Research Progress and Accomplishments · Suha Ülgen, Managing Director, Interactive Media and Geographic Information Systems T. Uzawa, Visiting Researcher, Department of Civil Engineering,

Accomplishments

A

Selection of

Papers

Chronicling

Technical

Achievements

of the

Multidisciplinary

Center for

Earthquake

Engineering

Research

1999 - 2000

Research Progress and

The Multidisciplinary Center for Earthquake Engineering Research (MCEER) is a national centerof excellence in advanced technology applications that is dedicated to the reduction of earthquakelosses nationwide. Headquartered at the University at Buffalo, State University of New York, the Centerwas originally established by the National Science Foundation (NSF) in 1986, as the National Centerfor Earthquake Engineering Research (NCEER).

Comprising a consortium of researchers from numerous disciplines and institutions throughoutthe United States, the Center’s mission is to reduce earthquake losses through research and the applica-tion of advanced technologies that improve engineering, pre-earthquake planning and post-earthquakerecovery strategies. Toward this end, the Center coordinates a nationwide program of multidisciplinaryteam research, education and outreach activities.

Funded principally by NSF, the State of New York and the Federal Highway Administration (FHWA),the Center derives additional support from the Federal Emergency Management Agency (FEMA), otherstate governments, academic institutions, foreign governments and private industry.

The Multidisciplinary Center for Earthquake Engineering Research

Research Progress and Accomplishments

1999 - 2000

Multidisciplinary Center for Earthquake Engineering ResearchUniversity at Buffalo, State University of New York

May 2000

MCEER-00-SP01Red Jacket Quadrangle, Buffalo, New York 14261

Tel: (716) 645-3391; Fax: (716) 645-3399; Email: [email protected] Wide Web: http://mceer.buffalo.edu

Foreword

by George C. Lee, Director,Multidisciplinary Center for Earthquake Engineering Research

iii

The research accomplishments of the Multidisciplinary Center for Earthquake Engineering Researchare as numerous as they are varied. Since the Center was established by the National Science Founda-tion (NSF) in 1986, its vision has been to help establish earthquake resilient communities throughoutthe United States and abroad. Over the past 14 years, our research and education programs haveannually supported more than 80 investigators throughout the country and the world, to work towardthis goal. Much has been accomplished, most notably in the areas of lifelines and protective systems,but our vision has not yet been fully realized.

Toward this end, we believe that the best way to achieve earthquake resilient communities in theshort-term is to invest in two highly focused system-integrated endeavors:

• the rehabilitation of critical infrastructure facilities such as hospitals and lifelines that societywill need and expect to be operational following an earthquake; and

• the improvement of emergency response and crisis management capabilities to ensure efficientresponse and prompt recovery following earthquakes.

Our research is conducted under the sponsorship of two major federal agencies, the National Sci-ence Foundation (NSF) Earthquake Engineering Research Centers Program and the Federal HighwayAdministration (FHWA), and the state of New York. Significant support is also derived from the FederalEmergency Management Agency (FEMA), other state governments, academic institutions, foreign gov-ernments and private industry. Together these resources are used to implement our research pro-grams.

This Research Progress and Accomplishments report is intended to provide the reader with anunderstanding of our current research efforts. The papers highlight research tasks that are in progressor have recently concluded, and provide those in the earthquake engineering community with aglimpse of the foci and direction that our programs are taking. We anticipate that this information willcontribute to the coordination and collaboration effort in earthquake engineering research nationallyand globally. The presentation is in descriptive form with preliminary observations and recommenda-tions, and an indication of future efforts is provided.

This past year was marked by two devastating earthquakes in Turkey and Taiwan. MCEER reconnais-sance teams visited both these areas, and several of the papers in this report describe their efforts tolearn from these tragic events. A few papers describe research that has been completed, most notablythe Federal Highway Administration-sponsored project on the seismic vulnerability of new highwayconstruction, while others describe work in progress. The authors identify the sponsors of the re-search, collaborative partners, related research tasks within MCEER’s various programs, links to re-search and implementation efforts outside MCEER’s program, and web site addresses for additionalinformation.

Although MCEER’s vision is to achieve earthquake resilient communities, it is the various profession-als in earthquake hazard mitigation who will collectively join forces to create them. We can providerelevant tools to these professionals, who include practicing engineers, policy makers, regulators andcode officials, facility and building owners, governmental entities, and other stakeholders who haveresponsibility for loss reduction decision making, and educate them on how to best use or apply these

iv

SocietySocietySociety

Owners ofCritical Facilities

• Hospitals Utilities Lifelines

EmergencyResponse and

Mitigation

Organizations(Local, State,

Federal)

U.S.

Government

Insurance and

Reinsurance

Industry

Advanced

Technology

Industries

Architects

Engineers

Planners

EarthquakeResilient

Communities

EarthquakeEarthquakeResilientResilient

CommunitiesCommunities

••

tools. Their decisions can encompass a range of actions, including the adoption of various new tech-nologies, the retrofitting of structures using improved techniques and approaches, and response andrecovery activities. This process, which moves from research through knowledge transfer, adoption,and implementation, is illustrated in the figure below. Each paper in this report highlights the endusers of the presented research or educational effort.

This report is the second in our annual compilation of research progress and accomplishments. It isavailable in both printed and electronic form (on our web site in PDF format at http://mceer.buffalo.edu/publications/default.asp, under Special Publications).

If you would like more information on any of the studies presented herein, or on other MCEERresearch or educational activities, you are encouraged to contact us by telephone at (716) 645-3391,facsimile (716) 645-3399, or email at [email protected].

Modeling Earthquake Impact on Urban Lifeline Systems: Advances and Integration 1Stephanie E. Chang (Coordinating Author), Adam Rose, Masanobu Shinozuka, Walter D.Svekla and Kathleen J. Tierney

Development of Fragility Information for Structures and Nonstructural Components 15Masanobu Shinozuka, Mircea Grigoriu (Coordinating Author), Anthony R. Ingraffea, SarahL. Billington, Peter Feenstra, Tsu T. Soong, Andrei M. Reinhorn and Emmanuel Maragakis

Damage to Critical Facilities Following the 921 Chi-Chi, Taiwan Earthquake 33Tsu T. Soong (Coordinating Author), George C. Yao and Chi-Chang Lin

Highway Bridge Seismic Design: Summary of FHWA/MCEER Project on Seismic 45Vulnerability of New Highway Construction

Ian M. Friedland

Ground Motion Prediction Methodologies for Eastern North America 63Apostolos S. Papageorgiou

Fiber Reinforced Composites for Advanced Seismic Performance of Water Supplies 71Thomas D. O’Rourke (Principal Author), James A. Mason, Ilker Tutuncu and Timothy Bond

The Marmara, Turkey Earthquake: Using Advanced Technology to Conduct 81Earthquake Reconnaissance

Ronald T. Eguchi (Coordinating Author), Charles K. Huyck, Bijan Houshmand, BabakMansouri, Masanobu Shinozuka, Fumio Yamazaki, Masashi Matsuoka and Suha Ülgen

Restoration Activities Following the Marmara, Turkey Earthquake of August 17, 1999 97Gary R. Webb

Human and Institutional Perspectives of the 921 Chi-Chi, Taiwan Earthquake 111George C. Lee and Chin-Hsiung Loh

Education and Educational Outreach: Using the Center Approach for Effective 121Knowledge Transfer

Andrea S. Dargush and George C. Lee

Graduate Professional Education in Earthquake Engineering: An Integrated Approach 131Andrei M. Reinhorn, Shahid Ahmad and Ernest Sternberg

Contents

v

Shahid Ahmad, Professor, Department of Civil, Structural and Environmental Engineering,University at Buffalo, State University of New York

Raul Barron-Corvera, Research Assistant, Department of Civil, Structural and EnvironmentalEngineering, University at Buffalo, State University of New York

Sarah L. Billington, Assistant Professor, School of Civil and Environmental Engineering, CornellUniversity

Timothy Bond, Laboratory Manager, School of Civil and Environmental Engineering, CornellUniversity

Ian G. Buckle, Professor, Department of Civil Engineering, University of Nevada, Reno

Stephanie E. Chang, Research Assistant Professor, Department of Geography, University ofWashington

Woon Hui Chong, Graduate Student, Department of Civil, Structural and EnvironmentalEngineering, University at Buffalo, State University of New York

Andrea S. Dargush, Assistant Director for Education and Research Administration,Multidisciplinary Center for Earthquake Engineering Research, University at Buffalo, StateUniversity of New York

Gang Dong, Post-doctoral Research Associate, Department of Civil, Structural and EnvironmentalEngineering, University at Buffalo, State University of New York

Ronald T. Eguchi, President, ImageCat, Inc.

Peter Feenstra, Senior Research Associate, School of Civil and Environmental Engineering,Cornell University

Maria Q. Feng, Associate Professor, Department of Civil and Environmental Engineering,University of California, Irvine

Ian M. Friedland, Associate Director for Development, Applied Technology Council

Mircea Grigoriu, Professor, School of Civil and Environmental Engineering, Cornell University

Benedikt Halldorsson, Graduate Student, Department of Civil, Structural and EnvironmentalEngineering, University at Buffalo, State University of New York

T. Han, Research Assistant, School of Civil and Environmental Engineering, Cornell University

Bijan Houshmand, Jet Propulsion Laboratory and Adjunct Associate Professor, ElectricalEngineering Department, University of California at Los Angeles

Charles K. Huyck, Vice President, ImageCat, Inc.

Anthony Ingraffea, Professor, School of Civil and Environmental Engineering, Cornell University

H. Kim, Visiting Professor, Department of Civil Engineering, University of Southern California;from Mokpo National University, Korea

George C. Lee, Director, Multidisciplinary Center for Earthquake Engineering Research, andSamuel P. Capen Professor of Engineering, University at Buffalo, State University of New York

Chi-Chang Lin, Professor and Chairman, Department of Civil Engineering, National Chung-HsingUniversity

Contributors

vii

viii

Chin-Hsiung Loh, Director, National Center for Research on Earthquake Engineering, NationalTaiwan University

Babak Mansouri, Ph.D. Candidate, Department of Civil Engineering, University of SouthernCalifornia

Emmanuel Maragakis, Professor and Chair, Department of Civil Engineering, University ofNevada, Reno

James A. Mason, Graduate Research Assistant, School of Civil and Environmental Engineering,Cornell University

Masashi Matsuoka, Chief Research Engineer, Earthquake Disaster Mitigation Research Center

Ronald Meis, Graduate Student, Department of Civil Engineering, University of Nevada, Reno

Ehab Mostafa, Research Assistant, School of Civil and Environmental Engineering, Cornell University

Thomas D. O’Rourke, Thomas R. Briggs Professor of Civil Engineering, School of Civil andEnvironmental Engineering, Cornell University

Apostolos S. Papageorgiou, Professor, Department of Civil, Structural and EnvironmentalEngineering, University at Buffalo, State University of New York

Andrei M. Reinhorn, Professor, Department of Civil, Structural and Environmental Engineering,University at Buffalo, State University of New York

Adam Rose, Professor and Head, Department of Energy, Environmental and Mineral Economics,The Pennsylvania State University

Christopher Roth, Research Assistant, School of Civil and Environmental Engineering, CornellUniversity

Masanobu Shinozuka, Fred Champion Chair in Civil Engineering, Department of CivilEngineering, University of Southern California

Raj Siddharthan, Professor, Department of Civil Engineering, University of Nevada, Reno

Tsu T. Soong, Samuel P. Capen Professor of Engineering Science, Department of Civil, Structuraland Environmental Engineering, University at Buffalo, State University of New York

Ernest Sternberg, Associate Professor, Department of Architecture and Planning, University atBuffalo, State University of New York

Walter D. Svekla, Graduate Research Assistant, Department of Geography, University of Washington

Kathleen J. Tierney, Co-Director, Disaster Research Center, and Professor, Department ofSociology, University of Delaware

Ilker Tutuncu, Graduate Research Assistant, School of Civil and Environmental Engineering,Cornell University

T. Ueda, Visiting Researcher, Department of Civil Engineering, University of Southern California;from Taisei Corp., Japan

Suha Ülgen, Managing Director, Interactive Media and Geographic Information Systems

T. Uzawa, Visiting Researcher, Department of Civil Engineering, University of Southern California;from Taisei Corp., Japan

Gary R. Webb, Post-doctoral Research Fellow, Disaster Research Center, and Department ofSociology, University of Delaware

Fumio Yamazaki, Assistant Professor, Earthquake Disaster Mitigation Research Center

George C. Yao, Associate Professor, Department of Architecture, National Cheng Kung University

1

Research Objectives

This research aims to develop and demonstrate an advanced, inte-grated earthquake loss estimation methodology for urban lifeline sys-tems. The methodology, which evaluates direct and indirect economiclosses from lifeline failures, provides a means for assessing both ex-pected losses from future earthquakes and potential loss reductionfrom mitigation alternatives. This effort builds on and coordinatesmultidisciplinary contributions from lifeline earthquake engineering,geography, sociology, and economics. An application is demonstratedfor the Memphis Light, Gas and Water Division (MLGW) water deliv-ery system.

In recent years, the massive losses caused by major urban earthquakes,combined with the increasing capabilities of computer technologies

such as geographic information systems (GIS), have led researchers andpractitioners to focus substantial energies on computerized methodolo-gies for estimating expected losses from future earthquake disasters. Atthe same time, new empirical data on the physical and economic effectsof recent earthquakes provide opportunities for re-evaluating, refining, andcalibrating loss estimation models. For discussions of current methodolo-gies and issues, see National Research Council (1999) and the 1997 Earth-quake Spectra special issue on loss estimation (including articles on theFEMA/NIBS HAZUS methodology).

The MCEER loss estimation research effort focuses on economic lossesassociated with earthquake-induced failures of critical urban lifeline systems,particularly water and electric power systems. It uses a multidisciplinary,coordinated approach to address some of the most salient challenges in thestate-of-the-art of earthquake loss estimation: capturing the systems responseof engineering and economic systems; calibrating models to limited empiri-cal data; improving our understanding of loss mechanisms, risk factors, andindirect economic loss; acknowledging loss estimate uncertainties; and,

Modeling Earthquake Impacton Urban Lifeline Systems:Advances and Integration

National Science Foundation,Earthquake EngineeringResearch Centers Program

Stephanie E. Chang,Research AssistantProfessor, Department ofGeography, University ofWashington

Adam Rose, Professor andHead, Department ofEnergy, Environmentaland Mineral Economics,Pennsylvania StateUniversity

Masanobu Shinozuka,Fred Champion Chair inCivil Engineering,Department of CivilEngineering, University ofSouthern California

Walter D. Svekla, GraduateResearch Assistant,Department of Geography,University of Washington

Kathleen J. Tierney,Co-director, DisasterResearch Center, andProfessor, Department ofSociology, University ofDelaware

by Stephanie E. Chang (Coordinating Author), Adam Rose,Masanobu Shinozuka, Walter D. Svekla and Kathleen J. Tierney

2

perhaps most fundamentally, inte-grating engineering and socioeco-nomic models in a balanced,insightful, and productive way.

This paper describes the method-ological approach and initial resultsof an application to the MLGW wa-ter delivery system for Memphisand Shelby County, Tennessee.Other ongoing efforts pertain to theMLGW electric power system andthe Los Angeles Department of Wa-ter and Power (LADWP) systems.Our ultimate objective is to dem-onstrate how the methodology canassist end users to assess and com-pare the potential benefits of dif-ferent pre- and post-earthquake lossreduction strategies.

Memphis Light, Gasand Water Case Study

Memphis and Shelby County(pop. 900,000) are at risk fromearthquakes originating in the NewMadrid Seismic Zone (NMSZ) in theCentral U.S. The NMSZ producedthe largest earthquakes in the re-corded history of the U.S. in thewinter of 1811-12, including at leastthree events with magnitude 8.0 orgreater. This paper presents initialfindings for three scenario earth-quakes (magnitudes 6.5, 7.0, and7.5, respectively) with epicenter atMarked Tree, Arkansas, some 55 kmnorthwest of downtown Memphis,in the southern arm of the NMSZ.

Scenario events with epicentral lo-cation elsewhere on the NMSZ willbe considered in further modeling.

Memphis Light, Gas and WaterDivision (MLGW) is the primarysupplier of water for Shelby County,with the exception of a few unin-corporated municipalities. Thewater source is an undergroundaquifer accessed by wells. Thewater delivery system consists ofa large low-pressure system andseveral high-pressure systems lo-cated on the outskirts of Memphiscity. The network includes about1,370 km of buried pipes and anumber of pumping stations, el-evated tanks, and booster pumps.

MethodologicalApproach

Previous research under the Na-tional Center for Earthquake Engi-neering Research (NCEER) haddeveloped a lifeline loss estimationmethodology and demonstrated itsapplication to the MLGW electricpower and water systems(Shinozuka et al., eds., 1998; Changet al., 1996; Rose et al., 1997).While that effort provides a foun-dation for the current work, nu-merous significant refinementshave been made. Highlights ofcurrent advances include:

• Model integration - “Seamless”integration of engineering and

Many types of users may benefit from this work. To the re-search community concerned with earthquake loss modelingand reduction, it provides an advanced methodology for evalu-ating the economic impacts of lifeline damage in earthquakes.To lifeline agencies and national, state and local governments,this research provides a means for evaluating the economicbenefits of potential lifeline loss reduction strategies.

Memphis Light, Gas and WaterDivision (MLGW)

3Modeling Earthquake Impact on Urban Lifeline Systems

direct economic loss models hasbeen achieved;

• Multiple earthquake sce-narios - Losses for differentpotential earthquakes are mod-eled, and a methodology is avail-able to incorporate these into aprobabilistic risk framework(Chang et al., forthcoming);

• Spatial modeling - AdvancedGIS capabilities have been imple-mented to refine the spatial reso-lution of analysis and moreeffectively use digital spatial data;

• Temporal modeling - Thetime dimension of loss has beenbetter captured through devel-oping a lifeline restorationmodel that incorporates post-disaster response parameters;

• Northridge data - Empiricaldata on business losses in theNorthridge earthquake havebeen used to refine direct andindirect loss models;

• Indirect loss modeling - Com-putable general equilibrium(CGE) approaches for estimatingindirect losses and business re-siliency effects have been devel-oped;

• Uncertainty - Uncertainties de-riving from both engineeringand economic models are takeninto account.

Figure 1 illustrates the frameworkfor implementing the loss estima-tion methodology for the MLGWwater delivery system. Individualearthquake scenarios are evaluatedfor events with different magni-tudes and locations. For each sce-nario earthquake, an integratedmodel (shaded box) is used to esti-mate damage and loss. Its core con-sists of an engineering modelpreviously developed by M.Shinozuka and associates to simu-late the expected damage and wa-ter service outage from majorearthquakes (Shinozuka et al.,1994). This model had been imple-mented in a simulation softwareprogram, Lifeline-W(II), for theMLGW case. Here, the Lifeline-W(II) model is expanded to a newmodel (LLW+E) that incorporatesa direct economic loss component.

The new model uses a MonteCarlo simulation approach toevaluate both physical damage and

Program 1: SeismicEvaluation and Retrofit ofLifeline Systems

• Loss EstimationMethodologies

• Memphis LifelineSystems Analysis

• Preliminary part ofmore comprehensivework for MCEER’sdemonstration projecton Memphis Lifelines

Program 2: Seismic Retrofitof Hospitals

• Cost-Benefit Studies

� Figure 1. Implementation Framework

4

economic loss. The model simu-lates damage to water systemcomponents, water leakage, andinitial water outage throughout thesystem. It then simulates how thesituation is improved on a weeklybasis as repairs are made. Direct eco-nomic loss from business interrup-tion is evaluated probabilistically atweekly intervals and summed overtime. This enables economic loss tobe estimated probabilistically in amanner consistent with engineeringdamage estimation. Direct loss esti-mates, water outage results, and in-dustry resiliency data are input intoa CGE model to calibrate and assessindirect economic loss.

Results can be summarized in“economic fragility curves” that in-dicate the exceedance probabilityof different levels of economic lossassociated with different levels ofearthquake severity (e.g., increasingearthquake magnitude, given afixed epicentral location). The eco-nomic fragility curves can then becombined with information on theoccurrence probabilities of earth-quakes of various severity levels.The ultimate result consists of anestimate of the expected annualloss — in terms of repair costs andeconomic output loss — from wa-ter disruption in future earth-quakes. Some of these conceptsand frameworks were developed byShinozuka and Eguchi (1997).

Note that the model is also ca-pable of evaluating the effects ofvarious loss reduction strategies.These include both pre-disaster miti-gation measures such as seismicallyupgrading vulnerable componentsor increasing network redundancy,as well as post-disaster mitigationssuch as implementing mutual aidrepair crew agreements or restoringdamaged facilities according to an

optimal restoration pattern. Themodel can therefore evaluate losses“with” and “without” mitigations toestimate the loss reduction benefitsof these actions.

Damage, Outage andRestoration

The MLGW water delivery systemis represented in the LLW+E modelby roughly 960 demand and supplynodes and 1,300 links. The modelestimates damage probabilisticallyusing a Monte Carlo simulation ap-proach (Shinozuka et al., 1994). Inparticular, the damageability of sys-tem components such as pipes isrepresented in terms of fragilitycurves that indicate the probabil-ity of failure for a given level ofground motion input, local soil con-dition, and component characteris-tics (e.g., pipe material anddiameter). Fragility curves forpumping stations were updatedusing results in Hwang et al. (1998).For each earthquake scenario, 100simulation cases of systemwidedamage were produced. Each simu-lation case represents a determin-istic damage pattern that is apossible outcome of the earth-quake; collectively, they representthe probabilistic damage outcomeof the event. Only damage fromground shaking is considered here.Other possible damage sourcessuch as soil liquefaction and elec-tric power disruption are ignored.For each simulation case of sce-nario damage, the model evaluatesthe loss of water service by under-taking a complex system f lowanalysis. The analysis translates pipedamage into water leakage andsolves for a new state of system equi-librium for the damaged state. Theratio of water flow in the damaged

“The restorationmodel specifiesa sequence ofrestorationbased onengineeringpriorities andobservations inthe Kobe andNorthridgedisasters.”


versus the intact state, evalu-ated at each of the demandnodes, provides the basic in-dicator of water outage thatis used in the economic lossmodel.

Figure 2 shows the initial(week 0) water outage re-sults for the M=7.0 scenario,before any restoration hastaken place. Results are av-eraged over the various simu-lated cases of the event. Inthe figure, node-level data onthe ratio of water flow in thedamaged versus intact con-ditions are spatially interpo-lated in GIS to yield anoutage surface. Dark shadings in-dicate areas of greatest water loss.Outage is more severe in the west-ern portion of the county.

Economic loss depends upon notonly the water outage that occursimmediately after the earthquake,but how and over what time periodwater service is restored to thecommunity. Restoration is modeledhere using a resource constraintapproach, which specifies the num-ber of repairs that can be made inany time period according to thenumber of repair personnel avail-able. Damage to large pipes re-quires more worker-days to repairthan damage to smaller pipes.Model parameters are based on asurvey of current lifeline restora-tion models and data from the Kobeearthquake (Chang et al., 1999). Inaddition, the restoration modelspecifies a sequence of restorationbased on engineering priorities andobservations in the Kobe andNorthridge disasters. Specifically,any damage to large pipes is re-stored first, in order to bring thetransmission “backbone” of the sys-tem online before repairing service

pipes. The restoration of smallerpipes then occurs in spatial se-quence, from census tracts with thelowest damage density (number ofpipe breaks per square kilometer)to those with the highest. Initialdamage patterns are updated on aweekly basis until repairs have beencompleted. At each weekly inter-val, system flow analysis is con-ducted on the updated network.(Note that alternative restorationpatterns can be simulated with ourmodel; for example, employmentlosses can be minimized by givingpriority to those sub-areas that uti-lize the lowest amounts of water,directly and indirectly, per worker(see Rose et al., 1997)).

Analysis Zones

The extent of economic loss de-pends not only on water outage,but also on how outage corre-sponds with the location of eco-nomic activity. Loss estimationmethodologies have typically usedcensus tracts, of which there are133 in Shelby County, as the unitof analysis. In contrast, the current

� Figure 2. Water Outage in Shelby County, M=7.0 Marked Tree Scenario (Week 0)

6

analysis defines approximately3,400 Analysis Zones using GISoverlay and intersection operationsintegrating two input datasets. Thefirst data set is Service Zones, a con-tiguous set of 971 polygons for thecounty, each representing the in-ferred service area of one node onthe water network. The ServiceZone polygons were defined byperforming a GIS Thiessen polygonoperation on the point distributionof water delivery nodes, implement-ing an assumption that a particularbusiness location will be served bythe closest node on the water de-livery network. The second datasetpertains to employment, distrib-uted into 515 Traffic Analysis Zone(TAZ) polygons. This informationwas provided by the Shelby CountyPlanning Department and is basedon U.S. Census Bureau data. Defin-ing Analysis Zones in this way rec-onciles point data on water outagewith polygon data on employment.It therefore allows implementationof the loss model at the same spa-tial resolution as that of the inputdata and maximizes use of informa-tion on the spatial variability of fac-tors contributing to loss.

Business Resiliency

The economic impact of the wa-ter loss to businesses is mitigated bytheir resiliency, or their ability towithstand temporary water disrup-tion. A 50 percent loss of water doesnot, in other words, necessarily re-duce economic output by 50 per-cent. Until recently, empiricalcalibration of resiliency has beenseverely limited by lack of appropri-ate data, and many loss estimationmethodologies use expert opiniondata for this purpose. In the previ-ous lifeline loss model developed in

Shinozuka et al., eds. (1997), resil-iency factors were calibrated on thebasis of survey data by K. Tierneyand colleagues on Memphis areabusinesses’ dependency on lifelines.Resiliency factors are defined as theremaining percentage output thatcould still be produced by a spe-cific industry in the event of totalwater outage. These data are lim-ited, however, because they are hy-pothetical — businesses had noprevious earthquake dislocationexperience on which to base theirresponses.

In the current study, new empiri-cal data from K. Tierney and col-leagues’ business survey of theNorthridge earthquake provide animportant source for more accu-rately calibrating business resiliency(Tierney, 1997; Tierney andDahlhamer, 1998). The principal dif-ficulty in using data from an actualdisaster in this way, however, is thatbusiness loss would have been in-fluenced by many sources of disrup-tion besides loss of water. Isolatingthe effects of water outage fromthose of building damage, electricpower disruption, transportationproblems, etc., is critical. A seconddifficulty consisted of limited samplesize. Although the Northridgedataset contained responses from1,110 businesses, less than one-fifthof them (207) had actually lost wa-ter service in Northridge. Moreover,only six businesses indicated thatloss of water was the main reasonfor their business closing tempo-rarily after the earthquake. Thesedata limitations were addressed bydividing the analysis into two steps— first looking at how disruptivewater outage was to businesses indifferent industries, and then evalu-ating how disruption from allsources impacted business activity.


Resiliency factors were developedfor 16 industries comprising the pri-vate sector economy and for threewater outage durations (less than1 week, 1 week, and 2 or moreweeks). Particularly relevant surveydata included information onwhether or not a business lost wa-ter, the duration of this outage, thedisruptiveness of this outage, thedisruptiveness of numerous othertypes of damage (e.g., building, trans-portation), whether or not the busi-ness closed for any length of time,and the most important reasons forthis closure.

Table 1 compares resiliency fac-tors developed from the Northridgedata and the Memphis business sur-vey, respectively, for selected indus-tries. Note that resiliency factorsrange from 0 (no resiliency) to 1(complete resiliency, or no depen-dency on water). The estimated fac-tors differ substantially, with theMemphis hypothetical resilien-cies consistently lower than theNorthridge actual resiliencies. Themost probable explanation is thatwithout having experienced a ma-jor earthquake, Memphis businessesunderestimated their own ability tocope with an emergency. Theanalysis reported in this paper useda mathematical average of the Mem-phis and Northridge resiliency fac-tors in the economic loss model.

Direct Economic Loss

Direct economic loss — definedhere as business interruption losscaused by water outage at the siteof production — results from a com-bination of outage pattern, its spa-tial coincidence with economicactivity, and business resiliency towater loss (for details, see Rose etal., 1997, and Chang et al., 1996). Akey innovation made here in mod-eling direct economic loss consistedof probabilistically simulating loss ina Monte Carlo approach rather thanestimating it deterministically. Spe-cifically, resiliency factors were ap-plied as a parameter indicating theprobability that a business or groupof businesses would close if it lostwater service. In this simulation,each specified industry (e.g., manu-facturing) in each Analysis Zone wasmodeled as either closed or operat-ing normally in each time period.(Partial closures were also allowedfor cases with partial water availabil-ity.) This probabilistic approach isimportant in that it enables consis-tent treatment of the engineeringand economic portions of the over-all model.

Indirect Economic Loss

Indirect economic loss is heredefined as business interruption

� Table 1. Survey-Based Economic Resiliency Factors for Water Outage (1 Week), Selected Industries

yrtsudnI

yevruSssenisuB

sihpmeM)retsasiDlacitehtopyH(

egdirhtroN)retsasiDlautcA(

noitcurtsnoC 03.0 46.0

gnirutcafunaM 71.0 05.0

edarTliateR 32.0 14.0

etatsElaeR/ecnarusnI/ecnaniF 12.0 43.0

secivreShtlaeH 81.0 42.0

8

deriving from interactions betweenbusinesses and industries throughchanges in input demands or out-put sales, including chain reactions.This study applies ComputableGeneral Equilibrium (CGE) meth-ods to assess indirect loss. CGEanalysis is the state-of-the-art in re-gional economic modeling, espe-cially for impact and policy analysis.It represents multi-market simula-tion models based on the simulta-neous optimizing behavior ofindividual consumers and firms,subject to economic account bal-ances and resource constraints (see,e.g., Shoven and Whalley, 1992). TheCGE formulation incorporatesmany of the best features of otherpopular model forms, but withoutmany of their limitations.

The basic CGE model representsan excellent framework for analyz-ing natural hazard impacts andpolicy responses. CGE models canbe finely disaggregated to distin-guish the various degrees of vulner-ability to hazards across sectors.The production functions are inclu-sive of all inputs, not just primaryfactors as in the case of many othereconomic models, which facilitatesidentification of materials short-ages. At the same time, CGE mod-els allow for the possibility of inputsubstitution, which mimics realworld responses beyond the veryshort-run in minimizing hazard im-pacts. They also allow for the sub-stitution of imported goods forregionally produced goods. CGEmodels are non-linear in form,thereby more closely reflecting realworld conditions, such as econo-mies of scale and non-linear dam-age functions. CGE models aremore capable of analyzing disjointchange than are model forms basedon time series data and which

therefore simply extrapolate thepast. They can also more readilyaccommodate engineering data ordata based on informed judgment.

Another set of CGE model advan-tages pertains to the important roleof prices and markets. Related tothis is explicit consideration of be-havioral response, and not justsimple optimization but also in-stances of bounded rationality. Thisapplies not only to consumptionbut also to mitigation and recoverybehavior.

Finally, CGE models are superiorto some other alternatives in mod-eling the role of lifeline infrastruc-ture services. A CGE model canplace a valuation on these services,even for public sector outputs thatare typically unpriced. This is morethan an academic exercise, since“shadow values” might serve as tem-porary prices to ration these ser-vices through the market ratherthan by administrative decree.

The methodology involvesrecalibrating the parameters ofthe CGE model for earthquakesimulations to yield direct businessdisruption losses consistent withempirical results of other MCEERresearchers, as referenced above.The parameter adjustments arelinked to specific real world ex-amples of business resiliency (e.g.,conservation, use of back-up sup-plies, increased substitutability).Different resiliency measures mayresult in the same direct output re-duction level, but have different im-plications for indirect impacts. Forexample, conservation lowers theeffective price of water, whichstimulates production if the priceof the final product is lowered ac-cordingly; however, it leads to a de-crease in direct and indirect inputsto providing water, which can have

“The analysis isone of the fewCGE simula-tions on anysubject that hasdistinguisheddirect andindirect effectsin detail.”


an economy-wide dampening ef-fect. On the other hand, substitu-tion of other inputs for water has acost-increasing effect, but stimu-lates direct and indirect demandsfor other inputs.

The production side of the CGEmodel used in this paper is com-posed of a multi-layered, or multi-tiered, constant elasticity ofsubstitution (CES) production func-tion for each sector. We explicitlyseparate water out as a major ag-gregate in the top tier of the pro-duction function. A summary ofadjustment types, linked to the pro-duction function layer and param-eters to which each relates and therecovery/reconstruction stage(time period) to which each is ap-plicable, is presented in Table 2. Theresiliency adjustments listed inTable 2 are incorporated into ouranalysis by altering the parameters,and, in one case, the variables in thesectoral production functions ofthe CGE model.

Results of the questionnaire sur-vey by K. Tierney and colleagues forthe Northridge earthquake areadapted to specify the resiliency towater disruption in the Memphiseconomy. Information on what ac-tions businesses took — if any —to remedy the loss of water pro-vided the basis for calibrating pa-rameters pertaining to adjustmentslisted in Table 2. For example, useof stored or boiled water, or “cre-ated supply” (presumably dug a wellor captured rain or riverine water)are considered back-up supplies.Use of bottled or hauled water rep-resents the substitution of other in-puts for pipeline water. Overall, theproportion of total explained resil-iency from the Memphis economywas apportioned to major adjust-ments in the following manner:

Conservation of water = 41.9%Increased substitutability = 47.7%Back-up supplies/costless = 8.1%Back-up supplies/cost-incurring = 2.3%

Total of all adjustments = 100%

� Table 2. Adjustments to Reduction in Water Availability to Business and Link to Formal Economic ProductionFunction Modeling

epyTtnemtsujdA retemaraP ytilibacilppA

retaWfonoitavresnoC.1 ygolonhceTretaW nuRgnoLotetaidemmI)gnisaerced(

stupnIsuoiraVfonoitavresnoC.2 ygolonhceTlareneG nuRgnoLotetaidemmI)tnatsnoc(

foytilibatutitsbuSdesaercnI.3retaWrofstupnIrehtO

yticitsalEnoitutitsbuSretaW nuRgnoLotnuRtrohSyreV)gnisaercni(

foytilibatutitsbuSdesaercnI.4otyticirtcelErofstupnIrehtO

retaWpmuP

yticirtcelEdnaretaWseiticitsalEnoitutitsbuS

nuRgnoLotetaidemmI

seilppuSpu-kcaB.5 rotcaF,ygolonhceTretaWserahS

Very Short RunotetaidemmI)tnatsnoc(

ecnatropmIretaW.6 1 ygolonhceTretaW nuRgnoLotetaidemmI)tnatsnoc(

esUyaD-fo-emiT.7 .A.N sruoh-ffO

ygolonhceTniegnahC.8 ygolonhceTllAottsoM )tnatsnoc(nuRgnoL

1 .retaweriuqertonodtahtnoitcudorpfostcepsaelbarapesotsrefeR

)gnisaercni(

10

Overall, the estimation of indirectlosses involves a multi-step proce-dure: (1) Extract the sectoral pro-duction functions from the CGEmodel and adjust parameters andvariables in them one at a time tomatch Chang’s direct loss estimates;(2) Reinsert the recalibrated sectoralproduction functions into the CGEmodel, reduce water supply to alevel consistent with Shinozuka andChang’s estimates, and compute to-tal regional losses; and (3) Subtractdirect losses from total losses to de-termine indirect losses.

FindingsInitial results from applying the

above methodology to the MLGW

water system have been encourag-ing and insightful. Figure 3 sum-marizes direct economic lossresults for the magnitude 6.5, 7.0,and 7.5 Marked Tree earthquakes,respectively. Loss values are aver-aged over 100 Monte Carlo simula-tions. Table 3 provides moredetailed results for these threeevents.

Figure 3 and Table 3 show that, asexpected, losses increase exponen-tially with magnitude since thethree events have identical epicen-tral locations. The table providesadditional insights. First, loss stan-dard deviations are very high rela-tive to mean values, indicatingsubstantial variation from one simu-lation case to the next for a givenscenario. Recall that this variationderives from both engineering andeconomic model uncertainty. Inthe case of the M=7.5 event, resultsare limited by the outcome that inthe first four weeks of system re-pairs, damage is sufficiently greatthat the system flow model fails tosolve, indicating that water outageis likely to be complete throughoutthe county. In this case, maximumeconomic loss (defined by resil-iency parameters) is assumed. Asnoted in Table 3, damage (numberof pipe breaks) and restorationtimes also increase exponentiallywith magnitude.

edutingaM 5.6 0.7 5.7

AGPegarevA 1 g41.0 g02.0 g92.0

of Pipe BreaksrebmuNegarevA 031 856 294,4

riapeRotskeeW 1 3-2 31-21 2

ssoLcimonocEtceriD 3 naeM- .lim24$ .lim631$ .lim214,2$

noitaiveDdradnat- S .lim36$ .lim99$ a/n 4

krowtennosknilllarevodegarevA14-1skeewniegatuoretawetelpmoC2

smrettuptuossorgnI3elbaliavatoN4

� Table 3. Loss Results from Monte Carlo Simulation, Marked Tree Earthquakes

� Figure 3. Mean Direct Economic Loss, Earthquakes at Marked Tree


However, closer inspection of thevariability in the individual simula-tion results showed that simulationcases with high dollar loss oftenhad similar numbers of pipe breaksas those with low loss. For theM=6.5 and 7.0 events, losses werepoorly correlated with number ofpipe breaks but highly correlatedwith the occurrence of pumpingstation failure(s). This insight sug-gests that in the MLGW case, pump-ing station retrofits would be moreeffective at reducing losses in mod-erate earthquakes than wholesalepipe upgrades. On the other hand,M=7.5 event losses appear to bedominated by the overwhelmingnumber of pipe breaks. Even afterpumping stations are repaired inthe initial week, system-wide failurepersists until three more weeks’worth of pipe repairs have beencompleted.

Results on indirect economic lossare summarized in Table 4. The ini-tial indirect loss results shown hereare based on one scenario charac-terized by a 7.0 magnitude earth-quake (average damage and directloss over 100 simulations), a one-week outage, and resiliency adjust-ments involving only thesubstitution of other inputs forwater utility services.

In Table 4, the sector labels on theleft-hand side refer to the economicproducing units of the MemphisCGE model. Direct water disrup-tion for each sector is presented incolumn 1, and sums to a 20.6 per-cent decline in water available inWeek 1. As noted in column 2, thisconstraint is binding on all but twosectors as the general equilibriumadjustments work themselves out.As discussed further below, nega-tive indirect effects on Construc-tion and Education are so great as

to reduce water demand even be-low the post-earthquake availabil-ity levels. Baseline output ispresented in Column 3 and reflectsthe relative prominence of sectorsin Shelby County (Memphisproper) economy and serves as areference point for the impact simu-lations. Note that the Water sector(11) gross output represents only0.113 percent of the regionaleconomy.

