estimates of water-supply restoration time to account for the 256   The HayWired Earthquake Scenario—Engineering Implications Table 27.  Hazus-MH estimate of Contra Costa and Alameda County, California, loss of water supply following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [%, percent] Analysis Hazus-MH1 This analysis2 Households without water following mainshock Day 3 Day 7 Day 30 854,738 853,731 845,534 98.53% 98.41% 97.47% 71% 70% 46% Day 1 855,207 98.58% 71% Day 90 762,299 87.87% 24% 1 Hazus-MH figures as reported by the software (Doug Bausch, written commun., Federal Emergency Management Agency, 2014), with its estimate of 867,495 total households. Estimates rounded to the nearest percent to reduce the appearance of excessive accuracy. Hazus-MH figures are as reported by the software. 2 400,000 EXPLANATION Water-service connections available This analysis Hazus-MH 300,000 200,000 100,000 0 1 10 100 1,000 Time after mainshock, in days Figure 41. Graph comparing the estimated restoration of water supply connections for East Bay Municipal Utility District between the HayWired scenario model in this analysis and the Hazus-MH model for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario (Seligson and others, this volume). differences between Hazus’ default assumptions of repair crew availability. SJWC estimates it can field 22 repair crews to respond to damage in its service area of 225,000 service connections, or approximately 1 crew per 10,000 service connections. Hazus-MH seems to equate households with service connections. With the inventory that FEMA enhanced for the San Francisco Bay region before the HayWired project, Hazus-MH estimates that Santa Clara County has 565,863 households, which it seems to equate with the number of service connections. To adjust Hazus’ default number of pipeline repair crews for purposes of equation 61, I take: q 1 565, 863 = × ×4 q0 100 10, 000 = 2.26 . EBMUD engineers agree with the HayWired scenario assumption that EBMUD can field 35 crews to respond to (72) damage in its system that provides water to 390,000 service connections. Those figures indicate approximately 1 crew per 11,000 service connections, versus SJWC’s estimate of as many as 1 crew per 9,000 service connections. I assume therefore that Alameda and Contra Costa Counties (EBMUD’s service area) have approximately 1 crew (4 workers) per 11,000 households, Santa Clara County (SJWC’s service area) has 1 crew per 9,000 households, and other counties have 1 crew per 10,000 households (an approximate mean of EBMUD and SJWC, in round numbers). Table 28 shows corresponding restoration-rate adjustment factors for repair-crew availability. The household-weighted average value of the repair-crew factor q/q0 is 1.37, although it is higher in the strongly shaken counties of Alameda and Santa Clara. To address lifeline interaction for the purposes of equation 67—the product term inside the summation—I take rate-limiting factors u and the flow of rate-limiting factors g as those proposed for EBMUD. That is, I assume that approximately half of services are in pumped pressure districts that require electricity, that electricity is restored within 1 week, and that there is a temporary fuel shortage between days 3 and 7. Hazus’ estimates of the number of service connections with water service at time tj (normalized by the number of households) are recapped in table 29. The assumption of complete restoration at day 210 is mine. Hazus-MH does not report level of service beyond 90 days, but I suggest that full restoration would likely be completed within 7 months. The table reports values of V̂ (t) in the sense of equation 61. Table  30 shows the restoration curves adjusted for repair-crew availability and lifeline interaction but using the restoration curves for Alameda, Contra Costa, and Santa Clara Counties as those calculated in the case studies under as-is conditions. Table 31 shows the restoration curves adjusted with a fuelmanagement plan in place in all counties. Alameda, Contra Costa, and Santa Clara Counties are as those calculated in the case studies with a fuel plan. Table 32 shows the restoration curves with emergency generators and fuel in all counties. Alameda, Contra Costa, and Santa Clara counties are as those calculated in the case studies under ideal-world conditions. The tables are illustrated in figure 42A through D. Note that, because Hazus-MH analyses were unavailable for lifelines subjected to aftershocks in the HayWired earthquake sequence, the restoration curves for counties other than Santa Clara, Alameda, and Contra Costa Counties do not reflect aftershock damage. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 257 Table 28.  Repair-crew adjustment factor, q/q0, for San Francisco Bay region, California, buried water-supply pipeline restoration. Table 29.  Hazus-MH unadjusted estimate of water-service restoration for San Francisco Bay region, California, counties following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [%, percent] County Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Sonoma Stanislaus Yolo Households 523,366 344,129 100,650 63,815 121,236 45,402 453,602 15,885 329,700 181,629 254,103 565,863 91,139 130,403 172,403 145,146 59,375 Crews 48 31 10 6 12 5 45 2 33 18 25 63 9 13 17 15 6 Workers 190 125 40 26 48 18 181 6 132 73 102 251 36 52 69 58 24 q/q0 1.90 1.25 0.40 0.26 0.48 0.18 1.81 0.06 1.32 0.73 1.02 2.51 0.36 0.52 0.69 0.58 0.24 County Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Sonoma Stanislaus Yolo Day 1 1% 3% 91% 98% 100% 100% 100% 98% 40% 100% 30% 11% 100% 98% 100% 100% 100% 3 1% 3% 98% 99% 100% 100% 100% 100% 52% 100% 34% 11% 100% 100% 100% 100% 100% 7 1% 3% 100% 100% 100% 100% 100% 100% 87% 100% 41% 12% 100% 100% 100% 100% 100% 30 1% 5% 100% 100% 100% 100% 100% 100% 100% 100% 100% 19% 100% 100% 100% 100% 100% 90 2% 28% 100% 100% 100% 100% 100% 100% 100% 100% 100% 76% 100% 100% 100% 100% 100% 210 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% Table 30.  Hazus-MH-based estimate of water-service restoration after adjusting for repair crew availability and lifeline interaction without a fuel-management plan for San Francisco Bay region, California, counties following the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Table 31.  Hazus-MH-based estimate of water-service restoration after adjusting for repair crew availability and lifeline interaction with a fuel-management plan for San Francisco Bay region, California, counties following the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. [Santa Clara, Alameda, and Contra Costa Counties are as calculated in case studies. %, percent] [Santa Clara, Alameda, and Contra Costa Counties are as calculated in case studies. %, percent] County Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Sonoma Stanislaus Yolo 1 29% 29% 91% 98% 100% 100% 100% 98% 40% 100% 30% 63% 100% 98% 100% 100% 100% 3 29% 29% 94% 98% 100% 100% 100% 98% 55% 100% 34% 70% 100% 99% 100% 100% 100% Day 7 30 30% 54% 30% 54% 95% 100% 98% 100% 100% 100% 100% 100% 100% 100% 100% 100% 67% 100% 100% 100% 35% 80% 73% 100% 100% 100% 99% 100% 100% 100% 100% 100% 100% 100% 90 76% 76% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 210 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% County Alameda Contra Costa Marin Merced Monterey Napa Sacramento San Benito San Francisco San Joaquin San Mateo Santa Clara Santa Cruz Solano Sonoma Stanislaus Yolo 1 32% 32% 91% 98% 100% 100% 100% 98% 40% 100% 30% 63% 100% 98% 100% 100% 100% 3 33% 33% 94% 98% 100% 100% 100% 98% 55% 100% 34% 70% 100% 99% 100% 100% 100% Day 7 30 35% 59% 35% 59% 100% 100% 99% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 41% 86% 73% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 90 81% 81% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 210 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 258   The HayWired Earthquake Scenario—Engineering Implications Table 32.  Hazus-MH-based estimate of water-service restoration after adjusting for repair crew availability and lifeline interaction with a fuel-management plan, emergency generators and fuel at all pumping stations, and brittle pipe replaced (ideal world) for San Francisco Bay region, California, counties following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [Santa Clara, Alameda, and Contra Costa Counties are as calculated in case studies. %, percent] County 1 Alameda 44% Contra Costa 44% Marin 91% Merced 98% Monterey 100% Napa 100% Sacramento 100% San Benito 98% San Francisco 40% San Joaquin 100% San Mateo 30% Santa Clara 71% Santa Cruz 100% Solano 98% Sonoma 100% Stanislaus 100% Yolo 100% Day 3 44% 44% 94% 98% 100% 100% 100% 98% 55% 100% 34% 78% 100% 99% 100% 100% 100% 7 47% 47% 100% 99% 100% 100% 100% 100% 100% 100% 41% 85% 100% 100% 100% 100% 100% 30 73% 73% 100% 100% 100% 100% 100% 100% 100% 100% 86% 100% 100% 100% 100% 100% 100% 90 97% 97% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 210 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% Conclusions This chapter introduces an analytical model of watersupply damage and restoration that can be implemented solely with GIS and spreadsheet software—no black-box, proprietary tool required. (Note: the damage part of the new model is not novel; the restoration part is.) Summary The analytical model of water-supply damage and restoration introduced in this chapter suggests that a Mw 7.0 Hayward Fault earthquake could damage buried pipeline networks that supply potable water to the East Bay part of the San Francisco Bay area to the extent that the average customer of the EBMUD would be without water for 6 weeks. Some EBMUD customers would be without water service for 6 months. The total loss of resilience (measured as the area above graph curves of number of customers receiving service versus time) totals 19 million service-days lost. That loss can be reduced by half if current efforts to replace brittle pipe are completed before the next large earthquake occurs—the replacement taking six decades at a reasonably aggressive rate of 1 percent per year. The loss can be reduced on the order of 200,000 service days by reducing or eliminating EBMUD’s dependence on commercial fuel supplies. The model suggests that a HayWired-like earthquake could cost the average customer of SJWC 4 days of lost water service, with a total loss of resilience equal to 940,000 service days. Implementing a fuel-management plan could reduce both calculations by about a quarter. If SJWC completes replacement of all cast-iron and asbestos-cement pipe before a significant earthquake occurs (about 25 years at current replacement rates), the as-is losses (no fuel-management plan of replacement of brittle pipe) would be reduced by about half. Both case-study utilities (EBMUD and SJWC) reviewed the study described in this chapter and find its results reasonable and in line with previous studies, their own judgment, and comparison with restoration of other water-supply systems in other earthquakes. Results of the present study are greatly at odds with restoration estimates produced by Hazus-MH, which estimates much-longer restoration times and much-greater loss of resilience. Innovations Introduced Here The methodology proposed here models damage and restoration of buried water-supply pipelines subject to earthquake shaking (called wave passage) and ground failure (liquefaction, landslide and surface rupture of a fault). The methodology assumes that the analyst already has maps of the earthquake excitation (especially PGV and ground-failure probability) and of the pipeline system in question. Many authors have proposed such models in the past. The present model may be unique in combining some unusual features: 1. It treats lifeline interaction and limited consumables by reducing the speed with which repairs are completed in relation to how important those upstream lifelines and other resources are to repair productivity. In the example of SJWC, it was estimated that lifeline interaction and limited consumables increase the loss of resilience (measured in terms of lost service days) by 25 percent. 2. The model considers aftershocks. 3. It can be evaluated either deterministically (with no uncertainty) or as stochastic model (accounting for major sources of uncertainty). 4. It can be carried out with a GIS system and a spreadsheet and does not require other special software such as Hazus-MH. Doing so provides the analyst more insight into the reasonableness of model results and underlying sources of damage and restoration delay. 5. It offers an approximate method to modify Hazus-MH lifeline damage and restoration-time estimates to account for lifeline interaction. 6. It does not require hydraulic analysis of the damaged system or the system as repairs proceed. That simplification necessarily involves a common but Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 259 series g(t) for damage to other lifelines that have not been modeled. questionable assumption relating break rate to loss of service, and it prevents the analyst from gaining important insight into variation in pressure throughout the system. The methodology is applied here to examine the effect of a large hypothetical but not exceedingly rare earthquake in the San Francisco Bay region on the buried pipeline networks of SJWC and EBMUD. Results tend to agree with operator judgment, 7. It mostly avoids reliance on expert opinion and unpublished data. Expert opinion seems to be required to quantify the rate-limiting factors u and the service time 100 75 EXPLANATION Alameda Contra Costa Santa Clara San Mateo San Francisco Marin San Benito and Solano Other counties Water-service connections available, in percent 50 25 A B C D 0 100 75 50 25 0 0 28 56 84 112 140 168 0 28 56 84 112 140 168 Time after mainshock, in days Figure 42.  Graphs showing restoration of water service connections for San Francisco Bay region, California, counties following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. A, Restoration curves according to initial Hazus-MH calculations. B, Restoration curves after adjusting for repair-crew availability and lifeline interaction. C, Restoration curves after all utilties have implemented a fuel-management plan. D, Restoration curves with fuelmanagement plan in place and all brittle pipe replaced (the ideal world). Curves in B, C, and D use the more-detailed case-study calculations for Contra Costa, Alameda, and Santa Clara counties instead of calculations from Hazus-MH. 260   The HayWired Earthquake Scenario—Engineering Implications various restoration measures observed in past earthquakes, and comparable aspects of other models. Like all other aspects of the HayWired earthquake scenario, the outcomes presented here will invariably differ in quantity, spatial distribution, and over time from whatever actually happens when (not if) a large earthquake next strikes on the Hayward Fault or other San Francisco Bay region fault. By preparing for the water-supply impacts of this hypothetical earthquake sequence, the region can better prepare for whatever real earthquake actually occurs. Research Needs The methodology proposed here mostly avoids reliance on opinion and judgment, but it would be practical to eliminate reliance on much of the opinion and judgment that remains. Presumably the u factors based on expert opinion could be replaced by compiling sufficient earthquake experience from utilities, perhaps by some survey analogous to that of Lund and Schiff (1991). The time series g(t) could be replaced by explicit modeling. It would be interesting to know if there were some theoretical justification for water service restoration following a power law as in equation 32, and whether or why the power should be approximately 2/3. It would be also desirable: • To examine more closely or replace the Hazus-MH formulation of the serviceability index. Can one relate water-pipeline break rate to the fraction of customers receiving various thresholds of flow, such as minimal flows for cooking and basic sanitary needs?; • To know whether and how a large mainshock degrades the seismic resistance of apparently undamaged pipe; • To add treatment of earthquake damage to other elements in a water-supply system, including tanks, tunnels, canals, valves, and reservoirs; References Cited Aagaard, B.T., Graves, R.W., Schwartz, D.P., Ponce, D.A., and Graymer, R.W., 2010a, Ground-motion modeling of Hayward Fault scenario earthquakes, part I; Construction of the suite of scenarios: Bulletin of the Seismological Society of America, v.100, no. 6, p. 2927–2944, https://doi. org/10.1785/0120090324. Aagaard, B.T., Graves, R.W., Rodgers, A., Brocher, T.M., Simpson, R.W., Dreger, D., Petersson, N.A., Larsen, S.C., Ma, S., and Jachens, R.C., 2010b, Ground-motion modeling of Hayward Fault scenario earthquakes, part II; Simulation of long-period and broadband ground motions: Bulletin of the Seismological Society of America, v. 100, no. 6, p. 2945–2977, https://doi. org/10.1785/0120090379. Aagaard, B.T., Lienkaemper, J.J., and Schwartz, D.P., 2012, Probabilistic estimates of surface coseismic slip and afterslip for Hayward Fault earthquakes: Bulletin of the Seismological Society of America, v.102, no. 3, p. 961–979, https://doi.org/10.1785/0120110200. Aagaard, B.T., Boatwright, J.L., Jones, J.L., MacDonald, Keith A. Porter, K.A., Wein, A.M., 2017, HayWired scenario mainshock ground motions, chap. C of Detweiler, S.T., and Wein, A.M., eds., The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., https://doi.org/10.3133/ sir20175013v1. American Society of Civil Engineers, 1984, Guidelines for the seismic design of oil and gas pipeline systems, committee on gas and liquid fuel lifeline: Reston, Va., American Society of Civil Engineers, 473 p. • Regarding the Lund and Schiff (1991) database, it would be desirable for water utilities to adopt a standard database for American Water Works Association, 2008, AWWA small systems pipe repair checklist, as published in July 2008 recording leaks and breaks, especially in earthquakes, so as Opflow Question of the Month: American Water Works to inform future improvements in mathematical functions Association, 7 p., accessed August 8, 2017, at http://www. for pipe-fragility; and awwa.org/Portals/0/files/resources/water%20knowledge/ rc%20small%20systems/piperepairchecklist.pdf. • To model the damage to water-supply systems subjected to the August 24, 2014, South Napa earthquake to see how Ang, A.H.S., and Tang, W.H., 1975, Probability concepts in well it agrees with experience. engineering planning and design, vol. 1—Basic Principles: John Wiley & Sons, New York, 409 p. Acknowledgments Applied Technology Council, 2012, Seismic performance assessment of buildings; volume 1—methodology: Washington, D.C., Federal Emergency Management The author thanks Jacob Walsh and James Wollbrinck of Agency, 278 p., accessed August 8, 2017, at https:// SJWC, along with Julia Halsne, Andrea Chen, Clifford Chan, www.fema.gov/media-library-data/1396495019848Xavier Irias, Roberts McMullin, Mike Ambrose, Devina Ojascastro, 0c9252aac91dd1854dc378feb9e69216/FEMAP-58_ and Serge Terentieff of EBMUD for their help constructing and Volume1_508.pdf. testing the models presented here and reviewing the report. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 261 Applied Technology Council, 1991, Seismic vulnerability and impact of disruption of lifelines in the conterminous United States (FEMA 224, ATC-25): Washington, D.C., Federal Emergency Management Agency, 439 p., accessed August 8, 2017, at https://www.fema.gov/media-librarydata/20130726-1452-20490-1854/fema_224.pdf. Ballantyne, D., Berg, E., Kennedy, J., Reneau, R., and Wu, D., 1990, Earthquake loss estimation modeling of Seattle water system: Seattle, Wash., Kennedy/Jenks/Chilton, Inc., report to U.S. Geological Survey, Grant No.14-08-000 I-G 1526. Bay Area Rapid Transit District, 2015, Remembering Loma Prieta: Bay Area Rapid Transit District web page, accessed August 8, 2017, at http://www.bart.gov/marketing/25-yearsafter-Loma-Prieta-quake. Bay City News, 2015, East by earthquake rupture more than a dozen water lines: Bay City News, Inc., August 18, 2015, accessed December 27, 2015, at http://blog.sfgate.com/ stew/2015/08/18/east-bay-earthquake-ruptures-more-than-adozen-water-lines/. Benjamin, J.R., and Cornell, C.A., 1970, Probability, statistics, and decision for civil engineers: Dover Books on Engineering, 704 p. Boore, D.M., Stewart, J.P., Seyhan, E., and Atkinson, G.M., 2014, NGA-West2 equations for predicting PGA, PGV, and 5% damped PSA for shallow crustal earthquakes: Earthquake Spectra, v. 30, no. 3, p. 1057–1085. 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, v. 19, no. 4, p. 733–752. California Water/Wastewater Agency Response Network, 2009, Mutual aid/assistance operational plan: California Water/Wastewater Agency Response Network, 67 p., accessed August 8, 2017, at https://goo.gl/kyT8qE. Contra Costa Local Agency Formation Commission, 2008, Section 9.0 East Bay Municipal Utility District Water and Wastewater Service, in West Contra Costa County water and wastewater MSR adopted 8/13/08: Martinez, Calif., Contra Costa Local Agency Formation Commission, 8 p., accessed October 26, 2015, at http://www.contracostalafco. org/municipal_service_reviews.htm. Detweiler, S.T., and Wein, A.M., eds., 2017, The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., https://doi.org/10.3133/sir20175013v1. East Bay Municipal Utility District, 2014, Napa’s neighbors to the rescue: East Bay Municipal Utility District Customer Pipeline, Nov.–Dec. 2014, 1 p., accessed August 8, 2017, at https://www.ebmud.com/files/1314/3197/7857/customerpipeline_0.pdf. Eidinger, J., 2001, Seismic fragility formulations for water systems: American Lifelines Alliance, 96 p., accessed August 8, 2017, at http://www.americanlifelinesalliance. com/pdf/Part_1_Guideline.pdf. Elvert, Kurt, 2015, Water supply and conservation: San Jose Water Supply Company presentation, accessed, July 25, 2017, at http://www.avca-sj.org/presentations/2015-07-13_ KurtElvert_min.pdf. Federal Emergency Management Agency, 2012, Hazus multihazard loss estimation methodology, earthquake model, Hazus®-MH 2.1 technical manual: Federal Emergency Management Agency, Mitigation Division, accessed July 18, 2017, 718 p., at https:// www.fema.gov/media-librarydata/20130726-1820-25045-6286/hzmh2_1_eq_tm.pdf. Felzer, K.R., Abercrombie, R.E., and Ekström, G., 2003, Secondary aftershocks and their importance for aftershock prediction: Bulletin of the Seismological Society of America, v. 93, no.4, p. 1433–1448. Chen, W.F., and Scawthorn, C.R., eds., 2003, Earthquake engineering handbook: New York, N.Y., CRC Press, 1480 p. Field, E.H., Biasi, G.P., Bird, P., Dawson, T.E., Felzer, K.R., Jackson, D.D., Johnson, K.M., Jordan, T.H., Madden, C., Michael, A.J., Milner, K.R., Page, M.T., Parsons, T., Powers, P.M., Shaw, B.E., Thatcher, W.R., Weldon, R.J., II, and Zeng, Y., 2013, Uniform California earthquake rupture forecast, version 3 (UCERF3)—The time-independent model: U.S. Geological Survey Open-File Report 2013– 1165, 97 p., California Geological Survey Special Report 228, and Southern California Earthquake Center Publication 1792, accessed August 8, 2017, at https://pubs.usgs.gov/ of/2013/1165/. City of Winnipeg, 2014, Water main breaks frequent asked questions (FAQs): City of Winnipeg, Canada, web page, accessed October 6, 2015, at http://www.winnipeg.ca/ waterandwaste/water/mainbreaks.stm. Honegger, D.G., and Eguchi, R.T., 1992, Determination of relative vulnerabilities to seismic damage for San Diego County Water Authority (SDCWA) water transmission pipelines: Costa Mesa, CA: EQE Engineering, 19 p. California Water/Wastewater Agency Response Network, n.d., CalWARN mutual assistance activated in response to Napa earthquake: California Water/Wastewater Agency Response Network web page, accessed August 8, 2017, at http:// prod.i-info.com/dashboard/Layout/41EFB108A3534676930 96E1D202485DB/about/aboutPublic/~events.htm. 262   The HayWired Earthquake Scenario—Engineering Implications Howard, R.A., 1990, From influence to relevance to knowledge, in Oliver R.M., and Smith, J.Q., eds., Influence diagrams, belief nets and decision analysis: New York, N.Y., John Wiley & Sons, p. 3–23. Hudnut, K.W., Brocher, T.M., Prentice, C.S., Boatwright, J., Brooks, B.A., Aagaard, B.T., Blair, J.L., Fletcher, J.B., Erdem, J.E., Wicks, C.W., Murray, J.R., Pollitz, F.F., Langbein, J., Svarc, J., Schwartz, D.P., Ponti, D.J., Hecker, S., DeLong, S., Rosa, C., Jones, B., Lamb, R., Rosinski, A., McCrink, T.P., Dawson, T.E., Seitz, G., Rubin, R.S., Glennie, C., Hauser, D., Ericksen, T., Mardock, D., Hoirup, D.F., and Bray, J.D., 2014, Key recovery factors for the August 24, 2014, South Napa earthquake: U.S. Geological Survey Open-File Report 2014– 1249, 51 p., accessed August 8, 2017, at https://doi.org/10.3133/ ofr20141249. Khater, M., and Grigoriu, M., 1989, Graphical demonstration of serviceability analysis: Reston, Va., American Society of Civil Engineers, Proceedings of the 5th International Conference on Structural Safety and Reliability, San Francisco, California, August 7–11, 1989, p. 525–532t. Kumaraswamy, P., 1980, A generalized probability density function for double-bounded random processes: Journal of Hydrology, v. 46, p. 79–88. Los Angeles Department of Water and Power, 2014, Westwood/Sunset trunk line repair update: Los Angeles Department of Water and Power press release, Sunday August 3, 2014, accessed July 11, 2017, at https://www. ladwpnews.com/westwood-sunset-trunk-line-repair-updatesunday-august-3-2014/. Isoyama, R., and Katayama, T., 1982, Reliability evaluation of water supply systems during earthquakes: University of Tokyo, Report of the Institute of Industrial Science, v. 30, no. 1, p. 1–64. Lilliefors, H., 1967, On the Kolmogorov-Smirnov test for normality with mean and variance unknown: Journal of the American Statistical Association, v. 62, no. 318, p. 399–402. Javanbarg, M., and Scawthorn, C., 2012, UILLIS—Urban Infrastructure and Lifelines Interactions of Systems: Proceedings of the 15th World Conference on Earthquake Engineering, September 2012, Lisbon Portugal, p. 10358– 10364. Lund, L., and Davis, C.A., 2005, Multihazard mitigation Los Angeles water system—A historical perspective, in Taylor, C., and VanMarcke, E., eds., Infrastructure risk management processes—Natural, accidental, and deliberate hazards: Reston, Va., American Society of Civil Engineers, p. 224–279. Jayaram, N., and Baker, J., 2009, Correlation model for spatially distributed ground-motion intensities: Earthquake Engineering and Structural Dynamics, v. 38, p. 1687–1708. Jeon, S.S., and O’Rourke, T.D., 2005, Northridge earthquake effects on pipelines and residential buildings: Bulletin of the Seismological Society of America, v. 95, no. 1, p. 294–318. Jones, J.L., Knudsen, K.L., and Wein, A.M., 2017, HayWired scenario mainshock—Liquefaction probability mapping, chap. E of Detweiler, S.T., and Wein, A.M., eds., The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., https://doi.org/10.3133/sir20175013v1. Jones, L.M., Bernknopf, R., Cox, D., Goltz, J., Hudnut, K., Mileti, D., Perry, S., Ponti, D., Porter, K., Reichle, M., Seligson, H., Shoaf, K., Treiman, J., and Wein, A., 2008, The ShakeOut scenario: U.S. Geological Survey Open-File Report 2008– 1150 and California Geological Survey Preliminary Report 25, 312 p. and appendixes, accessed April 12, 2017, at https://pubs. usgs.gov/of/2008/1150/. Lund, L., Davis, C.A., and Adams, M.L., 2005, Water system seismic performance 1994 Northridge-1995 Kobe earthquakes: American Water Works Association and Japan Water Works Association, Proceedings of the 4th Japan-U.S. Workshop on Seismic Measures for Water Supply, Kobe, Japan, January 2005, p. 19–30, accessed August 8, 2017, at http://www.waterrf.org/PublicReportLibrary/3156.pdf. Lund, L., and Schiff, A., 1991, TCLEE pipeline failure database: Reston, Va., American Society of Civil Engineers, Technical Council on Lifeline Earthquake Engineering, 36 p. Markov I., Grigoriu, M., and O’Rourke, T.D., 1994, An evaluation of seismic serviceability of water supply networks with application to San Francisco auxiliary water supply system: Buffalo N.Y., National Center for Earthquake Engineering Research report no. 94-0001, 154 p. Jones, M.C., 2009, Kumaraswamy’s distribution—A beta-type distribution with some tractability advantages: Statistical Methodology, v. 6, p. 70–81. McCrink, T.P., and Perez, F.G., 2017, HayWired scenario mainshock—Earthquake-induced landslide hazards, chap. F of Detweiler, S.T., and Wein, A.M., eds., The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., https://doi.org/10.3133/sir20175013v1. Kennedy, R.P., Chow, A.W., and Williamson, R.A., 1977, Fault movement effects on buried oil pipeline: Journal of Transportation Engineering, v. 103, no. TE5, p. 617–633. Mid-America Earthquake Center, 2006, MAEViz Software: Urbana, Ill., Multihazard Approach to Engineering Center, University of Illinois at Urbana-Champaign. Chapter N. A New Model of Water-Network Resilience, with Application to the HayWired Scenario 263 Nojima, N., and Kameda, H., 1991, Cross-impact analysis for lifeline interaction: American Society of Civil Engineering, Technical Council on Lifeline Earthquake Engineering, Proceedings of the 3rd U.S. Conference on Lifeline Earthquake Engineering, Los Angeles, California, August 22–23, 1991, p. 629–638. Ogata, Y., 1998, Space-time point-process models for earthquake occurrences: Annals of the Institute of Statistical Mathematics, v. 50, no. 2, p. 379–402. O’Rourke, M.J., 2003, Buried pipelines, chap. 23 of Chen, W.F., and Scawthorn, C.R., eds., New York, N.Y., CRC Press, 1480 p. O’Rourke M.J., and Ayala, G., 1993, Pipeline damage due to wave propagation: Reston, Va., American Society of Civil Engineers, Journal of Geotechnical Engineering, v. 119, no. 9, p. 1490–1498. O’Rourke, T.D., Jeon, S.S., Toprak, S., Cubrinovski, M., Hughes, M., van Ballegooy, S., and Bouziou, D., 2014, Earthquake response of underground pipeline networks in Christchurch, NZ: Earthquake Spectra, v. 30, no. 1, p. 183–204. O’Rourke, T.D., and Palmer, M.C., 1996, Earthquake performance of gas transmission pipelines: Earthquake Spectra, v. 12, no. 3, p. 493–527. O’Rourke, T.D., and Trautmann, C.H., 1980, Analytical modeling of buried pipeline response to permanent earthquake displacements: Ithaca, N.Y., Cornell University, School of Civil Engineering and Environmental Engineering, report no. 80–4, 102 p. Petersen, M.D., Moschetti, M.P., Powers, P.M., Mueller, C.S., Haller, K.M., Frankel, A.D., Zeng, Yuehua, Rezaeian, Sanaz, Harmsen, S.C., Boyd, O.S., Field, Ned, Chen, Rui, Rukstales, K.S., Luco, Nico, 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, 243 p., accessed July 11, 2017, at https://doi.org/10.3133/ ofr20141091. Porter, K.A., 2017a, A beginner’s guide to fragility, vulnerability, and risk: SPA Risk LLC and University of Colorado Boulder, 92 p., accessed January 18, 2018, at http://www.sparisk.com/pubs/Porter-beginners-guide.pdf. Porter, K.A., 2017b, HayWired scenario three-dimensional numerical ground-motion simulation maps, chap. H of Detweiler, S.T., and Wein, A.M., eds., The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., accessed July 11, 2017, at https://doi.org/10.3133/ sir20175013v1. Porter, K.A., and Sherrill, R., 2011, Utility performance panels in the ShakeOut scenario: Earthquake Spectra, v. 27, no. 2, p. 443–458, accessed July 11, 2017, at https://doi. org/10.1193/1.3584121. Porter, K., Wein, A., Alpers, C., Baez, A., Barnard, P., Carter, J., Corsi, A., Costner, J., Cox, D., Das, T., Dettinger, M., Done, J., Eadie, C., Eymann, M., Ferris, J., Gunturi, P., Hughes, M., Jarrett, R., Johnson, L., Dam Le-Griffin, H., Mitchell, D., Morman, S., Neiman, P., Olsen, A., Perry, S., Plumlee, G., Ralph, M., Reynolds, D., Rose, A., Schaefer, K., Serakos, J., Siembieda, W., Stock, J., Strong, D., Sue Wing, I., Tang, A., Thomas, P., Topping, K., and Wills, C.; Jones, L., chief scientist, Cox, D., project manager, 2011, Overview of the ARkStorm scenario: U.S. Geological Survey Open-File Report 2010–1312, 183 p. and appendixes, accessed April 10, 2017, at https://pubs. usgs.gov/of/2010/1312/. Prashar, Y., McMullin, R., Cain, B., and Irias, X., 2012, Pilot large diameter pipeline seismic fragility assessment, in Card, R.J., and Kenny, M.K., eds., Proceedings of the Pipelines 2012 Conference, Miami Beach, Florida, August 19–22, 2012: Reston, Va., American Society of Civil Engineers, p. 396–407. Rose, A., Wei, D., and Wein, A., 2011, Economic impacts of the ShakeOut earthquake scenario: Earthquake Spectra v. 27, no. 2, p. 521–538. San Francisco Lifelines Council, 2014, Lifelines interdependency study 1 report: San Francisco Office of the City Administrator, San Francisco, 56 p., accessed July 20, 2017, at http://sfgov.org/esip/sites/default/files/Documents/ homepage/LifelineCouncil%20Interdependency%20Study_ FINAL.pdf. San Francisco Public Utilities Commission, 2015, Quarterly report—Water Enterprise Capital Improvement Program—Local—January 2015–March 2015: San Francisco Public Utilities Commission, accessed July 20, 2017, at https://sfwater.org/modules/showdocument. aspx?documentid=7182. Scawthorn, C.R., 1993, Lifeline interaction and postearthquake functionality: Proceedings of the 5th U.S.-Japan Workshop on Earthquake Disaster Prevention for Lifeline Systems, Tsukuba Science City, Japan, October 26–27, 1992, p. 441–450. Scawthorn, C.R., 2008, The ShakeOut scenario supplemental study—Fire following earthquake: Denver Colo., SPA Risk LLC, 33 p., accessed August 9, 2017, at http://books. google.com/books?id=mDGrFAw5zqYC&lpg=PA1&dq=s hakeout%20fire%20following&pg=PA1#v=onepage&q&f =false. 264   The HayWired Earthquake Scenario—Engineering Implications Scawthorn, C.R., Porter, K.A, Khater, M., Seidel, D., Ballantyne, D., Taylor, H.T., Darragh, R.D., and Ng, C., 1992, Utility performance aspects, liquefaction study, Marina and Sullivan Marsh areas, San Francisco California, in Hamada, M., and O’Rourke, T., Proceedings from the fourth Japan-U.S. Workshop on Earthquake Resistant Design of Lifeline Facilities and Countermeasures for Soil Liquefaction, Tokai University Pacific Center, Honolulu, Hawaii, May 27–29, 1992, vol. 1: National Center for Earthquake Engineering Research, Technical Report NCEER-92-0019, p. 317–333, accessed August 9, 2017, at http://www.sparisk.com/pubs/Scawthorn1992-SF-Liquefaction-Study.pdf. Schiff, A., 1988, The Whittier Narrows, California earthquake of October 1, 1987—Response of lifelines and their effect on emergency response: Earthquake Spectra, v. 4, no. 2, p. 339–366. Seligson, H.A., Eguchi, R.T., Lund, L., and Taylor, C.E., 1991, Survey of 15 utility agencies serving the areas affected by the 1971 San Fernando and 1987 Whittier Narrows earthquakes: Los Angeles, Calif., Dames & Moore, Inc., report prepared for the National Science Foundation, 100 p. Small Business Administration Office of Advocacy, 2012,  Frequently asked questions about small business: Washington, D.C., Small Business Administration, accessed December 14, 2017, at https://www.sba.gov/sites/defaul/files/FAQ_ Sept_2012.pdf. Tabucchi, T., Brink, S., and Davidson, R., 2010, Simulation of post-earthquake water supply system restoration: Civil Engineering and Environmental Systems, v.  27, no. 4, accessed July 14, 2017, at http://dx.doi.org/10.1080/10286600902862615. Tabucchi, T.H.P., and Davidson, R.A., 2008, Post-earthquake restoration of the Los Angeles Water Supply System: University at Buffalo, State University of New York, Multidisciplinary Center for Earthquake Engineering Research, Technical Report MCEER-08-0008, 127 p., accessed July 14, 2017, at https://mceer.buffalo.edu/pdf/ report/08-0008.pdf. Terentieff, S., Chen, A. McMullin, R., Prashar, Y., and Irias, X.J., 2015, Emergency planning and response damage prediction modeling to mitigate interdependency impacts on water service restoration: Water Research Foundation, Proceedings 9th Water System Seismic Conference, Sendai, Japan October 14–16, 2015, p. 80–91, accessed August 9, 2017, at http://www.waterrf.org/PublicReportLibrary/4603. pdf. Tierney, K.J., 1995, Impacts of recent U.S. disasters on businesses—The 1993 Midwest floods and the 1994 Northridge earthquake: Proceedings of the National Center for Earthquake Engineering Research Conference on the Economic Impacts of Catastrophic Earthquakes—Anticipating the Unexpected, New York, N.Y., September 12 and 13, 1995, 52 p. Treiman, J., and Ponti, D., 2011, Estimating surface faulting impacts from the ShakeOut scenario earthquake: Earthquake Spectra, v. 27, no. 2, p. 315–330. Wein, A.M., Felzer, K.R., Jones, J.L., and Porter, K.A., 2017, HayWired scenario aftershock sequence, chap. G of Detweiler, S.T., and Wein, A.M., eds., The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., https://doi.org/10.3133/sir20175013v1. The HayWired Earthquake Scenario—Engineering Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q [Also see https://doi.org/10.3133/sir20175013] Chapter O Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock By Ibrahim M. Almufti,1 Carlos Molina-Hutt,2 Michael W. Mieler,1 Nicole A. Paul,1 and Chad R. Fusco1 Abstract The HayWired scenario examines a hypothetical earthquake (mainshock) with a moment magnitude (Mw) of 7.0 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of California’s San Francisco Bay area. The tall-building performance assessment for the scenario includes a structural analysis and downtime and loss estimates for three archetype buildings that generally represent typical tall-building construction in the San Francisco and Oakland downtown areas. Two of the archetype buildings are steel-frame office towers (20 story and 40 story) designed to 1970s-era building code and professional practice, and one is a reinforced-concrete residential tower (42 story) designed using the current state-of-the-art performance-based design approach. An inventory of tall buildings in San Francisco is also provided. Time histories of ground motions from the simulated mainshock fault rupture were provided by the U.S. Geological Survey (USGS). The simulated records closest to the San Francisco and Oakland downtown locations were selected to assess the performance of the archetype buildings. A nonlinear response-history analysis (NLRHA) of each building was done using LS-DYNA software to ascertain the expected building performance. From a structural-engineering perspective, new concrete residential buildings performed relatively well with minimal structural damage. However, older steel-frame buildings sustained structural damage, including widespread yielding and some fractures of the (pre-Northridge) moment-frame beam connections in the upper stories. This did not result in collapse nor significant residual drifts, which indicates that the steel-frame buildings may be repairable. The interstory drifts for all analyses were within the interstory drift limit (story height/50) allowed by current building codes under the design basis earthquake. Peak floor accelerations (as much as roughly the acceleration due to gravity at the Earth’s surface,  g) Arup, North America, Ltd. 1 University College, London. 2 were significant due to higher mode effects, particularly in the reinforced concrete tower. (The period, or mode of vibration, of a building is a dynamic property that typically refers to the time it takes for a building, if excited horizontally by ground shaking, to complete one cycle of sway back and forth.) A loss assessment was performed using the probabilistic approach (that is, Monte Carlo simulation) outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document. The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined in the 2013 REDi™ (Resilience-based Earthquake Design Initiative for the Next Generation of Buildings) rating system. The engineering-demand parameters from the NLRHA were used as input parameters to assess the likelihood that each building component (structural and nonstructural) would sustain a discrete state of damage (defined by fragility functions). The extent and severity of the expected damage is used to assess the likelihood that building reoccupancy and (or) functionality is hindered (by repair-class assignments defined in REDi) enabling the estimation of repair time and total downtime (to achieve reoccupancy and (or) functionality). It is noted that the estimation of downtime has considerable uncertainty due to a number of interdependent factors that are difficult to quantify. The downtime estimates include the potential for delays to the initiation of repairs (termed impeding factors), including contractor mobilization, financing repairs, and permitting. The estimates for the impeding factors are based on those quantified in REDi and our additional research. The time to achieve functionality is also dependent on utility disruption times estimated elsewhere for the HayWired mainshock, but these did not govern any of the scenarios considered. All scenarios resulted in significant downtime to achieve functional recovery, primarily due to nonstructural component damage. Existing steel-frame office towers had a median repair cost of 7.4–17.5 percent of replacement value, a median time to achieve reoccupancy of 186–250 days, and median time to achieve functional recovery of 242–288 days. New reinforced-concrete residential towers had a median repair cost of 3.1–5.1 percent of replacement value, a median time to achieve reoccupancy of 268   The HayWired Earthquake Scenario—Engineering Implications 121–139 days, and median time to achieve functional recovery of 224–245 days. The time to mobilize contractors (accounting for the bidding process, scarcity of contractors, and time to mobilize labor, material and equipment) often governed the total downtime. In the scenario, buildings located in Oakland generally sustained higher losses (see upper end of ranges provided in the table) than those in San Francisco due to the closer proximity to the Hayward Fault. Although the nonlinear analysis indicated that steel-frame buildings sustain some fractured beam connections, it is unlikely that an inspector would observe them because there are so few instances of this behavior. Although the older steel-frame buildings have welldocumented structural deficiencies, none of the analyses undertaken resulted in structural collapse. This is primarily due to the relatively low shaking intensity from the HayWired scenario mainshock as compared to the maximum considered earthquake (MCE) and even design level defined in modern building codes. A 55-percent probability of collapse at MCE for the same San Francisco archetype steel-frame buildings in this study (approximately five times the acceptable collapse limit in modern codes) was estimated in a previous work. It is important to note that this study is limited to one structurally regular archetype steel-frame building, whereas in reality, many older buildings exhibit structural irregularities such as setbacks (and some even lack corner columns) that may make them more collapse-prone. In addition, the assessment considers only one scenario ground-motion record and thus does not account for the variability of ground shaking important for assessing collapse risk. Therefore, the results should not necessarily be interpreted to mean that a large earthquake on the Hayward Fault would not cause any tall-building collapses. Even one tall building damaged to the point of near collapse in a downtown area could cause closure of a significant number of surrounding buildings, even if the surrounding buildings are undamaged. Introduction The HayWired scenario examines a hypothetical earthquake (mainshock) with a moment magnitude (Mw) of 7.0 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of California’s San Francisco Bay area. The U.S. Geological Survey (USGS) is currently leading an effort to study the implications of the scenario mainshock and aftershock sequence for the greater San Francisco Bay region. As part of this work, the USGS identified the need to assess the performance of the existing tall-building stock in downtown San Francisco and Oakland. We undertook an analysis of archetype tall buildings representative of the building stock to estimate damage levels, downtime, and repair costs for the HayWired mainshock scenario in San Francisco and Oakland. A 55-percent probability of collapse at the maximum considered earthquake (MCE) for the same San Francisco archetype steel-frame building in this study (approximately five times the acceptable collapse limit in modern codes) was estimated in a previous work (Molina-Hutt and others, 2015). Objectives This study describes the performance of three building types found to be representative of the tall-building stock in the respective cities’ downtown areas: • 1970s 40-story steel moment-resisting frame office building in San Francisco only. • 1970s 20-story steel moment-resisting frame office building in San Francisco and Oakland. • New 42-story reinforced-concrete core-only building designed to current code and Pacific Earthquake Engineering Research Center (PEER) (TBI Guidelines Working Group, 2010) in San Francisco and Oakland. (Although no new concrete towers of this height currently exist in Oakland, future development plans are likely to include them.) Structural analysis results include peak interstory drifts, residual drifts, racking drift (where applicable), floor accelerations, and coupling beam rotation (where applicable). Repair costs are expressed in terms of both absolute dollars and percentage of building replacement cost. A breakdown of repair costs by building components (both structural and nonstructural) is also provided for each building. Downtime estimates include repair time and downtime (the time to achieve either reoccupancy, functionality, or full recovery) in days. Downtime accounts for repair time plus impeding factors (such as the time required to mobilize contractors and engineers) and utility disruption. Report Structure Including the Introduction, this report is organized into seven sections—(1) Introduction; (2) HayWired Ground Motions, which describes the ground motions used in the nonlinear dynamic analyses, and provides a comparison of the HayWired scenario and the code design spectrum for each location; (3) Description of Archetype Buildings, Design and Analysis Assumptions, which describes the structural design—including the structural configuration, structural properties, and typical details—of the three archetype buildings, and the numerical modeling of the buildings; (4) Loss-Assessment Methodology, which describes the loss-assessment methodology; (5) Summary of Loss-Assessment Results, which provides a summary comparison of the lossassessment results; (6) Conclusion; and (7) References Cited. The report also has 12 appendixes. Appendix 1 provides lists of the structural and nonstructural building components in each archetype building, including component quantities, median engineering design parameters (EDPs) and dispersions for each damage state, and repair-class assignments defined in REDi™ (Resilience-based Earthquake Design Initiative for the Next Generation of Buildings). Appendixes 2 through 11 provide detailed structural analysis and loss assessment results for each building, as shown in table 1. Appendix 12 provides an inventory of tall building stock in San Francisco. Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 269 Table 1.  Contents of appendixes 2 through 12. Appendix 2 3 4 5 6 7 8 9 10 11 Building case study 40-story steel-frame building in San Francisco, baseline orientation 40-story steel-frame building in San Francisco, rotated orientation 20-story steel-frame building in San Francisco, baseline orientation 20-story steel-frame building in San Francisco, rotated orientation 20-story steel-frame building in Oakland, baseline orientation 20-story steel-frame building in Oakland, rotated orientation 42-story reinforced-concrete building in San Francisco, baseline orientation 42-story reinforced-concrete building in San Francisco, rotated orientation 42-story reinforced-concrete building in Oakland, baseline orientation 42-story reinforced-concrete building in Oakland, rotated orientation HayWired Ground Motions The performance of the archetype building is assessed at two locations for which ground-motion time histories have been provided for the HayWired mainshock—southwest of downtown San Francisco (37.775° N, 122.402° W) and downtown Oakland (37.804° N, 122.270° W). These record sets are USGS code number CT06075018000 and SF384, respectively (see Aagaard and others, 2010). Figure 1 illustrates these locations within the San Francisco Bay area. Out of the five ground-motion sets that USGS provided for Oakland, the one in downtown Oakland was selected, because this is where most tall buildings in Oakland are located. Five 122.5° Abbreviation S-SF-B-43 S-SF-R-43 S-SF-B-20 S-SF-R-20 S-OK-B-20 S-OK-R-20 C-SF-B-46 C-SF-R-46 C-OK-B-46 C-OK-R-46 ground-motion sets were also provided for San Francisco, but none were located in the Financial District, where most tall buildings are located. The ground-motion record located roughly 1.5 kilometers (km) southwest of the Financial District (see fig. 1) was selected because its time-averaged shear-wave velocity to a depth of 30 meters (VS30) value was most representative of the soil conditions (site class D of the International Building Code) in the Financial District. Time histories for the Oakland and San Francisco ground motions in the Mw 7.0 HayWired scenario mainshock are shown in figures 2 and 3, respectively. These are surface motions generated by a three-dimensional (3D) physics-based simulation for the scenario mainshock (Aagaard and others, 2017). 