Accomplishments Research Progress

Research Progress and Accomplishments 1997-1999

A Selection of Papers Chronicling Technical Achievements of the Multidisciplinary Center for Earthquake Engineering Research The Multidisciplinary Center for Earthquake Engineering Research

The Multidisciplinary Center for Earthquake Engineering Research (MCEER) is a national center of excellence in advanced technology applications that is dedicated to the reduction of earthquake losses nationwide. Headquartered at the University at Buffalo, State University of New York, the Center was originally established by the National Science Foundation (NSF) in 1986, as the National Center for Earthquake Engineering Research (NCEER). Comprising a consortium of researchers from numerous disciplines and institutions throughout the United States, the Center’s mission is to reduce earthquake losses through research and the application of advanced technologies that improve engineering, pre-earthquake planning and post-earthquake recovery strategies. Toward this end, the Center coordinates a nationwide program of multidisciplinary team 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), other state governments, academic institutions, foreign governments and private industry. Research Progress and Accomplishments 1997-1999

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

July 1999

Red Jacket Quadrangle, Buffalo, New York 14261 Tel: (716) 645-3391; Fax: (716) 645-3399; Email: [email protected] World Wide Web: http://mceer.buffalo.edu

Foreword

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

he research accomplishments of the Multidisciplinary Center for Earthquake Engineering Research Tare as numerous as they are varied. Since the Center was established by the National Science Foundation (NSF) in 1986, its vision has been to help establish earthquake resilient communities throughout the United States and abroad. Over the past 13 years, our research and education programs have annually supported more than 80 investigators throughout the country and the world, to work toward this 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 the short-term is to invest in two highly-focused system-integrated endeavors: • the rehabilitation of critical infrastructure facilities such as hospitals and lifelines that society will need and expect to be operational following an earthquake; and • the improvement of emergency response and crisis management capabilities to ensure efficient response and prompt recovery following earthquakes. Our research is conducted under the sponsorship of two major federal agencies, the NSF and the Federal Highway Administration (FHWA), and the state of New York. Significant support is also derived from the Federal Emergency Management Agency (FEMA), other state governments, academic institutions, foreign governments and private industry. Together these resources are used to implement our research programs, as shown below.

Earthquake Resilient Communities through Applications of Advanced Technologies

Infrastructures that Must be Earthquake Response Available/Operational following and Recovery an Earthquake

Hospitals Lifelines Urban Infrastructure

Pipelines Research Outcomes Bridges and Power Water Will Impact Retrofit Highways Network Strategies for: and Gas • Other emergency buildings • Industrial facilities

This Research Progress and Accomplishments report is intended to introduce the reader to and highlight some of MCEER’s research tasks that are currently in progress, and provide those in the earthquake engineering community with a glimpse of the foci and direction that our programs are taking. We anticipate that this information will contribute to the coordination and collaboration effort

iii in earthquake engineering research nationally and globally. The presentation is in descriptive form with preliminary observations and recommendations, and provides an indication of future efforts. The research studies represented in this report are in various stages of completion. A few papers describe efforts that have been completed and are now represented in codes, standards, and regional or national guidelines. Others describe work in progress. Each paper, whether it be on developing loss estimation techniques, construction of a benchmark model for repetitive testing, or the vulnerability analysis of the Los Angeles Department of Water and Power’s vast network, provides a snapshot of how MCEER accomplishes its multidisciplinary and team-oriented research. The authors identify the sponsors of the research, collaborative partners, related research tasks within MCEER’s various programs, and links to research and implementation efforts outside MCEER’s program. MCEER works with all members of the earthquake engineering community, including practicing engineers and other design professionals, policymakers, regulators and code officials, facility and building owners, governmental entities, and other stakeholders who have responsibility for loss reduction decision making. The end users of the presented research are also highlighted within each paper, which we hope will enhance MCEER’s partnership programs with industry, government agencies and others. This report is the first in what we anticipate will be an annual compilation of research progress and accomplishments. Future issues will include a brief overview of our strategic research plan to further enhance cooperative research efforts. The report is available in both printed and electronic form (on our web site in PDF format at http://mceer.buffalo.edu) If you would like more information on any of the studies presented herein, or on other MCEER research or educational activities, you are encouraged to contact us by telephone at (716) 645-3391, facsimile (716) 645-3399, or email at [email protected].

iv Contents

A New Application for Remotely Sensed Data: Construction of Building 1 Inventories Using Synthetic Aperture Radar Technology Ronald T. Eguchi, EQE International, Inc.; Bijan Houshmand, MCEER Consultant; Charles K. Huyck, EQE International, Inc; Masanobu Shinozuka, University of Southen California; and David M. Tralli, MCEER Consultant