Column 4 presents the directoutput losses that are estimated inthe CGE model before any resil-iency adjustment. Chang’s esti-mates of direct output losses, whichincorporate the extent of resiliency,are presented in Column 5. Ourdirect loss estimates are based oninput substitution elasticities (thelowest possible values that yieldedan equilibrium solution) presentedin the next to the last column inTable 4. Note that the CGE directloss estimates exceed those ofChang in every sector because theyomit all resiliency options exceptnormal input substitution. The fi-nal column shows the elasticitiesnecessary to incorporate resiliencymeasures for the CGE model resultsto be consistent with the Chang es-timates.

Our estimates of the indirect andtotal regional economic impacts ofthe water lifeline disruption arepresented in Columns 6 and 7.Overall, they yield a 5.1 percent in-direct reduction in regional grossoutput and an 19.0 percent totalreduction in regional gross outputfor the week. The former repre-sents $49.3 million ($2,563 million÷ 52) and the latter $184.6 million.

Some interesting aspects of indi-rect losses are indicated by Table 4.First, they are slightly more than

“Priorities forcontinuingwork includesensitivityanalysis,modeling otherNew MadridSeismic Zoneearthquakes,and evaluatingthe effects ofpre- and post-disaster lossreductionmeasures.”

12

Wat

er I

nput

Out

put

Out

put

Cha

nge

from

Wat

er O

utag

eEl

asti

city

(�

0)Se

ctor

Dir

ect

Dis

rupt

ion

Tota

lU

nuse

dB

asel

ine

(106 1

995)

Our

Dir

ect

w/o

Adj

ust.

2C

hang

'sD

irec

tIn

dire

ctTo

tal1

Wit

hout

Adj

ustm

ent

Wit

hA

djus

tmen

t

1A

gric

ultu

re-2

6.7%

-26.

7%12

8-2

1.0%

-11.

6%-1

1.5%

-23.

1%0.

0130

00.

0132

5

2M

inin

g-1

6.5%

-16.

5%20

-9.1

%-8

.5%

0.2%

-8.7

%0.

0124

00.

0124

1

3C

onst

ruct

ion

-25.

2%-3

3.0%

2782

-18.

3%-8

.3%

-24.

6%-3

2.9%

0.01

300

0.01

322

4Fo

od P

roce

ssin

g-1

9.5%

-19.

5%19

99-1

7.0%

-15.

8%-0

.3%

-16.

1%0.

0100

60.

0100

8

5M

anuf

actu

ring

-20.

1%-2

0.1%

8151

-14.

5%-1

4.0%

-3.0

%-1

7.0%

0.01

200

0.01

201

6Pe

trol

eum

Ref

inin

g-1

9.5%

-19.

5%51

1-1

7.4%

-15.

8%-0

.7%

-16.

5%0.

0094

50.

0094

8

7Tr

ansp

orta

tion

-10.

1%-1

0.1%

6235

-9.1

%-7

.8%

0.2%

-7.6

%0.

0115

00.

0115

3

8C

omm

unic

atio

n-2

2.2%

-22.

2%65

4-1

4.7%

-10.

8%-5

.5%

-16.

3%0.

0130

00.

0130

8

9El

ectr

icity

Ser

vice

s-2

2.2%

-22.

2%53

0-1

8.5%

-10.

8%-9

.5%

-20.

3%0.

0120

00.

0122

8

10G

as D

istr

ibut

ion

-22.

2%-2

2.2%

21-1

2.9%

-10.

8%-3

.7%

-14.

5%0.

0130

00.

0130

4

11W

ater

& S

anita

tion

-22.

2%-2

2.2%

57-1

9.1%

-10.

8%-9

.2%

-20.

0%0.

0130

00.

0134

3

12W

hole

sale

Tra

de-1

9.7%

-19.

7%45

81-1

2.7%

-12.

7%-1

.7%

-14.

4%0.

0130

00.

0130

0

13R

etai

l Tra

de-2

8.7%

-28.

7%41

48-2

3.1%

-14.

1%-1

0.1%

-24.

2%0.

0130

00.

0132

4

14Fi

nanc

e, In

sura

nce

& R

eal E

stat

e-2

4.7%

-24.

7%69

31-1

8.8%

-16.

5%-3

.2%

-19.

7%0.

0130

00.

0130

6

15Pe

rson

al S

ervi

ces

-23.

9%-2

3.9%

1546

-19.

0%-1

3.9%

-6.8

%-2

0.7%

0.01

300

0.01

316

16B

usin

ess

& P

rofe

ssio

nal S

ervi

ces

-22.

2%-2

2.2%

3917

-14.

9%-1

4.1%

-3.0

%-1

7.1%

0.01

300

0.01

302

17En

tert

ainm

ent S

ervi

ces

-30.

8%-3

0.8%

293

-25.

5%-1

5.3%

-11.

8%-2

7.1%

0.01

300

0.01

329

18H

ealth

Ser

vice

s-1

8.9%

-18.

9%31

43-2

2.9%

-22.

6%6.

8%-1

5.8%

0.01

050

0.01

051

19Ed

ucat

ion

-29.

1%-5

4.0%

1324

-20.

7%-1

6.1%

-37.

9%-5

4.0%

0.01

300

0.01

308

20G

over

nmen

t-2

6.9%

-26.

9%32

91-2

1.6%

-15.

7%-9

.1%

-24.

8%0.

0120

00.

0121

5

Tota

l-2

0.6%

-20.

8%50

263

-16.

8%-1

3.9%

-5.1

%-1

9.0%

1 Follo

win

g C

GE

sim

ulat

ion

2 From

par

tial e

quili

briu

m a

naly

sis

� T

able

4.

Part

ial a

nd G

ener

al E

quili

briu

m C

hang

es in

Out

put F

ollo

win

g a

Wat

er O

utag

e in

She

lby

Coun

ty (

1-w

eek

outa

ge fo

llow

ing

a 7.

0 m

agni

tude

ear

thqu

ake)


one-third the size of direct losses.In the context of an input-output(I-O) model, this would be a multi-plier of about 1.37. The ShelbyCounty overall output multiplier islarger than this, but the CGE modelincorporates many other factorsthat mute the uni-directional andlinear nature of the pure interde-pendence effect of the I-O model.For example, it is able to captureprice declines due to decreased in-termediate goods demand pressure,various substitutions aside fromthose relating to water, and variousincome, substitution and spendingconsiderations on the consumerside. Hence, we note that the twosectors suffering the largest indirectdecline are Education (not a neces-sity in the immediate aftermath ofan earthquake) and Construction (aleading indicator of economic ac-tivity). Note that once our modelfactors in post-earthquake recoveryand reconstruction, this would off-set the Construction decline. Inaddition, several sectors are char-acterized by positive or minimallynegative indirect effects, most no-tably basic necessities, such as FoodProcessing, Petroleum Refining,Transportation, and Health Services.

Note that we have modeled onlyone resiliency measure, albeit oneof the most important ones—in-creased substitutability of other in-puts for water utility services. Asdiscussed earlier, different types ofresiliency measures would generatedifferent types of indirect impacts.However, preliminary simulationsindicate the differences are smallsince the vast majority of indirectlosses stem from the direct outputreduction levels of producing sec-tors of the Memphis economy,which are basically the same no

matter what types of adjustmentsare modeled.

Overall, our results appear to bereasonable for the economy as awhole, for individual sectors, andfor individual impact stages (directand indirect). The analysis is oneof the few CGE simulations on anysubject that has distinguished directand indirect effects in detail. More-over, we have developed a method-ology that enables CGE users torecalibrate their models to makeuse of empirical data on individualparameters and direct impacts, suchas those associated with responsesto water lifeline disruptions follow-ing an earthquake.

ConclusionsThe MLGW water system appli-

cation demonstrates numerousmethodological advances in earth-quake loss estimation for lifelinesystems, ranging from integratedprobabilistic modeling of engi-neering and economic loss, to re-finements based on Northridgedata, to development of new Com-putable General Equilibrium mod-els for indirect loss evaluation.Priorities for continuing work in-clude sensitivity analysis, modelingother New Madrid Seismic Zoneearthquakes, and evaluating theeffects of pre- and post-disasterloss reduction measures. Ulti-mately, losses will be summarizedin the form of economic fragilitycurves that indicate the likelihoodof exceeding different levels of eco-nomic loss, with and without lossreduction actions. These results willbe combined with probabilistichazard information to estimate theexpected benefits of mitigationswhich can then be compared with

“Ultimately,losses will besummarized inthe form ofeconomicfragility curvesthat indicatethe likelihoodof exceedingdifferent levelsof economicloss, with andwithout lossreductionactions.”

14

their costs. This information canassist lifeline agencies in determin-ing which mitigation options andhow much mitigation to undertake.The methodological approach can

be implemented for other lifelines,such as electric power systems, andfor other seismically vulnerable ur-ban areas.

Chang, S.E., Seligson, H.A. and Eguchi, R.T., 1996, Estimation of the Economic Impactof Multiple Lifeline Disruption: Memphis Light, Gas and Water Division Case Study,Technical Report NCEER-96-0011, National Center for Earthquake EngineeringResearch, Buffalo, New York.

Chang, S.E., Shinozuka, M. and Moore, J.E. II, “Probabilistic Earthquake Scenarios:Extending Risk Analysis Methodologies to Spatially Distributed Systems,” EarthquakeSpectra, forthcoming.

Chang, S.E., Shinozuka, M. and Svekla, W., 1999, “Modeling Post-Disaster Urban LifelineRestoration,” in W.M. Elliott and P. McDonough, eds., Optimizing Post-EarthquakeLifeline System Reliability: Proceedings of the 5th U.S. Conference on LifelineEarthquake Engineering, ASCE Technical Council on Lifeline Earthquake EngineeringMonograph No. 16, pp. 602-611.

Earthquake Spectra, 1997, Special Theme Issue on Earthquake Loss Estimation, Vol.13, No. 4.

Hwang, H.H.M., Lin, H. and Shinozuka, M., 1998, “Seismic Performance Assessment of WaterDelivery Systems,” Journal of Infrastructure Systems, Vol. 4, No. 3, pp. 118-125.

National Research Council, 1999, The Impacts of Natural Disasters: A Framework forLoss Estimation, National Academy Press, Washington, D.C.

Rose, A., Benavides, J., Chang, S., Szczesniak, P. and Lim, D., 1997, “The RegionalEconomic Impact of an Earthquake: Direct and Indirect Effects of Electricity LifelineDisruptions,” Journal of Regional Science, Vol. 37, pp. 437-58.

Shinozuka, M. and Eguchi, R., 1997, Seismic Risk Analysis of Liquid Fuel Systems: AConceptual and Procedural Framework for Guidelines Development, NationalInstitute of Standards and Technology Report No. GCR 97-719, Gaithersburg,Maryland.

Shinozuka, M., Rose, A. and Eguchi, R. (eds)., 1998, Engineering and SocioeconomicImpacts of Earthquakes: An Analysis of Electricity Lifeline Disruptions in the NewMadrid Area, Monograph No. 2, Multidisciplinary Center for Earthquake EngineeringResearch, Buffalo, New York.

Shinozuka, M., Tanaka, S. and Koiwa, H., 1994, “Interaction of Lifeline Systems underEarthquake Conditions,” Proceedings of the 2nd China-U.S.-Japan TrilateralSymposium on Lifeline Earthquake Engineering, pp. 43-52.

Shoven, J.B. and Whalley, J., 1992, Applying General Equilibrium, Cambridge, U.K.:Cambridge University Press.

Tierney, K.J., 1997, “Business Impacts of the Northridge Earthquake,” Journal ofContingencies and Crisis Management, Vol. 5, pp. 87-97.

Tierney, K.J., and Dahlhamer, J.M., 1998, “Business Disruption, Preparedness, andRecovery: Lessons from the Northridge Earthquake,” pp. 171-178 in Proceedings ofthe NEHRP Conference and Workshop on Research on the Northridge, CaliforniaEarthquake of January 17, 1994, Vol. IV, Richmond, California: California Universitiesfor Research in Earthquake Engineering.

References

15

Research Objectives

Fragility curves are functions which represent the probability that agiven structure’s response to various seismic (conditions) excitationsexceeds performance limit states. As such, fragility curves are a measureof performance in probabilistic terms. The major objectives of MCEERresearch on fragility information are: (1) developing efficient and accu-rate methods to calculate fragility curves for structural and nonstructuralcomponents and for global assemblies (systems); (2) establishing fragilityinformation as a base for performance-based design, cost and loss assess-ment, and decision policies; (3) developing optimal strategies for seismicrehabilitation based on cost-benefit analysis; and (4) establishing rationalperformance limit states based on interaction between engineering andsocioeconomic issues.

This research focuses on methods for efficiently and accurately deter-mining fragility information for structural and nonstructural compo-

nents, and for global assemblies (systems). Fragility curves, giving the prob-ability of exceeding a specific limit state as a function of ground motionintensity, are commonly used to present the fragility information. Theycan be developed either for a specific system or component, or for a classof systems or components. Fragility curves can also be used to comparedifferent seismic rehabilitation techniques and to optimize the seismicdesign of structures.

There are two main approaches for generating fragility curves. The firstapproach is based on damage data obtained from field observations afteran earthquake or from experiments. The second approach is based onnumerical analysis of the structure, either through detailed time-historyanalysis or through simplified methods. There is no universally applicablebest method for calculating fragility curves. Different methods may bepreferred depending on the circumstances. For example, the field datamethod is particularly useful for characterizing the seismic performanceof a collection of similar structures. The numerical analysis method isuseful when fragility curves need to be developed for a particular struc-tural system with well-defined limit states.

Development of FragilityInformation for Structures andNonstructural Components

by Masanobu Shinozuka, Mircea Grigoriu (Coordinating Author), Anthony R. Ingraffea, SarahL. Billington, Peter Feenstra, Tsu T. Soong, Andrei M. Reinhorn and Emmanuel Maragakis


Federal HighwayAdministration

Masanobu Shinozuka,Professor, Department ofCivil Engineering,University of SouthernCalifornia, H. Kim, VisitingProfessor, Mokpo NationalUniversity, and T. Uzawaand T. Ueda, VisitingResearchers, Taisei Corp.

Maria Q. Feng, AssociateProfessor, Department ofCivil and EnvironmentalEngineering, University ofCalifornia, Irvine

Mircea Grigoriu, Professor,Anthony R. Ingraffea,Professor, Sarah L.Billington, AssistantProfessor, Peter Feenstra,Senior Research Associate,Christopher Roth, EhabMostafa and T. Han,Research Assistants, Schoolof Civil and EnvironmentalEngineering, CornellUniversity

Research Teamcontinued on page 2

16

In dealing with field data, two-parameter lognormal distributionfunctions are used to represent fra-gility curves with all the analytical,interpretational and other advan-tages associated with this tradi-tional form. In particular, when thisform is used, the randomness vs. un-certainty issue can be addressed ina manner as explored in early stud-ies dealing with nuclear powerplant equipment (e.g., Holman etal. 1987).

Shinozuka et al. (1999a) furtherinvestigated this issue in relation tofragility curves developed for struc-tural types categorized as in ATC-13 (1985). In this investigation, theyresurrected the interpretation thateach structural category representsa loose collection of sub-categoriesof structural types that are moresimilar in structural mechanics be-havior, and that the fragility curvefor each category is a lognormal dis-tribution function whose medianvalue has a probability distribution.The fragility curve with a realiza-tion of the median value representsthe randomness of the fragility fora corresponding sub-category ofthe structural type.

In this sense, the probability dis-tribution of the median is inter-preted as arising from theuncertainty. The fragility curve as-sociated with a specific medianvalue represents the randomness

of the fragility of a correspondingsub-category of the structural type.

In addition, an analytical expres-sion was derived for the log-stan-dard deviation of the fragility curvefor a structural category in termsof the median values and log-stan-dard deviations of constituent sub-categories of structural types. Thisprovided, for the first time, a clear,quantitative insight into this diffi-cult issue of randomness vs. uncer-tainty involved in the fragility curve.

In connection with damage datato support the development of fra-gility information for nonstructuralcomponents, a comprehensive non-structural damage database hasbeen developed (Kao et al., 1999).It contains nearly 3,000 entries en-compassing more than 50 earth-quakes from the 1964 Alaskaearthquake to the present. Infor-mation from various publications,including books, reports and peri-odicals, was gathered and re-corded, including a description ofdamage of the nonstructural com-ponent, information about thebuilding where the damage oc-curred, and strong ground motionrecords, when available.

The database is presented as a liv-ing document, which will continueto be updated and evolve. TheMCEER nonstructural componentsdatabase, in Access format, can bedownloaded from the publications

Fragility information can be used by design engineers, re-searchers, reliability experts, insurance experts and adminis-trators of critical systems such as hospitals and highwaynetworks. The information can be used to analyze, evaluateand improve the seismic performance of both structural andnonstructural systems. Optimal design and rehabilitation strat-egies can be identified. The techniques and computation toolsdeveloped as part of this research can also be used in under-graduate and graduate courses in earthquake engineering.

Tsu T. Soong, Samuel P.Capen Professor ofEngineering Science,Andrei M. Reinhorn,Professor, Raul Barron-Corvera, ResearchAssistant, and Woon HuiChong, Graduate Student,Department of Civil,Structural andEnvironmentalEngineering, University atBuffalo, State Universityof New York

Emmanuel Maragakis,Professor and Chair,Raj Siddharthan,Professor and RonaldMeis, GraduateResearcher, Department ofCivil Engineering,University of Nevada -Reno

17Fragility Information for Structures and Nonstructural Components

section of MCEER’s web site (http://mceer.buffalo.edu/publications/re-ports/docs/99-0014/default.html).

Field andExperimental Data

Suppose that damage data is avail-able from n earthquakes or experi-ments with characteristic peakground accelerations a

i, i=1, 2, 3,

…,n, with the convention a1 ≤ a

2 ≤

a3 ≤ ... < a

n. Let n

i > 1 be the num-

ber of structures of a certain classsubjected to earthquake i of PGAa

i and m

i ≤ n

i be the number of

these structures that exceeded aspecified limit state. The fraction, f

i

= mi / n

i, is an estimate of the prob-

ability of failure, or the fragility, ofthe class of structures consideredif subjected to earthquakes of peakground acceleration a

i. The fragil-

ity curve for this class of structurescan be obtained by fitting a distri-bution function to the data set {a

i,

fi}. Hence, fragility curves are in-

creasing functions of the peakground acceleration and take val-ues in [0,1].

Field data

This method of generating fragil-ity curves utilizes damage data col-lected in the field after earthquakesat different sites. The ground mo-tion intensity is determined at thelocation of each reported structureand the method outlined above canbe used to plot the fragility curves.However, more advanced statisticalmethods are usually employed todetermine the parameters of thefragility curve so that the goodnessof fit can be tested and statisticalconfidence intervals estimated.The empirical fragility curves, con-structed from the field damage

observation, play an indispensablerole in the fragility curve develop-ment study, because only they canbe used to calibrate the fragilitycurves developed analytically oreven experimentally if done underlaboratory conditions. The damagedata obtained from the 1999Marmara, Turkey and Chi-Chi, Tai-wan earthquakes are extremelyvaluable for the development of fra-gility curves in this sense.

In addition, as shown byShinozuka et al. (1999b), empiricalfragility curves for groups of struc-tures classified in terms of specificattributes (e.g., severity of skew-ness in the case of bridges) can bedeveloped easily on the basis ofthe field damage data to demon-strate the effect of the attribute dif-ference on fragility. Whileanalytical avenues are open for thistype of study, they would requirecomputational efforts of inhibitivedimension. The empirical distribu-tions would help develop practi-cal approximations based onanalysis under these circum-stances.

Fragility Curves for Caltrans’Expressway Bridges in Los AngelesCounty, California(Before Post-earthquake Retrofit)

Damage data was collected afterthe Northridge earthquake for1,998 of Caltrans’ expresswaybridges in Los Angeles County, Cali-fornia. Fragility curves were gener-ated by Shinozuka et al. (1999b) forfour damage states identified as “atleast minor,” “at least moderate,” “atleast major” and “collapse.” A two-parameter lognormal distributionfunction was calibrated to thesedamage data to develop fragilitycurves. The parameters of the dis-tribution that must be estimated are

CaltransHanshin Expressway Public

Corporation, HEPCGayle Johnson, EQE

InternationalMohammed Ettouney,

Weidlinger Associates

18

the median and the log-standarddeviation. The peak ground accel-eration (PGA) at each bridge sitewas used to represent the intensityof the seismic ground motion. Twomethods were developed to calcu-late fragility curves and establishconfidence intervals on thesecurves. Method 1 estimates a dif-ferent median and log-standard de-viation for each damage state andgroup of bridges. In Method 2, adifferent median is estimated foreach damage state and group of

bridges, but a single log-standarddeviation is estimated for all of thedistributions to prevent these fra-gility curves from intersectingeach other. Figures 1 and 2 showfragility curves generated for thefour damage states by Methods 1and 2, respectively. A similar studywas done by Basöz et al. (1998).

Fragility Curves for ReinforcedConcrete Bridge Columns alongKobe and Ikeda Route, Japan(Before Post-earthquake Retrofit)

A sample of 770 single-supportreinforced concrete columns alonga 40 km length of viaduct was con-sidered. All of the columns havesimilar geometry and reinforce-ment. Figure 3 defines the dam-age states considered for thecolumns. Figure 4 shows the fra-gility curves generated by Method1 by Shinozuka et al. (1999a,1999b).

Experimental data

The same approach used for thefield data method may be appliedto damage data obtained from ex-periments. In this case, the rangeof earthquake intensities and thestructure type can be controlled asrequired. However, experimentscan be expensive and the amountof damage data available will be lim-ited by the number of experimentsthat can be carried out. The follow-ing three examples illustrate the de-velopment of fragility curves usingthis method:

Sliding Fragility Curves ofUnrestrained Equipment inCritical Facilities

Seismic design of buildings hasbeen well developed and is beingcontinually updated and improved.Nonstructural components, such as

� Figure 2. Fragility curves for Caltrans’ bridges (Method 2)

> MINOR ( = 0.83g, = 0.82)C0 0�

> MODERATE ( = 1.07g, = 0.82)C0 0�

> MAJOR ( = 1.76g, = 0.82)C0 0�

COLLAPSE ( = 3.96g, = 0.82)C0 0�

1.0

0.8

0.6

0.4

0.2

0.0

0.0 0.2 0.4 0.6 0.8 1.0

PGA (G)

Pro

bab

ility

of

Exc

eed

ing

a D

amag

e S

tate

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.2 0.4 0.6 0.8 1.0

PGA (g)

> MINOR ( = 0.85g, = 0.84)C0 0�

> MODERATE ( = 0.96g, = 0.72)C0 0�

> MAJOR ( = 1.35g, = 0.65)C0 0�

COLLAPSE ( = 2.74g, = 0.67)C0 0�

Pro

bab

ility

of

Exc

eed

ing

a D

amag

e S

tate

� Figure 1. Fragility curves for Caltrans’ bridges (Method 1)


Dam

age

Mod

eA

sA

BC

DR

emar

ks

1. B

endi

ng d

amag

e at

grou

nd le

vel

Dam

age

thro

ugh

enti

re c

ross

sec

tion

Dam

age

mai

nly

attw

o op

posi

te s

ides

Dam

age

mai

nly

at o

nesi

deL

ight

cra

ckin

g an

dP

arti

al S

palli

ngN

o D

amag

eT

his

mod

eun

ltim

atel

ypr

oduc

es b

uckl

ing

of r

ebar

s, s

palli

ngan

d cr

ushi

ng o

f co

reco

ncre

te

2. C

ombi

ned

bend

ing

&sh

ear

dam

age

at g

roun

dle

vel

Inte

rnal

Dam

age

Dam

age

at t

wo

side

sD

amag

e m

ainl

y at

one

side

Lig

ht c

rack

ing

and

Par

tial

Spa

lling

No

dam

age

Ben

ding

and

she

arcr

acks

pro

gres

sw

ith

mor

e w

ide-

spre

ad s

palli

ng t

han

mod

el a

nd h

oops

deta

ched

fro

man

chor

age

3. C

ombi

ned

bend

ing

&sh

ear

dam

age

at t

he le

vel

of r

educ

tion

of

long

itud

inal

reba

rs

Inte

rnal

Dam

age

Inte

rnal

dam

age

Dam

age

mai

nly

at o

nesi

deP

arti

al d

amag

eN

o da

mag

eD

amag

e an

dco

llaps

e ar

eob

serv

ed a

t ab

out

the

loca

tion

(typ

ical

ly 4

-5m

abov

e gr

ound

) o

fre

duct

ion

oflo

ngit

udin

al r

ebar

s,ac

com

pany

ing

buck

ling

of r

ebar

san

d de

tach

ed h

oops

.

4. S

hear

dam

age

at g

roun

dle

vel

Dam

age

thro

ugh

enti

re c

ross

-sec

tion

Dam

age

thro

ugh

colu

mn

Par

tial

Dam

age

Lig

ht c

rack

ing*

No

Dam

age

Col

umns

wit

h lo

was

pect

rat

ion

shea

red

at 4

50 ang

le

* No

desc

ript

ion

prov

ided

in th

e or

igin

al

Dam

age

Stat

e

� F

igur

e 3.

Des

crip

tion

of s

tate

s of

dam

age

for

Has

hin

Expr

essw

ay C

orpo

ratio

n’s

brid

ge c

olum

ns

20

piping systems housed in buildings,are rarely designed with the samedegree of consideration. As a result,buildings that remain structurallysound after a strong earthquakeoften lose their operational capa-bilities due to damage to their non-structural components. In certainsituations, damage to nonstructuralcomponents can pose a greaterthreat to safety than damage to struc-tural components. The cost of dam-age to nonstructural componentscan also exceed the cost of struc-tural damage. For example, of thetotal damage to one building duringthe San Fernando earthquake in

1971, over 98% was nonstructural.Moreover, costly damage to non-structural components could occurin earthquakes of intensities lowenough to cause little or no struc-tural damage. This research devel-ops fragility curves for free-standingrigid equipment based on experi-mental data. To achieve this goal,the sliding motion of a rigid blockagainst the surface of a raised floorwas tested on a shaking table usingfive randomly chosen earthquaketime histories. Both horizontal andvertical accelerations were consid-ered in these experiments. Fivehorizontal peak ground accelera-tions (HPGA) were considered,namely 0.3 g, 0.4 g, 0.5 g, 0.6 g, and0.7 g. Four different scale factorswere used to represent the verticalpeak ground accelerations (VPGA)in terms of HPGA: 0, 1/4, 1/3, and1/2. Horizontal and vertical accel-eration measurements using accel-erometers were made at severallocations of the shaking table: theraised floor, and the free standingrigid block. The horizontal displace-ments of the block were measuredby the temposonic displacementtransducers as well as two perma-nent markers attached to the leftand right side on the surface facingthe sliding direction. Eight differ-ent relative displacement failure

� Figure 5. Experimental fragility curves for failure threshold = 1 inch

vpga/hpga=0vpga/hpga=1/4vpga/hpga=1/3vpga/hpga=1/2

1.2

1

0.8

0.6

0.2

0

Fra

gili

ty

0.4

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8HPGA. g

>C (minor) C0 = 0.47g, ζ0 = 0.59

>B (moderate) C0 = 0.69g, ζ0 = 0.45

>A (major) C0 = 0.80g, ζ0 = 0.43

1

0.9

0.8

0.7

0.6

0.4

0.3

0.2

0.1

0

0.5

0.2 0.4 0.6 0.8 1.2 1.4

PGA (g)

Pro

bab

ility

of

Exc

eed

ing

a D

amag

e S

tate

KOBE(RC- Single Column)

10

� Figure 4. Fragility curves for HEPC’s bridge columns (Method 1)


thresholds between 0.1 inch and3 inches were considered, and fra-gility curves were developed foreach of these thresholds. Figure 5shows the experimental fragilitycurves for failure threshold of oneinch for the four different ratiosbetween vertical and horizontalpeak ground accelerations (Chonget al., 2000).

Damage Assessment Curves forBuried Pipe Joints

Pipelines transporting water, gas,or volatile fuels are part of the in-frastructure “lifeline” system and arecritical to the viability and safety ofcommunities. Disruption to theselifelines can have disastrous resultssuch as the release of natural gasand flammable fuels, or restrictionof water required for fire-fighting.Pipelines have been shown to bevulnerable to damage and failuredue to seismic motions. The objec-tive of this study is to develop riskassessment charts for pipe jointsbased on experimental data. Twenty-two separate specimens of variouspipe materials and various jointtypes have been tested under cyclicaxial loading. The loading processinvolved the fabrication of a self-con-tained loading frame and the use ofan MTS 450 kips actuator. Severaldifferent failure modes, such as buck-ling and fracture of the barrel andspigot end, were observed.

Two forms of risk assessmentcharts have been developed. Thefirst form compares the pipe forcecapacity with a force level that maybe imposed on the pipe due to seis-mic motion. Two conditions mustbe satisfied for failure to occur: thelevel of force from seismic motionmust be greater than the pipe forcecapacity, and the level of force thatis able to be transferred from the

soil to the pipe surface by frictionmust be greater than the pipe forcecapacity. Given the pipe capacity,the two conditions can be ex-pressed in terms of earthquake andsoil parameters such as predomi-nant period of the earthquake (T

p ),

peak particle velocity (Vp ) and the

wave propagation velocity (c). Fig-ure 6 is a typical risk assessmentplot for one of the joints. The sec-ond form of risk assessment chartconsists of two separate plots defin-ing two separate considerations asshown in Figures 7 and 8. Figure 7

Pipeline Seismic Response Diagram

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

100 1000 10000c (ft/sec)

Tp

(se

c)

0.000.050.100.150.200.250.300.350.400.450.50

Vp

(ft

/sec

)

soft soil medium soil stiff soil

3

1

2 4

fail

no fail

no fail

no fail

AB

� Figure 6. Pipeline seismic response diagram

� Figure 7. Pipe joint capacity plot

[A] Pipe Joint Capacity Risk Plot

0

1

2

10 100 1000 10000 100000AEFj

curves of soil strain values: Vp/c

fail

no fail0.

001

0.00

05

0.00

01

0.00

005

0.00

5

Fai

lure

Lev

el

22

shows the joint force capacity fordifferent curves of imposed soilstrains, while Figure 8 is a plot ofthe friction transfer force for differ-ent curves of the earthquake wavelengths. Both plots must have a “fail”condition for the joint to be classi-fied as a “probable fail” condition.These plots are applicable for anypipe material and configuration aswell as any site and seismic condi-tions, and are not specific to anysingle scenario.

Numerical AnalysisNumerical analysis can be used

to generate fragility curves for struc-tures for which no earthquake fielddata exists, or for which experimen-tation would be prohibitively expen-sive. The analyses can be repeatedfor different ground motions andstructural configurations at relativelylittle cost. However, the results de-pend on the ground motion andstructural models used. Two differ-ent levels of numerical analysis havebeen considered: the simplified

capacity spectrum method, andmore detailed time-history analysis.

Several material models have beenconsidered for time-history analysis.Two enhanced constitutive modelshave been developed and imple-mented in DIANA to improve thecharacterization of the response ofconfined concrete and reinforce-ment under cyclic loads reinforce-ment (Kwan and Billington, 1999a,1999b). Two benchmark studies arebeing conducted to evaluate theaccuracy of DIANA in predictingresults of shake table experiments.The first benchmark study is a simu-lation of shake-table experiments ofa lightly reinforced concrete frame,and the second is an internationalbenchmark competition to simulatethe seismic response of structuralconcrete shear walls (Report I,1999).

Spectral Capacity Method forFragility Evaluation

The spectral capacity method(SCM) is a simplified method thatestimates the response of a struc-ture from spectrum demand andspectral capacity curves (Barronand Reinhorn, 2000). The spectrumdemand curve represents theground motion and is typically de-rived from the elastic accelerationresponse spectra of the motion.These spectra are converted to non-linear spectra (based on rules de-rived from inelastic spectraanalysis) plotted on spectral accel-eration, S

a , versus spectral displace-

ment, Sd

, axes, similar to ATC-40representations. The spectral ca-pacity curve represents the abilityof the structure to deform at vari-ous degrees of resistance. The spec-tral capacity can be approximated

[B] Pipe Joint Capacity Risk Plot

0

1

2

1.00E-06 1.00E-05 1.00E-04 1.00E-03 f

4 Fj

curves of wave length: Tp x c

fail

no fail10

0000

Fai

lure

Lev

el

5000

0

1000

0

5000

2000

00

� Figure 8. Pipe-soil friction transfer plots


from a “pushover” analysis in whichmonotonically increasing lateralloads are applied to the structureand the deformations are plottedagainst the load in the (S

d , S

a ) coor-

dinates following scaling. The basicidea of the SCM is the assumptionthat the expected median responseis determined by the intersectionof the spectral capacity and spec-trum demand curves. This intersec-tion is termed the expectedresponse point. If either or both ofthe curves are random, the re-sponse is random. This is concep-tually illustrated in Figure 9 byplotting mean plus/minus onestandard deviation curves for thecapacity and demand. The actualresponse is distributed in the inter-section range. The fragility curvesrepresent the probability-based re-lation between the expected re-sponse and the performance limits.

Various methods were developedto determine the probability thatthe response exceeds a given limitstate. One method has been sug-gested by Shinozuka et al. (2000)for the case of random groundmotion and deterministic struc-tures. The mean and mean plus/minus one standard deviation de-mand curves are plotted, and theintersection of these curves withthe deterministic capacity curve ofthe structure gives estimates of themean and mean plus/minus onestandard deviation of the response.A lognormal distribution is then fit-ted to these data points, allowingthe fragility to be calculated.

Barron and Reinhorn (2000) de-veloped a second method. In thisapproach, the structural responseis evaluated from inelastic nonlin-ear response spectrum and spec-tral capacity curves and theprobability distribution function of

spectral ordinates is obtained fromsimplified relationships. These re-lationships are functions of thestructural parameters and allow adirect assessment of the probabil-ity of exceeding the limit states,from which the fragility curves canbe constructed. This is a consis-tent methodology that explicitlyconsiders and quantifies uncertain-ties in both ground motion andstructural response. Sensitivityanalysis of the fragility curves canbe used directly to determine op-timum retrofit techniques, or it canbe combined with cost estimationsto obtain an economically baseddecision.

The advantage of the SCM is itssimplicity and ease of use. Notime-consuming time history analy-ses need be performed. However,it is a heuristic method and willonly give approximate answers.Shinozuka used the SCM methodto develop fragility curves for abridge in the Memphis area.Barron and Reinhorn (2000) gen-erated fragility curves for thePatterson building and a bridge,also in Memphis. The influence ofthe various structural parameters

“Damage tononstructuralcomponents canpose a greaterthreat to safetythan damage tostructuralcomponents.”

� Figure 9. Probability of exceeding limit-state

Mean demand

Displacement limit-stateand variability

Expected responseProbability of exceeding

limit state

Mean capacity

D i sp l acemen t

Ac

ce

lera

tio

n/g

24

such as the damping ratio, the yieldstrength level, the initial period, andthe post-yielding stiffness ratio wasconsidered in evaluating theprobabilistic response of nonlinearsystems. The fragility curves ob-tained from this study were moresensitive to the addition of damp-ing than to the changes in stiffnessand strength (see Figure 10), indi-cating the efficiency of increaseddamping in structures.

Time History Analysis

Time history analysis is used tosimulate the response of a structureto a given ground excitation. It isconsidered to be the most detailednumerical analysis method. Thecalculation of fragility curves canbe based on the same approach asin the section on Field and Experi-mental Data. The only difference

is that the data is obtained by cal-culation.

Sliding Fragility Curves ofUnrestrained Equipment inCritical Facilities

Fragility curves were constructedanalytically for the standing-freeblock described in the section onExperimental Data and consideredthe same eight failure thresholds.Ninety acceleration time historieswere generated using SIMQKE, anartificial motion generation pro-gram. The time histories werebased on a response spectrum fromthe 1997 NEHRP recommendedprovisions for seismic regulationsof new buildings and other struc-tures. Each of the ninety accelera-tion time history inputs was scaledto have eight different horizontalpeak ground accelerations between0.3 g and 1.0 g. Each of these eight

Pro

ba

bil

ity

of

Da

ma

ge

Pro

ba

bil

ity

of

Da

ma

ge

Pro

ba

bil

ity

of

Da

ma

ge

0.0

0.2

0.4

0.6

0.8

1.0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Qy/W =.30

Qy/W =.25

Qy/W =.20

2%-505%-5010%-50

H

M

I

r=4.4 5.3

r=2.3 2.

8

r=1.1 1.

3

PGA (g)

0.0

0.2

0.4

0.6

0.8

1.0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

To=0.6 s

To=0.5 s

To=0.4 s

2%-505%-5010%-50

H

MI

r=5.8 6.3 6.5

r=2.7 3.4 4.7

r=1.3 1.6

PGA (g)

0.0

0.2

0.4

0.6

0.8

1.0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

6 %

5 %

4 %

2%-505%-5010%-50

H

M

I

r=6.1 6.6 7.2

r=3.2 3.4 3.9

r=1.5 1.6 1.8

PGA (g)

0.0

0.2

0.4

0.6

0.8

1.0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

25 %r=1.5

5 %r=3.4

15 %r=2.0

2%-505%-5010%-50 PGA (g)

Influence of Suplemental Dampingin Moderated Damage

Pro

ba

bil

ity

of

Da

ma

ge

� Figure 10. Fragility sensitivity to change of structural parameters (a) stiffness changes; (b) strength changes; (c) inherentdamping estimation; (d) addition of supplemental damping (for moderate limit state only)


horizontal time histories was com-bined with four different verticalacceleration inputs, again scaled togive ratios between VPGA andHPGA of 0, 1/4, 1/3, and 1/2. Thetotal number of time history com-binations was 2,880. All of the timehistory combinations were re-peated for five different coefficientsof dynamic friction. Figure 11shows relative displacement andabsolute acceleration time historiesfor HPGA of 0.7 g, VPGA of 0.23 gand coefficient of dynamic frictionof 0.21. Fragility curves have alsobeen generated and plotted againstthe HPGA. Figure 12 shows one ofthe curves.

Fragility Curves for ReinforcedConcrete Bridges in the MemphisArea

Two representative bridges withprecast prestressed continuousdecks in the Memphis, Tennesseearea were used for fragility curvedevelopment by Shinozuka et al.(1999a, 1999b). Figure 13 showsthe plan, elevation and columncross-section of one of the bridges.The strength of the concrete f

c used

for the bridges was assumed to bedescribed by a normal distribution,while the yield strength f

y of the

reinforcing bars was described bya lognormal distribution. Other

parameters that could contribute tovariability of structural responsewere not considered in the analy-sis. Samples of ten bridges weregenerated for each bridge. The timehistories generated by Hwang andHuo (1996) were used for the

� Figure 11. Analytical Solution I

� Figure 12. Fragility Curves for Ud = 0.3; Failure Threshold = 1 inch

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1 1.2

HPGA. g

vpga/hpga = 0vpga/hpga = 1/4vpga/hpga = 1/3vpga/hpga = 1/2F

rag

ility

26

ground motion. Ten time historiesfor each of several moment magni-tude and distance combinationswere matched with the sample often bridges. The SAP 2000 finite el-ement code was used to calculatethe state of damage of each bridge.Bilinear hysteretic elements with-out strength or stiffness degenera-tion were used. The results fromSAP 2000 were validated for thebilinear behavior by analyzing thesame problem using ANSYS. Fig-ure 14 shows the fragility curves as-sociated with two states of damagefor the two bridges. Tests of good-ness of fit and estimation of confi-dence intervals have been done forthe bridges considered in the study.In this study, Shinozuka et al.