122.4° 122.3° Golden Gate Figure 1. Map of the central San Francisco Bay area, California, showing locations at which archetype buildings were assessed in San Francisco (code CT06075018000, see Aagaard and others, 2010) and downtown Oakland (code SF384, see Aagaard and others, 2010) for performance in ground-motions for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. SF384 37.8° CT06075018000 SAN PACIFIC CO CIS AN FR OCEAN 37.7° Y BA Base map ©2018 Google, Data SIO, NOAA, U.S. Navy, NGA, GEBCO, Data CSUMB SFML, CA OPC. 0 0 Area of map CALIF 1 2 1 3 2 4 5 KILOMETERS 3 MILES 270   The HayWired Earthquake Scenario—Engineering Implications North component Displacement, in meters Velocity, in meters per second Acceleration, in meters per second squared East component Time, in seconds Figure 2.  Graphs showing acceleration, velocity, and displacement time histories of the east-west and north-south components of the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario for downtown San Francisco, California (code ct06075018000, see Aagaard and others, 2010). Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 271 North component Displacement, in meters Velocity, in meters per second Acceleration, in meters per second squared East component Time, in seconds Figure 3.  Graphs showing acceleration, velocity, and displacement time histories of the east-west and north-south components of the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario for downtown Oakland, California (code SF384, see Aagaard and others, 2010). 272   The HayWired Earthquake Scenario—Engineering Implications Description of Archetype Buildings, Design, and Analysis Assumptions For comparison purposes, figures 4 and 5 show the ASCE 7–10 (American Society of Civil Engineers, 2010) design acceleration-response spectrum for the design-basis earthquake (DBE) scenario and the HayWired scenario ground-motion spectra selected for the Oakland and San Francisco sites, respectively. It can be observed that the HayWired groundmotion spectra selected for assessment have relatively low spectral accelerations near the fundamental period of the tall buildings (and consequently, relatively low spectral-displacement demands), although significant structural demands could be induced from higher mode effects.3 Moreover, because the epicenter of the HayWired mainshock is located in Oakland, there is less forward directivity for the cities of Oakland and San Francisco than if the epicenter was at either end of the Hayward Fault rupture, particularly compared to an epicenter at the south end of the rupture. Therefore, the HayWired mainshock ground motions may represent a nonconservative scenario for the buildings in this study. This section describes the building archetypes developed for this study of tall-building performance in the HayWired scenario mainshock. This includes a discussion of design and analysis assumptions. Steel Office Tower Forty-story and 20-story steel moment-resisting frame (MRF) office buildings designed according to the 1973 Uniform Building Code (International Conference of Building Officials, 1973) were selected as archetype buildings for this study. The archetype buildings are rectangular in plan and represent the state of design and construction practice from the mid-1970s to the mid-1980s. 1.2 1.0 Acceleration, in g Figure 4.  Graph showing ASCE 7–10 (American Society of Civil Engineers, 2010) design acceleration-response spectrum for its design-basis earthquake (DBE) versus ground-motion spectra for hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario for the representative site in Oakland, California. g, acceleration due to gravity at the Earth’s surface; N-S, north-south; E-W, east-west. EXPLANATION ASCE 7–10 (site class D) HayWired mainshock E-W HayWired mainshock N-S 0.8 0.6 0.4 0.2 0 0 1 2 3 4 5 6 Period, in seconds 1.0 ASCE 7–10 (site class D) HayWired mainshock E-W HayWired mainshock N-S 0.8 Acceleration, in g Figure 5.  Graph showing ASCE 7–10 (American Society of Civil Engineers, 2010) design acceleration-response spectrum for its design-basis earthquake (DBE) versus ground-motion spectra for hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario for the representative site in San Francisco, California. g, acceleration due to gravity at the Earth’s surface; N-S, north-south; E-W, east-west. EXPLANATION 0.6 0.4 0.2 0 0 1 2 3 4 5 6 Period, in seconds 3 The period (or mode of vibration) of a building is a dynamic property that typically refers to the time it takes for a building, if excited horizontally by ground shaking, to complete one cycle of sway back and forth. Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 273 Design (64 mm) metal deck supported by steel beams meeting ASTM standard A36 (see ASTM International, 2014) with a yield stress of 36 kilopounds per square inch (ksi) (248 megapascal, MPa). The steel columns are ASTM A572 (see ASTM International, 2015) with a yield stress of 42 ksi (290 MPa). As shown in figure 6, the archetype structural system consists of a space frame with 20–40 ft (6.1–12.2 m) spans using wide flange beams, built-up box columns, and welded beam-column connections. Typical story heights are 10 ft (3.0 m) for basement levels, 20 ft (6.1 m) at ground level (lobby), and 12.5 ft (3.8 m) for typical office levels as stated in Molina-Hutt and others (2016). Because the buildings are rectangular in plan, two analyses were performed in which the building was aligned in two different orientations (with grid lines aligned in the cardinal directions) in consideration that an earthquake produces different ground shaking intensity normal and parallel to a fault, respectively. These orientations are shown in figure 7. For the archetype buildings, because wind drift limits governed the MRF section sizes, beams and columns have low strength-utilization ratios under the building code prescribed seismic forces of the 1970s. Typical member sizes and connection details were verified against available existing building drawings. Consistent with these records, built-up box columns and wide flange beams were selected for the archetype buildings. A summary of the design section sizes of the 40-story steel MRF and of the 20-story MRF are provided in tables 2 and 3, respectively. More information on the design and analysis of the steel-frame buildings can be found in Molina-Hutt and others (2016). The design of the archetype steel-MRF buildings is in accordance with the provisions of the 1973 Uniform Building Code and 1973 Structural Engineers Association of California (SEAOC) Bluebook (Structural Engineers Association of California, 1973), which was commonly used to supplement minimum design requirements. The design criteria for the 1970s archetype buildings would have been equivalent, whether in Oakland or in San Francisco, according to the requirements of the Uniform Building Code (UBC) in 1973 (both locations fall within the same seismic and wind zones). The 40-story and 20-story archetype buildings represent the existing building stock in San Francisco, whereas the building stock in Oakland is generally limited in height to 20 stories. Based on examination of existing building drawings, the 40-story building layout consists of 38 levels of office space, 2 levels for mechanical equipment—one at mid-height and one at the top of the building—and three basement levels for parking. The overall height of the structure is 507.5 feet (ft) (154.7 meters, m) above ground and 30 ft (9.1 m) below grade. The 20-story building consists of 19 levels of office space—one level for mechanical equipment at the top of the building and one basement level for parking. The overall height of the structure is 267.5 ft (81.5 m) above ground and 10 ft (3.0 m) below grade. For both buildings, the enclosure is composed of precast concrete panels and glass windows. The floor system is composed of 3-inch (in.) (76 millimeters, mm) concrete slab over 2.5-in. A B D E 20’ 20’ 20’ MF MF MF MF 6 MF 120’ 4 MF MF MF MF MF MF 2 MF MF MF MF MF X 3 MF MF MF 20’ MF MF MF MF Y 5 MF 20’ MF MF MF MF MF 80’ MF 20’ MF 20’ MF 20’ MF MF MF Figure 6.  Diagrams showing archetype 40-story steelmoment-frame office building (A) plan and (B) isometric (from Molina-Hutt and others, 2016). MF, moment frame; ’, feet. 7 MF MF 20’ 20’ C B A 1 274   The HayWired Earthquake Scenario—Engineering Implications A B Baseline orientation Rotated orientation Figure 7.  Diagrams showing (A) baseline and (B) rotated orientations of the archetype 40-story steel-moment-frame office building shown in figure 6 (from Molina-Hutt and others, 2016). MF, moment frame; N-S, north-south; E-W, east-west; ’, feet. Table 2.  Lateral resisting system section sizes for the 40-story archetype steel-moment-frame office building (modified from MolinaHutt and others, 2016). [t, flange thickness; W, wide-flange beam width, in inches; lb/ft, pound per foot] Wide flange beams Level range, in feet Box columns Exterior short span (W×lb/ft) Interior short span (W×lb/ft) Interior long span (W×lb/ft) Interior, in inches Exterior short elevation (x), in inches Exterior long elevation (y), in inches Base to 10 W36×256 W36×282 W30×124 22×22, t=3 26×26, t=3 20×20, t=2.5 Base to 10 W33×169 W36×194 W27×84 20×20, t=2 26×26, t=2.5 20×20, t=2 11 to 20 W33×118 W33×169 W27×84 18×18, t=1 24×24, t=1.5 18×18, t=1 21 to 30 W24×62 W27×84 W24×76 18×18, t=0.75 24×24, t=1 18×18, t=0.75 30 to roof W36×256 W36×282 W30×124 22×22, t=3 26×26, t=3 20×20, t=2.5 Table 3.  Lateral resisting system section sizes for the 20-story archetype steel-moment-frame building (modified from Molina-Hutt and others, 2016). [t, flange thickness; W, wide-flange beam width, in inches; lb/ft, pound per foot] Wide flange beams Level range, in feet Box columns Interior, in inches Exterior short elevation (x), in inches Exterior long elevation (y), in inches Exterior short span (W×lb/ft) Interior short span (W×lb/ft) Interior long span (W×lb/ft) Base to 10 W30×148 W30×173 W30×211 22×22 t=2 22×22, t=2.5 22×22, t=1.5 11 to 20 W27×129 W27×146 W30×191 22×22 t=1.5 22×22, t=2.0 22×22, t=1 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 275 Typical Details Figure 8 shows some of the typical details frequently observed in existing building drawings. Welds in beam-column connections before 1994 had very low toughness, as evidenced by fractures observed in the Mw 6.7 1994 Northridge, California, earthquake (Bonowitz and Maison, 2003). It is assumed that these fracture-prone pre-Northridge moment connections are common in 1970s construction. Therefore, all beam-column connections were modeled as pre-Northridge moment connections. Column splices are typically located 4 ft above the floor level, approximately every three floors. Observed typical splice connection details consist of partial joint-penetration welds of roughly half the thickness of the smaller section being connected. When subject to tensile forces, these splices can only carry a fraction of the moment capacity and (or) axial-tension capacity of the smallest section size being connected. Furthermore, experimental tests on heavy steel section-welded splices have illustrated sudden failures with limited ductility (Bruneau and Mahin, 1990). Based on this evidence, column splice failures are considered in this assessment. Building Dynamic Properties The dynamic properties, including the fundamental and second mode period of vibration in each translational direction, from a modal analysis of the archetype steel-frame buildings is presented in table 4. The table shows that 80 percent of the modal mass is mobilized in the first two translational modes for the 40-story tower (that is, mode 1 and 4 in the x direction and mode 2 and 5 in the y direction), whereas 90 percent of the modal mass is mobilized in the first two translational modes for the 20-story tower. Finite Element Modeling The steel-moment-frame office buildings were modeled using LS-DYNA, an advanced general-purpose multiphysics simulation software package developed by Livermore Software Technology Corporation (2009). The use of LS-DYNA for performance-based seismic analysis has become more common, and several recent building projects in California have undergone rigorous seismic peer review by experts. A nonlinear response-history analysis, using an explicit solver— which accounts for secondary moment effects and nonlinear materials—was done using the ground motions discussed above in HayWired Ground Motions, which were applied at the base of the model. The model was subjected to the ground motions in conjunction with expected gravity loads, which include self-weight, superimposed dead loads, and 25 percent of the unreduced live loads. Fixed supports were assumed at the base of the structure. Soil-structure interaction was not considered. Table 4.  Dynamic properties in x and y directions of archetype 20- and 40-story steel-moment-frame buildings examined for response in the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [--, no data] Steel-frame building (40 story) Vibration mode A Effective/total mass, in percent Period, in seconds Steel-frame building (20 story) Period, in seconds Effective/total mass, in percent 1 5.62 (x) 60.79 (x) 2.33 (x) 75.15 (x) 2 3 4 5 5.29 (y) 2.86 (torsional) 1.86 1.66 60.80 (y) -- (torsional) 21.33 (x) 19.28 (y) 2.04 (y) 1.47 (torsional) 0.78 0.65 73.33 (y) -- (torsional) 13.44 (x) 15.74 (y) B C Figure 8.  Diagrams showing typical details observed in existing building drawings. A, Plan section of typical moment connection; B, elevation of typical moment connection; C, typical splice. t, flange thickness; ’, inch; ”, feet; Rad. Max., maximum radius. (From Molina-Hutt and others, 2016.) 276   The HayWired Earthquake Scenario—Engineering Implications For tall buildings with embedded basements in soft soils, the kinematic effects are likely to slightly reduce the peak floor accelerations associated with higher modes. The embedment depth is important and we note that older steel-frame buildings typically have one or two basement levels. A 2.5-percent critical damping was assumed based on work by PEER (TBI Guidelines Working Group, 2010). The steel-frame office buildings were modeled with the following element types: • One-dimensional (1D) lumped plasticity elements for the beams with nonlinear behavior intended to represent pre-Northridge connections which may fracture prematurely. Each beam connection was assigned a unique backbone curve based on the probability it would fracture at a given rotation based on experimental evidence (Molina-Hutt and others, 2016). • 1D lumped plasticity elements for the columns. • Two-dimensional (2D) elastic shell elements for the floors using stiffness modifiers. • Nonlinear rotational springs for the steel-panel zones. • Nonlinear rotational springs for the steel-column splices, with nonlinear behavior intended to represent the typical weld details that may fracture at demands lower than the column axial and moment strength. For full details on the steel-frame building modeling, see Molina-Hutt and others (2016). Reinforced-Concrete Residential Tower A 42-story reinforced-concrete core-only residential building was selected as one of the archetype buildings for this study, because it has become a prevalent construction type in San Francisco. The building design is intended to represent the current state of practice for tall buildings in San Francisco. Design The archetype reinforced-concrete building design is originally based on the PEER Task 12 study (Moehle and others, 2011) for an archetypical building in Los Angeles and was redesigned for San Francisco seismic demands by Tipler (2014) following the PEER Tall Building Initiative guidelines (TBI Guidelines Working Group, 2010). The general performances objectives for the structure are no different than modern code objectives—to provide “collapse-prevention” in the MCE, “life-safety” in a DBE, and minimal damage in serviceability-level earthquakes (SLE). The design guidelines set out in PEER (TBI Guidelines Working Group, 2010) require that the structure be evaluated through a NLRHA to explicitly verify that the building has a low probability of collapse under the MCE. An isometric view of the building is shown in figure 9. As above for the steel-frame buildings, two building Figure 9.  Diagram showing an isometric view of the archetype 42-story reinforced-concrete residentialbuilding analytical model. orientations were analyzed for the reinforced-concrete building. These orientations are shown in figure 10. The core-only lateral system (fig. 11) is designed such that energy is dissipated through two flexural yield mechanisms—(1) plastic hinges at the base of each wall pier and (2) the ends of the coupling beams up the height of the building. The reinforced-concrete residential building has 42 above-ground levels and 4 below-ground levels for parking. The building is 457 ft (139.3 m) tall above grade, including the roof bulkhead. The superstructure floor plate is 107.9×107.0  ft (32.9×32.6 m) in plan. Floor-to-floor heights are typically 9.7  ft (3.0 m). The substructure is 228.0×227.0  ft (69.5×69.2  m) in plan. The gravity system is 8 in (203 mm) thick prestressed concrete slab and reinforced concrete columns. Columns sizes typically range from 36 in. (914 mm) square at the ground to 18 in. (457 mm) square at the roof. Core wall thickness is 32  in. (813 mm) thick up to superstructure level 13 and 24  in. (610 mm) thick from superstructure level 13 to the roof. Coupling beam depth is 30 in. (762 mm) everywhere except the basement, where the beams are 34  in. (864 mm) deep. Core-wall design (nominal) strength is 8  ksi (55 MPa). Shear-wall steel-reinforcement nominal yield strength is 60 ksi (410 MPa). Coupling-beam steel-reinforcement nominal yield strength is 75 ksi (520 MPa). More information on the design of the reinforced-concrete residential building can be found in Tipler (2014). Note that Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 277 A B Baseline orientation Rotated orientation Figure 10.  Diagrams showing (A) baseline and (B) rotated orientations of the archetype 42-story reinforced-concrete residential-building analytical model (from Moehle and others, 2011). N-S, northsouth; E-W, east-west; ’, feet; ”, inches. A B Figure 11.  Diagrams showing (A) core-wall layout with dimensions and (B) pier labeling (right) for the archetype 42-story reinforced-concrete residential-building analytical model (from Tipler, 2014). m, meters. Tipler’s design was based on a San Francisco location, but we used it for Oakland (even though there are currently no 40-story towers in Oakland) to obtain a general understanding of how this type of building would perform. Building Dynamic Properties The dynamic properties of the archetype 42-story reinforcedconcrete residential building, using effective stiffness properties that account for flexural cracking, are shown in table 5. For further details see Tipler (2014). Table 5.  Dynamic properties of the archetype 42-story reinforcedconcrete residential-building analytical model (modified from Tipler, 2014). [DBE, design-basis earthquake (American Society of Civil Engineers, 2010), T1, first mode period in specified direction, T2, second mode period in specified direction] Period (DBE), in seconds Strong direction (y-axis) Weak direction (x-axis) T1=4.37 T1=5.27 T2=0.93 T2=1.10 278   The HayWired Earthquake Scenario—Engineering Implications Modeling Tipler (2014) modeled a reinforced-concrete building using LS-DYNA (Livermore Software Technology Corporation, 2009). We made some modifications to the LS-DYNA model to reflect updated knowledge of concrete-material modeling. A nonlinear response history analysis, using an explicit solver, which accounts for secondary moment effects and nonlinear materials, was undertaken with the ground motions discussed above in HayWired Ground Motions applied at the base of the model. The reinforced-concrete residential buildings were modeled with the following element types: • 1D distributed plasticity fiber beam elements for the shear walls, with fiber representing steel reinforcement, confined/unconfined concrete, and corresponding nonlinear material behavior. The use of fiber beam elements to represent core walls is industry practice. It enforces “plane sections remain plane” behavior which may underestimate the localized damage at the ends or corners of walls and may therefore underpredict losses. • 1D lumped plasticity elements for the coupling beams, which have been validated against testing at the University of California Los Angeles (UCLA) (Naish and others, 2009) to reproduce the hysteretic behavior. • 2D elastic shell elements for the floors using stiffness modifiers. • 2D elastic shell elements for the basement walls. • 1D elastic beam elements for the columns. The reinforced-concrete model was subjected to the ground motions in conjunction with expected gravity loads, which includes self-weight, superimposed dead loads, and 25 percent of the unreduced live loads. Fixed supports were assumed at the base of the structure and soil-structure interaction was not considered. For tall buildings with deep embedded basements in soft soils, the kinematic effects are likely to reduce the peak floor accelerations associated with higher modes. The embedment depth is important, and we note that newer reinforced-concrete buildings typically have three to five basement levels. It can be inferred that the damage to some acceleration-sensitive nonstructural components may be overestimated. A 2.5-percent critical damping was assumed based on work by PEER (TBI Guidelines Working Group, 2010). For full details on the reinforced-concrete building modeling, see Tipler (2014). Loss-Assessment Methodology The FEMA P-58 methodology (Applied Technology Council, 2012) was followed in to determine the risk metrics of interest for the buildings in this study. The fragility curves in the September 2016 update of FEMA P-58 volume 3 were used (Applied Technology Council, 2016). Using those results, the REDi™ downtime methodology (with some enhancements) was employed to calculate building downtime until reoccupancy and functional recovery is achieved (Almufti and Willford, 2013). Note that components which suffer only cosmetic damage do not contribute to the time required to achieve reoccupancy or functionality (only full recovery, which is not reported herein). These calculations are implemented using in-house software at Arup (Arup, North America, Ltd.), which runs Monte Carlo simulations (that is, realizations) for each building. The loss-assessment methodology in FEMA P-58 relates expected building movements to expected damage in individual components to expected consequences (for example, repair costs and repair times). The structural response from the ground motions for the HayWired mainshock was assumed to be ‘best estimate’ structural response. A modeling dispersion was applied to obtain variability in response, accounting for uncertainty in analysis model quality and construction quality. The modeling dispersion assumed was 0.35 for steel-frame buildings and 0.27 reinforced-concrete buildings based on guidance in FEMA P-58 (Applied Technology Council, 2012). A lognormal distribution was assumed for this purpose, where each floor, parameter type, and direction was assumed independent of one another. Using the “best estimate” structural response and modeling dispersion, each individual building underwent 1,000 Monte Carlo loss simulations to capture the uncertainty in building damage and consequences. This results in a distribution of repair-cost and repair-time estimates with corresponding probabilities of nonexceedance (often referred to as “confidence levels”) for the HayWired scenario mainshock and archetype buildings. As a result of this being a scenario study, the risk results have a much narrower range as compared with intensity-based studies. In those cases, the motion to motion variability adds a significant amount to the overall dispersion on structural response. REDi Downtime Methodology The REDi™ downtime methodology builds on the FEMA P-58 methodology to calculate the time required to achieve discrete downtime recovery states such as reoccupancy, functionality, and full recovery (Almufti and Willford, 2013). This is dependent on the extent and severity of damage to individual building components and their criticality in supporting occupancy or functionality. The REDi methodology maps each of the damaged components into repair classes, which indicate whether the extent and severity of damage to that component would hinder reoccupancy (repair class 3), functionality (repair class 2), and full recovery (repair class 1). Figure 12 provides an overview of the methodology, whereas figure 13 illustrates the aspects of the REDi downtime method for functional recovery. Probability of Observing Damage The FEMA P-58 methodology allows the estimation of the extent and severity of damage to specific building components. Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 279 Seismic hazard analysis Structural analysis Damage assessment Consequence analysis SA(T1) = 0.5 g Repair cost Repair time Downtime Figure 12.  Diagram showing overview of methodologies used for the loss assessment for tall buildings for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Methodologies— FEMA P-58 (Applied Technology Council, 2012); REDi (see Almufti and Willford, 2013). SA, spectral acceleration; T1, fundamental mode of vibration; g, acceleration due to gravity; PFA, peak floor acceleration; IDR, interstory drift ratio; P, probability; DS, damage state; EDP, engineering design parameter. Utilities Functional recovery Earthquake occurence Impeding factors Class 2 and 3 building component repairs Figure 13.  Illustration showing REDi™ (Almufti and Willford, 2013) downtime framework for assessing the functional recovery of tall buildings after the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. However, the impact of the damage on whether downtime is triggered depends to a great extent on whether the damage is observed or hidden (particularly true of structural components). The REDi methodology explicitly accounts for the probability that an inspector would observe a certain type of damage (and thus influence their decision how to tag a building) by assigning a 50-percent chance that an inspector would observe the damage when 10 percent of all components of that type are in the highest realized damage state. In this study, this assumption is particularly important for the pre-Northridge moment-frame connections in existing steel-frame buildings. Jurisdictions such as San Francisco do not have specific policies for post-earthquake inspection of the moment-frame connections and are unlikely to require such an inspection unless the building exhibited a lean (Laurence Kornfield, Earthquake Engineering Research Institute, written commun., 2017). However, many owners of older buildings retain structural engineers through the Building Occupancy Resumption Program (BORP) (Structural Engineers Association of Northern California, 2003). It likely that structural engineers, having become acutely aware of the deficiencies in such steel moment-frame connections, would recommend inspection of some connections (but not all). Section 3.3.3 of FEMA 352 (SAC Joint Venture, 2000) recommends that one connection in each framing line in each direction (east-west and north-south) is exposed for inspection under certain circumstances that would appear to be satisfied by the HayWired mainshock. For the archetype steel-frame building, with 6 bays (12 connections) in one direction and 4 in the other (8 connections), this is between 8.3 percent and 12.5 percent of all connections, respectively. This is relatively consistent with the 10-percent inspectionexposure assumption by REDi. In addition, the connections may be fire-proofed, which according to FEMA 352, may “tend to obscure many types of damage, unless the damage is very severe.” FEMA 352 also indicates that even for bare steel connections, certain types of damage “may be impossible to detect by visual observation alone.” This supports the REDi assumption that the likelihood of observing a fracture, having exposed the connection, is not 100 percent guaranteed. Because the number of beam fractures from the steel-frame building analysis in this study (see appendixes) is less than a few percent of all beam components in the building, it is unlikely that an inspector would observe one, and thus beam fractures do not contribute to downtime in the majority of realizations. 280   The HayWired Earthquake Scenario—Engineering Implications Impeding Factors Component Fragilities The time to make corresponding repairs to earthquakedamaged buildings is calculated based on FEMA P-58 but modified by REDi to consider likely construction repair sequences and labor allocation. Based on the type and extent of repairs, the total time to achieve reoccupancy or functionality must also consider impeding factors (that is, delays to the initiation of repairs). The following impeding factors are considered, the default values of which are generally taken from REDi but have been modified as indicated below: A full list of the archetype building-component quantities and fragility criteria can be found in appendix 1. Appendix 1 provides information on both structural components and information on nonstructural components. In addition to the default fragilities provided by FEMA P-58, we modified fragilities for elevators and precast facades. Two elevator components were modeled to capture two possible types of damage: • Postearthquake inspection. • Financing (assumes that the buildings are insured, based on the authors’ knowledge of earthquakeinsurance coverage for tall buildings in San Francisco). • Engineering mobilization and review. • Contractor mobilization (modified based on extent, type, and severity of damage and height of building; estimates based on survey of contractors administered by Arup). • Permitting. Utility disruption is also considered in the downtime, as described above under Utility Disruption. Input Parameters The basic building information used in the assessment methodology for the archetype tall-buildings included number of stories and square footage per floor. Sources for the assumed component types and corresponding quantities are provided in the appendixes. Building contents—such as modular workstations, bookcases, art pieces, and casework— were not included in the loss assessment. For the steel-frame buildings, heavy mechanical equipment (that is, cooling tower and chiller) capacities were modified based on input from mechanical engineering experts at Arup. The structural analysis results for given ground motions for the HayWired mainshock at a site were interpreted to be “best estimate” results about which a modelling dispersion was assumed to account for uncertainty in analytical model quality and assumed construction quality. Demolition Fragility The probability that a building might need to be demolished as a result of the HayWired mainshock was considered in our analysis. If a building has a peak residual interstory drift (RIDR) of 0.5 percent or less, there is a negligible chance the structure would need to be demolished. If a building has a peak RIDR of 1 percent, there is a 50-50 chance that it would need to be demolished. 1. Acceleration-sensitive component which is correlated to the elevator cabin damage.—This is modelled at the ground floor of the building because the FEMA P-58 fragility is a function of peak ground acceleration (PGA). It is our opinion that these are likely conservative for taller buildings, as the default FEMA P-58 fragilities were developed from shorter buildings in which the floor accelerations increased up the height (with roof accelerations likely to be 2–3 times PGA). The fragility does not account for the location of the elevator in the building, but presumably, the elevators that were damaged in observation were those at the upper stories. In a tall building, where the fundamental mode of vibration and higher mode effects contribute to the acceleration response, it is not necessarily the case that the PGA is amplified significantly. 2. RIDR sensitive component which is correlated to the shaft rail damage.—This is modeled at the superstructure level where the peak RIDR occurs, which varies from building to building. The fragility function for the precast facades used in the steel-frame buildings were developed based on the methodology in section 7.4 of FEMA P-58 (Applied Technology Council, 2012). The median interstory drift to cause damage was backcalculated by the gaps between façade panels and floor heights (determined through examination of drawings). The dispersion of 0.5 was adopted per section 7.4. Damage to façades is assumed to cause reoccupancy issues due to the risk of loose façade materials potentially falling and injuring passersby. Replacement Value The replacement value of steel-frame buildings is based on a class-5 rough cost estimate of the Association for the Advancement of Cost Engineering (AACE) and has an accuracy range of −5 to +30 percent as discussed in Molina-Hutt and others (2016). It includes all structure; exterior enclosure; mechanical, electrical, and plumbing (MEP) infrastructure; and partitions. It does not include demolition and site clearance. The replacement value of reinforced-concrete buildings is based on those of Tipler (2014), which used cost estimates from Moehle and others (2011) for most components, including structural elements. The cost of interior partitions and doors were provided by an experienced Arup estimator. Cost estimates for elevators and façade were obtained from vendors. The replacement value also includes MEP infrastructure and partitions. It does not include demolition and site clearance. Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 281 Utility Disruption The estimated building-restoration delays for due to disruptions of each of those utilities is shown in tables 6 and 7 for San Francisco and Oakland, respectively. Utility disruption times are considered when determining the time required to achieve building functionality. These are from other parts of the HayWired project (for example, Porter, Water-Network Resilience, this volume), except for natural gas. For natural gas, the values for utility disruption are based off of a study of several moderate to large magnitude earthquakes which have affected regions with modern infrastructure, including California, Japan, Chile, and New Zealand, as well as studies estimating utility disruption times with consideration of future earthquakes in the Western United States (Almufti and Willford, 2013). The utilities of interest include: Summary of Loss-Assessment Results The results of the loss assessment (repair cost and downtime estimates) for all 10 case-study buildings are discussed below. Both the median (50-percent confidence level) and probable maximum (90-percent confidence level) are presented. Because the HayWired mainshock is a scenario event, the difference between the median and 90th-percentile losses is small in comparison to the results from an intensity-based assessment involving many ground motions, where the motion to motion variability adds significantly to the overall dispersion. Figure 14 and table 8 show the results of the repair-cost assessment for the 10 buildings. Figure 15 and table 9 show the results of the downtime assessment for the 10 buildings. See the appendixes for detailed repair-cost and downtime results for each individual building. • Water • Natural gas • Electricity • Voice/data EXPLANATION Median reoccupancy Median functional 400 90th percentile reoccupancy 90th percentile functional 350 Downtime, in days Figure 14.  Graph showing median and 90th-percentile building-repair costs following the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario for all case study buildings (see table 1) as a percentage of replacement value. 300 250 200 150 100 50 OK -R - 46 6 -4 -B OK C- 46 C- C- SF - R- BSF C- ROK S- 46 20 20 K- B- 0 SO S- SF -R -2 20 F- B- -4 SS SF -R S- S- SF - B- 43 3 0 Case-study building abbreviation Table 6.  Estimated building-restoration delays due to utility disruption at the San Francisco, California, site following the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Utility Water1 Natural gas2 Electricity3 Voice/data4 50-percent restoration 3 days ~9 days 1 day 5 days Tail restoration 100 percent at 7 days 90 percent at ~33 days 99.5 percent at 30 days 100 percent at 7 days Porter (Water-Network Resilience, this volume). 1 Table 7.  Estimated building-restoration delays due to utility disruption at the Oakland, California, site following the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Utility Water1 Natural gas2 Electricity3 Voice/data4 50-percent restoration 30 days ~10 days 2 days 7 days Tail restoration 100 percent at 90 days 90 percent at ~36 days 96 percent at 30 days 100 percent at 30 days Porter (Water-Network Resilience, this volume). 1 Per REDi (Almufti and Willford, 2013) for peak ground velocities associated with the HayWired mainshock. Per REDi (Almufti and Willford, 2013) for peak ground velocities associated with the HayWired mainshock. 3 Preliminary Hazus-MH (Federal Emergency Management Agency, 2012) estimate run for the HayWired scenario (Doug Bausch, written commun., Federal Emergency Management Agency, 2014). 3 Preliminary Hazus-MH (Federal Emergency Management Agency, 2012) estimate run for the HayWired scenario (Doug Bausch, written commun., Federal Emergency Management Agency, 2014). 4 Preliminary estimate provided by John Erichsen (oral commun., 2016 HayWired Telecommunications Workshop). 4 Preliminary estimate provided by John Erichsen (oral commun., 2016 HayWired Telecommunications Workshop). 2 2 282   The HayWired Earthquake Scenario—Engineering Implications Table 8.  Summary of total building-repair costs following the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario for each case-study building (see table 1). Case-study building abbreviation S-SF-B-43 S-SF-R-43 S-SF-B-20 S-SF-R-20 S-OK-B-20 S-OK-R-20 C-SF-B-46 C-SF-R-46 C-OK-B-46 C-OK-R-46 Median total repair cost, in dollars 15,057,000 13,512,000 5,138,200 5,687,900 12,172,000 11,510,000 5,517,497 9,023,409 8,604,872 8,864,100 Percentage of replacement value 10.8 9.7 7.4 8.2 17.5 16.5 3.1 5.1 4.9 5.0 90th-percentile total repair cost, in dollars 17,132,000 15,690,958 6,592,184 7,261,380 14,395,814 13,065,046 6,470,705 9,839,828 9,393,212 10,829,600 Percentage of replacement value 12.3 11.3 9.5 10.4 20.7 18.7 3.7 5.6 5.3 6.2 Table 9.  Summary of building reoccupancy and functional-repair-time results following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario for each case-study building (see table 1). Case-study building abbreviation S-SF-B-43 S-SF-R-43 S-SF-B-20 S-SF-R-20 S-OK-B-20 S-OK-R-20 C-SF-B-46 C-SF-R-46 C-OK-B-46 C-OK-R-46 Median repair time Reoccupancy, Functional, in in days days 41 45 37 39 20 29 22 33 54 92 54 82 3 15 6 16 5 16 6 16 90th-percentile repair time Reoccupancy, in Functional, in days days 61 126 54 102 39 145 47 189 116 272 97 237 5 27 11 26 9 26 11 26 EXPLANATION Median reoccupancy Median functional 400 90th percentile reoccupancy 90th percentile functional 300 250 200 150 100 50 Case-study building abbreviation 6 46 RCOK - OK -B -4 C- CSF - C- SF -B -4 R46 6 -2 0 OK -R S- 0 20 BSOK - SF -R -2 20 S- BSSF - -4 -R SF S- SF -B -4 3 3 0 S- Figure 15.  Graph showing building downtime results in days following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario for all case-study buildings (see table 1). Downtime, in days 350 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 283 Table 10.  Summary of building reoccupancy and functional downtime results following the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario for each case-study building (see table 1). Case-study building abbreviation S-SF-B-43 S-SF-R-43 S-SF-B-20 S-SF-R-20 S-OK-B-20 S-OK-R-20 C-SF-B-46 C-SF-R-46 C-OK-B-46 C-OK-R-46 Median repair time Reoccupancy, in Functional, in days days 248 288 250 288 189 242 186 251 231 273 224 269 126 224 121 239 130 233 139 245 Conclusion This study used state of the art structural analysis and risk assessment to examine tall-building performance in the Mw 7.0 mainshock of the HayWired earthquake scenario. It is important to recognize that existing (and new) tall buildings have largely been untested by real earthquakes of this magnitude; thus the analytical methods have not been confirmed. The building-downtime estimates in particular rely on multiple assumptions that have not yet been robustly validated, including: • The fragility functions and underlying methodology in FEMA P-58 (Applied Technology Council, 2012) are suitable for the purposes of predicting repair time. • The severity and extent of damage to specified building components are the basis for the type of downtime incurred (that is, reoccupancy, functionality, or full recovery). • The length of time before repairs can be initiated (that is, impeding factors) • The time it takes to make repairs (dependent on the availability of laborers and the specific repair sequence schedule). In many respects, the loss results seem reasonable. The financial losses for the new reinforced-concrete buildings are significantly lower than for the existing steel-frame buildings; the reoccupancy time for the new reinforcedconcrete buildings are roughly half those of the steel-frame buildings; and the losses for buildings located in Oakland are higher than those in San Francisco, primarily due to the much larger ground accelerations for HayWired mainshock in Oakland. However, the time to achieve functional recovery is relatively similar between the new and old buildings, which on the face of it seems suspicious. In studying the results in 90th-percentile repair time Reoccupancy, in Functional, in days days 375 388 364 390 316 364 304 361 344 385 333 371 194 328 213 545 198 346 223 359 more detail, this seems plausible as new reinforced-concrete buildings experience higher peak floor accelerations than the steel-frame buildings, causing more nonstructural damage to acceleration-sensitive components that support building functions. Contractor mobilization times, which are dependent on the height of the building and the type and severity of component damage (primarily nonstructural for both building types), are similarly large and govern the total downtime for each building type. Of course, in the event that the few beam fractures in the steel-frame buildings are observed by an inspector, the estimated downtimes between new and old buildings would significantly widen. This would suggest that just a slightly larger earthquake scenario than that modeled in the HayWired scenario (perhaps at the design level or greater) would pose a greater risk to existing tall steel-frame buildings than new reinforced-concrete tall buildings. In other words, existing and new tall buildings may perform relatively similarly up to a certain seismic demand (that is, before widespread fractures in steel-frame buildings), but the performance would diverge considerably after that point, with the steel-frame buildings more prone to irreparable damage and collapse. This is supported by the high collapse-risk rates of existing tall steel-frame buildings (Molina-Hutt and others, 2016) relative to the code objectives for new tall buildings. Acknowledgments The report was enhanced in response to comprehensive review by Terry Paret (Wiss, Janey, and Eistner Associates, Inc.; WJE), Andrew Shuck (WJE), Tony Yang (University of British Columbia), and Erol Kalkan (USGS). The study was funded by USGS Land Change Science Program (reference number G16PS00702; USGS HayWired Scenario—Tall Building Performance). 284   The HayWired Earthquake Scenario—Engineering Implications References Cited Aagaard, B.T., Boatwright, J.L., Jones, J.L., MacDonald, T.G, Porter, K.A., and Wein, A.M., 2017, HayWired scenario mainshock ground motions, chap. C of Detweiler, S.T., and Wein, A.M., eds., The HayWired earthquake scenario— Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., https://doi. org/10.3133/sir20175013v1. Aagaard, B.T., Graves, R.W., Rodgers, A., Brocher, T.M., Simpson, R.W., Dreger, D., Petersson, N.A., Larsen, S.C., Ma, S., and Jachens, R.C., 2010, Ground-motion modeling of Hayward Fault scenario earthquakes, part II—Simulation of long-period and broadband ground motions: Bulletin of the Seismological Society of America, v. 100, no. 6, p. 2945–2977, https://doi.org/10.1785/0120090379. Almufti, I., and Willford, M., 2013, REDi™ rating system— Resilience-based earthquake design initiative for the next generation of buildings (version 1.0): Arup, Ltd., 68 p., accessed February 8, 2018, at http://usrc.org/files/technicalresource/ REDi_Final%20Version%201.0_October%202013.pdf. American Society of Civil Engineers, 2010, Minimum design loads for buildings and other structures: Reston, Va., American Society of Civil Engineers, ASCE/SEI 7–10, 608 p. American Society of Civil Engineers, 2014, Seismic evaluation and retrofit of existing buildings: Reston, Va., American Society of Civil Engineers, ASCE/SEI 41–13, 518 p., https://doi. org/10.1061/9780784412855. Applied Technology Council, 2012, Seismic performance assessment of buildings, including PACT software and background documents: Federal Emergency Management Agency, FEMA P-58, v. 1–3, accessed January 31, 2018, https:// www.fema.gov/media-library/assets/documents/90380. ASTM International, 2014, ASTM A36/A36M-14, Standard specification for carbon structural steel: West Conshohocken, Pa., ASTM International, https://doi.org/10.1520/A0036_ A0036M-14. ASTM International, 2015, ASTM A572/A572M-15, standard specification for high-strength low-alloy columbium-vanadium structural steel: West Conshohocken, Pa., ASTM International, https://doi.org/10.1520/A0572_A0572M-15. Bonowitz, D., and Maison, B., 2003, Northridge welded steel-moment-frame damage data and its use for rapid loss estimation: Earthquake Spectra, v. 19, no. 2, p. 335–364. Charette, R.P., and Marshall, H.E., 1999, UNIFORMAT II elemental classification for building specifications, cost estimating, and cost analysis: U.S. Department of Commerce, National Institute of Standards Technology Interagency Report 6389, 109 p., accessed March 12, 2018, at https://arc-solutions. org/wp-content/uploads/2012/03/Charette-Marshall-1999UNIFORMAT-II-Elemental-Classification....pdf. Federal Emergency Management Agency, 2012, Hazus multihazard loss estimation methodology, earthquake model, Hazus®-MH 2.1 technical manual: Federal Emergency Management Agency, Mitigation Division, accessed July 18, 2017, 718 p., at https://www.fema.gov/media-librarydata/20130726-1820-25045-6286/hzmh2_1_eq_tm.pdf. International Conference of Building Officials, 1973, Uniform Building Code: Whittier, Calif., International Conference of Building Officials, 700 p., accessed January 31, 2018, at http://digitalassets.lib.berkeley.edu/ubc/UBC_1973.pdf. Livermore Software Technology Corporation, 2009, LSDYNA keyword user’s manual—Volume 1 (ver. 971, release 4 beta): Livermore, Calif., Livermore Software Technology Corporation, 2435 p., accessed January 31, 2018, at http://www.dynasupport.com/manuals/ls-dynamanuals/LS-DYNA_971_R4_manual_k-beta-June2009.pdf. Moehle, J., Bozorgnia, Y., Jayaram, N., Jones, P., Rahnama, M., Shome, N., Tuna, Z., Wallace, J., Yang, T., and Zareian, F., 2011, Case studies of the seismic performance of tall buildings designed by alternative means, task 12 report for the Tall Buildings Initiative: Berkeley, Calif., Pacific Earthquake Engineering Research Center, University of California, no. 2011/05, 596 p., accessed January 31, 2018, at https://peer.berkeley.edu/publications/peer_reports/ reports_2011/webPEER-2011-05-TBI_Task12.pdf. Molina Hutt, C., 2017, Risk-based seismic performance assessment of existing tall steel framed buildings: London, UK., University College London, Ph.D. dissertation, 438 p., accessed February 28, 2018, at http://discovery. ucl.ac.uk/10040499/7/Molina%20Hutt_10040499_thesis_ redacted.pdf. Molina-Hutt, C., Deierlein, G., Almufti, I., and Willford, M., 2015, Risk-based seismic performance assessment of existing tall steel-framed buildings in San Francisco: Society for Earthquake and Civil Engineering Dynamics, 2015 Conference—Earthquake Risk and Engineering Towards a Resilient World, July 9–10, 2015, Cambridge, U.K., 11 p., accessed January 31, 2018, at http://www. seced.org.uk/images/newsletters/MOLINA%20HUTT,%20 DEIERLEIN,%20ALMUFTI,%20WILLFORD.pdf. Molina-Hutt, C., Ibrahim, A., Willford, M., and Deierlein, G., 2016, Seismic loss and downtime assessment of existing tall steel-framed buildings and strategies for increased resilience: Journal of Structural Engineering, v. 142, no. 8, https://doi.org/10.1061/(ASCE)ST.1943541X.0001314. Naish, D., Wallace, J.W., Fry, J.A., and Klemencic, R., 2009, Experimental evaluation and analytical modeling of ACE 318-05/08 reinforced concrete coupling beams subjected to reversed cyclic loading: University of California at Los Angeles, Structural and Geotechnical Engineering Laboratory, report 2009–06, 109 p. Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 285 SAC Joint Venture, 2000, Recommended Post-Earthquake Evaluation and Repair Criteria for Welded Steel Moment Frame Buildings: Federal Emergency Management Agency, FEMA-352, 200 p., accessed January 31, 2018, at https://www.fema.gov/medialibrary-data/20130726-1444-20490-4440/fema-352.pdf. TBI Guidelines Working Group, 2010, Tall Buildings Initiative— Guidelines for performance-based seismic design of tall buildings (ver. 1.0): Berkeley, Calif., Pacific Earthquake Engineering Research Center, University of California, no. 2010/05, 104 p., accessed January 31, 2018, at http://peer.berkeley.edu/publications/peer_ reports/reports_2010/web_PEER2010_05_GUIDELINES.pdf. Structural Engineers Association of California, 1973, Recommended lateral force requirements and commentary: Sacramento, Calif., Structural Engineers Association of California. Tipler, J.F., 2014, Seismic resilience of tall buildings—benchmarking performance and quantifying improvements: Stanford University master’s thesis, 91 p., accessed January 30, 2018, at https:// stacks.stanford.edu/file/druid:xh842sm8488/Thesis%20Final%20 Jennisie%20Tipler-augmented.pdf. Structural Engineers Association of Northern California, 2003, Building occupancy resumption program (BORP): Structural Engineers Association of Northern California, accessed February 28, 2018, at http://sfdbi. org/borp-guidelines-engineers. Yang, T.Y., Moehle, J., Stojadinovic, B., and Der Kiureghian, A., 2009, Performance evaluation of structural systems—Theory and implementation: Journal of Structural Engineering, v. 135, no. 10, p. 1146–1154. Appendix 1. Building Structural and Nonstructural Components The tables below list the following information for the three building archetypes in the baseline orientation (40-story steelframe building, 20-story steel-frame building, and reinforcedconcrete building): • Component quantities. • Component National Institute of Standards and Technology Interagency Reports (NISTIR) fragility classification number (shown as “NISTIR” in tables). • Component units, as specified by the NISTIR fragility. • The source of the component quantity; sources are: • The building design (for example, counting explicitly the number of base plates in the building design). • National Institute of Standards and Technology (NIST) Normative Quantity estimation tool packaged with FEMA’s Performance Assessment Calculation Tool (PACT; see Applied Technology Council, 2012) (marked as simply “Norm Qty” in the tables). • Moehle and others (2011), which documents the original structural design of reinforced concrete building before the San Francisco redesign by Tipler (2014). • Arup estimator. • Component medians and dispersions for each damage state. Tables 11–13 show structural components for each archetype building (40-story steel-frame building, 20-story steel-frame building, and reinforced-concrete building, respectively). Tables 14–16 show nonstructural components for each building type. Components for buildings with rotated orientation are generally the same as listed below, but with the east-west and north-south quantities flipped. L10–L17 L1–L9 B1 B2 B3 Floor Steel column base plates, column W>300 plf Bolted shear tab gravity connections Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≥W30 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≥W30 Bolted shear tab gravity connections Welded column splices, column W>300 plf Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≥W30 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≥W30 Bolted shear tab gravity connections Welded column splices, column W>300 plf ** Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≥W30 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≥W30 Bolted shear tab gravity connections Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≤W27 Component 72 0 24 B1035.052 B1031.001 B1035.041 8 B1035.042 24 B1035.052 72 13 8 B1035.042 B1031.001 B1031.021c 72 13 24 B1035.052 B1031.001 B1031.021c 72 8 13 E-W qty B1031.001 B1035.042 B1031.011c NISTIR 0 6 32 10 0 13 32 10 0 13 32 0 10 13 N-S qty Each Each Each Each Each Each Each Each Each Each Each Each Each Each Units DS2 median (dispersion) [repair class] 0.07 (0.4) rad [RC3] DS3 median (dispersion) [repair class] 0.1 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.11 (0.4) rad [RC3] NA 0.11 (0.4) rad [RC3] NA 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.08 (0.4) rad [RC3] 0.05 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.08 (0.4) rad [RC3] 0.05 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] Building design 0.04 (0.4) rad [RC3] 0.08 (0.4) rad [RC3] 0.11 (0.4) rad [RC3] Building design 0.017 (0.4) rad 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] [RC3] Building design Building design Building design 0.04 (0.4) rad [RC3] Building design 0.02 (0.4) rad [RC3] Building design Building design Building design 0.04 (0.4) rad [RC3] Building design 0.02 (0.4) rad [RC3] Building design Building design 0.04 (0.4) rad [RC3] 0.08 (0.4) rad [RC3] 0.11 (0.4) rad [RC3] Building design 0.017 (0.4) rad 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] [RC3] Source DS1 median (dispersion) [repair class] Building design 0.04 (0.4) rad [RC3] [NISTIR, National Institute of Standards and Technology Interagency Reports. E-W qty, east-west quantity per floor; N-S qty, north-south quantity per floor—note that components with only one quantity listed are nondirectional. DS, damage state. Other abbreviations: A, area; HVAC, heating, ventilation, and air conditioning; IGU, insulating glass unit; FT IGU, fully tempered insulating glass unit; Ip, component importance factor; NA, not applicable; Norm Qty, National Institute of Standards and Technology Normative Quantity estimation tool packaged with Federal Emergency Management Agency’s Performance Assessment Calculation Tool; OSHPD, California Office of Statewide Planning and Development; RC, REDi™ repair class (Almufti and Willford, 2013); SDC, ASCE 7–10 seismic design criteria (American Society of Civil Engineers, 2010); SSG, structural silicone glazing; VAV, variable air volume; Vg, velocity of gravity; VHB™ SGT, 3M™ VHB™ structural glazing tape; Vo, velocity of object; W, wide-flange beam; WUF-B, welded unreinforced flange-bolted web. Unit abbreviations: A, ampere; ft2, square foot; ft3/min, cubic foot per minute; g, acceleration due to gravity; in., inch; kVA, kilo-volt-ampere; lin ft, linear foot; mm, millimeter; plf, per linear foot] Table 11.  Forty-story steel-frame building structural components. Structural Components 286   The HayWired Earthquake Scenario—Engineering Implications Component 1 8 0 24 13 72 8 5 0 8 0 24 72 8 5 8 24 B1035.051 B1035.052 B1031.021c B1031.001 B1031.021b B1031.021c B1035.041 B1035.042 B1035.051 B1035.052 B1031.001 B1031.021b B1031.021c B1035.041 B1035.051 E-W qty B1035.042 NISTIR 32 10 5 0 8 20 12 4 6 5 0 8 13 20 12 4 N-S qty Each Each Each Each Each Each Each Each Each Each Each Each Each Each Each Each Units 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] DS1 median (dispersion) [repair class] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] Building design Building design 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] Building design 0.02 (0.4) rad [RC3] Building design 0.04 (0.4) rad [RC3] Building design 0.02 (0.4) rad [RC3] Building design Building design Building design Building design Building design 0.02 (0.4) rad [RC3] Building design 0.04 (0.4) rad [RC3] Building design 0.02 (0.4) rad [RC3] Building design 0.02 (0.4) rad [RC3] Building design Building design Building design Source Column splices only located on the following floors: L3, L6, L9, L12, L15, L18, L21, L24, L27, L30, L33, and L36. Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≥W30 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≤W27 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≥W30 Welded column splices1, column W>300 plf L18–L29 Bolted shear tab gravity connections Welded column splices1, column 150 plf300 plf Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≤W27 Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≥W30 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≤W27 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≥W30 L30–Roof Bolted shear tab gravity connections Welded column splices1, column 150 plf300 plf Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≤W27 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≤W27 Floor Table 11.—Continued NA 0.11 (0.4) rad [RC3] NA NA NA 0.11 (0.4) rad [RC3] NA 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.05 (0.4) rad [RC3] 0.08 (0.4) rad [RC3] 0.05 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.05 (0.4) rad [RC3] 0.08 (0.4) rad [RC3] 0.05 (0.4) rad [RC3] 0.05 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] DS2 median DS3 median (dispersion) (dispersion) [repair class] [repair class] 0.025 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 287 24 B1031.021c B1035.042 B1035.052 Welded column splices1, column W>300 plf Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≥W30 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≥W30 Bolted shear tab gravity connections 8 0 24 0 5 8 B1035.041 B1035.042 B1035.051 B1035.052 B1031.021b B1031.021c 72 13 72 1 Column splices only located on the following floors: L3, L6, L9, L12, L15, L18, and L21. Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≤W27 Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≥W30 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≤W27 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≥W30 Welded column splices1, column 150 plf300 plf 8 B1031.001 L2–L10 B1031.001 24 B1035.052 L11–Roof 8 B1035.042 Pre-Northridge WUF-B beam-column joint, beam one side of column, beam depth≥W30 Pre-Northridge WUF-B beam-column joint, beam both sides of column, beam depth≥W30 Bolted shear tab gravity connections 72 B1031.001 Bolted shear tab gravity connections 13 E-W qty L1 B1031.011c NISTIR Steel column base plates, column W>300 plf Component B1 Floor 8 5 12 20 6 4 0 32 10 13 0 32 10 0 13 Source Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design Each Building design N-S qty Units 0.02 (0.4) rad [RC3] 0.02 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.04 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.04 (0.4) rad [RC3] 0.02 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] DS1 median (dispersion) [repair class] 0.04 (0.4) rad [RC3] 0.04 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] 0.017 (0.4) rad [RC3] DS3 median (dispersion) [repair class] 0.1 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] NA 0.05 (0.4) rad [RC3] 0.05 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] NA NA 0.03 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.03 (0.4) rad [RC3] 0.08 (0.4) rad [RC3] 0.11 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.05 (0.4) rad [RC3] 0.08 (0.4) rad [RC3] 0.11 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.025 (0.4) rad [RC3] 0.08 (0.4) rad [RC3] 0.11 (0.4) rad [RC3] DS2 median (dispersion) [repair class] 0.07 (0.4) rad [RC3] [NISTIR, National Institute of Standards and Technology Interagency Reports. E-W qty, east-west quantity per floor; N-S qty, north-south quantity per floor—note that components with only one quantity listed are nondirectional. DS, damage state. Other abbreviations: A, area; HVAC, heating, ventilation, and air conditioning; IGU, insulating glass unit; FT IGU, fully tempered insulating glass unit; Ip, component importance factor; NA, not applicable; Norm Qty, National Institute of Standards and Technology Normative Quantity estimation tool packaged with Federal Emergency Management Agency’s Performance Assessment Calculation Tool; OSHPD, California Office of Statewide Planning and Development; RC, REDi™ repair class (Almufti and Willford, 2013)—1=reoccupancy, 2=functional recovery, and 3=full recovery; SDC, ASCE 7–10 seismic design criteria (American Society of Civil Engineers, 2010); SSG, structural silicone glazing; VAV, variable air volume; Vg, velocity of gravity; VHB™ SGT, 3M™ VHB™ structural glazing tape; Vo, velocity of object; W, wide-flange beam; WUF-B, welded unreinforced flange-bolted web. Unit abbreviations: A, ampere; ft2, square foot; ft3/min, cubic foot per minute; g, acceleration due to gravity; in., inch; kVA, kilo-volt-ampere; lin ft, linear foot; mm, millimeter; plf, per linear foot] Table 12.  Twenty-story steel-frame building structural components. 288   The HayWired Earthquake Scenario—Engineering Implications L1 B1 B3–B2 B4 Floor Concrete link beam, diagonally reinforced, aspect ratio between 1.0 and 2.0, beam>24 in. wide and depth<30 in. Slender concrete wall, 30 in. thick, 12 ft high, 15 ft long Slender concrete wall, 30 in. thick, 12 ft high, 30 ft long Post-tensioned concrete flat slabscolumns with shear reinforcing 024 in. wide and depth<30 in. Slender concrete wall, 30 in. thick, 12 ft high, 15 ft long Slender concrete wall, 30 in. thick, 12 ft high, 30 ft long Post-tensioned concrete flat slabscolumns with shear reinforcing 024 in. wide and depth<30 in. Slender concrete wall, 30 in. thick, 12 ft high, 15 ft long Slender concrete wall, 30 in. thick, 12 ft high, 30 ft long Post-tensioned concrete flat slabscolumns with shear reinforcing 024 in. wide and depth<30 in. Rectangular low aspect ratio concrete walls 18–24 in. thick with double curtain and heights up to 15 ft Post-tensioned concrete flat slabscolumns with shear reinforcing 024 in. wide and depth<30 in. Post-tensioned concrete flat slabscolumns with shear reinforcing 024 in. wide and depth<30 in. Table 13.—Continued 28 2 B1042.021a 2 B1042.021a B1049.031 28 2 B1042.021a B1049.031 28 B1049.031 4.02 3.49 B1044.113 B1044.021 0.53 2 E-W qty B1044.111 B1042.021a NISTIR 4 4 5.23 4 0 5.23 4 N-S qty Each Each Each Each 144 ft2 Each Each 144 ft2 144 ft2 Each Units Building design Building design Building design Building design Building design Building design Building design Building design Building design Building design Source 0.04 (0.5) rad [RC3] 0.0352 (0.44) g [RC3] 0.0179 (0.38) g [RC1] 0.0352 (0.44) g [RC3] 0.0179 (0.38) g [RC1] 0.028 (0.5) rad [RC3] 0.04 (0.5) rad [RC3] 0.028 (0.5) rad [RC3] 0.0109 (0.3) rad [RC3] 0.0352 (0.44) g [RC3] 0.0179 (0.38) g [RC1] 0.0055 (0.36) rad [RC1] 0.0128 (0.35) rad [RC3] 0.0128 (0.35) rad [RC3] 0.04 (0.5) rad [RC3] DS2 median (dispersion) [repair class] 0.0352 (0.44) g [RC3] 0.0093 (0.5) rad [RC3] 0.0093 (0.5) rad [RC3] 0.028 (0.5) rad [RC3] DS1 median (dispersion) [repair class] 0.0179 (0.38) g [RC1] 0.0543 (0.95) g [RC3] NA 0.0543 (0.95) g [RC3] NA 0.013 (0.36) rad [RC3] 0.0543 (0.95) g [RC3] 0.0186 (0.45) rad [RC3] 0.0186 (0.45) rad [RC3] NA DS3 median (dispersion) [repair class] 0.0543 (0.95) g [RC3] 290   The HayWired Earthquake Scenario—Engineering Implications B3–B1 NA Floor Cold or hot potable water piping (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—large diameter welded steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A,B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A,B, BRACING FRAGILITY HVAC galvanized sheet-metal ducting less than 6 ft2 in cross-sectional area, SDC A or B HVAC galvanized sheet-metal ducting—6 ft2 crosssectional area or greater, SDC A or B HVAC drops/diffusers in suspended ceilings—no independent safety wires, SDC A or B VAV box with in-line coil, SDC A or B Traction elevator—applies to most California installations 1976 or later, most Western States installations 1982 or later and most other U.S. installations 1998 or later Traction elevator —applies to most California installations 1976 or later, most Western States installations 1982 or later and most other U.S. installations 1998 or later Prefabricated steel stair with steel treads and landings with no seismic joint Independent pendant lighting—nonseismic Component 0.144 0.854 0.854 0.336 0.547 0.547 0.72 0.192 8.64 6.72 D2022.011a D2022.011b D2022.021a D2031.021a D2031.021b D3041.011a D3041.012a D3041.031a D3041.041a 144 D2021.021a C3034.001 0.8 12 D1014.011_ridr C2011.001b 12 E-W qty D1014.011 NISTIR 1.2 N-S qty 10 Units 10 Units 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft Each Each Each Each Units Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Source 1.5 (0.5) g [RC1] 2.25 (0.4) g [RC1] 1.2 (0.5) g [RC3] 1.5 (0.4) g [RC3] 1.5 (0.4) g [RC3] 1.3 (0.4) g [RC3] 1.9 (0.4) g [RC2] 0.005 (0.6) rad [RC1] 0.6 (0.4) g [RC3] 1.5 (0.4) g [RC1] 0.55 (0.5) g [RC1] 1.2 (0.5) g [RC3] 0.002 (0.3) rad [RC2] DS1 median (dispersion) [repair class 0.002 (0.3) rad [RC2] NA 2.4 (0.5) g [RC3] 2.25 (0.4) g [RC3] 2.25 (0.4) g [RC3] NA 2.6 (0.5) g [RC3] NA NA 2.6 (0.4) g [RC3] 1.1 (0.5) g [RC3] 2.4 (0.5) g [RC3] NA NA NA NA NA NA NA NA NA 0.028 (0.45) rad [RC3] NA 0.017 (0.6) rad [RC3] NA DS2 median DS3 median (dispersion) (dispersion) [repair class] [repair class] 0.005 (0.3) NA rad [RC2] 0.005 (0.3) NA rad [RC2] [NISTIR, National Institute of Standards and Technology Interagency Reports. E-W qty, east-west quantity per floor; N-S qty, north-south quantity per floor—note that components with only one quantity listed are nondirectional. DS, damage state. Other abbreviations: A, area; HVAC, heating, ventilation, and air conditioning; IGU, insulating glass unit; FT IGU, fully tempered insulating glass unit; Ip, component importance factor; NA, not applicable; Norm Qty, National Institute of Standards and Technology Normative Quantity estimation tool packaged with Federal Emergency Management Agency’s Performance Assessment Calculation Tool; OSHPD, California Office of Statewide Planning and Development; RC, REDi™ repair class (Almufti and Willford, 2013)—1=reoccupancy, 2=functional recovery, and 3=full recovery; SDC, ASCE 7–10 seismic design criteria (American Society of Civil Engineers, 2010); SSG, structural silicone glazing; VAV, variable air volume; Vg, velocity of gravity; VHB™ SGT, 3M™ VHB™ structural glazing tape; Vo, velocity of object; W, wide-flange beam; WUF-B, welded unreinforced flange-bolted web. Unit abbreviations: A, ampere; ft2, square foot; ft3/min, cubic foot per minute; g, acceleration due to gravity; in., inch; kVA, kilo-volt-ampere; lin ft, linear foot; mm, millimeter; plf, per linear foot] Table 14.  Forty-story steel-frame building nonstructural components. Nonstructural Components Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 291 L1 Floor Cold or hot potable water piping (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—Large Diameter Welded Steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A,B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A,B, BRACING FRAGILITY HVAC galvanized sheet-metal ducting less than 6 ft2 in cross-sectional area, SDC A or B HVAC galvanized sheet-metal ducting—6 ft2 crosssectional area or greater, SDC A or B HVAC Drops/Diffusers in suspended ceilings—No independent safety wires, SDC A or B VAV box with in-line coil, SDC A or B Suspended ceiling, SDC A, B, area (A): 2502.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—large diameter welded steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, BRACING FRAGILITY HVAC galvanized sheet-metal ducting less than 6 ft2 in cross-sectional area, SDC A or B HVAC galvanized sheet-metal ducting—6 ft2 crosssectional area or greater, SDC A or B Suspended ceiling, SDC A, B, area (A): 2502.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—large diameter welded steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A,B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, BRACING FRAGILITY HVAC galvanized sheet-metal ducting less than 6 ft2 in cross-sectional area, SDC A or B HVAC galvanized sheet-metal ducting—6 ft2 crosssectional area or greater, SDC A or B HVAC drops/diffusers in suspended ceilings—No independent safety wires, SDC A or B Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—no bracing, SDC A or B, PIPING FRAGILITY Fire sprinkler drop standard threaded steel—dropping into unbraced lay-in tile soft ceiling—6 ft long drop maximum, SDC A or B Low voltage switchgear—capacity: 100 to <350 A— unanchored equipment that is not vibration isolated— EQUIPMENT FRAGILITY only Precast concrete panels 4.5 inches thick—in plane deformation Prefabricated steel stair with steel treads and landings with no seismic joint. Suspended ceiling, SDC A, B, area (A): 2502.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—large diameter welded steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Suspended ceiling, SDC A, B, area (A): 2502.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—large diameter welded steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, PIPING FRAGILITY Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—no bracing, SDC A or B, PIPING FRAGILITY Fire sprinkler drop standard threaded steel—dropping into unbraced lay-in tile soft ceiling—6 ft long drop maximum, SDC A or B Low voltage switchgear—capacity: 100 to <350 A— unanchored equipment that is not vibration isolated— EQUIPMENT FRAGILITY only Precast concrete panels 4.5 inches thick—in plane deformation Prefabricated steel stair with steel treads and landings with no seismic joint. Suspended ceiling, SDC A, B, area (A): 2502.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—large diameter welded steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, BRACING FRAGILITY HVAC galvanized sheet-metal ducting less than 6 ft2 in cross-sectional area, SDC A or B HVAC galvanized sheet-metal ducting—6 ft2 crosssectional area or greater, SDC A or B HVAC drops/diffusers in suspended ceilings—no independent safety wires, SDC A or B VAV box with in-line coil, SDC A or B Component 0.8 144 0.144 0.854 0.854 0.336 0.547 0.547 0.72 0.192 8.64 6.72 C3034.001 D2021.021a D2022.011a D2022.011b D2022.021a D2031.021a D2031.021b D3041.011a D3041.012a D3041.031a D3041.041a 6 D1014.011_ridr C2011.001b 6 E-W qty D1014.011 NISTIR 1.2 N-S qty 10 Units 10 Units 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft Each 1,000 lin ft Each Each Each Units Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty 0.005 (0.6) rad [RC1] Norm qty Norm qty Norm qty Norm qty Source 0.005 (0.3) rad [RC2] DS2 median (dispersion) [repair class] 0.005 (0.3) rad [RC2] NA NA NA NA 1.5 (0.5) g [RC1] 2.6 (0.5) g [RC3] NA 2.25 (0.4) g [RC1] 1.2 (0.5) g [RC3] 2.4 (0.5) g [RC3] 2.25 (0.4) g [RC3] 2.25 (0.4) g [RC3] NA NA 1.5 (0.4) g [RC3] 1.9 (0.4) g [RC2] 1.3 (0.4) g [RC3] NA NA NA NA 1.2 (0.5) g [RC3] 2.4 (0.5) g [RC3] 1.5 (0.4) g [RC3] NA NA NA NA NA DS3 median (dispersion) [repair class] NA 1.1 (0.5) g [RC3] 0.55 (0.5) g [RC1] 0.017 (0.6) rad 0.028 (0.45) rad [RC3] [RC3] NA 0.6 (0.4) g [RC3] 1.5 (0.4) g [RC1] 2.6 (0.4) g [RC3] 0.002 (0.3) rad [RC2] DS1 median (dispersion) [repair class] 0.002 (0.3) rad [RC2] [NISTIR, National Institute of Standards and Technology Interagency Reports. E-W qty, east-west quantity per floor; N-S qty, north-south quantity per floor—note that components with only one quantity listed are nondirectional. DS, damage state. Other abbreviations: A, area; HVAC, heating, ventilation, and air conditioning; IGU, insulating glass unit; FT IGU, fully tempered insulating glass unit; Ip, component importance factor; NA, not applicable; Norm Qty, National Institute of Standards and Technology Normative Quantity estimation tool packaged with Federal Emergency Management Agency’s Performance Assessment Calculation Tool; OSHPD, California Office of Statewide Planning and Development; RC, Redi™ repair class (Almufti and Willford, 2013)—1=reoccupancy, 2=functional recovery, and 3=full recovery; SDC, ASCE 7–10 seismic design criteria (American Society of Civil Engineers, 2010); SSG, structural silicone glazing; VAV, variable air volume; Vg, velocity of gravity; VHB™ SGT, 3M™ VHB™ structural glazing tape; Vo, velocity of object; W, wide-flange beam; WUF-B, welded unreinforced flange-bolted web. Unit abbreviations: A, ampere; ft2, square foot; ft3/min, cubic foot per minute; g, acceleration due to gravity; in., inch; kVA, kilo-volt-ampere; lin ft, linear foot; mm, millimeter; plf, per linear foot] Table 15.  Twenty-story steel-frame building nonstructural components. 298   The HayWired Earthquake Scenario—Engineering Implications L1 Floor 0.8 72 14.4 144 0.144 0.854 0.854 0.336 0.547 0.547 0.72 0.192 8.64 6.72 C2011.001b C3027.001 C3032.001b C3034.001 D2021.021a D2022.011a D2022.011b D2022.021a D2031.021a D2031.021b D3041.011a D3041.012a D3041.031a D3041.041a 1 D5012.021a 5.332 0.864 D4011.031a B2011.201a 1.92 D4011.021a Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—no bracing, SDC A or B, PIPING FRAGILITY Fire sprinkler drop standard threaded steel—dropping into unbraced lay-in tile soft ceiling—6 ft long drop maximum, SDC A or B Low voltage switchgear—capacity: 100 to <350 A—unanchored equipment that is not vibration isolated—EQUIPMENT FRAGILITY only Precast concrete panels 4.5 inches thick—in plane deformation Prefabricated steel stair with steel treads and landings with no seismic joint Raised access floor, nonseismically rated Suspended ceiling, SDC A, B, area (A): 2502.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—large diameter welded steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, BRACING FRAGILITY HVAC galvanized sheet-metal ducting less than 6 ft2 in cross-sectional area, SDC A or B HVAC galvanized sheet-metal ducting—6 ft2 crosssectional area or greater, SDC A or B HVAC drops/diffusers in suspended ceilings—no independent safety wires, SDC A or B VAV box with in-line coil, SDC A or B E-W qty NISTIR Component Table 15.—Continued 1.2 7.998 N-S qty 10 Units 10 Units 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft Each 1,000 lin ft Each Panel 600 ft2 Each 390 ft2 225 A 100 Units 1,000 lin ft Units Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty 0.005 (0.5) rad [RC3] 0.005 (0.6) rad [RC1] Norm qty Norm qty Norm qty Norm qty Norm qty Source NA NA NA NA 1.5 (0.5) g [RC1] 2.6 (0.5) g [RC3] NA 2.25 (0.4) g [RC1] 1.2 (0.5) g [RC3] 2.4 (0.5) g [RC3] 2.25 (0.4) g [RC3] 2.25 (0.4) g [RC3] NA NA 1.5 (0.4) g [RC3] 1.9 (0.4) g [RC2] 1.3 (0.4) g [RC3] NA NA NA NA 1.2 (0.5) g [RC3] 2.4 (0.5) g [RC3] 1.5 (0.4) g [RC3] NA NA 1.69 (0.25) g [RC3] NA NA 0.028 (0.45) rad NA NA NA DS3 median (dispersion) [repair class] NA 1.1 (0.5) g [RC3] 0.55 (0.5) g [RC1] 0.017 (0.6) rad 0.028 (0.45) rad [RC3] [RC3] NA 0.5 (0.5) g [RC2] 1.01 (0.25) g 1.45 (0.25) g [RC1] [RC3] NA 0.6 (0.4) g [RC3] 1.5 (0.4) g [RC1] 2.6 (0.4) g [RC3] NA NA 2.4 (0.4) g [RC3] NA 0.95 (0.4) g [RC3] 0.75 (0.4) g [RC2] DS1 median DS2 median (dispersion) (dispersion) [repair class] [repair class] 1.1 (0.4) g [RC2] 2.4 (0.5) g [RC3] Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 299 L2–L19 Floor 3.84 0.8 0.29 72 14.4 144 0.144 0.854 0.854 0.336 0.547 0.547 0.72 0.192 C1011.001a C2011.001b C3011.001a C3027.001 C3032.001b C3034.001 D2021.021a D2022.011a D2022.011b D2022.021a D2031.021a D2031.021b D3041.011a D3041.012a 1 D5012.021a 5.332 0.864 D4011.031a B2011.201a 1.92 D4011.021a Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—no bracing, SDC A or B, PIPING FRAGILITY Fire sprinkler drop standard threaded steel—dropping into unbraced lay-in tile soft ceiling—6 ft long drop maximum, SDC A or B Low voltage switchgear—capacity: 100 to <350 A—unanchored equipment that is not vibration isolated—EQUIPMENT FRAGILITY only Precast concrete panels 4.5 inches thick—in plane deformation Wall partition, type: gypsum with metal studs, full height, fixed below, fixed above Prefabricated steel stair with steel treads and landings with no seismic joint. Wall partition, type: gypsum and wallpaper, full height, fixed below, fixed above Raised access floor, nonseismically rated. Suspended ceiling, SDC A, B, area (A): 2502.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—large diameter welded steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, BRACING FRAGILITY HVAC galvanized sheet-metal ducting less than 6 ft2 in cross-sectional area, SDC A or B HVAC galvanized sheet-metal ducting—6 ft2 crosssectional area or greater, SDC A or B E-W qty NISTIR Component Table 15.—Continued 0.435 1.2 5.76 7.998 N-S qty 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft Each 1,000 lin ft Each Panel 600 ft2 900 ft2 Each 900 ft2 390 ft2 225 A 100 Units 1,000 lin ft Units 0.95 (0.4) g [RC3] NA 0.75 (0.4) g [RC2] 2.4 (0.4) g [RC3] DS1 median DS2 median (dispersion) (dispersion) [repair class] [repair class] 1.1 (0.4) g [RC2] 2.4 (0.5) g [RC3] Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty NA NA NA NA 2.25 (0.4) g [RC1] 1.2 (0.5) g [RC3] 2.4 (0.5) g [RC3] 1.5 (0.4) g [RC3] NA NA 1.5 (0.5) g [RC1] 2.6 (0.5) g [RC3] 1.5 (0.4) g [RC3] NA 1.2 (0.5) g [RC3] 2.4 (0.5) g [RC3] 2.25 (0.4) g [RC3] 2.25 (0.4) g [RC3] NA NA 1.69 (0.25) g [RC3] NA NA NA NA NA NA NA NA DS3 median (dispersion) [repair class] NA 1.1 (0.5) g [RC3] 0.55 (0.5) g [RC1] 0.005 (0.5) NA NA rad [RC3] 0.005 (0.4) 0.01 (0.3) rad 0.021 (0.2) rad rad [RC1] [RC1] [RC2] 0.005 (0.6) 0.017 (0.6) rad 0.028 (0.45) rad rad [RC1] [RC3] [RC3] 0.0021 (0.6) NA NA rad [RC1] Norm qty 0.5 (0.5) g [RC2] NA Norm qty 1.01 (0.25) g 1.45 (0.25) g [RC1] [RC3] Norm qty 0.6 (0.4) g [RC3] NA Norm qty 1.5 (0.4) g [RC1] 2.6 (0.4) g [RC3] Norm qty Norm qty Norm qty Source 300   The HayWired Earthquake Scenario—Engineering Implications L20 Floor 0.8 14.4 144 0.144 0.854 0.854 0.336 0.547 0.547 3 0.72 C3032.001b C3034.001 D2021.021a D2022.011a D2022.011b D2022.021a D2031.021a D2031.021b D3031.011c D3041.011a 1 D5012.021a C2011.001b 0.864 D4011.031a 5.332 6.72 1.92 D3041.041a D4011.021a B2011.201a 8.64 D3041.031a HVAC drops/diffusers in suspended ceilings—No independent safety wires, SDC A or B VAV box with in-line coil, SDC A or B Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—no bracing, SDC A or B, PIPING FRAGILITY Fire sprinkler drop standard threaded steel—dropping into unbraced lay-in tile soft ceiling—6 ft long drop maximum, SDC A or B Low voltage switchgear—capacity: 100 to <350 A—unanchored equipment that is not vibration isolated—EQUIPMENT FRAGILITY only Precast concrete panels 4.5 in. thick—in plane deformation Prefabricated steel stair with steel treads and landings with no seismic joint. Suspended ceiling, SDC A, B, area (A): 2502.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC A or B, PIPING FRAGILITY Hot-water piping—small diameter threaded steel— (diameter≤2.5 in.), SDC A or B, BRACING FRAGILITY Hot-water piping—large diameter welded steel— (diameter>2.5 in.), SDC A or B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC A, B, BRACING FRAGILITY Chiller—capacity: 350 to <750 tons—unanchored equipment that is not vibration isolated—equipment fragility only HVAC galvanized sheet-metal ducting less than 6 ft2 in cross-sectional area, SDC A or B E-W qty NISTIR Component Table 15.—Continued 1.2 7.998 N-S qty 1,000 lin ft 490 tons 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft Each 1,000 lin ft 600 ft2 Each 390 ft2 225 A 10 Units 10 Units 1,000 lin ft 10 Units Units Norm qty Arup estimator Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty 0.005 (0.5) rad [RC3] 0.005 (0.6) rad [RC1] Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Source DS2 median (dispersion) [repair class] NA NA NA NA NA NA 1.2 (0.5) g [RC3] 2.4 (0.5) g [RC3] 1.5 (0.5) g [RC1] 2.6 (0.5) g [RC3] NA 2.25 (0.4) g [RC1] 1.2 (0.5) g [RC3] 2.4 (0.5) g [RC3] NA 2.25 (0.4) g [RC3] 0.2 (0.4) g [RC3] 1.5 (0.4) g [RC3] NA NA 1.69 (0.25) g [RC3] NA NA 0.028 (0.45) rad NA NA NA NA NA DS3 median (dispersion) [repair class] NA 1.1 (0.5) g [RC3] 0.55 (0.5) g [RC1] 0.017 (0.6) rad 0.028 (0.45) rad [RC3] [RC3] 1.01 (0.25) g 1.45 (0.25) g [RC1] [RC3] NA 0.6 (0.4) g [RC3] 1.5 (0.4) g [RC1] 2.6 (0.4) g [RC3] NA NA 2.4 (0.4) g [RC3] NA 0.95 (0.4) g [RC3] 0.75 (0.4) g [RC2] NA 1.9 (0.4) g [RC2] 1.1 (0.4) g [RC2] 2.4 (0.5) g [RC3] DS1 median (dispersion) [repair class] 1.3 (0.4) g [RC3] Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 301 Roof Floor 0.864 8.5 1 D4011.031a D5012.013a D5012.021a 1 1.92 D4011.021a D5012.021a 6.72 6.5 D3041.041a D3052.011d 2 8.64 D3041.031a D3031.021d 0.192 D3041.012a HVAC galvanized sheet-metal ducting—6 ft2 crosssectional area or greater, SDC A or B HVAC Drops/Diffusers in suspended ceilings—No independent safety wires, SDC A or B VAV box with in-line coil, SDC A or B Air handling unit—capacity: 25,000 to <40,000 ft3/ min—unanchored equipment that is not vibration isolated—EQUIPMENT FRAGILITY only Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—no bracing, SDC A or B, PIPING FRAGILITY Fire sprinkler drop standard threaded steel—dropping into unbraced lay-in tile soft ceiling—6 ft long drop maximum, SDC A or B Motor control center—capacity: all —unanchored equipment that is not vibration isolated— EQUIPMENT FRAGILITY only Low voltage switchgear—capacity: 100 to <350 A—unanchored equipment that is not vibration isolated—EQUIPMENT FRAGILITY only Cooling tower—capacity: 750 to <1,000 tons— unanchored equipment that is not vibration isolated—EQUIPMENT FRAGILITY only Low voltage switchgear—capacity: 100 to <350 A—unanchored equipment that is not vibration isolated—EQUIPMENT FRAGILITY only E-W qty NISTIR Component Table 15.—Continued N-S qty 225 A 850 tons 225 A Each 10 Units 1,000 lin ft 10 Units 30,000 ft3/ min 10 Units 1,000 lin ft Units Norm qty Arup estimator Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Source NA NA 1.9 (0.4) g [RC2] 0.25 (0.4) g [RC3] 0.95 (0.4) g [RC3] NA NA NA NA 0.75 (0.4) g [RC2] 0.73 (0.45) g [RC3] 2.4 (0.4) g [RC3] 0.5 (0.4) g [RC3] 2.4 (0.4) g [RC3] 1.1 (0.4) g [RC2] 2.4 (0.5) g [RC3] 1.3 (0.4) g [RC3] DS2 median (dispersion) [repair class] 2.25 (0.4) g [RC3] NA DS1 median (dispersion) [repair class] 1.5 (0.4) g [RC3] NA NA NA NA NA NA NA NA NA DS3 median (dispersion) [repair class] NA 302   The HayWired Earthquake Scenario—Engineering Implications B3–B2 B4 NA Floor Fire sprinkler drop standard threaded steel—no ceiling—6 ft long drop maximum, SDC D, E, or F Traction elevator —applies to most California installations 1976 or later, most Western States installations 1982 or later and most other U.S. installations 1998 or later Traction elevator—applies to most California installations 1976 or later, most Western States installations 1982 or later and most other U.S. installations 1998 or later Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—with designed bracing, SDC D, E, or F (OSHPD or similar), PIPING FRAGILITY Fire sprinkler drop standard threaded steel—no ceiling—6 ft long drop maximum, SDC D, E, or F Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—with designed bracing, SDC D, E, or F (OSHPD or similar), PIPING FRAGILITY Component D4011.073a 4.14 5.69 4.14 D4011.073a D4011.024a 5.69 6 D1014.011_ridr D4011.024a 6 E-W qty D1014.011 NISTIR N-S qty 100 Each 1,000 lin ft 100 Each 1,000 lin ft Each Each Units Norm qty Norm qty Norm qty Norm qty Norm qty Norm qty Source 0.005 (0.3) rad [RC2] 3.4 (0.4) g [RC3] 3 (0.4) g [RC3] 3.4 (0.4) g [RC3] 3 (0.4) g [RC3] 1.9 (0.4) g [RC2] 2.6 (0.4) g [RC2] 1.9 (0.4) g [RC2] 2.6 (0.4) g [RC2] DS2 median (dispersion) [repair class] 0.005 (0.3) rad [RC2] 0.002 (0.3) rad [RC2] DS1 median (dispersion) [repair class] 0.002 (0.3) rad [RC2] NA NA NA NA NA DS3 median (dispersion) [repair class] NA [NISTIR, National Institute of Standards and Technology Interagency Reports. E-W qty, east-west quantity per floor; N-S qty, north-south quantity per floor—note that components with only one quantity listed are nondirectional. DS, damage state. Other abbreviations: A, area; HVAC, heating, ventilation, and air conditioning; IGU, insulating glass unit; FT IGU, fully tempered insulating glass unit; Ip, component importance factor; NA, not applicable; Norm Qty, National Institute of Standards and Technology Normative Quantity estimation tool packaged with Federal Emergency Management Agency’s Performance Assessment Calculation Tool; OSHPD, California Office of Statewide Planning and Development; RC, REDi™ repair class (Almufti and Willford, 2013)—1=reoccupancy, 2=functional recovery, and 3=full recovery; SDC, ASCE 7–10 seismic design criteria (American Society of Civil Engineers, 2010); SSG, structural silicone glazing; VAV, variable air volume; Vg, velocity of gravity; VHB™ SGT, 3M™ VHB™ structural glazing tape; Vo, velocity of object; W, wide-flange beam; WUF-B, welded unreinforced flange-bolted web. Unit abbreviations: A, ampere; ft2, square foot; ft3/min, cubic foot per minute; g, acceleration due to gravity; in., inch; kVA, kilo-volt-ampere; lin ft, linear foot; mm, millimeter; plf, per linear foot] Table 16.  Reinforced-concrete building nonstructural components. Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 303 B1 Floor Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—with designed bracing, SDC D, E, or F (OSHPD or similar), PIPING FRAGILITY Chiller—capacity: 100 to <350 tons—Equipment is hard anchored— EQUIPMENT FRAGILITY only Chiller—capacity: 100 to <350 tons—vibration isolated equipment that is not snubbed or restrained—ANCHORAGE FRAGILITY only Air handling unit—capacity: 5,000 to <10,000 ft3/ min—equipment is hard anchored— EQUIPMENT FRAGILITY only Air handling unit—capacity: 5,000 to <10,000 ft3/ min—equipment is hard anchored—ANCHORAGE FRAGILITY only Air handling unit—capacity: 10,000 to <25,000 ft3/ min—equipment is hard anchored— EQUIPMENT FRAGILITY only Air handling unit—capacity: 10,000 to <25,000 ft3/ min—equipment is hard anchored— ANCHORAGE FRAGILITY only Component Table 16.—Continued 1 1 1 1 D3052.013e D3052.013d D3052.013h D3052.013g 5.69 2 D3031.012d D4011.024a 2 E-W qty D3031.013e NISTIR N-S qty 1,000 lin ft 20,000 ft3/min 20,000 ft3/min 8,000 ft3/min 8,000 ft3/min 250 tons 250 tons Units Norm qty Norm qty Arup estimator Arup estimator Arup estimator Arup estimator Arup estimator Source DS2 median (dispersion) [repair class] NA NA NA NA NA NA 3.4 (0.4) g [RC3] DS1 median (dispersion) [repair class] 0.72 (0.2) g [RC2] 0.4 (0.5) g [RC3] 1.54 (0.6) g [RC2] 0.4 (0.5) g [RC3] 1.54 (0.6) g [RC2] 0.4 (0.5) g [RC3] 1.9 (0.4) g [RC2] NA NA NA NA NA NA DS3 median (dispersion) [repair class] NA 304   The HayWired Earthquake Scenario—Engineering Implications Floor 4.14 1 1 1 1 2 2 1 D4011.073a D5011.013k D5011.013j D5012.023h D5012.023g D5012.023k D5012.023j D5012.033k Fire sprinkler drop standard threaded steel—no ceiling—6 ft long drop maximum, SDC D, E, or F Transformer/primary service—capacity: 750 to 1,500 kVA—equipment is hard anchored— EQUIPMENT FRAGILITY only Transformer/primary service—capacity: 750 to 1,500 kVA—equipment is hard anchored— ANCHORAGE FRAGILITY only Low voltage switchgear— capacity: 750 to <1,200 A—equipment is hard anchored—EQUIPMENT FRAGILITY only Low voltage switchgear— capacity: 750 to <1,200 A—equipment is hard anchored—ANCHORAGE FRAGILITY only Low voltage switchgear— capacity: 1,200 to 2,000 A—equipment is hard anchored—EQUIPMENT FRAGILITY only Low voltage switchgear— capacity: 1,200 to 2,000 A—equipment is hard anchored—ANCHORAGE FRAGILITY only Distribution panel—capacity: 1,200 to 2,000 A—equipment is hard anchored— EQUIPMENT FRAGILITY only E-W qty NISTIR Component Table 16.—Continued N-S qty 1,600 A 1,600 A 1,600 A 800 A 800 A 1,000 kVa 1,000 kVa 100 Each Units Arup estimator Arup estimator Arup estimator Arup estimator Arup estimator Arup estimator Arup estimator Arup estimator Source DS2 median (dispersion) [repair class] 3 (0.4) g [RC3] NA NA NA NA NA NA NA DS1 median (dispersion) [repair class] 2.6 (0.4) g [RC2] 3.05 (0.5) g [RC2] 0.4 (0.5) g [RC3] 2.4 (0.4) g [RC2] 0.4 (0.5) g [RC3] 2.4 (0.4) g [RC2] 0.4 (0.5) g [RC3] 3.05 (0.4) g [RC2] NA NA NA NA NA NA NA DS3 median (dispersion) [repair class] NA Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 305 L1 Floor 0.02 0.02 0.88 0.88 0.06 0.06 0.54 0.54 D2021.023a D2021.023b D2022.013a D2022.013b D2022.023a D2022.023b D2031.023a D2031.023b 174 C3034.001 1 42 D5012.033j Distribution panel—capacity: 1,200 to 2,000 A—equipment is hard anchored—ANCHORAGE FRAGILITY only Suspended ceiling, SDC D, E (Ip=1.0), area (A) < 250, vertical and lateral support Independent pendant lighting— nonseismic Cold-water piping (diameter>2.5 in.), SDC D, E, F, PIPING FRAGILITY Cold-water piping (diameter>2.5 in.), SDC D,E,F, BRACING FRAGILITY Hot-water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC D, E, or F, PIPING FRAGILITY Hot-water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC D, E, or F, BRACING FRAGILITY Hot-water piping—large diameter welded steel—(diameter>2.5 in.), SDC D, E, or F, PIPING FRAGILITY Hot-water piping—large diameter welded steel—(diameter>2.5 in.), SDC D, E, or F, BRACING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC D, E, F, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC D, E, F, BRACING FRAGILITY E-W qty C3032.003a NISTIR Component Table 16.—Continued N-S qty 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft Each 250 ft2 1,600 A Units Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Norm qty Norm qty Arup estimator Source DS2 median (dispersion) [repair class] NA 1.95 (0.3) g [RC3] NA 4.1 (0.4) g [RC3] 2.25 (0.4) g [RC3] 1.1 (0.5) g [RC3] NA 4.1 (0.5) g [RC3] 2.25 (0.5) g [RC3] NA NA DS1 median (dispersion) [repair class] 0.4 (0.5) g [RC3] 1.6 (0.3) g [RC1] 0.6 (0.4) g [RC3] 2.25 (0.4) g [RC1] 1.5 (0.4) g [RC3] 0.55 (0.5) g [RC1] 2.25 (0.5) g [RC3] 2.25 (0.5) g [RC1] 1.5 (0.5) g [RC3] 3 (0.5) g [RC1] 2.25 (0.5) g [RC3] NA NA NA NA NA NA NA NA NA 2.07 (0.3) g [RC3] DS3 median (dispersion) [repair class] NA 306   The HayWired Earthquake Scenario—Engineering Implications Floor 0.45 0.45 0.06 0.06 0.06 0.58 8.2 2.08 D2051.013a D2051.013b D2061.023a D2061.023b D3041.011c D3041.012c D3041.031b D4011.024a Chilled water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC D, E, or F, PIPING FRAGILITY Chilled water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC D, E, or F, BRACING FRAGILITY Steam piping—large diameter welded steel—(diameter>2.5 in.), SDC D, E, or F, PIPING FRAGILITY Steam piping—large diameter welded steel—(diameter>2.5 in.), SDC D, E, or F, BRACING FRAGILITY HVAC galvanized sheet-metal ducting—6 ft2 cross-sectional area or greater, SDC D, E, or F HVAC galvanized sheet-metal ducting—6 ft2 cross-sectional area or greater, SDC D, E, or F HVAC Drops/Diffusers in suspended ceilings—No independent safety wires, SDC C Fire sprinkler water piping—horizontal mains and branches—old style Victaulic—thin wall steel—with designed bracing, SDC D, E, or F (OSHPD or similar), PIPING FRAGILITY E-W qty NISTIR Component Table 16.—Continued N-S qty 1,000 lin ft 100 Each 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft Units Norm qty Moehle and others, 2011 Norm qty Norm qty Norm qty Norm qty Moehle and others, 2011 Moehle and others, 2011 Source DS2 median (dispersion) [repair class] 1.1 (0.5) g [RC3] NA 4.1 (0.4) g [RC3] 2.25 (0.4) g [RC3] 4.5 (0.4) g [RC3] 4.5 (0.4) g [RC3] NA 3.4 (0.4) g [RC3] DS1 median (dispersion) [repair class] 0.55 (0.5) g [RC1] 2.25 (0.5) g [RC3] 2.25 (0.4) g [RC1] 1.5 (0.4) g [RC3] 3.75 (0.4) g [RC3] 3.75 (0.4) g [RC3] 1.3 (0.4) g [RC3] 1.9 (0.4) g [RC2] NA NA NA NA NA NA NA DS3 median (dispersion) [repair class] NA Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 307 L2–L42 Floor 0.02 1.66 D2021.013b D2022.013a 0.79 C3011.001a 0.02 2 C2011.001b D2021.013a 2.62 C1011.001a 1.86 68.9 D4011.053a Fire sprinkler drop standard threaded steel—dropping into braced lay-in tile soft ceiling—6 ft long drop maximum, SDC D, E, or F Curtain walls—unitized curtain wall (also generic unitized curtain wall), configuration: symmetric insulating glass units (dual-pane, equal-thickness IGU), lamination: not laminated, glass type: full tempered, details: 1-1/4 in. (32 mm) FT IGU [1/4 in. (6 mm) inner and outer panes], 4-sided SSG, VHB™ SGT; glass-frame clearance=0.43 in. (11 mm); aspect ratio=varies; sealant=wet Wall partition, type: gypsum with metal studs, full height, fixed below, fixed above Prefabricated steel stair with steel treads and landings with no seismic joint Wall partition, type: gypsum and wallpaper, full height, fixed below, fixed above Cold-water piping (diameter>2.5 in.), SDC D, E, F, PIPING FRAGILITY Cold-water piping (diameter>2.5 in.), SDC D, E, F, BRACING FRAGILITY Hot-water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC D, E, or F, PIPING FRAGILITY E-W qty B2022.201 NISTIR Component Table 16.—Continued 0.79 2.62 68.9 N-S qty 1,000 lin ft 1,000 lin ft 1,000 lin ft 100 lin ft Each 100 lin ft 30 ft2 100 Each Units Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Norm qty Moehle and others, 2011 Norm qty Moehle and others, 2011 Norm qty Source 0.021 (0.2) rad [RC2] 0.04 (0.3) rad [RC3] DS3 median (dispersion) [repair class] NA NA 4.1 (0.4) g [RC3] 2.25 (0.4) g [RC3] 1.1 (0.5) g [RC3] 2.25 (0.4) g [RC1] 1.5 (0.4) g [RC3] 0.55 (0.5) g [RC1] NA NA NA NA 0.017 (0.6) rad [RC3] 0.028 (0.45) rad [RC3] 0.01 (0.3) rad [RC1] 0.03 (0.3) rad [RC3] DS2 median (dispersion) [repair class] 2.25 (0.4) g [RC3] 0.0021 (0.6) rad [RC1] 0.005 (0.6) rad [RC1] 0.005 (0.4) rad [RC1] 0.01 (0.3) rad [RC2] DS1 median (dispersion) [repair class] 1.5 (0.4) g [RC2] 308   The HayWired Earthquake Scenario—Engineering Implications Floor 1.66 0.06 0.06 1.43 1.43 1.47 1.47 0.6 0.58 8.2 D2022.013b D2022.023a D2022.023b D2031.023a D2031.023b D2051.013a D2051.013b D3041.001c D3041.011c D3041.032c Hot-water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC D, E, or F, BRACING FRAGILITY Hot-water piping—large diameter welded steel—(diameter>2.5 in.), SDC D, E, or F, PIPING FRAGILITY Hot-water piping—large diameter welded steel—(diameter>2.5 in.), SDC D, E, or F, BRACING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC D, E, F, PIPING FRAGILITY Sanitary-waste piping—cast iron with bell and spigot couplings, SDC D, E, F, BRACING FRAGILITY Chilled water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC D, E, or F, PIPING FRAGILITY Chilled water piping—small diameter threaded steel—(diameter≤2.5 in.), SDC D, E, or F, BRACING FRAGILITY HVAC in-line fan, fan independently supported and vibration isolators, SDC D, E, F HVAC galvanized sheet-metal ducting less than 6 ft2 in cross-sectional area, SDC D, E, or F HVAC drops/diffusers without ceilings—supported by ducting only—No independent safety wires, SDC D, E, or F E-W qty NISTIR Component Table 16.—Continued N-S qty 100 Each 1,000 lin ft 10 Each 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft 1,000 lin ft Units Moehle and others, 2011 Norm qty Arup estimator Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Moehle and others, 2011 Source DS2 median (dispersion) [repair class] NA 4.1 (0.5) g [RC3] 2.25 (0.5) g [RC3] NA NA 1.1 (0.5) g [RC3] NA 2.6 (0.4) g [RC3] 2.25 (0.4) g [RC3] NA DS1 median (dispersion) [repair class] 2.25 (0.5) g [RC3] 2.25 (0.5) g [RC1] 1.5 (0.5) g [RC3] 3 (0.5) g [RC1] 2.25 (0.5) g [RC3] 0.55 (0.5) g [RC1] 2.25 (0.5) g [RC3] 2.25 (0.4) g [RC2] 1.5 (0.4) g [RC3] 1.5 (0.4) g [RC3] NA NA NA NA NA NA NA NA NA DS3 median (dispersion) [repair class] NA Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 309 Floor NISTIR VAV box with in-line coil, D3041.041b SDC C Fire sprinkler water pipD4011.024a ing—horizontal mains and branches—old style Victaulic—thin wall steel—with designed bracing, SDC D, E, or F (OSHPD or similar), PIPING FRAGILITY Fire sprinkler drop standard D4011.073a threaded steel—no ceiling—6 ft long drop maximum, SDC D, E, or F Curtain walls—unitized curtain B2022.201 wall (also generic unitized curtain wall), configuration: symmetric insulating glass units (dual-pane, equal-thickness IGU), lamination: not laminated, glass type: full tempered, details: 1-1/4 in. (32 mm) FT IGU [1/4 in. (6 mm) inner and outer panes], 4-sided SSG, VHB™ SGT; glass-frame clearance=0.43 in. (11 mm); aspect ratio=varies; sealant=wet Wall partition, type: gypsum C1011.001a-racking with metal studs, full height, fixed below, fixed above Prefabricated steel stair with C2011.001b steel treads and landings with no seismic joint. Wall partition, type: gypsum C3011.001a-racking and wallpaper, full height, fixed below, fixed above Wall partition, type: gypsum C3011.002a-racking and ceramic tile, full height, fixed below, fixed above Component Table 16.—Continued 4.51 4.51 2.38 3.43 2.38 3.43 2 68.9 100 lin ft 100 lin ft Each 100 lin ft 30 ft2 100 Each 1.39 68.9 1,000 lin ft 2.55 Units 10 Each N-S qty 0.6 E-W qty 3 (0.4) g [RC3] 2.6 (0.4) g [RC2] 0.0021 (0.6) rad [RC1] 0.005 (0.6) rad [RC1] 0.005 (0.4) rad [RC1] 0.021 (0.2) rad [RC2] 0.04 (0.3) rad [RC3] NA NA DS3 median (dispersion) [repair class] NA 0.0071 (0.45) rad [RC1] NA NA NA 0.017 (0.6) rad [RC3] 0.028 (0.45) rad [RC3] 0.01 (0.3) rad [RC1] 0.03 (0.3) rad [RC3] 3.4 (0.4) g [RC3] 1.9 (0.4) g [RC2] 0.01 (0.3) rad [RC2] DS2 median (dispersion) [repair class] NA DS1 median (dispersion) [repair class] 1.9 (0.4) g [RC2] Moehle and 0.0021 (0.6) rad [RC1] others, 2011 Norm qty Moehle and others, 2011 Norm qty Moehle and others, 2011 Norm qty Norm qty Norm qty Source 310   The HayWired Earthquake Scenario—Engineering Implications Roof Floor 3 4 4 1 1 D3031.023g D3041.103b D3041.103a D3052.013h D3052.013g 1 3 D5012.023b Low voltage switchgear— capacity: 100 to <350 A—equipment is hard anchored—EQUIPMENT FRAGILITY only Cooling tower—capacity: 350 to <750 tons—equipment is hard anchored—EQUIPMENT FRAGILITY only Cooling tower—capacity: 350 to <750 tons—equipment is hard anchored—EQUIPMENT FRAGILITY only HVAC fan—capacity: all— equipment is hard anchored—EQUIPMENT FRAGILITY only HVAC fan—capacity: all— equipment is hard anchored—ANCHORAGE FRAGILITY only Air handling unit—capacity: 10,000 to <25,000 ft3/ min—equipment is hard anchored— EQUIPMENT FRAGILITY only Air Handling Unit—Capacity: 10,000 to <25,000 ft3/ min—Equipment is hard anchored—ANCHORAGE FRAGILITY only E-W qty D3031.023h NISTIR Component Table 16.