Improving Earthquake Loss Estimation: Review, Assessment and Extension of 13 Loss Estimation Methodologies Kathleen J. Tierney, University of Delaware; Stephanie E. Chang and Ronald T. Eguchi, EQE International, Inc.; Adam Rose, Pennsylvania State University; and Masanobu Shinozuka, University of Southern California

Benchmark Models for Experimental Calibration of Seismic Fragility of Buildings 29 Andrei M. Reinhorn, Michael C. Constantinou and Dyah Kusumastuti, University at Buffalo, State University of New York

Development of a Semi-Active Structural Control System 35 George C. Lee, Zhong Liang and Mai Tong, University at Buffalo, State University of New York

GIS Characterization of the Los Angeles Water Supply, Earthquake 45 Effects and Pipeline Damage Thomas D. O’Rourke, Selcuk Toprak and Sang-Soo Jeon, Cornell University

Axial Behavior Characteristics of Pipe Joints Under Static Loading 55 Emmanuel Maragakis, Raj Siddharthan and Ronald Meis, University of Nevada, Reno

Seismic Performance Analysis of Electric Power Systems 61 Masanobu Shinozuka and Tsen-Chung Cheng, University of Southern California; Maria Q. Feng, University of California, Irvine; and Sheng-Taur Mau, New Jersey Institute of Technology

National Representation of Seismic Hazard and Ground Motion for Highway Facilities 71 Maurice Power and Shyh-Jeng Chiou, Geomatrix Consultants, Inc. and Ronald L. Mayes, Dynamic Isolation Systems, Inc., for the Applied Technology Council

Updating Assessment Procedures and Developing a Screening Guide for Liquefaction 81 T. Leslie Youd, Brigham Young University

v Fragility Curve Development for Assessing the Seismic Vulnerability of 91 Highway Bridges John B. Mander, University at Buffalo, State University of New York

Changes in the New AASHTO Guide Specifications for Seismic Isolation Design 101 Michael C. Constantinou,University at Buffalo, State University of New York

Seismic Retrofitting Manuals for Highway Systems 111 Ian M. Friedland, Multidisciplinary Center for Earthquake Engineering Research and Ian G. Buckle, University of Auckland

vi A New Application for Remotely Sensed Data: Construction of Building Inventories Using Synthetic Aperture Radar Technology

by Ronald T. Eguchi, EQE International, Inc.; Bijan Houshmand, MCEER Consultant; Charles K. Huyck, EQE International, Inc.; Masanobu Shinozuka, University of Southern California; and David M. Tralli, MCEER Consultant

Research Objectives The MCEER research team is attempting to use Synthetic Aperature Radar MCEER/NSF (SAR) technology in combination with digital elevation models (DEM) to create building inventories for highly urbanized areas. If successful, these techniques could revolutionize the way in which structural inventory data are compiled for large cities. One particular application that is especially relevant to the MCEER program is the development of building inventory data for loss estimation modeling. Current methods of inventory development are often expensive to apply, can result in incomplete datasets, and are generally not standardized. Because of these shortcomings, these methods are employed Mike Bullock, Intermap only periodically, thus rendering the data static during its application. Technologies Ltd. We intend to explore the use of airborne SAR data, along with other re- Stephanie E. Chang, EQE motely sensed data, to construct building inventories for loss estimation mod- International, Inc. eling. In our study, we have two specific research objectives: Masashi Matsuoka, Earth- 1. Using a combination of remote sensing technologies, discriminate with a quake Disaster Mitigation high degree of confidence the difference between the built and natural Research Center, Japan Bryan Mercer, Intermap environment. Technologies Ltd. 2. With a moderate level of reliability, quantify the important structural and Ken Rath, Intermap economic parameters associated with large-scale urban and suburban de- Technologies Ltd. velopments (building heights, floor areas, replacement values, material or Kathleen J. Tierney, Disaster structural types, and usage). Research Center, University This research program focuses on methodology development, data analysis of Delaware and fusion, and application to the Los Angeles area. The ultimate goal of the Marc Wride, Intermap Technologies Ltd. research is to improve loss estimation modeling by creating more accurate Fumio Yamazaki, Earth- building inventories. A related program area – damage detection modeling – is quake Disaster Mitigation discussed in another paper (see Improving Earthquake Loss Estimation: Re- Research Center, Japan view, Assessment and Extension of Loss Estimation Methodologies).

mages, measurements and other data obtained from the vantage point Iof high altitude aircraft and satellites have recently entered the technology arsenal of disaster specialists. Technologies such as synthetic aperture radar (or more commonly known as SAR) have been instrumental in measuring the movement of the earth’s crust due to large- and medium-sized earthquakes. Minute displacements on the order of several centimeters