(1999b) also developed fragilitycurves using spectral acceleration(SA), peak ground velocity (PGV),spectral velocity (SV) and spectralintensity (SI) as measures of groundmotion intensity for comparison,and preliminary observations weremade; however, a more elaboratestudy is currently underway.

Fragility Curves for a Water Tankin a Hospital

A hypothetical 22,000 gallon wa-ter tank located on the 20th floor ofa hospital in New York City wasused to illustrate the developmentof fragility curves for secondarysystems based on the results oftime-history analyses. The hospi-tal is a steel frame building with

“The damagedata obtainedfrom the 1999Marmara,Turkeyand Chi-Chi,Taiwanearthquakes areextremelyvaluable for thedevelopment offragility curves.”

� Figure 13. A representative Memphis bridge


partial moment resistant connec-tions and supported by pile foun-dations. Figure 15 shows the plazaand the middle tower of the hospi-tal. Figure 16 shows the locationof the water tank on a schematicplan of the hospital.

The water tank has dimensionsof 20.3 x 16.8 x 4.8 ft. It is sup-ported at its corners by four verti-cal legs, each of height 6.83 ft. Theground excitation was assumed tobe a zero-mean band limitedGaussian white noise (BLWN). Fra-gility curves were generated forfour limit states. The limit statesconsidered were d > d

cr , where d

is drift of the water tank legs andd

cr takes the values of 6, 12, 24 and

36% of the height of the legs. Theprimary structure was assumed toremain linear elastic under theground motion excitations. Thebody of the tank was modeled as arigid body. No feedback from thesecondary system to the primarystructure was considered. The in-put to the water tank is modeledby the corresponding floor accel-eration spectral density defined by

analogy with the pseudo accelera-tion floor response spectrum. Theonly source of randomness is theseismic input. Figure 17 shows thesolution sequence.

The primary structure was ana-lyzed separately neglecting the in-teraction with the secondarysystem. The responses at the attach-ment points are used as the inputto the secondary system. The pri-mary structure is a linear filter sothat the power spectral densityfunction for the response at the at-

tachment points can be calcu-lated (Soong and Grigoriu,1993). The function is then usedto generate floor accelerationsfor the secondary system. 240realizations of the input were

� Figure 15. Plaza and themiddle tower of the hospital

� Figure 16. Schematic plan of the hospital

� Figure 14. Fragility curves for bridge 1 and 2

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.2 0.4 0.6 0.8 1.0

PGA(g)P

rob

abili

ty o

f E

xcee

din

g D

amag

e S

tate

Bridge1, MINOR ( = 0.20g, = 0.18)C0 0�

Bridge1, MAJOR ( = 0.26g, = 0.13)C0 0�

Bridge1, MAJOR

Bridge2, MINOR ( = 0.23g, = 0.20)C0 0�

Bridge2, MAJOR ( = 0.30g, = 0.31)C0 0�

28

generated for twelve values ofwhite noise intensity g

o . The sec-

ondary system was studied usingDIANA under each of the excita-tions. The body of the tank was as-sumed to be rigid so thatdeformations were concentrated inthe four legs. Both geometrical andmaterial nonlinearities were consid-ered in the analysis. The materialmodel of the four legs was von Misesplasticity with strain hardening.The maximum drift of the legs, d,was calculated for each simulation.For a given limit state, d

cr , and a spe-

cific value of go , the fragility P (d

cr ,

go ) was calculated as the proportion

of the samples with maximum drift,d, exceeding d

cr .

P (dcr, g

o ) = N

ex (d

cr , g

o ) / N (g

o ),

where N (go) is the total number

of samples generated for the giveng

o. A lognormal distribution was

then fitted to the points to give the

fragility curves. The procedure wasrepeated after retrofitting the tankwith diagonal braces. The fragilitycurves for the system before and af-ter retrofitting are shown in Figure18. It is clear that the retrofit im-proves the seismic performance ofthe system.

System FragilityA structural or nonstructural sys-

tem typically consists of many indi-vidual components connectedtogether in a certain way. The fra-gility of the system depends on (1)the fragility of the individual com-ponents and (2) the way in whichthe components are connected.The fragility of the system can bevery different from the fragilities ofthe components. It is thus essen-tial to evaluate the system fragilityin addition to the individual com-ponent fragilities.

BLWN CWNResponse

of

secondary system

FragilityCurves

Secondary System

(Nonlinear Filter)

Primary Structure

Linear Filter

Statistical Data

Processing

FragilityCurves

� Figure 17. Solution sequence

� Figure 18. Fragility curves of the secondary system

0

0.2

0.4

0.6

0.8

1

0 0.1 0.2 0.3 0.4 0.5 0.6go

Fra

gilit

yC

A BD

A: Drift > 6% B: Drift >12%C: Drift > 24% D: Drift > 36%

___ Bare System ___ Braced System


Seismic Risk Assessment of LosAngeles Area ExpresswayNetwork

A seismic risk analysis was per-formed (Shinozuka, et al. 1999b) onthe Los Angeles area expresswaynetwork under a postulated mag-nitude 7.1 Elysian Park earthquake.For this purpose, the families of fra-gility curves for “at least minor,” “atleast moderate,” “at least major”and “collapse” states of damage de-veloped in the section on FieldData were used. States of damagefor all 2,225 Caltrans’ bridges inLos Angeles and Orange Countywere simulated.

The state of damage simulatedfor each bridge was quantified us-ing a “bridge damage index” (BDI).The BDI is 0.1, 0.3, 0.75 and 1.0for the “minor,” “moderate,” “ma-jor” and “collapse” damage states,respectively. The state of link dam-age was then quantified by makinguse of a “link damage index” (LDI)which was computed for each linkas the square root of the sum of thesquares of BDI values assigned toall bridges on the link under con-sideration. The LDI value was thentranslated into link traffic flow ca-pacity. In this study, it was consid-ered reasonable to assume that thecapacity is 100% (relative to thecase with no damaged bridges onthe link) if LDI < 0.5 (no link dam-age), 75% if 0.5 < LDI < 1.0 (minorlink damage), 50% if 1.0 < LDI < 1.5(moderate link damage) and 25% if1.5 > LDI (major link damage).

In an on-going research project,a computer code “USC-EPEDAT” isbeing developed to perform thesimulation of states of link damageand hence, network damage, effi-ciently (each simulation takes less

than 10 seconds with a 300 MHzor faster PC). Figure 19 depicts theresult averaged over 10 simulations.The information contained in Fig-ure 19 can be used to support de-cision-making for post-earthquakeresponse activities in near real-time.Figure 20 depicts the state of theexpressway network damage un-der the assumption that eachbridge has appropriate seismic ret-rofitting so that the median param-eter of the lognormal fragilitycurve is increased by 50%. Figure20 clearly indicates that the retro-fitting improves the performance

� Figure 19. Averaged network damage under postulated Elysian Parkearthquake (10 simulations)

� Figure 20. Averaged network damage under postulated Elysian Parkearthquake (10 simulations on retrofitted network)

30

of the network. Such a compari-son makes it possible to evaluatethe cost-effectiveness of differentseismic retrofit strategies by addi-tional cost-benefit analysis.

San Francisco High-rise FireSuppression System

Logic trees are frequently used torepresent and analyze structuraland nonstructural systems. An ex-ample of such a tree is shown inFigure 21 for a San Francisco high-rise fire suppression system

(Grigoriu and Waisman, 1998). Thetree consists of the componentsalong with “gates” describing thelogical connectivity of the compo-nents. For example, in Figure 21, thewater supply will function if one orboth of the city water and the20,000 gallon tank function. Theentire system will function only ifall three of the water supply, wa-ter pumps and piping function.Equations have been derived forcalculating the probability of fail-ure of the system from the prob-ability of failure of the individual

“Buildings thatremain struc-turally soundafter a strongearthquakeoften lose theiroperationalcapabilities dueto damage totheir non-structuralcomponents.”

� Figure 21. Logic tree for San Francisco high-rise fire suppression system

FireSuppression

Piping

CityWater

20,000GallonTank

NormalBuilding

Power

NewEmergencyGenerator

ElectricPumps (2)

DayTank

StartSystem Valves

55GallonDrums

Piping

DieselDrivenPumps

AND GATE

OR GATE

Component above gate functionsif all components below function

Component above gate functionsif any components below function

Symbol Name Meaning

WaterPumps

WaterSupply

Subsystem 1 Subsystem 2


Nonstructural DamageDatabasehttp://mceer.buffalo.edu/

publications/reports/docs/99-0014/default.html

(a) (b)

0.0 0.5 1.0 1.5 2.0

PGA (g)

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0PGA (g)

0.0

0.2

0.4

0.6

0.8

1.0

pcw=0.8385

pgt=0.1071

� Figure 22. Fragility curve for 20,000 gallon tank

(a) (b)

0.0 0.2 0.4 0.6 0.8 1.0

PGA (g)

0.0

0.2

0.4

0.6

0.8

1.0

RS

F

0.7885

0.5 0.0 0.2 0.4 0.6 0.8 1.0

PGA (g)

0.0

0.2

0.4

0.6

0.8

1.0

PS

F

0.2115

0.5

� Figure 23. System fragility curve

components. These equations canbe used to generate a fragilitycurve for the system from the fra-gility curves of the components(Figures 22 and 23). The logic treemethod can also be used for sensi-tivity analysis to identify the criti-cal components of the system, andto determine confidence intervalsfor the system fragility (Roth,1999).

ConclusionsThis paper outlines common

methods for generating fragilitycurves, explores some features andlimitations of each method, and il-lustrates the methods by means ofexamples. The methods are basedon field data, experiments, and nu-merical calculation.

In the future, the techniques de-scribed herein will be extended, re-fined and made more accessible toengineers through the developmentof suitable computation programs.The statistical data analysis methodwill be applied to data from the1999 earthquakes in Taiwan andTurkey. A computer program isplanned for the simplified SpectralCapacity Method (SCM), while themore detailed programs for timehistory analysis will be improvedthrough the addition of sensitivityanalysis, adaptive time step algo-rithms, more robust material mod-els, and parallelization. Anotherresearch direction will be in theuse of fragility curves for cost-ben-efit analysis and systems analysis.Alternative ground motion inten-sity measures to PGA will also beconsidered.

32

Applied Technology Council (ATC), 1985, Earthquake Damage Evaluation Data forCalifornia, Report ATC-13, Applied Technology Council, Redwood City, California.

Barron, R., and Reinhorn, A., 2000, Spectral Evaluation of Seismic Fragility ofStructures, Technical Report, Multidisciplinary Center for Earthquake EngineeringResearch, Buffalo, NY, (in review).

Basöz, N., and Kiremidjian, A.S., 1998, Evaluation of Bridge Damage Data from theLoma Prieta and Northridge, California Earthquake, Technical Report MCEER-98-0004, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, NY.

Chong, W.H. and Soong, T.T., 2000, Sliding Fragility of Unrestrained Equipment inCritical Facilities, Technical Report MCEER-00-00xx, Multidisciplinary Center forEarthquake Engineering Research, Buffalo, NY.

Grigoriu, M. and Waisman, F., 1998, “Seismic Reliability and Performance ofNonstructural Components,” Proceedings of the Seminar on Seismic Design,Retrofit and Performance of Nonstructural Components, ATC 29-1, AppliedTechnology Council, Redwood City, CA.

Holman, G.S., Chou, C.K., Shipway, G.D. and Glozman, V., 1987, Compact FragilityResearch Program, Phase I Demonstration Tests, NUREG/CR-4900.

Hwang, H.M. and Huo, J.R., 1996, Simulation of Earthquake Acceleration TimeHistories, Center for Earthquake Research and Information, The University ofMemphis, Technical Report.

Kao, A., Soong, T.T. and Vender, A., 1999, Nonstructural Damage Database, TechnicalReport MCEER-99-0014, Multidisciplinary Center for Earthquake EngineeringResearch, Buffalo, NY.

Kwan, W.P. and Billington, S.L., 1999a, “Cyclic FE Analyses of Structural Concrete I:Material Model Evaluation,” ASCE Journal of Structural Engineering, (in review).

Kwan, W.P. and Billington, S.L., 1999b, “Cyclic FE Analyses of Structural Concrete II:Simulation of Experiments,” ASCE Journal of Structural Engineering, (in review).

Report I, 1999, “Specimen and Loading Characteristics,” Specification for the ParticipantsReport, organized by Commissariat À L’energie Atomique, Electricite De France,Camus 3 International Benchmark, August.

Roth, C., 1999, “Logic Tree Analysis of Secondary Nonstructural Systems withIndependent Components,” Earthquake Spectra, Vol. 15, No. 3.

Shinozuka, M., Feng, M.Q., Lee, J.H. and Nagaruma, T., 1999a, “Statistical Analysis ofFragility Curves,” Proceedings of the Asian-Pacific Symposium on StructuralReliability and its Application (APSSRA 99), Keynote Paper, Taipei, Taiwan,Republic of China, February 1-3, Journal of Engineering Mechanics, ASCE, (acceptedfor publication).

Shinozuka, M., Feng, M., Kim, H., Uzawa, T. and Ueda, T., 1999b, Statistical Analysisof Bridge Fragility Curves, Technical Report, Multidisciplinary Center forEarthquake Engineering, Buffalo, NY, (in review).

Soong, T.T. and Grigoriu, M., 1993, Random Vibration of Mechanical and StructuralSystems, PTR Prentice-Hall, New Jersey.

Zhu, Z. and Soong, T.T., 1998, “Toppling Fragility of Unrestrained Equipment,”Earthquake Spectra, Vol. 14, No. 4.

References

33


Research Objectives

As part of research collaboration between MCEER and the NationalCenter for Research in Earthquake Engineering (NCREE) in Taiwan, ateam of MCEER/NCREE researchers undertook a reconnaissance mis-sion shortly after the 921 Chi-Chi earthquake occurred in Taiwan at1:47 a.m. on September 21, 1999. A major objective of this mission wasto assess earthquake-induced damage, to document lessons learned, andto identify short-term strategies for post-earthquake restorations. Thispaper, a portion of the MCEER/NCREE Reconnaissance Report (MCEER/NCREE, 2000), summarizes preliminary findings on the damaging ef-fects of the earthquake on critical facilities.

Critical facilities include hospitals and health care facilities; schools;police, fire and emergency response stations; key government facili-

ties; and key industries. They provide lifesaving functions and render emer-gency assistance to communities when a disaster strikes. It is thusparticularly important that every effort be made to insure their safety andfunctionality during and after a disaster. In this paper, damages to some ofthese facilities due to the 921 Chi-Chi earthquake are assessed, togetherwith their probable causes and impact. Possible corrective actions andresearch needs in mitigating the effect of a similar future disaster event onthese critical facilities are addressed.

It is noted that damage information on many of these facilities is still beingcollected and processed at the time of this writing. Therefore, they should beconsidered incomplete and preliminary as reported in this section.

HospitalsAccording to available information, there are 4,375 health care facilities

within the six-county seismic affected zone, of which 165 are hospitals.Damage to hospitals can be grouped into the following three categories:(1) minor structural damage and minor nonstructural damage, (2) partial

Damage to Critical FacilitiesFollowing the 921 Chi-Chi,Taiwan Earthquake

by Tsu T. Soong (Coordinating Author), George C. Yao and C.C. Lin

National Center for Researchon Earthquake Engineering,National Taiwan University

Tsu T. Soong, Samuel P.Capen Professor ofEngineering Science,Department of Civil,Structural andEnvironmentalEngineering, University atBuffalo, State University ofNew York

George C. Yao, AssociateProfessor, Department ofArchitecture, NationalCheng Kung University

Chi-Chang Lin, Professor andChairman, Department ofCivil Engineering, NationalChung-Hsing University

34

structural damage but serious non-structural damage, and (3) seriousstructural damage. In the first cat-egory, evacuation of patients andstaff was not required and the hos-pitals were capable of performingemergency care to earthquake vic-tims. In the second, hospitals wererendered non-serviceable and timewas required for restoration. Thistype of damage was prevalent in theareas close to the epicenter. For ex-ample, within the critical 48-hourperiod after the earthquake, close to1,000 beds were lost for patient carein Nantou county alone. Seriousstructural damage, or the third cat-egory, was also evident in regionsclose to the epicenter, where majorhospitals were closed, requiring ei-ther demolition or major repair.

While structural damage waswidespread in the six-county af-fected area, nonstructural damagewas found to be a major factor ad-versely affecting functionality ofmajor hospitals. Common occur-rences included fallen interior wallsand ceilings, toppling, sliding orcollision of medical and non-medi-cal equipment, overturning of wa-ter and oxygen tanks, interruptionof emergency power, and floodingdue to pipe breakage. For a closerscrutiny, several hospitals were cho-sen for more in-depth site visits. Inwhat follows, damage investigationsmade on October 5, 1999 at threemajor hospitals are summarized.

The observations described beloware based on interior as well as ex-terior damage inspections and oninterviews with hospital officials.

Christian Hospital, Puli

A major facility in Puli and sur-rounding communities, the Chris-tian Hospital is a 400-bed facility ofreinforced concrete (RC) construc-tion consisting of a new (about ayear old) section and an old (about20 years old) section (see Figure 1).

Damage Summary

• New section sustained consider-able damage from the mainevent on 9/21/99. Out of safetyconcerns primarily due to non-structural damage (water dam-age, equipment failure, etc.), thebuilding was evacuated withpatients housed in tents on thehospital grounds. The buildingwas immediately inspected andconsidered safe. Upon interiorcleaning, patients were returnedto the building.

• The building suffered significantnonstructural damage againfrom the 9/27/99 M6.8 after-shock. Patients were again evacu-ated and housed in temporarytrailers with considerably re-duced capacity (about 50 beds),with overflow transferred toother area hospitals.

MCEER Program 1: SeismicEvaluation and Retrofit ofLifeline Systems

MCEER Program 2: SeismicRetrofit of Hospitals

MCEER Program 3: EmergencyResponse and Recovery

This research focused broadly on the damage to numerouscritical facilities caused by the 921 Chi-Chi, Taiwan earthquake.It is anticipated that numerous groups including the earthquakeresearch community; seismic code committees; regulatory offi-cials; and administration and technical staff of critical facilitiesdealing with seismic vulnerability and rehabilitation will havean interest in the issues discussed.

35Damage to Critical Facilities Following the 921 Chi-Chi, Taiwan Earthquake

• The first floor of the building re-mained open and was being usedfor emergency care, patient reg-istration and processing, com-mand post, and other necessaryfunctions (see Figure 2).

Consequences and Impact

• A major part of the hospital wasnon-serviceable primarily due tononstructural damage.

• Drastically reduced capacity(10% of original) at a time whendemand was the highest.

• Trauma to patients through tworelocations.

• Drastically reduced services dueto equipment damage.

• The lack of an earthquake emer-gency management plan prob-ably made the situation worse.

RestorationRestoration was underway. It was

estimated that the interior would

be restored and serviceable withintwo weeks.

Veterans Hospital, Puli

The Veteran’s Hospital is anothermajor hospital located in Puli. It isa 450-bed facility with two mainreinforced concrete (RC) buildings

Photographs by T.T. Soong

� Figure 1. The Christian Hospital sustained considerable damage. Photos show (left) partof the exterior damage to the hospital; and (right) interior damage to partitions andceilings.

Photograph by T.T. Soong

� Figure 2. The first floor of the Christian Hospital remainedopen though the hospital suffered extensive interior damage.

Multidisciplinary Centerfor EarthquakeEngineering Research

http://mceer.buffalo.edu/research/taiwaneq9_99/default.html

36

(the Medical Center and the Admin-istration Center), built about threeyears ago and several older (about25 years old) and smaller buildings(see Figures 3 and 4).

Damage Summary

• New buildings sustained consid-erable damage from the mainevent. The Medical Building(Bldg. 1) was closed and the pa-tients in the AdministrationBuilding (Bldg. 2), along withthose in Bldg. 1, were eithermoved to the older buildings ortransferred to other VA hospi-

tals. About 220 patients re-mained at the hospital.

• Considerable nonstructural dam-age in Bldgs. 1 and 2, includingpower failure1, water damage,and equipment damage. Bldg. 1also sustained considerablestructural damage, probably dueto a lack of ductile detailing.


• A major part of the hospital wasnon-serviceable due to bothstructural and nonstructuraldamage.

• Drastically reduced capacity(50% of original) at a time whendemand was the highest.

• Trauma to patients due to evacu-ation.

• Drastically reduced services dueto equipment damage.

• As in the case of the ChristianHospital, no earthquake emer-gency management plan ap-peared to be in place at the timeof the earthquake.

RestorationWhether Bldg. 1 was to be demol-

ished or repaired remained to bedetermined. Bldg. 2 was expected

to be repaired within twoweeks.

� Figure 4. The Veterans Hospital sustained bothexterior damage (left) and interior damage (right)to the newest buildings of the Medical Center.

Photographs by M. Bruneau

Photograph by M. Bruneau

� Figure 3. An overview model of the Veterans Hospital in Puli. Thetwo white buildings are the newest and both sustained consider-able damage.


Shiu-Tuan Hospital, Tsushan

The Shiu-Tuan Hospital is a 9-story, two year old reinforced con-crete (RC) building that has a400-bed capacity. It is privatelyowned and is the largest in Nantoucounty. The structure is situatedabout 120 m from the Che-Lung-PuFault with an uplift of approxi-mately one meter at the site.

Damage Summary

• Structurally intact, it sufferedconsiderable nonstructural dam-age as in the case of the othertwo hospitals (see Figures 5 and6). Interior damage was mostsevere at the second- and third-floor levels where, unfortunately,some of the major facilities, suchas operating and recoveryrooms, were located. Patientswere moved to open hospitalground and subsequently trans-ferred to other hospitals.

• Hospital closed.


• Trauma to patients due to evacu-ation and reallocation.

• Hospital closed, making the larg-est hospital in this vicinity un-available to patients andearthquake victims.

• Seven patients died due to stop-page of life-support system.

RestorationRepair was underway and the

process was expected to take oneto two months. Funds for the re-pair remained to be found.

Photograph by G.C. Yao

� Figure 6. Damage to an exterior wall of the Shiu-TuanHospital in Nantou county

Photographs by G.C. Yao

� Figure 5. Interior damage in the Shiu-Tuan Hospital.Shown are (left) fallen brick inside the hospital and(right) a damaged interior glass brick wall.

38

SummaryThe damage to the three hospitals

and its impact underscores the im-portance of securing medical equip-ment and protecting patients andstaff from falling debris and over-turning objects. As demonstrated inthe case studies above, hospitals canbe rendered non-serviceable andlives of patients can be lost due tofailure of life support equipment incritical care areas. Table 1 is a com-pilation of nonstructural damage inthe three hospitals highlighted inthis paper which, incidentally, couldhave been reduced or even avoidedwith inexpensive and easilyimplementable protective measures.

SchoolsSchool buildings sustained severe

damage, reaching as far as the cityof Taipei, 150 km from the epicen-ter. As in the 1998 Chia-Yi/Ruei-Liearthquake, the severity of damageto school buildings, as demon-strated in Table 2, again exceededthat of other structures due prima-rily to the commonality of theirweaknesses in construction. Thecommon problems associated withschool buildings appeared to be, onthe one hand, short-column effectswhich led to shear failure in col-umns and, on the other, eccentric-ity of most school buildingsassociated with cantilevered corri-dors at upper floors (see Figure 7).It is estimated that restoration, re-pair and reconstruction costs asso-ciated with school buildings canreach US$ 1.3 billion (Ministry ofEducation, 1999a).

According to a recent accountingmade available by the Ministry ofEducation, a total of 786 schoolswere damaged by the earthquakeand its aftershocks as listed in Table3, of which 51 suffered completecollapse. Damage was heavily

noitutitsnIfoepyT latoT degamaD )%(oitaRegamaD

segelloCdnaseitisrevinU 63 33 7.19

snoitutitsnIlacinhceT 89 83 8.83

seitisrevinUlamroN 31 8 5.16

sloohcShgiH 242 36 0.62

sloohcSelddiM 517 861 5.32

sloohcSyratnemelE 755,2 884 1.91

degatnavdasiDehtrofsloohcS 02 4 0.02

latoT 186,3 208 8.12

� Table 2. Extent of Damage to Schools

Ministry of Education, 1999b

noitaucavEdnanoitpursiDfoesuaC naitsirhC snareteV nauT-uihS

egatuOrewoPpukcaB X X

egatuOylppuSretaW X X X

egatuOecivreSsaG X X

egamaDrotavelE X X

eruliaFsnoitacinummoC X X X

sirbeDgnillaF X X X

egakaeLretaW,gnipiPnekorB X X X

seruliaFegarohcnACAVH X X X

egamaDtnempiuqElacinahceM X X X

sknaTegarotSdiuqiL,saGfognilppoT X X X

egamaDtnempiuqElacideM X X X

ecalPnitoNnalPnoitaucavEycnegremE X X X

� Table 1. Nonstructural Damage in Three Surveyed Hospitals


ytnuoC/ytiC 1 seitisrevinUsegelloCdna

setutitsnIlacinhceTsloohcShgiHdna

elddiMsloohcS

yratnemelEsloohcS

latoT

ytiCiepiaT 8 8 8 34 76

ytnuoCiepiaT 2 1 12 25 67

ytnuoCnaL-iY 0 3 1 4 8

ytnuoCnauY-uoT 2 1 1 7 11

ytiCuhcnisH 2 0 5 9 61

ytnuoCuhcnisH 0 1 2 01 31

ytnuoCiL-uiM 2 4 02 95 58

ytiCgnuhciaT 11 9 91 73 67

ytnuoCgnuhciaT 3 51 41 93 17

ytnuoCuotnaN 6 11 01 24 96

ytnuoCawH-gnahC 3 01 62 64 58

ytnuoCniL-uY 1 6 01 23 94

ytiCiY-aihC 0 9 7 61 23

ytnuoCiY-aihC 2 2 7 53 64

ytiCnaniaT 1 0 01 81 92

ytnuoCnaniaT 1 2 0 0 3

srehtO 3 1 6 93 94

latoT 74 2 38 2 861 884 687 2

1 .seitnuocehtnihtiwseiticniesohtedulcnitonodseitnuocrofsrebmuN2 .secruosnoitamrofnitnereffidoteudylbaborpera3-4dna2-4selbaTneewtebseicnetsisnocnI

� Table 3. Damage to Schools

Ministry of Education, 1999a

Photographs by T.T. Soong

� Figure 7. Typical damage to schools due to short-column effect and eccentricity

40

concentrated in Nantou andTaichung counties as illustrated inTable 4. In Nantou county, for ex-ample, 139 out of 186 elementaryand middle schools, or approxi-mately 75%, suffered damage seri-ous enough that they had to beclosed. This situation not only af-fected the education of students,but also made them unusable asevacuation and emergency re-sponse centers.

Police and FireStations

As in the case of schools and otherpublic buildings, police stations andemergency response centers alsosustained severe damage in the

affected region (for ex-ample, see Figures 8and 9). Damage reportforms were sent tothese units by the Na-tional Center for Re-search in EarthquakeEngineering (NCREE)investigators andthose returned to date

are summarized in Table 5.

Key IndustrialFacilities

Of particular interest to the inter-national business community wasimpact of the Chi-Chi earthquakeon the output at the Hsinchu’s Sci-ence Based Industrial Park, whereabout 30 firms produce a significantpercentage of the world’s semicon-ductors and silicon wafers. Dam-age to this facility and its globalimpact are the focus of this section.

Hsinchu’s Industrial Park, situatedabout 110 km from the epicenter,houses approximately 239 hightechnology firms that have impor-tant links to the world’s computer

ytnuoC espalloClatoT dnaespalloClaitraPegamaDlarutcurtsnoN

latoT

uotnaN 03 901 931

gnuhciaT 11 23 34

seitnuoCgnirobhgieN 01 14 15

� Table 4. Severity of Elementary and Middle School Damage

Ministry of Education, 1999a


� Figure 9. Heavily damaged Puli Police Station


ytnuoC/ytiC

egamaDfoytireveS

laitraProlatoTroespalloCgninrutrevO

egamaDsuoireSnoitilomeDgniriuqeR(

)tiforteRro

egamaDetaredoMtiforteRgniriuqeR(

)elbariapeRro

thgiLegamaD

)elbariapeR(

oNegamaD

ytnuoCuotnaN 5 1 1 1 0

ytnuoCgnuhciaT -- -- -- -- --

ytiCgnuhciaT 0 0 0 41 0

ytnuoCiL-uiM 1 0 2 1 4

ytnuoCawH-gnahC 0 2 3 0 0

ytnuoCniL-uY -- -- -- -- --

� Table 5 Interim Damage Summary Pertaining to Police and Fire Stations

NCREE, 1999a


� Figure 8. Total collapse of Puli Town Hall

and communications industry.Based on the types of products theyproduce, they can be grouped intothe following: Integrated Circuit:95, Computers and Peripherals: 44,Telecommunications: 36, Electro-optical: 35, Automation: 16, and Bio-technology: 13.

Power to the entire island wasinterrupted due to damage to theelectrical transmission network andswitching stations close to the epi-center, due to high-priority user sta-tus at the Hsinchu’s Industrial Park.Power to the Park was restored to

full capacity at 500,000 KV on Sep-tember 25, 1999, four days after theearthquake. Even so, productionloss at the facility was estimated tobe around US$ 400 million, most ofwhich incurred at the semiconduc-tor and silicon wafer productionfacilities.

Overall damage at the facility hasbeen light in comparison withthose closer to the epicenter. Again,nonstructural and equipment dam-age stood out, including fallen ceil-ings, cracked walls and partitions,shear failure of columns, piping

42

breakage, and equipment damage.Most of this damage was repairedrapidly and the entire industrialcomplex has been restored to itspre-earthquake production level.Table 6 lists damage survey resultsbased on the returned survey formsto date.

General Observationsand Lessons Learned

Seismicity

Heavy damage to critical facilitiesin areas close to the epicenter wascertainly in large part attributableto the unanticipated high level ofground shaking in the region. Forexample, Nantou and Taichungcounties, where most of the dam-age occurred, are located in seis-mic zone 2 with a design peakground acceleration (PGA) speci-fied at 0.23 g. On the other hand,the actual recorded PGAs in the re-gion were in general higher than0.35 g, and were as high as 0.92 g.Even accounting for an importancefactor of 1.25 for public buildings,the design PGA values were consid-erably below those actually experi-enced, causing widespread damageto constructed facilities.

Structural Damage

Beyond high seismicity, severalimportant factors contributing toobserved structural damage tobuildings, including critical facili-ties, have been identified and in-clude:

• Short-column effects leading tocolumn shear failure.

• Eccentricity due to cantileveredcorridors.

• Lack of ductile detailing.• Column failures due to inad-

equately spaced stirrups, rein-forcements, and splices.

• Soft-story induced failures.• Unstable foundations and

ground uplift.

However, in spite of these com-mon ills, structural damage to criti-cal facilities appeared to be heavierthan those in other sectors. It hasbeen speculated that this may bedue to the nature of the biddingprocess associated with the con-struction of public and governmentowned buildings, where fixed-pricedesign/built contracts are practiced.As reported in local news broad-casts, this practice invites abuse andencourages contractors to utilizesubstandard construction materialsand circumvent accepted engineer-ing practices in order to completetheir projects under budget.

� Table 6 Damage Survey at Hsinchu Industrial Park

171:denruteRsmroF

101:egamaDoN%95:latoTfotnecreP

07:detropeRegamaD%14:latoTfotnecreP

egamaD thgiL etaredoM suoireS

sllaWniskcarC 55 11 0

sroolFdemrofeD 8 3 0

snmuloCniraehS 5 0 0

NCREE, 1999b


Nonstructural Damage

As highlighted in several parts ofthis paper, the impact of nonstruc-tural damage on the loss of func-tionality of critical facilities hasbeen significant, leading to their in-ability to perform emergency andlifesaving services and, tragically,loss of lives. While seismic codesexist in Taiwan for buildings andbridges, there appears to be an ab-sence of rational seismic provisionsfor nonstructural components.Ironically, seismic performance ofnonstructural components can besubstantially improved using rathersimple and inexpensive means.

Research Needs andRecommendations

The outstanding issue identifiedby other team members fromMCEER and NCREE related to con-struction practices associated withpublic buildings needs to be crit-ically reviewed. Revisions and

References

Lee, G.C. and Loh, C-H., Editors, 2000, The Chi-Chi Taiwan Earthquake of September 21,1999: Reconnaissance Report, MCEER-00-0003, in press.

Ministry of Education, 1999a, Interim Damage Report, Taipei, Taiwan (in Chinese).

Ministry of Education, 1999b, Private Communications.

National Center for Research in Earthquake Engineering (NCREE), 1999a, Interim DamageReport, Taipei, Taiwan (in Chinese).

National Center for Research in Earthquake Engineering (NCREE), 1999b, Damage Reporton 921 Ji-Ji Earthquake (Draft), Report No. NCREE-99-033, December, Taipei, Taiwan(in Chinese).

Endnotes1 The emergency generators also failed. They were located on the second floor of a separate

building and, due to amplified acceleration on that floor, major components broke looseand rendered them inoperable.

modifications appear to be neces-sary in order to insure quality engi-neering and quality construction inthe future.

Also of critical importance is thenonstructural issue. Stringent seis-mic design and installation guide-lines need to be in place to insurenot only structural integrity, butalso functionality of critical facili-ties, which require protecting non-structural components, as well asstructures, from seismic damageunder strong ground shaking asexperienced in the Chi-Chi earth-quake. A systematic developmentof these guidelines involves the fol-lowing:

• Review and improve currentdesign and installation practicesin nonstructural components.

• Develop effective retrofit strate-gies for nonstructural compo-nents in existing criticalfacilities.

• Develop effective implement-ational procedures for existingfacilities and new constructions.

45

Research Objectives

The FHWA-sponsored project titled Seismic Vulnerability of New High-way Construction (MCEER Project 112), which was completed in 1998,performed studies on the seismic design and vulnerability analysis ofhighway bridges, tunnels, and retaining structures. Extensive researchwas conducted to provide revisions and improvements to current de-sign and detailing approaches and national design specifications forhighway bridges. The program included both analytical and experi-mental studies, and addressed seismic hazard exposure and ground mo-tion input for the U.S. highway system; foundation design and soilbehavior; structural importance, analysis, and response; structural de-sign issues and details; and structural design criteria.

In the fall of 1992, MCEER commenced work on a comprehensive re-search program sponsored by the Federal Highway Administration to

evaluate the seismic vulnerability of new and existing highway construc-tion. One part of this two-contract program, Seismic Vulnerability of NewHighway Construction (Project 112), resulted in a series of special studiesrelated to seismic design of highway bridges, tunnels, and retaining struc-tures, in order to develop technical information upon which new seismicdesign approaches and details, and future specifications, could be based.The project was prompted in part because significant progress had beenmade over the last two decades in several key areas, including improvedknowledge of: (1) seismic hazard and risk throughout the United States,(2) geotechnical earthquake engineering, and (3) seismically resistantdesign. At the time Project 112 was initiated, however, there were stillmany gaps in basic knowledge, and some of the recently-developed in-formation and data required additional study before they could be ap-plied directly to highway engineering applications nationwide.Consequently, Project 112 resulted in a series of analytical and experi-mental studies related to the seismic analysis, design, and performanceof bridges, tunnels, and foundations.

Highway Bridge Seismic Design:Summary of FHWA/MCEER Project on SeismicVulnerability of New Highway Construction

by Ian M. Friedland

Federal HighwayAdministration

Project Management

George C. Lee, Director,Multidisciplinary Centerfor EarthquakeEngineering Research

Ian G. Buckle, Professor,Department of CivilEngineering, University ofNevada, Reno; formerlyDeputy Director,Multidisciplinary Centerfor EarthquakeEngineering Research

Ian M. Friedland, AssociateDirector for Development,Applied TechnologyCouncil; formerly AssistantDirector for Bridges andHighways, MultidisciplinaryCenter for EarthquakeEngineering Research

46

The research conducted underProject 112 had a national focus andwas intended, in large part, to ad-dress differences in seismicity,bridge types, and typical designdetails between eastern or centralU.S. bridges, and those that had beenpreviously studied in California andthe western U.S. In particular, un-like the western U.S., design strate-gies used in the eastern and centralU.S. need to reflect the statisticalprobability that an earthquake sig-nificantly larger than the “design”earthquake can occur. In manycases, it was noted that Californiadesign practice required significantmodification before being imple-mented in the eastern and centralU.S. due to these differences in seis-micity and bridge construction type.

A range of special studies werecarried out that encompassed re-search on: seismic hazard; founda-tion properties, soil properties, andsoil response; and the response ofstructures and systems (see Figure1). These studies were conductedby a consortium of researchers, co-ordinated by MCEER. The consor-tium included a variety of academicinstitutions and consulting engi-neering firms, bringing togethermore than 20 earthquake and

bridge engineers and scientists.This consortium provided a balancebetween researchers and practicingprofessionals from the eastern, cen-tral, and western U.S.

It is anticipated that currentspecifications for the seismic de-sign of bridges will be revised, andthat new seismic design guidelineswill be prepared for other highwaysystem components, in part on thebasis of this work.

This paper summarizes some ofthe important results of the re-search conducted under the pro-gram. It draws heavily from areview of the program’s researchresults and expected impacts(Rojahn et al., 1999) and discussesissues raised in the report.

Seismic HazardExposure and GroundMotion Input

The research in the area of seis-mic hazard exposure focused on theevaluation of alternative approachesfor portraying and representing thenational seismic hazard exposure inthe U.S., quantifying and developingan understanding of the effects ofspatial variation of ground motion

• University at Buffalo

• Applied Technology Council

• Brigham Young University

• Dynamic IsolationSystems Inc.

• Earth Mechanics Inc.

• Geomatrix Consultants Inc.

• Imbsen & Associates Inc.

• Modjeski and MastersConsulting Engineers

• Princeton University

• Rensselaer PolytechnicInstitute

• University of Nevada, Reno

• University of SouthernCalifornia

It is anticipated that the results of this program will be consid-ered in future design specification development work. Specifi-cally, the AASHTO-sponsored National Cooperative HighwayResearch Program (NCHRP) initiated NCHRP Project 12-49, “De-velopment of Comprehensive Bridge Specifications and Commen-tary” in the fall of 1998. The objective of NCHRP Project 12-49,which is being conducted by a joint venture between MCEER andthe Applied Technology Council, is to develop new bridge seis-mic design specifications, commentary, and design examples,which can be incorporated into the AASHTO LRFD Bridge De-sign Specifications in the near future. Much of the basis for thespecification changes that will be recommended under NCHRPProject 12-49 are expected to be drawn from the results of thework conducted under this FHWA contract.

47Seismic Vulnerability of New Highway Construction

� Figure 1. Special Studies Conducted by MCEER under Project 112

on the performance of highwaystructures, and the development ofinelastic design spectra for assessinginelastic deformation demands.