—Continued N-S qty 20,000 ft3/min 20,000 ft3/min Each Each 500 tons 500 tons 225 A Units Arup estimator Arup estimator Arup estimator Arup estimator Arup estimator Arup estimator Norm qty Source DS2 median (dispersion) [repair class] NA NA NA NA NA NA NA DS1 median (dispersion) [repair class] 2.4 (0.4) g [RC2] 1.52 (0.4) g [RC2] 1.2 (0.5) g [RC3] 4.8 (0.6) g [RC2] 1.2 (0.5) g [RC3] 1.54 (0.6) g [RC2] 0.4 (0.5) g [RC3] NA NA NA NA NA NA DS3 median (dispersion) [repair class] NA Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 311 312   The HayWired Earthquake Scenario—Engineering Implications Appendix 2. S-SF-B-43—40-Story Steel-Frame Building in San Francisco (Baseline Orientation) S-SF-B-43 Description This appendix summarizes the results of interest for the HayWired mainshock from the structural analysis and loss assessment of building S-SF-B-43—a 40-story steel-frame office building in San Francisco with the baseline orientation shown in figure 7. Engineering-Demand Parameters The “best estimate” EDPs were obtained from the NLRHA. To capture uncertainty associated with modeling and construction quality, the EDPs for each realization in the loss assessment follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building S-SF-B-43 sees low-moderate acceleration demands, with peak nondirectional median floor acceleration for the whole building at 0.63 g. Peak median interstory drift ratio for the whole building is 1 percent, and the peak residual interstory Direction 1 45 drift ratio is 0.04 percent. As shown in figures 16 and 17, the peak drifts occur around level 34 (31st superstructure floor). This is because beam fracturing and yielding is concentrated in the top third of the building as shown in figure 18. This concentration of damage is due to 1970s design procedure that distributed seismic design forces up the building height according to solely the first mode translational response. In addition, wind forces were applied as uniform up the height rather than in an inverted triangle as done in practice today. It should be noted that beam yielding does not necessarily mean repair is required. Conversely, the column performance for this building is good, with nearly all columns remaining elastic. All column splices remain elastic. Realized peak floor-acceleration demands for the building are shown in figure 19. Loss-Assessment Results The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document Direction 2 Nondirectional Figure 16.  Graphs showing realized peak interstory drift ratio demands in building S-SF-B-43 (40-story steel-frame building in San Francisco, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 4 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 .005 .01 .015 .02 0 .005 .01 .015 Story drift ratio .02 0 .005 .01 .015 .02 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 313 Direction 1 45 Nondirectional Direction 2 Figure 17.  Graphs showing realized peak residual interstory drift ratio demands in building S-SF-B-43 (40story steel-frame building in San Francisco, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 4 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 0.2 0.4 0.6 0.8 1×10-3 0 0.2 0.4 0.6 0.8 1×10-3 0 0.2 0.4 0.6 0.8 1×10-3 Maximum residual drift A B EXPLANATION Elastic Yielded Fractured Figure 18.  Diagrams showing beam performance for sample (A) long and (B) short elevations of building S-SF-B-43 (40-story steel-frame building in San Francisco, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 4 is the ground floor. 314   The HayWired Earthquake Scenario—Engineering Implications Direction 1 Direction 1 Direction 2 Direction 2 45 40 40 3535 35 35 3030 30 30 2525 25 25 2020 20 20 1515 15 15 1010 EXPLANATION10 10 55 5 5 00 Percentile 10 2510%ile 5025%ile 50%ile 7575%ile 9090%ile 00 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 11 0 00 0.2 0.2 0.4 0.4 0.6 0.6 Acceleration [g] Nondirectional Non-directional 45 4040 Floor Level Floor level 4545 0.8 0.8 11 0 00 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 Figure 19. Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building S-SF-B-43 (40-story steel-frame building in San Francisco, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 4 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 11 Acceleration, in g (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). building contents if they are known. Figure 21 shows the contribution of building component groups to realized median total repair cost. Table 17 shows realized median and 90thpercentile repair time and total downtime, and table 18 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. Damaged Components Table 17.  Realized median and 90th-percentile repair time and total downtime for building S-SF-B-43 (40-story steel-frame building in San Francisco, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The probability that a component type in building S-SFB-43 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 20. Repair Costs, Repair Time, Downtime, and Impending Factors The median repair cost for building S-SF-B-43 is 10.8 percent of the total building replacement value, or $15.1 million. The 90th-percentile total repair cost is 12.3 percent of the total building replacement value, or $17.1 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based on a construction cost estimate, including at minimum all structural and nonstructural components plus the value of damageable [REDi (Almufti and Willford, 2013)] Repair time Median repair time Median total downtime 90th-percentile repair time 90th-percentile downtime REDi repair class Functional Reoccupancy, recovery, in days in days 41 45 Full recovery, in days 52 248 288 292 61 126 132 374 388 395 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 315 EXPLANATION REDi repair class Reoccupancy Full recovery Functional Building component Low voltage switchgear (unanchored) Motor control center (unanchored) Fire sprinkler drops Fire sprinkler water piping (piping) Air handling unit (unanchored) HVAC drops/diffusers HVAC ducting HVAC ducting Cooling tower (unanchored) Chiller (unanchored) Sanitary waste piping (bracing) Heating hot water piping, large diameter (piping) Heating hot water piping, small diameter (bracing) Heating hot water piping, small diameter (piping) Potable water piping, large diameter (piping) Traction elevator Pendant lighting Suspended ceilings Raised access floors Wall coverings Prefabricated steel stairs (no seismic joint) Gypsum wall partitions Precast concrete cladding Pre-Northridge beam-column joints Pre-Northridge beam-column joints Figure 20.  Graph showing the percentage of realizations in which a building component type in building S-SF-B-43 (40-story steel-frame building in San Francisco, California; baseline orientation) incurs damage from the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)—reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning. 0 20 40 60 80 100 Percentage of realizations in which a repair class is triggered 6 5 EXPLANATION Building component group Air handling unit (unanchored) 23 53 Chiller (unanchored) Precast concrete cladding Traction elevator Gypsum wall partitions Other Pendant lighting 6 4 3 Figure 21.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building S-SF-B-43 (40-story steel-frame building in San Francisco, California; baseline orientation) due to the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. 316   The HayWired Earthquake Scenario—Engineering Implications Table 18.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building S-SF-B-43 (40-story steel-frame building in San Francisco, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Inspection Financing Engineer mobilization Median disruption, in days 5 55 0 Contractor mobilization Permitting 253 0 90th-percentile disruption, in days 10 200 0 354 0 Comment None. None. The median engineer mobilization time is zero even though some beam-joint joints are redamaged because REDi (Almufti and Willford, 2013) assumes the likelihood of an inspector seeing structural damage decreases as the percentage of structural components damaged decreases. None. The median permitting time is zero even though some beamjoint joints are redamaged because REDi (Almufti and Willford, 2013) assumes the likelihood of an inspector seeing structural damage decreases as the percentage of structural components damaged decreases. Appendix 3. S-SF-R-43—40-Story Steel-Frame Building in San Francisco (Rotated Orientation) S-SF-R-43 Description This appendix summarizes the results of interest for the HayWired mainshock from the structural analysis and loss assessment of building S-SF-R-43—a 40-story steel-frame office building in San Francisco with the rotated orientation shown in figure 7. Results are shown in figures 22–24. Engineering-Demand Parameters The simulation of EDPs follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building S-SF-R-43 sees low-moderate acceleration demands, with peak nondirectional median floor acceleration for the whole building at 0.56 g. Peak median interstory drift ratio for the whole building is 0.7 percent, and the peak residual interstory drift ratio is 0.025 percent. The peak interstory drift ratio is governed by the ground-floor frames. All beams and columns remain elastic. All column splices remain elastic. Loss-Assessment Results The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). Damaged Components The probability that a component type in building S-SF-R-43 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 25. Repair Costs, Repair Time, Downtime, and Impending Factors The median total repair cost for building S-SF-R-43 is 9.7 percent of the total building replacement value, or $13.5 million. The 90th-percentile total repair cost is 11.3 percent of the total building replacement value, or $15.7 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based on a construction cost estimate including at minimum all structural and nonstructural components plus the value of damageable building contents if they are known. Figure 26 shows the contribution of building component groups to realized median total repair cost. Table  19 shows realized median and 90th-percentile repair time and total downtime, and table 20 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 317 Direction Direction 1 1 Direction Direction22 45 40 40 35 35 35 35 30 30 30 30 25 25 25 25 20 20 20 20 15 15 15 15 10 10 10 55 5 00 0 0 0.005 .005 0.01 .01 0.015 .015 0 EXPLANATION Percentile 10 25 10%ile 25%ile 50 50%ile 75 75%ile 90%ile 90 0 0.005 .005 Nondirectional Non-directional 45 40 40 Floor Level Floor level 45 45 0.01 .01 0.015 .015 Story Drift Ratio Figure 22.  Graphs showing realized peak interstory drift ratio demands in building S-SF-R-43 (40-story steel-frame building in San Francisco, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 4 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 10 5 0 00 0.005 .005 0.01 .01 0.015 .015 Story drift ratio Direction 1 45 Direction 2 Nondirectional Figure 23.  Graphs showing realized peak residual interstory drift ratio demands in building S-SF-R-43 (40-story steel-frame building in San Francisco, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 4 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 0.5 1 1.5 2 2.5×10-4 0 0.5 1 1.5 2 Maximum residual drift 2.5×10-4 0 0.5 1 1.5 2 2.5×10-4 318   The HayWired Earthquake Scenario—Engineering Implications Direction 1 Direction 1 Direction 2 Direction 2 45 40 40 35 35 35 35 30 30 30 30 25 25 25 25 20 20 20 20 15 15 15 15 10 10 10 55 5 00 0 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 1 0 EXPLANATION Percentile 10 10%ile 25 25%ile 50 50%ile 75 75%ile 90%ile 90 0 0.2 0.2 0.4 0.4 0.6 0.6 Acceleration [g] Nondirectional Non-directional 45 40 40 Floor Level Floor level 45 45 0.8 0.8 1 Figure 24.  Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building S-SF-R-43 (40-story steel-frame building in San Francisco, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 4 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 10 5 0 00 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 1 Acceleration, in g EXPLANATION REDi repair class Reoccupancy Functional Full recovery Low voltage switchgear (unanchored) Figure 25.  Graph showing the percentage of realizations in which a building component type in building S-SF-R-43 (40-story steel-frame building in San Francisco, California; rotated orientation) incurs damage from the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)— reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning. Motor control center (unanchored) Fire sprinkler drops Fire sprinkler water piping (piping) Air handling unit (unanchored) Cooling tower (unanchored) Building component Chiller (unanchored) Sanitary waste piping (bracing) Heating hot water piping, large diameter (piping) Heating hot water piping, small diameter (bracing) Heating hot water piping, small diameter (piping) Traction elevator Pendant lighting Suspended ceilings Raised access floors Wall coverings Prefabricated steel stairs (no seismic joint) Gypsum wall partitions Precast concrete cladding 0 20 40 60 80 Percentage of realizations in which a repair class is triggered 100 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 319 EXPLANATION 7 Building component group 6 Air handling unit (unanchored) Chiller (unanchored) Gypsum wall partitions Pendant lighting Precast concrete cladding 21 Traction elevator 55 Other Figure 26.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building S-SF-R-43 (40-story steel-frame building in San Francisco, California; rotated orientation) due to the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. 5 4 2 Table 19.  Realized median and 90th-percentile repair time and total downtime for building S-SF-R-43 (40-story steel-frame building in San Francisco, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [REDi (Almufti and Willford, 2013)] Repair time Median repair time Median total downtime 90th-percentile repair time 90th-percentile downtime Reoccupancy, in days 37 250 54 364 REDi repair class Functional recovery, in days 39 288 102 390 Full recovery, in days 44 292 110 398 Table 20.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building S-SF-R-43 (40-story steel-frame building in San Francisco, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Median disruption, in days 90th-percentile disruption, in days Comment Inspection 5 10 None. Financing 80 222 None. 0 0 255 366 0 0 Engineer mobilization Contractor mobilization Permitting The median engineer mobilization is zero because no structural components are damaged. None. No structural damage. 320   The HayWired Earthquake Scenario—Engineering Implications Appendix 4. S-SF-B-20—20-Story Steel-Frame Building in San Francisco (Baseline Orientation) S-SF-B-20 Description and beams remain elastic for this building. All column splices remain elastic. This appendix summarizes the results for the HayWired mainshock of interest from the structural analysis and loss assessment of building S-SF-B-20—a 20-story steel-frame office building in San Francisco with the baseline orientation shown in figure 7. Results are shown in figures 27–29. Loss-Assessment Results The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). Engineering-Demand Parameters The simulation of EDPs follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building S-SF-B-20 sees low-moderate acceleration demands, with peak nondirectional median floor acceleration for the whole building at 0.63 g, slightly higher than the 40-story San Francisco building. Peak median interstory drift ratio for the whole building is 0.45 percent, and the peak residual interstory drift ratio is 0.006 percent. All columns Direction Direction 1 1 2525 Damaged Components The probability that a component type in building S-SFB-20 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 30. Direction Direction 2 2 25 Nondirectional Non-directional 25 20 20 1515 15 15 10 10 5 5 Floor Level Floor level 2020 Figure 27.  Graphs showing realized peak interstory drift ratio demands in building S-SF-B-20 (20-story steel-frame office building in San Francisco, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 1010 EXPLANATION Percentile 10 10%ile 25 25%ile 50 50%ile 75 75%ile 90%ile 90 55 00 0 0.002 .002 0.004 .004 0.006 .008 0.008 .006 0.01 .01 0 00 0.002 .002 0.004 .006 0.006 .008 0.008 .004 Story Drift Ratio Story drift ratio 0.01 .01 0 0 0 0.002 .002 0.004 .004 0.006 .006 0.008 .008 0.01 .01 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 321 Direction 1 25 Direction 2 Nondirectional Figure 28.  Graphs showing realized peak residual interstory drift ratio demands in building S-SF-B-20 (20-story steel-frame office building in San Francisco, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 20 Floor level 15 10 EXPLANATION 5 0 Percentile 10 25 50 75 90 0 0.5 1 1.5×10-4 0 0.5 1 1.5×10-4 0 0.5 1 1.5×10-4 Maximum residual drift Direction 1 Direction 1 Direction 2 Direction 2 25 Nondirectional Non-directional 25 2020 20 20 1515 15 15 10 10 Figure 29.  Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building S-SF-B-20 (20-story steel-frame office building in San Francisco, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. Floor Level Floor level 2525 1010 EXPLANATION Percentile 10 25 10%ile 25%ile50 50%ile75 75%ile 90%ile90 55 00 0 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 1 5 0 5 0 0.2 0.2 0.4 0.4 0.6 0.6 Acceleration [g] Acceleration, in g 0.8 0.8 1 0 00 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 1 1 322   The HayWired Earthquake Scenario—Engineering Implications EXPLANATION Building component REDi repair class Reoccupancy Functional Full recovery Figure 30.  Graph showing the percentage of realizations in which a building component type in building S-SF-B-20 (20-story steel-frame office building in San Francisco, California; baseline orientation) incurs damage from the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)— reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning. Fire sprinkler drops Fire sprinkler water piping (piping) Air handling unit HVAC drops/diffusers HVAC ducting HVAC ducting Cooling tower Chiller Sanitary waste piping (bracing) Heating hot water piping, large diameter (piping) Heating hot water piping, small diameter (bracing) Heating hot water piping, small diameter (piping) Potable water piping, large diameter (piping) Traction elevator Pendant lighting Suspended ceilings Raised access floors Wall coverings Prefabricated steel stairs (no seismic joint) Gypsum wall partitions Precast concrete cladding 0 20 40 60 80 100 Percentage of realizations in which a repair class is triggered Repair Costs, Repair Time, Downtime, and Impending Factors The median total repair cost for building S-SF-B-20 is 9.7 percent of the total building replacement value, or $5.1 million. The 90th-percentile total repair cost is 9.5 percent of the total building replacement value, or $6.6 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based on a construction cost estimate including at minimum all structural and nonstructural components plus the value of damageable building contents if they are known. Figure 31 shows the contribution of building component groups to realized median total repair cost. Table 21 shows realized median and 90th-percentile repair time and total downtime, and table 22 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. Table 21.  Realized median and 90th-percentile repair time and total downtime for building S-SF-B-20 (20-story steel-frame office building in San Francisco, California; baseline orientation) due to the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. [REDi (Almufti and Willford, 2013)] Repair time Median repair time Median total downtime 90th-percentile repair time 90th-percentile downtime Reoccupancy, in days 20 189 39 316 REDi repair class Functional recovery, in days 29 242 145 364 Full recovery, in days 35 243 159 364 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 323 11 10 EXPLANATION Building component group Gypsum wall partitions 3 Pendant lighting 3 Precast concrete cladding Traction elevator Other 73 Figure 31.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building S-SF-B-20 (20-story steel-frame office building in San Francisco, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Table 22.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building S-SF-B-20 (20-story steelframe office building in San Francisco, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Inspection Financing Engineer mobilization Contractor mobilization Permitting Median disruption, in days 5 101 0 90th-percentile disruption, in days 10 239 0 229 0 347 0 Comment None. None. The median engineer mobilization is zero because no structural components are damaged. None. No structural damage. 324   The HayWired Earthquake Scenario—Engineering Implications Appendix 5. S-SF-R-20—20-Story Steel-Frame Building in San Francisco (Rotated Orientation) S-SF-R-20 Description This appendix summarizes the results of interest for the HayWired mainshock from the structural analysis and loss assessment of building S-SF-R-20—a 20-story steel-frame office building in San Francisco with the rotated orientation shown in figure 7. Results are shown in figures 32–34. Engineering-Demand Parameters The simulation of EDPs follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building S-SF-R-20 sees low-moderate acceleration demands, with peak nondirectional median floor acceleration for the whole building at 0.68 g, slightly higher than the 40-story San Francisco building. Peak median interstory drift ratio for the whole building is 0.60 percent and the peak residual interstory drift ratio is 0.0075 percent. All columns Direction Direction 1 1 25 Direction Direction 2 2 Loss-Assessment Results The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). Damaged Components The probability that a component type in building S-SF-R-20 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 35. Nondirectional Non-directional 25 2020 20 20 1515 15 15 10 10 5 5 Figure 32.  Graphs showing realized peak interstory drift ratio demands in building S-SF-R-20 (20-story steel-frame office building in San Francisco, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. Floor Level Floor level 2525 and beams remain elastic for this building. All column splices remain elastic. 1010 EXPLANATION Percentile 10 25 10%ile 25%ile 50 50%ile 75 75%ile 90%ile 90 55 00 00 0.002 .002 0.004 .006 0.006 .008 0.008 .004 0.01 .01 0 00 0.002 .004 0.004 .006 0.006 .008 0.008 .002 Story Drift Ratio Story drift ratio 0.01 .01 0 0 0 0.002 .002 0.004 .004 0.006 .006 0.008 .008 0.01 .01 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 325 Direction 1 25 Direction 2 Nondirectional Figure 33.  Graphs showing realized peak residual interstory drift ratio demands in building S-SF-R-20 (20-story steel-frame office building in San Francisco, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 20 Floor level 15 10 EXPLANATION 5 0 Percentile 10 25 50 75 90 0 0.5 1 1.5×10-4 0 0.5 1 1.5×10-4 0 0.5 1 1.5×10-4 Maximum residual drift Direction 1 Direction 1 Direction 2 Direction 2 25 Nondirectional Non-directional 25 20 20 20 20 15 15 15 15 10 10 5 5 Figure 34.  Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building S-SF-R-20 (20-story steel-frame office building in San Francisco, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. Floor Level Floor level 25 25 10 10 EXPLANATION 55 00 Percentile 10%ile 10 25%ile 25 50%ile 75%ile 50 90%ile 75 90 0 0 0.5 0.5 11 1.5 1.5 0 0 0 0.5 0.5 1 Acceleration [g] Acceleration, in g 1.5 0 00 0.5 0.5 1 1 1.5 1.5 326   The HayWired Earthquake Scenario—Engineering Implications Building component EXPLANATION REDi repair class Reoccupancy Functional Full recovery Motor control center Fire sprinkler drops Fire sprinkler water piping (piping) Air handling unit HVAC drops/diffusers HVAC ducting HVAC ducting Cooling tower Chiller Sanitary waste piping (bracing) Sanitary waste piping (piping) Heating hot water piping, large diameter (piping) Heating hot water piping, small diameter (bracing) Heating hot water piping, small diameter (piping) Potable water piping, large diameter (piping) Traction elevator Pendant lighting Suspended ceilings Raised access floors Wall coverings Prefabricated steel stairs (no seismic joint) Gypsum wall partitions Precast concrete cladding 0 20 40 60 80 Figure 35.  Graph showing the percentage of realizations in which a building component type in building S-SF-R-20 (20-story steel-frame office building in San Francisco, California; rotated orientation) incurs damage from the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)—reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning. 100 Percentage of realizations in which a repair class is triggered Repair Costs, Repair Time, Downtime, and Impending Factors The median total repair cost in building S-SF-R-20 is 8.2 percent of the total building replacement value, or $5.7 million. The 90th-percentile total repair cost is 10.4 percent of the total building replacement value, or $7.3 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based on a construction cost estimate including at minimum all structural and nonstructural components plus the value of damageable building contents if they are known. Figure 36 shows the contribution of building component groups to realized median total repair cost. Table 23 shows realized median and 90th-percentile repair time and total downtime, and table 24 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. Table 23.  Realized median and 90th-percentile repair time and total downtime for building S-SF-R-20 (20-story steel-frame office building in San Francisco, California; rotated orientation) due to the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. [REDi (Almufti and Willford, 2013)] REDi repair class Repair time Median repair time Median total downtime Reoccupancy, in days Functional recovery, in days Full recovery, in days 23 33 41 186 251 253 90th-percentile repair time 47 189 199 90th-percentile downtime 304 361 363 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 327 13 EXPLANATION Building component group Gypsum wall partitions 11 Pendant lighting Precast concrete cladding 4 Traction elevator 3 Other 69 Figure 36.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building S-SF-R-20 (20-story steel-frame office building in San Francisco, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Table 24.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building S-SF-R-20 (20-story steel-frame office building in San Francisco, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Inspection Financing Engineer mobilization Contractor mobilization Permitting Median disruption, in days 5 93 0 90th-percentile disruption, in days 10 229 0 237 0 344 0 Comment None. None. The median engineer mobilization is zero because no structural components are damaged. None. No structural damage. 328   The HayWired Earthquake Scenario—Engineering Implications Appendix 6. S-OK-B-20—20-Story Steel-Frame Building in Oakland (Baseline Orientation) S-OK-B-20 Description This appendix summarizes the results of interest for the HayWired mainshock from the structural analysis and loss assessment of building S-OK-B-20—a 20-story steel-frame office building in Oakland with the baseline orientation shown in figure 7. Loss Assessment Results Engineering-Demand Parameters The simulation of EDPs follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building S-OK-B-20 sees moderate acceleration demands, with peak nondirectional median floor acceleration for the whole building at 1.04 g, significantly higher than the San Francisco buildings. Peak median interstory drift ratio for the whole building is 1.5 percent and the peak residual interstory drift ratio is 0.1 percent. As shown in figure 37, the peak residual drift occurs in the building’s long direction (north-south) around level 10. This is because, as shown in figure 38, the beams in the long direction see yielding and fracturing around the mid-height of the building. It should be noted that beam yielding doesn’t necessarily mean repair is Direction 1 25 required. Conversely, the columns all remain elastic during the analysis. All column splices remain elastic. Realized peak interstory drift ratio demands are shown in figure 39, and realized peak floor-acceleration demands for the building are shown in figure 40. The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). Damaged Components The probability that a component type in building S-OK-B-20 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 41. Direction 2 Nondirectional Figure 37.  Graphs showing realized peak residual interstory drift ratio demands in building S-OK-B-20 (20-story steel-frame office building in Oakland, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 20 Floor level 15 10 EXPLANATION Percentile 10 25 50 75 90 5 0 0 .005 .01 .015 .02 .025 .03 0 .005 .01 .015 .02 .025 .03 Story drift ratio 0 .005 .01 .015 .02 .025 .03 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 329 A B Figure 38.  Diagrams showing beam performance for sample (A) long and (B) short elevations of building S-OK-B-20 (20-story steel-frame office building in Oakland, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. EXPLANATION Elastic Yielded Fractured Direction 1 25 Direction 2 Nondirectional Figure 39.  Graphs showing realized peak interstory drift ratio demands in building S-OK-B-20 (20story steel-frame office building in Oakland, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 20 Floor level 15 10 EXPLANATION 5 0 Percentile 10 25 50 75 90 0 0.5 1 1.5 2×10-3 0 0.5 1 1.5 Maximum residual drift 2×10-3 0 0.5 1 1.5 2×10-3 330   The HayWired Earthquake Scenario—Engineering Implications Direction 1 Direction 1 Direction 2 Direction 2 25 Nondirectional Non-directional 25 20 20 20 20 15 15 15 15 10 10 10 10 55 5 Figure 40.  Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building S-OK-B-20 (20-story steel-frame office building in Oakland, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. Floor Level Floor level 25 25 EXPLANATION 00 00 0.5 0.5 1 1 1.5 1.5 22 0 Percentile 10 10%ile 25 25%ile 50 50%ile 75 75%ile 90%ile 90 0 0.5 0 0.5 1 1 Acceleration [g] 1.5 2 1.5 2 5 0 0 0 0.5 1 0.5 1 1.5 1.5 2 2 Acceleration, in g Building component EXPLANATION REDi repair class Reoccupancy Functional Full recovery Switchgear 100–350 A Motor control center Fire sprinkler drops Fire sprinkler water piping (piping) Air handling unit Variable air volume boxes HVAC drops/diffusers HVAC ducting HVAC ducting Cooling tower Chiller Sanitary waste piping (bracing) Sanitary waste piping (piping) Heating hot water piping, large diameter (piping) Heating hot water piping, small diameter (bracing) Heating hot water piping, small diameter (piping) Potable water piping, large diameter (piping) Traction elevator Traction elevator Pendant lighting Suspended ceilings Raised access floors Wall coverings Prefabricated steel stairs (no seismic joint) Gypsum wall partitions Precast concrete cladding Pre-Northridge beam-column joints Pre-Northridge beam-column joints Pre-Northridge beam-column joints Pre-Northridge beam-column joints Welded column splices Welded column splices 0 20 40 60 80 Percentage of realizations in which a repair class is triggered Figure 41.  Graph showing the percentage of realizations in which a building component type in building S-OK-B-20 (20-story steel-frame office building in Oakland, California; baseline orientation) incurs damage from the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)—reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning; A, ampere. 100 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 331 Repair Costs, Repair Time, Downtime, and Impending Factors The median total repair cost for building S-OK-B-20 is 17.5 percent of the total building replacement value, or $12.2 million. The 90th-percentile total repair cost is 20.7 percent of the total building replacement value, or $14.4 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based on a construction cost estimate including at minimum all structural and nonstructural components plus the value of damageable building contents if they are known. Figure 42 shows the contribution of building component groups to realized median total repair cost. Table 25 shows realized median and 90th-percentile repair time and total downtime, and table 26 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. 9 EXPLANATION Building component group Chiller 16 Suspended ceilings Gypsum wall partitions Traction elevator Pendant lighting Other Precast concrete cladding 58 8 Figure 42.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building S-OK-B-20 (20-story steel-frame office building in Oakland, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. 4 2 3 Table 25.  Realized median and 90th-percentile repair time and total downtime for building S-OK-B-20 (20-story steel-frame office building in Oakland, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [REDi (Almufti and Willford, 2013)] Repair time Median repair time Median total downtime 90th-percentile repair time 90th-percentile downtime Reoccupancy, in days 54 231 116 343 REDi repair class Functional recovery, in days 92 273 272 385 Full recovery, in days 100 277 274 388 Table 26.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building S-OK-B-20 (20-story steel-frame office building in Oakland, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Inspection Financing Engineer mobilization Contractor mobilization Permitting Median disruption, in days 5 45 46 254 33 90th-percentile disruption, in days 10 189 123 363 74 Comment None. None. Engineer mobilization is triggered by structural damage. None. Permitting is triggered by structural damage. 332   The HayWired Earthquake Scenario—Engineering Implications Appendix 7. S-OK-R-20—20-Story Steel-Frame Building in Oakland (Rotated Orientation) S-OK-R-20 Description columns all remain elastic during the analysis. All column splices remain elastic. This appendix summarizes the results of interest for the HayWired mainshock from the structural analysis and loss assessment of building S-OK-R-20—a 20-story steel-frame office building in Oakland with the rotated orientation shown in figure 7. Results are shown in figures 43–45. Loss-Assessment Results Engineering-Demand Parameters The simulation of EDPs follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building S-OK-R-20 sees moderate acceleration demands, with peak nondirectional median floor acceleration for the whole building at 1.06 g, significantly higher than the San Francisco buildings. Peak median interstory drift ratio for the whole building is 1.0 percent, and the peak residual interstory drift ratio is 0.007 percent. Roughly a quarter of the beams either yield or fracture on average. Conversely, the Direction 1 Direction 1 Direction 2 Direction 2 25 Damaged Components The probability that a component type in building S-OK-R-20 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 46. Nondirectional Non-directional 25 2020 20 20 1515 15 15 10 10 Figure 43.  Graphs showing realized peak interstory drift ratio demands in building S-OK-R-20 (20-story steel-frame office building in Oakland, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. Floor Level Floor level 2525 The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). 1010 EXPLANATION 55 00 0 Percentile 10 10%ile 25 25%ile 50 50%ile 75%ile 75 90%ile 90 5 0.005 .005 0.01 .01 0.015 .015 0.02 .02 0 0 0.005 .005 0.01 .01 0.015 .015 Story Drift Ratio Story drift ratio 5 0.02 .02 0 0 0 0.005 .005 0.01 .01 0.015 .015 0.02 .02 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 333 Direction 1 Direction 1 Direction 2 Direction 2 25 Nondirectional Non-directional 25 2020 20 20 1515 15 15 10 10 Figure 44.  Graphs showing realized peak residual interstory drift ratio demands in building S-OK-R-20 (20-story steel-frame office building in Oakland, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. Floor Level Floor level 2525 1010 EXPLANATION 55 00 00 0.5 0.5 11 -4 1.5 1.5×10 10 -4 Direction 1 Direction 1 25 25 0 00 0.5 0.5 11 Max Residual Drift 5 -4 1.5 1.5×10 10 -4 0 00 0.5 0.5 11 -4 1.5 1.5×10 10 -4 Maximum residual drift Direction 2 Direction 2 25 Nondirectional Non-directional 25 20 20 20 20 15 15 15 15 10 10 Figure 45.  Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building S-OK-R-20 (20-story steel-frame office building in Oakland, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 2 is the ground floor. Direction 1 is east-west and direction 2 is north-south. Floor Level Floor level Percentile 10 25 10%ile 25%ile 50 50%ile 75 75%ile 90%ile 90 5 10 10 EXPLANATION 55 00 Percentile 10 25 50 75 90 5 0 0.5 0.5 11 1.5 1.5 2 0 00 0.5 0.5 11 Acceleration [g] 1.5 1.5 Acceleration, in g 22 5 10%ile 25%ile 50%ile 75%ile 90%ile 0 00 0.5 0.5 11 1.5 1.5 2 334   The HayWired Earthquake Scenario—Engineering Implications Building component EXPLANATION REDi repair class Reoccupancy Functional Full recovery Switchgear 100–350 A Motor control center Fire sprinkler drops Fire sprinkler water piping (piping) Air handling unit Variable air volume boxes HVAC drops/diffusers HVAC ducting HVAC ducting Cooling tower Chiller Sanitary waste piping (bracing) Sanitary waste piping (piping) Heating hot water piping, large diameter (piping) Heating hot water piping, small diameter (bracing) Heating hot water piping, small diameter (piping) Potable water piping, large diameter (piping) Traction elevator Pendant lighting Suspended ceilings Raised access floors Wall coverings Prefabricated steel stairs (no seismic joint) Gypsum wall partitions Precast concrete cladding Pre-Northridge beam-column joints Pre-Northridge beam-column joints Pre-Northridge beam-column joints Pre-Northridge beam-column joints Welded column splices Welded column splices 0 20 40 60 Figure 46.  Graph showing the percentage of realizations in which a building component type in building S-OK-R-20 (20-story steel-frame office building in Oakland, California; rotated orientation) incurs damage from the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)—reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning; A, ampere. 80 100 Percentage of realizations in which a repair class is triggered Repair Costs, Repair Time, Downtime, and Impending Factors The median total repair cost for building S-OK-R-20 is 16.5 percent of the total building replacement value, or $11.5 million. The 90th-percentile total repair cost is 18.7 percent of the total building replacement value, or $13.1 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based on a construction cost estimate including at minimum all structural and nonstructural components plus the value of damageable building contents if they are known. Figure 47 shows the contribution of building component groups to realized median total repair cost. Table 27 shows realized median and 90th-percentile repair time and total downtime, and table 28 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. Table 27.  Realized median and 90th-percentile repair time and total downtime for building S-OK-R-20 (20-story steel-frame office building in Oakland, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [REDi (Almufti and Willford, 2013)] REDi repair class Repair time Median repair time Median total downtime Reoccupancy, in days Functional recovery, in days Full recovery, in days 54 82 91 224 269 274 90th-percentile repair time 97 237 241 90th-percentile downtime 333 371 378 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 335 10 EXPLANATION 15 Building component group Chiller Gypsum wall partitions Pendant lighting Precast concrete cladding 54 8 Suspended ceilings Traction elevator Other 4 6 3 Figure 47.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building S-OK-R-20 (20-story steel-frame office building in Oakland, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Table 28.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building S-OK-R-20 (20-story steel-frame office building in Oakland, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Median disruption, in days 90th-percentile disruption, in days Comment Inspection 5 10 None. Financing 41 175 None. Engineer mobilization Contractor mobilization Permitting 0 101 249 348 0 67 Engineer mobilization is triggered by structural damage. None. Permitting is triggered by structural damage. 336   The HayWired Earthquake Scenario—Engineering Implications Appendix 8. C-SF-B-46—42-Story Reinforced-Concrete Building in San Francisco (Baseline Orientation) C-SF-B-46 Description This appendix summarizes the results of interest for the HayWired mainshock from the structural analysis and loss assessment of building C-SF-B-46—a 42-story reinforced-concrete residential building in San Francisco with the baseline orientation shown in figure 10. Results are shown in figures 48–53. Engineering-Demand Parameters The simulation of EDPs follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building C-SF-B-46 sees low demands, with a peak interstory drift (IDR) for whole building of just 0.4 percent. Peak floor acceleration is 0.9 g. Peak coupling beam rotation is 0.006 radian (rad), far below 0.05–0.06 rad, the point at which significant shear strength degradation occurs. The core walls see no crushing. The core-wall rebar experiences little yielding, all occurring at the base of the core walls. Direction 1 50 The damage in partitions and slab-column joints is better correlated with racking drift than IDR. Therefore, racking drift was used as the EDP for these components. Racking drift is different from IDR in that it excludes rigid body rotation and includes vertical racking resulting from the relative vertical movement between the core walls and the perimeter columns. There are a few interesting things to note about the demands. First, figure 51 clearly shows the plastic hinge zone in the core wall at the ground floor level, as evidenced by the large spike in the wall rotation. However, this spike is modest, in an absolute sense, at about 0.00115 rad median. This compares favorably with the acceptable plastic hinge rotation per ASCE 41–13 (American Society of Civil Engineers, 2014) of 0.001–0.002 rad for unconfined walls. Second, figure 53 shows that the coupling beam rotations are fairly constant up the height of the building with the exception of a dip in rotations at level 24 (superstructure floor 20), due to a reduction in the core wall reinforcement ratio at level 26 by about half. This reduction in core wall reinforcement is also evidenced by the small spikes in core-wall rotation in figure 52. Coupling beam rotation demands were enveloped for all beams on each floor and thus, beam directionality was not considered. Despite this conservatism, beam rotations were very low and had virtually no impact on the loss and downtime assessment. Nondirectional Direction 2 Figure 48.  Graphs showing realized peak interstory drift ratio demands in building C-SF-B-46 (42-story reinforced-concrete residential building in San Francisco, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 2 4 6 8×10-3 0 2 4 6 Story drift ratio 8×10-3 0 2 4 6 8×10-3 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 337 Direction 1 50 Direction 2 Nondirectional Figure 49.  Graphs showing realized peak residual interstory drift ratio demands in building C-SF-B-46 (42story reinforced-concrete residential building in San Francisco, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear responsehistory analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 EXPLANATION Floor level 30 Percentile 10 25 50 75 90 25 20 15 10 5 0 0 0.5 1 1.5 2 2.5×10-4 0 0.5 1 1.5 2 2.5×10-4 0 0.5 1 1.5 2 2.5×10-4 Maximum residual drift 50 Direction 1 Direction 2 Nondirectional Figure 50.  Graphs showing realized peak racking drift ratio demands in building C-SF-B-46 (42-story reinforced-concrete residential building in San Francisco, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 .005 .01 .015 0 .005 .01 Racking drift .015 0 .005 .01 .015 338   The HayWired Earthquake Scenario—Engineering Implications Direction 1 Direction 1 Direction 2 Direction 2 50 45 45 40 40 40 40 35 35 35 35 30 30 30 30 25 25 20 20 15 15 10 10 5 5 25 25 20 20 EXPLANATION Percentile 10 25 50 75 90 15 15 10 10 10%ile 25%ile 50%ile 75%ile 90%ile 55 00 00 2 4 66 -4 8 8×10 10 -4 0 0 44 66 Effective Drift Effective drift Direction 1 50 2 Nondirectional Non-directional 50 45 45 Floor Level Floor level 50 50 -4 8 8×10 10 -4 0 0 0 22 44 66 Figure 51.  Graphs showing realized peak wall-rotation demands (radians) in building C-SF-B-46 (42-story reinforcedconcrete residential building in San Francisco, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. -4 8 8×10 10 -4 Nondirectional Direction 2 Figure 52.  Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building C-SF-B-46 (42-story reinforcedconcrete residential building in San Francisco, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 0.5 1 1.5 0 0.5 1 Acceleration, in g 1.5 0 0.5 1 1.5 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 339 Direction 1 Direction 1 Direction 2 Direction 2 50 Nondirectional Non-directional 50 45 45 45 45 40 40 40 40 35 35 35 35 30 30 30 30 25 25 25 25 20 20 20 20 15 15 15 15 10 10 5 5 Floor Level Floor level 50 50 Figure 53.  Graphs showing realized peak coupling-beam rotation demands (radians) in building C-SF-B-46 (42-story reinforced-concrete residential building in San Francisco, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. EXPLANATION Percentile 10 25 50 75 90 10 10 55 00 00 0.005 .005 0.01 .01 0.015 .015 0 00 0.005 .005 0.01 .01 Link Beam Chord Rotation 0.015 .015 0 10%ile 25%ile 50%ile 75%ile 90%ile 00 0.005 .005 0.01 .01 0.015 .015 Link beam chord rotation Loss-Assessment Results The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). Damaged Components The probability that a component type in building C-SF-B-46 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 54. Repair Costs, Repair Time, Downtime, and Impending Factors The median total repair cost for building C-SF-B-46 is 3.1 percent of the total building replacement value, or $5.5 million. The 90th-percentile total repair cost is 3.7 percent of the total building replacement value, or $6.5 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based off a construction cost estimate, including at minimum all structural and nonstructural components plus the value of damageable building contents if they are known. The total repair cost is dominated by wall partitions because there are a large number of partitions in a residential building. Figure 55 shows the contribution of building component groups to realized median total repair cost. Table 29 shows realized median and 90th-percentile repair time and total downtime, and table 30 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. 340   The HayWired Earthquake Scenario—Engineering Implications EXPLANATION REDi repair class Reoccupancy Functional Full recovery AHU 10,000–25,000 CFM AHU 5,000– <10,000 CFM Figure 54.  Graph showing the percentage of realizations in which a building component type in building C-SF-B-46 (42-story reinforced-concrete residential building in San Francisco, California; baseline orientation) incurs damage from the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)— reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning; AHU, air handling unit; CFM, cubic feet per minute. HVAC fan Variable air volume box Building component HVAC drops without ceilings HVAC drops in suspended ceilings HVAC Sm ducting Cooling tower Chiller Chilled water piping Large diameter hot water piping bracing Small diameter hot water piping Cold water piping bracing Elevator Pendant lighting Wall partition with tile Wall partition with wallpaper Stair Wall partition 0 20 40 60 80 100 Percentage of realizations in which a repair class is triggered Table 29.  Realized median and 90th-percentile repair time and total downtime for building C-SF-B-46 (42-story reinforced-concrete residential building in San Francisco, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [REDi (Almufti and Willford, 2013)] REDi repair class Repair time Median repair time Median total downtime 90th-percentile repair time 90th-percentile downtime Reoccupancy, in days Functional recovery, in days Full recovery, in days 3 15 109 126 224 323 5 27 155 194 328 434 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 341 2 20 EXPLANATION Building component group Elevator Wall partition 7 Wall partition with tile Wall partition with wallpaper 61 Other 10 Figure 55.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building C-SF-B-46 (42-story reinforced-concrete residential building in San Francisco, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Table 30.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building C-SF-B-46 (42story reinforced-concrete residential building in San Francisco, California; baseline orientation) due to the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Median disruption, in days 90th-percentile disruption, in days Comment Inspection 5 10 Financing 0 0 Total repair cost projections are <5 percent of the total replacement cost for median and 90th percentile, so it is assumed that the owner readily has funds available for these repairs. Engineer mobilization 0 0 No structural damage. 210 316 0 0 Contractor mobilization Permitting None. None. No structural damage. 342   The HayWired Earthquake Scenario—Engineering Implications Appendix 9. C-SF-R-46—42-Story Reinforced-Concrete Building in San Francisco (Rotated Orientation) C-SF-R-46 Description This appendix summarizes the results of interest for the HayWired mainshock from the structural analysis and loss assessment of building C-SF-R-46—a 42-story reinforced-concrete residential building in San Francisco with the rotated orientation shown in figure 10. Results are shown in figures 56–61. Engineering-Demand Parameters The simulation of EDPs follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building C-SF-R-46 sees low demands, with a peak interstory drift (IDR) for whole building of just 0.7 percent. Peak floor acceleration is 1.09 g. Peak coupling beam rotation is 0.009 rad, far below 0.05–0.06 rad, the point at which significant shear strength degradation occurs. The core walls see no crushing. The core-wall rebar experiences little yielding, all occurring at the base of the core walls. Direction 1 50 The damage in partitions and slab-column joints is better correlated with racking drift than IDR. Therefore, racking drift was used as the EDP for these components. Racking drift is different from IDR in that it excludes rigid body rotation and includes vertical racking resulting from the relative vertical movement between the core walls and the perimeter columns. There are a few interesting things to note about the demands. First, figure 59 clearly shows the plastic hinge zone in the core wall at the ground floor level, as evidenced by the large spike in the wall rotation. However, this spike is modest, in an absolute sense, at about 0.00115 rad median. This compares favorably with the acceptable plastic hinge rotation per ASCE 41–13 (American Society of Civil Engineers, 2014) of 0.001–0.002 rad for unconfined walls. Second, figure 61 shows that the coupling beam rotations are fairly constant up the height of the building with the exception of a dip in rotations at level 17 (superstructure floor 13), where the core wall and coupling beam width abruptly changes from 32 to 24 in. Coupling beam rotation demands were enveloped for all beams on each floor and thus, beam directionality was not considered. Despite this conservatism, beam rotations were very low and had virtually no impact on the loss assessment. Nondirectional Direction 2 Figure 56.  