1 have been measured by comparing created from data produced by an radar data taken before and after the airborne SAR system operated by event. By measuring the difference Intermap Technologies; Figure 2 is a • Loss Estimation Method- in return times of the radar signal geographic information system ologies, K.J. Tierney, to and from the target for paired im- (GIS) map layer of taxable proper- University of Delaware, ages, scientists can detect whether ties for the same area. A quick re- S. Chang, EQE International, the ground has moved either closer view of both images suggests similar A. Rose, Pennsylvania State or further away from the sensor, i.e., regional characteristics, even though University and satellite. This technique of radar in- each image was developed from an M. Shinozuka, University entirely different data source. The of Southern California terferometry has been successfully applied in measuring the large-scale top image was created using an au- regional movement of the Los An- tomated SAR processing algorithm. geles basin during the 1992 Landers Data collected from a single flight earthquake (Massonnet et al., 1993) over Los Angeles (May 1998) was and the 1995 Kobe earthquake processed in a matter of hours, re- (Ozawa et al., 1997). sulting in digital elevation models This capability to measure eleva- and reflectance or intensity data. In- tion changes can serve other use- tensities vary throughout the imaged region depending upon look angle ful purposes. SAR technology has from the sensor, the type of mate- demonstrated significant value in rial being imaged, and other factors, creating digital elevation models which may vary from day to day (DEM) for non-urban areas. Using (e.g., moisture content and condi- airborne SAR technology with dual tion of vegetation or foliage). The sensors, it is possible to create ac- bottom image (Figure 2) was created curate DEMs inexpensively and on over a period of several months. a regular basis. Presently, this is the Data from the Los Angeles County most popular application of air- Tax Assessor’s Office were used in borne SAR technology. identifying building locations, types, sizes, heights and uses. Unfortu- Technical Summary nately, only data on taxable properties were available leaving many Figures 1 and 2 show two images government buildings and facilities of the same geographical area. Vis- out of the compilation. This lack of ible in both images is the downtown information results in what appears Los Angeles area with its freeways, to be open areas or zones in Figure streets and open areas. Figure 1 rep- 2. Furthermore, freeways and other resents a SAR (intensity) image infrastructure systems, which

Building inventories are essential for all loss estimation models. Although differences may exist in exactly how to characterize building types, all loss estimation models require an estimate of number of buildings or total square footage. Users of building inventory methodologies include loss estimation modelers, government agencies, and various private-sector groups, including insurance companies, real estate organizations, and financial institutions.

2 age, type, usage, and story Downtown height. As mentioned earlier, “Without Los Angeles the most significant draw- accurate back is the expensive nature knowledge of of the data. In the example what is exposed (see Figure 2), roughly four during a large person-months were required to collect, process and earthquake, it validate data on 1.7 million may be structures in Los Angeles impossible to County. measure the Los Angeles Understanding the nature full extent of River of the built environment will the disaster.” 110 also be key in developing rapid post-earthquake dam- Data Source: Intermap Technologies, May 1998 age assessment tools. Without ■ Figure 1. SAR Intensity Image of Downtown Los accurate knowledge of what Angeles is exposed during a large event, it may be impossible to measure the full extent of the disaster. In a separate section of this report, we discuss how remotely sensed data can be used to quickly identify areas of severe damage. Using interferometric techniques, we are confident that significant damage (e.g., burnt areas, collapsed houses, and fallen bridges) can be identified by comparing pre- and post-event imagery. The paper begins with an introduction to some basic Data Source: LA Co. Tax Assessor’s Office, 1994 radar imagery principles. For ■ Figure 2. GIS Map Layer of Building Locations most readers, the idea of radar should not be new, however, its application to disaster appear in the SAR image, do not ap- management may be. For this rea- pear in Figure 2. son, we define basic terms and at- At the present time, most detailed tempt to describe their usage loss estimation models use data within the context of urban appli- from tax assessor files to create cations. The importance of describ- building inventories. Although rec- ing the surface of the earth in terms ognized as not being complete, of roughness, vegetation cover, they often represent the best avail- height profile, and level of develop- able data to develop information on ment will be key in understanding number of buildings, building size, how to separate the built from the