Representation of SeismicHazard Exposure

Research in the area of seismichazard exposure representationwas conducted in order to:

• explore a number of importantissues involved in national

representations of seismicground motions for design ofhighway facilities;

• recommend future directions fornational seismic ground motionrepresentation, especially for usein nationally applicable guide-lines and specifications such asthe AASHTO seismic design pro-visions for bridges; and

• identify areas where further de-velopment and/or research areneeded to define ground motion

• Seismic Vulnerability ofExisting Construction(Project 106)

• Seismic Vulnerability of theNational Highway System(TEA-21 Project)

• National CooperativeHighway Research Program(NCHRP) Project 12-49

SEISMIC VULNERABILITY OF NEW HIGHWAY CONSTRUCTIONFHWA Contract DTFH61-92-C-00112

TASK A: PROJECT ADMINISTRATION & HIGHWAY SEISMIC RESEARCH COUNCIL

TASK B TASK D1 TASK D8 TASK D6 TASK D4

TASK D2 TASK D5 TASK C TASK D3

TASK D9

TASK D7: STRUCTURAL RESPONSE

TASK E: IMPACT ASSESSMENT

ExistingDesignCriteriaReview

DuctilityRequire-ments

StructuralAnalysis

SpatialVariation

SoilsBehavior

andLiquefaction

StructureImportance

SeismicDetailing

HazardExposureReview

FoundationsandSoil-

StructureInteraction

NationalHazard

Exposure

48

representation for the design ofguidelines and specifications.

The ground motion issues thathave emerged in recent years aspotentially important to the designof highway facilities and that wereconsidered in this work includedconsideration of the following:

• What should be the basis for thenational seismic hazard portrayalof highway facilities, and howshould this be implemented interms of design values?

• Can or should energy or dura-tion be used in a design proce-dure?

• How should site effects be char-acterized for design?

• Should vertical ground motionsbe specified for design?

• Should near-source ground mo-tions be specified for design?

The following summarize the keyelements of each issue and conclu-sions of the research.

Seismic Hazard PortrayalIn 1996, the U.S. Geological Sur-

vey (USGS) developed new seismicground shaking maps for the con-tiguous U.S. These maps depict con-tours of peak ground acceleration(PGA) and spectral accelerations(SA) at 0.2, 0.3, and 1.0 second (for5% critical damping) of ground mo-tions on rock for probabilities ofexceedance (PE) of 10%, 5%, and 2%in 50 years, corresponding to returnperiods of approximately 500, 1000,and 2500 years, respectively.

The research considered whetherthe new USGS maps should replaceor update the maps currently inAASHTO, which were developed bythe USGS in 1988. The key issueregarding whether the new USGSmaps should provide a basis for thenational seismic hazard portrayalfor highway facilities is the degreeto which they provide a scientifi-cally improved representation ofseismic ground motion. Based onan analysis of the process of devel-oping the maps, the inputs to themapping, and the resulting mapvalues, it was concluded that thesenew maps represent a major stepforward in the characterization ofnational seismic ground motion.The maps are in substantially bet-ter agreement with current scien-tific understanding of seismicsources and ground motion attenu-ation throughout the U.S. than thecurrent AASHTO maps. It wastherefore concluded that the newUSGS maps should provide the ba-sis for a new national seismic haz-ard portrayal for highway facilities.

The issue of an appropriate prob-ability level or return period for de-sign ground motions based on thenew USGS maps was also examined.Analyses were presented showingthe effect of probability level orreturn period on ground motionsand comparisons of ground mo-tions from the new USGS maps andthe current AASHTO maps. The re-search recommended that, for de-sign of highway facilities againstcollapse, consideration should be

Design and PerformanceCriteria• Review Existing Design

Criteria and Philosophies,C. Rojahn, R. Mayes,I. Buckle

• Impact Assessment andStrawman Guidelines forthe Seismic Design ofHighway Bridges,C. Rojahn, R. Mayes,I. Buckle

Seismic Hazardand Exposure• Compile and Evaluate Maps

and Other Representations,and Summarize AlternativeStrategies for Portraying theNational Hazard Exposureof the Highway System,M. Power

• Recommended Approachfor Portraying the NationalHazard Exposure,M. Power

Ductility Requirements• Establish Representative

Pier Types forComprehensive Study –Eastern & Western U.S.,J. Kulicki, R. Imbsen

• Physical and AnalyticalModeling to Derive OverallInelastic Response of BridgePiers, J. Mander

• Derive Inelastic DesignSpectra, R. Imbsen

Structure Importance• Evaluation of Structure

Importance, J. Kulicki

Technical Report: Seismic Hazard Exposure• Proceedings of the FHWA/MCEER Workshop on the National Representation of

Seismic Ground Motion for New and Existing Highway Facilities, edited by I.M.Friedland, M.S. Power and R.L. Mayes, NCEER-97-0010.


given to adopting probability lev-els for design ground motions thatare lower than the 10% probabilityof exceedance in 50 years that is cur-rently in AASHTO. This is consistentwith proposed revisions to the 1997NEHRP provisions for buildings, inwhich the new USGS maps for aprobability of exceedance of 2% in50 years (2,500 year return period)have been adopted as a collapse-prevention design basis. It was de-termined that, from a deterministicstandpoint, the 2%/50 year mapsprovided a good representation ofthe acceleration and force levelsthat had occurred in the 1800s inboth the New Madrid seismic zoneand in Charleston, South Carolina.

Consideration of Energy orDuration

At the present time, the energyor duration of ground motions isnot explicitly recognized in thedesign process for bridges or build-ings, yet many engineers are of the

opinion that the performance of astructure may be significantly af-fected by these parameters, in ad-dition to the response spectralcharacteristics of the ground mo-tion. As a result, it was concludedthat some measure of the energyof ground motions is important tothe response of a bridge, but, atpresent, there is no accepted designprocedure to account for energy. Itwas recommended that research inthis area should be continued to de-velop energy-based design methodsthat can supplement current elas-tic-response-spectrum-based designmethods. It was also concludedthat energy rather than duration isthe fundamental parameter affect-ing structural behavior.

Characterization of Site EffectsAt a Site Effects Workshop held

in 1992 at the University of South-ern California (USC), a revised quan-tification of site effects on responsespectra and revised definitions of

Foundations and Soil-Structure Interaction• Compile Data and Identify

Key Issues, I.P. Lam

• Abutment and Pile FootingStudies by CentrifugeTesting, R. Dobry

• Develop Analysis andDesign Procedures forAbutments, I.P. Lam,G. Martin

• Develop Analysis andDesign Procedures forRetaining Structures,R. Richards, K. Fishman

• Develop Analysis andDesign Procedures for PileFootings, I.P. Lam,G. Martin

• Develop Analysis andDesign Procedures forDrilled Shafts, I.P. Lam,G. Martin

• Develop Analysis andDesign Procedures forSpread Footings,G. Gazetas

• Performance andSensitivity Evaluation, andGuideline Development,P. Lam, R. Dobry,G. Martin

Soil Behavior andLiquefaction• Site Response Effects,

R. Dobry

• Identification ofLiquefaction Potential,T.L. Youd

• Development ofLiquefaction MitigationMethodologies, G. Martin

• Design Recommendationsfor Site Response andLiquefaction Mitigation,G. Martin

-120˚

-120˚

-110˚

-110˚

-100˚

-100˚

-90˚

-90˚

-80˚

-80˚

-70˚

-70˚

30˚ 30˚

40˚ 40˚

50˚ 50˚

2

2

4

4

4

4

4 66

6

6

6

6

6

6

8

8

8

8

8

8

8

8

10

10

10

10

10

10

10

10

10

10

20

20

20

20

20

20

20

20

2020

60

60

60

80

8080

Peak Acceleration (%g) with 2% Probability of Exceedance in 50 Years

site: NEHRP B-C boundary

U.S. Geological SurveyNational Seismic Hazard Mapping Project

-120˚

-120˚

-110˚

-110˚

-100˚

-100˚

-90˚

-90˚

-80˚

-80˚

-70˚

-70˚

30˚ 30˚

40˚ 40˚

50˚ 50˚

Nov. 1996

U.S. Geological Survey

� Figure 2. Peak ground accelerations represented as a percentage of g with a 2% probability ofexceedance in 50 years

50

site categories were proposed. Sub-sequently, these revised site factorsand site categories were adoptedinto the 1994 NEHRP provisionsand the 1997 Uniform BuildingCode (UBC). Since the develop-ment of these revised site factors,two significant earthquakes oc-curred (the 1994 Northridge and1995 Kobe earthquakes) whichprovided substantial additional datafor evaluating site effects on groundmotions, and research using thesedata has been conducted.

The site factors and site catego-ries in the current AASHTO specifi-cations are those that weresuperseded by the USC Workshoprecommendations in the NEHRPProvisions and the UBC. Under thisresearch, the question was whetherthe USC Workshop recommenda-tions should be utilized in charac-terizing ground motions forhighway facilities design andwhether they should be modifiedto reflect new data and new knowl-edge since the 1992 Workshop. Themost significant differences be-tween the USC Workshop recom-mendations and the previous sitefactors (those currently in AASHTO)are: (1) the revised site factors in-clude separate sets of factors for theshort-period and long-period partsof the response spectrum, whereasthe previous site factors were onlyfor the long-period part; (2) the re-vised site factors are dependent onrather than independent of inten-sity of ground shaking, reflectingsoil nonlinear response; and (3) therevised site factors are larger (i.e.,show a greater soil response ampli-fication) than the previous factorsat low levels of shaking, and are ap-propriate to the lower-seismicityregions in the U.S.

It was found that the post-Northridge and post-Kobe earth-quake research conducted to datewas supportive of the site factorsderived in the 1992 USC Workshop,although revisions to these factorsmight be considered as further re-search findings on site effects be-come available.

Vertical Ground MotionsAt present, the AASHTO specifi-

cations do not contain explicit re-quirements to design for verticalaccelerations. Ground motion datafrom many earthquakes in the past20 years have shown that, in thenear-source region, very high short-period vertical spectral accelera-tions can occur. For near-sourcemoderate-to-large magnitudeearthquakes, the rule-of-thumb ra-tio of 2/3 between vertical andhorizontal spectra is a poor descrip-tor of vertical ground motions. Atshort periods, the vertical-to-hori-zontal spectral ratios can substan-tially exceed unity, whereas at longperiods, a ratio of two-thirds may beconservative. It was demonstratedthat the profession’s current under-standing and ability to characterizenear-source vertical ground motionsis good, especially in the western U.S.where the near-source region is bet-ter defined (i.e., near mapped-activefaults). It was also demonstrated thathigh vertical accelerations as may beexperienced in the near-source re-gion can significantly impact bridgeresponse and design requirementsin some cases. On the basis of thesefindings, it was concluded that ver-tical ground motions should be con-sidered in bridge design in higherseismic zones for certain types ofbridge construction. However, spe-cific design criteria and procedures

Special SeismicDetailing• Capacity Detailing of

Columns, Walls, and Piersfor Ductility and Shear,J. Mander, R. Imbsen,J. Kulicki, M. Saiidi

• Capacity Detailing ofMembers to Ensure ElasticBehavior, J. Mander,R. Imbsen, J. Kulicki

• Detailing for StructuralMovements - Bridges,R. Imbsen, J. Kulicki

• Detailing for StructuralMovements - Tunnels,M. Power

• Structural Steel and Steel/Concrete Interface Details,J. Kulicki

Spatial Variation ofGround Motion• Spatial Variation of

Ground Motion,M. Shinozuka, G.Deodatis

Structural Response• Effects of Vertical

Acceleration on StructuralResponse, M. Button

Structural Analysis• Review Existing Analytical

Methods, and Identify andRecommend AnalyticalProcedures Appropriate forEach Structure Categoryand Hazard Exposure,I. Buckle, J. Mander


still need to be developed for cer-tain bridge types.

Near-source Ground MotionsThe characteristics and the ef-

fects of near-source ground mo-tions on bridge response wereexamined. As the distance to anearthquake source decreases, theintensity of ground motions in-creases, and this increase in groundmotion intensity is incorporated innew USGS maps. However, in addi-tion to their higher intensity, near-source ground motions havecertain unique characteristics thatare not found at greater distances.The most significant characteristicappears to be a large pulse of long-period ground motions when anearthquake rupture propagates to-ward a site. Furthermore, this pulseis larger in the direction perpen-dicular to the strike of the fault thanin the direction parallel to the strike.This characteristic of near-sourceground motions has been observedin many earthquakes, including mostrecently in the Northridge, Kobe,Turkey and Taiwan earthquakes.Preliminary analyses of bridge re-sponse indicate that near-sourceground motions may impose unusu-ally large displacement demands onbridge structures. It was thereforeconcluded that traditional groundmotion characterizations (i.e., re-sponse spectra) may not be ad-equate in describing near-sourceground motions, because thepulsive character of these motionsmay be more damaging than indi-cated by the response spectra ofthe motions. Recommendations

include the need for additional re-search to evaluate more fully theeffects of near-source ground mo-tions on bridge response and toincorporate these effects in codedesign procedures. Until adequateprocedures are developed, consid-eration should be given to evalu-ating bridge response usingsite-specific analyses with repre-sentative near-source accelerationtime histories.

Spatial Variation of GroundMotion

The objective of the research inthis area was to develop proceduresfor determining spectrum compat-ible time histories that adequatelyrepresent spatial variations inground motion including the ef-fects of different soil conditions.The procedures were then used toexamine the effects of spatial vari-ability on critical response quanti-ties for typical structures.

The methodology used a spectralrepresentation to simulate stochas-tic vector processes having compo-nents corresponding to differentlocations on the ground surface. Aniterative scheme was used to gen-erate time histories compatiblewith prescribed response spectra,coherency, and duration of motion.Analysis results for eight examplebridges were tabulated, showingthe relative ductility demand ratiofor column flexure due to seismicwave propagation spatial effects. Ingeneral, there was about a 10%maximum increase when using

Summary Report: Spatial Variation of Ground Motion• Effect of Spatial Variation of Ground Motion on Highway Structures, by M.

Shinozuka and G. Deodatis, Agency Final Report.

52

linear analysis, and a 25% maximumincrease when using nonlinearanalysis for bridges up to 1000 feetin length. Results were also tabu-lated for relative opening and clos-ing at expansion joints for bridgeswith superstructure hinges. In gen-eral, the relative joint openingmovement was up to two timeswhen using either linear or nonlin-ear analysis for bridges up to 1000feet in length. However, the con-ducted analyses were too limited inscope on which to base specificguidance for a large variety ofbridge types and conditions at thistime, and it was recommended thatfurther studies are required beforecode language could be developed.

Inelastic Design Spectrum

The research in this area had theobjective of developing inelastic re-sponse spectrum which would al-low designers to assess the inelasticdeformation demands, ultimatelyleading to improved seismic perfor-mance for new bridge construction.The spectrum are being derived for

nationwide use, accom-modating different seis-mic environments andsite soil conditions. Theyare also being devel-oped for design applica-tions, by accounting forscattering and variabili-ties that exist in realearthquake ground mo-tions and for nonlinearstructural response.

Under the currentprogram, the researchhas not progressed tothe point where its re-sults are ready forimplementation. Whencomplete, it is likely tohave a major impact on

seismic design code requirementsfor bridges, as inelastic spectra willbe one of the key elements in a dis-placement-based or energy-baseddesign procedure. Future work inthis area should include proceduresfor determining inelastic spectra ata specific site. The current state ofresearch provides an approximatemethod that starts with an elasticspectrum rather than time history;as time histories for the eastern U.S.are currently lacking, this approachwill have an obvious appeal.

Foundation Designand Soil Behavior

Research tasks in this area inves-tigated and improved criteria forthe design and analysis of majorfoundation elements includingabutments, retaining walls, pile andspread footings, and drilled shafts.In addition, work was performed onsoil liquefaction and lateral spreadidentification and mitigation.

� Figure 3. Research in spatial variation of groundmotion included case studies of the SR14/I-5interchange and other bridges damaged during theNorthridge, California earthquake. Further research isneeded to define the importance of these effects and todevelop simplified design procedures for a broadrange of bridge configurations.

Photographs courtesy of I.G. Buckle


Abutments and Retaining Walls

Research on bridge abutmentsand retaining walls focused on mod-eling alternatives, clarifying the pro-cess of design for service loadsversus seismic loading, and provid-ing simplified approaches for designthat incorporate key issues affect-ing seismic response. The researchalso attempted to provide a newprocedure for determining the seis-mic displacements of abutmentsand retaining walls founded onspread footings, which differ fromcurrent procedures by addressingmixed-mode behavior (i.e., includ-ing rotation due to bearing capacitymovement and sliding response).Both experimental and analyticalstudies were conducted; the experi-mental studies included sand-boxexperiments on a shaking table andcentrifuge models. Results of thisresearch included:

• Development of a simplifiedprocedure for estimating abut-ment stiffness. A key element ofthis approach is determining theportion of the wall that can berelied on to mobilize backfill re-sistance.

• Extending current AASHTO de-sign procedures to the more gen-eral case of translation androtation of walls and abutments.The results are presented in amanner that will allow the meth-ods to be easily introduced intoa future code revision. The newprocedures will be of greatestuse for free-standing gravitywalls and for active mode abut-ment and wall movements.

• Consideration of passive loadingconditions for walls and abut-ments. Current AASHTO provi-sions only address active loadingconditions. Since passive load-ing can result in forces that areup to 30 times those for activeconditions, there is a strong pos-sibility that many bridges willnot develop passive resistancewithout abutment damage.

Pile and Spread Footings, andPile Groups

Studies on pile footings, spreadfootings, and pile groups includedexperimental and analytical re-search which was intended to: (a)provide improved understanding of

Technical and Summary Reports: Foundation Design and Soil Behavior• Foundations and Soils - Compile Data and Identify Key Issues, by I.P. Lam,

Agency Final Report.• Centrifuge Modeling of Cyclic Lateral Response of Pile-Cap Systems and Seat-

Type Abutments in Dry Sands, by A. Gadre and R. Dobry, MCEER-98-0010.• Modeling of Bridge Abutments in Seismic Response Analysis of Highway Bridges,

I.P. Lam and M. Kapuskar, Agency Final Report.• Seismic Analysis and Design of Bridge Abutments Considering Sliding and Ro-

tation, by K.L. Fishman and R. Richards, Jr., NCEER-97-0009.• Modeling of Pile Footings and Drilled Shafts for Seismic Design, by I.P. Lam, M.

Kapuskar and D. Chaudhuri, MCEER-98-0018.• Development of Analysis and Design Procedures for Spread Footings, by G.

Gazetas, G. Milonakis and A. Nikolaou, Agency Final Report.• Synthesis Report on Foundation Stiffness and Sensitivity Evaluation on Bridge

Response, by I. P. Lam, G.R. Martin, and M. Kapuskar, Agency Final Report.

MCEER Highway Projecthttp://mceer.buffalo.edu/

research/HighwayPrj/default.html

MCEER Publicationshttp://mceer.buffalo.edu/

publications/default.asp

54

the lateral response of pile-capfoundations; (b) evaluate the influ-ence of modeling parameters onestimated displacement and forcedemands; (c) recommend methodsfor characterizing the stiffness of pilefootings; (d) quantify the importanceof radiation damping and kinematicinteraction on response; and (e)evaluate conditions under whichuplift becomes significant and howbest to model uplift in a design pro-cedure. Results from these studiesincluded the following:

• For pile-cap systems, the re-search demonstrated that designprocedures should use simpleadditions for the contributionfrom the base, side and active/passive ends when estimatingthe lateral capacity of embeddedspread footings in dense sand,along with elastic solutions withan equivalent linear soil shearmodulus at shallow depths toestimate the secant stiffness ofthe footing. This effectively con-firms that existing procedurescan be used to obtain reasonableapproximations of pile-cap foun-dation response, as long as con-sideration is given to the levelsof deformation and embedmentfor the system.

• Axial and lateral loading re-sponse and stiffness character-istics are important parametersfor the design of single piles andpile groups, although such infor-mation is not currently ad-dressed in the AASHTOprovisions. Axial response oftencontrols rocking response of apile group. New proceduresand simplified stiffness chartsare provided for determiningthe lateral load-deflection char-acteristics of single piles andgroups.

• Nonlinear load-deflection analy-ses illustrate the sensitivity of re-sults to uncertainties in p-ystiffness, gapping, pile-head fixity,bending stiffness parameters, andembedment effects. The analyseshave demonstrated that load-de-formation response is more af-fected by input variations than bythe moment within the pile.

• For spread footings without up-lift, the research demonstratedthat (a) ignoring soil-structureinteraction reduces the funda-mental period of the system, re-sulting in higher accelerations;(b) increasing the effectivenessof embedment increases radia-tion damping and reduces thefundamental period of the sys-tem; and (c) neglecting radia-tion damping has only a minoreffect on the system. Uplift ofthe spread footing results in asofter mode of vibration for thesystem, with increasing funda-mental period as the amount ofuplift increases.

Drilled Shafts

Research on drilled shafts wasconducted in order to provide infor-mation on the influence of model-ing procedures on the response ofthe structure, evaluate the effects ofmodeling on estimated displace-ment and force demands on thefoundation, and to summarize meth-ods for characterizing the responseof drilled shaft foundations, includ-ing their limitations. Results of thiswork included the following:

• Foundation stiffness has beenshown as a key parameter andcontributor to the dynamic re-sponse of the structure, necessi-tating realistic estimates andappropriate integration into a

“Among themajor studiesconductedunder thisproject was areview, syn-thesis, andimprovement torecent devel-opments insimplifiedprocedures forevaluating theliquefactionresistance ofsoils, and thecompilationand evaluationof case studiesand proceduresfor groundliquefactionmitigation.”


detailed structural analysis. Theresponse of a soil-foundation sys-tem to load is nonlinear; how-ever, for practical purposes, anequivalent linear representationis normally used.

• Guidance is provided on the de-velopment of equivalent linearand nonlinear stiffness values,and the importance and sensi-tivity of foundation geometryand boundary conditions at theshaft head are identified. A keyconclusion is that realistic rep-resentation of pile-head fixitycan lead to a much more eco-nomical design.

• The p-y approach is recognizedas the most common method ofanalyzing the nonlinear re-sponse of the shaft to lateralload. Parameters that must beconsidered include the effects ofsoil property variation, degrada-tion effects, embedment, gap-ping, and scour effects.

Liquefaction Processes andLiquefaction MitigationMethodologies

A significant amount of researchwas conducted under the MCEER

Highway Project on liquefactionprocesses, screening for liquefactionpotential, and the development and/or improvement of liquefaction miti-gation methodologies. Much of thiswork was conducted under a com-panion FHWA contract on the seis-mic vulnerability of existingtransportation infrastructure (FHWAContract DTFH61-92-C-00106), butmuch of it is appropriate for bothnew design and existing structureevaluation. Among the major stud-ies conducted under this projectwas a review, synthesis, and im-provement to recent developmentsin simplified procedures for evalu-ating the liquefaction resistance ofsoils, and the compilation and evalu-ation of case studies and proce-dures for ground liquefactionmitigation. Results of this researchincluded:

• Identification of a consensussimplified procedure for evalu-ating liquefaction resistance.Minor modifications for the de-termination of the stress reduc-tion factor used in the calculationof the cyclic stress ratio wererecommended, which allow thestress reduction factor to be

� Figure 4. Researchers reviewed three procedures for representing foundation stiffness in the modeling process: (a)coupled foundation stiffness matrix, (b) equivalent cantilever approach; and (c) uncoupled base-spring model.

56

calculated to depths greater than30 meters.

• Identification of the latest pro-cedures for determining cyclicresistance ratios (CRR) usingcone penetration test (CPT) pro-cedures. One of the primaryadvantages of CPT is the consis-tency and repeatability of themethod. Plots for determiningthe liquefaction resistance di-rectly from CPT data, rather thanconverting to an equivalent stan-dard penetration test (SPT) N-value, are presented. Proceduresare also provided for correctingCPT data based on overburdenpressures, fines contents, and forthin layers.

• Plots for determining CRR fromshear wave velocity data havebeen prepared, and proceduresfor correcting shear wave veloc-ity data due to overburden stressand fines content are explicitlygiven.

• Methods which have been em-ployed successfully for liquefac-tion mitigation include deepdynamic compaction, deep vi-bratory densification, graveldrains, permeation grouting, re-placement grouting, soil mixing,and micro blasting. Parametersand limitations for each of theseapproaches are summarized,including typical treatmentdepths and applicable soil types.

• Flow charts for assessing grounddeformations for pre- and post-treatment conditions were de-veloped. These are accompaniedby recommendations for pre-ferred ground improvementsmethods based on differing siteconditions.

• Development of rapid screen-ing procedures for liquefactionsusceptibility of soils and foun-dations.

Structural Importance,Analysis and Response

Studies were conducted in orderto provide a definition of structuralimportance, which is necessary inthe development of design and per-formance criteria, and to evaluatemethods of analysis and structuralresponse. Structural importanceanalysis response studies also pro-vided a synthesis of current systemsand details commonly used to pro-vide acceptable seismic perfor-mance in various states and regions.Among the findings of these stud-ies were the following:

• Provisions employed by the Cali-fornia Department of Transpor-tation (Caltrans) were generallymore rigorous than those usedby the majority of states (whoprimarily used current AASHTOprovisions). However, adoption

Technical and Summary Reports: Soil Behavior and Liquefaction• Site Factors and Site Categories in Seismic Codes, by R. Dobry, R. Ramos and M.

Power, MCEER-99-0010.• Proceedings of the NCEER Workshop on Evaluation of Liquefaction Resistance

of Soils, edited by T.L. Youd and I.M. Idriss, NCEER-97-0022.• Development of Liquefaction Mitigation Methodologies/Ground Densification

Methods, by G. Martin, Agency Final Report.• Design Recommendations for Site Response and Liquefaction, by G. Martin,

Agency Final Report.

“Structuralimportance,analysis, andresponse studiesprovided asynthesis ofcurrent systemsand detailscommonly usedto provideacceptableseismicperformance invarious statesand regions.”


of Caltrans’ design provisionsnationwide would likely compli-cate designs and add to con-struction cost; this may beunjustified for many low-to-mod-erate seismic hazard states. Inaddition, if Caltrans’ experienceis to be adopted nationally, someadjustments are required in or-der to accommodate bridgetypes and details commonlyused elsewhere.

• Studies that were conducted onthe application of advancedmodeling methods for concretebridge components provided acomputer program which deter-mines moment-curvature andforce-deflection characteristicsfor reinforced concrete columns;excellent correlation was ob-tained between analytical andexperimental test results forthese components.

• A refined model to simulate thehysteretic behavior of confinedand unconfined concrete inboth cyclic compression andtension was developed. Themodel includes consideration ofthe nature of degradation withinpartial hysteresis looping andthe transition between openingand closing cracks.

• A study on energy and fatiguedemands on bridge columns re-sulted in design recommenda-tions for the assessment offatigue failure in reinforcingsteel, based on the results ofnonlinear dynamic analyses.

This methodology incorporatestraditional strength and ductilityconsiderations with the fatiguedemands. Based on parametricstudies, it was concluded that lowcycle fatigue demand is bothearthquake and hysteretic modeldependent.

• Based on an examination of ex-isting and proposed methods forquantifying bridge importance,a specific method was selectedand tested against a database ofbridge information commonlyavailable within the FHWA’s Na-tional Bridge Inventory. Onelimitation of the study is that itdeliberately avoided addressingpolitical and economic issuesrelated to bridge seismic designcriteria and highway networkconsiderations.

• Following the Northridge earth-quake, concerns were raised asto the role vertical accelerationsmay have played in causingdamage to one or more bridges.In a study conducted to investi-gate the effects of vertical ac-celeration on bridge response,preliminary results indicate thatvertical components of groundmotion could have a significanteffect on bridge response forstructures within 10 km of thefault, and even within 20 – 30km for certain conditions. Theresults of this study will be con-troversial when publicized; how-ever, a far too limited study wasconducted (only six example

Technical Reports: Structural Importance, Analysis and Response• Effect of Vertical Ground Motions on the Structural Response of Highway

Bridges, by M. Button, C. Cronin and R. Mayes, MCEER-99-0007.• Methodologies for Evaluating the Importance of Highway Bridges, by A. Tho-

mas, S. Eshenaur and J. Kulicki, MCEER-98-0002.

58

bridges were analyzed) in orderto provide definitive guidance atthis time.

• In a study which investigated theapplicability of simplified analy-sis methods to various types andconfigurations of bridges, a num-ber of design and analysis limita-tions were identified. Parametersevaluated included curvature,span length ratio, pier height,skew and span connectivity.Based on the analyses, a definitionfor “regular” bridges and forwhich simplified methods areappropriate was developed. Ingeneral, regular bridges musthave three or fewer spans, varia-tion of mass distribution betweenadjacent spans varying by lessthan 50%, a maximum ratio be-tween adjacent pier stiffnesses inthe longitudinal and transversedirections not greater than 4.0,and a subtended angle in plan notgreater than 90°.

Structural DesignIssues and Details

A number of studies were con-ducted in order to improve designprocedures and structural detailingfor highway structures, but the fo-cus was primarily on bridges; onestudy also examined movementdetailing for tunnels. These studieslooked at issues of capacity detail-ing for ductility, elastic behavior, andmovements. Among the results ofthis research were the following:

• A design concept termed Dam-age Avoidance Design (DAD) wasdeveloped which attempts toavoid plastic hinging in columns,thereby avoiding loss of servicefor important bridges followinga major earthquake. The conceptevaluated details which providefor rocking of columns and piers,which rotate about their ends butare restrained from collapsethrough gravity and the optional

Photographs courtesy of J. Mander

� Figure 5. Researchers developed a new design concept called Control and Repairability of Damage (CARD) for bridge column hinges.The CARD method controls damage while accommodating large earthquake-induced deformations.

Undamaged specimen (prior to testing) Damaged specimen (after testing)


use of central unbonded posttensioning in the column core.

• A second design concepttermed Control and Repairabil-ity of Damage (CARD) was alsodeveloped, which providedstructural and construction de-tails for reinforced concrete col-umns that provide replaceableor renewable sacrificial plastichinge zone components. In thisconcept, the hinge zones are de-liberately weakened and re-gions outside the hinge zones

are detailed to be stronger thanthe sacrificial (fuse) zone; the re-maining elements of the struc-ture then remain elastic duringstrong earthquakes.

• In a study on transverse reinforc-ing requirements for concretebridge columns and pier walls,it was found that the currentAASHTO requirements could belowered by up to 50% while stillachieving displacement ductili-ties of 4 to 7 for bridges in lowto moderate seismic zones. An

Technical and Summary Reports: Structural Design Issues and Details• Application of Simplified Methods of Analysis to the Seismic Design of Bridges,

by J.H. Kim, M.R. Button, J.B. Mander and I.G. Buckle, Agency Final Report.• Establish Representative Pier Types for Comprehensive Study: Eastern U.S., by

J. Kulicki and Z. Prucz, NCEER-96-0005.• Establish Representative Pier Types for Comprehensive Study: Western U.S., by

R.A. Imbsen, R.A. Schamber, and T.A. Osterkamp, NCEER-96-0006.• Seismic Resistance of Bridge Piers Based on Damage Avoidance Design, by J.B.

Mander and C.T. Cheng, NCEER-97-0014.• Seismic Design of Bridge Columns Based on Control and Repairability of Dam-

age, by C.T. Cheng and J.B. Mander, NCEER-97-0013.• Capacity Design and Fatigue Analysis of Confined Concrete Columns, by A. Dutta

and J.B. Mander, MCEER-98-0007.• Capacity Design of Bridge Piers and the Analysis of Overstrength, by J.B. Mander,

A. Dutta, and P. Goel, MCEER-98-0003.• Seismic Energy Based Fatigue Damage Analysis of Bridge Columns: Part I –

Evaluation of Seismic Capacity, by G.A. Chang and J.B. Mander, NCEER-94-0006.• Seismic Energy Based Fatigue Damage Analysis of Bridge Columns: Part II –

Evaluation of Seismic Demand, by G.A. Chang and J.B. Mander, NCEER-94-0013.• Ductility of Rectangular Reinforced Concrete Bridge Columns with Moderate Con-

finement, by N. Wehbe, M. Saiidi, D. Sanders and B. Douglas, NCEER-96-0003.• Capacity Detailing of Members to Ensure Elastic Behavior, by R.A. Imbsen, R.A.

Schamber, and M. Quest, Agency Final Report.• Capacity Detailing of Members to Ensure Elastic Behavior - Steel Pile-to-Cap

Connections, by P. Ritchie and J. M. Kulicki, Agency Final Report.• Structural Steel and Steel/Concrete Interface Details for Bridges, by P. Ritchie,

N. Kauhl and J. Kulicki, MCEER-98-0006.• Structural Details to Accommodate Seismic Movements of Highway Bridges and

Retaining Walls, R.A. Imbsen, R.A. Schamber, E. Thorkildsen, A. Kartoum, B.T.Martin, T.N. Rosser and J.M. Kulicki, NCEER-97-0007.

• Derivation of Inelastic Design Spectrum, by W. D. Liu, R. Imbsen, X. D. Chen andA. Neuenhofer, Agency Final Report.

• Summary and Evaluation of Procedures for the Seismic Design of Tunnels., by M. S.Power, D. Rosidi, J. Kaneshiro, S. D. Gilstrap, and S.-J. Chiou, Agency Final Report.

“Studies instructuraldesign issuesand detailslooked atcapacitydetailing forductility, elasticbehavior, andmovements.”

60

important aspect of this workwas that the end anchorages fortransverse steel hoops must bemaintained for the reinforcing tobe effective; 90° bends on J-hookswere found to be inadequate.

• Research was conducted onmoment overstrength capacityin reinforced concrete bridgecolumns, and a simplifiedmethod for determining columnoverstrength was developed. Theupper-bound overstrength fac-tors developed in this task vali-date prescriptive overstrengthfactors recommended in ATC-32,but also indicate that some fac-tors in current Caltrans andAASHTO provisions may be toolow.

• A synthesis was conducted ondetails commonly used to ac-commodate expected move-ments on bridges and retainingwalls in the eastern and westernU.S. Based on the synthesis, de-sign and detailing recommenda-tions were made in order toprovided the basis for improvedbridge design standards. Thespecific elements considered inthis effort included restrainingdevices, sacrificial elements, pas-sive energy dissipation devices,and isolation bearings. A similareffort was conducted on move-ment criteria and detailing fortunnels.

• For steel superstructures, a num-ber of issues were considered,including ductility based on

cross-section configuration, ap-plicability of eccentrically-braced frames, details whichallow for easy repair of steel sec-tions following a moderate tolarge earthquake, anchor boltperformance under lateral upliftloads, and economical momentconnection details betweensteel superstructures and con-crete substructures.

Implementation ofResearch Results

In the case of highway bridges,seismic design provisions are con-tained in the two AASHTO bridgespecifications: LRFD Bridge DesignSpecifications (2nd Edition, 1998)and Division I-A, Seismic Design, ofthe Standard Specifications forHighway Bridges (16th Edition,1996). In 1997, AASHTO recog-nized that there had been many im-portant advances in the knowledgeof earthquake hazard, bridge seismicperformance, and design and detail-ing. As a result, AASHTO chargedthe Transportation Research Board’sAASHTO-sponsored National Coop-erative Highway Research Program(NCHRP) with conducting a projectto update and revise the bridge seis-mic design specifications containedin the LRFD Bridge Design Specifi-cations.

NCHRP Project 12-49, “Compre-hensive Specification for the SeismicDesign of Bridges,” was initiated in

Technical Reports: Implementation of Research Results• Impact Assessment of Selected MCEER Highway Project Research on the Seismic

Design of Highway Structures, by C. Rojahn, R. Mayes, D.G. Anderson, J.H. Clark,D’Appolonia Engineering, S. Gloyd and R.V. Nutt, MCEER-99-0009.

• Seismic Design Criteria for Bridges and Other Structures, by C. Rojahn, R. Mayes,D.G. Anderson, J. Clark, J.H. Hom, R.V. Nutt and M.J. O’Rourke, NCEER-97-0002.


August 1998. The objective ofNCHRP Project 12-49 is to developnew specifications for the seismicdesign of highway bridges, whichcan be incorporated into the LRFDBridge Design Specifications.These new specifications will benationally applicable with provi-sions for all seismic zones. The re-sults of research currently inprogress or recently completed,along with current demonstratedpractice, are the principal re-sources for this project, especiallywith respect to the research con-ducted under the FHWA-sponsoredMCEER Highway Project.

The design criteria being devel-oped under NCHRP Project 12-49will address the following: (1)strength-based and displacement-based design philosophies; (2)single- and dual-level performancecriteria; (3) acceleration hazardmaps and spectral ordinate maps;(4) spatial variation effects; (5) ef-fects of vertical acceleration; (6) siteamplification factors; (7) inelasticspectra and use of response modi-fication factors; (8) equivalent staticnonlinear analysis methods; (9)modeling of soil-structure interac-tion and structural discontinuitiesat expansion joints; (10) durationof the seismic event; and (11) de-sign and detailing requirements forboth steel and concrete super- andsubstructures.

A joint venture of the AppliedTechnology Council and MCEERwas selected by the NCHRP to con-duct Project 12-49. In the firstphase of the project, the basic phi-losophy for the new specificationshas been developed, along with rec-ommendations regarding represen-tation of seismic hazard for design,and minimum design and perfor-mance criteria for typical highwaybridges and bridge components.

The work conducted previously byMCEER, Caltrans, and others leaddirectly to the development of thisnew specification’s philosophicalframework.

The time schedule for NCHRPProject calls for the completion ofa first draft of the specification andcommentary by the end of 1999,and subsequent drafts and revisionscompleted during 2000. The targetdate for submission of a recom-mended specification to AASHTOby NCHRP is early 2001.

ConclusionAs a result of a research program

sponsored by the Federal HighwayAdministration, researchers work-ing for the Multidisciplinary Centerfor Earthquake Engineering Re-search have developed a number ofanalytical tools, methods of analy-sis, structural design details, andspecification recommendations ap-propriate for seismic design of high-way systems and structures. Theprimary focus of this work has beenon highway bridges, but some re-search on tunnels and retainingstructures was also performed. Theprogram has also resulted in recom-mendations regarding the represen-tation of seismic hazard in futuredesign codes, the performance andimprovement of soils under seismicshaking, and an improved under-standing of the behavior of struc-tural systems and componentsunder seismically-induced forcesand displacements. In addition, itis likely that additional recommen-dations regarding the use of a per-formance-based design philosophyand dual-level design and perfor-mance criteria will be made toAASHTO as a result of this work.

“MCEER hasdeveloped anumber ofanalyticaltools, methodsof analysis,structuraldesign details,andspecificationrecommendationsappropriate forseismic designof highwaysystems andstructures.”

62

Rojahn, C., Mayes, R., Anderson, D.G., Clark, J.H., D’Appolonia Engineering, Goyd, S. andNutt, R.V., 1999, Impact Assessment of Selected MCEER Highway Project Researchon the Seismic Design of Highway Structures, MCEER-99-0009, April.

MCEER, 1999, Seismic Vulnerability of New Highway Construction, Final Report/Extended Technical Summary, Federal Highway Administration.