Graphs showing realized peak interstory drift ratio demands in building C-SF-R-46 (42-story reinforcedconcrete residential building in San Francisco, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 .005 .01 .015 0 .005 .01 Story drift ratio .015 0 .005 .01 .015 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 343 Direction 1 50 Direction 2 Nondirectional Figure 57.  Graphs showing realized peak residual interstory drift ratio demands in building C-SF-R-46 (42-story reinforcedconcrete residential building in San Francisco, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 EXPLANATION 35 Percentile 10 25 50 75 90 Floor level 30 25 20 15 10 5 0 1 2 3×10-4 0 1 2 3×10-4 0 1 2 3×10-4 Maximum residual drift 50 Direction 1 Nondirectional Direction 2 Figure 58.  Graphs showing realized peak racking drift ratio demands in building C-SF-R-46 (42-story reinforced-concrete residential building in San Francisco, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 .005 .01 .015 .02 0 .005 .01 .015 Racking drift .02 0 .005 .01 .015 .02 344   The HayWired Earthquake Scenario—Engineering Implications 50 Direction 1 Nondirectional Direction 2 Figure 59.  Graphs showing realized peak wall-rotation demands (radians) in building C-SF-R-46 (42-story reinforcedconcrete residential building in San Francisco, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION Percentile 10 25 50 75 90 10 5 0 0 0.5 1 1.5 2×10-3 0 0.5 1 1.5 2×10-3 0 0.5 1 1.5 2×10-3 Effective drift Direction 1 50 Direction 2 Nondirectional Figure 60.  Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building C-SF-R-46 (42-story reinforcedconcrete residential building in San Francisco, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 0.5 1 1.5 2 0 0.5 1 1.5 Acceleration, in g 2 0 0.5 1 1.5 2 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 345 Direction 1 Direction 1 Direction 2 Direction 2 50 45 45 40 40 40 35 35 35 35 30 30 30 25 25 20 20 25 20 20 EXPLANATION Percentile 10 15 25 50 75 10 90 15 15 10 10 10%ile 25%ile 50%ile 75%ile 90%ile 55 00 00 0.005 .005 0.01 .01 0.015 .015 Figure 61.  Graphs showing realized peak coupling-beam rotation demands (radians) in building C-SF-R-46 (42-story reinforced-concrete residential building in San Francisco, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 15 10 5 0 Nondirectional Non-directional 50 45 Floor Level Floor level 50 5 00 0.005 .005 0.01 .01 Link Beam Chord Rotation 0.015 .015 0 00 0.005 .005 0.01 .01 0.015 .015 Link beam chord rotation Loss-Assessment Results The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). Damaged Components The probability that a component type in building C-SF-R-46 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 62. Repair Costs, Repair Time, Downtime, and Impending Factors The median total repair cost for building C-SF-R-46 is 5.0 percent of the total building replacement value, or $8.9 million. The 90th-percentile total repair cost is 6.1 percent of the total building replacement value, or $10.8 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based off a construction cost estimate, including at minimum all structural and nonstructural components plus the value of damageable building contents if they are known. The total repair cost is dominated by wall partitions because there are a large number of partitions in a residential building. Figure 63 shows the contribution of building component groups to realized median total repair cost. Table 31 shows realized median and 90th-percentile repair time and total downtime, and table 32 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. 346   The HayWired Earthquake Scenario—Engineering Implications Building component EXPLANATION REDi repair class Reoccupancy Functional Full recovery Switchgear 1,200–2,000 A Fire sprinkler drop with ceiling Fire sprinkler piping-OSPHD bracing AHU 10,000–25,000 CFM AHU 5,000– <10,000 CFM HVAC fan Variable air volume box HVAC drops without ceilings HVAC drops in suspended ceilings HVAC small ducting HVAC in line fan Cooling tower Chiller Steam piping bracing Steam piping Chilled water piping bracing Chilled water piping Sanitary waste piping bracing Large diameter hot water piping bracing Large diameter hot water piping Small diameter hot water piping bracing Small diameter hot water piping Cold water piping bracing Cold water piping Elevator Pendant lighting Suspended ceiling Wall partition with tile Wall partition with wallpaper Stair Wall partition Coupling beam 0 20 40 60 Figure 62.  Graph showing the percentage of realizations in which a building component type in building C-SF-R-46 (42-story reinforcedconcrete residential building in San Francisco, California; rotated orientation) incurs damage from the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)—reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning; AHU, air handling unit; CFM, cubic feet per minute; A, ampere; OSPHD; California Office of Statewide Health Planning and Development. 80 100 Percentage of realizations in which a repair class is triggered Table 31.  Realized median and 90th-percentile repair time and total downtime for building C-SF-R-46 (42-story reinforced-concrete residential building in San Francisco, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [REDi (Almufti and Willford, 2013)] REDi repair class Repair time Median repair time Reoccupancy, in days Functional recovery, in days Full recovery, in days 6 16 183 136 239 414 90th-percentile repair time 11 26 251 90th-percentile downtime 213 350 539 Median total downtime Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 347 2 17 EXPLANATION Building component group Chiller 2 Elevator Wall partition Wall partition with tile 10 Wall partition with wallpaper 62 Other 7 Figure 63.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building C-SF-R-46 (42-story reinforced-concrete residential building in San Francisco, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Table 32.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building C-SF-R-46 (42-story reinforced-concrete residential building in San Francisco, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Median disruption, in days 90th-percentile disruption, in days Comment Inspection 5 10 None. Financing 59 198 None. 0 0 223 335 0 0 Engineer mobilization Contractor mobilization Permitting No structural damage. None. No structural damage. 348   The HayWired Earthquake Scenario—Engineering Implications Appendix 10. C-OK-B-46—42-Story Reinforced-Concrete Building in Oakland (Baseline Orientation) C-OK-B-46 Description This appendix summarizes the results of interest for the HayWired mainshock from the structural analysis and loss assessment of building C-OK-B-46—a 42-story reinforced-concrete residential building in Oakland with the baseline orientation shown in figure 10. Results are shown in figures 64–69. Engineering-Demand Parameters The simulation of EDPs follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building C-OK-B-46 sees low demands, with a peak interstory drift (IDR) for whole building of just 0.7 percent. Peak floor acceleration is 1.06 g. Peak coupling beam rotation is 0.009 rad, far below 0.05–0.06 rad, the point at which significant shear strength degradation occurs. The core walls see no crushing, with a peak compressive strain of 0.0017. The core-wall rebar experiences little yielding, with a maximum tensile strain at the base of the core walls of 0.003. Direction 1 50 The damage in partitions and slab-column joints is better correlated with racking drift than IDR. Therefore, racking drift was used as the EDP for these components. Racking drift is different from IDR in that it excludes rigid body rotation and includes vertical racking resulting from the relative vertical movement between the core walls and the perimeter columns. There are a few interesting things to note about the demands. First, figure 67 clearly shows the plastic hinge zone in the core wall at the ground floor level, as evidenced by the large spike in the wall rotation. However, this spike is modest, in an absolute sense, at about 0.00105 rad median. This compares favorably with the acceptable plastic hinge rotation per ASCE 41–13 (American Society of Civil Engineers, 2014) of 0.001– 0.002 rad for unconfined walls. Second, figure 69 shows that the coupling beam rotations are fairly constant up the height of the building. This is expected as the coupling beam sizes are the same for almost the full height of the building, with only the width changing with the core wall width one-third up the building height. Coupling beam rotation demands were enveloped for all beams on each floor and thus, beam directionality was not considered. Despite this conservatism, beam rotations were very low and had virtually no impact on the loss assessment. Nondirectional Direction 2 Figure 64.  Graphs showing realized peak interstory drift ratio demands in building C-OK-B-46 (42-story reinforcedconcrete residential building in Oakland, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 .005 .01 .015 0 .005 .01 Story drift ratio .015 0 .005 .01 .015 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 349 Direction 1 Direction 1 Direction 2 Direction 2 50 Direction 3 Non-directional 50 45 45 45 45 40 40 40 40 35 35 35 35 30 30 30 30 25 25 25 25 20 20 20 20 15 15 15 15 10 10 5 5 Floor Level Floor level 50 50 Figure 65.  Graphs showing realized peak residual interstory drift ratio demands in building C-OK-B-46 (42-story reinforcedconcrete residential building in Oakland, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. EXPLANATION Percentile 10 25 10%ile 25%ile 50 50%ile 75 75%ile 90 90%ile 10 10 5 00 00 0.5 0.5 11 1.5 1.5 2 2 -4 2.5 2.5×10 10 -4 0 00 11 1.5 1.5 Max Residual Drift 22 -4 2.5 2.5×10 10 -4 0 00 0.5 0.5 11 1.5 1.5 2 Maximum residual drift Direction 1 50 0.5 0.5 -4 2.5 2.5×10 10 -4 Nondirectional Direction 2 Figure 66.  Graphs showing realized peak racking drift ratio demands in building C-OK-B-46 (42-story reinforced-concrete residential building in Oakland, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION Percentile 10 25 50 75 90 10 5 0 0 .005 .01 .015 0 .005 .01 Racking drift .015 0 .005 .01 .015 350   The HayWired Earthquake Scenario—Engineering Implications Direction 1 50 Nondirectional Direction 2 Figure 67.  Graphs showing realized peak wall-rotation demands (radians) in building C-OK-B-46 (42-story reinforcedconcrete residential building in Oakland, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 0.5 1 1.5 2×10-3 0 0.5 1 1.5 2×10-3 0 0.5 1 1.5 2×10-3 Effective drift Direction 1 50 Direction 2 Nondirectional Figure 68.  Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building C-OK-B-46 (42-story reinforcedconcrete residential building in Oakland, California; baseline orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION Percentile 10 25 50 75 90 10 5 0 0 0.5 1 1.5 0 0.5 1 Acceleration, in g 1.5 0 0.5 1 1.5 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 351 Direction 1 50 Direction 2 Nondirectional Figure 69.  Graphs showing realized peak coupling-beam rotation demands (radians) in building C-OK-B-46 (42story reinforced concrete residential building in Oakland, California; baseline orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 EXPLANATION Percentile 10 25 50 75 90 15 10 5 0 0 .005 .01 .015 .02 0 .005 .01 .015 .02 0 .005 .01 .015 .02 Link beam chord rotation Loss-Assessment Results The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). Damaged Components The probability that a component type in building C-OKB-46 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 70. Repair Costs, Repair Time, Downtime, and Impending Factors The median total repair cost for building C-OK-B-46 is 4.9 percent of the total building replacement value, or $8.6 million. The 90th-percentile total repair cost is 5.3 percent of the total building replacement value, or $9.4 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based off a construction cost estimate, including at minimum all structural and nonstructural components plus the value of damageable building contents if they are known. The total repair cost is dominated by wall partitions because there are a large number of partitions in a residential building. Figure 71 shows the contribution of building component groups to realized median total repair cost. Table 33 shows realized median and 90th-percentile repair time and total downtime, and table 34 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. 352   The HayWired Earthquake Scenario—Engineering Implications Building component EXPLANATION REDi repair class Reoccupancy Functional Full recovery Switchgear 1,200–2,000 A Fire sprinkler drop with ceiling AHU 10,000–25,000 CFM AHU 5,000– <10,000 CFM HVAC fan Variable air volume box HVAC drops without ceilings HVAC drops in suspended ceilings HVAC small ducting Cooling tower Chiller Steam piping bracing Steam piping Chilled water piping bracing Chilled water piping Sanitary waste piping bracing Large diameter hot water piping bracing Large diameter hot water piping Small diameter hot water piping bracing Small diameter hot water piping Cold water piping bracing Elevator Pendant lighting Suspended ceiling Wall partition with tile Wall partition with wallpaper Stair Wall partition Coupling beam 0 20 40 60 Figure 70.  Graph showing the percentage of realizations in which a building component type in building C-OK-B-46 (42-story reinforced-concrete residential building in Oakland, California; baseline orientation) incurs damage from the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)— reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning; AHU, air handling unit; CFM, cubic feet per minute; A, ampere. 80 100 Percentage of realizations in which a repair class is triggered Table 33.  Realized median and 90th-percentile repair time and total downtime for building C-OK-B-46 (42-story reinforced-concrete residential building in Oakland, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario [REDi (Almufti and Willford, 2013)] REDi repair class Repair time Median repair time Median total downtime 90th-percentile repair time 90th-percentile downtime Reoccupancy, in days Functional recovery, in days Full recovery, in days 5 16 173 130 233 396 9 26 242 198 346 527 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 353 2 17 EXPLANATION Building component group Chiller 3 Elevator Wall partition 10 Wall partition with tile Wall partition with wallpaper 62 Other 7 Figure 71.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building C-OK-B-46 (42-story reinforced-concrete residential building in Oakland, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Table 34.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building C-OK-B-46 (42-story reinforced-concrete residential building in Oakland, California; baseline orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Median disruption, in days 90th-percentile disruption, in days Inspection 5 10 Financing 0 162 Engineer mobilization Contractor mobilization Permitting 0 0 219 329 0 0 Comment None. Total repair cost projections are <5 percent of the total replacement cost for median, so it is assumed that the owner readily has funds available for these repairs. No structural damage. None. No structural damage. 354   The HayWired Earthquake Scenario—Engineering Implications Appendix 11. C-OK-R-46—42-Story Reinforced-Concrete Building in Oakland (Rotated Orientation) C-OK-R-46 Description This appendix summarizes the results of interest for the HayWired mainshock from the structural analysis and loss assessment of building C-OK-R-46—a 42-story reinforced-concrete residential building in Oakland with the rotated orientation shown in figure 10. Results are shown in figures 72–77. Engineering-Demand Parameters The simulation of engineering-demand parameters follows the method recommended by FEMA P-58 (Applied Technology Council, 2012) for buildings with nonlinear response history analysis results available. This algorithm was developed by Yang and others (2009). Building C-OK-R-46 sees low demands, with a peak interstory drift (IDR) for whole building of just 0.7 percent. Peak floor acceleration is 1.09 g. Peak coupling beam rotation is 0.009 rad, far below 0.05–0.06 rad, the point at which significant shear strength degradation occurs. The core walls see no crushing. The core-wall rebar experiences little yielding, all occurring at the base of the core walls. Direction 1 50 The damage in partitions and slab-column joints is better correlated with racking drift than IDR. Therefore, racking drift was used as the EDP for these components. Racking drift is different from IDR in that it excludes rigid body rotation and includes vertical racking resulting from the relative vertical movement between the core walls and the perimeter columns. There are a few interesting things to note about the demands. First, figure 75 clearly shows the plastic hinge zone in the core wall at the ground floor level, as evidenced by the large spike in the wall rotation. However, this spike is modest, in an absolute sense, at about 0.00115 rad median. This compares favorably with the acceptable plastic hinge rotation per ASCE 41–13 (American Society of Civil Engineers, 2014) of 0.001–0.002 rad for unconfined walls. Second, figure 77 shows that the coupling beam rotations are fairly constant up the height of the building with the exception of a dip in rotations at level 17 (superstructure floor 13), where the core wall and coupling beam width abruptly changes from 32 to 24 in. Coupling beam rotation demands were enveloped for all beams on each floor and thus, beam directionality was not considered. Despite this conservatism, beam rotations were very low and had virtually no impact on the loss assessment. Nondirectional Direction 2 Figure 72.  Graphs showing realized peak interstory drift ratio demands in building C-OK-R-46 (42-story reinforcedconcrete residential building in Oakland, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 .005 .01 .015 0 .005 .01 Story drift ratio .015 0 .005 .01 .015 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 355 Direction 1 50 Direction 2 Direction 3 Figure 73.  Graphs showing realized peak residual interstory drift ratio demands in building C-OK-R-46 (42story reinforced-concrete residential building in Oakland, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION Percentile 10 25 50 75 90 10 5 0 0 1 2 3×10-4 1 0 2 3×10-4 0 1 2 3×10-4 Maximum residual drift Direction 1 50 Nondirectional Direction 2 Figure 74.  Graphs showing realized peak racking drift ratio demands in building C-OK-R-46 (42-story reinforced-concrete residential building in Oakland, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 2 4 6 8×10-4 0 2 4 6 Racking drift 8×10-4 0 2 4 6 8×10-4 356   The HayWired Earthquake Scenario—Engineering Implications Direction 1 50 Nondirectional Direction 2 Figure 75.  Graphs showing realized peak wall-rotation demands (radians) in building C-OK-R-46 (42-story reinforcedconcrete residential building in Oakland, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION Percentile 10 25 50 75 90 10 5 0 0 0.5 1 1.5 2×10-3 0 0.5 1 1.5 2×10-3 0 0.5 1 1.5 2×10-3 Effective drift Direction 1 50 Nondirectional Direction 2 Figure 76.  Graphs showing realized peak floor-acceleration demands (relative to acceleration due to gravity, g) in building C-OK-R-46 (42-story reinforcedconcrete residential building in Oakland, California; rotated orientation) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 15 EXPLANATION 10 Percentile 10 25 50 75 90 5 0 0 0.5 1 1.5 2 0 0.5 1 1.5 Acceleration, in g 2 0 0.5 1 1.5 2 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 357 Direction 1 50 Nondirectional Direction 2 Figure 77.  Graphs showing realized peak coupling-beam rotation demands (radians) in building C-OK-R-46 (42story reinforced-concrete residential building in Oakland, California; rotated orientation) for the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. The 50th percentile is taken directly from the nonlinear response-history analysis. Due to the explicit modeling of the substructure, level 5 is the ground floor. Direction 1 is east-west and direction 2 is north-south. 45 40 35 Floor level 30 25 20 EXPLANATION Percentile 10 25 50 75 90 15 10 5 0 0 .005 .01 .015 0 .005 .01 .015 0 .005 .01 .015 Link beam chord rotation Loss-Assessment Results The loss assessment is based on a Monte Carlo simulation with 1,000 realizations. The loss assessment was performed using the probabilistic approach outlined in the Federal Emergency Management Agency’s (FEMA) P-58 document (Applied Technology Council, 2012). The likely costs to repair and (or) replace damaged components are calculated based solely on FEMA P-58, whereas repair time and downtime are estimated using the FEMA P-58 based methodology outlined REDi (Almufti and Willford, 2013). Damaged Components The probability that a component type in building C-OKR-46 incurs damage that hinders either reoccupancy, functionality, or full recovery was examined in the realizations. The results are shown in figure 78. Repair Costs, Repair Time, Downtime, and Impending Factors The median total repair cost for building C-OK-R-46 is 5.0 percent of the total building replacement value, or $8.9 million. The 90th-percentile total repair cost is 6.2 percent of the total building replacement value, or $10.8 million. For this study, the total building replacement value is defined as the hard costs only required to replace the building, based off a construction cost estimate, including at minimum all structural and nonstructural components plus the value of damageable building contents if they are known. The total repair cost is dominated by wall partitions because there are a large number of partitions in a residential building. Figure 79 shows the contribution of building component groups to realized median total repair cost. Table 35 shows realized median and 90th-percentile repair time and total downtime, and table 36 shows realized median and 90th-percentile delays due to impeding factors to functional recovery. 358   The HayWired Earthquake Scenario—Engineering Implications Building component EXPLANATION REDi repair class Reoccupancy Functional Full recovery Switchgear 1,200–2,000 A Fire sprinkler drop with ceiling AHU 10,000–25,000 CFM AHU 5,000– <10,000 CFM HVAC fan Variable air volume box HVAC drops without ceilings HVAC drops in suspended ceilings HVAC small ducting Cooling tower Chiller Steam piping bracing Steam piping Chilled water piping bracing Chilled water piping Sanitary waste piping bracing Large diameter hot water piping bracing Large diameter hot water piping Small diameter hot water piping bracing Small diameter hot water piping Cold water piping bracing Elevator Pendant lighting Suspended ceiling Wall partition with tile Wall partition with wallpaper Stair Wall partition Coupling beam 0 20 40 60 80 Figure 78.  Graph showing the percentage of realizations in which a building component type in building C-OK-R-46 (42-story reinforced-concrete residential building in Oakland, California; rotated orientation) incurs damage from the hypothetical momentmagnitude-7.0 mainshock of the HayWired earthquake scenario. Damaged components are assigned to a given REDi repair class (Almufti and Willford, 2013)— reoccupancy, functional recovery, or full recovery. HVAC, heating, ventilation, air conditioning; AHU, air handling unit; CFM, cubic feet per minute; A, ampere. 100 Percentage of realizations in which a repair class is triggered Table 35.  Realized median and 90th-percentile repair time and total downtime for building C-OK-R-46 (42-story reinforced-concrete residential building in Oakland, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. [REDi (Almufti and Willford, 2013)] REDi repair class Repair time Median repair time Median total downtime Reoccupancy, in days Functional recovery, in days Full recovery, in days 6 16 184 139 245 415 90th-percentile repair time 11 26 251 90th-percentile downtime 222 359 555 Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 359 2 17 EXPLANATION Building component group Chiller 2 Elevator Wall partition Wall partition with tile 10 Wall partition with wallpaper 62 Other 7 Figure 79.  Pie chart showing the percentage contribution of building component groups to realized median total repair cost for building C-OK-R-46 (42-story reinforced-concrete residential building in Oakland, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Table 36.  Realized median and 90th-percentile delays due to impeding factors to functional recovery for building C-OK-R-46 (42-story reinforced-concrete residential building in Oakland, California; rotated orientation) due to the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Impeding factor Median disruption, in days 90th-percentile disruption, in days Comment Inspection 5 11 None. Financing 59 204 None. Engineer mobilization Contractor mobilization Permitting 0 0 229 342 0 0 No structural damage. None. No structural damage. 360   The HayWired Earthquake Scenario—Engineering Implications Appendix 12. Inventory of Existing Tall-Building Stock in San Francisco 50 Figure 80 summarizes this information (Molina-Hutt and others, 2017). This list should not be considered as definitive because it has not been verified by the San Francisco Department of Building Inspection and it was developed approximately 7 or 8 years ago. We are not aware of a similar database for tall buildings in Oakland. A 40 30 20 10 –2 –2 00 00 9 0 0 20 01 91 19 81 –1 99 0 19 19 71 –1 98 0 –1 97 0 61 19 51 –1 96 0 19 19 41 –1 95 0 –1 94 1 31 19 19 21 –1 93 0 92 –1 11 19 19 00 –1 91 0 0 Year range 40 Number of buildings Figure 80.  Graphs showing (A) number of tall buildings built in San Francisco, California, per decade between 1900 and 2010 and (B) type of lateral load-resisting system for tall buildings built in the city between 1960 and 1990 (Modified from MolinaHutt and others, 2016). Number of buildings constructed A list of tall buildings (more than 50 m/~160 ft in height) in San Francisco was compiled by the Structural Engineers Association of Northern California (SEAONC) (see MolinaHutt, 2017, table 37). There are roughly 230 buildings that fit this criterion. The building list is ordered by the year of construction and the type of lateral load-resisting system. EXPLANATION B Steel moment-resisting frame Other system Unknown system 30 20 10 0 <20 20–25 26–30 31–35 36–40 41–45 >45 Number of stories Table 37.  List of tall buildings (more than 50 meters/~160 feet in height) in San Francisco, California, compiled by the Structural Engineers Association of Northern California (SEAONC) (see Molina-Hutt, 2017). [MF, moment frame; CBF, concentrically braced frame; EBF, eccentrically braced frame; RC, reinforced concrete; NA, not applicable; --, no data] List number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Name Ritz-Carlton Club and Residences Mills Building Ferry Building Central Tower One Kearny Street Building The Merchants Exchange The Westin St. Francis [The Westin St. Francis] Whittel Building One Sixth Street Maxwell Hotel Humboldt Bank Building Adam Grant Building 209 Post Building Campton Place Hotel Hobart Building Street number Street 690 220 1 703 1 465 335 Market Montgomery Ferry Market Kearny California Powell 166 1 386 785 114 209 340 582 Geary 6th Geary Market Sansome Post Stockton Market Height, in meters 95 52 75 91 54 69 60 60 57 51 85 64 55 53 87 24 10 12 21 12 15 13 Year completed 1889 1892 1898 1898 1902 1904 1904 Lateral loadresisting system -Steel MF ------ 16 15 12 19 14 13 16 21 1907 1908 1908 1908 1908 1909 1913 1914 --------- Stories Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 361 Table 37.—Continued List number 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 Name 56 57 58 59 60 The Chancellor Hotel San Francisco City Hall 115 Sansome Street Southern Pacific Building 300 Montgomery JH Dollar Building Commercial Union Assurance Building Alexander Building 225 Bush Street 605 Market Street Huntington Hotel Pacific Gas & Electric Headquarters Kensington Park Hotel Bank of the Orient Building Pacific Bell Building Serrano Hotel The Mark Hopkins Hotel Omni San Francisco Hotel Clift Hotel Marines’ Memorial Club and Hotel Crown Tower Apartments 220 Sansome Street Hunter-Dulin Building 1090 Chestnut Co-op 945 Green Street Clay-Jones Apartments Russ Building Medico Dental Building Sir Francis Drake Hotel Shell Building McAllister Tower Apartments Hamilton Apartments 450 Sutter Cathedral Apartments Bellaire Tower Pacific National Bank Clarion Hotel Cosmo Pacific Coast Stock Exchange Tower Mills Tower [The Mills Building] Bureau of Citizenship and Immigration Building 1000 Green Apartments UCSF Medical Center Parnassus Medical Sciences Building Equitable Life One Bush Plaza 61 62 63 64 65 Industrial Indemnity Building Philip Burton Federal Building Bethlehem Steel Company HQ International Building Green Hill Tower Street number Street 433 1 115 1 300 351 315 155 225 605 1075 245 450 233 140 405 999 500 491 to 499 450 666 220 111 1090 945 1250 235 490 450 100 100 631 450 1201 1101 333 to 341 761 155 220 444 Powell Carlton B Goodlett Sansome Market Montgomery California Montgomery Montgomery Bush Market California Market Post Sansome New Montgomery Taylor California California Geary Post Post Sansome Sutter Chestnut Green Jones Montgomery Post Powell Bush McAllister O’Farrell Sutter California Green Montgomery Post Sansome Bush Washington Height, in meters 59 94 61 65 65 73 94 60 100 61 52 78 62 53 133 56 93 66 64 66 55 66 94 53 53 70 133 64 96 115 94 64 105 74 77 93 60 60 92 67 15 4 13 12 12 16 16 15 22 15 12 18 14 13 26 16 20 15 15 12 16 16 22 13 14 21 32 16 22 29 28 18 26 19 20 18 16 13 22 16 Year completed 1914 1915 1915 1916 1917 1920 1921 1921 1922 1922 1924 1924 1924 1924 1925 1925 1926 1926 1926 1926 1926 1926 1926 1927 1927 1927 1927 1928 1928 1929 1929 1929 1929 1930 1930 1930 1930 1930 1931 1944 Lateral loadresisting system --------Steel MF --Steel MF --Steel MF ----------------------------Steel MF Dual-system steel CBF/EBF; steel MF ------ Stories 1000 505 513 120 1 Green Parnassus Parnassus Montgomery Bush 51 77 70 108 94 16 18 17 25 20 1950 1954 1954 1955 1959 245 450 100 601 1070 California Golden Gate California California Green 70 95 52 107 65 17 21 13 22 21 1959 1959 1960 1961 1961 362   The HayWired Earthquake Scenario—Engineering Implications Table 37.—Continued List Name number 66 The Comstock 67 Fairmont Hotel Tower [The Fairmont San Francisco] 68 Grosvenor Suites 69 66 Cleary Court 70 10 Miller 71 Nob Hill Community Apartments 72 Hartford Building 73 One Maritime Plaza 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 Street number Street Height, in meters 55 99 16 29 Year completed 1961 1962 Lateral loadresisting system --- Stories 1333 950 Jones Mason 899 66 10 1170 650 300 Pine Cleary Miller Sacramento California Clay 70 61 70 61 142 121 20 18 22 19 34 27 1962 1963 1963 1963 1964 1964 1100 555 2200 405 Gough Market Sacramento Davis 66 95 65 80   22 20 25 1964 1964 1964 1965 Davis 67 22 1965 ----Steel MF Dual-system steel CBF/EBF; steel MF ---Dual-system RC wall/steel MF -- Gough Green Jackson 91 96 80 27 32 25 1965 1965 1965 ---- 111 Pine 76 19 1965 Royal Towers Archstone Fox Plaza Beal Bank Building Golden Gateway Center [Golden Gateway Center] Bechtel Building Bank of California Building 44 Montgomery Fontana West Fontana East Pacific Bell—Pine Street Building 1750 1390 180 550 Taylor Market Sansome Battery 101 108 76 67 29 29 17 22 1965 1966 1966 1967 RC core walls, RC gravity system --Steel MF -- 50 400 44 1050 1050 555 Beale California Montgomery North Point North Point Pine 100 95 172 80 80 88 23 22 43 18 18 16 1967 1967 1967 1967 1967 1967 Insurance Center Building 425 California Street 555 California Street One California The Sequoias McKesson Plaza Donatello Hotel Pacific Gas & Electric Building One Embarcadero Center [Embarcadero Center] Hilton Financial District Hilton Hotel San Francisco 475 Sansome Street 450 425 555 1 1400 1 501 77 355 Sansome California California California Geary Post Post Beale Clay 93 109 237 134 80 161 54 150 173 19 26 52 32 25 38 15 34 45 1967 1968 1969 1969 1969 1969 1969 1971 1971 750 333 475 Kearny O’Farrell Sansome 111 150 86 30 46 21 1971 1971 1971 50 California Street Transamerica Pyramid [Transamerica Center] 100 Pine Center 50 600 California Montgomery 148 260 37 48 1972 1972 --RC core walls, steel gravity -Steel MF 100 Pine 145 33 1972 Steel MF Carillon Tower 555 Market Street [Market Center] Pacific Heights Towers Macondray House [Golden Gateway Center] Golden Gateway Center [Golden Gateway Center] Cathedral Hill Tower The Summit Buckelew House [Golden Gateway Center] 111 Pine Street 440 1200 999 155 Steel MF -Steel MF --RC core walls, steel gravity system -Steel MF Steel MF --Steel MF -Steel MF Steel MF Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 363 Table 37.—Continued List Name number 108 The Westin St. Francis Hotel 109 Grand Hyatt San Francisco 110 San Francisco Marriott Union Square or Crowne Plaza 111 Holiday Inn 112 Hyatt Regency 113 211 Main Street 114 First Market Tower 115 425 Market Street 116 Twelve Hundred California 117 Two Embarcadero Center [Embarcadero Center] 118 221 Main Street 119 California Automobile Association Building 120 Chevron Tower [Market Center] 121 Hinode Tower 122 Spear Tower [One Market Plaza] 123 Steuart Tower [One Market Plaza] 124 California Building 125 Three Embarcadero Center[Embarcadero Center] 126 Bank of America Computer Center 127 1275 Market Street 128 Gramercy Towers 129 Bechtel Building 130 601 Montgomery Street 131 Shaklee Terraces 132 333 Market Street 133 595 Market Street 134 Bank of the West 135 22 4th Street 136 201 California 137 Two Transamerica Center 138 Providian Financial Building 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 101 California Street Four Embarcadero Center [Embarcadero Center] Telesis Tower 353 Sacramento 150 Spear 1 Ecker Square Montgomery Washington Tower 100 Spear Street Westin San Francisco Hotel—Market Street Renaissance Parc 55 101 Montgomery [101 Montgomery] United Commercial Bank Citicorp Center 50 Fremont Center 456 Montgomery Plaza Street number Street Height, in meters 120 108 95 32 35 29 Year completed 1972 1972 1972 Lateral loadresisting system ---- Stories 335 345 480 Powell Stockton Sutter 1500 5 211 525 425 1200 255 Van Ness Embarcadero Main Market Market California Clay 88 85 67 161 160 88 126 26 20 17 39 38 27 30 1972 1973 1973 1973 1973 1974 1974 --Steel MF Steel MF Steel MF -Steel MF 221 100 Main Van Ness 64 122 16 29 1974 1974 Steel MF Steel MF 575 1615 1 1 350 155 Market Sutter Market Market California Clay 175 55 172 111 99 126 40 15 43 27 23 31 1975 1975 1976 1976 1977 1977 Steel MF -Steel MF Steel MF -Steel MF Market Market California Fremont Montgomery Market Market Market Montgomery 4th California Sansome Mission 88 81 61 145 77 164 144 125 98 67 72 80 127 21 17 17 34 20 38 33 30 24 17 17 20 30 1977 1977 1978 1978 1978 1979 1979 1979 1979 1980 1980 1980 1981 101 55 California Clay 183 174 48 45 1982 1982 1 353 150 1 655 100 50 Montgomery Sacramento Spear Ecker Montgomery Spear 3rd 152 95 79 85 91 83 114 38 23 18 18 26 22 34 1982 1982 1982 1983 1983 1983 1984 ---MF -Steel MF Steel MF Steel MF --Steel MF Steel moment frame Dual-system steel CBF/EBF; steel MF Steel MF Dual-system steel CBF/EBF; steel MF Steel MF --Steel MF -Steel MF -- 55 101 555 1 50 456 Cyril Magnin Montgomery Montgomery Sansome Fremont Montgomery 107 123 86 168 183 115 32 28 18 43 43 26 1984 1984 1984 1984 1985 1985 -Steel MF -Steel MF Steel MF -- 1455 1275 1177 45 601 444 333 595 180 22 201 505 201 364   The HayWired Earthquake Scenario—Engineering Implications Table 37.—Continued List number 154 155 156 157 158 159 160 161 162 163 Street number Street 160 Spear Building Spear Street Terrace 333 Bush Street 345 California Center 301 Howard Street 88 Kearny Street 135 Main Street 123 Mission Street Continental Center 33 New Montgomery 160 201 333 345 301 88 135 123 250 33 Spear Spear Bush California Howard Kearny Main Mission Montgomery New Montgomery Height, in meters 78 75 151 212 92 94 90 124 69 65 164 165 90 New Montgomery 580 California 90 580 New Montgomery California 166 167 168 169 170 171 172 173 174 175 176 177 178 179 Hawthorne Plaza Central Plaza 388 Market Hotel Nikko Hilton San Francisco Hotel JW Marriott Hotel Stevenson Place Park Hyatt 100 First Plaza 505 Montgomery 49 Stevenson Street San Francisco Marriott Embarcadero West [Embarcadero Center] One Daniel Burnham Court West 75 455 388 222 333 500 71 333 100 505 49 55 275 1 180 181 182 183 184 185 186 187 188 189 190 191 192 88 Howard Street Fillmore Center I 101 Spear Street Hills Plaza 222 Second Street 235 Pine Street 634 Sansome Street 600 California Street Post International PacBell Center San Francisco Towers 101 Second Street Second Street Towers 193 194 W Hotel Avalon Towers North [Avalon Towers by the B..] Avalon Towers South [Avalon Towers by the Bay] 150 California 195 196 Name 88 1755 101 345 222 235 634 600 1377 611 1661 101 246 19 18 43 48 23 22 23 29 17 20 Year completed 1985 1985 1986 1986 1986 1986 1986 1986 1986 1986 65 107 15 23 1986 1987 Hawthorne Market Market Mason O’Farrell Post Stevenson Battery First Montgomery Stevenson 4th Battery Daniel Burnham 85 97 94 90 106 70 103 80 136 100 61 133 123 62 20 23 24 28 22 20 28 25 27 24 15 39 34 18 1987 1987 1987 1987 1987 1987 1987 1988 1988 1988 1988 1989 1989 1989 Howard O’Farrell Spear Spear 2nd Pine Sansome California Post Folsom Pine 2nd 2nd 95 64 95 75 69 110 63 85 60 80 53 108 58 24 20 24 19   26 16 22 14 20 13 26 17 1989 1989 1989 1989 1990 1990 1990 1992 1993 1995 1997 1999 1999 Stories 181 388 3rd Beale 96 76 33 20 1999 1999 388 Beale 76 20 1999 150 California 101 24 2000 Lateral loadresisting system --Steel MF Steel moment frame Steel MF ----Dual-system steel CBF/EBF; steel MF -Dual-system steel CBF/EBF; steel MF ----Steel MF ------Steel MF Steel MF Dual-system RC wall/RC MF ---------Steel MF --Dual-system RC wall/RC MF -Dual-system RC wall/RC MF Dual-system RC wall/RC MF Dual-system steel CBF/EBF; steel MF Chapter O. Case Studies of Tall-Building Structural Analyses and Downtime and Loss Assessment for the HayWired Scenario Mainshock 365 Table 37.—Continued List Name number 197 199 Fremont Street Street number Street 199 Fremont Height, in meters 111 27 Year completed 2000 Stories 198 199 Hiram W Johnson State Building Courtyard San Francisco Downtown 455 299 Golden Gate 2nd 58 62 14 18 2000 2001 200 201 202 The Brannan II The Brannan I Gap Building 229 219 2 Brannan Brannan Folsom 65 65 84 18 18 14 2001 2001 2001 203 Four Seasons Hotel 757 or 735 Market 121 40 2001 204 205 55 Second Street Bridgeview 55 400 2nd Beale 101 87 25 26 2002 2002 206 207 208 209 Brannan III JPMorgan Chase Building The Paramount The Beacon West Brannan Mission Mission King 66 128 128 57 18 31 40 16 2002 2002 2002 2003 210 Avalon at Mission Bay 255 King 58 17 2003 211 212 213 214 355 333 125 848 1st 1st 3rd Kearny 81 66 148 59 26 21 42 15 2004 2004 2005 2005 215 216 The Metropolitan I [The Metropolitan] The Metropolitan II [The Metropolitan] St. Regis San Francisco International Hotel and St. Mary Catholic Center The Watermark Avalon at Mission Bay IIA 501 301 Beale King 73 58 22 17 2006 2006 217 218 219 San Francisco Federal Building SoMa Grand One Rincon Hill, South [One Rincon Hill] 1000 1146 to 1160 425 Mission Mission 1st 71 71 184 18 23 54 2007 2007 2008 220 221 222 223 224 Arterra Radiance I InterContinental San Francisco 555 Mission Street Argenta 320 325 868 555 1 Berry China Basin Howard Mission Polk 55 65 104 140 68 16 16 32 33 20 2008 2008 2008 2008 2008 225 226 The Infinity, Phase I [The Infinity] Millennium Tower [Millennium Tower] 300 301 Spear Mission 107 197 37 58 2008 2009 227 The Infinity, Phase II [The Infinity] 300 Spear 137 41 2009 228 Health Sciences West 513 Parnassus 64 16 NA 229 230 231 232 Health Sciences East Fillmore Center II 680 Folsom 350 Mission Street 513 1510 680 350 Parnassus Eddy Folsom Mission 64 55 52 168 16 18 13   NA NA NA NA 239 560 680 250 to 266 Lateral loadresisting system Dual-system steel CBF/EBF; steel MF -RC core walls, RC gravity RC MF RC MF Dual-system RC wall/RC MF Dual-system steel CBF/EBF; steel MF Steel MF Dual-system RC wall/RC MF RC MF -Others Dual-system RC wall/RC MF RC core walls, RC gravity -----RC core walls RC gravity --RC core walls with outrigger ---Steel MF RC core walls, RC gravity -Dual-system RC wall/RC MF RC core walls, RC gravity RC core walls, RC gravity --Steel MF -- The HayWired Earthquake Scenario—Engineering Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q [Also see https://doi.org/10.3133/sir20175013] Chapter P Fire Following the HayWired Scenario Mainshock By Charles Scawthorn1 Abstract Fire following earthquake is a significant problem in California. This chapter discusses potential losses arising from fires following the HayWired earthquake scenario, a hypothetical moment magnitude (Mw) 7.0 earthquake (mainshock) occurring on April 18, 2018, at 4:18 p.m., on the Hayward Fault in the east bay part of the San Francisco Bay area. The earthquake causes Modified Mercalli Intensities of VI–X in the greater San Francisco Bay region, with very strong shaking along the fault in the densely urbanized east bay. Weather conditions are typical for the season, with strong onshore winds in the afternoon, subsiding to calm in the evening. Fire following earthquake is a highly nonlinear process, modeling of which does not have great precision and is such that, in many cases, the only clear result is differentiation between situations of a few small fires versus major conflagration. For the Mw 7.0 scenario mainshock, it is estimated that approximately 668 ignitions will occur requiring the response of a fire engine. The first responding engine will not be able to adequately contain approximately 450 of these fires, such that in Alameda, Contra Costa, and Santa Clara Counties, dozens to hundreds of large fires are likely to merge into numerous conflagrations destroying tens of city blocks, with several of these potentially merging into one or several super conflagrations destroying hundreds of city blocks. Under the assumed scenario conditions, it is estimated that the about 450 large fires will result in an ultimate burned area of approximately 79 million square feet of residential and commercial building floor area, equivalent to more than 52,000 single SPA Risk LLC. 1 Usage of the term conflagration varies within the fire service (and interestingly, does not appear in the 1,449-page National Fire Protection Association’s Glossary of Terms, 2013 Edition; http://www.nfpa.org/~/media/ files/codes-and-standards/glossary-of-terms/glossary_of_terms_2013.pdf). It has previously been defined by the author (Scawthorn and others, 2005) as “. . . in the urban context, a conflagration usually denotes a large fire that spreads across one or more city streets.” 2 family dwellings. The fires following the scenario mainshock would be directly responsible for the loss of hundreds of lives, a total building replacement value of almost $16 billion, and property losses approaching $30 billion (2014 dollars). This loss is virtually fully insured and would be one of the largest single-loss events in the history of the insurance industry. Other economic impacts include the loss of perhaps $1 billion in local tax revenues. A number of opportunities exist for mitigating this problem, including greatly enhancing the postearthquake supply of water for firefighting and the mandatory use of automated gas shut-off valves, or seismic shut-off meters, in densely built areas. Introduction The HayWired earthquake scenario examines a hypothetical moment magnitude (Mw) 7.0 earthquake (mainshock) occurring on April 18, 2018, at 4:18 p.m., on the Hayward Fault in the east bay part of the San Francisco Bay area. This chapter discusses the potential for fire in the bay region after the mainshock. “Fire following earthquake” refers to a series of events or a stochastic process initiated by a large earthquake. Fires occur following all earthquakes that significantly shake a human settlement but are generally only a significant problem in large metropolitan areas predominantly composed of densely spaced wood buildings. In such circumstances, multiple simultaneous ignitions can lead to catastrophic conflagrations2 that may be the dominant agent of damage. Example regions vulnerable to such conflagrations include Japan, New Zealand, parts of Southeast Asia, and western North America. A large earthquake, such as a Mw 7.0 event on the Hayward Fault in the San Francisco Bay area (or comparable events in southern California, Washington’s Puget Sound region, or the lower mainland of British Columbia), combines all of the requisite factors for major conflagrations that, depending on circumstances, can be uniquely catastrophic, such as the fire following the Mw 7.8 Great 1906 San Francisco, California, earthquake. 368   The HayWired Earthquake Scenario—Engineering Implications Purpose The purpose of this chapter is to quantitatively describe fires following a hypothetical Mw 7.0 earthquake on the Hayward Fault, with primary emphasis for assisting emergency planning. The HayWired scenario occurs on Wednesday, April 18, 2018, at 4:18 p.m., with average April weather conditions. This analysis is intended to be realistic and not a “worst-case” scenario, and addresses the following questions: • What is a realistic scenario of ignitions, fire growth, and spread? • How will ignitions be reported after an earthquake, how will fire departments respond, and what factors will influence the spread of fires? What mutual-aid agreements are in place and how will they be activated? • How will damage to telecommunications, water supply, and roadways affect response? • What, if any, effective mitigation actions have been undertaken elsewhere that might be practical in the San Francisco Bay region in addition to those already taken? • What are the limitations of the fire-following-earthquake scenario and what research would provide a more realistic, perhaps more challenging or detailed, scenario? Background Large fires, measured in terms of square miles of burned area, have not been unique to fires following earthquakes— indeed the great fires of London (1666) and Chicago (1871) are only the most noteworthy of a long succession of nonearthquake related urban conflagrations. Large urban conflagrations were common in 19th century America, which allowed the National Board of Fire Underwriters (1905) to state the following with some confidence: In fact, San Francisco has violated all underwriting traditions and precedent by not burning up. That it has not done so is largely due to the vigilance of the fire department, which cannot be relied upon indefinitely to stave off the inevitable. Although the 1906 San Francisco earthquake had major geological effects and damaged many buildings, it was the ensuing fire that resulted in 80 percent of the total damage—a fire foreseen and expected, irrespective of an earthquake. As the fire service was professionalized in the 20th century— with improvements in equipment, communications, training, and organization—large urban conflagrations tended to become much less common (National Commission on Fire Prevention and Control, 1973). However, they were not entirely eliminated, as witnessed in the San Francisco Bay area in the 1991 East Bay Hills Fire, when 3,500 buildings were destroyed in a matter of hours. The two largest peacetime, urban conflagrations in history have been fires following earthquakes—1906 San Francisco (Mw 7.9) and 1923 Tokyo (Mw 7.9) earthquakes. In Tokyo, the fires caused the great majority of the 140,000 fatalities. Much larger wildland fires also occur and continue to be a major source of loss in places such as southern California almost every year. However, historical earthquakes have not caused major wildland fires. Although the combination of professionalized fire services, improved water supply, and better building practices has largely eliminated nonearthquake-related large urban conflagrations in the United States, fire following earthquakes is still a concern. This is owing to the correlated effects of a large earthquake simultaneously causing numerous ignitions, degrading building fire-resistive features, dropping pressure in water-supply mains, and overwhelming communications and transportation routes, thus allowing some fires to quickly grow into conflagrations that outstrip local resources. It is not sufficiently appreciated that the key to modern fire protection is a well-drilled, rapid response by professional firefighters in the early stages of structural fires, arriving in time to suppress a fire while that is still relatively feasible. For example, a typical response goal for urban fire departments is 4 minutes (from time of report to arrival) for a single ignition. If suppression is delayed, owing to either delayed response or lack of water, a single structural fire can quickly spread to neighboring buildings and grow to the point where an entire municipality’s fire resources are required, and perhaps even assistance from neighboring communities. This is for a single ignition. Most fire departments are not sized or equipped to cope with the fires following a major earthquake. A major earthquake and its associated fires is a low probability event for which, although having very high potential consequences, it may not be feasible to adequately prepare. There are exceptions to this; the Cities of San Francisco, Los Angeles, and Vallejo Fire Departments (California) and Vancouver (British Columbia) Fire and Rescue Services have all undertaken special measures, which are discussed below. Scenario Earthquake and Prevailing Conditions This section summarizes the seismological aspects and affected region for the HayWired scenario. The focus is primarily on the fire-related aspects of the scenario. Chapter P. Fire Following the HayWired Scenario Mainshock 369 Rupture Segment, Magnitude, and Intensity The HayWired scenario Mw 7.0 mainshock on the Hayward Fault affects the entire San Francisco Bay region (fig. 1). Seismological aspects of the scenario are discussed in Detweiler and Wein (2017). Peak ground acceleration (PGA) and Modified Mercalli Intensity (MMI) distributions were developed by Aagaard and others (2017) for this project and furnished for this report (fig. 2). Noteworthy are the high MMI (VIII–X) along the fault in the entire east bay. Affected Region Ten San Francisco Bay region counties affected by the scenario mainshock were analyzed for fire following earthquake. The region is densely urbanized (fig. 3), and the total affected population is approximately 7.7 million people (table 1; California Department of Finance, 2014), with population density as shown in figure 4. Exposure Building exposure data for the San Francisco Bay area was provided by the Federal Emergency Management Agency Table 1.  