Constructing Building Inventories Using SAR Technology3 natural environment. Next, we de- toward the target, with each pulse For more scribe important considerations in having a duration of about 10-50 details on SAR relating SAR information to build- microseconds. Typical bandwidths and IFSAR, see ing inventory data. As the reader fall in the range of 10 to 200 MHz. will see, characterizing buildings by Higher bandwidths correspond to NASA’s Jet their footprint and height are the ini- finer resolutions of the image. Propulsion tial steps in developing usable data The term synthetic aperture ra- Laboratory’s for building inventories. dar, or SAR, refers to the technique imaging radar Although we are in the midst of used to simulate a long antenna by home page at our research program, we do vali- combining signals (echos) received http://southport. date in this paper several important by the sensor as it moves along a jpl.nasa.gov or modeling assumptions. In particu- particular flight track. In the case the Environ- lar, we show that by using remotely of SAR, both phase and amplitude mental Research sensed data and a multi-spectral information are used, in contrast to classification scheme, we can in fact conventional radar that uses only Institute of identify the footprints of small and amplitude. This allows a larger ap- Michigan at large buildings. We also demon- erture to be synthesized, thus allow- http://erim.org. strate that it is possible to quantify ing for higher resolution images. the heights of buildings by using SAR imaging is possible using both DEM data from airborne SAR sur- airborne (jet) and spaceborne (sat- veys. Both of these validations are ellite) platforms. For example, in the extremely important in construct- U.S., airborne systems are operated ing an automated scheme for build- by the National Aeronautics and ing inventory development. Finally, Space Administration (NASA) with we provide a glimpse of future re- its Air-SAR system, and by Intermap search tasks. Ultimately, we intend Technologies, through its Star-3 sys- to describe the composition of ur- tem. Satellite-based systems that of- ban areas by a set of SAR “signa- fer SAR capabilities include Radarsat, tures.” These signatures will ERS-1 and ERS-2, and J-ERS. These distinguish between different devel- systems should be distinguished opment types (residential, commer- from those that primarily offer opti- cial, industrial and mixed), various cal imaging, e.g., SPOT and Landsat. levels of density and possibly dif- The products from SAR imaging fall ferent building materials. into three basic categories: elevation data, reflectance or intensity data, and correlation information. Basic Principles about Elevation information is possible Radar, SAR and IFSAR when IFSAR data are collected. IFSAR refers to interferometric syn- Radar imaging works by sending thetic aperture radar and is usually microwave pulses toward a target performed using an airborne or and measuring the return time and spaceborne platform with dual an- intensity that is reflected back to tennae. A SAR image is formed for the sensor or antenna. Normally, each receiving antenna with each wavelengths are on the order of 1 image having an associated magni- cm to 1 m, which corresponds to a tude and phase. The difference in frequency range of about 30 GHz return times from the two antennae to 300 MHz. For an imaging radar appear as a difference in phase for system, about 1500 high-power each pixel imaged. These differences pulses per second are transmitted will vary from pixel to pixel and this

4 Surface

Flat Surface Forest Mountains Rough City Surface

Radar Image

■ Figure 3. Radar Image versus Ground Surface information eventually characterizes Figure 3 provides an illustration of the height differences on the ground. radar images based on different sur- See http://southport.jpl.nasa.gov face conditions. or http://erim.org for more details Another important concept in un- on SAR and IFSAR. The reader is also derstanding the properties of the referred to Rodriguez and Martin earth’s surface is multi- and hyper- (1992) and Gabriel and Goldstein spectral analysis. It is possible using (1988) for more information on in- a variety of different sensors (SPOT terferometry. High-Resolution Visible multi-spec- When a radar image is viewed, tral data, SPOT panchromatic imag- what is seen is a mosaic of dots of ery, Landsat-TM, and U.S. Geological varying shades and intensities. Each Survey aerial photographs) to clas- pixel represents a radar backscat- sify the land cover of an area. Each ter for that area on the ground. De- sensor offers a range of visible and in some cases, infrared bands that pending upon the roughness of the characterize the chemical composi- surface, the moisture level of that tion of the earth’s materials and veg- area, or the “look” angle from the etation (Goetz et al., 1985; Smith et sensor, the processed image may be al., 1990). Taxonomy schemes that light or dark. Light or bright areas have been verified though field sur- represent high backscatter, or areas veys often allow classification of which return a large fraction of the vegetation by type using the above radar energy back to the sensor; data. These same procedures are dark areas represent low backscat- useful in separating buildings from ter, or areas which tend to reflect surrounding vegetation. much of the energy away from the In the next subsection, we will dis- sensor. Flat areas typically result in cuss how remotely sensed data can low backscatter, while areas con- be used to: 1) separate buildings taining extensive vegetation tend to from the natural environment and be high backscatterers. Depending 2) quantify the heights of buildings. upon the orientation of buildings We will demonstrate these con- and streets relative to the sensor, cepts using data collected for the backscattering may be high or low. Los Angeles area.