Friedland, I.M., Mayes, Ronald L., Yen, Phillip and O’Fallon, John, 2000, “Highway BridgeSeismic Design: How Current Research May Affect Future Design Practice,” TRB BridgeConference, Paper Number 5B0042.

References

63

Research Objectives

The objectives of this task are both to conduct research on seismichazards, and to provide relevant input on the expected levels of thesehazards to other tasks. Other tasks requiring this input include thosedealing with inventory, fragility curves, rehabilitation strategies and dem-onstration projects. The corresponding input is provided in various for-mats depending on the intended use: either as peak ground motionparameters and/or response spectral values for given magnitude, epicen-tral distance and site conditions; or as time histories for scenario earth-quakes that are selected based on the disaggregated seismic hazardmapped by the U.S. Geological Survey (Frankel, 1995; Harmsen, et al.1999) and used in the NEHRP Recommended Provisions for the Devel-opment of Seismic Regulations for New Buildings (BSSC, 1998).

Prediction of the seismic hazard in tectonic regions of moderate-to-lowseismicity is a difficult task because of the paucity of data that are

available regarding such regions. Specifically, the problem of earthquakeground motion prediction in Eastern North America (ENA) is hindered bytwo factors: (1) the causative structures of seismicity in ENA are largelyunknown, and (2) the recorded strong motion database is very limited (ifit exists at all), especially for moderate to large magnitude (M

W ≥ 6) events

virtually for all epicentral distances. For these reasons, prediction of strongground motion in ENA makes the use of well-founded physical modelsimperative. These models should, among other things, provide the meansto make extrapolations to large magnitudes and/or short distances withconfidence. [This is in contrast to the western U.S. (WUS), where theabundance of recorded strong motion data makes prediction of groundmotion by empirical methods a viable procedure].

Another distinct feature of the tectonic region of ENA is related to itsscattering and absorption characteristics. As pointed out by Aki (1982),because of the extremely low attenuation in ENA as compared to California(an order of magnitude difference in the quality factor Q at around 1 Hz),

Ground Motion PredictionMethodologies forEastern North America

by Apostolos S. Papageorgiou


Apostolos Papageorgiou,Professor, Department ofCivil, Structural andEnvironmentalEngineering, University atBuffalo, State University ofNew York

Gang Dong, Post-doctoralResearch Associate,Department of Civil,Structural andEnvironmentalEngineering, University atBuffalo, State University ofNew York

Benedikt Halldorsson,Graduate Student,Department of Civil,Structural andEnvironmentalEngineering, University atBuffalo, State University ofNew York

64

seismic waves traveling along mul-titudes of paths through the hetero-geneous lithosphere of the Eartharrive at an observation site withsignificant amplitude. This effecttends to make the duration of shak-ing longer, although the source du-ration for ENA earthquake sourcesis shorter. It is therefore very im-portant for ground motion synthe-sis purposes to study the scatteringand absorption characteristics ofthe lithosphere in ENA.

Strong MotionSynthesis Techniques

Our task is the synthesis of strongground motion input over the en-tire frequency range of engineeringinterest. There are two approachesfor modeling earthquake strongmotion:

1. The Stochastic (Engineering)Approach, according to which,earthquake motion (accelera-tion) is modeled as Gaussiannoise with a spectrum that iseither empirical [e.g., band-lim-ited white-noise (e.g., Cornell,1964; Shinozuka and Sato, 1967),Kanai-Tajimi type of spectrum(e.g., Housner and Jennings,1964) or variations of it such asthe Clough-Penzien type of spec-trum (Clough and Penzien,1993)], or a spectrum that is

based on a physical model (suchas the “Specific Barrier Model”)of the earthquake source. Thisapproach is expedient and there-fore cost-effective, and has beenextensively used in the past byengineers (using empirical spec-tra) and recently by seismolo-gists (using spectra derived fromphysical models of the source).The intent of this approach tostrong motion simulation is tocapture the essential character-istics of high-frequency motionat an average site from an aver-age earthquake of specified size.Phrasing this differently, theaccelerograms artificially gener-ated using the Engineering Ap-proach do not represent anyspecific earthquake, but embodycertain average properties ofpast earthquakes of a given mag-nitude.

2. The Kinematic Modeling Ap-proach was developed by seis-mologists. In this approach, therupture process is modeled bypostulating a slip function on afault plane and then using theElastodynamic RepresentationTheorem to compute the mo-tion. There are several variantforms of this approach depend-ing on whether the slip func-tion (i.e., the function thatdescribes the evolution of slipon the fault plane) and/or the

The user community for this research is both academic re-searchers and practicing engineers who may use the seismicinput generated by the synthesis techniques that are developedunder this task for a variety of applications. These includeground motions for scenario earthquakes, for developing fra-gility curves and in specifying ground motion input for criticalfacilities (such as hospitals) located in the eastern U.S.

MCEER Program 1: SeismicEvaluation and Retrofit ofLifeline Systems

MCEER Program 2: SeismicRetrofit of Hospitals

65Ground Motion Prediction Methodologies for Eastern North America

Green functions are syntheticor empirical. The KinematicModeling Approach involves theprediction of motions from afault that has specific dimen-sions and orientation in a speci-fied geologic setting. As such,this approach more accuratelyreflects the various wave propa-gation phenomena and is usefulfor site-specific simulations.

For both approaches, we haveadopted the “Specific BarrierModel,” proposed and developed byPapageorgiou and Aki (1983a, b), torepresent the earthquake source.Various source models have beenproposed in the literature for ENA[such as the “�2-model” (Boore andAtkinson, 1987), the “multiple epi-sode/fractional stress drop model”by Haddon (1995,1996), and the“two corner frequency empiricalspectral model” by Atkinson(1993)]. The advantage of the Spe-cific Barrier Model is that it pro-vides the most complete, yet

parsimonious, self-consistent de-scription of the faulting processesthat are responsible for the genera-tion of the high frequencies, and atthe same time provides a clear andunambiguous way of how to distrib-ute the seismic moment on the faultplane. The latter requirement isnecessary for the implementationof the Kinematic Modeling Ap-proach described above.

According to the Specific BarrierModel, the seismic source consistsof an aggregate of circular cracks(sub-events) of equal diameter 2�

0,

filling up a rectangular fault planeof length L and width W, as shownin Figure 1. As the rupture frontsweeps the fault plane with the“sweeping velocity” V, a stressdrop �� (referred to as the “localstress drop”) takes place in eachcrack starting from its center orfrom a point of its periphery andspreading with a “spreading veloc-ity” v. The region between the cir-cular cracks represents barriers left

ρ0

RuptureVelocity V

W

L

� Figure 1. The Specific Barrier Model consists of an aggregate of circular cracks of equal diameter distributed on arectangular area

66

unbroken after the passage of therupture front. The ruptures of indi-vidual cracks are statistically as-sumed to take place independently.Thus, the Specific Barrier Modelis a hybrid of deterministic and sto-chastic models and is described byfive parameters, namely, L, W, V ( =v), 2�

0 and ��. Rise time � (i.e.,

the time that it takes for a point onthe fault plane to complete its slip),which is also a very importantsource parameter in kinematic mod-eling, may be estimated as follows:� ≈ (1 ~ 2)(v/2�

0).

After we calibrated the SpecificBarrier Model using the availablestrong motion data base (Figure 2),we used the source spectrum (i.e.,the spectrum of the elastic wavesradiated by the source before thesehave been modified by the propa-gation path and site effects) of themodel (Papageorgiou and Aki,1985; Papageorgiou, 1988) in com-bination with Random VibrationTheory (or Stochastic ProcessTheory) (e.g., Boore, 1983) to predict

expected peak values of variousmeasures of strong ground motionuseful in aseismic design such aspeak acceleration a

max, peak veloc-

ity vmax

, peak displacement dmax

,Spectral Acceleration SA, PseudoSpectral Velocity PSV, etc., for anycombination of magnitude, epicen-tral distance and site conditions(Figure 3). The amplification func-tions to account for the differentsite conditions were selected con-sistent with the site classes of theNEHRP Provisions (Boore andJoyner, 1997). The advantage of theabove approach is that the ex-pected peak values mentionedabove are obtained without havingto resort to the generation of nu-merous realizations and then aver-age the results. These results shouldbe useful, for example, to all theresearchers working on Loss Esti-mation.

We have used the source spec-trum of the Specific Barrier Modelin the Stochastic Approach describ-ed above to synthesize realizations

� Figure 2. Comparison of predicted and observed 5% damped pseudo-velocity response spectra (PSV) for 1 Hz and2 Hz oscillators

101 102 103

104

103

102

101

100

101

r (km)

PS

V (

cm/s

)

1 Hz

101 102 103

104

103

102

101

100

101

r (km)

PS

V (

cm/s

)

2 Hz

a) b)


of ground motion for a given mag-nitude, epicentral distance and siteconditions. An example of such arealization is shown in Figure 4 fora M

w 6.5 earthquake event at an epi-

central distance R = 35 km on hardrock.

The stochastic synthesis ofaccelerograms, as was originallyused by engineers or even in itsmost refined form proposed by seis-mologists (e.g., Boore, 1983), hasthe following limitations: (1) Themodel is based on a point source;(2) the model is not adequate forsimulating the long-period near-field effect expected in the near-source region from the slip on thefault; and (3) the model provides adescription of the temporal varia-tion of ground motion of a singlepoint of the ground but cannot pro-vide a description of the spatialvariability of ground motion which � Figure 4. Synthetic acceleration, velocity and displacement time histories for a

moment magnitude Mw 6.5 and epicentral distance R = 35 km

0 5 10 15 20200

100

0

100

200

Synthetic acceleration, velocity and displacement. Mw=6.5, R=35 km.

Acc

. [cm

/s2 ]

0 5 10 15 2010

5

0

5

10

Vel

. [cm

/s]

0 5 10 15 204

2

0

2

4

Dis

p. [

cm]

Time [sec]

is necessary forthe analysis of ex-tended structures(e.g., pipelines,tunnels, bridges,dams).

Therefore, inparallel to theabove develop-ments, we have ini-tiated work on theKinematic Model-ing Approach. Pre-liminary results forthe 1988 Saguenayearthquake arevery encouraging.We are currentlyinvestigating andtesting variousmodels that havebeen proposed inthe field of Sto-chastic Seismol-ogy (Sato and

Fehler, 1998) to model scattering ef-fects that are so important in ENA.

� Figure 3. Predictions of peak ground acceleration and 5%damped pseudo-velocity response spectra (PSV) at threeoscillator frequencies of 0.5 Hz, 2 Hz and 10 Hz, for momentmagnitude Mw

of 5, 6 and 7

101 102 103

103

102

101

100

101

102

r (km)P

SV

(cm

/s)

PSV (ξ=0.05)

0.5 Hz

Mw=7

Mw=6

Mw=5

101 102 103

101

100

101

102

103

104

r (km)

PG

A (

cm/s

/s)

Peak Ground Motion (PGA)

amax

Mw=7Mw=6Mw=5

101 102 103

103

102

101

100

101

102

r (km)

PS

V (

cm/s

)

PSV (ξ=0.05)

10 HzMw=7

Mw=6Mw=5

101 102 103

103

102

101

100

101

102

r (km)

PS

V (

cm/s

)

PSV (ξ=0.05)

2 Hz

Mw=7

Mw=6

Mw=5

a) b)

c) d)

68

Eventually, a suite of time historieswill be synthesized and made di-rectly available for use by consult-ants (i.e., a suite of “standard”MCEER time histories analogous tothose generated for California bythe SAC Project).

ConclusionsAs part of the development ef-

forts related to the Kinematic Mod-eling Approach discussed above,we have initiated the followingsubtasks:

1. Scattering effects of the lithos-phere: Scattering effects are avery important consideration inthe synthesis of ground motionin Eastern North America. Ourobjective is to synthesize resultsdeveloped in the field of Sto-chastic Seismology with tech-niques developed in the field ofApplied Stochastic Processes.

2. Sub-event Models: We have madesubstantial progress in develop-ing closed-form mathematicalexpressions for the far-field ra-diation of new kinematic mod-els to represent the sub-eventsof the Specific Barrier Model.

These new kinematic modelsaccount more realistically fordirectivity effects.

3. Validation: Any simulationmethod should be validated bycomparing synthetic seismo-grams against recorded ones. Wehave initiated such validationcomparisons using the 1988Saguenay earthquake event as acase study.

4. Near-source ground motions:There are two competing physi-cal effects that affect near-sourceground motions in EasternNorth America (ENA): earth-quake sources in ENA are char-acterized by higher local stressdrops and shorter rise times ascompared to correspondingmotions in California. Thus, ENAnear-source “killer pulses” areexpected to be stronger and ofhigher frequency (i.e., shorterduration). On the other hand,ENA earthquake sources appearto occur at greater depths andthus source-to-station distance(and consequently geometricattenuation) is greater. It is ofgreat practical importance toinvestigate which of the abovetwo effects dominates.

Aki, K., 1982, “Strong motion prediction using mathematical modeling techniques,”Bulletin of the Seismological Society of America, Vol. 72, No. 6, pp. S29-S41.

Atkinson, G.M., 1993, “Earthquake source spectra in Eastern North America,” Bulletinof the Seismological Society of America, Vol. 83, No. 6, pp. 1778-1798.

Boore, D.M., 1983, “Stochastic simulation of high-frequency ground motions based onseismological models of the radiated spectra,” Bulletin of the Seismological Societyof America, Vol. 73, No. 6, pp. 1865-1894.

References


Boore, D.M. and Atkinson, G.M., 1987, “Stochastic prediction of ground motion andspectral response parameters and hard-rock sites in eastern North America,”Bulletin of the Seismological Society of America, Vol. 77, No. 2, pp. 440-467.

Boore, D.M. and Joyner, W.B., 1997, “Site amplifications for generic rock sites,”Bulletin of the Seismological Society of America, Vol. 87, No. 2, pp. 327-341.

Building Seismic Safety Council, 1998, 1997 Edition - NEHRP RecommendedProvisions for the Development of Seismic Regulations for New Buildings,Federal Emergency Management Agency, Washington, D.C., FEMA 302.

Clough, R.W. and Penzien, J., 1993, Dynamics of Structures, Second edition, McGraw-Hill Inc, pp. 648.

Cornell, C.A., 1964, Stochastic Process Models in Structural Engineering, TechnicalReport 34, Department of Civil Engineering, Stanford University, Stanford, CA.

Frankel, A., 1995, “Mapping Seismic Hazard in the Central and Eastern United States,”Seismological Research Letters, Vol. 66, No. 4, pp. 8-21.

Haddon, R.A.W., 1995, “Modeling of source rupture characteristics for the Saguenay,earthquake of November 1988,” Bulletin of the Seismological Society of America,Vol. 85, No. 2, pp. 525-551.

Haddon, R.A.W., 1996, “Earthquake source spectra in Eastern North America,” Bulletinof the Seismological Society of America, Vol. 86, No. 5, pp. 1300-1313.

Harmsen, S., Perkins, D. and Frankel, A., 1999, “Deaggregation of Probabilistic GroundMotions in the Central and Eastern United States,” Bulletin of the SeimologicalSociety of America, Vol. 89, No. 1, pp. 1-13.

Housner, G.W. and Jennings, P.C., 1964, “Generation of artificial earthquakes,” ASCEJournal of the Engineering Mechanics Division, Vol. 90, pp.113-150.

Papageorgiou, A.S. and Aki, K., 1983, “A specific barrier model for the quantitativedescription of inhomogeneous faulting and the prediction of strong ground motion,I. Description of the model,” Bulletin of the Seismological Society of America, Vol.73, No. 3, pp. 693-722.

Papageorgiou, A.S. and Aki, K., 1983, “A specific barrier model for the quantitativedescription of inhomogeneous faulting and the prediction of strong ground motion.Part II, Applications of the model,” Bulletin of the Seismological Society ofAmerica, Vol. 73, No. 4, pp. 953-978.

Papageorgiou, A.S. and Aki, K., 1985, “Scaling law of far-field spectra based onobserved parameters of the specific barrier model,” Pure Applied Geophysics, Vol.123, pp. 354-374.

Papageorgiou, A.S., 1988, “On two characteristic frequencies of acceleration spectra:Patch corner frequency and fmax,” Bulletin of the Seismological Society ofAmerica, Vol. 78, No. 2, pp. 509-529.

Sato, H. & Fehler, M.C., 1998, Seismic wave propagation and scattering in theheterogeneous earth, Springer-Verlag New York Inc., pp. 308.

Shinozuka, M., and Sato, Y., 1967, “Simulation of nonstationary random processes,”ASCE Journal of the Engineering Mechanics Division, Vol. 93, No. 1, pp. 11-40.

71

Research Objectives

The overall goal of this research is to achieve substantial improve-ment in seismic reliability of water supply systems through advancedtechnologies, notably fiber reinforced composites (FRC’s) to strengthenthe welded slip joints of critical steel trunk lines. The research objec-tives are: 1) acquisition of full-scale welded slip joint specimens forlaboratory testing, 2) development of simplified shell and 3-D FEM ana-lytical models for performance of welded slip joints under compres-sive load, 3) full-scale testing of welded slip joints without FRC’s, 4)refinement and validation of analytical models, 5) full-scale testing ofwelded slip joints with FRC’s, 6) development and validation of analyti-cal models for compressive load performance of FRC-reinforced weldedslip joints, and 7) implementation of FRC reinforcement at Los AngelesDepartment of Water and Power (LADWP) and other water utilities.

In addition, the seismic performance of internal FRC linings will beassessed. This research will take advantage of work already supported bygas utility companies on the chemical and mechanical aging of FRC lin-ings used to rehabilitate cast iron distribution mains. The improved per-formance of cast iron pipelines strengthened with FRC linings to bothtransient and permanent ground deformations during an earthquake willbe quantified through full-scale tests on industry supplied specimens.

The 1994 Northridge earthquake resulted in the most extensive dam-age to a U.S. water supply system since the 1906 San Francisco earth-

quake. Los Angeles Department of Water and Power and MetropolitanWater District (MWD) trunk lines (nominal pipe diameter greater than600 mm) were damaged at 74 locations, and the LADWP distribution sys-tem required repairs at 1,013 locations. An analysis of Northridge earth-quake performance shows that approximately 60% of critical trunk linedamage in the Los Angeles Department of Water and Power (LADWP) sys-tem occurred because of compressive failure at welded slip joints.

Fiber Reinforced Compositesfor Advanced SeismicPerformance of Water Supplies

by Thomas D. O’Rourke (Principal Author), James A. Mason,Ilker Tutuncu and Timothy Bond

Thomas D. O’Rourke,Thomas R. Briggs Professorof Engineering, School ofCivil and EnvironmentalEngineering, CornellUniversity

Timothy Bond, LaboratoryManager, School of Civiland EnvironmentalEngineering, CornellUniversity

James A. Mason and IlkerTutuncu, GraduateResearch Assistants, Schoolof Civil and EnvironmentalEngineering, CornellUniversity


Los Angeles Department ofWater and Power

72

A welded slip jointis fabricated by insert-ing the straight end ofone pipe into the bellend of another andjoining the two sec-tions with a circum-ferential fillet weld.The bell end is cre-ated by the pipemanufacturer by in-serting a mandrel inone end of a straightpipe section, and ex-panding the steel intoa flared, or bell casing. Large diam-eter pipelines can be constructedrapidly and economically in thefield by joining the bell and spigot(straight) ends of connecting pipesegments.

Figure 1 shows a compressive fail-ure at a welded slip joint on theGranada Trunk Line, a 1,245 mm (49in.) diameter steel pipeline with 6.4mm (1/4 in.) wall thickness thatfailed during the Northridge earth-quake because of lateral groundmovement triggered by liquefac-tion near the intersection of BalboaBoulevard and Rinaldi Street in theSan Fernando Valley. Similar com-pressive failures were observed in

trunk lines during the 1971 SanFernando earthquake and in theadjacent 1,727 mm (68 in.) diam-eter, (9.5 mm (3/8 in.) wall thick-ness) Rinaldi Trunk Line during theNorthridge earthquake. Loss ofboth the Granada and Rinaldi TrunkLines cut off water to tens of thou-sands of customers in the SanFernando Valley for several days.

As illustrated in Figure 2, failureof welded slip joints can be initi-ated by compressive forces that in-duce buckling and outwarddeformation at the location of maxi-mum curvature in the bell casing.Compressive forces sufficient to failwelded slip joints can be generated

The users of research results include: 1) water utilities, suchas the Los Angeles Department of Water and Power (LADWP),East Bay Municipal Utility District (EBMUD), Memphis Light, Gasand Water, and many other utilities operating in areas vulner-able to earthquakes; 2) private enterprises, such as engineeringfirms, construction contractors, and specialists in the designand distribution of advanced materials; and 3) communities thatdepend on water supply and the engineering and constructionservices needed to ensure reliable, cost effective service. Re-search on advanced materials, such as FRC’s, not only results insubstantial improvements in seismic performance, but also pro-vides advanced materials technology needed for the rehabilita-tion of U.S. civil infrastructure, under daily conditions of agingand environmental deterioration.

• LADWP contributedextensive data sets, maps,and information aboutsystem facilities. Thedepartment has alsocontributed speciallyfabricated 305 to 915 mmdiameter (12 and 36 in.)steel pipe specimens thatwere shipped to CornellUniversity for testing.

• EBMUD has agreed toprovide technical advice,oversight, and in-houseservices to validate the FRCtechnology and improveapplication procedures inthe field.

• Taylor Devices, Inc. isproviding the use of the 1.5million lb. load test frameand facilities at their NorthTonawanda, New Yorkoffices.

• Structural PreservationSystems, Inc., Hanover, MD,is providing engineering,FRC materials, andinstallation personnel.

• Master Builders, Inc,Cleveland, Ohio, isproviding engineering, FRCmaterials, and installationpersonnel.

• R.J. Watson, Inc., EastAmherst, New York, isproviding engineering, FRCmaterials, and installationpersonnel.

• Fyfe Company, San Diego,California, is providingengineering, FRCmaterials, and installationpersonnel.

• The research work was

� Figure 1. Compressive failure of the 1,245 mm Granada TrunkLine

73Fiber Reinforced Composites for Advanced Seismic Performance of Water Supplies

by near source pulses of high par-ticle velocity as well as permanentground deformation generated bysurface faulting, liquefaction, andlandslides.

Fiber Reinforced Composite(FRC) wrapping can be used to con-fine the welded slip joint againstoutward deformation of the bell,(see Figure 2). This type of reinforc-ing not only can be used to retrofitexisting welded slip joints, but tostrengthen new joints during fabri-cation in the field. As new pipelinesare constructed, the pipeline surfaceat and adjacent to the fillet weldneeds to be wrapped to provide cor-rosion protection. If the wrap usedto protect against corrosion can alsostrengthen the pipeline against com-pressive failure, then the seismicperformance of the critical trunklines can be enhanced at relativelylittle additional cost.

Research at MCEER has focusedon developing robust analyticalmodels that can emulate bucklingin straight pipe and welded slipjoint sections. The research also in-cludes a comprehensivelaboratory test program toquantify the improve-ments in axial load capac-ity achieved with FRCwrap relative to that of un-wrapped specimens. Thelaboratory tests are alsoused to qualify and validatethe analytical models.

A comprehensive studyof analytical modeling withDIANA and ABAQUSwas undertaken, andABAQUS was adopted asthe finite element method(FEM) package of choicefor evaluating welded slipjoint and pipeline per-formance. Benchmark

analytical studies were performedwith ABAQUS, and the sensitivityof the solutions to initial imperfec-tions was investigated by bifurca-tion buckling techniques. The mostappropriate element type chosenwas a linear, four node, reduced in-tegration shell element. Circumfer-ential weld representation wasfound to have a significant effect onthe finite element solutions, and anappropriate weld representationtechnique was developed for theanalytical work with a techniqueknown as multi-point constraints(MPC). The MPC technique con-nects nodes around the circumfer-ence of the bell to their counterpartson the inserted straight pipe. A rigidweld is then simulated by enforcingidentical degrees of freedom in eachpair of connecting nodes.

Figure 3 shows profile views ofthe FEM mesh in the vicinity of a305 mm (12 in.) diameter weldedslip joint before and after the ap-plication of axial compressive load.The FEM mesh which was config-ured to simulate the 610 mm

coordinated with MCEER-supported studies of theLADWP electric powersystem supervised by M.Shinozuka, University ofSouthern California.

• New York Gas Group, NewYork City, New York.

• In-Situ Form Technologies,Inc. St. Louis, Missouri.

• Miller Pipeline Co.,Columbus, Ohio.

� Figure 2. Compressive response of welded slip-joint and FRC reinforcement

74

(24 in.) long experimental speci-mens, was modeled as a quartersection to take advantage of thesymmetrical loading condition andgeometry. Based on mesh refine-ment studies, 7,260 elements wereused with 66 equi-dimensional el-ements along the circumference ofeach quarter section in the regionof refined mesh coverage.

Analytical studies were per-formed for a prismatic section ofpipe with no welded slip joint sothat the analytical results could becompared with published solu-tions and experimental measure-ments. The buckling limit of aprismatic, or straight, pipe sectionestablishes the maximum capacityof the pipe. The upper bound ofperformance with FRC wrappingis the buckling limit of the straightpipe. If FRC strengthening of awelded slip joint increases the loadcarrying capacity to the bucklinglimit of a straight pipe section, thenthe FRC technology has been suc-cessful in achieving the maximumpossible improvement.

Specimens of 305 and 610 mm(12 and 24 in.) diameter weldedslip joints were fabricated by

LADWP and shipped to the GeorgeWinter Structures Laboratory forStructural Engineering Research, atCornell University. Full coopera-tion and support from LADWP wasestablished early in the experimen-tal process allowing for immediatefeedback between the two partiesin project definition and direction.The choice of specimen size (di-ameter and pipe wall thickness)and joint construction (exterior vs.interior welded slip joint) has haddirect input from LADWP to re-flect their concerns for existingand future inventory. In addition,close coordination with FRC de-signers and contractors has pro-gressed in parallel with testing.Laboratory tests have been per-formed on FRC reinforced as-builtspecimens that were prepared bytwo vendor teams participating inthe experimental program. Thevendor teams are: 1) StructuralPreservation Systems, Inc, Balti-more, MD, and Master Builders,Inc., Cleveland OH, and 2) Fyfe,Inc., La Jolla, CA, and R.J. Watson,Inc., Buffalo, NY.

The types of tests that have beenperformed on 305 mm (12 in.) di-ameter pipe are: 1) compressionof a prismatic section, 2) compres-sion of unreinforced welded slip-joints, and 3) compression ofFRC-reinforced welded slip-joints.As previously discussed, the per-formance of the prismatic sectionestablishes the baseline, i.e., themaximum, for comparison with allother test results.

The experimental testing facili-ties consisted of a custom de-signed MTS load frame, with a loadcapacity of 2,700 kN (600 kips).Data was acquired by a personalcomputer controlling several dataacquisition systems. Axial forces

• Seismic Reliability Analysisand Retrofit Method forSouthern California PowerSystems, T.C. Cheng,University of SouthernCalifornia.

• Rehabilitation Strategies forLifelines: LADWP PowerSystems, S.T. Mau, NewJersey Institute ofTechnology, T.C. Cheng andM. Shinozuka, Universityof Southern California.

� Figure 3. Finite element mesh of welded slip-joint


were cycled on sections of strain-gaged pipe at levels needed to seatthe specimen, and at approximately30 and 60 percent maximum capac-ity. The pipe then was loaded tothe yield capacity and cycled, afterwhich axial load was applied as faras the hydraulic ram could move[100 mm (4 in.)].

Figure 4 shows the prismatic sec-tion in the test frame. For this testa total of 48-strain gage rosetteswere bonded to the pipe, 24 on theoutside and 24 on the inside match-ing the location and orientation ofthe associated gage on the outside.

Many researchers (e.g., Donnelland Wan, 1950) have shown thatinitial imperfections due to themanufacturing and handling havea strong influence on the bucklinglimit. Imperfections were mea-sured systematically across the pris-matic pipe specimen on a 25 mm(1 in.) grid, utilizing a digital dialgage. The periodicity of the imper-fections was analyzed with fast Fou-rier transform functions, and the

imperfection spectra generated wasused to model the imperfectionamplitude and wavelengths. Themaximum imperfection amplitudewas approximately three percent ofthe pipe wall thickness. Closeagreement between the experimen-tally observed buckling pattern andnumerically simulated pattern wasachieved.

Figure 5 shows the axial load vs.displacement plots for the pris-matic test specimen and the nu-merical simulation. There is aremarkably close agreement be-tween the experimental and analyti-cal results. The peak predicted andmeasured loads are 2,200 and 2,175kN (495 and 489 kips), respectively,which are two to three percentgreater than the theoretical yieldload of the specimen. As shown bythe inset images, there is closeagreement with respect to the lo-cation of buckling in the experi-mental and analytical results.

Figure 6 shows the axial load vs.displacement plots for a welded slipjoint specimen and the numericalsimulation. Again, there is remark-ably close agreement between the

� Figure 4. Prismatic pipe specimen in loadframe

� Figure 5. Comparison of test and FEM results for prismatic pipe

“Research atMCEER hasfocused ondevelopingrobust anal-ytical modelsthat can emulatebuckling instraight pipeand welded slipjoint sections.”

76

force vs. displacement relationshipsand patterns of buckling.

Figure 7 shows the axial load vs.displacement plots of the prismaticand the welded slip joint speci-mens. The maximum loads carriedby the welded slip joints were1,670 and 1,740 kN (375 and 390kips), which are between 77 and 79percent of the buckling limit of theprismatic pipe section.

Figure 8 shows the axial load vs.displacement plots for the weldedslip joint specimen, Slip Joint 3, and

the FRC wrapped slip joint speci-mens prepared by the SPS/MB (FRCWrapped-2) and Fyfe/Watson (FRCWrapped-1) teams. In each case,the FRC wrap resulted in an in-creased capacity of approximately25 percent. When the wrappedspecimen results are compared withthose of the prismatic section, it canbe seen that FRC strengtheningachieves a compressive capacity vir-tually equal to the buckling limit ofa straight pipe section. Moreover, asthe inset images in Figure 7 and Fig-ure 8 show, the FRC wrapped speci-men “FRC Wrapped–2” failed bybuckling at the same location andmanner as the prismatic specimen.

The close agreement betweenthe analytical and experimental re-sults indicates that the model de-veloped in the research work isrobust and sufficiently reliable forevaluating the response of weldedslip joints with different geom-etries. Figure 9 shows the analyti-cal results of welded slip joint axialload capacity, expressed as P

R, the

ratio of maximum to theoreticalyield load, plotted relative to pipediameter-to-thickness ratio, D/t. Theresults for various pipe diametersare plotted. The yield load is simplythe product of the steel compressiveyield stress and cross-sectional areaof the pipe. As discussed previously,the yield load is very close to thebuckling capacity of a prismaticpipe section so that it may be takenas the maximum achievable capac-ity of the pipe.

The analytical results in Figure 9allow one to scale the current find-ings to larger D/t ratios, represen-tative of larger diameter pipe. Forexample, the Granada and RinaldiTrunk Lines described previously,have D/t ratios 160 to 180. Weldedslip joints in this D/t range would� Figure 7. Comparison of prismatic and welded slip-joint performance

� Figure 6. Comparison of test and FEM results for a welded slip-joint


be expected to mobilize only about50 percent of the maximum axialcapacity of the pipe. Hence, FRCstrengthening can result in nearlya 100 percent increase in compres-sive load capacity of a pipe withthese geometric characteristics.

The FEM results are also plottedrelative to the results from a sim-plified shell model developed byTawfik and O’Rourke (1985). Al-though the simplified shell modelprovides a good representation forD/t ≥ 300, it tends to underpredictthe axial compressive capacity forthe D/t range most frequently en-countered in water supply trunklines (75 ≤ D/t ≤ 200).

Conclusions andFuture Research Needs

The full-scale tests being per-formed at Cornell University are fo-cused on the effectiveness of FRCstrengthening for steel pipeline withwelded slip joints. The experimen-tal and numerical simulation workin progress will clarify: 1) influenceof diameter to pipe wall thickness(D/t) on overall performance, 2) sur-face irregularities and their effectson performance, 3) local bucklingdeformation of exterior weldedjoints verses interior welded joints,and 4) different vendor FRC designsand installation procedures. As-con-structed welded slip joints (bellhousing and spigot) are tested to in-vestigate the reduction of axial com-pressive capacity due to the joint.Participating vendors wrap the jointregions of test specimens with pro-prietary FRC materials. All speci-mens are tested to large plasticstrains that investigate the failureprocess through successive localbuckling modes.

The experimental program in-cludes tests of pipes ranging in di-ameter from 305 to 914 mm withD/t ratios from 48 to 250. Bothstatic and dynamic tests are inprogress, and an experimentalevaluation of performance undercompressive and tensile forces will

� Figure 8. Comparison of welded slip-joint and FRC reinforced test specimenperformance

� Figure 9. Welded slip joint capacity as a function of diameter-to-wallthickness ratio

78

be made. At the request of LADWP,an assessment will be made of theperformance of slip joints with ex-terior and interior fillet welds at aD/t ratio identical to that of the new2.44 m (96 in.) diameter City TrunkLine currently in design.

Large diameter pipes (762 and914 mm) will be tested at the Tay-lor Devices facilities in NorthTonawanda, New York. Taylor De-vices will donate the use of itsequipment at no charge, except fortime required by Taylor personnelto assist in test preparation. This rep-resents exceptional cost and timesavings relative to previous plansfor fabricating the appropriate test-ing equipment at Cornell. It also rep-resents an excellent example ofindustry cooperation in the MCEERresearch effort. The capacity of the

load frame at their facilities isroughly 6,700 kN (1500 kips). Thisload can be applied in either com-pression or tension. The test ma-chine is capable of rapid loadingwith velocities similar to those re-corded during the Northridgeearthquake. Figure 10 shows theframe being used for testing a largeseismic damping device.

A similar approach with inter-nally applied FRCs for improvingthe seismic performance of gas andwater distribution mains is beingpursued. In this case, MCEER re-searchers are taking advantage ofresearch already in progress andfunded by the New York GasGroup (NYGAS) at Cornell. Thiswork involves mechanical andchemical aging tests on full-scalespecimens of cast iron (CI) distri-bution pipe prepared and linedwith FRCs under the supervisionof gas utility personnel and con-tractors specializing in FRC liningtechnologies. Specimens of castiron gas mains, provided by theNew York Gas Group, have beenlined with the FRCs and tested un-der conditions in which the castiron pipe has ruptured with the in-ternal lining intact. The tests in-volved the simulation of verticaloffsets and relative rotations offractured pipe lengths consistentwith the type of deformation ex-perienced in liquefied ground. Thelining was able to sustain low-pres-sure service without disruption.

Because this research involvesthe qualification of FRC linings fordamaged pipe under cyclic andpermanent deformations, the re-sults can be readily adapted toevaluate the seismic performanceof FRC-reinforced distributionmains. Additional tests are planned

� Figure 10. 1.5 million lb. load frame at the Taylor DevicesTesting Facility


to load the FRC-lined pipes to fail-ure and to assess dynamic deforma-tions consistent with near fieldstrong motion records. Coopera-tion in this testing has been ob-tained from the New York GasGroup, In-situ Form TechnologiesInc., St. Louis, Missouri and theMiller Pipeline Co., Columbus,Ohio.

To evaluate the impact of FRCtechnology, improvements in system

References

Donnell, L.H. and Wan, C.C., 1950, “Effect of imperfections on buckling of thin cylindersand columns under axial compression,” Journal of Applied Mechanics, Vol. 17, pp.73-83.

Tawfik, M.S. and O’Rourke, T.D., 1985, “Load-carrying capacity of welded slip joints,”Journal of Pressure Vessel Technology, Transactions of the ASME, Vol.107, pp 36-43.

reliability will be evaluated bymeans of probabilistic hydraulicnetwork simulations. These analy-ses will be used to quantify the ef-fects on system performance ofpipe and welded slip jointsstrengthened by FRCs. The systemsanalyses will take advantage of theinventory development and GISmodeling of the Los Angeles watersupply being performed concur-rently with the FRC research.

81

Research Objectives

1. To conduct high level reconnaissance using satellite imagery, Differ-ential Global Positioning Systems (GPS), and in-field GPS-GIS inter-faces

2. To validate damage information contained in early U.S. State Depart-ment Damage Maps

3. To serve as a model case study for exploring the use of remotelysensed data for post-earthquake damage assessment.

4. To reinforce the collaborative activities between MCEER and EDM

The Marmara earthquake occurred during a time of unprecedented tech-nological development. Post-disaster information and data that once

took months to generate were developed within a matter of days in thisevent. Furthermore, the ability to comprehensively understand the mean-ing of these data was significantly enhanced because of the use of sophis-ticated database management programs and geographical relationalalgorithms, e.g., geographic information systems, (GIS). In the most gen-eral sense, we were able to literally map the effects of the earthquake in“real time.”

One area that has benefited immensely from this technology explosionis earth observation and mapping (see Figure 1). From over a dozen differ-ent platforms, we are able to view and quantify with surprising precisionthe different properties of the earth’s surface. Topographic features areclearly seen from low earth-orbiting optical satellites. Urban areas are alsovisible in these images. In addition, using an analysis technique called“interferometry,” it is possible to detect minute changes in the earth’s sur-face by comparing a series of radar images taken at different times. It isfrom the perspective of testing the use of advanced technologies for post-earthquake reconnaissance that we provide our analysis of the Marmaraearthquake.

The following discussion offers a preliminary report on a joint recon-naissance effort between MCEER and the Earthquake Disaster Mitigation(EDM) Research Center in Miki, Japan. This reconnaissance took place

The Marmara, Turkey Earthquake:Using Advanced Technology toConduct Earthquake Reconnaissance

by Ronald T. Eguchi (Coordinating Author), Charles K. Huyck, Bijan Houshmand, BabakMansouri, Masanobu Shinozuka, Fumio Yamazaki, Masashi Matsuoka and Suha Ülgen

National Science Foundation,Earthquake EngineeringResearch Center Program

Ronald T. Eguchi, President,and Charles K. Huyck,Vice President, ImageCat,Inc.; formerly of EQEInternational, Inc.

Bijan Houshmand, JetPropulsion Laboratory andAdjunct AssociateProfessor, ElectricalEngineering Department,University of California atLos Angeles

Babak Mansouri, Ph.D.Candidate, andMasanobu Shinozuka,Fred Champion Chair inCivil Engineering,Department of CivilEngineering, University ofSouthern California

Gary Webb, Post-doctoralResearch Fellow, DisasterResearch Center, andDepartment of Sociology,University of Delaware

82

approximately one and a halfmonths – between September 28th

through October 4th - after the di-sastrous main shock of the Marmaraearthquake. The team leaders wereRonald T. Eguchi of ImageCat, Inc.(formerly of EQE International),who led the MCEER team and Pro-fessor Fumio Yamazaki of TokyoUniversity, who led the EDM team.The collaboration with EDM hasbeen ongoing since the 1994Northridge earthquake in the U.S.and the Kobe earthquake in Japan

(1995). Both Centers have com-mitted substantial resources toexplore the use of advanced tech-nologies for natural disaster man-agement. The investigation of theMarmara earthquake representsthe latest collaboration betweenthese two organizations.