Counties and populations in the San Francisco Bay region, California, affected in the HayWired earthquake scenario. County Estimated population (2014)1 Alameda 1,573,254 Contra Costa 1,087,008 Marin 255,846 Napa 139,255 San Francisco 836,620 San Mateo 745,193 Santa Clara 1,868,558 Santa Cruz 271,595 Solano 424,233 Sonoma 490,486 Total 7,692,048 From California Department of Finance (2014). 1 based on Hazus-MH (Federal Emergency Management Agency, 2012) building inventory. There is a total building floor area of 5.77 billion square feet, with an estimated value (structure only) of approximately $1.15 trillion, distributed as shown in figure 5. Oakland SA N NF RAN CISCO BAY PACIFIC O CE AN I Not felt None II–III IV EXPLANATION V Weak Light Moderate None None Very light VI VII VIII IX X+ Strong Very strong Severe Extreme Violent Moderate/ Light Moderate Heavy Very heavy Heavy Base map from Google Earth; image Landsat/Copernicus; data LDEO-Columbia, NSF, NOAA; data SIO, NOAA, U.S., Navy, NGA, GEBCO, 2015. 0 0 10 5 20 10 30 KILOMETERS 15 MILES Figure 1.  Satellite image of the San Francisco Bay region, California, overlaid with a U.S. Geological Survey ShakeMap for the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario on the Hayward Fault (fault rupture shown by bright red line). (Mainshock data from Aagaard and others, 2017.) 370   The HayWired Earthquake Scenario—Engineering Implications 123° 122° 121° A Figure 2.  Maps of the San Francisco Bay region, California, showing (A) peak ground acceleration (PGA) and (B) instrumental intensity (approximately Modified 38° Mercalli Intensity) for the hypothetical magnitude-7.0 PACIFIC mainshock of the OCEAN HayWired earthquake scenario on the Hayward Fault. g, acceleration EXPLANATION due to gravity. SF, San Peak ground Francisco. (Mainshock acceleration (g) data from Aagaard and 0.1 others, 2017.) 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 37° Solano SF Sacramento San Francisco Bay San Mateo Santa Cruz 0 Area CALIF of map 123° 10 0 20 30 10 20 122° 50 KILOMETERS 40 30 MILES 121° B Solano Sacramento 38° PACIFIC OCEAN EXPLANATION 37° Instrumental intensity I II III IV V VI VII VIII IX X SF San Francisco Bay San Mateo Santa Cruz Area of map CALIF 0 0 10 20 10 30 20 40 50 KILOMETERS 30 MILES Chapter P. Fire Following the HayWired Scenario Mainshock 371 San Jose Oakland AY SAN FRANCISCO B N San Francisco PACIFIC OCEAN 0 Base map modified from Google Earth, 2015. 0 10 5 20 10 30 KILOMETERS 15 MILES Figure 3.  Satellite image of San Francisco Bay region, California. The region is densely urbanized, with a population approximately 7.7 million people (California Department of Finance, 2014). Fire Protection More than 500 fire stations were considered in the analysis of fire protection (fig. 6). In the most heavily impacted area there are a total of 229 fire engines potentially available immediately following the HayWired mainshock. Although many jurisdictions have seismically retrofitted fire stations (and other critical infrastructure), the functionality of a significant number of fire stations is still questionable (fig. 7). According to Bello and others (2006), an Earthquake Engineering Research Institute survey of these fire stations in 2006: . . . indicated that average peak ground acceleration are [sic] 0.5 g, and 52 percent of the stations are in areas mapped as moderate to very high liquefaction susceptibility with 102 stations being located within State designated Seismic Hazard Zone of Required Investigation for liquefaction or landsliding. More than 60 volunteers conducted walk-through field surveys of about 100 stations. In terms of life safety considerations, based on construction type, age and assessment of vulnerability, 42 percent of fire stations are in moderate to high-risk categories. In terms of functionality of the fire stations, based on a subset (293 stations) for which information was available, 67 percent were of moderate to high risk of not functioning after an earthquake. Based on these results, it is recommended that those fire stations at higher risk be evaluated and retrofitted such that life safety and vulnerability are improved before the next large earthquake occurs. Each fire station in the affected region was allocated an immediate area using a Voronoi diagram3 as an approximation 3 For this analysis, the Voronoi diagram was a partition of the region into polygons, each side of which was a line equidistant from the nearest two fire stations. (For an explanation of Voronoi diagrams see https://en.wikipedia.org/ wiki/Voronoi_diagram.) of the station’s primary response area (fig. 8). The subsequent analysis is based on these primary response areas. Time of Day Time of day is relevant in that more human activity occurs during waking hours, typically resulting in higher ignition rates at those times. The HayWired mainshock is specified as occurring Wednesday, April 18, 2018, at 4:18 p.m. However, the specific time of occurrence is not considered in this analysis. Wind and Humidity Weather can affect fire growth and spread, as well as the direction and distance at which communities are affected by hazardous material release. Important meteorological parameters include windspeed, wind direction, temperature, rain, and humidity. For purposes of estimating fire effects in the HayWired scenario, average April conditions were assumed to apply, based on data for the period 1974–2012 (WeatherSpark, 2014). Average conditions for the three San Francisco Bay area international airports are shown in table 2 and figure 9. In the case of precipitation, the most common condition is reported (for example, no rain), along with the probability of precipitation at some point in the day and the most common form of precipitation when it does rain. In the case of wind direction, the most common direction is tabulated. Humidity is reported as average daily low and high. In April, wind conditions are typically created by a trough of low pressure east of the bay area, which draws in strong, westerly, cooler and more humid air from the ocean in the afternoon, subsiding to more calm conditions in evening. An example of this is shown in figure 10 for April 18, 2012, in which major streaklines are shown at 4 p.m. and 5 p.m., with much shorter streaklines at 9 p.m. Cumulative distribution functions for windspeeds for 4 p.m., 372   The HayWired Earthquake Scenario—Engineering Implications 122° 123° A I PAC FIC OC 38° N EA Figure 4.  Maps of the San Francisco Bay region, California, showing population density for (A) the 10-county region and (B) the area adjacent to the fault rupture for the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario. The length of the Hayward Fault ruptured in the scenario is shown on the maps. persons/ km2, persons per square kilometer. San Francisco 0–1 Oakland Y BA 1–10 O SC CI AN FR SAN EXPLANATION Population density (persons/km2) 10–1,000 1,000–10,000 San Jose >10,000 HayWired rupture 0 37° 122.75° 38° 10 0 20 10 30 Area of map CALIF 40 KILOMETERS 20 MILES 122.5° 122.25° B C IFI PAC N EA OC Oakland San Francisco SAN EXPLANATION CO CIS AN FR Population density (persons/km2) 0–1 1–10 Y BA 10–1,000 1,000–10,000 >10,000 HayWired rupture 37.5° 0 0 Area of map 10 5 CALIF 20 KILOMETERS 10 MILES 121° Chapter P. Fire Following the HayWired Scenario Mainshock 373 122.5° 122° Figure 5.  Map showing building density of San Francisco Bay region, California. The length of the Hayward Fault ruptured in the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario is shown on the map. Total floor area ratio, total building floor area in a census block divided by the area of the census block. 38° IFIC PAC San Francisco Oakland SAN AN O SC CI AN FR E OC EXPLANATION Y BA 37° Total floor area ratio <0.5 0.5–1 1–2 2–7 7–20 20–50 >50 HayWired rupture San Jose Area CALIF of map 0 10 0 5 122.5° 38° 20 KILOMETERS 10 MILES 122° PACIFIC OCEAN EXPLANATION O Y BA 5 SC 0 10 I NC 0 A FR 37.5° SA N PGA (g) 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Fire station HayWired rupture Area CALIF of map 20 KILOMETERS 10 MILES Figure 6.  Map of San Francisco Bay area, California, fire stations overlaid on peak ground acceleration (PGA) for the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario. The length of the Hayward Fault ruptured in scenario is shown on the map. g, acceleration due to gravity. (Mainshock data from Aagaard and others, 2017.) 374   The HayWired Earthquake Scenario—Engineering Implications Figure 7. Three-dimensional graph showing the number of fire stations in San Francisco Bay area, California, counties that are at low, moderate, and high risk of earthquake damage in the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario (data from Bello and others, 2006). 40 35 30 25 20 Solano (13) Santa Clara (64) San Mateo (32) 15 10 San Francisco (38) Napa (7) 5 Marin (18) Contra Costa (48) 0 Low Moderate 122.5° Alameda (66) High 122° 38° C FI CI PA EA OC N SAN O SC CI AN FR EXPLANATION Y BA Fire stations HayWired rupture 37.5° Area CALIF of map 0 0 10 5 20 KILOMETERS 10 MILES Figure 8.  Map of San Francisco Bay area, California, fire stations with associated Voronoi areas. The primary response area for each fire station was approximated by a Voronoi diagram. The length of the Hayward Fault ruptured in the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario is shown on the map. Chapter P. Fire Following the HayWired Scenario Mainshock 375 A B Figure 9.  Maps and charts of average temperatures on April 18 at (A) San Francisco International Airport and (B) Hayward Executive Airport in the San Francisco Bay area, California. Average windspeeds are noted. (Images from WeatherSpark, 2014; http://www.weatherspark.com, used with permission.) N N N EXPLANATION 0 Windspeed, in nautical miles per hour 0–4.9 15–19.9 20–24.9 5–9.9 10–14.9 +25 0 Area of map 10 20 10 30 20 40 50 KILOMETERS 30 MILES CALIF Figure 10.  Maps showing wind streaklines (arrows) for April 18, 2012 (4/18/2012), at 4 p.m. (left), 5 p.m. (center), and 9 p.m. (right), typical of April wind conditions in the San Francisco Bay region, California. Strong westerly winds in late afternoon subside in evening. Windspeed was measured 10 meters above surface elevation. PST, Pacific Standard Time. (Images from San Jose State University, 2014.) 376   The HayWired Earthquake Scenario—Engineering Implications Table 2.  Average wind conditions in April at San Francisco Bay area, California, major airports. [Weather data from WeatherSpark (2014); mi/hr, miles per hour; °F, degrees Fahrenheit; NW, northwest; W, west] Airport (city/identifier) Windspeed (mi/hr) Direction Temperature (°F) 7 12 10 NW W W 50–65 50–65 50–65 San Jose (SJC) San Francisco (SFO) Oakland (OAK) 5 p.m., and 9 p.m. for the years 2000–2012 are shown in figure 11, and indicate significant variability of the stronger afternoon winds, with consistently calmer conditions later in the evening. However, a reverse of the typical summertime weather pattern can occur, consisting of occasional intense katabatic winds, locally sometimes termed “Diablo winds.” These are hot, dry, offshore winds from the northeast that sometimes occur in the San Francisco Bay region during the spring and fall. These winds differ from the more familiar Southern California Santa Ana winds, and are created by the combination of strong inland high pressure at the surface, strongly sinking air aloft, and lower pressure off the California coast. The air descending from aloft, as well as from the Coast Ranges, compresses at sea level, where it warms as much as 20 degrees Fahrenheit (°F) (11 degrees Celsius, °C), and loses humidity. If the pressure gradient is large enough, the dry offshore wind can become quite strong with gusts reaching speeds of 40 miles per hour (64 kilometers per hour) or higher, particularly along and in the lee of the ridges of the Coast Ranges, where warm, dry surface air from the windward eastern side is drawn up and over the ridgelines (fig. 12). Such winds were major factors in the 1923 Berkeley and 1991 East Bay Hills Fires (discussed below). This effect is especially significant as it can enhance the updraft generated by large wildland or urban fires. The pattern of windspeeds and direction used for the scenario was the more typical westerly wind subsiding in the evening, rather than the more dangerous Diablo-wind scenario. Light rain (percent chance) 19 28 22 Percent humidity 42–93 52–88 56–92 Experience with Large Fires in the San Francisco Bay Region The Great 1906 San Francisco earthquake and fire is the archetypical fire following earthquake event. It is so familiar and well documented that we will not spend much time on it here. Simply put, it was the largest peacetime urban fire in history at the time, only exceeded since by the 1923 Tokyo earthquake and fire. The 1906 earthquake and resulting fires caused an estimated 3,000 Windspeed cumulative distribution function 1.0 0.9 0.8 0.7 N 0.6 EXPLANATION 0.5 4 p.m. 5 p.m. 9 p.m. 0.4 0.3 Windspeed, in nautical miles per hour 0–4.9 15–19.9 20–24.9 5–9.9 10–14.9 +25 0.2 0.1 0 0 10 20 30 40 0 EXPLANATION 50 0 Area of map 10 20 10 30 20 40 50 KILOMETERS 30 MILES CALIF 60 Windspeed, in miles per hour Figure 11.  Graph showing cumulative distribution function of windspeeds in the central San Francisco Bay region, California, for 4 p.m., 5 p.m., and 9 p.m. for the years 2000–2012. (Data from San Jose State University, 2014.) Figure 12.  Map showing wind streaklines (arrows) for September 13, 2003 (9/13/2003), at 10 a.m., typical of Diablo wind conditions in the San Francisco Bay region, California. Windspeed was measured at 10 meters above surface elevation. PST, Pacific Standard Time. (Images from San Jose State University, 2014.) Chapter P. Fire Following the HayWired Scenario Mainshock 377 deaths and $524 million (1906 dollars) in property loss. Fires that ignited in San Francisco soon after the onset of the earthquake burned for 3 days because of the lack of water to control them. The damage in San Francisco was devastating and 28,000 buildings were destroyed, although 80 percent of the damage was caused by the fire rather than the shaking (fig. 13). Fires also intensified the losses in 1906 at Fort Bragg and Santa Rosa (see Scawthorn and O’Rourke, 1989; Scawthorn and others, 2005; Scawthorn and others, 2006). Beyond the 1906 San Francisco earthquake and fire, the San Francisco Bay region has a long history of conflagrations (fig. 14), owing in large part to the Diablo winds discussed above. According to the Hills Emergency Forum (2005), over the period from 1923 to 1991 the east bay has averaged a 585-acre fire every 5 years, destroying on average 266 homes (2005). However, most of these building losses occurred in only two of these fires—the 1923 and 1991 events (fig. 15, table 3). It should also be noted that almost all of these fires occurred in autumn (which is typically the region’s greatest fire-risk season), in contrast to the scenario being considered here. Three recent fires in the San Francisco Bay region are worthy of mention: • On September 9, 2010, a buried, high-pressure, 30-inch steel natural-gas pipeline exploded in a residential neighborhood in San Bruno, California, near San Francisco. The explosion and ensuing fire killed 8 people and injured 58. It destroyed 38 homes and damaged an additional 70. During the first 50 hours following the incident, more than 500 firefighters and 90 firefighting apparatus responded, involving 42 fire agencies. The total cost of the disaster was estimated to be approximately $1.6 billion (Davidson and others, 2012). • The Mission Bay fire was a five-alarm fire that occurred shortly before 5 p.m. on March 11, 2014, in the Mission Bay neighborhood of San Francisco, California. The conflagration appeared to completely destroy block 5, a 172-unit building, part of Mega Blocks 360, a $227 million apartment complex being developed by San Francisco-based BRE Properties, Inc., at China Basin Street and Fourth Street (San Francisco Chronicle, 2014). The San Francisco Fire Department needed a large amount of resources to combat the fire, including the city’s auxiliary water-supply system. • On the night of October 8, 2017, Diablo winds started and drove widespread wildfires in the northern San Francisco Bay region counties of Napa, Sonoma, and Solano. The fires killed at least 43 people, destroyed 8,900 homes and other structures, and burned 164,000 acres. More than 10,000 firefighters responded to the fires (Wikipedia, 2017). (Note that because of its recency, this fire was not considered further in this chapter.) Figure 13.  Map of fires (orange and yellow diamonds) caused by the moment-magnitude-7.8 1906 San Francisco, California, earthquake and area burned in the great conflagration that followed (orange). Ignition data from Scawthorn and O’Rourke (1989) and Scawthorn and others (2005). Area of map CALIF 1 0 0 0.5 2 KILOMETERS 1 MILE N 378   The HayWired Earthquake Scenario—Engineering Implications Table 3.  List of some large, historical fires driven by Diablo winds in the east bay part of the San Francisco Bay area, California (data from Hills Emergency Forum, 2005; California Governor’s Office of Emergency Services, 2013; Routley, [n.d.]; National Board of Fire Underwriters, 1923. [--, no data] Month/year Fire name/location Deaths Structures destroyed Acres burned September 1923 November 1933 September 1946 City of Berkeley Joaquin Miller (Redwood Road) Buckingham Boulevard/Norfolk Road Leona Hillside Oakland Hills East Bay Hills 0 1 0 584 20 homes 0 130 1,000 1,000 0 0 25 2 homes 37 homes and 21 damaged 3,354 homes and 456 apartments 1,200 204 1,600 N SA San Pablo Reservoir RD . AK PE LY D. IZZ LV GR B 1923 RSITY UNIVE Y ASHB 80 Briones Reservoir AM OD BL PA 1961 1980 1905 24 1946 1970 Lafayette Reservoir 1937 1991 1955 1929 24 BRO AD WA Y October 1960 September 1970 October 1991 RE 1940 DW 13 1933 880 OO D RD . San Leandro Reservoir 1961 1991 580 E IN YL SK SAN FRANCISCO BAY . VD BL 1931 1968 1960 Lake Chabot CA MA 880 UR RTH Area of map D. BLV N CALIF 0 0 1 2 1 3 4 2 5 KILOMETERS 3 MILES Figure 14.  Map of fires in the east bay part of the San Francisco Bay area, California, from 1923 to 1991. Note that colors are only used to differentiate among areas burned by fires. (Modified from Hills Emergency Forum, 2005.) Estimated damage, in Ignition cause billions of U.S. dollars -Smoker -Smoker -Arson and rekindle -Unknown -Arson 1.5 Rekindle Chapter P. Fire Following the HayWired Scenario Mainshock 379 122.3° 122.25° Area of map 37.89° 1923 Berkeley Fire ALAMEDA COUNTY CALIF 37.87° 1991 East Bay Hills Fire 37.85° 37.83° 0 0 1 0.5 2 KILOMETERS 1 MILE Figure 15.  Map of final burned areas (dark orange) for 1923 Berkeley and 1991 East Bay Hills Fires in the east bay part of the San Francisco Bay area, California. Fire Following Earthquake Aspects of the Scenario This section presents the analysis underlying the estimation of fires and losses likely to result from the HayWired scenario mainshock. The section discusses modeling of fire following earthquake, ignitions, initial response, fire spread, and performance of lifelines (for example, utilities and transportation). Modeling of Fire Following Earthquake A full probabilistic methodology for analysis of fire following earthquake was developed in the late 1970s (Scawthorn and others, 1981) and has been applied to major cities in western North America (Scawthorn and Khater, 1992). Scawthorn and others (2005) summarizes modeling for fire following earthquake, so only a brief review is presented here. In summary, the steps in the process of fire following earthquake are shown in figure 16: • Occurrence of the earthquake—causing damage to buildings and contents, even if the damage is as simple as objects (such as candles or lamps) falling over. • Ignition—whether a structure has been damaged or not, ignitions can occur as a result of earthquakes. The sources of ignitions are numerous, including overturned heat sources, abraded and shorted electrical wiring, spilled chemicals having exothermic reactions, and friction from objects rubbing together. • Discovery—at some point, the fire resulting from the ignition will be discovered, if it has not self-extinguished (this aspect is discussed in more detail below). In the confusion following an earthquake, the discovery may take longer than it might otherwise. • Report—if it is not possible for people discovering a fire to immediately extinguish it, fire department response will be required. For a fire department to respond, a report has to be made to the fire department. Communications-system malfunction and congestion may delay many reports. • Response—a fire department then has to respond but may be delayed by responding to nonfire emergencies (for example, building collapse) and by transportation disruptions. • Suppression—a fire department then has to suppress the fire. If the fire department is successful, they move on to the next incident. If not successful, they continue to attempt to control the fire, but it can spread and become a conflagration. Success or failure hinges on numerous factors, including the functionality of the water-supply system, building construction and density, and weather conditions such as wind and 380   The HayWired Earthquake Scenario—Engineering Implications humidity. If the fire department is unable to contain the fire, the process ends when the fuel is exhausted or when the fire reaches a firebreak. Earthquake Structural/nonstructural damage Ignition Discovery Report Fire department response Communication infrastructure damage/saturation Transportation impediments Water supply damage/malfunction Stop Ignitions Suppression Yes No Spread/conflagration This fire-following-earthquake process is also shown in figure 17, which is a fire department operations time line. A rapid response is essential to reduce losses from fire following earthquake. Fire following earthquake is not a linear process, and modeling it is not very precise—in many cases, only a few small fires versus a major conflagration can be distinguished. Building construction and density, wind, humidity, vegetation... Firebreak/fuel exhaustion Postearthquake ignition rates in the United States have been studied by a number of investigators (Lee and others, 2008) with the most recent and relevant algorithms for estimating postearthquake ignition rates being developed by Davidson (2009a,b) and SPA Risk LLC (2009), both of which are considered here. Davidson (2009b) conducted an exhaustive selection to evaluate 48 potential covariates, of which model A.NB2 is: Stop Figure 16.  Flow chart of the fire-following-earthquake process (from Scawthorn and others, 2005). ln(µ ) = β 0 + βii II + βtbldg ln(tbldg)+ β%CIT x%CIT + β dens xdens , (1) where μ II is ignitions per census tract, is the instrumental intensity of the earthquake,4 4 Modified Mercalli Intensity (MMI) and Instrumental Intensity (II) are used synonymously here. EXPLANATION D Fire discovery Fire growth—Each stacked line indicates one additional engine needed R Fire report Fire response 7 Ignition Suppression Figure 17.  Chart of fire department (FD) operations timeline when responding to fires following an earthquake. Horizontal axis is time, beginning at time of earthquake. Horizontal bars depict development of fires, from ignition through growth or increasing size (size is indicated by width or number of horizontal bars). (From Scawthorn and others, 2005.) Chapter P. Fire Following the HayWired Scenario Mainshock 381 x%CIT tbldg x%URM xdens is the percent of land area that is commercial, industrial, or transportation, is the total building area in thousands of square meters, is the percent of building area that is unreinforced masonry (URM), is the population density (persons per square kilometer), and parameter (β) estimates are β0= –15.42, βii=1.13, β%CIT= –32.48, βtbldg=0.85, β%URM=27.72, and βdens=0.0000453. The SPA Risk LLC (2009) relation used the Davidson (2009b) dataset restricted to census tracts that (1) fell within jurisdictions for which ignition data was available, (2) experienced MMI≥VI, and (3) had population densities greater than 3,000 per square kilometer. Using this approach, relatively simple regressions to model postearthquake ignitions were developed: Ignitions per million square feet of building floor  area= –0.029444PGA+0.581895PGA2, (2) Ignitions per million square feet of building floor area=1.0449–0.338MMI+0.0277MMI 2, (3) where PGA is the peak ground acceleration of the earthquake, relative to the acceleration due to gravity at the Earth’s surface (g). Of the two ignition regressions, the one in equation 1 requires more data, some of which may not be readily available (for example, percentage of URM building).5 A comparison of the two ignition models is shown in figure 18A, where equation 1 is plotted using median values (standard deviations in parentheses) for tbldg=244.7 (164), x%CIT=0.027 (0.016), x%URM = 0.013 (0.01), and xdens=3,445 (4,048) as provided in Davidson (2009b), and equation 3 (SPA Risk LLC, 2009) is plotted in black using 2.6 million square feet of building floor area per census tract. Dotted lines in the figure are equation 1 plus and minus one standard deviation (determined by way of numerical simulation). Figure 18B and C are similar, except that the variable x%CIT in Davidson’s (2009b) equation, representing the percentage of land area employed for commercial, industrial, and transportation (CIT) purposes, is varied by plus and minus one sigma (sigma of x%CIT), with equation 3 remaining the same in all plots. In figure 18A, it can be seen that the median SPA Risk LLC (2009) model is higher than the Davidson (2009b) model, by a factor of 2.8 at MMI VI and 2.3 at MMI VIII, while actually being lower (0.93) at MMI X. In figure 18B, 5 Davidson (2009b) used default data from Hazus-MH MR2 (Federal Emergency Management Agency, 2003) for building floor area and unreinforced masonry (URM) estimates. An issue exists with the use of “URM” default data since most URM buildings in California have been retrofitted, so whether such buildings are now unreinforced is unclear. Ding and others (2008) have examined the Hazus-MH MR2 building inventory data (in general, in the context of flood) and found it to have significant inaccuracies. That being said, at the regional level the database can be useful, and Davidson’s use of it was innovative. corresponding to lower CIT land use (more representative of residential areas), the two models are in closer agreement, whereas C (representative of higher CIT uses) shows a somewhat greater difference of the two models. Equation 2 was used to estimate the total number of ignitions for the HayWired mainshock, resulting in a mean estimate of 668 ignitions, as shown in figure 19 and table 4. Ninety percent of the ignitions are confined to three counties— Alameda, Contra Costa and Santa Clara—with Alameda County alone having 53 percent of all ignitions. These are only ignitions that require fire department response; there may be other, typically minor, ignitions that are suppressed immediately by citizens, which are often not reported. Of the approximately 668 total ignitions, it is estimated 453 of these will grow to be large fires (defined as fires exceeding the capacity of the first arriving engine). The cause of these ignitions will likely be similar to causes following the 1994 Mw 6.7 Northridge, California, earthquake, which is the best U.S. dataset for fire following a recent earthquake; about half of all ignitions would be electrical, a quarter gas related, and the remainder owing to a variety of causes, including chemical reactions (table 5). Also, on the basis of the Northridge experience, nearly half of all ignitions would typically occur in single-family residential dwellings, with another 26 percent in multifamily residential dwellings—that is, about 70 percent of all ignitions occur in residential dwellings (Scawthorn and others, 1998). Ignitions in educational facilities would be a small percentage of the total (3 percent in Northridge), and most of these would be a result of the exothermic reactions of spilled chemicals in chemistry laboratories. A particular concern is the large number of oil refineries, tank farms, and related facilities in the northern bay area. These facilities refine one-third of the gasoline used west of the Rocky Mountains. When strongly shaken, oil refineries and tank farms have typically had large fires, which have burned for days. Examples include the Showa Refinery fire following the Mw 7.6 1964 Niigata, Japan, earthquake (Kawasumi, 1968), the Tüpraçs Refinery fire following the Mw 7.6 1999 İzmit, Turkey, earthquake (Scawthorn, 2000), and the Idemitsukosan Hokkaido Refinery fire following the Mw 8.3 2003 Tokachi-Oki, Japan, earthquake (Scawthorn and others, 2005). Initial Response This section discusses the initial response to ignitions following an earthquake. Reporting of fires is particularly crucial, yet problematic, following an earthquake. Citizen Response Initially, citizens will respond to the approximately 668 ignitions requiring fire-department response in the HayWired scenario. When they realize suppressing the fires is beyond their capabilities, they will attempt to contact emergency 382   The HayWired Earthquake Scenario—Engineering Implications 10.00 A EXPLANATION Equation 1 median value medianX%CIT = 0.027 1.00 Equation 1 median plus standard deviation 0.10 Equation 1 median minus standard deviation Ignition rate per census tract 0.01 Equation 3 0 10.00 B X%CIT−1X%CITsigma = 0.014 1.00 0.10 0.01 0 10.00 C X%CIT−1X%CITsigma = 0.0395 1.00 0.10 0.01 0 V VI VII IX VIII X Modified Mercalli Intensity Figure 18.  Graphs comparing two regressions (equations 1 and 3, see text) used to model postearthquake ignitions per census tract. A, Graph of equation 1 (Davidson 2009b, A.NB2) plotted in red using median values (standard deviations in parentheses) for tbldg=244.7 (164), x%CIT=0.027 (0.016), x%URM=0.013 (0.01), and xdens=3445 (4048) as provided in Davidson (2009b), and equation 3 (SPA Risk LLC, 2009) is plotted in black using 2.6 million square feet of building floor area per census tract. Dashed lines in the graphs are equation 1 plus and minus one standard deviation (determined by way of numerical simulation). Equation 3 (SPA Risk LLC, 2009) is plotted in black. B and C are similar, except that the variable x%CIT in Davidson’s (2009b) equation, representing the percentage of land area used for commercial, industrial and transportation (CIT) purposes, is varied by plus and minus one sigma (sigma of x%CIT), with equation 3 remaining the same in all plots. It can be seen that the median SPA Risk LLC (2009) model is higher than the Davidson (2009b) model by a factor of 2.8 at Modified Mercalli Intensity (MMI) VI and 2.3 at MMI VIII, while actually being lower (0.93) at MMI X. In B, corresponding to lower land use for commercial, industrial, or transportation purposes (CIT) (more representative of residential areas), the two models are in closer agreement, whereas C (representative of higher CIT uses) shows a somewhat greater difference in the two models. Table 4.  Estimated ignitions and damage from the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario on April 18, 2018, at 4:18 p.m. (breezy conditions and moderate humidity). [--, no data; TFA, total floor area] Exposed building TFA Ignitions Large fires Conflagrations (multiblock fires) Final burned TFA, in millions of square feet Alameda 1,853 352 279 198 49 $9,710 4 53 Contra Costa 1,480 123 60 43 10 $2,103 1 18 342 23 14 10 2 $500 1.1 4 County Marin Napa Final burned loss, in millions of 2014 U.S. dollars ($) Percent burned Percent of total losses 90 27 19 13 3 $651 5.3 4 San Francisco 817 21 5 4 1 $177 0 3 San Mateo 576 19 15 11 3 $519 1 3 Santa Clara 1,610 83 56 40 10 $1,940 Santa Cruz 96 1 -- -- -- 338 12 Solano Sonoma Total 4 3 1 1 12 -- 0.01 0 $142 0.4 2 38 7 0 0 0 $13 0.3 1 7,241 668 453 321 79 $15,755 2 100 Chapter P. Fire Following the HayWired Scenario Mainshock 383 services by telephone, because street fire-alarm pull boxes have largely disappeared from the U.S. urban landscape. Attempts to report fires by calling 9-1-1 will likely be unsuccessful, owing to congestion of the system and overwhelmed 9-1-1 dispatch centers. Citizens may then go in person to the nearest fire station, but such “still alarms” will largely be futile because the fire companies will have already responded (self-dispatched) to the nearest fire, if not dispatched by 9-1-1. Experience shows that citizens on scene will respond rationally (Van Anne and Scawthorn, 1994), rescuing as many people as possible and protecting neighboring buildings (exposures). Water supply from mains (discussed below) will often be unavailable. Table 5.  General sources of ignition after the momentmagnitude-6.7 1994 Northridge, California, earthquake. (ignition data from Scawthorn and others, 1998). Source of ignition Percentage of ignitions Electrical 56 Gas-related 26 Other 18 123° 122° San Pablo Bay 38° C PA IF C I OC EA N San co cis an Fr y Ba EXPLANATION Ignitions 37.5° <0.1 0.1–0.5 0.5–1.0 1.0–2.0 2.0–5.0 >5.0 HayWired rupture Area of map CALIF Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. 0 0 10 20 10 30 40 KILOMETERS 20 MILES Figure 19.  Map of San Francisco Bay region, California, showing estimated number of ignitions within fire station primary response areas (see fig. 8) following the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario. Green indicates a small likelihood of ignition and dark red indicates five or more ignitions per area. The length of the Hayward Fault ruptured in the scenario is shown on the map. 384   The HayWired Earthquake Scenario—Engineering Implications Reporting As noted above, 9-1-1 dispatch centers will be overwhelmed and doing as much as possible to triage events and dispatch resources after the HayWired mainshock. Reports of fires during the initial period will be haphazard. Most fire departments do not have their own helicopters, and reporting by television news helicopters will be a valuable resource for a few major incidents, but not most. The first knowledge the San Francisco Emergency Operations Center had of the Marina fire following the 1989 Mw 6.9 Loma Prieta earthquake was from television news reports (despite several fire companies having responded). Quickly gaining an accurate and complete awareness of fires following an earthquake remains a challenge. Fire Department Initial Response The initial response of fire companies and personnel in the region of the HayWired scenario will be to protect themselves during violent shaking, and as soon as possible, open fire-station doors and remove firefighting apparatus (such as pumpers and ladder trucks). Different fire departments have somewhat varying earthquake procedures, but in general companies will remove firefighting apparatus to a predesignated location (often simply in front of the fire station), check the station for damage, and perform a radio check. By this time, typically within 5 minutes, they will either have self-dispatched to an observed smoke column, responded to a citizen still alarm, or been instructed to mobilize with other fire companies into a strike team. Local fire department resources will be completely committed, and in need of assistance from outside of the San Francisco Bay region. The primary needs will be personnel, additional hose, hard suction hose (that is, hose that does not collapse when used to draft water from a source that is not already under pressure), firefighting foam, light equipment (gloves, hand tools, self-contained breathing apparatus [SCBA]), and heavy equipment (cranes, bulldozers, backhoes). Additional fire apparatus (pumpers and ladder trucks) will not be the primary need, but will still prove useful as extraregional strike teams arrive. In the initial stage, personnel needs may be significantly supplemented by the Community Emergency Response Team (CERT) program, but will be more significantly strengthened by the recall of off-duty, trained firefighters. Off-duty personnel can be expected to have doubled staffing within 3–6 hours after the HayWired mainshock, and tripled it within 12–24 hours. How these personnel join their fire companies will be an issue, and there will be some inefficiencies as personnel join first available companies. Nevertheless, arrival of off-duty personnel will be very important to relieve on-duty personnel nearing physical limits. Fire Spread This analysis assumes that after the HayWired mainshock all fire-service resources will initially focus on firefighting, leaving search and rescue, hazmat response, and other emergencies until fires are brought under control. The initial 668 ignitions will not all develop into large fires. Nevertheless, the normal 4-minute structural-fire response time will most likely be delayed. This delayed response, owing primarily to delayed reporting and dispatch, will result in many fires having grown such that a multiengine capacity is needed on arrival. Especially in low humidity conditions, an ignition that has not been suppressed can become a room-sized fire within several minutes and grow into a fully involved, single-family structural fire within several more minutes. To protect neighboring buildings, typically two or more companies are needed. If only one fire company is available, it is possible but unlikely that it might be able to protect two exposures using a monitor (water cannon) and hand line (fire hose) with civilian assistance. In fire following earthquake modeling, fires that have grown to exceed one engine company’s capabilities are termed large fires. The number of large fires for the HayWired mainshock is estimated based on several rules, including (1) availability of water for firefighting within each fire-response area and (2) ratio of ignitions to fire engines within each county (the latter to account for limited mutual aid), resulting in an estimate of 453 large fires (table 3). The large number of ignitions developing into large fires is a result of the high earthquake shaking intensities in the east bay combined with fuel provided by the high-density of wood construction between San Francisco Bay and hills to the east (East Bay Hills). Lifelines The performance of lifelines, such as water supply, gas, electric power, communications, and transportation, is integral to the firefighting process during fire following earthquake. A detailed discussion of lifeline performance for this scenario is beyond the scope of this report, which only briefly discusses selected lifelines with regard to fire following earthquake. Water Supply Water supply would be severely impacted by an earthquake like the HayWired scenario mainshock (see Porter, Water Supply, this volume). A significant part of the San Francisco Bay area’s water derives from the Sierra Nevada and is conveyed by several major canals and aqueducts, particularly the Mokelumne and Hetch Hetchy Aqueducts (fig. 20). In the last few decades, earthquake hazards mitigation has been largely focused on assuring delivery of water from these distant sources to the bay area. Major seismic retrofit programs have been completed by the East Bay Municipal Utilities District (EBMUD), Contra Costa Water District, and Marin Municipal Water District and are ongoing for the Santa Clara Valley Water District and the Hetch Hetchy system, which is owned by the City of San Francisco and serves that city as well as much of the west and south bay area (fig. 21). These retrofit programs have focused on the dams, tanks, and major transmission lines; however, most of these Chapter P. Fire Following the HayWired Scenario Mainshock 385 water operators have found that significant upgrading of their extensive water distribution systems is beyond available resources. As a result, extensive portions of the water distribution systems are very vulnerable and likely to sustain a number of breaks in a large earthquake. The following was noted in a recent study by the Association of Bay Area Governments (2010): . . . 68.1% of critical water system facilities . . . are exposed to extremely high shaking levels (peak ground accelerations, PGA, of greater than 60% g with a 10% chance of being exceeded in the next 50 years) . . . 95.2% of pipelines are estimated to be exposed to high shaking levels (PGA >40% g), and 62.8% are exposed to extremely high shaking levels (PGA >60% g) . . . [the Association of Bay Area Governments] has estimated that there could be, for example, 6,000–10,000 water pipeline breaks or major leaks in an earthquake on the Hayward fault (compared to 507 in the Loma Prieta earthquake) . . . Pipe breaks in the 1989 Loma Prieta earthquake are shown in figure 22. Owing to their proximity to the Hayward Fault, east bay water distributions are particularly vulnerable (East Bay Municipal Utility District, 2011): . . . earthquake hazard information . . . with more detailed information on materials and design of these facilities, and pipeline materials and connections associated with EBMUD, were used to estimate the problems associated with District facilities in a 1994 study. At that time, EBMUD estimated that, should an earthquake occur on the Hayward fault EBMUD customers could have expected: 0 0 Area of map 20 10 40 20 60 KILOMETERS 30 MILES CALIF Figure 20.  East Bay Municipal Utility District (EBMUD) map showing major water-supply systems of the San Francisco Bay area, California. A significant part of the bay area’s water comes from reservoirs in the Sierra Nevada. The Mokelumne Aqueduct supplies much of the water to the EBMUD service area, and the San Francisco Public Utilities Commission’s (SFPUC) Hetch Hetchy system primarily conveys water to San Francisco and the west and south bay. (From East Bay Municipal Utility District, 2017). 386   The HayWired Earthquake Scenario—Engineering Implications 122.5° 122° 121.5° 38° PACIFIC OCEAN Z7 37° Area of map CALIF 0 0 10 5 20 10 30 15 40 KILOMETERS 20 MILES EXPLANATION Major water districts Alameda County Water District (ACWD) Contra Costa Water District (CCWD) East Bay Municipal Utility District (EBMUD) Marin Municipal Water District MMWD) Napa Water Division (NWD) San Jose Water Company (SJWC) Zone 7 (Z7) HayWired rupture Water supply reservoirs (acre-feet) <10,000 10,000–50,000 San Francisco/Bay Area Water Supply and Conservation District (SF/BAWSCA) 50,000–100,000 Santa Clara Valley Water District (SCVWD) >100,000 Figure 21.  Map of major water districts and water-supply reservoirs in the San Francisco Bay region, California. The length of the Hayward Fault ruptured in the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario is shown on the map. (Water district and reservoir data from Bay Area Water Supply and Conservation Agency, [n.d.]; California Department of Water Resources, [n.d.]; and Datahub, [n.d.]) Chapter P. Fire Following the HayWired Scenario Mainshock 387 122° 123° 38° MARIN PA CIF CONTRA COSTA IC N EA SAN OC SAN FRANCISCO FR AN CI O SC ALAMEDA BA Y EXPLANATION PGA (g) 0.1 SAN MATEO 0.2 0.3 0.4 0.5 0.6 0.7 SANTA CLARA 0.8 0.9 SANTA CRUZ 1.0 Water-pipe breaks 37° 0 0 20 10 Area of map CALIF 40 KILOMETERS 20 MILES Figure 22.  Map of the San Francisco Bay region, California, showing breaks in water-distribution pipes from the moment-magnitude-6.9 1989 Loma Prieta earthquake overlaid on peak ground acceleration (PGA) for the event. g, acceleration due to gravity. (Data from Lund and Schiff, 1992.) • Water cut off immediately to 63 percent of customers, including hospitals and disaster centers; • Major damage to 65 water reservoirs and about 87 pumping plants that would require months, or even years, to repair; • Loss of water for fire hydrants and increased fire risk; • An estimated impact of $1.2 billion (in 1994 dollars) to the regional economy owing to fire damage and lack of water; and • More than 5,500 pipelines serving homes and businesses to break; • A likelihood of untreated drinking water resulting from damage to four of six treatment plants; • EBMUD’s most critical water conduit, the Claremont Tunnel, to be cut off west of the Oakland/Berkeley hills—affecting 70 percent of EBMUD customers; • Lack of water weeks after an earthquake, with some customers lacking service for as long as six months afterwards. . . . As a result of the 1994 water system study, EBMUD developed a $189 million Capital Improvement program that, between 1995 and 2007, resulted in a system-wide mitigation of these impacts with the goal of providing an improved post-earthquake 388   The HayWired Earthquake Scenario—Engineering Implications functional water system with no redundancies. . . . In addition, portable equipment, such as pumps, hoses and generators, required to maintain operations following a disaster, has been procured. A number of other facilities still require seismic upgrades. . . Finally, roadway and building damage in EBMUD’s service area may result in delays in recovery that may necessitate on-going communication with service vehicles to ensure that repairs to pipelines and critical facilities are completed in a timely manner. Although dating from 2011, these estimates (for distribution piping damage) actually rely on analyses developed in 1994. However, although key facilities such as the Claremont Water Tunnel, which crosses the Hayward Fault in Alameda County, have been improved, little has changed since 1994 regarding distribution piping, and the situation remains largely the same today (EBMUD, oral commun., October 30, 2014). To examine the impacts of this situation following the HayWired mainshock, two sources of information were used to estimate the number and pattern of distribution pipe breaks and leaks. Data on pipe breaks and leaks from Porter (Water Supply, this volume) was used for one of the main water-distribution service areas affected by this earthquake, that of EBMUD. Outside of the EBMUD service area, a more approximate method was used to estimate water-main breaks and leaks, which consisted of assuming an “average” water main was under each street, and basing damage to water-distribution 122.5° networks on that assumption. Sections of pipe in zones of high liquefaction susceptibility are shown in figure 23. Based on this data, the HayWired mainshock devastates the water-supply infrastructure in the affected region, causing a total of about 9,400 buried water mains to require repairs,6 owing to a combination of fault rupture, shaking, and permanent ground displacement. The result is a lack of water supply to most hydrants in the east bay (fig. 24). Without water infrastructure, firefighters will have to resort to alternative water sources, which in many cases require hard suction hose. Hard suction hose is a specific type of fire hose that allows a fire engine to create a vacuum in order to draft water from a source that is not pressurized (such as a swimming pool, river, or bay; fig. 25). The hose is reinforced with embedded metal rings to be circumferentially rigid so as to withstand an external pressure (such as internal vacuum). In the United States, the National Fire Protection Association specifies hard suction hose as standard equipment for class-A fire engines. However, in recent years some fire departments have adopted a practice of keeping hard suction hose in fire stations rather than carried on their engines. A limited survey of San Francisco Bay region fire departments conducted as part of this study found only about one-third of the departments could be confirmed as carrying hard suction hose on their engines. 6 The estimate of 9,400 buried water mains requiring repairs is the total from Porter (Water Supply, this volume) combined with the estimate in this paper based on street lengths. 122° A B 0 0 38° Area of B 1 2 3 1 2 4 5 KILOMETERS 3 MILES Berkeley San Francisco Emeryville Oakland 37.5° Area of map CALIF 0 10 0 5 20 KILOMETERS N Alameda 10 MILES Figure 23.  