Constructing Building Inventories Using SAR Technology5 Key Considerations in type is wood-frame construction, accounting for all but 0.05 percent Relating SAR of the total. However, for story Parameters to heights greater than four stories, the Building Inventory inventory is comprised of either steel, concrete or brick/concrete Information block/other concrete buildings. For There are two important consid- buildings in the 4 to 7 story height erations in translating SAR data into category, it is more likely that the building inventory information: 1) building type is brick/concrete. For outlining building footprints and 2) buildings greater than 7 stories, it characterizing building heights. is likely that the material type With these two parameters, it is would be some kind of steel con- possible to construct enough in- struction. A more careful analysis formation to quantify total square by use (residential, commercial, and footage and possibly building con- industrial) may help to refine these struction type. With total square distributions further. footage, one can estimate the re- Later stages of our research will placement cost of a building, thus focus on the classification of con- providing a key piece of informa- struction based on development tion in characterizing exposure. By patterns, building densities, build- knowing the height of a building, ing sizes, roof types and other fac- one might infer a particular build- tors that may characterize building ing material type (e.g., steel, con- construction practices around the crete, wood, etc.) or construction country. system for some specific region or The next two sections discuss area. how SAR data, along with other re- Table 1 shows the distribution of motely sensed information, can be buildings in Los Angeles County by used to define building footprints story height and material type. The and heights. Later research reports data source for this information was will summarize how measurement the Los Angeles County Tax of these two parameters can be Assessor’s Office; therefore, non-tax- translated into information and data able properties such as government required by most loss estimation buildings are not included in this methodologies. summary. The table shows that the most predominant construction Building Footprints

■ from Multi-Spectral Table 1. Number of Buildings by Height and Material Type Imagery Data fo.oN -dooW -leetS etercnoC ,kcirB latoT latoT% seirotS emarF emarF etercnoC In the last several years, there kcolB have emerged special imagery 3-0 16159,776, 764 879 5093,2 1599,137, 9.99 tools that allow classification of 7-4 06173 152 187 473 20.0 the earth’s materials based on +8 00119 162 57336 20.0 spectral information. These tools slatotbuS 12159,776, 1470, 1841, 5526,2 1097,237, 01 can distinguish roads from grassy latoT% 9638.6 070. 040. 300. 100 01 Source: Los Angeles County Tax Assessor’s Office

6 yards, roofs of buildings from sur- Figure 4 shows two images. The rounding foliage, and trees and first image, Figure 4a, is an aerial “The bushes of varying types and sizes. photograph of a residential area in percentage of We are beginning to investigate the the city of Santa Monica, California. area covered by use of these tools in separating built The second image, Figure 4b, is a building and natural objects in an urban set- map derived from the aerial data footprints can ting. The data sources for this re- using a classification scheme con- be quantified view include aerial photographs, tained within ENVI, an imaging pro- optical imagery (SPOT), and SAR cessing software, that distinguishes relatively easily intensity data. Each of these data sets roof types from other objects, e.g., using GIS has major benefits and some draw- roads, bushes, trees, etc. It is clear technology” backs. In the case of aerial photo- from the comparison that large graphs, we are limited to the three trees are easily recognized, road- visual bands, i.e., red, green and blue. ways can be distinguished from Satellite optical data include some yards and improved properties, and near-infrared bands, however, the that the outlines of buildings are present resolution of this data is on reasonably clear. The significant the order of 10 to 20 m. The same advantage of Figure 4b is that the resolution problem also applies, to information used to derive the im- some extent, to the SAR data. age can be translated into vector It is anticipated, however, that the data that are usable in GIS systems. technology will develop so that In fact, the percentage of area cov- high-resolution (1 m) optical and ered by building footprints can be multi-spectral data will be available quantified relatively easily using GIS via satellite in the next few years. technology. Several commercial satellite compa- The extraction of building foot- nies have plans to launch these prints can be complicated by a types of sensors in the year 2000. number of factors. For example, At that point, the options for defin- nearby trees often obscure the out- ing accurate building footprints will line of a building by covering part be numerous and less expensive. of the roof. Symmetrically-shaped

a. Aerial Photo b. Classification by Spectral Properties ■ Figure 4. Classification of Building Shapes based on Multi-Spectral Analysis of Aerial Photographs