The following sections discuss thepurpose of the trip and the meet-ings that were held, the new tech-nologies that were used during thisreconnaissance trip, a “thumbnail”sketch of specific in-field studies

Serkan Bozkurt, InteractiveMedia and GeographicInformation Systems(IMAGINS)

Turgay Türker, TürkerEngineering (StructuralEngineering)

Suha Ülgen, InteractiveMedia and GeographicInformation Systems(IMAGINS)

Fumio Yamazaki andMasashi Matsuoka,Earthquake DisasterMitigation Research Center

Image provided to EQE International by the European Space Agency under a cooperative researchagreement established for this earthquake. Image details: RGB mapping of 721, replacing the red band

with the far infrared band. A Gaussian filter was applied to maximize visibility of the smoke.

� Figure 1. Landsat 5 Image of I·zmit Bay taken on August 19, 1999. Note the fire and smoke plumeoriginating from the Tüpraçs Refinery, seen at the center of the figure.

The data collected during this reconnaissance will benefit vari-ous user groups. First, this information will help researchersvalidate new damage detection models based on remote sensingtechnologies. Second, the lessons learned during this trip willhelp to improve earthquake reconnaissance techniques and pro-cedures by encouraging the use of new and advanced technolo-gies. Of particular significance is the contribution that advancedGIS-GPS interface systems have in recording damage informa-tion in real-time. Finally, the results of this research will ulti-mately help future emergency responders by providing morereliable methods of assessing post-earthquake damage. Assess-ing damage sooner will allow responders to act more quicklyand more effectively in deploying limited resources.

83Using Advanced Technology to Conduct Earthquake Reconnaissance

that were conducted, and finally, theusefulness of these advanced tech-nologies in assessing damage fromthis devastating event.

ItineraryThe itinerary for the trip was es-

tablished several weeks before de-parting for Turkey. The tripconsisted of meetings with Turkishresearchers and investigators, andbrief field visits to several of thehardest hit areas. The details of thefield visits are discussed in moredetail later in this section. Providedbelow are brief summaries of themeetings that were held during thefirst two days of this trip. Table 1summarizes the itinerary for thistrip.

During our visit with ProfessorMustafa Erdik at the Kandilli Obser-vatory and Earthquake ResearchCenter of Bogaziçi University, theresearch team was able to ask gen-eral questions about the extent ofdamage to western Turkey. Wefound that although the earthquakewas initially named after one of thecities closest to the main shock (i.e.,I·zmit), damage in this area was notas severe as other areas furtheraway from the epicenter. We weretold that Gölcük (located on thesouthern side of I·zmit Bay androughly 80 kilometers east ofI·stanbul) and Adapazarı (locatedroughly 125 kilometers east ofI· stanbul) had experienced farmore damage than the town ofI·zmit.

We were also shown prelimi-nary ground motion records fromthis event, learning that the peakground acceleration in Adapazarıreached about 40 percent g. Atthe time of our visit, a number ofportable instruments were being

installed in order to record groundmotions from large aftershocks.Before leaving this facility, theMCEER/EDM team was given a tourof the Kandilli Observatory Labo-ratory where we viewed otherground motion data that was beingcollected.

At TÜBI·TAK (Turkish Scientificand Technical Research Institute)Marmara Research Center, severaldifferent meetings were held. Thefirst meeting was with the EarthSciences Research Institute wherethe team met with Dr. M. NamıkYalçın, the Director of the EarthSciences Institute at TÜBI· TAKMarmara Research Center, and Dr.Semih Ergintav, a member of theEarth Sciences Research Institute.The purpose of this meeting wasto determine the availability of GPSinformation for the earthquake. Theteam was particularly interested incollecting data on post-earthquakedisplacements for towns that werehardest hit by this earthquake. Thisinformation would be used to com-pare permanent ground displace-ments calculated using SyntheticAperture Radar (SAR) with thosederived from the continuous GPSnetwork. We understood that acontinuous GPS system was in

9/30/99

Meeting/VisitDate

g

TÜBITAK Marmara Research Center: Earth Sciences Research Institute

ekauqhtraEdnayrotavresbOillidnaK,kidrEafatsuM•ytisrevinUiçizaoB,etutitsnIhcraeseR

ralıcvAottisiVdleiF•

10/1/99 •

TÜBITAK Marmara Research Center: Space Technologies Group

•

nemyeSottisiVdleiF•kucloGottisiVdleiF•

10/2/99 ırazapadAottisiVdleiF•

10/3/99 nemyeSottisiVdleiF•kucloGottisiVdleiF•

·

·

� Table 1. Itinerary

Program 3: EmergencyResponse and Recovery

84

place at the time of the earthquakeand that displacements at severalsites were being monitored. We alsounderstood that the primary inter-est in these data was to map theco-seismic slip of the fault, and notthe relative tectonic movement ofthe region. The results of this workwere presented at the AnnualAmerican Geophysical Union meet-ing that was held in San Franciscoin December 1999.

The second meeting was heldwith the Space Technologies Group

at TÜBI·TAK. The investigation teammet with Dr. Hülya Yıldırım, the Di-rector of Remote Sensing and Geo-graphic Information Systems at theMarmara Research Center. Beforethe earthquake, this group was in-volved with numerous environmen-tal and agricultural studies. Onestudy involved the preparation ofwatershed maps for the five prov-inces surrounding I·zmit, which isin the province of Kocaeli. After theAugust 1999 earthquake, this groupwas assigned the responsibility of

TÜBI·TAK

� Figure 2. Landuse Map developed from SPOT image after earthquake (July 7, 1999). Thedifferent colors/shades represent various levels of development or open areas. The circlesurrounds the town of Gölcük. Note that the harbor is located just to the left of the circle.

TÜBI·TAK

� Figure 3. Image of Gölcük created after the earthquake. (SPOT image taken on 8/20/99). Notethat the white or light areas identify city blocks where significant damage to buildings occurred.

Multidisciplinary Centerfor EarthquakeEngineering Researchhttp://mceer.buffalo.edu/research/turkeyeq/default.html

Bogaziçi University,Kandilli Observatoryand Earthquake ResearchInstitutehttp://www.koeri.boun.edu.tr

Earthquake DisasterMitigation ResearchCenterhttp://www.miki.riken.go.jp


producing GIS maps that docu-mented damage in the I·zmit area.While lacking any prior earthquakeexperience, this group immediatelybegan integrating available GISmaps with satellite imagery to iden-tify those urban regions that weremost affected by this event.

One analysis, which was based ona comparison of pre- and post-earthquake SPOT1 images, showedlarge areas in Gölcük that wereclearly affected by the earthquake.Several of these images are shownin Figures 2 and 3. In addition, othersatellite-derived images showed ex-tensive areas along the shore nearGölcük that were inundated as aresult of ground subsidence causedby liquefaction. Figure 4 shows anaerial view of the inundation area.

Field InvestigationsDuring this trip, the MCEER/EDM

team was able to survey earthquakedamage in four areas: Avcılar,Seymen, Gölcük and Adapazarı.Where possible, we applied GPStechnology linked with GIS systemsto record damage information. Inaddition, satellite imagery and aerialphotographs were available forsome areas. The following sectionsdiscuss the data that were collected,the technologies that were used inrecording this information, and theanalyses that have been performed– since our return – to interpret theeffects of this earthquake. Beforediscussing these field investigations,we discuss very briefly the tech-nologies that were used on this trip.

New Technologies

Radar Images. The MCEER/EDM team was fortunate to obtainnumerous satellite images of the

I·zmit area prior to its departure.Through a cooperative researchagreement with the EuropeanSpace Agency (ESA), EQE Interna-tional received a series of radarimages, both before and after theearthquake, via the ERS-1 and ERS-2satellites. A tabulation of these datais listed in Table 2. As will be dis-cussed later, several of these scenes(i.e., images) are being used to cre-ate interferograms that will hope-fully identify areas of significantdamage.

The project team also receivedpost-earthquake Radarsat images ofwestern Turkey. These, however,were received after the team re-turned from Turkey. We have yet toprocess this information.

Optical Satellite Images. Inaddition to the radar data, ESA alsoprovided Landsat 5 images takenseveral days after the earthquake.Figure 1 shows a Landsat 5 imageof the I·zmit Bay region. Visible inthis image is the fire that occurredat the Tüpraçs Oil Refinery, approxi-mately 70 kilometers southeast ofI·stanbul.

Photograph by D. Andrews

� Figure 4. Aerial view of earthquake damage in Gölcük. This area was affected byground subsidence caused by liquefaction.

86

Photograph by R. Eguchi

� Figure 5. Vehicle used to collect damage/GPS data in Avcılar and Adapazarı. Note theantennae being held by front passenger.

� Table 2. ESA Synthetic Aperture Radar (SAR) Data and Images

)yy/dd/mm(etaD kcarT emarF tibrO tcudorP

06/07/95 336

336

336

336

336

064

293

064

157

157

293

336

336

064

064

157

157

2781

2781

2781

2781

2781

2781

2781+2 nodes

2781

819-4 nodes

819-4 nodes

2781+2 nodes

2781

2781

2781

2781

819-4 nodes

819-4 nodes

20364

00691

18226

19228

20230

20459

20688

20960

42229

22556

22692

42408

22735

42637

22964

42730

23057

10/15/98

12/24/98

03/04/99

03/20/99

04/05/99

04/24/99*

08/12/99

08/13/99

08/17/99

08/23/99

08/25/99

08/26/99

09/10/99*

09/11/99

09/16/99

09/17/99

EARTHQUAKE

Note: RAW refers to unprocessed data; SLC is single look complex; and PRI is precision averaging. Asterisks refer to scenes that were used to create interferograms for this event.

06/08/95

CLSroWAR

CLSroWAR

CLSroWAR

CLSroWAR

IRProCLSroWAR

CLSroWAR

IRP

CLSroWAR

IRProCLSroWAR

IRProCLSroWAR

IRP

IRProCLSroWAR

IRProCLSroWAR

CLSroWAR

CLSroWAR

CLSroWAR

CLSroWAR

ESA


Aerial Photographs. Someaerial photos were taken of the af-fected areas; these, however, haveyet to be widely distributed. Onephoto was published in a localI·stanbul newspaper. The photowas scanned by one of our collabo-rators in Turkey (IMAGINS) and wasused during our field survey of theAvcılar area. This is discussed fur-ther in the next section.

Global Positioning Systems.Two separate GPS systems wereused in the field during this inves-tigation. The first was a high-preci-sion single-frequency 12-channelNovAtel GPS receiver. This systemwas generally used while drivingthrough the different study areas,see Figure 5. A second system – ahandheld Magellan unit – was usedwhen field studies were conductedon foot.

Real-time GPS-GIS Interface.One of the major improvements indocumenting the effects of thisearthquake was the use of a real-time GPS-GIS system. Using the GPSsystems mentioned above, theMCEER/EDM team was able to as-sociate damage information, includ-ing photographs, with accurategeographical coordinates. Wherethere was real-time Differential GPSbroadcast data (FM-RDS), which wasthe case in Avcılar, the positionalaccuracy was one meter or less.When using the handheld Magellanunit, the accuracy level dropped toplus or minus 30 meters. Figure 6shows the equipment setup forthis GPS-GIS interface. One of theadvantages to using this systemwas that the investigation teamwas able to create reasonably ac-curate road maps when none wereavailable. These maps were createdby plotting GPS coordinates (whilein transit) directly onto Landsat 5

imagery. The interface between theNovAtel GPS receiver and theMapInfo GIS software was providedby GeoTracker from Blue MarbleGeographics.

Avcılar

Avcılar is located in the south-western part of I·stanbul borderingthe Marmara Sea. Although locatedroughly 80 kilometers from theepicenter of the August 17, 1999earthquake, there was substantial –but isolated – damage to multi-storyresidential buildings. At the time ofour visit, most of the severely dam-aged structures had already beendemolished. Since there was verylittle to record, other than the loca-tion of these demolished buildings,the team opted to use this time asan opportunity to refine and cali-brate our GPS-GIS data collectionsystem.

Figure 7 shows an aerial photo ofthe part of Avcılar that we focusedon. Visible on this photograph –scanned from a newspaper article– is the shoreline facing theMarmara Sea, at the bottom of thefigure.

Photograph by M. Matsuoka

� Figure 6. Real-time GPS-GIS setup for in-field data collectionand processing.

88

In this part of Avcılar, there wereabout a dozen buildings that expe-rienced significant damage, result-ing in complete collapse of thestructure, or substantial damagerequiring demolition of the build-ing. While traveling to each of thesesites, the investigation team utilizedthe real-time Differential GPS-GISsystem. With accuracy levels withinone meter, we were able to recordthe precise location of eachof the demolished buildings.This information will be usedto assess whether these par-ticular sites can be recog-nized from either satelliteimagery or aerial photos.Most other structures in thisarea experienced little, if anydamage.

Seymen

Seymen is located on thesouthern side of I·zmit Bay,just east of Gölcük (40o 42’

N latitude and 29o 54’ E longitude).This particular area was of interestto the MCEER/EDM team because1) damage to buildings (the area iscomposed of entirely residentialapartment buildings) varied widelyranging from slight to completecollapse; 2) the area was unoccu-pied at the time of our visits, and 3)the buildings were situated in large,open areas. From the standpointof detecting damage from space-borne systems, this area would beideal. Figures 8 through 10 showsome of the buildings where de-tailed data were collected by the in-vestigation team.

As part of an ongoing researchproject, researchers at the Universityof Southern California (Mansouriand Shinozuka) are exploring waysin which SAR data can be used tocharacterize earthquake damage tobuildings. Simulation algorithmsare being developed that willmodel such failures as tilting, firstfloor collapse and complete build-ing failure (i.e., massive pancak-ing). During this reconnaissancetrip, Mansouri collected detailedmeasurements on several of thebuildings that had collapsed dur-ing the earthquake. All of the

IMAGINS

� Figure 7. Aerial photograph of the city of Avcılar. Shown on this figure are sites(identified by small circles) visited by the MCEER/EDM team in September 1999.


� Figure 8. Isolated building in Seymen. This buildingexperienced first floor collapse and well as damage tothe top floor. This building should be visible from aerialphotos; it may be detectable using radar pre- and post-event images.



� Figure 9. Near complete collapse of a five-story concretestructure. Should be visible from both aerial photos andSAR imagery.


� Figure 10. Severe damage to multi-story apartment buildings.Lower floors have “pancaked” and building is not tilted. Maybe visible from aerial photos; collapse of lower floors may bedetectable from radar imagery.

buildings were constructed of re-inforced concrete with rectangularfootprints and had tiled roofs. Be-cause of the orientation of thesebuildings (two rows of six buildingseach) and because of the largespaces between buildings, this sitewas considered an ideal one fromthe standpoint of testing or validat-ing their analytical models.

As indicated earlier, SAR imagesof this area – both before and afterthe earthquake – have been ob-tained from ESA and RADARSAT.Pre- and post-event SAR data is pre-ferred for detecting damage aftercatastrophic events. Using SARdata, it is possible to create imagesat night or through cloud cover.Also, because of its ability to facili-tate change detection analysis, SARtechnology is considered the mosteffective technology in attemptingto “quantify” the effects of large di-sasters.

Over the next year, Mansouri andShinozuka will be calibrating theirsimulation algorithms by testingthem against the Seymen data.Some of the questions that they willattempt to answer are: (1) whatdamage conditions or failure modesare detectable using coarse resolu-tion SAR data, (2) can damage beaccurately simulated using thesenew simulation algorithms, and (3)how can these findings be ex-tended to assess damage to largerareas.

Adapazarı

The town of Adapazarı is locatedabout 125 kilometers east ofI·stanbul. The town sits on a recentlakebed, and suffered extensive liq-uefaction during this earthquake.This was one of the most heavilydamaged areas in this event; damage

to multi-story apartment buildingswas severe, many commercial struc-tures were also seriously damaged.

The MCEER/EDM team spent anentire day in Adapazarı attemptingto collect damage information fromlocal authorities. Despite the factthat the requests were made almosttwo months after the earthquake,there was still very little data avail-able on number of damaged build-ings, whether structures were safeor unsafe, and how many peoplehad perished in this earthquake. Be-cause of this, the investigation team

90

decided to conduct its own surveyof Adapazarı.

One of the primary objectives ofthis trip was to collect enough in-formation to validate a series ofearthquake damage maps that wereproduced by the U.S. State Depart-ment shortly after the earthquake.These maps appeared on theInternet on the Kandilli Observa-tory website. In total, the U.S. StateDepartment produced five maps:Yalova, Seymen, Derence, Adapazarıand West Gölcük.

One example of the type of mapdistributed by the State Departmentis seen in Figure 11 for the Ada-pazarı area. The map identifies ar-eas of catastrophic and extensivedamage. This map was put on theInternet on August 20, 1999, threedays after the earthquake. Accord-ing to officials at the State Depart-ment, high-resolution satelliteimages were used to classify thetown into these two damage cat-egories. Our purpose in validatingthese maps was to create a “groundtruth” database that could eventu-ally be used to calibrate a series of

interferograms developed from thisearthquake.

Unfortunately, the scale and pro-jection of these maps was a littlemisleading. When compared di-rectly with Landsat images – whichwere in a valid geographical projec-tion – many of the obvious features(roads, rivers, and city boundaries)– did not match. The State Depart-ment maps appeared to present adistorted image of the affected ar-eas.

Since returning from Turkey, EQEspent considerable time trying to“warp” the images so that the StateDepartment maps actually coin-cided with the major land featuresand roadways of the region. As itturns out, it is possible to approxi-mately fit these maps to the correctprojections by overlaying the mapsonto available Landsat images and“warping” the map to match obvi-ous landmarks (e.g., roads, rivers,major intersections, etc.) The nextsection discusses how the MCEER/EDM team attempted to validate thedamage maps.

Figure 12 shows a Landsat 5 im-age of the Adapazarı area. Shown onthis same figure is the route takenby the investigation team on Octo-ber 2nd. It is interesting to note thatat the time of our visit, there ap-peared to be no publicly availablemaps of the city.2 Therefore, ouronly means of tracking our routewas to plot the latitude/longitudeof the van – as determined from theportable GPS system – directly ontoa Landsat image of Adapazarı. Al-though it took some time to set up,this system was invaluable in assign-ing geographical coordinates tospecific damage sites.

The route, shown in Figure 12 asa black line, began near the Mosque

U.S. State Department

� Figure 11. Earthquake damage map produced by the U.S. State Departmentfor the Adapazarı region (August 20, 1999) - uncorrected.


(center of the figure) and pro-ceeded in a counterclockwise di-rection. Also noted on Figure 12are color-coded circles that repre-sent various levels of site damage(on a city block level), ranging fromnone or slight to catastrophic dam-age. Unlike the Avcılar survey, wedid not have access to DifferentialGPS. Therefore, some wavering ofthe route trace is noted. In theory,the coordinates that are registeredare within a block of the actual lo-cation.

To help select the best route totake, we enlisted the services of agovernment worker who was famil-iar with many of the city’s recon-struction projects. With her help,we were able to map out a routethat took us through the mostheavily damaged areas, as well asother areas (generally outside of

town) which did not suffer muchdamage. Most importantly, thisroute took us in and out of thoseareas that were classified as cata-strophic and extensive damage bythe U.S. State Department.

In addition to assigning damagelevels to each block, we attemptedto record this information via (1)digital still cameras, and (2) digitalvideo camera.3 Also recorded vialaptop computer were writtencomments regarding our damageobservations at specific sites. Someof these comments appear on Fig-ure 12. Note on the figure, the areadesignated as “Circle Area.” This wasone of the most devastated areas inthe city.

Figure 13 shows two ERS radarimages of the Adapazarı area, takenbefore and after the earthquake. Asin Figure 12, the team’s route is seen

� Figure 12. Landsat 5 image showing Adapazarı (light area in center of figure) and the route takenby the MCEER/EDM team on October 2, 1999 (route shown as a continuous black line). Also seenare individual damage assignments that were made along the route. Note: A damage index of 1 isassociated with an observation of slight to no damage; an index of 5 reflects catastrophic damage.

“As part of anongoingresearch project,researchers atthe Universityof SouthernCalifornia areexploring waysin which SARdata can beused to char-acterizeearthquakedamage tobuildings.”

92

in these figures as a string of smallcircles. Unlike the previous image,which was optical, the images inFigure 13 were derived from pro-cessed SAR data and show up aspixels of varying intensity. Eachpixel in the raw data set representsa rectangular area of approximately4 meters by 21 meters, depending

on the terrain. Since each of theseimages was taken from the samesatellite track, the view or positionof the image is similar.

To process these data, EQE im-ported the single-look complex im-ages into the ENVI (the Environmentfor Visualizing Images) software, aspecial imaging processing programdesigned specifically for remotelysensed data. The single-look com-plex images provide the highestresolution possible using a SAR im-aging system. The resolution or pixelsize is dependent on a number offactors including radar hardware pa-rameters. For our data set, the pixelsize is about 4 meters in the azi-muthal direction (i.e., the directionof the radar platform) and 21 metersin the cross direction. In order toreduce the inherent noise in the ra-dar image (e.g., speckle noise), weapplied an averaging filter in the azi-muthal direction. This process is nor-mally referred to as multi-lookaveraging. This reduction in noise,however, comes at the expense ofreduced spatial resolution. The endproduct, after this processing, is a 21meter by 21 meter pixel.

We anticipate that with this reso-lution, it will be difficult to recog-nize from these images physicalstructures having dimensions ofless that 20 meters. Therefore, un-less a structure takes up several pix-els, it may be difficult to assess anychange between before- and after-earthquake images. We are also in-vestigating a “correlation” approachto detect pixel level changes; how-ever, these analyses are not com-plete at this time.

The two images that are shownin Figure 13 were taken almost fivemonths apart. The first image wastaken in late spring (4/24/99); thesecond image was taken less than

(a)

(b)

� Figure 13. Intensity images of Adapazarı based on before and after earthquakeSAR data (Data Source: ESA). The light areas correspond to built-up areas,which generally demonstrate a higher degree of reflectance or brightness. Amedian and Frost filter were applied to reduce white noise and maintain edges.


one month after the August 17th

earthquake (9/10/99). Generally,there would be image differencescaused by seasonal change. How-ever, for this area, temperature dif-ferences within that time span areexpected to be mild (55 °F in latespring to 70+ °F in summer); aver-age monthly rainfall between thesetwo periods is also expected to varyonly slightly.

In general, pixel intensity willdepend upon vegetation level, thedensity of buildings and theirshapes, moisture levels, season, etc.Brighter areas are those that reflectmore of the radar energy sent downfrom the sensor. In Figures 13a and13b, the brighter areas are associ-ated with the built-up area of Ada-pazarı. Both images were exportedto MapInfo where they were geo-referenced to the Landsat 5 data.Since the resolution of these imagesis fairly coarse, we generally seeonly the major features of the re-gion, e.g., major roadways andstreets and some very large build-ings. A cursory examination of bothimages does show some differencesbetween the two figures. In orderto view these changes in detail, we

must concentrate on smaller por-tions of these images.

In order to identify any significantchanges between the two images,we had to isolate part of the previ-ous figures and concentrate onthose areas that the team spentsome time investigating while inthe field. Once such area is the“Circle Area” which shows up at thetail end of the route (see Figure 12for approximate location). To ex-amine the changes between thetwo images, we drew a circle of0.25-mile radius around this area.Comparison of the two images re-veals that the after-earthquake im-age (Figure 14b) contains fewerbright spots. As was explained be-fore, many of the buildings in thisarea had either completely failed(total collapse) or were severelydamaged (many tilted buildings). Itis expected that this type of dam-age would result in more scatter-ing and thus, duller images.

Figure 15 shows two photo-graphs of damage in the “CircleArea.” The first photograph (Figure15a) is taken looking west along themain road. The second photographis taken from the same location,

(a) (b)

� Figure 14. Close-up of “Circle Area” in Adapazarı (see Figure 12 for location). Note the designations (a) and (b) in Figure14b which refer to photographs (a) and (b) in Figure 15. North is up. Raw data supplied by ESA.

94

looking northeast. Note that manyof the damaged buildings have beentorn down and the debris hauledaway (Figure 15a). Since the sec-ond image was taken about onemonth after the earthquake, it ispossible that the bare ground wasbeing imaged at that time. At anyrate, the second image was able topick up this change.

In the next several months, wewill be examining these data inmore detail. In addition to the“Circle Area,” we hope to examineother sites where we have collectedfield data on the earthquake. Oneuseful source is a digital video thatwas taken on the second trip toAdapazarı in November. By exam-ining these images in more detail,and perhaps, by creating a series ofcorrelation or coherence maps, wecan begin to explain the meaningof these image changes. In addition,we hope to return to Turkey to col-lect other data that may help toquantify the extent of damage tothis town.

SummaryAlthough largely untested in

earthquakes, all of the new tech-nologies employed during our re-connaissance proved to beinvaluable in documenting the ef-fects from this earthquake. Listedbelow are some of the lessonslearned while performing this in-vestigation.

1. The Landsat images proved to beinvaluable in calibrating on-ground data, such as the loca-tions of major roads, theboundaries of urban regions, andin some cases, the location oflarge buildings, i.e., buildingswith large footprints. Becausethe resolution was fairly coarse(30 meters), it was difficult touse these data to detect or quan-tify earthquake damage. Theseimages, however, were easilyprocessed and were availablesoon after earthquake.

2. The ERS-1 and ERS-2 data provedto be useful once the MCEER/

(a) (b)

Photographs by R. Eguchi

� Figure 15 . Damage to apartment buildings in the “Circle Area” of Adapazarı. Left image is view (a) in Figure 14b. Right imageis view (b) in Figure 14b.


EDM team left the field and re-turned home. Although thesedata were available before thetrip, it was difficult to processthis information because of mapregistration problems, and be-cause the data contained morethan just image data. We are nowbeginning to work with thesedata to explore how useful theyare in detecting damage throughinterferometric techniques. Wespeculate that correlation or co-herence maps developed froman analysis of pre- and post-earthquake images will result inthe best use of these data andcould possibly detect areaswhere significant damage (e.g.,collapsed buildings) has oc-curred.

3. The use of GPS equipment wasessential in documenting theactivities of the team. Precise co-ordinates were established forimportant damage sites. This in-formation was crucial in relatingsatellite imagery data to on-ground observations. Also, whenconnected to the portablelaptop computer, many criticalanalyses were possible in real-time. There was also a signifi-cant difference in results whenDifferential GPS was available.In Avcılar, geographical coordi-nates were accurate to withinone meter. In Adapazarı, where

Differential GPS measurementswere not possible, coordinateswere accurate to within 30meters. This difference could becritical when attempting todocument damage to individualbuildings.

4. Having access to in-field GIS soft-ware made the documentationprocess extremely efficient.Many of the records – such asthe Adapazarı survey – wouldnot have been possible had it notbeen for the GPS-GIS setup. Theactual survey that was discussedin the previous section took lessthan 4 hours. If paper maps orother manual methods of docu-menting damage had been em-ployed, it would have taken atleast several days to accomplishwhat was done in half a day us-ing these new mapping tech-nologies.

5. One important piece of equip-ment – which was used by in-vestigation team members forthe first time – was a digital cam-era. The advantages to usingsuch a camera is that the imagesthat are taken are immediatelyviewable, they can be down-loaded immediately for transmis-sion to some other site, andwhen connected to a GPS unit(which was not done on thistrip), could produce more reli-able documentation of an event.

1 SPOT is a French company that provides satellite imagery data throughout the world.SPOT stands for Satellite Pour l’Observation de la Terra.

2 As it turns out, one of the guides that we used during this half-day trek had a touristmap of the city, which she kindly turned over to us as we departed.

3 A digital video camera was used on a second trip to Adapazarı on November 20,1999.

Endnotes

96

The MCEER/EDM team would like to acknowledge the Multidisciplinary Center forEarthquake Engineering Research for its generous support of this investigation. Inaddition, we would like to thank the following individuals and organizations for theirhelp in making this a unique and productive trip. Without this support, many ofthe activities mentioned above would not have been possible. They are: Dr. BetlemRosich, ESA/ESRIN; Mr. Serkan Bozkurt, IMAGINS; Mr. Turgay Türker, TürkerEngineering; Professor Mustafa Erdik, Kandilli Observatory and Earthquake ResearchInstitute; Dr. M. Namık Yalçın, Dr. Semih Ergintav, and Dr. Hülya Yıldırım, TÜBI·TAK;Mr. I·smail Barıçs, Mayor of Gölcük; Mr. Larry Roeder, U.S. State Department; and Dr.Charles Scawthorn and Ms. Hope A. Seligson, EQE International.

Acknowledgments

97

Research Objectives

The primary objective of the research is to provide detailed informa-tion on various social aspects of the Marmara earthquake. In particular,the research focuses on activities aimed at restoring basic social func-tions to the impacted region during the early recovery phase of thedisaster. This paper examines three of those functions: (1) housing, (2)education, and (3) health care. The paper concludes by discussing nu-merous future research needs that stem from this earthquake in theareas of disaster preparedness, emergency response, social recovery,and mitigation.

Restoration Activities Followingthe Marmara, Turkey Earthquakeof August 17, 1999


By any standard or definition, the earthquake that struck northwesternTurkey on August 17, 1999 was a major disaster. Measuring 7.4 on the

Richter scale, the earthquake was centered near the cities of I·zmit, Göl-cük, and Adapazarı. It damaged or destroyed over 200,000 buildings, lefthundreds of thousands of people homeless, and, according to official esti-mates, resulted in the deaths of nearly 17,000 people. The earthquake alsohad a major impact on large industrial facilities in the region, and esti-mates of its economic impacts vary between 5 billion and 10 billion U.S.dollars. While preliminary estimates of the economic costs associated withthe earthquake vary widely, actual costs will likely be substantial given thesheer magnitude of the event. Because the earthquake occurred in a largelyurban and industrialized area, it resulted in widespread physical damageand severe social and economic disruptions.

This paper describes activities that were initiated to restore social rou-tines to the impacted region during the early recovery phase of the disas-ter. Three basic social functions are described: (1) housing, (2) education,and (3) health care. The first part of this paper describes some of themajor housing issues that arose following the earthquake, including diffi-culties associated with estimating the number of homeless and the estab-lishment of large “tent cities.” The second section addresses the issue ofrestoring education, particularly in Gölcük, and discusses the challenge ofdetermining when it was appropriate to resume school and deciding how

by Gary R. Webb

Gary R. Webb, Post-doctoralResearch Fellow, DisasterResearch Center, andDepartment of Sociology,University of Delaware

Ronald T. Eguchi, President,and Charles K. Huyck,Vice President, ImageCat,Inc., formerly of EQEInternational

Babak Mansouri, Ph.D.Candidate, Department ofCivil Engineering,University of SouthernCalifornia

98

to best do that. The third sectiondescribes how some hospitals inthe region were impacted and howthey responded following theearthquake.

Housing and theEarthquake

During the MCEER team’s trip toTurkey (September 28 to October5, about six weeks after the earth-quake occurred), the most promi-nent and salient social aspect of theevent was housing. Because theearthquake was so physically de-structive, it displaced an enormousnumber of people from theirhomes, all of whom needed alter-native living arrangements. As willbe discussed below, however, forvarious reasons it is difficult toknow exactly how many peoplewere left without homes in theearthquake’s aftermath.

Following major disasters like theone in Turkey, the provision of tem-porary sheltering and housing istypically a priority in early at-tempts to restore normal socialfunctioning. While in the first fewhours and days after a major eventthe focus is likely to be on imme-diate response activities such assearch and rescue and the deliveryof emergency medical services, the

sheltering and housing process alsotypically begins fairly quickly. Interms of characterizing the socialaspects of disasters, housing is acrucial component of the entireprocess.

Estimating the Number ofHomeless

The task of estimating the num-ber of tent cities that were estab-lished after the earthquake and thenumber of people living in themproved to be a major challenge.Some estimates suggest that theearthquake destroyed or badly dam-aged 120,000 housing units, leavingas many as 600,000 people withouthomes. Other estimates suggestthat approximately 120,000 peopleare living in 200 tent cities through-out the region. In either case, thenumber of people left homeless inthis disaster is very large, and it willbe important to generate more ac-curate estimates as plans are devel-oped for more permanent livingarrangements.

The need for more accurate esti-mates is heightened due to the factthat winter can be bitter in the im-pacted region. Because most of thetents in which people were livingat the time of the team’s visit werenot adequate for extreme winter

Because this research focuses broadly on the social aspects ofthe Marmara earthquake, it is anticipated that numerous groupswill have an interest in the issues discussed. Researchers in theUnited States, for example, can use the material presented hereas a basis for making cross-cultural comparisons of social re-sponses to disaster. The research will also be useful to policymakers and other officials who must make crucial decisionsabout disaster response priorities and recovery alternatives.Emergency management officials and relief workers from vol-untary associations, government agencies, and non-governmentorganizations will also benefit from the research findings.

Faruk Birtek, Professor,Department of Sociology,Bogaziçi University,I·stanbul, Turkey

Ayçse Akalın, GraduateStudent, Department ofSociology, BogaziçiUniversity, I·stanbul,Turkey

Elif Kale, Graduate Student,Department of Sociology,Bogaziçi University,I·stanbul, Turkey

99Restoration Activities following the Marmara, Turkey Earthquake

weather, plans were being dis-cussed to bring in stronger wintertents and some pre-fabricated build-ings. Therefore, as officials beganmaking housing arrangements, theycould have benefited from an ac-curate estimate of the number ofpeople living in tent cities.

In many U.S. disasters, it is notuncommon for officials to drasti-cally over-estimate public housingneeds because they sometimes donot recognize that many of thosewho are displaced go to live withfriends or relatives whose homeswere not destroyed. It is likely thatsimilar patterns occurred in re-sponse to the Turkey earthquakeand that these patterns may havecomplicated census-taking efforts.For example, in the mountains sur-rounding Gölcük and Adapazarı,two cities that were very heavilydamaged, there are many small vil-lages from which people migratedto live in the larger cities. And manypeople from other parts of thecountry that may be much furtheraway have migrated to these moreurbanized and industrialized citiesto find work. Following the earth-quake, it was not known how manypeople returned to their places oforigin to live with friends or rela-tives and exactly how many peopleremained in the two cities. It mayhave been easier for people fromthe surrounding mountain villagesto return home, whereas peoplefrom more distant places in Turkeymay have been less likely to leavethe area after the earthquake.

In either case, officials do knowthat there has been some migration,but they do not know how much.For example, a health official inAdapazarı indicated that prior tothe earthquake, approximately200,000 people lived in the center

of the city. After the earthquake, thisofficial estimated that only about50,000 to 70,000 remained in thecity. Similarly, an official in Gölcük,which had a population of about75,000 prior to the earthquake, in-dicated that about half that manyremained in the city after the earth-quake.

In addition to internal migrationpatterns and survivors’ reliance onexisting social networks of support,there are other reasons why it isdifficult to officially estimate thenumber of people left homeless bythe earthquake. For example, an-other major impediment to obtain-ing an accurate census is that manypeople (exactly how many is notknown) whose homes were notbadly damaged are neverthelessreluctant to reenter their buildings.Since the earthquake occurred,there have been several major af-tershocks that have instilled hesi-tancy on the part of survivors.

Additionally, some officials indi-cated that although several groupsand organizations are developingcounts for various purposes, theyare not coordinating those effortsclosely enough. For example, somegroups are taking counts in orderto make arrangements for the de-livery of mental health services, andothers may be trying to order ap-propriate amounts of certain sup-plies. With so much activity goingon, however, it is very difficult forthese various groups to collaboratewith each other and coordinatetheir efforts. The result, then, is thatvarious groups and organizationsare taking counts for their own pur-poses, and these numbers are notbeing shared.

In most disaster situations, re-search has shown that both inter- andintra-organizational coordination are

Program 2: Seismic Retrofit ofHospitals


100

often difficult to achieve becausecircumstances change rapidly andbecause numerous organizations(many of which have no mandateor responsibility for emergencies)become involved in the overallcommunity response (Dynes 1970).In situations where various organi-zations are not familiar with one an-other and lack established patternsof interaction and coordination, itis not uncommon for these kindsof problems and issues to arise.

Three Types of Tent Cities

Although it was not possible atthe time of the team’s visit to knowexactly how many tent cities ex-isted and how many people wereliving in them, it was possible todescribe the tent cities and howresidents adjusted to living in them.Basically, displaced people in theimpacted area who have not soughtshelter in other locations were liv-ing in three different types of tentcities: (1) those organized by themilitary, (2) those organized bynon-government organizations andprivate corporations, and (3) thosethat are informally organized. Inreality, it is difficult to classify indi-vidual tent cities because there is

some overlap among these generaltypes. For example, a tent city or-ganized and run by the military mayalso offer some services to residentsthat are performed by a voluntaryor non-government organization.Similarly, a tent city organized by aprivate corporation may integratenon-government organizations intoits service delivery system and relyon military personnel to provide se-curity. In a very general sense, how-ever, it is useful to organize thenumerous tent cities into thesethree general types.

In terms of size, the largest tentcities seem to be those that are or-ganized either by the military or byprivate corporations or non-govern-ment organizations. At one of themilitary-run tent cities in Gölcük,for example, 3,000 people are liv-ing in tents that cover a large landarea (shown in Figure 1). Anothertent city in Gölcük set up by a largemanufacturer in the area housesapproximately 3,700 people. Infor-mal tent cities, which are comprisedmainly of neighborhood groups liv-ing outside in tents near their homes(which may or may not be badlydamaged), are scattered throughoutthe region and tend to be compara-tively small (see Figure 2). It wasdifficult at the time of the visit toascertain the proportion of peopleliving in the various types of tentcities for the same reasons dis-cussed above. For example, the cri-sis response center in Gölcükreported the existence of 12 tentcities in Gölcük, but one adminis-trator knew of at least 21 differenttent cities. A more accurate countmay improve the efficiency and ef-fectiveness of the delivery ofneeded supplies, and it may ulti-mately help in the arrangement of

Photograph by G. Webb

� Figure 1. A military-run tent city in Gölcük

Multidisciplinary Centerfor EarthquakeEngineering Research

http://mceer.buffalo.edu/research/turkeyeq/default.html


adequate provisions for the future,particularly for winter.

A more accurate census of tentcities and people living in themwould also make it possible to com-pare the different types along sev-eral dimensions. For example, onthe one hand, a clear benefit of theinformally organized tent cities isthat they allow primary socialgroups (i.e., families, extended fami-lies, and close friends) to live neareach other and rely on each otherfor social support. While adminis-trators of the other two types of tentcities have tried to keep these group-ings intact, they are less able to doso as these tent cities increase dra-matically in size and as space be-comes less plentiful in them. On theother hand, the larger military- andvolunteer-run tent cities may be ableto offer a wider range of services toresidents, including large kitchens,pharmacies, counseling services, en-tertainment, and kindergarten forsmall children. For example, at oneof the large military-run tent citiesin Gölcük, a civic group fromI·stanbul (which had no prior in-volvement in disasters) establisheda kindergarten for young people.