Maps of the San Francisco Bay region, California, showing water mains in areas of high liquefaction susceptibility in an earthquake (water mains were presumed to be under each road). A, Overview map of the San Francisco Bay region; B, detail map of parts of the Cities of, Berkeley, Emeryville, Oakland, and Alameda. Water-main susceptibility to liquefaction—red, very high; light red, high; yellow, moderate; pink, low. (Roads from U.S. Census Bureau, 2015; liquefaction data for most of the region from Witter and others, 2006, which omits San Francsico County.) Chapter P. Fire Following the HayWired Scenario Mainshock 389 123° 122° 38° PAC IF IC O CE AN Figure 24.  Map of the San Francisco Bay region, California, showing likelihood of the availability of water service within fire station Voronoi areas (a proxy for a fire station’s response area, see fig. 8) following the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario. Red areas approach zero likelihood of water service. The length of the Hayward Fault ruptured in the scenario is shown on the map. EXPLANATION Water supply factor <0.10 0.10–0.20 0.20–0.30 0.30–0.40 0.40–0.50 0.50–0.60 0.60–0.70 0.70–0.80 Area CALIF of map 0.80–0.90 >0.90 37° HayWired rupture 0 0 20 10 40 KILOMETERS 20 MILES Gas and Liquid Fuels Gas and liquid fuel are used throughout many modern cities; there are buried major transmission lines (fig. 26) with associated terminals, refineries, and tank farms. A rupture of one large gas or liquid-fuel transmission line can be catastrophic and require the resources of a major fire department to respond. Similarly, a major petroleum refinery fire requires a major response, which may not be possible in the immediate aftermath of an earthquake. The San Francisco Bay area has five major petroleum refineries, which constitute 40 percent of California’s refining capacity. These refineries are concentrated at the north end of the HayWired scenario fault rupture. In the Mw 7.0 scenario mainshock, these crucial refineries will experience severe shaking such that at least one (and possibly several) refineries will have major fires that may burn for several days, as has occurred in the past few decades in large earthquakes near refineries, such as the Mw 8.3 2003 Tokachi-Oki, Japan, and Mw 7.6 1999 İzmit, Turkey, earthquakes. In the bay area, and very significantly, gas distribution pipes underlie nearly every street, with connections to nearly every building. Ignitions from these sources typically account for about 25 percent of the total number of fire-following-earthquake ignitions. Figure 25.  Photograph of a San Francisco Fire Department, California, engine and firefighters using a hard-suction hose to draft water from a cistern (photograph by Charles Scawthorn). 390   The HayWired Earthquake Scenario—Engineering Implications 122°30' 122° 122°30' 122° 38° Oakland F SAN RA IS O C I SC NC C AN BA Y O 37°30' Oakland San Francisco FR SAN San Francisco BA Y A B EXPLANATION EXPLANATION Peak ground acceleration (g) 0.1 0.2 0.3 0.4 0.5 0.6 Liquefaction susceptibility 0.7 0.8 0.9 1.0 Very high High Pipelines Moderate Low Pipelines Gas pipeline Liquid-fuel pipeline Gas pipeline Liquid-fuel pipeline HayWired rupture HayWired rupture 0 Area of map CALIF 0 20 MILES 20 KILOMETERS Figure 26.  Maps showing gas and liquid-fuel transmission pipelines in the San Francisco Bay region, California. A, Pipelines overlaid on scenario peak ground acceleration distribution for the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario on the Hayward Fault. B, Pipelines overlaid on zones of high liquefaction susceptibility. Length of fault rupture in the HayWired scenario shown by black lines. g, acceleration due to gravity. (Pipeline data from U.S. Department of Transportation, 2015; mainshock data from Aagaard and others, 2017; liquefaction data for most of the region from Witter and others, 2006, which omits San Francisco County.) Communications Communications systems, particularly telephone networks, will sustain some damage but perhaps not enough to reduce functionality following the HayWired mainshock. However, congestion will reduce functionality to a great degree, for several hours or more. This lack of telephone service will result in delayed reporting of fires, with consequences as discussed above. Transportation The transportation system most relevant to fire following earthquake is the road network, which is most vulnerable at bridge crossings. The California Department of Transportation (Caltrans) has nearly completed a major seismic review and retrofit of all bridges under its purview (California Department of Transportation, 2014). Although the local road and highway networks are sufficiently dense in most places that redundant pathways exist within the San Francisco Bay region, heavy traffic following a major earthquake could significantly impede emergency responders. Emergency strike teams arriving from outside of the bay region may also be delayed owing to traffic disruptions at several chokepoints at the boundaries of the region, including U.S. Route 101 north of the Golden Gate Bridge and south of San Jose and westbound Interstates 80 and 580. Chapter P. Fire Following the HayWired Scenario Mainshock 391 Regional and State Response The HayWired scenario mainshock primarily affects California Governor’s Office of Emergency Services (Cal OES) region II (fig. 27). The most likely sources of regional resources will be a number of strike teams assembled by Cal OES from the Central Valley, arriving in the affected region within 6–24 hours. Some of these will be brush rigs (wildland fire engines specifically designed to assist in fighting wildfires), which are more suited to wildland than urban-structural fires. By the time of their arrival in the region affected by the HayWired mainshock, the issue will be large fires that have grown into conflagrations, constituting a much larger challenge. Outside of region II, Cal OES is likely to stage a number of strike teams, drawn generally from southern California and the Central Valley. Assembling 100 strike teams, consisting of approximately 500 pumpers and other firefighting apparatus, as well as firefighters, is easily within Cal OES capability, and several times this number of people and equipment can be managed if necessary. Within about 12 hours of notification, 100 strike teams can arrive at staging areas, with probably another 100 teams arriving during the next week. In our analysis, however, mutual aid will be largely ineffective in the immediate period following the HayWired mainshock, owing to the following factors: • Access: • The east bay hills are quite steep, with relatively narrow and winding roads that hinder access. • The hills are also heavily vegetated which, combined with prevailing winds and topography, will greatly enhance fire spread and impede firefighting. • Supplying water to higher elevations in the hills will be very difficult. • Limited access to the San Francisco Bay area. Final Burned Area The 453 large fires estimated to follow the HayWired scenario mainshock will be spread over a large area of varying building density and availability of water for firefighting. The number of large fires that will grow into conflagrations, and the ultimate extent of the final burned area, will depend on the building density, weather conditions, initial unfought size of the fire before fire department response, number of responding fire • Delayed response time to fire scene: • Fire departments in the San Francisco Bay area (for example, peninsular and Walnut Creek-Concord area) will conserve resources and not be able to respond quickly to the east bay. • Mutual aid will have to come from farther afield (northern California, southern California, and the Central Valley), requiring at least several hours, and will be arriving at night in blackout conditions (owing to wide-scale failure of electric power). • Water shortages: • Water-tanker truck refills will be at some distance from fires, resulting in delays. Although a few fire departments (Berkeley, Oakland, Vallejo, and San Francisco) have portable water-supply systems (PWSS), these are currently inadequate for the demands that will be placed on them. • Aerial firefighting effectiveness in urban areas is currently unknown. • Firefighting foam is a “force-multiplier,” greatly increasing the effectiveness of a hose stream. However, current local fire-department supplies of foam are limited. Figure 27.  Map showing California Governor’s Office of Emergency Services mutual-aid regions (from California Governor’s Office of Emergency Services, 2017). 392   The HayWired Earthquake Scenario—Engineering Implications engines and water supply available for firefighting associated with each large fire. Under the assumed scenario conditions, it is estimated that of the 453 large fires, about 321 will grow to a size such that they will spread beyond the city block of origin (in other words, become a conflagration), with the final burned area then largely dependent on fires crossing streets and other firebreaks. Based on the probability of fire crossings, the estimated final burnt area is approximately 79 million square feet of residential and commercial building floor area equivalent to more than 52,000 single-family dwellings.7 This loss is equivalent to a total building replacement value of almost 7 An average single-family equivalent dwelling is 1,500 square feet of residential or commercial occupancy floor area, and this measure is used to normalize and communicate overall building losses in a readily comprehensible way. A loss of 1.5 million square feet of residential and commercial building for example is equivalent to 1,000 single-family dwellings. Most people can more readily interpret the loss of 1,000 houses than 1.5 million square feet of floor area. 8 Based on a replacement cost of $200 per square foot. Note this is a conservative estimate of replacement cost. Hogan (2014) estimates that construction in San Francisco can cost $300 per square foot, not counting subsidies, permits, and selling expenses. $16 billion8 (2014 dollars), representing about 2 percent of the entire exposed value (fig. 28, table 4), with most of the loss concentrated in Alameda County. Under the assumed wind and humidity conditions during the HayWired mainshock, the areas of most concern for fire following earthquake are parts of Alameda and Santa Clara Counties, where large areas of relatively uniform, dense, low-rise buildings provide a fuel bed such that dozens to hundreds of large fires are likely to merge into many dozens of conflagrations, destroying tens of city blocks. Two particular concerns exist in this regard—(1) if Diablo winds are present (which is not assumed in this scenario), losses could be much larger; and (2) if extremely calm conditions exist (which is also not assumed in this scenario), a symmetric wind pattern could develop where uprising air from conflagrations draws air inward (an example of the stack effect) to create a selfsustaining feedback situation (commonly termed a firestorm), which can be very destructive. Although relatively unlikely, this potential should not be ignored. The first concern is a larger mass conflagration, fed by higher winds; the second is potentially much worse. Both would be catastrophic. 123° 122° San Pablo Bay 38° C PA C IF I OC EA N San co cis an Fr Figure 28.  Map of the San Francisco Bay region, California, showing final burned-area losses (in millions of 2014 U.S. dollars) from fire following earthquake after the hypothetical magnitude-7.0 mainshock of the HayWired earthquake scenario. Areas shown are fire station Voronoi areas (a proxy for a fire station’s response area, see fig. 8). The length of the Hayward Fault ruptured in the scenario is shown on the map. EXPLANATION 37.5° y Ba Fire losses, in millions of dollars 0 1–10 10–50 50–100 100–500 500+ HayWired rupture Area of map CALIF Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. 0 0 10 20 10 30 40 KILOMETERS 20 MILES Chapter P. Fire Following the HayWired Scenario Mainshock 393 Another major concern is the very large concentration of high-rise buildings in the financial district of San Francisco. Firefighting under postearthquake conditions in more than one of these buildings could be beyond the resources of the San Francisco Fire Department, so the loss of several high-rises is quite possible. • More than 100 ignitions (each) following the 1971 San Fernando, California (Mw 6.6); 1994 Northridge, California (Mw 6.7); and 1995 Kobe, Japan (Mw 6.9), earthquakes. • At least 348 ignitions, more than any other earthquake in history, occurring in the 2011 Tohoku, Japan (Mw 9.0), earthquake and tsunami (Anderson and others, 2016) Uncertainty, Verification, and Validation • The 1991 East Bay Hills Fire (East Bay Hills Fire Operations Review Group, 1992), a massive conflagration centered in the study region that overwhelmed fire and water agencies. There is considerable uncertainty in the estimates of ignitions and final burned area presented above. The United States has been very fortunate in having few large earthquakes within urban areas in the past 50 or more years, so the experience database for ignitions following earthquakes is relatively sparse and has significant uncertainty, as can be seen in the confidence bands shown in figure 18, which span an order of magnitude. Moreover, the majority of data for ignitions (178 of 244 or 73 percent) are drawn from very early morning earthquakes, a time of day associated with low normal fire occurrence (U.S. Fire Administration, 2008). A full exploration of uncertainty is beyond the scope of the present study, but the number of ignitions estimated to follow the HayWired mainshock (668) could vary by hundreds, depending on many factors. Regarding verification (accuracy of the estimate) and validation (meeting the intended need) of estimates of ignitions and final burned area in the HayWired scenario, a large earthquake like the Mw 7.0 mainshock is a rare event, and the postearthquake fire situation even rarer, so that verification and validation is very challenging. Verification is particularly difficult, owing to the sparsity of data and experience. Qualitatively, the following experiences tend to support the scenario losses presented above: • Precedent—several events support the potential for large postearthquake losses, including in the HayWired study region: • Quantitatively, the methods in this study were used to hindcast (estimate): • Ignitions in previous California earthquakes (tables 6–8 and fig. 29) with reasonable agreement. The base data is dataset A in Davidson (2009a). • Large fires for the 1989 Loma Prieta (Mw 6.9) and 2014 South Napa (Mw 6.0), California, earthquakes, the only events for which sufficient data on all aspects (fire resources, firefighting water availability, and so forth) was available. Although this quantitative verification is quite limited, it tends to confirm the reasonableness of the estimates and also illustrate the uncertainty. Both the SPA Risk LLC (2009) and Davidson (2009b) ignition models produce enough fires following the HayWired mainshock to overwhelm currently available firefighting resources in the San Francisco Bay region, so the conclusions of this chapter would be the same regardless of which model was used. Validation (meeting the intended need) for fire following earthquake in the HayWired scenario is also a challenge but toward this end the above methodology and findings were presented to a workshop on fire following earthquake, held at the University of California, Berkeley, Richmond Field Station on October 29, 2014. The workshop was attended by 76 personnel, representing 31 fire departments • Catastrophic fires following the 1906 San Francisco (Mw 7.8) and 1923 Tokyo (Mw 7.9) earthquakes. Table 6.  Summary count of ignition data from California earthquakes since 1971 (SPA Risk LLC, 2009, 2014). Earthquake 1971 San Fernando Number of ignitions 91 Date source1 Unpublished data 1983 Coalinga 3 Scawthorn (1984) 1984 Morgan Hill 6 Scawthorn (1985) 1986 North Palm Springs 1 Earthquake Engineering Research Institute (1986) 1987 Whittier Narrows 20 Wiggins (1988) 1989 Loma Prieta 36 Mohammadi and others (1992); Scawthorn (1991) 1994 Northridge 81 Scawthorn and others (1997) 2014 Napa Total number of ignitions 1 See SPA Risk LLC (2009, 2014) for detailed references. 6 244 SPA Risk LLC (2014) 394   The HayWired Earthquake Scenario—Engineering Implications Table 7.  Hindcast (estimated) ignitions for selected California earthquakes (1984–2014), using equations from Davidson (2009b) and SPA Risk LLC (2009) (see discussion in text and equations 1 and 3, respectively). [NA, not applicable] Observed ignitions 41 1 13 36 81 6 Earthquake 1984 Morgan Hill 1986 North Palm Springs 1987 Whittier 1989 Loma Prieta 1994 Northridge 2014 Napa Davidson (2009b; model A.NB2) estimated ignitions 1.2 2.1 22.2 29.5 99.0 NA SPA Risk LLC (2009) estimated ignitions 4.0 4.1 72.1 15.9 166.4 6.24 1 There were four structural ignitions in Morgan Hill and two in San Jose in the 1984 earthquake. The total of six is indicated in table 6. For validation, only Morgan Hill was modeled, so table 7 shows only four observed ignitions. and emergency response agencies. The workshop was subsequently independently evaluated by Allison Madera and others (Natural Hazards Center, University of Colorado Boulder, written commun., 2016), who found “almost all of the survey respondents (95.8 percent) indicated that they believed the HayWired scenario accurately represented what a fire following earthquake incident might look like in the San Francisco Bay Area.” Table 8.  Observed and hindcast (estimated) large fires for selected northern California earthquakes. Fire type Observed Estimated 1989 Loma Prieta earthquake1 Total ignitions 31 24 Large Fires 12 Negligible Conflagrations 1? Negligible 2014 Napa earthquake Total ignitions 6 6.24 Large Fires 1 Negligible Conflagrations ? Negligible Impacts of Fire Following Earthquake This section discusses the human and economic impacts of fire following earthquake. Not well understood, but discussed here, is the major impact such events can have on the insurance industry. Based on 1990 census population (dataset A in Davidson, 2009a). 1 Estimated number of ignitions 200 180 160 EXPLANATION Davidson (2009b) SPA Risk LLC (2009b) A B 100 140 120 100 10 80 60 line t agr erfec fp line o 40 nt eeme c erfe of p ent em re t ag 20 0 20 40 60 80 100 1 1 10 100 Observed number of ignitions Figure 29.  Observed and hindcast (estimated) ignitions for selected California earthquakes (1984–2014), using equations from Davidson (2009b) and SPA Risk LLC (2009) (see discussion in text and equations 1 and 3, respectively; see table 7 for data). A, Number of ignitions plotted on arithmetic axes. B, Number of ignitions plotted on log-log axes. Chapter P. Fire Following the HayWired Scenario Mainshock 395 Human Impacts Estimating the fatalities associated with the fires following the HayWired mainshock is very problematic. A very simple approach is taken here; in the 1991 East Bay Hills Fire, which destroyed approximately 3,500 dwellings, 25 people perished. The building losses projected here are approximately 20 times larger. In proportion, there would be hundreds of deaths caused by fire following the scenario mainshock. Such an approach is admittedly very simplistic and does not account for the potential overwhelming of the regional emergency medical capacity in a large earthquake, as opposed to the isolated nature of the 1991 fire. Injuries would probably be an order of magnitude greater. For the HayWired scenario, an estimated 500,000 to 1 million people will need shelter as a result of fire following earthquake. Economic and Insurance Impacts Regarding the estimated $16 billion value of the structures burned by fire following earthquake in the HayWired scenario, the value of contents and other improvements (for example, landscaping) will add to this loss. For example, the contents of residences are commonly insured to 70 percent of the replacement cost of the building, so content loss could realistically amount to an additional $11 billion. An additional loss is loss of use; that is, the people normally living in these destroyed buildings (or conducting business in them) must find other accommodations, which most likely would not be available in the San Francisco Bay region given the impact of the scenario mainshock. This loss, termed “additional living expenses” by the insurance industry, can be consequential, equivalent to many tens of billions of dollars. Accounting for this can be difficult; if people who have lost their dwellings are housed in hotels at insurance company expense, the loss is simply the hotel bill. If people are forced to live in tents following the mainshock, at public expense, there may be no bill.9 In such a situation, people have not paid for their tents, and cannot therefore claim against the insurance company for a financial loss. However, they have lost value in services (of their house) approximately equivalent to the rental value of their house (minus the rental value of the tent) but would not be compensated for those losses. Nevertheless, this is a loss that should be accounted for, overall. One approximation is to estimate the additional living expenses in proportion to the typical limit of liability for homeowner’s insurance—20 percent of the replacement cost of the buildings, which for the HayWired scenario is about $3 billion. Because virtually all buildings and contents in the United States are insured for fire, and U.S. insurance contracts include losses from fire following earthquake under the fire policy, the direct fire-following-earthquake losses for the HayWired mainshock are likely to result in a loss approaching $30 billion of insurance claims. Because $30 billion amounts to nearly 6 percent of the gross domestic product of the San Francisco Bay region—and shaking, liquefaction, and landslide-related 9 Note that public authorities may attempt to recoup their expenses, if the sheltered people are insured. damage adds to the demands for construction services—it is likely that demand surge will occur (the temporary increase in construction costs following major natural disasters). Losses of this magnitude are probably sustainable by the U.S. insurance industry (the $60 billion in insured claims arising from the September 11, 2001, attacks were handled without great strain). The 1991 East Bay Hills Fire, in which 3,500 homes were lost, at the time resulted in about $1 billion in insured losses—the event projected here is 23 years of inflation later and about 60 times as large. In summary, losses from fire following earthquake are likely to be the largest part of the insured losses in the scenario event, and would be one of the largest single-loss events in the history of the insurance industry. Another aspect of the economic impacts is the loss of real-estate tax revenues. A loss of tens of billions of dollars in the value of property improvements is likely to result in perhaps a decrease of a billion dollars in regional real-estate tax revenues for several years, directly attributable to fire following earthquake. Mitigation of Fire Following Earthquake Mitigation of fire following earthquake has been extensively discussed elsewhere (Scawthorn and others, 2005). Only some limited observations specific to the HayWired scenario are provided here. Fire-Service Opportunities The fire service in California is perhaps the most experienced in the world in dealing with large conflagrations, owing to the wildland fires recurring annually in the region. Fire departments have also been relatively diligent in preparing for a large earthquake—the CERT program is a model in that regard. However, the following opportunities for improvement are noted: • Improvements are needed in the ability to more quickly assess the event and facilitate fire incident reporting. Reconnaissance using unmanned aerial vehicles, as well as cellular text-messaging incident reports directly to a 9-1-1 portal, could be developed and operationalized. • Alternative water sources need to be better identified and access and water movement capabilities enhanced. Hard suction hoses could be carried on all engines. Large diameter hose (LDH) systems, comparable to San Francisco Fire Department’s PWSS (Scawthorn and others, 2006), could be developed on a regional basis. In this regard and as part of the HayWired scenario project, the earlier mentioned October 29, 2014, workshop was held at the University of California’s Richmond Field Station. The four existing PWSSs, belonging to Berkeley, Oakland, San Francisco, and Vallejo Fire Departments 396   The HayWired Earthquake Scenario—Engineering Implications (fig. 30) were brought together and used in a joint exercise for the first time. • A regional multidisciplinary task force could be formed within the fire service, to examine urban conflagration potential in more detail. Water-Service Opportunities Water-service providers in California have worked to prepare for a major earthquake, but more can still be done (Scawthorn, 2011a,b). One overriding issue with regard to fire following earthquake is that water agencies typically are not institutionally responsible for fire protection. That is, although they provide hydrants, if the hydrants fail to supply water, the water agency is not responsible. Therefore, water-system upgrades are typically more oriented to maintenance of customer service and minimizing direct damage to the system than to maximizing water-supply reliability. A mandate could be developed to make water agencies more responsive to this need. Given the realities of the limited water supply in California, this may be unlikely to occur, but should at least be raised for discussion. A real way in which water agencies could be more responsive to the problem of fire following earthquake is if each agency were to configure and upgrade their system so as to provide a “backbone” system of water mains of high seismic reliability, which would both help ensure the reliability of water services to communities and provide fire departments with sources to draw water from to suppress a conflagration using an LDH system. This entire aspect is discussed in more detail in Scawthorn (2011a,b). Building-Standards Opportunities Since the 1906 San Francisco earthquake, significant progress has been made in making buildings more earthquake and fire resistant, yet there are still opportunities for improvement. For example, residential fire sprinklers are now required by many communities for new construction (at a cost less than the carpeting), but generally there are no requirements for existing homes (where the cost is significantly higher). Similarly, seismic retrofitting of existing buildings is increasingly being considered for older commercial buildings, but very few communities have requirements for existing single-family homes. Seismic retrofitting would reduce the number of postearthquake ignitions. Both seismic retrofitting and installation of fire sprinklers could be more widely mandated for existing buildings. Energy-Industry Opportunities The gas industry could contribute significantly to reducing the fire following earthquake by developing a program to either install automated gas shut-off valves (fig. 31) or redesign gas meters to have seismic shutoffs, particularly in densely built up areas. If the number of ignitions could be reduced by 25 percent, the number of large fires would be decreased in greater proportion and the total losses further reduced. For example, the City of Los Angeles Fire Department has shown leadership in seeking legislation to require gas shut-off valves. Note that the gas industry in Japan moved to do this proactively following the 1995 Kobe earthquake. Figure 30.  Photograph of four portable water-supply systems, belonging to the Berkeley, Oakland, San Francisco, and Vallejo Fire Departments, at the edge of San Francisco Bay in Berkeley, California, on October 29, 2014. This was the first time the four systems were brought together and used in a joint exercise. (Photograph by Charles Scawthorn.) Chapter P. Fire Following the HayWired Scenario Mainshock 397 Figure 31.  Photograph of an automatic gas shutoff valve installed after the meter on a gas-service line (photograph by Charles Scawthorn). In regard to electricity, opportunities to reduce fire following earthquake are problematic. Electric power often fails in large earthquakes, owing to automatic system trips, as well as damage to the system. However, the power failure usually takes several seconds, during which power is a source of many ignitions. Certain electric appliances (such as those with heating elements) can still cause fires even after power is cut. Large-scale intentional curtailment of electric power would be problematic, because some communications systems and other essential equipment would not be useable. Petroleum refineries and related facilities in the San Francisco Bay area are likely to sustain major fires in the HayWired scenario. The degree of earthquake preparedness of these facilities is generally unclear and may need to be reviewed. Conclusion That fire following earthquake is a significant problem in California is confirmed historically, by recent events, and by analysis. The Mw 7.0 mainshock of the HayWired earthquake scenario is estimated to result in approximately 668 ignitions, such that in Alameda, Contra Costa, and Santa Clara Counties dozens to hundreds of large fires are likely to merge into numerous conflagrations destroying tens of city blocks, with several of these potentially merging into one or more super conflagrations, destroying hundreds of city blocks. The ultimate burned area is estimated to total 79 million square feet of residential and commercial building floor area, equivalent to more than 52,000 single-family dwellings, with property losses approaching $30 billion. This loss is virtually fully insured and would be one of the largest single-loss events in the history of the insurance industry. Other economic impacts include the loss of perhaps $1 billion in local tax revenues. A number of opportunities exist for mitigating fire following earthquake, including greatly enhancing the potential postearthquake supply of water for firefighting and the use of automated gas shut-off valves, or seismic shut-off meters, in densely built areas. References Cited Aagaard, B.T., Boatwright, J.L., Jones, J.L., MacDonald, Porter, K.A., Wein, A.M., 2017, HayWired scenario mainshock ground motions, chap. C of Detweiler, S.T., and Wein, A.M., eds., The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., accessed November 16, 2017, at https://doi.org/10.3133/ sir20175013v1. Anderson, D., Davidson, R.A., Himoto, K., and Scawthorn, C., 2016, Statistical modeling of fire occurrence using data from the Tohoku, Japan earthquake and tsunami: Risk Analysis, v. 36, no. 2, p. 378–395. Association of Bay Area Governments, 2010, Taming natural disasters—Multi-jurisdictional local hazard mitigation plan for the San Francisco Bay area (2010 update of 2005 plan): Association of Bay Area Governments Resilience Program web page, accessed April 28, 2015, at http://resilience.abag.ca.gov/ wp-content/documents/ThePlan-Chapters-Intro.pdf. Bay Area Water Supply and Conservation Agency, [n.d.], Member agency map: Bay Area Water Supply and Conservation Agency web page, accessed February 22, 2018, at http://bawsca.org/ members/map. Bello, M., Cole, C., Knudsen, K.L., Turner, F., Parker, D., and Bott, J., 2006, San Francisco Bay area fire stations— Seismic risk assessment, in 100th Anniversary Earthquake Conference—Conference on Earthquake Engineering, 8th, San Francisco, Calif., April 18–22, 2006, Proceedings: Oakland, Calif., Earthquake Engineering Research Institute, paper no. 001662, 11 p. California Department of Finance, 2014, E-1 population estimates for cities, counties, and the state—January 1, 2013 and 2014: California Department of Finance web page, accessed April 28, 2015, at http://www.dof.ca.gov/research/demographic/reports/ estimates/e-1/view.php. 398   The HayWired Earthquake Scenario—Engineering Implications California Department of Transportation, 2014, Seismic retrofit program: California Department of Transportation web page, accessed April 28, 2015, at http://www.dot. ca.gov/hq/paffairs/about/retrofit.htm. California Department of Water Resources, [n.d.], Water management planning tool: California Department of Water Resources website, accessed February 22, 2018, at https:// gis.water.ca.gov/app/boundaries/. California Governor’s Office of Emergency Services, 2013, California multi-hazard mitigation plan: California Governor’s Office of Emergency Services, section 5.4, p. 250, accessed April 28, 2015, at http://www.caloes.ca.gov/ cal-oes-divisions/hazard-mitigation/hazard-mitigationplanning/state-hazard-mitigation-plan/. California Governor’s Office of Emergency Services, 2017, Regions: California Governor’s Office of Emergency Services web page, accessed Dec 31, 2017, at http://www. caloes.ca.gov/cal-oes-divisions/fire-rescue/regions. Datahub, [n.d.], California water district boundaries: Datahub web page, accessed February 22, 2018, at https://old. datahub.io/dataset/california-water-district-boundaries. Davidson, R.A., 2009a, Generalized linear (mixed) models of post-earthquake ignitions: University of Buffalo, State University of New York, Multidisciplinary Center for Extreme Event Research, no. MCEER-09-0004, 124 p. East Bay Municipal Utility District, 2011, Annex to 2010 Association of Bay Area Governments local hazard mitigation plan; taming natural disasters—East Bay Municipal Utility District: East Bay Municipal Utility District, 42 p., accessed April 28, 2015, at http://resilience.abag.ca.gov/wp-content/ documents/2010LHMP/EBMUD-Annex-2011.pdf. East Bay Municipal Utility District, 2017, Water supply: East Bay Municipal Utility District web page, accessed November 16, 2017, at http://www.ebmud.com/water-and-drought/about-yourwater/water-supply/. Federal Emergency Management Agency, 2003, Hazus multihazard loss estimation methodology, earthquake model, Hazus®-MH 2.1 technical manual: Federal Emergency Management Agency, Mitigation Division, 718 p. [Also available at https:// www.fema.gov/media-librarydata/20130726-1820-25045-6286/hzmh2_1_eq_tm.pdf.] Hills Emergency Forum, 2005, Fires in the Oakland–Berkeley Hills: Hills Emergency Forum web page, accessed April 28, 2015, at http://www.hillsemergencyforum.org/docs/fire%20 history%20eastbay%20hills.pdf. Hogan, Mark, 2014, The real costs of building housing: The Urbanist web page, accessed April 28, 2015, at http://www.spur. org/publications/article/2014-02-11/real-costs-building-housing. Kawasumi, H., ed., 1968, General report on the Niigata Earthquake of 1964: Tokyo, Tokyo Electrical Engineering College Press, 648 p. Davidson, R., 2009b, Modeling postearthquake fire ignitions using generalized linear (mixed) models: Journal of Infrastructure Systems, v. 15, p. 351–360. Lee, S., Davidson, R., Ohnishi, N., and Scawthorn, C., 2008, Fire following earthquake—Reviewing the state-of-the-art of modeling: Earthquake Spectra, v. 24, no. 4, p. 933–967. Davidson, R.A., Kendra, J., Li, Sizheng, Long, L.C., McEntire, D.A., Scawthorn, C., and Kelly, J., 2012, San Bruno California, September 9, 2010 gas pipeline explosion and fire: Newark, Del., University of Delaware, Final Project Report no. 56, 201 p., accessed April 28, 2015, at http://udspace.udel.edu/handle/19716/11337. Lund, L.V., and Schiff, A.J., 1992, TCLEE pipeline failure database: New York, Technical Council on Lifeline Earthquake Engineering, American Society of Civil Engineers. Detweiler, S.T., and Wein, A.M., eds., 2017, The HayWired earthquake scenario—Earthquake hazards: U.S. Geological Survey Scientific Investigations Report 2017–5013–A–H, 126 p., accessed November 16, 2017, at https://doi. org/10.3133/sir20175013v1. National Board of Fire Underwriters, 1923 Report on the Berkeley, California conflagration of September 17, 1923: New York, National Board of Fire Underwriters, Committee on Fire Prevention and Engineering Standards, 9 p. Ding, A., White, J.F., Ullman, P.W., and Fashokun, A.O., 2008, Evaluation of HAZUS-MH flood model with local data and other program: Natural Hazards Review, v. 9, no. 1, p. 20–28. East Bay Hills Fire Operations Review Group, 1992, The East Bay Hills Fire—Multi-agency review of the October 1991 fire in the Oakland/Berkeley Hills: California Office of Emergency Services, 74 p. National Board of Fire Underwriters, 1905, Report of National Board of Fire Underwriters by its Committee of Twenty on the City of San Francisco, Cal.: New York, 64 p. National Commission on Fire Prevention and Control, 1973, America burning—The report of the National Commission on Fire Prevention and Control: Washington, D.C., U.S. Government Printing Office, 191 p. Routley, J.G., [n.d.], The East Bay Hills Fire, OaklandBerkeley, California: Federal Emergency Management Agency, U.S. Fire Administration, USFA-TR-060, 130 p., accessed November 16, 2017, at https://www.usfa.fema.gov/ downloads/pdf/publications/tr-060.pdf. Chapter P. Fire Following the HayWired Scenario Mainshock 399 San Francisco Chronicle, 2014, Huge San Francisco fire destroys six-story building: San Francisco Chronicle SFGATE website, accessed November 14, 2017, at http://www.sfgate. com/bayarea/article/Huge-San-Francisco-fire-destroys-sixstory-5308589.php. San Jose State University, 2014, San Francisco Bay wind archives: San Jose State University WeatherSpark website, accessed October 21, 2014, at http://www.met.sjsu.edu/cgi-bin/wind/ new_windarchive.cgi?data=obs;year=2012;month=4;day=18;h our=16. Scawthorn, C., 2000, ed., The Marmara, Turkey earthquake of August 17, 1999: University at Buffalo, State University of New York, Multidisciplinary Center for Earthquake Engineering Research Reconnaissance Report 00-0001, 169 p., accessed accessed April 28, 2015, at https://mceer.buffalo.edu/ pdf/report/00-0001.pdf. Scawthorn, C., 2011a, Water supply in regard to fire following earthquake: Pacific Earthquake Engineering Research Center PEER Report 2011/08, 173 p., accessed accessed April 28, 2015, at http://peer.berkeley.edu/publications/peer_reports/ reports_2011/webPEER-2011-08-Scawthorn.pdf. Scawthorn, C., 2011b, Fire following earthquake: Pacific Earthquake Engineering Research Center graphic handout, 4  p., accessed accessed April 28, 2015, at http://peer.berkeley. edu/publications/peer_reports/reports_2011/Fire Following Earthquake-online-view-layout-sm.pdf. Scawthorn, C., Cowell, A.D., and Borden, F., 1998, Fire-related aspects of the Northridge Earthquake: National Institute of Standards and Technology Grant/Contract Report 98-743, 165 p. Scawthorn, C., Eidinger, J.M., and Schiff, A.J., 2005, Fire following earthquake: Technical Council on Lifeline Earthquake Engineering Monograph no. 26, 345 p. Scawthorn, C., and Khater M., 1992, Fire following earthquake— Conflagration potential in the greater Los Angeles, San Francisco, Seattle and Memphis areas: EQE International, prepared for the Natural Disaster Coalition, Boston. Scawthorn, C., and O’Rourke, T.D., 1989, Effects of ground failure on water supply and fire following earthquake—The 1906 San Francisco earthquake, in O’Rourke, T.D., and Hamada, M., eds., U.S.-Japan Workshop on Liquefaction, Large Ground Deformation and Their Effects on Lifelines, 2nd, September 26–29, 1989, Proceedings: Buffalo, N.Y., National Center for Earthquake Engineering Research, Technical Report 89-0032, p. 16–35. Scawthorn, C., O’Rourke, T.D., and Blackburn, F.T., 2006, The 1906 San Francisco earthquake and fire—Enduring lessons for fire protection and water supply: Earthquake Spectra, v. 22, no. S2, p. 135–158. Scawthorn, C., Yamada, Y., and Iemura, H., 1981, A model for urban post-earthquake fire hazard: Disasters, v. 5, no. 2, p. 125–132. SPA Risk LLC, 2009, Enhancements in Hazus-MH, Fire following earthquake—Task 3, Updated Ignition Equation: SPA Risk LLC, 74 p. SPA Risk LLC, 2014, 24 August 2014 South Napa Mw 6 Earthquake Reconnaissance Report: SPA Risk LLC, 63 p., accessed accessed April 28, 2015, at http://www.sparisk.com/ pubs/SPA-2014-Napa-Report.pdf. U.S. Census Bureau, 2015, TIGER/line shapefiles: U.S. Census Bureau, accessed April 2015, at https://www.census.gov/geo/ maps-data/data/tiger-line.html. U.S. Department of Transportation, 2015, National Pipeline Mapping System public viewer: National U.S. Department of Transportation website, accessed accessed April 28, 2015, at https://www.npms.phmsa.dot.gov/PublicViewer/. U.S. Fire Administration, 2008, Residential structure and building fires: U.S. Fire Administration, Federal Emergency Management Agency, 84 p. Van Anne, C., and Scawthorn, C., eds., 1994, The Loma Prieta, California, earthquake of October 17, 1989—Fire police transportation and hazardous materials, chap. C of Mileti, D.S., coord., Societal Response: U.S. Geological Survey Professional Paper 1553, 44 p., accessed April 28, 2015, at https://pubs.usgs. gov/pp/pp1553/. WeatherSpark, 2014, The typical weather anywhere on Earth: WeatherSpark website, accessed October 21, 2014, at http:// www.weatherspark.com. Wikipedia, 2017, October 2017 northern California wildfires: Wikipedia web page, accessed November 14, 2017, at https:// en.wikipedia.org/wiki/October_2017_Northern_California_ wildfires. Witter, R.C., Knudsen, K.L., Sowers, J.M., Wentworth, C.M., Koehler, R.D., and Randolph, C.E., 2006, Maps of Quaternary deposits and liquefaction susceptibility in the central San Francisco Bay region, California, part 3—Description of mapping and liquefaction interpretation: U.S. Geological Survey Open-File Report 2006–1037, ver. 1.1., https://pubs. usgs.gov/of/2006/1037/. The HayWired Earthquake Scenario—Engineering Implications Edited by Shane T. Detweiler and Anne M. Wein Scientific Investigations Report 2017–5013–I–Q [Also see https://doi.org/10.3133/sir20175013] Chapter Q How Many Injuries Can Be Avoided in the HayWired Scenario Through Earthquake Early Warning and Drop, Cover, and Hold On? By Keith A. Porter1 and Jamie L. Jones2 Abstract One of many potential benefits of earthquake early warning (EEW) is the reduction in injuries because people have more time to perform the self-protective series of actions called “drop, cover, and hold on” (DCHO). We offer an initial estimate of the potential benefit of EEW and DCHO in the HayWired earthquake scenario, both in terms of avoided injuries and the acceptable cost to avoid those injuries from the perspective of the U.S. Government. The HayWired scenario examines a hypothetical earthquake (mainshock) with a moment magnitude (Mw) of 7.0 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of California’s San Francisco Bay area. Using estimates of EEW warning time in the HayWired scenario, new observations of the time required to DCHO, and Hazus-MH (Federal Emergency Management Agency, 2012) estimates of the number of nonfatal injuries in the scenario, we estimate that out of 18,000 estimated nonfatal injuries in the HayWired scenario, as many as 1,500 people who would otherwise be injured could complete the DCHO actions before the arrival of strong motion and avoid injury. The estimated value of these prevented injuries is approximately $300 million. This is not the same as the probabilistic benefit of EEW and DCHO, because it is conditioned on the occurrence of a single earthquake among many whose occurrence is uncertain. However, it is a useful index of the potential benefit of EEW. The figures—1,500 avoided injuries and $300 million value of statistical injuries avoided—represent upper bounds because we do not know how many people could realistically receive EEW and how effectively DCHO prevents injuries. Introduction The HayWired scenario examines a hypothetical earthquake (mainshock) with a moment magnitude (Mw) of 7.0 University of Colorado Boulder. 1 U.S. Geological Survey. 2 occurring on April 18, 2018, at 4:18 p.m. on the Hayward Fault in the east bay part of California’s San Francisco Bay area. The hypothetical earthquake has its epicenter in Oakland, and strong ground shaking from the scenario causes a wide range of severe impacts throughout the greater bay region. Background Earthquake early warning (EEW) can provide several seconds advanced warning before the arrival of earthquake waves that cause strong ground motion. In that time, a person who receives the warning can take self-protective action. As of this writing, the American Red Cross, Earthquake Country Alliance, and others recommend a sequence of actions to protect oneself called “drop, cover, and hold on” (DCHO). To DCHO, a person drops to hands and knees on the floor, covers their head and neck with their arms, and if there is sturdy furniture such as a table nearby under which to take shelter, crawls there and holds on to the table legs to ensure that the cover does not slide away from the person. DCHO is also used as shorthand to cover a variety of additional advice for what to do during earthquake shaking in a variety of situations, such as outdoors, in a vehicle, and in a stadium. The Earthquake Country Alliance advises people to DCHO as a means to reduce the risk of injury, especially from falling objects (see, for example, https://earthquakecountry.org/step5/). If DCHO reduces the risk of injury, then using EEW to get to cover sooner, before earthquake shaking arrives and causes items to fall, should further reduce injuries in earthquakes. To our knowledge no one has quantified the efficacy of DCHO; we do not know by how much DCHO reduces earthquake injury risk. We can establish an upper-bound estimate of its efficacy by assuming that DCHO prevents all earthquake injuries associated with falling, building contents, building nonstructural components, and actions such as jumping out of windows or attempting to catch falling objects. 