Constructing Building Inventories Using SAR Technology7 areas, such as yards or asphalt ar- collected from a higher elevation, eas, that often look like the tops of displacement of the building foot- buildings may also complicate the print due to perspective is very classification of building footprints. slight. However, the current reso- Therefore, to outline building foot- lution of the SPOT data is 10 m for prints using only visual information the panchromatic images and 20 m may not be the most effective ap- for the multi-spectral data. In the proach, particularly if large areas are future, this may not be a problem to be evaluated and checked. How- with the high-resolution satellite ever, as a first approximation of data that will be available. footprint area, these data are judged to be adequate. In addition to aerial photographs, Building Height we are investigating the use of Information from SAR SPOT imagery (both panchromatic Imagery and multi-spectral) to classify the footprints of buildings. The aerial The concept of measuring the photographs are useful data in clas- heights of buildings has been dem- sifying the footprints of short build- onstrated recently by Houshmand ings. Since aerial photos are (1996) and Hepner, Houshmand, collected at a relatively low eleva- Kulikov and Bryant (1998) for the tion, tall buildings which are not di- Los Angeles area. They were able rectly under the flight path appear to show that by using IFSAR and to lean in these photos (see Figure Airborne Visible/Infrared Imaging 5a). Automatically extracting the Spectrometer (AVIRIS) hyper-spec- footprints of tall buildings using tral imagery that it was possible to aerial data leads to a horizontal dis- not only differentiate the complex placement corresponding to the urban landscape of Los Angeles but visible portion of the side of the also create three-dimensional mod- building. Because satellite data are els of very tall buildings. By using

1 2 3 N Tower 4 5 a

e Look ∆

a. Aerial Photo b. Pixel Elevations from SAR ■ Figure 5. Aerial Photo and SAR Elevation Data for Federal Building in West Los Angeles. Note that the Tower in Figure 5a is several meters higher than the rest of the building.

8 the hyper-spectral imagery to mask Intermap Technologies Ltd. located surfaces adjacent to tall structures, in Alberta, Canada and Englewood, they were able to determine the Colorado. These data were col- baseline topography around build- lected during a flight in late May of ings and the footprints of tall struc- 1998 and the general “look” or ob- tures. servation angle is seen in Figure 5b. Where our study enhances this The posting of data for this flight earlier effort is in addressing all was 2.5 m. The measurable heights types of development, including in Figure 5b are seen in quadrants densely compacted, low-rise resi- 2c and 3c. In 2c, we are picking up dential construction, and in creat- the elevation of the top of the building a methodology that can be ing. In 3c, we are measuring the applied over extremely broad areas, elevation of the tower. In 3e, and (e.g., large cities). In order to build 5c and 5d, we are picking up equip- on this earlier effort, we have ap- ment located on top of the parking plied the techniques described in structure, which are observable Hebner et al. (1998) to our demon- from the aerial photograph in Fig- stration areas. At the present time, ure 5a. Note that the elevations we are concentrating on four areas: shown in the legend have not been 1) the Wilshire corridor between normalized to the actual ground Westwood and Santa Monica, 2) the height. Northridge area which experi- What is also evident from this fig- enced damage after the 1994 earth- ure is that we are picking up eleva- quake, 3) Downtown Los Angeles, tion data for only part of the and 4) the University of Southern building, i.e., quadrants 2c and 3c. California campus. The white areas of the figure indi- Figure 5 shows an aerial view of cate parts of the image where no the Federal Building located on data have been collected. These “no Wilshire Blvd. near the 405 Freeway. data” areas include quadrants 3a, 3b The building is essentially a rectan- and 4a. Because of the observation gular-shaped building with a large or look angle of the sensor, the tower located on the southern tower in quadrant 3c is obstructing boundary. The tower is roughly 10 our view of these areas. These ar- meters higher than the rest of the eas are referred to as “shadow” building. The overall height of the zones. In addition, there are “no building (excluding the tower) is data” zones in quadrants 1d, 1e, 2d over 70 meters. Although the out- and 2e. The reason for this prob- line in Figure 5 suggests an “H” lem is a phenomenon called “lay- shape footprint for the building, over.” Layover refers to a situation only the top portion of outline where imaging is often compli- where the tower is located is the cated by multiple signals overlay- actual building. The lower part of ing onto the same pixel. This the “H” footprint surrounds possi- phenomenon is more of a problem bly a parking structure. with tall structures. In any event, In Figure 5b, we observe for the there are enough elevation obser- same outline the pixel elevations as vations in Figure 5b to quantify the determined through analysis of air- maximum height of the tower and borne SAR data. The airborne SAR the building. data for this study was provided by