This involvement of non-emer-gency organizations in providingservices after the earthquake issimilar to what occurs in many U.S.disasters, and it is an issue thatshould be explored further throughcross-cultural and cross-societalcomparisons of disaster responses.If there are important differencesin the type and quality of servicesoffered at the various types of tentcities, and if there are certain ben-efits and limitations to each of them,they should be used as lessons forfuture disasters when mass num-bers of people must be temporarilyrelocated.

Adjusting to Daily Living in theTent Cities

Following major disasters like theone in Turkey, the establishment oftent cities serves several importantfunctions: it provides necessaryshelter from the elements; it re-es-tablishes and reaffirms communityand collective solidarity; and it be-gins to provide a stable base fromwhich people can start to restoretheir daily routines. In some of thelarge military- and volunteer-runtent cities, for example, meals areserved at certain times each day,residents engage in routine reli-gious rituals and perform basic rou-tines like doing laundry, youngpeople play soccer and attend kin-dergarten, and some adults leaveeach morning to go to work. Justacross the street from one of thetent cities, a market has emergedthat provides residents a place togo to purchase basic items. And lo-cal bus companies have altered theirroutes to provide transportationfrom the tent cities to various pointsthroughout the city. All of these ex-amples illustrate the point that whensocial routines are severely dis-rupted, individuals, groups, and or-ganizations improvise and adapt in


� Figure 2. An informal tent city

102

creative ways as they attempt to re-store normalcy to the social order(Bosworth and Kreps 1986; Krepsand Bosworth 1993; Quarantelli1996; Webb 1998).

One of the most important issuesthat individuals andfamilies face underthese circumstances isthe challenge of mak-ing a temporary livingarrangement into ahome that provides allits members withsafety and comfort. Ina social psychologicalsense, this means re-building or re-estab-lishing an attachmentto place that provides

security and stability in daily inter-actions. When a major disaster dis-rupts a group’s attachment to place,its members interact to develop anew definition of the situation thatgives meaning to their experiencesand guides them through their in-teractions with others.

One of the most noticeable thingsabout life in a large tent city is thatresidents have almost no privacy. Inthis kind of living arrangement, vir-tually every aspect of an individual’slife is on public display. To mini-mize or alleviate that problem, resi-dents often adapt in some verycreative ways. For example, occu-pants of a tent will make additionsthat divide it into two separatespheres: a front area where they cansit and talk with others, and a backarea where they sleep and preparefor the day. Figure 3 nicely illus-trates how this is accomplished. InFigure 4, a tent is shown that is “un-der construction,” and Figure 5shows a finished product that ac-tually resembles a small house.

In a sociological sense, these in-novations are very meaningfulbecause, although there are impor-tant cultural differences, the sepa-ration of public and private spheresis a crucial component of social life.In their interactions with others,


� Figure 5. A finished product


� Figure 4. A tent under construction


� Figure 3. A “renovated tent”


individuals make decisions aboutwhat things to openly present inthe front stage region and whatthings to display only in the backstage region (Goffman 1959). In avery basic sense, for example, hous-ing units in many cultures are de-signed with a living area intendedto be on display to guests and asleeping area that is typically notopenly displayed. By making thesekinds of additions to their tents,many of which are fairly elaborate,residents of the tent cities are re-defining a fundamental aspect oftheir social lives.

Residents are also engaged in theprocess of building new social re-lationships and a sense of commu-nity. For example, as shown inFigure 6, the residents of a particu-lar row of tents in a very large tentcity in Gölcük gave themselves astreet name and erected a sign bear-ing the words “Save Me Street”(translated). This example nicely il-lustrates how under conditions ofextreme stress people rely on eachother and the relationships theyhave to give meaning to their ex-periences. Similarly, in other cases,community members often spraypaint graffiti after disasters to ex-press either messages of hope andcollective solidarity, discontent

with the official re-sponse, or simplyconvey basic infor-mation. As shownin Figure 7, resi-dents of one tentcity stretched largewhite bannersacross a fence sur-rounding a play-ground and paintedvarious things onthem, including theslogan “Let’s not

Forget Gölcük.”These expressions of solidarity

and hope may account for some ofthe debate about the delivery ofmental health services followingmajor disasters (Quarantelli 1985).In many U.S. disasters it is often as-sumed that these services will bewidely needed, but in many casessurvivors do not seek out that kindof assistance. At some of the tentcities in Turkey, there have also beensome concerns that residents are notutilizing mental health services tothe degree that they should. Theremay be two reasons why disastersurvivors do not always seek men-tal health services after a majorevent: first, they may find comfortand support in their interactionswith significant others who have


� Figure 7. Graffiti in Gölcük


� Figure 6. A “street sign” in a tent city

“Some estimatessuggest that theearthquakedestroyed orbadly damaged120,000housing units,leaving asmany as600,000 peoplewithout homes.”

104

also experienced the event; and sec-ond, some of the mental health con-sequences may not necessarily benegative. As some of the examplesabove show, when a community hasbeen severely disrupted, its mem-bers rely on existing social relation-ships and newly formed ones incoping with the emergency andbeginning to restore normalcy totheir lives. Sometimes disasters mayenhance, if only temporarily, thecollective solidarity felt by mem-bers of a community because theyall share a common experience.Clearly, there is a need for more re-search on the mental health conse-quences of disasters, and theearthquake in Turkey provides asetting where important cross-cul-tural and cross-societal compari-sons can be made.

Restoration ofEducation After theEarthquake

Another basic social institutionthat is often disrupted in a disasterand that must be restored, is edu-cation. In addition to providingyoung people knowledge theyneed to become adult members ofa society, schools also serve the cru-cial function of keeping a substan-tial portion of a populationoccupied and on a rigorous dailyschedule. When that schedule isinterrupted, a certain amount of am-biguity and confusion is created, soofficials typically try to resumeschool as quickly as possible. Theirconcern is often not only to get stu-dents back in school for learningpurposes, but also to give structureand meaning to young people’slives in a period of confusion anddisruption. The restoration of daily

activities, including school for chil-dren, is a crucial part of communityresponse to and recovery from a di-saster.

The earthquake in Turkey oc-curred almost one month beforeschools across the country wereoriginally scheduled to begin onSeptember 15. Schools in many ar-eas did begin as scheduled, butwhen a major aftershock occurredon that same day, all school open-ings were indefinitely postponed.Ultimately, schools in areas such asI·stanbul that did not sustain heavydamage began operations on Mon-day, October 4. Even that causedsome controversy because manyparents in those areas did not un-derstand why their children werebeing held out of school for so long.In more heavily damaged regions,such as Gölcük, it was hoped thatschool could begin in early Novem-ber, significantly later than origi-nally planned.

There are several reasons why theopening of schools in Gölcük wasdelayed for such a lengthy periodof time. First, some of the schoolbuildings themselves sustainedheavy damage, so pre-fabricatedstructures or large tents would beneeded to conduct classes. Second,many teachers, students, and par-ents expressed major concernsabout reentering even those schoolbuildings that had not been badlydamaged in the earthquake. Finally,in Gölcük it was not known howmany students in the area would bereturning to school. As was dis-cussed above in relation to hous-ing issues, migration patternsstrongly affected the population ofstudents in the most severely im-pacted areas, and that made it diffi-cult to estimate the number ofreturning students for whom plans

“When acommunity hasbeen severelydisrupted, itsmembers rely onexisting socialrelationshipsand newlyformed ones incoping with theemergency andbeginning torestorenormalcy totheir lives.”


should be made. Many studentswere believed to have returnedwith their families to either villagesin the surrounding mountains fromwhich they came or to more dis-tant parts of the country to livewith relatives or close friends. Aschool official in Gölcük, for ex-ample, estimated that only 10,000of the area’s 28,000 students wouldreturn to school.

To facilitate the opening ofschools in the most heavily dam-aged regions, the Turkish govern-ment commissioned one of themajor universities to become in-volved. According to some of thoseinvolved in the project, the commis-sion has two major goals: first, toget an accurate estimate of thenumber of students who will be re-turning to school; and, second, toconduct focus groups with teach-ers, students, and parents to betterunderstand their anxiety about re-entering school buildings that werenot damaged.

Some school officials, however,indicated that this approach maynot be the most efficient and effec-tive way to go about restoringschool to the most badly damagedareas. For example, one official sug-gested that it may be more produc-tive to establish schools throughoutthe region in large tents and seehow many students report. If thedemand were to exceed the num-ber of tent schools established, thenadditional ones could be set up.This case illustrates that a majordisaster can disrupt even the mostbasic social institutions and that keyparticipants often have differingperspectives on how to respond. Insome disasters, such as the recentearthquake in Taiwan, schools re-sume fairly quickly, but there are

certainly times when a prompt res-toration is not possible. Thus, thereare important lessons to be learnedfrom the recent earthquakes interms of understanding barriers tothe restoration of basic social insti-tutions and identifying strategiesthat are particularly effective.

Health Care Facilitiesand the Earthquake

Another basic social function thatis sometimes disrupted in majordisasters is health care. These dis-ruptions occur either because thenumber of fatalities and injuriesexceeds the health care system’scapacity to respond or the healthcare system itself sustains physicaldamage which affects its ability todeliver services. In either case, thisis another area in which the socialsystem often becomes very flexibleand adaptive in meeting heightenedemergency demands.

In both I·zmit and Adapazarı, sev-eral hospital buildings experiencedmajor physical damage during theearthquake. At a hospital in I·zmit,for example, two buildings and thepedestrian walkway connectingthem sustained serious damagefrom the shaking. Immediately af-ter the earthquake, medical staffwere forced to evacuate the build-ing and move existing patients out-side. As people began bringing theinjured to the hospital’s emergencydepartment, it quickly became con-gested and overcrowded. To allevi-ate the crowding, hospital staffbegan assembling the injured in aschool yard across the street, sort-ing and tagging them by severity ofinjury, and transporting them toother regional facilities. Similarly,existing patients who could not

“Schools servethe crucialfunction ofkeeping asubstantialportion of apopulationoccupied andon a rigorousdaily schedule.”

106

simply be discharged early werealso transferred to other facilities.

What is most interesting aboutthis particular hospital is that evenfive weeks after the earthquake,staff still had not yet reentered thebuildings. On two occasions theytried to reenter, but when major af-tershocks occurred, they returnedoutside. Because they were forcedto set up operations in tents in theparking lot (shown in Figure 8), thehospital has been unable to resumeits normal functioning. For ex-ample, minor injuries are treated,basic exams are conducted, andmedications are dispensed, but it isnot possible to perform major medi-cal procedures. This case showsthat even emergency relevant orga-nizations such as hospitals canthemselves be impacted by disas-ters, and, when they are, they mustbecome flexible and adaptive un-der the circumstances.

Similarly, a major hospital in Ada-pazarı sustained extensive physicaldamage to its buildings. In particu-lar, two of the facility’s five build-ings were damaged, forcing staffmembers to move existing patientsoutside. As was the case in the pre-vious example, existing patients andnew arrivals were assembled out-side, sorted and tagged by severity

of injury, and transferred to otherregional hospitals or field hospitalsset up by international relief orga-nizations. According to one official,this process created some confu-sion at the hospital because staffmembers were not able to docu-ment all of the victims who wereseen and transferred. The receiv-ing hospitals later created detailedlists and sent them back to this fa-cility, but the delay in that processcreated some confusion as con-cerned people came to the hospi-tal looking for their friends orrelatives. This official also indicatedthat for a short period of time therewas a shortage of trained medicalstaff in the immediate aftermath ofthe earthquake. That problem wasresolved, however, as volunteersquickly began arriving. The sameofficial pointed out that this staffshortage existed in Adapazarı evenbefore the earthquake.

While staff members have reen-tered the facility in Adapazarı, thehospital still had not resumed itsnormal functioning at the time theresearch team visited. For example,major medical procedures stillwere not being performed at thehospital, and staff from other cit-ies who came to volunteer werestill living in tents in the parkinglot. At the time the team visited,the hospital in Adapazarı had fo-cused its activities on providingbroader public health services. Forexample, officials began producingand distributing brochures andpamphlets that describe how totreat water, prepare food, and avoidbacterial diseases. In addition, lo-cal health officials have been in-volved in monitoring the city’ssupply of clean water, much ofwhich was being hauled in by tank-ers from a nearby lake. One official


� Figure 8. A hospital in I·zmit


gladly reported that there havebeen no major outbreaks of bacte-rial diseases in the region, in largepart because of the activities under-taken by staff members at the hos-pital.

These examples of hospitals inTurkey highlight two importantpoints. First, they clearly show thathealth care facilities, which are usu-ally assumed to be operational inmass emergency situations, cansometimes experience physicalimpacts themselves. And, second,when hospitals are impacted bydisasters, their basic structures andfunctions are often altered to meetheightened demands created bythe emergency. Although these al-terations and innovations are oftenfunctional and adaptive, they maysometimes be dysfunctional andmaladaptive. In either case, it can-not be assumed that hospitals willalways be operational in the after-math of a major disaster event. Infact, surprisingly little research hasbeen done that documents exactlyhow prepared hospitals are for di-sasters and how they actually func-tion when disasters do occur.Clearly, this is an area where muchmore research is needed, and theearthquake in Turkey provides asituation where cross-cultural andcross-societal comparisons can bemade in describing and understand-ing how hospitals function understress.

Future ResearchNeeds

This section draws out some ofthe future research needs that werementioned in the previous sectionsand presents others that were notexplicitly stated. There are several

areas in which the earthquake inTurkey provides a setting in whichinteresting cross-cultural and cross-societal comparisons can be madealong several dimensions. First, ata very basic and descriptive level,it would be useful to do further re-search on the various organizationsthat have become involved in theresponse to and recovery from theearthquake. As was described in theprevious sections, various organiza-tions, many of which have no de-fined disaster responsibilities, havebecome involved, and many ofthose that do have disaster respon-sibilities, such as hospitals, have sig-nificantly altered their basicstructures and functions. This kindof organizational innovation andadaptation has often been docu-mented in studies of U.S. disasters,so there is a unique opportunity tomake comparisons with the situa-tion in Turkey. These studies caneither document how specifictypes of organizations, such ashospitals, responded to the earth-quake or focus on the issue of in-ter-organizational coordinationbetween various types of organiza-tions. This kind of research can bereadily translated into lessonslearned and ultimately usefully in-tegrated into the practice of emer-gency management.

The broader issue of social recov-ery is another area in which impor-tant cross-cultural comparisons canbe made. Several studies havelooked at the process of household(Bolin 1994) and business recovery(Dahlhamer and Tierney 1998;Tierney and Dahlhamer 1998;Webb, Tierney and Dahlhamerforthcoming) in the U.S., so theearthquake in Turkey presents anopportunity for comparative re-search. There are several important

“Existingpatients andnew arrivalswere assembledoutside, sortedand tagged byseverity ofinjury, andtransferred toother regionalhospitals orfield hospitalsset up byinternationalrelieforganizations.”

108

areas in which the earthquake inTurkey can improve our broaderunderstanding of community re-covery as a social process (Nigg1995). For example, future studiesmight document the impact of thisevent on the collective memory ofthe people who experienced it andassess the degree to which it mayor may not affect subsequent miti-gation decisions aimed at reducingthe impacts of future disasters.Along those same lines, it will beinteresting to measure the local, re-gional, national, and internationaleconomic impacts of the earth-quake and monitor the progress ofeconomic recovery. As one ex-ample of this, officials in Gölcükhave already expressed differingviews of how best to promote eco-nomic recovery—some want torebuild the existing downtownbusiness district, while others wantto relocate it away from the sea andcloser to the mountains. Thesekinds of perspectives and debateswill likely intensify in the comingmonths, and that process should bestudied.

In addition to studying the processof economic recovery, there is a tre-mendous amount to be learned fromstudying the transition of earth-quake survivors from temporary topermanent housing. On a practicallevel, there are valuable lessons tobe learned from this case as officialstry to place tens or even hundredsof thousands of people in more per-manent living arrangements. And, ona conceptual level, it will be impor-tant to understand how individuals,families, and groups construct mean-ing in their daily lives in the tent cit-ies and beyond and how theyre-establish their attachment toplace under conditions of such ex-treme uncertainty.

Future research should also bedone to explore the many politicalimplications of the earthquake(Sylves 1998). For example, somecommentators have suggested thatthe earthquake has promoted a cer-tain critical sentiment among Turk-ish citizens and that for the firsttime they are speaking out againsttheir government and criticizing itsresponse to the disaster. Othershave suggested that the sympa-thetic outpouring of internationalrelief reflects improved relationsbetween Turkey and other nations.Whether these changes were in-duced by the earthquake or simplyaccelerated by it, the political di-mensions of this disaster will alsobe important to consider.

Finally, the recent earthquake inTurkey also provides a setting inwhich to assess the applicabilityand utility of advanced damage as-sessment technologies and loss es-timation methodologies (see theMCEER reconnaissance report onthe earthquake, Scawthorn 2000).Moreover, there is a need for re-search that describes what tech-nologies have been employed inresponding to and recovering fromthe earthquake, assesses their util-ity, and identifies areas in whichtechnologies can be improved toenhance response capabilities.

Whether the knowledge gainedfrom that research is used to pro-mote disaster preparedness, en-hance emergency response,facilitate social recovery, or suggestcertain mitigation measures, thereis a tremendous amount to belearned from this event. Ultimately,the lessons learned from this earth-quake should be translated intomeasures that either reduce theimpacts of future disasters or im-prove societal responses to them.

“Ultimately, thelessons learnedfrom thisearthquakeshould betranslated intomeasures thateither reducethe impacts offuture disastersor improvesocietalresponses tothem.”


Bolin, R., 1994, “Postdisaster Sheltering and Housing: Social Processes in Response andRecovery,” in R. Dynes and K. Tierney (eds), Disasters, Collective Behavior, andSocial Organization, Newark, DE, University of Delaware Press.

Bosworth, S. and Kreps, G., 1986, “Structure as Process: Organization and Role,”American Sociological Review, Vol. 51, pp. 699-716.

Dahlhamer, J. and Tierney, K., 1998, “Rebounding from Disruptive Events: BusinessRecovery Following the Northridge Earthquake,” Sociological Spectrum, Vol. 18, pp.121-141.

Dynes, R., 1970, Organized Behavior in Disaster, Lexington, MA, D.C. Heath.

Fritz, C., 1961, “Disasters,” in R. Merton and R. Nisbet (eds), Social Problems, NewYork, Harcourt Brace.

Goffman, E., 1959, The Presentation of Self in Everyday Life, Garden City, NY,Doubleday.

Kreps, G. and Bosworth, S., 1993, “Disaster, Organizing, and Role Enactment: AStructural Approach,” American Journal of Sociology, Vol. 99, pp. 428-463.

Nigg, J.N., 1995, “Disaster Recovery as a Social Process,” Wellington after the Quake:The Challenge of Rebuilding, Wellington, New Zealand, The EarthquakeCommission, pp. 81-92.

Quarantelli, E., 1985, “An Assessment of Conflicting Views on Mental Health: TheConsequences of Traumatic Events,” in C. Figley (ed), Trauma and its Wake: TheTreatment of Post-traumatic Stress, New York, Brunner/Mazel, pp. 173-215.

Quarantelli, E., 1996, “Emergent Behavior and Groups in the Crisis Time of Disasters,”in K. Kwan (ed), Individuality and Social Control: Essays in Honor of TamotsuShibutani, Greenwich, CT, JAI Press, pp. 47-68.

Scawthorn, C., editor, 2000, The Marmara, Turkey Earthquake of August 17, 1999:Reconnaissance Report, Technical Report MCEER-00-0001, Multidisciplinary Centerfor Earthquake Engineering Research, Buffalo, New York.

Sylves, R., 1998, The Political and Policy Basis of Emergency Management,Emmitsburg, MD, Emergency Management Institute.

Tierney, K. and Dahlhamer, J., 1998, “Business Disruption, Preparedness, and Recovery:Lessons from the Northridge Earthquake,” in Proceedings of the NEHRP Conferenceand Workshop on Research on the Northridge, California Earthquake of January17, 1994, Richmond, CA, California Universities for Research in EarthquakeEngineering, pp. 171-178.

Webb, G., 1998, Role Enactment in Disaster: Reconciling Structuralist andInteractionist Conceptions of Role, Ph.D. Dissertation, Newark, DE, University ofDelaware.

Webb, G., Tierney, K. and Dahlhamer, J., Forthcoming, “Businesses and Disasters:Empirical Patterns and Unanswered Questions,” Natural Hazards Review.

References

111

Research Objectives

MCEER and the National Center for Research on Earthquake Engi-neering (NCREE) have had a cooperative agreement to carry out funda-mental earthquake engineering research in areas of mutual interest since1992. Shortly after the devastating Chi-Chi earthquake of September21, 1999, the Directors of the two Centers planned a joint MCEER-NCREE workshop to identify important short-term strategies and ac-tions for post-earthquake restoration and identify research needs. InApril 2000, a second workshop was held between NCREE, MCEER,the Pacific Earthquake Engineering Research (PEER) Center at theUniversity of California, Berkeley, and the Office of National Scienceand Technology Hazard Mitigation of the Taiwan government, to de-velop a highly focused Center-to-Center research program. The pro-gram will incorporate the vast amount of reconnaissance informationgathered in Taiwan and apply it to specific problem-focused researchalready underway at the four Centers. This plan, together with obser-vations of the authors with respect to the societal and governmentresponses, is offered herein.

On Tuesday, September 21, 1999, a devastating earthquake struck thecentral region of Taiwan. This earthquake became known as the 921

earthquake or the “Ji-Ji” or “Chi-Chi” earthquake. The magnitude of the 921earthquake was M

S = 7.6 (Richter scale) or M

L = 7.3 (the system used in

Taiwan). There were ten aftershocks greater than magnitude 6. Of these,an M

L = 6.8 occurred about 30 hours and 120 hours after the main shock,

respectively. An ML = 5.3 aftershock was recorded as late as 260 hours

later causing collapses of already damaged structures. As of October 8, thedeath toll was more than 2,350. Over 8,700 people were injured, anddozens remained missing. Approximately 10,000 buildings/homes col-lapsed and over 7,000 more were damaged.

For the past eight years, MCEER and the National Center for Research onEarthquake Engineering (NCREE) have had a research collaboration agree-ment to carry out fundamental earthquake engineering research in areas

Human and InstitutionalPerspectives of the 921Chi-Chi, Taiwan Earthquake

by George C. Lee and Chin-Hsiung Loh


New York StateUniversity at Buffalo, State

University of New YorkFederal Emergency

Management Agency

George C. Lee, Director,Multidisciplinary Centerfor EarthquakeEngineering Research,University of Buffalo, StateUniversity of New York

Chin-Hsiung Loh, Director,National Center forResearch on EarthquakeEngineering, NationalTaiwan University

Jack Moehle, Director, PacificEarthquake EngineeringCenter, University ofCalifornia, Berkeley

Chin-Lien Yen, Director,Office of National Scienceand Technology HazardMitigation

112

of mutual interest. A num-ber of joint studies in thegeneral area of seismic re-sponse control are inprogress and last year, twoadditional research projectswere added (remote sens-ing applications, and pro-tective systems for bridges).

Under the umbrella ofcontinuing cooperative re-search, the authors dis-cussed the best way to joinforces following the 921earthquake. They decided tohold an MCEER-NCREE workshop,which took place on October 3-5,1999 in Taiwan. The purpose of theworkshop was to identify impor-tant short-term strategies/actionsfor post-earthquake restorationand research needs, including spe-cific cooperative projects for inves-tigators from both centers to workas teams based on the 921 experi-ence. Many reports have since beenpublished by NCREE on a varietyof technical areas (in Chinese) andMCEER has a reconnaissance reportin press at this time (Lee and Loh,2000), which focuses on both tech-nical and societal issues.

This growing body of knowledgeand its vast potential promptedMCEER to examine whether or notour research results, which are car-ried out in concert with achieving

our vision of creating earthquakeresilient communities, could help inTaiwan; and if so, how it could bestbe accomplished. To explore thispossibility in greater detail, a sec-ond workshop was organized inApril 2000 in Taiwan and includedtwo more centers, the PacificEarthquake Engineering Research(PEER) Center at the University ofCalifornia, Berkeley, and the Officeof National Science and Technol-ogy Hazard Mitigation of the Tai-wan government. The Directorsfrom each of the four Centers helddiscussions to further define andclarify how to bring their collectivestrengths together to make this vi-sion a reality. The result was theestablishment of a three-year Cen-ter-to-Center research program.The specific research projects to

Post-earthquake information is valuable for the enhancementand validation of existing models, methods and practices. It isanticipated that the results of this research effort will be usedby earthquake hazard mitigation experts for this purpose, andwill ultimately strengthen our existing knowledge base in a widevariety of areas. On the Taiwan side, one focus will be to vali-date HAZ-Taiwan, to develop more accurate loss estimation meth-odologies. For both the U.S. and Taiwan, new knowledge gainedwill be applicable to code improvements and implementationin both countries.


� Figure 1. Surface faulting caused major damage to theShih-kang Dam

Program 1: SeismicEvaluation and Retrofit ofLifeline Systems

Program 2: Seismic Retrofitof Hospitals


113Human and Institutional Perspectives of the 921 Chi-Chi, Taiwan Earthquake

be carried out under this programare described later in this paper.

The first part of this paper sum-marizes the observations and re-flections of the authors shortlyafter the earthquake with respectto the societal and government re-sponses of the 921 earthquake.The authors are of the opinion thatthis earthquake not only destroyeda segment of Taiwan’s physicallandscape, but also made a signifi-cant impact on the society andgovernment. Taiwan, like the U.S.,is lucky because it hasn’t experi-enced a major destructive earth-quake with large death tolls inrecent memory. This 921 earth-quake presents a reminder and anopportunity for the people andgovernment in Taiwan to begin aserious effort to establish more re-silient communities against futureearthquakes. The second part ofthe paper describes the new Cen-ter-to-Center research program.

Initial Observationsand Reflections

The Public

Earthquake ground motions arefelt by people living throughoutTaiwan. Thus, the term “earthquake”is a familiar one. However, an earth-quake of the magnitude of 921 hasnot happened in recent history. Tomany people, an earthquakeamounted to the swaying of build-ings and the development of crackson a wall. Occasionally, the roof ofsome houses collapsed. The build-ing code in Taiwan provides for rea-sonable design guidelines (the mostrecent update to the guidelines wasmade in 1997). Based on historical

data and measurements made bystrong motion instrumentation pro-grams, the Taichung/Nantou area isclassified as a region of moderateintensity with a design peak hori-zontal acceleration of 0.23 g. The921 earthquake generated a hori-zontal force of more than four timesthis maximum design criterion. Itis thus easy to see why so manybuildings and bridges collapsed.Psychologically, the public in thearea was accustomed to earth-quakes, but not one of such destruc-tive magnitude, occurring inbuilt-up areas.

The public showed tremendousspirit as they worked together tosave lives and help each other withthe basic needs to survive in thedays following the earthquake. Af-ter a day or two, many began tocomplain that the government wastoo slow in its rescue and relief ef-forts. As a few more days passedand people were forced to acceptthe loss of a loved one whose bodyhad not yet been located, thesecomplaints of ineptitude and inef-ficiency were understandably inten-sified. Nonetheless, it seemed thatthe government was actually quiteswift and effectual in its response


� Figure 2. Damage to buildings and bridges was widespreadthroughout the epicentral area

Multidisciplinary Centerfor EarthquakeEngineering Researchhttp://mceer.buffalo.edu/research/taiwaneq9_99/default.html

National Center forResearch on EarthquakeEngineeringhttp://www.ncree.gov.tw

Pacific EarthquakeEngineering ResearchCenterhttp://peer.berkeley.edu/index.html

114

to the event, given its magnitude.At the same time, many peoplepraised the efforts of military per-sonnel and international emer-gency response teams, even thoughsuch help was limited in its effec-tiveness by the scarcity of criticalinformation such as local area maps,building blueprints, and other suchdata. The disaster management ef-fort at the regional and local levelwas clearly unprepared for this di-saster.

In speaking with a variety of in-dividuals, one can see that thisearthquake has had an enormousimpact on the way the public viewsthe importance of building safetyand location of both workplace andresidences. In the short three-dayvisit, questions regarding these is-sues were the most frequentlyraised by the general public. Nowis the time in Taiwan to emphasizepublic education about earthquakehazards and mitigation measures.A well-educated public will affectimprovements in policy regardingmitigation and preparedness forearthquake and other natural haz-ards. In the past, real estate prop-erties for many are the means to

become rich. The landslides, thedisappearance of the lake (reser-voir), and the interrupted skylinesin the city caused by the 921 earth-quake had elevated the awarenessof the public to treasure the smallisland shared by 22 million people.One may expect that environmen-tal conservation and protection willbe emphasized. Activities such asillegal pumping of fresh groundwater for growing seafood (sinkingland surface level) will be con-demned by the public.

The Government

On the national level, the govern-ment seems to have respondedwell. It was certainly not possibleto satisfy all those affected by theearthquake. However, many of thecomplaints stemmed from lack ofpreparedness rather than lack ofemergency action. Within hours ofthe initial main impact, the nationalgovernment announced policies forrelief, short-term restoration andan organized interagency structurefor efficient execution of rescue ef-forts. By September 28th (oneweek later), there were 17 majorpolicies implemented, includinghotlines, information and healthcenters, temporary housing, disas-ter relief funds and materials, andothers.

However, a lack of earthquakedisaster preparedness at both thenational and local levels was evi-dent. To varying degrees, this stateof affairs exists everywhere in theworld with respect to unexpectednatural disasters. Because the oc-currence of devastating earth-quakes is probabilistic in nature,the consequences of such eventsare often not taken seriously by


� Figure 3. Newly constructed buildings suffered severe damage

“An effectiveinstitutionalstructure forearthquakehazardmitigation isneeded todevelop well-preparedcommunities.”


either the government or its con-stituents. In recent decades, theprofessional communities and gov-ernment in Taiwan have made sig-nificant progress to mitigateearthquake hazards by, for ex-ample, funding the Strong MotionInstrumentation Programs at theCentral Weather Bureau of the Min-istry of Transportation and Commu-nication, and funding earthquakeand earthquake engineering re-search projects by the National Sci-ence Council, including theestablishment of NCREE. The Minis-try of the Interior and the variousstructural engineering professionalorganizations have also been ac-tive in updating building codes,and the Ministry of Education hasbeen investing in human resourcesdevelopment and earthquake engi-neering facilities at the universities.Additionally, a National Science andTechnology Program for hazardmitigation was established severalyears ago to coordinate the devel-opment of national hazard mitiga-tion strategies.

All these efforts have been car-ried out by many talented research-ers and administrators. They nowbeg the question, “What differencedid these programs make in thecommunities where the earth-quake struck?” Other than buildingcode improvements for recently-built structures, very little can besaid about how the investment oftax money improved the prepared-ness and resiliency of the commu-nities. It seems that a systemsapproach involving multiple agen-cies and professionals at all levelsfrom national to local must be de-signed and implemented. An effec-tive institutional structure forearthquake hazard mitigation is

needed to develop well-preparedcommunities. It is important, how-ever, to distinguish between a com-prehensive block diagram ofrelevant components (which iseasy to draw) and a properly func-tioning hierarchy of agencies(which makes decisions at eachlevel in a manner consistent withthe overall system objectives).

Lessons Learned andRecommendations forPossible Actions

An earthquake resilient commu-nity should have three elements atits core:

• Properly developed codes forthe physical infrastructure andhigh quality professional prac-tice in planning, design, con-struction and maintenance.

• An informed and participatingpublic.

• An institutional infrastructuresystem prepared for mitigationand response.

All three of these points requirelong-term sustained commitmentfrom both the government and itscitizens.

The 921 earthquake offers achance to learn from “real worldexperience.” Many issues related toearthquake engineering, from boththe research and practical sides,have been addressed in other re-ports co-authored by investigatorsfrom both NCREE and MCEER. Sev-eral reconnaissance teams have alsoissued technical observation re-ports (for examples, the NSF-sup-ported reconnaissance team, theEERI reconnaissance team, and oth-ers). In this section, the authors

“Within hoursof the initialmain impact,the nationalgovernmentannouncedpolicies forrelief, short-termrestoration andan organizedinteragencystructure forefficientexecution ofrescue efforts.”

116

reflect on their observations of thedamage from the 921 earthquakeand its effect on the local people,their community and governmentinfrastructure, and the level of pub-lic knowledge on issues of earth-quake hazard preparedness. Otherreconnaissance reports have paidspecial attention to earthquake en-gineering research opportunitiesand the importance of long-termprofessional practice dealing withthe physical world. The followingobservations are made from thetotal perspective involving human,institutional, and physical infra-structure system. They are offeredfor the public and the governmentin Taiwan as they face the restora-tion challenge after the 921 earth-quake.

1. If the epicenter of the 921 earth-quake had been located just 50miles either north or south ofthe Taichung/Nantou area, thedevastation to Taiwan’s economyand the quality of life wouldhave been much worse. To thenorth, the high-tech industrialpark in Hsin-Chu and the politi-cal and economic center Taipeiwould have been struck; to thesouth, the center of heavy in-dustry and manufacturingKaoHsiung could have been de-stroyed or seriously damaged.Eventually, these areas will be hitwith a major earthquake. Theopportunity exists today to carryout careful loss estimation andrisk assessment studies for theseareas. Ground motion informa-tion, geotechnical and structuraldesign information all exist insufficient quantities to conductcredible analyses of possibleearthquake scenarios. These re-sults could have a significant im-pact on the general public,

elected officials and other deci-sion-makers and stakeholders inTaiwan. Many individuals andorganizations have gained finan-cially from the booming real es-tate market of the past severaldecades – these groups willsurely be supportive of such stud-ies while the 921 earthquake isfresh in the population’s collec-tive memory. This type of studywould allow for some quantifi-cation of the vulnerability ofcritical regions in Taiwan andcould serve as the focus for asustained effort in public educa-tion.

Many individuals directly im-pacted by the earthquake re-quire psychological help, forwhich the government hasimplemented a program. Often,survivors of critical events (drunkdriving car crash, recovery froma terminal disease, and other trau-matic events) become the bestcrusaders. These individuals maybe provided with adequate un-derstanding of the issues involv-ing earthquake preparedness sothat they can contribute to thepublic education task.

2. There is an immediate need forreliable methods to evaluate theextent of damage to a structureso that proper decisions can bemade with regard to its retrofit/repair/replacement. More than7,000 damaged buildings re-main standing in and around theepicenter and they all needcritical assessment of the dam-age sustained. This is a signifi-cant opportunity to beginaccumulating the knowledgeabout “building damage” by de-veloping an “expert system” orstandardized system of measures

“A well-educatedpublic willaffect improve-ments in policyregardingmitigation andpreparednessfor earthquakeand othernaturalhazards.”


for non-destructive buildingevaluation. Of high priority isthe evaluation of essential in-frastructure buildings such ascommand centers, hospitals,manufacturing complexes andcritical lifeline systems such aswater, electrical power net-works and bridges. An additionalresearch effort to explore ad-vanced technologies for deploy-ment and implementation ofemergency response, communi-cation and rescue is also appro-priate at this point, based on thelessons learned. There are manyother long-term research oppor-tunities in earthquake engineer-ing that will not be addressed inthis article.

3. The current emergency manage-ment and restoration organiza-tional structure should bereplaced gradually by a long-term institutional infrastructurewhich involves agencies at allgovernment levels concerningall types of hazards. But beyonda simple box diagram of the hi-erarchy, such a system requiresthoughtful implementation.One very important element isthe appointment of the properindividuals at key positions inthe various agencies (an Emer-gency Response Corps - theERC). In an emergency situation,these individuals of the ERC arethe connecting nodes of the sys-tem of agencies. They must bewell versed in and loyal to theoverall strategic and tactical aimsof the system because they maybe called upon to make deci-sions on short notice withoutthe ability to consult either theirsuperiors or their subordinates.These individuals would need to

meet regularly, say twice a year,to review the emergency oper-ating plan that should exist, andto update their coordinatedefforts in mitigation and emer-gency preparedness. An institu-tional infrastructure for multiplehazard mitigation and responsemay be organized differentlyconsistent with a country’s ownsystem and culture. The systemin the United States (Congres-sional hearings and actions, theNEHRP agencies, the leadagency FEMA and its regionaloffice, etc., and how they func-tion) can be used as a startingpoint for development.

In general, a functioning institu-tional infrastructure system ismuch more difficult to establishthan to reconstruct the physicalinfrastructure system. The lattermay be targeted for completionin three or five years if resourcesare available. Emergency re-sponse and short-term actionsrequire a top-down approach.But for long-term re-es-tablishment, the top-down approach mustbe coupled with thebottom-up efforts ofthe participation of awell-educated public.In the opinion of theauthors, the centralgovernment respondedswiftly and in an orga-nized manner with rea-sonable policies to helpthe affected people.However, preparednessat the local and regionallevels was very inad-equate, due to the lackof emergency aware-ness and relief plans. A


� Figure 4. Reliable methods to evaluate theextent of damage are needed

“A functioninginstitutionalinfrastructuresystem is muchmore difficultto establishthan toreconstruct thephysicalinfrastructuresystem.”

118

workable system to empowerlocal government and a strategicsystem that ties together localand national government agen-cies during times of crisis isneeded to avoid potential bottle-necks that could impede theimplementation of Taiwan’s five-year reconstruction plan. Ad-dressing these issues is one ofthe major challenges ahead forpost-921 revitalization.

4. One of the most pressing issuesin short-term restoration is thatof construction quality. This hasalways been an ill-defined factorthat makes the evaluation of ex-isting damaged facilities moredifficult. It also becomes a fac-tor of importance in the timeimmediately following an earth-quake, as reconstruction begins.Other issues such as public edu-cation, research, institutional ef-fectiveness and building codeimprovements are longer-termefforts. However, working withthe real estate and constructionindustry can and should beginimmediately with the restorationefforts. A workshop might beorganized to review the current

practice in building inspections(see item 2 above) and construc-tion monitoring. Additionalguidelines or recommendationswould be issued as necessary.Building inspection and con-struction quality assuranceshould be examined from theoverall perspective of planning,design, construction, decision-making processes (in the case ofpublic works) and cost.

5. The short-term and long-term re-search needs identified by the921 earthquake indicate thatmost of the research programsof MCEER and NCREE, particu-larly those involving current andpotential joint MCEER-NCREEresearch efforts in: (a) loss esti-mation and risk assessment, (b)developing evaluation and retro-fit strategies for critical facilities(water and electric power net-works, medical facilities andbridges) and (c) application ofadvanced technologies in struc-tural response mitigation andemergency responses, can ben-efit from the real world experi-ence of the 921 earthquake. Atthe same time, these efforts canmake a contribution to the state-of-the-art of earthquake engi-neering practice both in Taiwanand U.S. We look forward to asuccess story resulting from thiscenter-to-center cooperation en-hanced by the 921 earthquake.

Opportunities forCollaborativeResearch Projects

As noted in the introductory sec-tion, MCEER and NCREE have hada long-term cooperative agreementsince 1992. Over the past several


� Figure 5. Failure of non-ductile details was frequently observed


years, many workshops and discus-sions have been carried out to iden-tify joint research and informationexchange between the two Cen-ters. At the time of the 921 earth-quake, some of these researchactivities were already underway.