402   The HayWired Earthquake Scenario—Engineering Implications Objectives Our objectives in this study are twofold. First, estimate the benefit of earthquake early warning in terms of the number of human injuries that could be avoided if the HayWired mainshock were to occur after the population of the San Francisco Bay region were all trained and drilled on DCHO and equipped with EEW. Second, estimate the economic value of the avoided injuries. We do not examine other applications of EEW such as stopping trains. Literature Review EEW has been in development since at least 1995, among its many goals is reducing deaths and injuries (see Anderson and others, 1995; Lee and others, 1996; or Gasparini and others, 2007). EEW has been implemented in Japan and is available there as a free app for Android and iOS devices, called Yurekuru Call (Sung, 2011). The U.S. Geological Survey (USGS), along with a coalition of university partners, is developing and testing an EEW system called ShakeAlert for the West Coast of the United States (Burkett and others, 2014). If EEW can aid people in avoiding injuries, it is possible to assign an economic value to the avoided injuries. Since 1993, the U.S. Government has been required to demonstrate the cost effectiveness of new regulations, including those intended to enhance public safety (Clinton, 1993). To meet this requirement an acceptable cost to avoid statistical deaths and nonfatal injuries had to be established. “Statistical deaths and injuries” means deaths and injuries to unknown persons at some uncertain future date, as opposed to the value of a particular person’s life or present or past injuries. Different agencies of the U.S. Government assign these values differently but produce generally similar values. The U.S. Department of Transportation recently assigned a value of $9.1 million per statistical fatality avoided in 2013 U.S. dollars (USD), lesser values for nonfatal injuries, and an inflation factor of 1.18 percent per year for years after 2013 (U.S. Department of Transportation, 2014). The degree of the nonfatal injuries is measured using the Abbreviated Injury Scale (AIS) of the Association for the Advancement of Automotive Medicine (Gennarelli and Wodzin, 2005). In Natural Hazard Mitigation Saves—An Independent Study to Assess the Future Savings From Mitigation Activities (Multihazard Mitigation Council, 2005), Porter and colleagues employed Hazus-MH’s (Federal Emergency Management Agency, 2012) estimates of deaths and injuries to estimate the cost effectiveness of natural-hazard mitigation, producing a value of $4 saved for every $1 invested by the Federal Emergency Management Agency (FEMA) on natural-hazard mitigation. To produce this value, the Hazus-MH 4-level injury severity scale was related to the 6-level scale of the AIS, using the definitions for each scale (Multihazard Mitigation Council, 2005). The potential benefits of avoiding earthquake injuries are great. Porter and others (2006) show that the injuries experienced in the Mw 6.7 January 17, 1994, Northridge earthquake in southern California had an economic value of $2–3 billion in 2005 USD—that is, if all of the injuries in the 1994 Northridge earthquake could have been prevented, an expense of $2–3 billion by the U.S. Government would have been justifiable. That work drew on Shoaf and others (1998), who offer statistics on the number, severity, and causes of injuries in several earthquakes, including the 1994 Northridge earthquake. They show that 55 percent of injuries in that earthquake were caused by nonstructural damage, 22 percent by the “physical force of the earthquake,” 12 percent by behavior such as jumping out a window, 1 percent by “structural objects,” and the 10 percent by other causes. These statistics can provide insight into the fraction of injuries that might realistically be prevented or avoided by DCHO. Johnston and others (2014) provide analogous statistics for the Mw 7.1 September 4, 2010, Darfield and Mw 6.3 February 22, 2011, Christchurch, New Zealand, earthquakes. Practicing DCHO is a key component of earthquake drills during annual ShakeOut exercises, which are held around the world. Practicing DCHO is the first item covered at ShakeOut. org (https://www.shakeout.org), which coordinates global ShakeOut registration and provides instructions and other resources. McBride and others (2014) evaluated the 2012 New Zealand ShakeOut and reported on the degree to which participants performed DCHO: Over 60 percent of people seen by observers actively participated in drop, cover, and hold on. Of those who did not participate, disability and age (too young and too old) were reported to have been factors preventing participation. McBride and others (2014) also reported that embarrassment played a role in some people not participating. Becker and others (2017) reported higher participation in DCHO in the 2015 New Zealand ShakeOut, with 65 percent of respondents indicating that everyone they saw was performing DCHO and approximately 35 percent of people responding “not everybody” or “no-one.” The authors reported a 10-percent drop in people’s reluctance to practice DCHO because of disability and suggest that messaging from the New Zealand Ministry of Civil Defence and Emergency Management may have helped in addressing concerns about how to perform DCHO with impaired mobility. It is also possible that experiencing the 2012 New Zealand ShakeOut may have helped people be less embarrassed about performing DCHO during the 2015 ShakeOut. Lindell and others (2016) offer insight into what people do in actual earthquakes, having surveyed 257 people who experienced the 2011 Christchurch, New Zealand, earthquake and 332 people who experienced the 2011 Tohoku, Japan, earthquake. A considerable part of both groups of respondents reported their initial response to shaking was to freeze in place (38 percent of Christchurch respondents) rather than taking Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   403 self-protective actions such as DCHO (only 17 percent of Christchurch respondents). However, it is noteworthy that as of 2011, New Zealand had not yet participated in a ShakeOut drill (which tends to promote training and practice of DCHO), so it is perhaps to be expected that few people completed DCHO. Lambie and others (2017) used closed-circuit television recordings to observe the reactions of 213 people during the 2011 Christchurch earthquake; none performed DCHO. DCHO would prevent few if any of them, so the EEW+DCHO benefit would be considered to accrue only from the reduction in nonfatal injuries from causes other than structural collapse. We estimate B in any particular location i as the product of three quantities: • I=number of nonfatal injuries under as-is conditions (without EEW), • F(t)=fraction of occupants who could in principle DCHO after receiving EEW and before arrival of strong motion t seconds later, and Methodology The New Zealand and Tohoku experiences suggest that before ShakeOut drills became common, few people actually completed DCHO in an earthquake and that even during ShakeOut drills some people do not attempt to DCHO. An EEW system that recommends DCHO might increase DCHO participation in two ways—(1) by giving warning time so that people can complete DCHO before strong shaking begins (and while they still can do it) and (2) by reminding people what to do as part of the warning message. Suppose that training, EEW messaging, and other preparations were able to encourage action in an earthquake, such that nearly everyone would at least attempt DCHO. What might be the upper bound of benefit from combining DCHO and EEW, in terms of reduced injuries in an actual earthquake? To estimate the number of avoided injuries, one can estimate the number of people who can complete DCHO actions after they receive an EEW message and before strong ground motion arrives at their location. That estimate requires an estimate of the time it takes people to complete DCHO, which we calculated from a survey of more than 400 people who took online training and then timed themselves performing the actions. The estimate also requires one to estimate the number of people who would have been injured without EEW and DCHO. We estimated those quantities on a geographic basis using Hazus-MH. Finally, the estimate of avoided injuries requires an estimate of what fraction of injuries can be avoided by successfully completing DCHO. Estimating that quantity is problematic. It seems prudent to assume that DCHO cannot help much in the event of building collapse but that DCHO can prevent some large fraction of nonfatal injuries in the absence of collapse; we make a reasonable guess. Details of the methodology follow. Estimating Avoided Injuries Our goal is to produce an upper-bound estimate of how many fewer people would be nonfatally injured in the HayWired scenario if everyone in the San Francisco Bay region received EEW and had been well trained and drilled on DCHO. We refer to the reduction as the injury-prevention benefit from the combination of EEW and DCHO, or EEW+DCHO. We denote the benefit by B. We associate earthquake deaths entirely with structural collapse and assume that • f=fraction of nonfatal injuries that could be avoided by DCHO. Then the upper bound of avoided injuries in a particular earthquake can be estimated as: N B = ∑ I i × F (t i ) × f , i=1 (1) where Ii denotes the number of nonfatal injuries estimated in geographic location i (for example, a census tract or ZIP code) and F(ti) denotes the fraction of people in location i who could take effective self-protective action within the available warning time ti. The multiplicands can be estimated as follows—Ii is estimated using Hazus-MH (Federal Emergency Management Agency, 2012), and ti is the warning time at location i (the time between the EEW alert and arrival of S waves). The warning time calculations during a fault rupture are explained in appendix 2. Equation 2 estimates the warning time at location i: ti = Ri − tL , V (2) where Ri is hypocentral distance to location i in kilometers, considering the 8-kilometer (km) focal depth of the HayWired mainshock; V is the shear-wave velocity in rock (3.4 kilometers per second, km/s); and tL is latency time (from earthquake nucleation to transmission of the warning—5 seconds, s). F(ti) can be estimated as the cumulative distribution function of DCHO reaction time from a population survey evaluated at ti, the warning time for location i. The fraction of nonfatal injuries that could be avoided by DCHO, f, can be estimated using data compiled from injuries suffered in the Northridge and possibly other earthquakes. Shoaf and others (1998) provide the necessary data—55 percent of injuries result from nonstructural objects, 22 percent from earthquake force, and 12 percent from behavior. It seems reasonable that effective DCHO could prevent injuries from nonstructural objects such as a bookcase or television, by shielding the person from falling or sliding nonstructural objects. It also seems reasonable that DCHO could prevent injuries associated with earthquake force by having the person drop to the floor on hands and knees and therefore avoid being thrown by the force of the earthquake. Finally, DCHO is a behavior that substitutes for injurious ones such as jumping out of windows or trying to catch falling objects—two 404   The HayWired Earthquake Scenario—Engineering Implications examples of behavior-related injuries in the Northridge earthquake. Summing these injuries, it seems DCHO could conceivably prevent 0.55+0.22+0.12=89 percent of injuries, suggesting f=0.89. The remaining 11 percent of injuries in the Northridge earthquake were associated with structural objects (1 percent) and other causes (10 percent). Although some furnishings have supported structural objects in past earthquakes (for example, steel desks in a school building), it seems conservative to assume that this is not a general case, so we omit the 1 percent of injuries associated with structural objects from potential benefits of DCHO. Shoaf and others, (1998) provide no additional detail regarding the remaining 10 percent of injuries associated with other causes; we assume that DCHO would not prevent these either. We are aware of no research on the effectiveness of DCHO, so f=0.89 should be thought of as an upper bound on the number of nonfatal injuries that would in practice be avoided by DCHO. The figure f=0.89 with its two significant figures may give a false impression of precision. It is probably only meaningful to one significant figure, such as, f=0.9, but common engineering practice calls for carrying an extra significant figure to reduce cumulative rounding errors. Note that Johnston and others (2014) categorize causes of injuries differently from Shoaf and others (1998) and lump together injuries that occurred during an earthquake and in its aftermath, such as helping others or injuries caused by glass. As a result, it is problematic to estimate f using Johnston and others (2014) data. Depending on what one includes and excludes, one can estimate f from these data to be between 69 percent and 99 percent. The range overlaps and therefore does not contradict the value derived using Shoaf and others, (1998). Table 1.  Acceptable cost to avoid a statistical injury, calculated using Multihazard Mitigation Council (2005). Estimating the Acceptable Cost to Avoid Injuries 6. Demographic questions. One can calculate the acceptable cost to avoid statistical injuries, B2, as suggested in Multihazard Mitigation Council (2005) as follows: In an initial trial study, volunteers were recruited through Twitter feeds and social media accounts. The initial study yielded only 65 responses, so the earlier sample was ignored and more than 500 participants were recruited through Qualtrics Panels, a paid service provided to the University of Colorado Boulder. Subsequent findings refer only to responses from Qualtrics Panels. 3 B2 = ∑ Bi ×V j , j=1 (3) where Bj denotes the number of avoided injuries of severity j and Vj denotes the U.S. Government’s acceptable cost to avoid a statistical injury of severity j. We take values of statistical injuries avoided from U.S. Department of Transportation (2014), inflate them to 2015 USD as instructed there, and map them to Hazus-MH injury severity levels using Multihazard Mitigation Council’s (2005) table F-5 mapping 2. Table 1 presents those figures, rounded to two significant figures to avoid the appearance of excessive precision. Hazus injury severity1 Acceptable cost per avoided injury, in 2015 U.S. Dollars2 1. Basic medical aid by paraprofessionals 28,000 2. More than 1 but not life threatening 660,000 3. Life threatening but not immediately fatal 3,700,000 4. Fatal 9,400,000 Federal Emergency Management Agency (2012). 1 U.S. Department of Transportation (2014). 2 appendix 1. The protocol was approved by the university’s Institutional Review Board (IRB) on November 17, 2015. The survey instrument has six parts: 1. An introduction that explains the purpose of the survey, its procedures, risks, benefits, confidentiality, and consent to participate. 2. Training materials, text and brief YouTube videos, on how to DCHO. 3. Instructions on where and how to time oneself performing DCHO. 4. A question to determine the setting where the volunteer performed the exercise, in terms of Hazus-MH occupancy classification. 5. A series of questions to examine how well the training material worked. Findings Because physical dexterity and proximity to protective furniture differs from person to person and by time of day, one should expect that people will take different amounts of time to complete DCHO. Survey results bear out that expectation. Survey Design Distribution of DCHO Reaction Time The University of Colorado Boulder developed a datacollection protocol to collect statistics about DCHO reaction time through a web-based survey instrument copied to Data on DCHO reaction times were collected using the web-based survey instrument shown in appendix 1 between December 18 and 23, 2015. As of this writing, we collected Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   405 1. It often reasonably agrees with observations of realworld variables, as it does here. 2. Like many real-world variables, it can only take on a positive value and has a specifiable median and logarithmic standard deviation. 3. It assumes the least knowledge about the variable in question, conditioned on the value of the median and logarithmic standard deviation. 4. Tradition—engineers have used the lognormal at least since the 1980s to characterize earthquake damage to building components. With the survey we can estimate that reaction time is approximately lognormally distributed with median 8.8 s and standard deviation of the natural logarithm equal to 0.40, as shown in figure 1. The lognormal appears to be a reasonable approximation of the sample data provided by the study participants according to a Lilliefors (1967) goodness-of-fit test. Respondents were aged between 18 and 95, with a mean age of 35. The majority of respondents (57 percent) were female, versus 50.8 percent in the general U.S. population in 2014 (U.S. Census, 2015). Most (69 percent) described their race or ethnicity as white/Caucasian (73.6 percent of the 1.00 Cumulative probability data from a large sample—525 people completed the survey out of 638 to whom the survey was sent, a response rate of 82 percent. We refer to these data as round-1 surveyed reaction times, in case future surveys are performed. Many of the data appear to reflect an incorrect understanding of when to stop counting or data-entry typographic errors—38 respondents reported times in excess of 20 s, some as long as 200 s. Omitting these 38 responses, the remaining 487 responses are shown in figure 1 along with a lognormal distribution fit to the data. The nontechnical reader can interpret figure 1 as follows. The dashed, stairstep line represents the survey respondents’ reaction time. For example, 10 percent of them were able to DCHO in 5 s or less (see how the dashed line passes through x=5 s, y=0.10), 50 percent in 9 s or less (x=9 s, y=0.50), and 90 percent in 15 s or less (x=15 s, y=0.90). The smooth S-shaped curve is a parametric distribution called the lognormal cumulative distribution function. It approximates the stairstep line with a more convenient mathematical equation. It is closely related to the familiar bell-shaped curve called the normal or Gaussian distribution. The lognormal cumulative distribution function has two variables that determine its shape. One of the variables, called the median, adjusts the x-value associated with the midpoint of the curve (the x-value associated with y=0.50). The other variable, called the standard deviation of the natural logarithm of the variable (or logarithmic standard deviation), adjusts the width of the S-shaped curve. The reader should understand that lognormal may be the parametric probability distribution more commonly used than any other in earthquake engineering. Engineers use it for any of several reasons: 0.75 0.50 EXPLANATION 0.25 Reported reaction times LN(θ=8.8 seconds, β=0.4) 0.00 0 5 10 15 20 Drop, cover, and hold on reaction time, in seconds Figure 1.  Graph of cumulative probability for “drop, cover, and hold on” (DCHO) reaction times. DCHO reaction times were collected from volunteers using the web-based survey instrument shown in appendix 1. Red dotted line shows the distribution of reported reaction times for 487 respondents (38 respondents who reported DCHO times in excess of 20 seconds were omitted). Black curved line shows the lognormal distribution—LN, lognormal; θ, median, β, standard deviation. U.S. population), 13 percent described their race or ethnicity as African American (14 percent of the U.S. population), 10 percent described their race or ethnicity as Hispanic (17 percent of the U.S. population), 8 percent described their race or ethnicity as Asian (6 percent of the U.S. population), 3 percent described their race or ethnicity as Native American (2 percent of the U.S. population), 1 percent described their race or ethnicity as Pacific Islander (same as the U.S. population), and 4 percent described their race or ethnicity as other (respondents were allowed to select all that applied). The majority (56 percent) have at most some college education (U.S. population is 61 percent), 33 percent have a 2- or 4-year college degree (U.S. population is 28 percent), and 11 percent had a masters, professional, or doctoral degree (U.S. population is 10 percent) (U.S. population education attainment percentages from U.S. Census Bureau, 2015a). Respondents were slightly poorer than the U.S. population in general—53 percent report a combined household income less than $40,000 per year (U.S. median household income is $53,657) and 91 percent report a combined household income less than $110,000 per year (90th percentile among the U.S. population is $155,000) (U.S. general population household income data from U.S. Census Bureau, 2015b). 406   The HayWired Earthquake Scenario—Engineering Implications Calculation of Benefits We now apply the foregoing procedures to evaluate the benefits of the combination of EEW and DCHO in the HayWired scenario mainshock. Warning time ti of equation 2 is shown for the mainshock in figure 2. See appendix 2 of this chapter for the derivation of figure 2. We can now apply equations 1 and 3. We compiled Ii by census tract from the Hazus-MH (Federal Emergency Management Agency, 2012) analysis of the HayWired mainshock. We calculated Ri at each census tract centroid, took V=3.4 km/s and tL as 5 s, (as discussed in appendix 2), and idealized F(t) as lognormally distributed with median reaction time equal to 8.8 s and logarithmic standard deviation equal to 0.40, so that equation 4 now follows: ⎛ ln (t / 8.8sec ) ⎞ F (t ) = Φ ⎜ ⎟. 0.40 ⎝ ⎠ (4) The total number of nonfatal injuries from the HayWired scenario mainshock is estimated to be 18,1303; EEW could avoid as many as 1,468 of these, or 8 percent (table 2). Dollar values in table 2 are rounded to two significant figures. The data in table 2 suggests that if EEW were fully implemented and everyone in the San Francisco Bay region had been trained and drilled on DCHO before the HayWired mainshock occurred, as many as ~1,500 people who would otherwise be injured would have time to complete DCHO before the arrival of strong motion and avoid nonfatal injury. (Table 2 shows a calculated figure of 1,468 avoided injuries, but we use “about 1,500” to avoid the appearance of excessive accuracy.) The acceptable cost to avoid those injuries is about $300 million. Recall that the results in table 2 assume that completing DCHO effectively avoids 89 percent of injuries, which was acknowledged as an upper bound. The actual benefit would be lower in proportion to the ratio of actual nonfatal injuries avoided to the upper bound. Note also that some of the injuries shown in the first column would also be avoided by DCHO without EEW, because the Hazus injury model predates widespread training in DCHO. Perhaps DCHO would be less effective in the absence of EEW because people would be trying to take action during strong motion and would be injured before successfully completing the DCHO actions. We do not speculate on injuries avoided by DCHO without EEW. EEW provides more advanced warning the farther the recipient is from the earthquake focus (fig. 3A), but injuries tend to be concentrated close to the focus. It is in the middle ground (where EEW provides at least some warning time but shaking is still strong enough to threaten life safety) that EEW combined with “drop, cover, and hold on” actions has the greatest potential to reduce injuries, as shown in figure 3B. Conclusions We do not know how effective DCHO is in preventing earthquake injuries, nor do we know how much time decisionmaking adds to DCHO reaction time. However, to begin to estimate the benefit of EEW and DCHO, we assume that training and drilling can reduce decision time to near zero relative to reaction time. We further neglect the benefit of DCHO without EEW and assume that DCHO before the arrival of strong motion can prevent almost all (f=89 percent) of nonfatal earthquake injuries. With all of these simplifying assumptions, we can estimate an upper bound to the benefit of the combination of EEW and DCHO. If everyone in the San Francisco Bay region were trained and drilled in DCHO and received EEW before the arrival of the HayWired scenario mainshock, the additional warning time provided by EEW would be sufficient for about 1,500 of 18,000 people who would otherwise be injured to take DCHO actions that successfully prevent their Table 2.  Upper-bound number and value of avoided injuries from earthquake early warning combined with “drop, cover, and hold on” actions in the San Francisco Bay region, California, in the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. Acceptable cost per avoided injury, in 2015 U.S. Dollars2 Acceptable cost to avoid, in millions of 2015 U.S. Dollars Number of people injured Maximum number of avoided injuries 14,081 1,216 28,000 34 2. More than 1 but not life threatening 3,491 218 660,000 144 3. Life threatening but not immediately fatal 558 34 3,700,000 127 Hazus injury severity1 1. Basic medical aid by paraprofessionals 4. Fatal Total 971 0 9,400,000 0 19,101 1,468 Not applicable 305 Federal Emergency Management Agency (2012). 1 U.S. Department of Transportation (2014). 2 3 The Hazus-MH analysis estimated damages for shaking from the HayWired mainshock, as well as liquefaction, and these numbers were combined. The number differs from the 16,000 injuries reported in Seligson and others (this volume) for the mainshock in the context of the HayWired earthquake sequence, because only the shaking hazard data was consistently available across all earthquakes in the sequence. Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   407 123° 122° 121° Sacramento 20 seconds PA Area of map CIF N a p a CALIFORNIA IC S o n o m a OC EA N Stockton 38° conds 0 se San Francisco Oakland P EXPLANATION S a n J o a q u i n a Block group density— residents per square kilometer c Alameda ic if < 2,845 c e a 4,230–7,072 n 7,073–12,912 Modesto Fremont O 2,845–4,229 Livermore Hayward 5 s e c o nds S t a n i s l a u s 12,913–24,904 24,905–49,531 49,532–100,101 100,102–203,949 Warning Time, in seconds 0 10 seconds San Jose S a n M a t e o S a n t a C r u z 37° 25 City County boundary Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary and population data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. M e r c e d 25 seconds M o n t e r e y S a n B e n i t o 0 0 10 10 20 MILES 20 KILOMETERS Figure 2.  Map of the San Francisco Bay region, California, with contours of warning time (ti ; see equation 2) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario and block group density (residents per square kilometer). 408   The HayWired Earthquake Scenario—Engineering Implications 123° 122° A Area of map CALIFORNIA Stockton 38° Oakland San Francisco Alameda Livermore N SA O SC CI AN FR PA Hayward C IF IC Fremont C E Y BA O A N San Jose EXPLANATION Percent avoided injuries per census tract as a result of using earthquake early warning < 20% 20–40% 37° 40–60% 60–80% 80–90% No avoided injuries No warning zone City Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W, latitude of origin, 0.0° N. 0 0 10 10 20 MILES 20 KILOMETERS Figure 3.  Maps of San Francisco Bay region, California, showing injuries potentially avoided because of earthquake early warning combined with “drop, cover, and hold on” actions in the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. A, Fraction of injuries potentially avoided as a result of using earthquake early warning; B, upper bound number of injuries avoided by census tract as a result of using earthquake early warning. %, percent. Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   409 123° 122° B Area of map CALIFORNIA Stockton 38° Oakland San Francisco Alameda SA N PA C IF IC FR AN CI SC O Livermore Hayward O BA Y Fremont C E A N San Jose EXPLANATION Number of avoided injuries per census tract as a result of using earthquake early warning 1–10 10–20 20–40 37° 40–80 80–160 160–212 No avoided injuries No warning zone City Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W, latitude of origin, 0.0° N. Figure 3.—Continued 0 0 10 10 20 MILES 20 KILOMETERS 410   The HayWired Earthquake Scenario—Engineering Implications injury before the arrival of strong motion. The U.S. Government would value mitigation measures to avoid that number of nonfatal injuries at approximately $300 million (2015 USD). Limitations and Research Needs The intent of this report is not to say that the probabilistic benefit of EEW and DCHO is $300 million, for several reasons. That figure is an upper bound, not an expected value, and it is conditioned on the occurrence of a single earthquake, when in fact the future date of such an earthquake is uncertain, and there are many other possible earthquakes where EEW and DCHO would contribute to probabilistic benefit. However, the figures of 1,500 injuries and $300 million are useful to understand the potential magnitude of the benefits of EEW and DCHO. There does not appear to be any published research or other evidence on the effectiveness of DCHO to prevent injuries. For example, we do not know how many injuries related to nonstructural objects, earthquake force, or human behavior would actually be avoided by DCHO. Other important, unanswered questions include the following—What fraction of fatal injuries could be avoided by DCHO? How much would reaction times differ in the real event as opposed to the calm setting of the survey? What fraction of injuries can be avoided without EEW, that is, if people DCHO when they begin to feel strong motion? How long after the initiation of strong motion do injuries occur? (Presumably the answer to this last question depends on the severity of motion.) One can imagine laboratory experiments using human simulacra (crash-test dummies), finite-element analysis, and other means to explore these questions, but such experiments do not appear to have been carried out. According to a National Science Foundation (NSF) program officer, NSF does not appear to have a program to address earthquake-induced injuries (David Mendonca, NSF, written commun., December 8, 2015), nor can we find record of a relevant program within the National Institutes of Health. References Cited Anderson, S., Kobara, S., Mathis, B., Rosing, D., and Shafrir, E., 1995, SYNERGIES—A vision of information products working together, in Miller, J., ed., Conference companion on human factors in computing systems: New York, Association for Computing Machinery, p. 423–424. Becker, J.S., Coomer, M., Potter, S.H., McBride, S.K., Lambie, E., Johnston, D.M., Cheesman, B., Guard, J., and Walker, A., 2016, Evaluating New Zealand’s ShakeOut national earthquake drills—A comparative analysis of the 2012 and 2015 events: Proceedings, 2016 New Zealand Society for Earthquake Engineering (NZSEE) Conference, Christchurch NZ, April 1–3, 2016. Burkett, E.R., Given, D.G., and Jones, L.M., 2014, ShakeAlert—An earthquake early warning system for the United States West Coast: U.S. Geological Survey Fact Sheet 2014–3083, 4 p., accessed December 20, 2015, at https://doi.org/10.3133/fs20143083. Clinton, W.J., 1993, Executive Order 12866 of September 30, 1993—Regulatory Planning and Review: Washington, D.C., Federal Register, v. 58, no. 190, accessed December 20, 2015, at http://www.archives.gov/federal_register/executive_orders/pdf/12866.pdf. Federal Emergency Management Agency, 2012, Hazus multihazard loss estimation methodology, earthquake model, Hazus®-MH 2.1 technical manual: Federal Emergency Management Agency, Mitigation Division, accessed July 18, 2017, 718 p., at https://www.fema.gov/media-librarydata/20130726-1820-25045-6286/hzmh2_1_eq_tm.pdf. Gennarelli, T.A., and Wodzin, E., eds., 2005, Abbreviated injury scale (AIS) 2005: Barrington, Ill., Association for the Advancement of Automotive Medicine, 68 p. Gasparini, P., Manfredi, G., and Zschau, J., 2007, Earthquake early warning systems: Berlin, Springer, 350 p. Johnston, D., Standring, S., Ronan, K., Lindell, M., Wilson, T., Cousins, J., Aldridge, E., Ardagh, M.W., Deely, J.M., Jensen, S., Kirsch, T., and Bissell, R., 2014, The 2010/2011 Canterbury earthquakes—Context and cause of injury: Natural Hazards, v. 73, no. 2, p. 627–637, accessed December 20, 2015, at https://doi.org/10.1007/s11069-014-1094-7. Lee, W.H.K., Shin, T.C., and Teng, T.L., 1996, Design and implementation of earthquake early warning systems in Taiwan: 11th World Conference on Earthquake Engineering, Acapulco, Mexico, paper no. 2133, accessed December 20, 2015, at http://www.iitk.ac.in/nicee/wcee/article/11_2133. PDF. Lilliefors, H.W., 1967, On the Kolmogorov-Smirnov test for normality with mean and variance unknown: Journal of the American Statistical Association, v. 62, no. 318, p. 399– 402, accessed December 20, 2015, at https://doi.org/10.108 0/01621459.1967.10482916. Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   411 Lindell, M.K., Prater, C.S., Wu, H.C., Huang, S.-K., Johnston, D.M., Becker, J.S., and Shiroshita, H., 2016, Immediate behavioural responses to earthquakes in Christchurch, New Zealand, and Hitachi, Japan: Disasters, v. 40, no. 1, p. 85−111. McBride, S.K., Becker, J.M., Coomer, M.A., Tipler, K., and Johnston, D.M., 2014, New Zealand ShakeOut observation evaluation report—A summary of initial findings: GNS Science Report 2013/61, Institute of Geological and Nuclear Sciences Ltd, Lower Hutt, New Zealand, 39 p. Multihazard Mitigation Council, 2005, Natural hazard mitigation saves—An independent study to assess the future savings from mitigation activities—vol. 1 and 2: Washington, D.C., National Institute of Building Sciences, 161 p., accessed December 20, 2015, at http://www.nibs. org/?page=mmc_projects#nhms. Porter, K.A., Shoaf, K., and Seligson, H., 2006, Value of injuries in the Northridge earthquake: Earthquake Spectra, v. 22, no. 2, p. 555–563. Shoaf, K.I., Sareen, H.R., Nguyen, L.H., and Bourque, L.B., 1998, Injuries as a result of California earthquakes in the past decade: Disasters, v. 22, no. 3, p. 218–235. Sung, S.J., 2011, How can we use mobile apps for disaster communications in Taiwan—Problems and possible practice: 8th International Telecommunications Society (ITS) AsiaPacific Regional Conference, Taiwan, 26–28 June, 2011, accessed December 20, 2015, at https://www.econstor.eu/bits tream/10419/52323/1/67297973X.pdf. U.S. Census Bureau, 2015a, Educational attainment in the United States—2014: U.S. Census Bureau web page, accessed January 4, 2016, at https://www.census.gov/data/ tables/2014/demo/educational-attainment/cps-detailed-tables. html. U.S. Census Bureau, 2015b, Current population survey (CPS): U.S. Census Bureau web page, accessed January 4, 2016, at http://www.census.gov/hhes/www/cpstables/032015/hhinc/ hinc01_000.htm. U.S. Census, 2015, QuickFacts—United States: U.S. Census Quick Facts website, accessed December 20, 2015, at https:// www.census.gov/quickfacts/. U.S. Department of Transportation, 2014, Guidance on treatment of the economic value of a statistical life (VSL), in U.S. Department of Transportation Analyses—2014 adjustment: Washington, D.C., U.S. Department of Transportation, accessed November 30, 2015, at https://www.transportation. gov/sites/dot.gov/files/docs/VSL_Guidance_2014.pdf. 412   The HayWired Earthquake Scenario—Engineering Implications Appendix 1. Survey of DCHO Reaction Times The following survey instrument was successfully administered by the University of Colorado Boulder to study participants using the Qualtrics Panels service in December, 2015. The protocol was approved by the university’s Institutional Review Board on November 17, 2015. Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   413 414   The HayWired Earthquake Scenario—Engineering Implications [Image used in original survey instrument not shown] Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   415 [Image used in original survey instrument not shown] 416 The HayWired Earthquake Scenario?Engineering Implications Outdoors: Move to a clear area if you can safely do so; avoid power lines, trees, signs, buildings, vehicles, and other hazards. Driving: Pull over to the side of the road, stop, and set the parking brake. Avoid overpasses, bridges, power lines, signs and other hazards. Stay inside the vehicle until the shaking is over. If a power line falls on the car, stay inside until a trained person removes the wire. in a stadium or theater: Stay at your seat and protect your head and neck with your arms. Don't try to leave until the shaking is over. Then walk out slowly watching for anything that could fall in the aftershocks. Near the shore: Drop, cover, and hold on until the shaking stops. Estimate how long the shaking lasts. If severe shaking lasts 20 seconds or more, immediately evacuate to high ground as a tsunami might have been generated by the earthquake. Move inland 3 kilometers (2 miles) or to land that is at least 30 meters (100 feet) above sea level immediately. Don't wait for officials to issue a warning. Walk quickly, rather than drive, to avoid traffic, debris and other hazards. Below a dam: Dams can fail during a major earthquake. Catastrophic failure is unlikely, but ifyou live from a dam, you should know flood-zone information and have prepared an evacuation plan. Now, please watch these two videos, and then click "Next." Each video will open a new browser window or tab. When done, navigate back to this one. 1. If you are near a study desk or table 2. If there is no sturdy desk or table. Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   417 418 The HayWired Earthquake Scenario?Engineering Implications Store Warehouse Shop or service station Professional or technical services office Bank or financial institution -. Hospital -. Medical office or clinic Entertainment or recreation (for example, a restaurant or bar) Theater Parking lot or garage Industrial facility -. Heavy industry Light industry -. Food, drugs, or chemical factory Metals or minerals processing ., . High technology factory Construction office Agriculture .. Agricultural facility, such as a farm or ranch Religion or nonprofit Church, mosque, synagogue, food pantry, or other nonprofit Government General services, such as a government office . Emergency response, such as a police or fire station Education -. School (pre?K to 12) . College or university, other than dormitory Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   419 420 The HayWired Earthquake Scenario?Engineering Implications shaking from an earthquake occurs while I am in bed. (Please mark only one answer.) Neither Agree nor Strongly Disagree Disagree Disagree Agree Strongly Agree In the event of strong shaking from an earthquake while I am in bed, my self- protective actions should be: (Please mark only one answer.) Find an area outside the building that is clear of power lines and trees and go there. Drop out of bed to the floor, cover my head and neck under something sturdy, and hold on to the sturdy furniture or to my head and neck. . Stay in bed, hold on, and protect my head with a pillow. . Take cover near an exterior wall that doesn't have any windows. The instructions I read and videos I viewed made clear to me what to do if strong shaking from an earthquake occurs while I am in a high-rise building. (Please mark only one answer.) Neither Agree nor Strongly Disagree Disagree Disagree Agree Strongly Agree In the event of strong shaking from an earthquake while I am in a high-rise building, my self-protective actions should be: (Please mark only one answer.) Take the stairs to the first floor, get out of the building, and move to an area outside that is clear of power lines and trees. Drop to the floor, cover my head and neck under something sturdy, and hold on to the sturdy furniture or to my head and neck. Take the stairs to the first floor, get out of the building, and drive to a place of safety as directed by emergency managers. Take cover near an exterior wall that doesn't have any windows. Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   421 422 The HayWired Earthquake Scenario?Engineering Implications The instructions I read and videos I viewed made clear to me what to do if strong shaking from an earthquake occurs while I am in a stadium or theater. (Please mark only one answer.) Neither Agree nor Strongly Disagree Disagree Disagree Agree Strongly Agree In the event of strong shaking from an earthquake while I am in a stadium or theater, my self-protective actions should be: (Please mark only one answer.) Find an area outside that is clear of power lines and trees and go there. a Drop to the floor, cover my head and neck, and hold on to the seat or to my head and neck. Find an evacuation route and drive to a place of safety as directed by emergency managers. Stay in my seat and protect my head and neck with my arms. The instructions I read and videos I viewed made clear to me what to do if an earthquake occurs while I am near the ocean shore. (Please mark only one answer.) Neither Agree nor Strongly Disagree Disagree Disagree Agree Strongly Agree In the event of strong shaking from an earthquake when I am near the ocean shore, my self-protective actions should be: (Please mark only one answer.) Find an area that is clear of power lines and trees and go there. . Drop to the ground, cover my head and neck, and hold on to my neck and head until the shaking stops. Then evacuate to high ground. Find an evacuation route and drive to a place of safety as directed by emergency managers. Find a nearby building. Go inside and drop to the floor, cover my head and neck, and hold on to something sturdy. Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   423 424 The HayWired Earthquake Scenario?Engineering Implications materials more helpful? Demographic questions, to check how representative respondents are of the general population. What year were you born? What is your gender? Male Female Which of the following best describes your race or ethnicity? (Check all that apply.) White/Caucasian African American Hispanic Asian Native American Pacific Islander Other What is the highest level of education you have received? Less than High School Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   425 426 The HayWired Earthquake Scenario?Engineering Implications Powered by Qualtrics Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   427 Appendix 2. HayWired Mainshock Earthquake Early Warning Time Calculation By Elizabeth S. Cochrane,1 Anne M. Wein,1 Erin R. Burkett,1 Douglas D. Given,1 and Keith Porter2 We estimate the earthquake early warning (EEW) capabilities for the magnitude (M) 7.0 HayWired scenario mainshock based on the performance of the ShakeAlert (https:// www.shakealert.org) demonstration system during the M 4.0 Piedmont, California, earthquake on August 17, 2015. The Piedmont earthquake epicenter lies only 6 kilometers (km) from the epicenter of the HayWired scenario mainshock. Therefore, it is a reasonable proxy for our purposes here, because the initial alert latency (the delay between earthquake origin time and the time at which the first alert is issued) is primarily dependent on the local station density. The latency between the earthquake initiation and the initial ShakeAlert alert time was 4.8 seconds (s) for the 2015 Piedmont earthquake. The alert latency includes (1) the time for the P-wave to reach the surface and travel to at least four seismic stations, (2) waveform data to be transmitted to the processing centers (generally 1 s or less for California Integrated Seismic Network, CISN, stations close to the Piedmont earthquake), and (3) computation of the magnitude and location of the earthquake. The 3 km difference in hypocentral depth between the Piedmont earthquake (4.9 km) and the HayWired scenario earthquake (8 km) would add approximately 0.5 s to the P-wave travel time to the Earth’s surface—using an average speed of P-waves in California of 6 kilometers per second (km/s). Thus, we estimate that the initial alert time for the HayWired scenario mainshock would be approximately 5.3 s after the earthquake origin time—that is, after the rupture begins on the Hayward Fault. For the purposes of this scenario, we make a series of simplifying assumptions. We assume the length of the waveform segments that were used for the estimation of the Piedmont earthquake magnitude are similar to those used for characterization of the HayWired scenario mainshock. For the Piedmont earthquake, the closest station is at a hypocentral distance of 6.2 km, with a source-to-station P-wave travel time of about 1 s. Assuming a network transmission latency of 1 s, there was about 3 s of waveform available to analyze at this station before an alert was issued. In ShakeAlert, magnitude estimates can be made using waveform data from a single station. A source duration of about 3 s (as captured in the first 3 s of the P wave) would provide an initial magnitude estimate of at least M5.5 (for example, Meier and others, 2016). So, for the HayWired scenario mainshock, we estimate that an initial alert would be issued for a M5.5 earthquake at 5.3 s after the earthquake origin and that the magnitude estimate would U.S. Geological Survey. 1 University of Colorado Boulder. 2 subsequently be updated as the rupture continued to grow into a M7.0 earthquake. ShakeAlert plans to distribute alerts to the public when the observed ground motions are consistent with an earthquake that is above a minimum magnitude (for example, M>4.5), and alerts will be sent to regions expected to experience ground shaking above a minimum predicted Modified Mercalli Intensity (MMI) (for example, MMI≥II). For the HayWired scenario, we assume that when the first alert is sent the initial estimated magnitude would be M5.5, and alerts would be issued to regions expected to experience MMI greater than or equal to II. This corresponds to a region inside of which users are expected to feel ground shaking, with or without resulting damage. Using a standard intensity relation (Atkinson and others, 2014) to calculate the initial alerting region, we find that for a M5.5 earthquake, the area that encompasses MMI II or larger ground shaking extends to approximately 250 km. We estimate the range of warning times for this initial alerting region from the time it takes for S-waves traveling at a velocity of 3.4 km/s to travel distances between 0 and 250 km, minus the alert latency time (5.3 s). At the time of the initial alert, issued 5.3 s after the earthquake origin time, S-waves would have traveled a hypocentral distance of 17 km (or 15 km epicentral distance). This circular region defined by a 15-km radius around the epicenter defines a region of no warning for S-wave shaking. Note that we neglect the time it takes to distribute an alert to users. At epicentral distances greater than 15 km, a user of the ShakeAlert system would receive an alert before the arrival of S-waves. In the initial alert region, warning times range from no warning to more than 60 s of warning, assuming a constant S-wave velocity and no alert distribution latencies. This initial alert region encompasses the area that will eventually experience MMI III or greater shaking when the final magnitude (M7.0) is reached. Given that damage is unlikely to occur at shaking levels less than this, we do not extend this scenario through the entire rupture evolution. However, we expect that as the earthquake magnitude grows the alerting region will be also expanded in the seconds after the initial alert until the final scenario magnitude is reached. Assuming ShakeAlert algorithms correctly estimate the final earthquake magnitude of M7.0, then the alerting region will extend to epicentral distances of about 575 km. Here, we assume a point source that reflects the current capability of the ShakeAlert system, but a more realistic line source (finite rupture) EEW algorithm is currently under development for ShakeAlert and would result in a noncircular alert region. 428   The HayWired Earthquake Scenario—Engineering Implications We note that earthquake-generated P-waves tend to shake the ground vertically, whereas S-waves tend to shake the ground side-to-side, horizontally. P-waves tend to have smaller amplitudes than S-waves. Buildings, bridges, and other infrastructure tend to be stronger in resisting vertical motion than horizontal motion. For these reasons, the slower, later-arriving, horizontally shaking, and stronger S-waves tend to cause more damage to buildings, bridges, and other infrastructure than do the faster but weaker, vertically shaking P-waves. That is why, for EEW purposes, the warning time is typically measured between the alert time and the arrival of S-waves at a site. However, it is important to note that, particularly for sites close to the rupture, shaking from the P-wave can also be strong, and thus the time available to take action may be shorter (Meier, 2017; Minson and others, 2018). Also, warning times may be longer for cases where the shaking threshold is exceeded after S-wave arrival (Meier, 2017; Minson and others, 2018). The available warning time at a given location depends on a variety of factors, including (1) the time it takes for the rupture to grow (the final magnitudes of large earthquakes that rupture over a period of several to tens of seconds cannot be predicted in advance), (2) the magnitude and (or) shaking-intensity threshold for which an alert is generated, (3) the frequency at which alerts are updated (as the rupture evolves), (4) the time required to distribute alerts, and (5) how far the recipient of the warning is from the hypocenter (Minson and others, 2018). Table 3 catalogs the shaking intensities and warning times that an EEW alert could provide until S-wave arrival for major cities affected by the hypothetical M7.0 HayWired rupture. Figure 4 shows the HayWired instrumental intensity map and contours of the warning time until S-wave arrival. References Cited Atkinson, G.M., Worden, B., and Wald, D.J., 2014, Intensity prediction equations for North America: Bulletin of the Seismological Society of America, v. 104, no. 6, p. 3084– 3093, doi: 10.1785/0120140178. Meier, M.-A., Heaton, T., and Clinton, J., 2016, Evidence for universal earthquake rupture initiation behavior: Geophysical Research Letters, v. 43, no. 15, p. 7991–7996, doi: 10.1002/2016GL070081. Meier, M.-A., 2017, How “good” are real-time ground motion predictions from earthquake early warning systems: Journal of Geophysical Research—Solid Earth, v. 122, no. 7, p. 5561–5577, doi: 10.1002/2017JB014025. Minson, S.E., Meier, M.-A., Baltay, A.S., Hanks, T.C., and Cochran, E.S., 2018, The limits of earthquake early warning—Timeliness of ground motion estimates: Science Advances, v. 4, no. 3, doi: 10.1126/sciadv.aaq0504. Table 3.  Estimated earthquake early warning times and shaking intensity at select locations in the San Francisco Bay region, California, for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario. City Latitude Longitude Oakland 37.80 122.27 Berkeley 37.87 122.27 Hayward 37.67 122.08 San Francisco 37.78 122.42 San Mateo 37.55 122.31 Fremont 37.55 Vallejo 38.11 Redwood City San Rafael Epicentral distance, in miles Warning time, in seconds Shaking intensity1 8 0.0 VIII 7 11 0.0 IX 11 17 0.7 IX 13 21 1.8 VII 19 30 4.3 VII 122.99 21 33 5.2 IX 122.24 21 35 5.8 VII 37.48 122.24 23 36 6.0 VII 37.97 122.53 23 36 6.0 VII Livermore 37.68 122.77 24 39 6.9 VIII San Jose 37.34 121.89 36 58 12.4 VIII Modified Mercalli Intensity. 1 5 Epicentral distance, in kilometers Chapter Q. How Many Injuries Can Be Avoided Through Earthquake Early Warning and Drop, Cover, and Hold On?   429 123° 122° 121° Sacramento 20 seconds PA CIF Area of map N a p a IC S o n o m a CALIFORNIA OC EA N Stockton 38° nds eco 0s P EXPLANATION San Francisco Oakland a c Alameda ic if Modified Mercalli Intensity Livermore Fremont O c e V (moderate) Modesto Hayward III (weak) IV (light) S a n J o a q u i n a n VI (strong) 5 s e co n d s S t a n i s l a u s VII (very strong) VIII (severe) IX (violent) X+ (extreme) Warning time, in seconds 0 1 0 s e c onds S a n M a t e o San Jose S a n t a C l a r a S a n t a C r u z 37° 25 City County boundary Hydrology from U.S. Geological Survey National Hydrography Dataset, 2016. Boundary data from U.S. Census Bureau TIGER data, 2016. North American Datum of 1983 UTM 10N projection. Central meridian, 123° W., latitude of origin, 0.0° N. S a n B e n i t o 25 seconds M e r c e d M o n t e r e y 0 0 10 10 20 MILES 20 KILOMETERS Figure 4.  Map of the San Francisco Bay region, California, showing contours of estimated earthquake early warning times on a shaking intensity map (Modified Mercalli Intensity) for the hypothetical moment-magnitude-7.0 mainshock of the HayWired earthquake scenario on the Hayward Fault. For the scenario, the initial alerting region extends 250 kilometers from the earthquake epicenter, which is outside of the boundaries of the figure. Note that strong shaking can also occur at the time of the P-wave arrival, particularly close to the fault rupture, which would reduce available time for taking protective and risk-reduction actions. Menlo Park Publishing Service Center, California Manuscript approval date February 28, 2018 Edited by James W. Hendley II, Regan Austin, Scott Darling, Monica Erdman, Katherine Jacques, and Jessica Dyke Design and layout by Cory Hurd and Vivian Nguyen Detweiler and Wein, editors—The HayWired Earthquake Scenario—Engineering Implications—Scientific Investigations Report 2017–5013–I–Q ISSN 2328-0328 (online) https://doi.org/10.3133/sir20175013v2