Constructing Building Inventories Using SAR Technology9 We are now in the process of which represent varying degrees quantifying the heights of smaller of development and land usage. buildings, such as dwellings. In these We will “ground truth” parts of cases, because the footprints are each study area to validate our smaller, it is more difficult to pick classification scheme. Where pos- up reliable heights. This problem is sible, we will evaluate the efficacy further complicated by the fact that of different datasets in establish- there may be other nearby struc- ing accurate footprint boundaries tures (e.g., trees) that could account and building heights. for higher elevation data. Prelimi- • Creating a regional index to mea- nary indications are that using this sure building density and devel- technique of matching building foot- opment. Preliminary research print areas with maximum elevation using tax assessor’s data indicate data for residential areas leads to a that depending upon develop- success rate (i.e., gauging the true ment type (residential, commer- story height of residential buildings) cial, industrial or mixed), a small of about 60 percent. We are trying city unit (1/2 mile by 1/2 mile to improve this percentage by refin- grid cell) will have a unique building our techniques for extracting ing height profile or signature. height data from SAR images. That is, if one were to count up In the next section, we present fu- the number of buildings in vari- ture research plans for validating ous height categories and create these techniques for larger urban ar- a cumulative histogram of total eas. We hope to complete these vali- number of buildings by story dations by the end of this year and height, one would see vastly dif- then move on to post-earthquake ferent trends for residential, com- damage assessment techniques. mercial and industrial areas. If these trends could be standardized so that they represent a mea- Future Research Plans sure of the total volume of floor In the next year, we plan to pursue space in an area, they could be a number of research activities. Most used as an alternative to existing of these will concentrate on refining measures of building inventory. Depending on our success in cre- the methods described above so that • ating these regional building in- they are applicable to a wider range dices, our next step is to test of buildings and development con- whether it is possible to recog- ditions. In all cases, however, the ba- nize these “building height signa- sic modeling assumption remains the tures” using remote sensed data. same, i.e., we can define the physical If successful, we will have dem- dimensions of a building by quanti- onstrated an entirely new set of fying its footprint and maximum methods for creating building in- height. Some of the more notable ac- ventories for large regional areas. tivities include: For loss estimation studies, this • Validating our modeling assump- would be a major breakthrough tions (footprint and height deter- since it would then be possible minations) for a wider range of to economically create a building development types. We currently inventory database for any part have four major study areas of the country.

10 References

Gabriel, A.K. and Goldstein, R.M., (1988), “Repeat Pass Interferometry,” International Journal of Remote Sensing, Vol. 9, pp. 857-872. Goetz, A.F. et al., (1985), “Imaging Spectrometry for Earth Remote Sensing,” Science, Vol. 228, pp. 1147-1153. Hepner, G.F., Houshmand, B., Kulikov, I. and Bryant, N., (1998), “Investigation of the Integration of AVIRIS and IFSAR for Urban Analysis,” Photogrammetric Engineering & Remote Sensing, Vol. 64, No. 8, August, pp. 813-820. Houshmand, B., (1996), “Urban IFSAR Modeling,” Union of Radio Science Institute Symposium, Baltimore, Maryland, June. Massonnet, D., Rossi, M., Carmona, C., Adragna, F., Peltzer, G., Felgl, K. and Rabaute, T., (1993), “The Displacement Field of the Landers Earthquake Mapped by Radar Interferometry,” Nature, Vol. 364, July 8. Mercer, J.B., (1998), “Radar-Derived DEMs for Urban Areas,” ISPRS Commission IV Symposium, Stuttgart, September. Ozawa, S., Murakami, M., Fujiwara, S. and Tobita, M., (1997), “Synthetic Aperture Radar Interferogram of the 1995 Kobe Earthquake and its Geodetic Inversion,” Geophysical Research Letters, Vol. 24, No. 18, September, pp. 2327-2330. Rodriguez, E. and Martin, J.M., (1992), “Theory and Design of Interferometric Synthetic Aperture Radar,” IEEE Proceedings, Vol. 139, No. 2, April, pp.147-159. Smith, M.O. et al., (1990), “Vegetation in Deserts: A Regional Measure of Abundance from Multispectral Images,” Remote Sensing of the Environment, Vol. 31, pp. 1-26.

Constructing Building Inventories Using SAR Technology11

Improving Earthquake Loss Estimation: Review, Assessment and Extension of Loss Estimation Methodologies

by Kathleen J. Tierney, University of Delaware; Stephanie E. Chang and Ronald T. Eguchi, EQE International; Adam Rose, Pennsylvania State University; and Masanobu Shinozuka, University of Southern California

Research Objectives The overall goals of the research are to link engineering system vul- MCEER/NSF nerability analysis with the best-available techniques of economic analysis, to produce integrated models of physical and economic loss, and to improve the efficiency and effectiveness of loss estimation methods through the application of advanced and emerging technologies. This program of research focuses both on general methodological refine- ments and model improvement and on applications that will support MCEER’s power and water lifeline and hospital demonstration projects.