A special workshop was heldshortly after the 921 earthquakeduring the first week of October1999 in Taipei, organized by the di-rectors of the two Centers. In viewof the observations made followingthe devastating earthquake, work-shop participants reexamined cur-rent research efforts, and prioritywas given to those issues that mayanswer questions related toTaiwan’s short-term restoration ef-forts. These projects include:

1. Seismic retrofit (retrofit strategy)of transformer-bushing systems:system evaluation and analysis,develop retrofit options.

2. Seismic response, system iden-tification, non-destructive evalu-ation and response modificationtechnologies for hospitals, manu-facturing facilities, and other fa-cilities; bridges; and lifelinesystems (electric power).

3. GIS-integrated technologies (in-cluding optical and RADAR-based remote sensing) fordamage assessment from a sys-tem viewpoint.

4. 3-D time domain characteriza-tion of ground acceleration at apoint and experimental valida-tion of Tong-Lee kinematic for-mulation.

More recently, a Center-to-Centercooperative research program wasinitiated, including MCEER, NCREEand two additional Centers, thePacific Earthquake Engineering Re-search Center (PEER) at the Univer-sity of California at Berkeley in the

U.S., and the Office of National Sci-ence and Technology Hazard Miti-gation in Taiwan. The proposedthree-year effort aims to take advan-tage of areas of overlapping inter-est and strengths of the fourCenters, and to capitalize on theextensive results collected by theresearch team in Taiwan’s recon-naissance efforts. Two major areasof emphasis were identified:

• The analysis of new informationto enhance model validation,and to develop a better under-standing that will lead to a new,more accurate knowledge base.

• Code improvements and imple-mentation specific to Taiwandeveloped by earthquake hazardmitigation professionals.

These types of efforts are alreadyunderway and funded through theresearch programs of the four Cen-ters. The proposed focus of theCenter-to-Center research programis as follows:

1. Ground motion attenuation, siteeffects, spatial variation and vali-dation.


� Figure 6. Damage to the Veteran’s Hospital in Puli reduced its capacity to about50% at a time when demand was highest.

120

2. Development of retrofit strate-gies for buildings shown to bevulnerable by the Chi-Chi earth-quake. This includes two parts:development of specific retrofitideas for 1-3 story and 8-12 storybuildings; and development ofevaluation and retrofit strategiesfor hospitals and selected manu-facturing facilities including con-tents. This effort is an extensionof existing projects on structuralcontrol technologies.

Lee, G.C. and Loh, C-H., editors, 2000, The Chi-Chi Taiwan Earthquake of September21, 1999: Reconnaissance Report, MCEER-00-0003, in press.

References

3. Development of evaluation andretrofit strategies for electricpower (extension of existingproject) water systems; and sys-tem analysis of electric powerand water systems (extension ofexisting project).

4. HAZUS and HAZ-Taiwan Pro-gram (with Chi-Chi earthquakedata, develop new system-re-lated loss estimation methods forHAZ-Taiwan).

5. Social and economic issues.

121

Research Objectives

The vision of the MCEER education program is to educate potentialusers through formal and nontraditional programs. Users include butare not limited to researchers, students, practicing professionals, publicofficials, organizational decision-makers and other beneficiaries of newdiscoveries and applications in earthquake studies. The objective is toincrease awareness, improve safety and advance earthquake loss reduc-tion activities in government, private industry, and the public at large.

In 1986, the National Science Foundation established the National Cen-ter for Earthquake Engineering Research (NCEER) to carry out systems

integrated studies in earthquake hazard mitigation that would yield resultsthat could not be accomplished by using the individual investigators ap-proach. The success of NCEER for 10 years has resulted in an expansion ofthe center approach in earthquake engineering research. In 1997, NSFawarded three earthquake engineering research centers. NCEER contin-ued its efforts with a name change to MCEER in 1998.

One of the most important efforts in pursuing the “center approach” isthe organized earthquake engineering educational effort, which would bedifficult to carry out with individual investigators. This paper not only sum-marizes the NCEER/MCEER programs of the past 14 years, but also exam-ines formal degree programs and other types of nontraditional efforts. Inaccordance with the nature of the effort, both successful and challengingefforts will be highlighted. Special emphasis will also be given to orga-nized efforts that require a center approach for development and imple-mentation. Among the activities to be highlighted in this paper are:

• K-12 efforts• Undergraduate participation• Master of Engineering program in earthquake engineering• Professional and public education

Effective transfer of knowledge is influenced by many external variableswhich affect the eventual application of research findings. Experts in this

Education and Educational Outreach:Using the Center Approach forEffective Knowledge Transfer

by Andrea S. Dargush and George C. Lee,Multidisciplinary Center for Earthquake Engineering Research


State of New YorkFederal Emergency

Management AgencyNew York State Library

Association

George C. Lee, Director,Multidisciplinary Centerfor Earthquake EngineeringResearch and Samuel P.Capen Professor ofEngineering, University atBuffalo, State University ofNew York

Andrea S. Dargush, AssistantDirector for Education andResearch Administration,Multidisciplinary Centerfor EarthquakeEngineering Research

Ernest Sternberg, AssociateProfessor, Department ofArchitecture and UrbanPlanning, University atBuffalo, State University ofNew York

B.F. Spencer, Jr., Professorand Director of GraduateStudies, Department ofCivil Engineering andGeological Sciences,University of Notre Dame

122

area of communication agree thatone of the most effective vehiclesfor knowledge utilization is throughsocial interaction (Lagorio et al.,1991), involving researchers andusers from many different disci-plines to maximize exchange of in-formation on respective needs,potential applications and their limi-tations. The principal assumptionsare that the greater the systematicorganization and coordination be-tween information user and re-searcher, the more likely theknowledge will be used. Further,the greater the number and varietyof end user communities, the morelikely the user is to be innovativeand to use new ideas.

Because MCEER serves a widerange of end users with varyingneeds and educational and profes-sional backgrounds, it is imperativethat the educational efforts of theCenter build on the collectivestrengths of its researchers, pack-aging knowledge in various waysto optimally educate these indiv-iduals. MCEER’s systems approachto research and education thusprovides an ideal platform to de-velop and carry out its educationand educational outreach activities,which are highlighted in Table 1.Because of the importance ofhaving a center-wide platform, afive-member committee on ed-ucation-research interface was

established and chaired by theCenter’s director. The five membersare Research Committee membersfrom the Center’s core institutions:University at Buffalo, Cornell Uni-versity, University of Delaware, Uni-versity of Southern California andUniversity of Nevada/Reno.

All educational activities are over-seen by this committee and are co-ordinated by the Assistant Directorfor Education and Research Admin-istration. Representatives fromthese institutions also play an advi-sory role in identifying future ac-tivities. As efforts to electronicallylink the affiliated institutions to-gether continue, more educationalactivities common to the core uni-versities are expected.

Achievements and NewUndertakings

K-12 Education

The Center began its K-12 activi-ties in 1988, with a series of work-shops to raise the awareness ofteachers and administrators aboutearthquake hazard and the riskposed to a school population by adamaging earthquake. Basic con-cepts of earth science and engineer-ing were presented, accompaniedby necessary emergency responseand social counseling procedures.

MCEER works with variousorganizations to enhance theimpact of its educationalactivities. These partnersinclude:

• Earthquake InformationProviders Group (EqIP)

• Federal EmergencyManagement Agency(FEMA)

• IRIS (Institutions inResearch in Seismology)

• Mid-America EarthquakeCenter (MAE)

• Pacific EarthquakeEngineering Center (PEER)

The end users of MCEER education and educational outreachactivities span a wide range of audiences, beginning with K-12students, teachers and parents, university students, and extend-ing to government officials, public and private business execu-tives, practicing professionals, and researchers advancing thestate of existing knowledge. In summary, the public-at-largecan be considered to be MCEER’s end user community.

123Education and Educational Outreach

A highlight of these activities wasrealized in 1997, when an earth-quake education program to pro-mote awareness and preparednesswas held in Anchorage, Alaska. Inspite of its high seismic hazards, nounilateral mitigation/response planfor schools existed within the dis-trict before this initiative.

A comprehensive listing of exist-ing earthquake educational materi-als was initiated in 1989. TheBibliography of Earthquake Edu-cation Materials (Ross, 1989) hasbeen widely distributed and hasbeen revised several times. The lat-est revision of the document –which includes both paper and elec-tronic reference material citations –will include an evaluative compo-nent. A panel of educators will as-sess materials for effectiveness andgrade-appropriateness. When com-pleted, the document will also beproduced on CD-ROM to facilitate

more expedient searching and re-trieval.

Single-day seminars for teachershave also been held over the years,to expose them to basic concepts,available materials and resources,and potential instructional ideas.This effort will be expanded in 2000,through the creation of an annualsummer Teachers Institute, whichwill add to the normal seminar con-tent by exposing them to earth-quake research and thus stimulatingideas for creative educational ap-proaches. It is well documentedthat teachers, like other profession-als, benefit from continuing educa-tion and that exposure to researchenhances appreciation of the scien-tific method (Voss, 1999).

An important aspect of K-12 Edu-cational activities is the inclusionof underrepresented minorities. Acritical objective of NSF is to pro-mote interest among precollege

tcudorP/ytivitcA sucoF

ecneiduAtegraT

1 2 3 4 5

stropeRlacinhceT scipothcraeserREECM x x x

srettelsweN seitivitca,setadpuhcraeserREECM x x x x x

spohskroW egnahcxeegdelwonk-fo-etatS x x

secnerefnoC hcraeserfonoitacilppa,weiveR x x x

sgnifeirB,sranimeS tseretnilacinhcetfoseussitnerruC x x x

sesruoCtrohS dnahcraesertnaveleryllanoisseforPsnoitacilppa

x x

s;margorPretupmoCerawtfoS

ytilibarenluv,sisylanangiseDatadnoitomdnuorg,stnemssessa

ssecca

x x

ssenerawAcilbuP noitacudeksirdnadrazaH x x x x

slairetaMlanoitacudE ekauqhtraenamorfslootgnihcaeTevitcepsrep

x x x

noitamrofnIdesab-beW ,sknil,noitamrofniylemiTsecnerefer

x x x x x

1 Academia 2 Practice 3 Policy 4 Informed Lay Audience 5 General 6 Precollege 7 University Students

x

x

x

x

x

x

x

x

x

x

x x

6 7

� Table 1. MCEER’s Traditional Technology Transfer Activities

“It is welldocumented thatteachers, likeotherprofessionals,benefit fromcontinuingeducation andthat exposureto researchenhancesappreciation ofthe scientificmethod.”

124

students in Science, Mathematics,Engineering and Technology(SMET). MCEER embraces thisobjective and endeavors to encour-age this interest among individualswho do not traditionally enter thesefields. This is carried out throughoutreach to such groups as BEAM(Buffalo-area Engineering Activitiesfor Minorities), ASCE Future CitiesProgram, the Girl Scouts and BoyScouts of America, Urban League,and others.

While not a precollege project, aparticular effort to involve minori-ties was carried out under the aus-pices of MCEER’s New YorkCity-area Consortium for Earth-quake-loss Mitigation (NYCEM)project. An element of the studyrequired field validation of buildinginventory data being used in anearthquake loss estimation of NewYork City, under FEMA sponsorship.MCEER contributed funding to theactivity to support the City Univer-sity of New York (CUNY), a minor-ity-serving institution, to allowCUNY students to carry out theneeded field survey. This effort wassuccessfully carried out under theguidance of Professor GeorgeMylonakis of CUNY.

Undergraduate Activities

Undergraduates have alwaysbeen a part of the MCEER educa-tional process. Because mostMCEER researchers are also teach-ing faculty, they integrate many oftheir findings and approaches intotraditional undergraduate andgraduate studies. Students are thenexposed to state-of-the-art knowl-edge and research developmentsthrough such course offerings asstructures, mechanics, and others.

To further enhance student ex-posure to MCEER research, theCenter has entered a collaborativeenterprise with its sister Earth-quake Engineering Research Cen-ters (EERCs), the Mid-AmericaEarthquake Center (MAE) and thePacific Earthquake Engineering Re-search Center (PEER), to organizethe Tricenter Research Experi-ences for Undergraduates (REU)program. The program invites un-dergraduate students from U.S. uni-versities to engage in EERCfaculty-supervised research activi-ties. The multi-year program willconclude annually with an REU sym-posium, which will require all par-ticipating students to makepresentations on their research andwill offer lectures on issues key todeveloping engineers – ethics andcommunications.

Graduate Activities

Graduate education has tradition-ally been a prominent element ofMCEER educational activities. Fac-ulty members have supported nu-merous graduate students to assistin their MCEER-sponsored activities.Hundreds of students have matricu-lated with the help of MCEER fac-ulty involvement and advisement,and have progressed to significantpositions at universities, public agen-cies and private businesses. Manyof their technical efforts have beendocumented in the NCEER/MCEERTechnical Report Series.

Masters of Engineering Program inEarthquake Engineering

In 1998, encouraged and partiallysupported by MCEER, the Depart-ment of Civil, Structural and Envi-ronmental Engineering at theUniversity of Buffalo developed a

MCEER educationalactivities are naturalextensions of its ongoingresearch activities. Forexample:

• The work the Center hasdone in the area ofstructural control providesan engaging focal topic formany groups from K-12 topractice, because of itsinnovative approach.

• People at many levels areinterested in theapplication of pre-existingtechnologies, for example,in the area of defense, tothe solution of earthquakeengineering designproblems.

• Relating earthquakeproblems from amultidisciplinaryperspective stimulatesinterest in individualsbeyond limited technicalcontexts.

• The Center’s studies inEmergency Response willlead to an educationaleffort to assist emergencyresponders in the effectiveuse of decision supportsystems.

• Work on retrofittechnologies for improvedpipeline performance areof interest to the utilitycommunity.

• MCEER’s hospital projectwill be developingimportant guidelines onthe performance ofbuildings and criticalnonstructural componentswhich will be very useful tohospital administrators.


Master’s of Engineering Program inEarthquake Engineering. The in-tent of this specialized degree isto provide post-graduate trainingfor students wishing to improvetheir knowledge base in earthquakeengineering. It is a design and prac-tice-oriented program suitable forstudents planning to pursue a pro-fessional career in consulting, indus-try or government service. Inaddition to traditional courses instructural dynamics, plastic analysisand design, concrete and steel struc-tures and construction estimating, aseminar on social and economic as-pects of earthquake engineering ispart of the curriculum. This course,offered for the first time in Spring2000, featured speakers from manydifferent disciplines and professions,many of whom were linked to theclass remotely. More information onthis course, developed by ProfessorErnest Sternberg, as well as the de-gree program itself, may be found in

Graduate Professional Educationin Earthquake Engineering: An In-tegrated Approach in this report.

Virtual Laboratory Tools forEarthquake EngineeringExperiments

Professor B.F. Spencer carried outa separately funded MCEER educa-tion project at Notre Dame Univer-sity to develop learning tools to beused in the Masters of Engineeringcourses in structural dynamics. Asuite of Virtual Laboratory earth-quake engineering experimentswere developed using Java to pro-vide an interactive means of instruc-tion on structural performanceunder seismic conditions. Twomodules were developed, consis-tent with MCEER’s emphasis on ad-vanced technologies, to enactsimulations under varying condi-tions. The first module, “StructuralControl using Tuned Mass Dampers(TMD) and Hybrid Mass Dampers

• The incorporation ofMCEER research findingsinto undergraduate andgraduate curricularemains a constanteducational vehicle, andhas led in many instancesto rewardingundergraduateexperiences, new coursesand new degree programs.This important interfacebetween research andeducation is critical to theCenter’s success and isincreasingly encouraged.

� Figure 1. The Virtual Laboratory for Earthquake Engineering provides an interactive framework for studentsand others to gain fundamental understanding of a wide range of topics in earthquake engineering

126

(HMD),” considers a single-degree-of-freedom building modelsubjected to various historicalearthquake records. The moduleallows users to change system pa-rameters and design TMD and HMDfor structural response modifica-tion. A snapshot of this module canbe seen in Figure 1.

The second module, “Base Isola-tion,” considers a two-degree-of-freedom building model with baseisolation, which can be animated byvarious historical isolation earth-quake records. Again, users maychange system parameters and de-sign base isolators to modify struc-tural performance.

Both modules were mounted onthe Notre Dame web site, andhotlinked to MCEER’s web site toprovide wide access to the mod-ules. They may be found at the fol-lowing URL’s: http://www.nd.edu/~quake/java/animation.html,and http://www.nd.edu/~quake/java/isolation/animation.html.

Each module is accompanied bytechnical background informationfor the user, as well as suggestedexercises and references. They areintended to provide students with agreater understanding and apprecia-tion of simulation techniques. Thisis especially useful for students atinstitutions without sophisticatedexperimental facilities. These activi-ties may be expanded as MCEER’sresearch program in User Networksfor Seismic Assessment and Retro-fit of Critical Facilities becomes bet-ter developed. They may alsoprovide insight into how such ac-tivities might be designed at a largerscale, such as through the new NSFinitiative, Network for EarthquakeEngineering Simulation.

Student Leadership CouncilAs part of its charge from the NSF

EEC Division, MCEER has estab-lished its own EERC Student Lead-ership Council (SLC). BecauseMCEER is a multi-institutional orga-nization, SLC subcouncils are es-tablished at institutions withgreater numbers of MCEER stu-dents. The balance of students atremaining universities will be incor-porated into an at-large chapter, al-lowing them equal access to theSLC activities.

The SLC is designed to increaseinteraction between EERC-fundedfaculty and associated students, en-hancing student exposure to re-search and the concurrent process.It is also intended to encourage net-working with students at otherEERC–affiliate institutions and toimprove needed skills in commu-nication and other areas. Increasedinteractions between SLC studentsand members of MCEER’s industrypartnership program will providestudents and practitioners with avaluable interface.

Professional and PublicEducation

Professional andContinuing Education

Throughout its lifetime, NCEER/MCEER has had constant interac-tion with engineering professionals.It became apparent that as newtechnologies were developed foruse in seismic design and construc-tion, there was an increasing needfor professionals to become betterinformed about the technologiesand their potential for application.

In 1996, NCEER launched a formalshort course program for practicing

MCEER’sprofessionaland continuingeducationprogram focuseson informingprofessionalsabout newtechnologiesand theirpotential forapplication.


engineers, called PACE (Professionaland Continuing Education). The ini-tial three courses focused on passiveenergy dissipation systems and theiruse in the design of building for seis-mic and wind retrofit.

The course was supplement-ed by an MCEER monograph(Constantinou et al., 1998) on thesame subject, which presentedtechnical information in a fashionthat would be easily understood bythe professional non-expert. Themonograph continues to be inhigh demand and is an excellenteducational tool.

The course itself was well-re-ceived in areas where local engi-neers had a better appreciation forthe potential of the technology.This became an important lessonto learn. While MCEER researchhad great potential for knowledgetransfer, it needed to be carefullytailored to the needs, expectations,and educational background of theaudience.

The second topic to be featuredin the PACE program was a pilotcourse on the seismic retrofittingof highway bridges, which wasbased on research and a resultingmanual (FHWA, 1995) done underthe Center’s Federal Highway Ad-ministration contract. Because thisinformation was immediately rel-evant to state transportation engi-neers, it was a very well attendedevent.

In NCEER’s transition to MCEER,course delivery has been slowed,but the Center has continued itsmomentum to develop newcourses which can help practicingengineers stay abreast of new de-sign and code developments. Morespecific, corporate-driven coursesare likely to be offered as MCEER’s

industrial partnership programgrows. A major direction is the ap-plication of advanced technologiesin performance-based engineering.

Public Education ActivitiesMCEER education and outreach to

the public-at-large serves as a foun-dation to many of the program’sother activities. In communicatinginformation about seismic hazards,risk and mitigation approaches, itmust be carefully formulated. Mes-sages need to be representedclearly and credibly, presented inmany formats to appeal to differentlearning styles, and easily accessible(Nathe et al., 1999). The audiencesmay include average homeowners,public officials, concerned citizens,and students - all with a differentneed to know. To reach them,MCEER’s efforts in this area haveused many vehicles to convey in-formation - popular press, museumexhibits, public briefings and factsheets.

MCEER Web Site and InformationService

Perhaps our most dynamicmechanism for conveying timelyinformation and reference assis-tance is the MCEER web site. De-veloped and maintained by theMCEER Information Service, theweb site is a key repository ofmany of MCEER’s activities andproducts. Technical reports may bepartially viewed and ordered, andthe Center’s two newsletters aremounted there. In addition, count-less links and references, “frequently-asked-questions,” and educationalexercises make it a comprehensivesource of information (see Figure2). It is a highly visited site, withnearly 300 outside visitors eachday. It is now undergoing additional

The MCEERInformationService hostsEqNet, the website of theEarthquakeInformationProvider’sGroup, whichprovides one-stop access toearthquakeinformationresources andlinks to qualityearthquakeinformationweb sites.

128

enhancements to make it evenmore accessible - particularly tousers in the disabled community.

Beyond its management of theweb site, the Information Servicehas a national and internationalreputation for excellence in keyearthquake reference acquisitionsand in thorough reference assis-tance. The searchable Quakeline®database of earthquake-related pub-lications is the only one its kind,featuring fugitive reference cita-tions on earthquake engineeringand related subjects that might notbe otherwise catalogued orarchived.

Because of their expertise, the In-formation Service was selected bythe National Science Foundation tolead an effort to unify the informa-tion provision activities of manyother natural disaster organizations.The program, called EqIP (Earth-quake Information Providersgroup), seeks to improve access touseful earthquake information andreduce redundancies through a glo-bal web site, EqNet. EqNet func-tions as sort of a clearinghouse - agateway to the numerous other or-ganization web sites which exist.EqIP’s members include federalagencies, universities, not for profitagencies, and private research or-ganizations.

New EndeavorsAmong the new emphases MCEER

plans to pursue in the coming yearsis the increased utilization of ad-vanced communications technolo-gies for information sharing andexchange. It is our goal to be ableto link MCEER-affiliated institutionsinto one network that will allowboth researchers and students ac-cess to teaching and research go-ing on throughout the consortium.Students involved in disaster stud-ies need increased exposure toknowledge of other disciplines thataddress earthquake problems. Re-mote learning capabilities, whenused responsibly and effectively, canhelp make this type of holisticlearning experience a reality.

As a cooperative effort, MCEER isworking with the Mid-America Cen-ter and the Pacific Earthquake En-gineering Center to developgraduate school modules in earth-quake-related studies. Each centerwill build on their own respectivetechnical expertise to formulate six-to eight-week stand-alone modulesthat can then be exchanged withother institutions. As more modulesare developed it is hoped that acomplete program will be devel-oped that might serve as a templatefor a national earthquake-engineer-ing curriculum that would be madewidely available. The first modulesare to be completed in 2000.

ConclusionsIt has been satisfying to watch the

Center’s educational program be-come more robust and responsiveto the needs of information-seekers.This has only been possible due tothe collective technical strength

Multidisciplinary Centerfor EarthquakeEngineering Researchhttp://mceer.buffalo.edu

New York City Consortiumfor Earthquake-lossMitigation (NYCEM)http://nycem.org

Virtual Laboratory Toolsfor Earthquake Engin-eering Experimentshttp://www.nd.edu/~quake/java/animation.htmlhttp://www.nd.edu/~quake/java/isolation/animation.html.

EQNethttp://www.eqnet.org

Other limited accessweb sites:

• MCEER investigators andstudents can test newsoftware to evaluate thestructural sensitivity ofbuildings subjected toseismic ground motion.

• MCEER ResearchExperiences forUndergraduates Program

• MCEER-wide StudentLeadership Councilmembers

� Figure 2. Basic illustration of earthquakeground motion


Constantinou, M.C., Soong, T.T. and Dargush, G.F., 1998, Passive Energy DissipationSystems for Structural Design and Retrofit, MCEER-98-MN01, Multidisciplinary Centerfor Earthquake Engineering Research, Buffalo, New York.

Federal Highway Administration (FHWA), 1995, Seismic Retrofitting Manual forHighway Bridges, Report No. FHWA-RD-94-052, U.S. Department of Transportation,Federal Highway Administration, McLean, Virginia.

Lagorio, H.J., Olson, R.A., Scott, S. and Shefner, J., 1991, Knowledge Transfer inEarthquake Hazard Reduction: A Challenge for Practitioners and Researchers,Center for Environmental Design Research Publication No. #CEDR-01-91, Universityof California, Berkeley, CA.

Lee, George C. and Dargush, Andrea S., 1995, Engineering Concept to Application:Experiences of NCEER, Proceedings, Research Transformed into Practice:Implementation of NSF Research, Arlington, VA, pp. 550-561.

Nathe, Sarah, Gori, Paula, Greene, Marjorie, Lemersal, Elizabeth and Mileti, Dennis, 1999,Public Education for Earthquake Hazards, Natural Hazards Observer, No. 2,November.

Ross, K.E.K., 1989, Bibliography of Earthquake Education Materials, NCEER-89-0010,Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Voss, Howard G., 1999, Testimony, Hearing on K-12 Mathematics and ScienceEducation - Finding, Training and Keeping Good Teachers, U.S. House ofRepresentatives Committee on Science, Joint Session with the Committee onEducation and the Work Force, June 10.

References

contributed by its many research-ers and the systematically-inte-grated approach which MCEER/NCEER has historically used forboth research and education. Thegreat challenge we face in trying tocommunicate rather abstract con-cepts of risk and mitigation cannotbe handled effectively by a singleinvestigator.

As MCEER research continues tobear fruit and the interaction be-tween research and education in-creases, the educational outcomesfor the public will be enriched.Education programs with an im-pact are essential for MCEER toachieve its vision of disaster resil-ient communities.

131

Research Objectives

The master of engineering (M.Eng.) program at the University atBuffalo is a comprehensive educational endeavor dedicated to train-ing professionals in specialty fields (structures, geotechnical, environ-mental or construction engineering) while providing a wide view ofthe profession, including management, economics and socioeconomicaspects. The newest addition to the curriculum is training in earth-quake engineering, which is inspired by the needs of and researchwork conducted at MCEER. A required comprehensive team projectprovides the student with the opportunity to observe the integrationof the various specialties and subjects, while strengthening the in-dividual specialty focus. In this program, the variety and uniquenessof subjects related to earthquake engineering and structural dynam-ics help produce unparalleled professionals.

The earthquake engineering program at the University at Buffalo devel-oped over the last two decades from a single course presentation

to a comprehensive program offering Master of Science (M.S.), Masterof Engineering (M.Eng.), Doctor of Philosophy (Ph.D.) degrees, and mostrecently, a combined Bachelor/Master of Engineering (B.S./M.Eng.) de-gree. The focus of this education program is to prepare professionalswho can lead design and decision processes to develop earthquake-safecommunities. Moreover, the program is intended to prepare profession-als who can further develop themselves as technologies advancethroughout their careers.

The catalyst for the development of these degree programs was theNational Center for Earthquake Engineering Research from 1986-1996,and the renewed MCEER from 1997. The program began from theneed to train students for their research work and then expanded toinclude the transfer of the newest research developments. Most re-cently, due to the multidisciplinary aspect of the earthquake engineer-ing problem and new understanding of societal problems, the programhas further expanded its horizons to include an introductory seminar

Graduate Professional Educationin Earthquake Engineering:An Integrated Approach

by Andrei M. Reinhorn, Shahid Ahmad and Ernest Sternberg


University at Buffalo, StateUniversity of New York

Shahid Ahmad, Professor,Department of Civil,Structural andEnvironmentalEngineering, University atBuffalo, State University ofNew York

Andrei M. Reinhorn,Professor, Department ofCivil, Structural andEnvironmentalEngineering, University atBuffalo, State University ofNew York

Ernest Sternberg, AssociateProfessor of Planning,School of Architecture andPlanning, University atBuffalo, State University ofNew York

132

on socioeconomic aspects of di-sasters with a focus on earth-quakes, as described later in thispaper. With strong support fromMCEER and private contributions,the program has developed, underthe guidance of the third author, amultidisciplinary seminar usingteleconferencing and expertisefrom prominent specialists fromaround the country and the Uni-versity at Buffalo. Moreover, theprogram provides expertise forshort continuing education courses,seminars and consultations.

This paper describes the contentof the earthquake engineeringeducation program and empha-sizes the new multidisciplinaryseminar developed in the last year.The program may serve as a pro-totype for other programs or pro-vide a possibility for exchange ofcontent with other institutions.

Educational ProgramOverview

The educational program isbased on several core courses,taught every year, that introducethe students to the basic sourcesand effects of earthquakes onstructures and foundations. Thesecourses are taught based on theprinciples of structural dynamicsand wave propagation, i.e.: (1)structural dynamics and earth-quake engineering sequence of

two courses, (2) structural sys-tems and construction technologywith economics content, (3)geotechnical aspects and founda-tion design. In addition, a varietyof specialty courses are taughtevery other year, allowing limitedspecialization for M.S. and M.Eng.students and a more thorougheducation for the B.S./M.Eng. andPh.D. students. The specialtycourses include: (1) random vi-brations, (2) advanced topics inengineering seismology, (3) plasticanalysis and design, (4) advancedtopics in reinforced concrete,steel and timber design, (5)aseismic base isolations, (6) struc-tural control - damping systems(active and passive), (7) experi-mental methods in structural dy-namics and (8) advanced topicsin structure analysis (finite ele-ments and boundary elements).Several topics, such as base-isola-tion design, structural control andstructural experimentation, areunique to this program. Thesetopics have been developed asresult of the research of the fac-ulty and students. In fact, in mostof the classes, the students areexposed to combined analyticaland experimental methods usingshaking table (see Figure 1) test-ing, in particular in the experimen-tation course. The latest develop-ment in the program is the addi-tion of the seminar on socio-eco-nomic aspects of disasters.

The users of this educational program are full-time stu-dents and distance learning engineers, enrolled on limitedbasis. With the development of networking technology, usergroups can be extended to professionals already in prac-tice.

Dr. Serafim Arzoumanidis,Steinman BoyntonGronquist and Birdsall,New York, New York

Dr. Fazard Naeim, VoluntaryConsultant for the M. Eng.Program, John A. Martinand Associates, Los Angeles,California

133Graduate Professional Education in Earthquake Engineering

The program leading to theM.Eng. degree is more rigidlystructured and emphasizes thepractical aspects of engineeringsubjects. The current and futuresuggested curriculum is shownin Table 1. The highlights of theM.Eng. program are (1) exposingthe students to multidisciplinaryaspects of the design and con-struction process along with thesocioeconomic concerns through

the seminar highlighted be-low, and (2) the preparationof a major group project ex-posing the students to allconstruction and manage-ment issues.

The model of the groupproject is a version of theproject concept developedfor the last twenty-five yearsat Cornell University by thelate Professor Peter Gergely(former member of the Execu-tive Committee of NCEER).The project is performedjointly by groups of students,specialists in different areas(structures, construction, foun-dation, earthquake engineering,management, architectural en-gineering), working togetherto design and “deliver” a turn-

key project for a bridge or alarge building (sports complex,airport building, or other largestructures). Students are super-vised during the year by severalfaculty and during critical studiosessions by practicing consult-ants.

It should be noted that recently,a direct B.S./M.Eng. path was ap-proved by the University at Buf-falo administration, which allows

Course Designation CIE 505, Social andEconomic Aspects ofDisaster

Seminar Web Pagehttp://mceer.buffalo.edu/courses/cie505/default.html

� Figure 1. Shaking table experiment used inStructural Control and Experimentation courses

retsemeSllaF retsemeSgnirpS

deriuqeR deriuqeR

IgnEekauqhtraE&scimanyDlarutcurtS915EIC IIgnEekauqhtraE&scimanyDlarutcurtS916EIC

smetsySlarutcurtSfonoitcurtsnoCdnangiseD725EIC ranimeSgnireenignEekauqhtraEE505EIC

)stiderc3(tcejorPgnireenignE755EIC )stiderc3(tcejorPgnireenignE855EIC

:)gniwollofehtmorfowttceles(sevitcelE :)gniwollofehtmorfowttceles(sevitcelE

serutcurtSleetSecnamrofrePhgiH425EIC ngiseD&sisylanAcitsalP125EIC

serutcurtSetercnoCecnamrofrePhgiH525EIC scimanyDdnuoF&rgnEekauqhtraE435EIC

stnemelEetiniF625EIC rgnElarutcurtSnisdohteMlatnemirepxE616EIC

snoitarbiVmodnaR025EIC noitalosIesaBcimsiesA526EIC

lortnoClarutcurtS626EIC

� Table 1. Requirements for the M.Eng. degree

134

exceptional senior students to en-roll and take specialty courses intheir last undergraduate year, thuscompleting a larger sequence ofcourses than the graduate onlyprogram. The first group of stu-dents is about to graduate andtheir performance and feedbackwill be valuable to the earthquakeengineering educators.

Social and EconomicAspects of Disaster:Focus on Earthquakes- A Graduate Seminar

During the Spring 2000 semes-ter, and with MCEER support, theUniversity of Buffalo’s Departmentof Civil, Structural, and Environ-mental Engineering offered to stu-dents an innovative new graduateprofessional learning opportunity:a multidisciplinary seminar onnatural disasters, with a focus onearthquakes. The objective wasto give them a broad backgroundof the complex institutionalframeworks and diverse methodsthey would have to work with as

earthquake engineers. To achievethis objective, the seminar madeheavy use of remotely-sited lectur-ers from around the U.S. Theymade presentations to studentsand responded to student ques-tions through interactive videotechnology.

First-year enrollment was limitedto a small number of students inkeeping with the experimental na-ture of the seminar. Ten masters’students were registered: five fromthe M.Eng. program in earthquakeengineering, four from urban plan-ning, and one from architecture.In addition, the interactive videolectures were routinely publicized,bringing in several guests (facultyand graduate students) each ses-sion.

The seminar sought to combinethe advantages of teleconferenc-ing technology—which broughtleading earthquake and disasterspecialists to the classroom fromremote locations—with the valuesof on-site workshop learning ledby the instructor. Of the seminar’s14 sessions, four were conductedby the instructor, two by on-siteguest faculty from the Universityat Buffalo, and the remaining eightby guest lecturers appearingthrough interactive video.

The remotely-sited video guestsallowed students to interact di-rectly with some of the leadingfigures in multidisciplinary as-pects of earthquake engineering,including Howard Kunreuther(University of Pennsylvania) oncost-benefit analysis and KathleenTierney (University of Delaware)on organizational aspects of disas-ters. One guest was not an indi-vidual faculty member but anorganization: the New York State

� Figure 2. Nine members of the New York State Emergency ManagementOrganization gave short presentations and a real-time virtual tour of theiremergency operations center via interactive video

135Graduate Professional Education in Earthquake Engineering

Seminar Schedule & Guest Lecturers 2000

“Earthquakes as Compared to Other Disasters,” Ernest Sternberg,School of Architecture and Planning, University of Buffalo, StateUniversity of New York

“Disaster Planning,” Ernest Sternberg, School of Architecture andPlanning, University of Buffalo, State University of New York

“Earthquake Effects on Hospitals,” Chris Tokas, California Office ofStatewide Health Planning and Development

“Information Sources for Earthquake Research,” Laura Taddeo,MCEER Information Service

“Hazard Assessment through HAZUS,” Mike Tantala, Department ofCivil Engineering, Princeton University

“Risk-Cost-Benefit Analysis,” Howard Kunreuther, Wharton School,University of Pennsylvania

“Individual Behavior in Buildings During Emergencies,” Ed Steinfeld,School of Architecture and Planning, University of Buffalo, StateUniversity of New York

“Administrative and Organizational Issues in Disaster,” KathleenTierney, Disaster Research Center, University of Delaware

“Mitigation Techniques and Code Making,” Rob Olshanksy, Univer-sity of Illinois at Urbana-Champaign

“Politics and Policies for Disaster Mitigation,” NYS Emergency Man-agement Office personnel

“Information Technology Applications for Disaster Management,”Ron Eguchi, ImageCat, Inc., formerly of EQE International

Special Project Day to Prepare Final Project Report, ErnestSternberg, School of Architecture and Planning, University ofBuffalo, State University of New York

“Ethics of Disaster Planning and Engineering,” Ernest Sternberg,School of Architecture and Planning, University of Buffalo, StateUniversity of New York

Final project presentation

Emergency Management Organiza-tion. On April 2, nine members ofits staff gave short presentationson various aspects of the agency’sfunctioning and included a real-time virtual tour of its emergencyoperations center, with studentsasking questions from Buffalo asthe camera was walked aroundthe facility in Albany (see Figure2). For each of these sessions,students had to prepare throughassigned readings and requiredquestioning and participation.

Seminar Project

The seminar sought to balance,on the one hand, the interactionswith remotely-sited guests onmultiple topics with, on the otherhand, one focused on-site projectdeveloped directly with the in-structor. The project is an inven-tory of hospital evacuations in theU.S. from 1970 to 1999. Usingmultiple databases and telephoneinterviews, students prepared aunique resource: a list of hospitalevacuations, causes of evacuation,

The implementation of theseminar requiredcollaboration not just withremotely sited guests, but alsoamong divisions of theUniversity at Buffalo. Theseminar was formally offeredby the Department of Civil,Structural, and EnvironmentalEngineering, in the School ofEngineering and AppliedScience, and supported byMCEER. However, since theseminar was purposefullymeant to incorporate diverseperspectives, the coordinatinginstructor was a facultymember in the Department ofPlanning of the School ofArchitecture and Planning. Theseminar also receivedextensive technical assistancefrom the distance learningoperations of the University’sMillard Fillmore College.

136

duration, numbers evacuated, andcasualties, if any. The results willgive hospital administrators andpolicy makers evidence by whichto judge how much priority togive to internal disaster plansand to disaster mitigation in thefacility.

The focus on hospitals gave stu-dents a central theme throughoutthe seminar, a theme on whichthey could ask questions of, anddraw insights from, speakers rep-resenting many points of view.The project also gave students apreliminary introduction to meth-ods of social and economic re-search. The final report may be

issued as an MCEER publicationor submitted to an academicjournal.

Future Directions

The plans for Spring 2001 areto make the seminar simulta-neously available at several univer-sities affiliated with MCEER.Complex arrangements will haveto be worked out on incentivesfor guest instructors, students’ reg-istration for credits at home insti-tutions, fees to cooperatinginstitutions, and technical assis-tance for bridging several cam-puses.

Acknowledgements

This report was prepared by the Multidisciplinary Center for Earthquake Engineering Research througha grant from the National Science Foundation Earthquake Engineering Research Centers Program, acontract from the Federal Highway Administration, and funding from New York State and other spon-sors.

The material herein is based upon work supported in whole or in part by the National ScienceFoundation, Federal Highway Administration, New York State and other sponsors. Opinions, findings,conclusions or recommendations expressed in this publication do not necessarily reflect the views ofthese sponsors or the Research Foundation of the State University of New York.

Production Staff

Jane E. Stoyle, EditorJenna L. Tyson, Layout and DesignHector A. Velasco, Illustration

Documents

Research Progress and Accomplishments · Suha Ülgen, Managing Director, Interactive Media and Geographic Information Systems T. Uzawa, Visiting Researcher, Department of Civil Engineering,