purred in part by the rising economic costs of natural disasters, there Shas been a dramatic increase in efforts aimed at estimating the direct and indirect losses caused by earthquakes. For example, in 1997 the journal Earthquake Spectra devoted a special issue to loss estimation. Papers appearing in that publication ranged from cost-benefit analyses of structural rehabilitation strategies (D’Ayala et al., 1997) to the development of real-time earthquake damage assessment tools (Eguchi et al., 1997). In 1998, MCEER published a monograph addressing the physical and socioeconomic impacts of earthquake-induced electrical power disruption in the central U.S. (Shinozuka, Rose, and Eguchi, 1998). More recently, the National Research Council Committee on Assessing the Costs of Natural Disasters published a report outlining a framework for loss estimation (National Research Council, 1999). The HAZUS methodology, developed by the National Institute of Building Sciences with funding from the Fed- eral Emergency Management Agency, is currently one of the best-known set of loss estimation techniques (National Institute of Building Sciences, 1997). Further advances in loss estimation research have been facilitated by new geographic information system (GIS) mapping techniques, as well as by the growing body of empirical data on the physical and economic effects of recent earthquakes. Providing better estimates of potential earthquake losses is extremely challenging, because of shortages in the kinds of empirical data that are

13 needed for more accurate estimates, Advanced Technologies our limited understanding of the mechanisms through which losses in Real-Time Damage • MCEER investigator Ronald are generated and of the risk factors Assessment Eguchi has established associated with loss, and the uncer- This phase of MCEER’s research close working relationships tainties that enter into loss calcula- focuses on improving loss estima- with the California tions at various stages. The MCEER Governor’s Office of tion for earthquake preparedness, loss estimation research team has at- Emergency Services and response, and mitigation through tempted to address these problems with the City of Los Angeles, the application of new technolo- by systematically reviewing selected which have used damage gies that collect and analyze infor- assessment and decision loss estimation methodologies, iden- mation on the built environment support systems developed tifying areas where improvements more efficiently, rapidly, and eco- by EQE in responding to the are needed, and conducting new re- nomically. The availability of accu- 1994 Northridge earth- search on earthquake losses. These quake, in emergency rate and timely information on investigations center on four inter- response training, and in post-event damage is one of the related topics: mitigation decision most critical factors influencing the making. • use of advanced technologies for effectiveness of post-disaster re- • In related work, with real-time damage and loss esti- sponse efforts. The rapid deploy- funding from the joint mation; ment of resources where they are federal-state TriNet • measurement and estimation of most needed cannot take place un- program, EQE, the direct physical and economic less a comprehensive picture of University of Delaware losses; Disaster Research Center, damage is available. The concept • identification of risk factors for and the University of of real-time damage assessment in California, Los Angeles have business losses, including both the U.S. began roughly eight years recently begun to under- physical and business interrup- ago with the introduction of the take research on the tion losses; and CUBE (Caltech-USGS-Broadcast of feasibility of implementing • estimation of indirect or induced Earthquakes) system in southern a real-time earthquake economic losses California (see CUBE, 1992). About warning system for the southern California region. This paper briefly summarizes four years ago, a similar system, REDI work that has been undertaken to (Rapid Earthquake Data Integration) • MCEER investigator Adam date in each of these four areas. was set up in northern California. Rose has been working with Colorado State University economist Harold Cochrane on Both probabilistic and scenario-based loss estimates are be- improving methods for ing used as planning tools in the pre-earthquake context, e.g., indirect loss modeling. to provide forecasts of likely physical impacts. Loss estimation • Members of the loss methodologies are also being applied to aid mitigation decision estimation group also work making through making it possible to determine the cost-effec- closely with Howard tiveness of alternative mitigation strategies. With the advent of Kunreuther, an MCEER real-time damage- and loss-estimation tools, loss estimation investigator at the Wharton methodologies also have the potential for use in guiding emer- School, who will be gency response and early recovery activities, such as search undertaking benefit/cost and rescue, the provision of emergency shelter, and decision analyses for alternative mitigation measures. making with respect to lifeline restoration. Users of loss estimation research and techniques include federal, state, and local policy makers and planners, the emergency management community, and various private-sector groups, particularly those in the financial, insurance, and real estate sectors.

14 Future developments, as part of the the total slip (magnitude and dir- CUBE/TriNet program, will include ection) across all strands, warps, real-time ground motion maps and and other areas across the fault a real-time warning system based zone. • Los Angeles Water and on early detection of earthquakes. Synthetic Aperture Radar (SAR) Power Demonstration Although they provide valuable is another promising technology Project earthquake information, the CUBE that extends the applicability of and REDI systems stop short of es- satellite-based or airborne systems • Hospital Demonstration Project timating the damaging effects of to post-earthquake analysis. When earthquakes. To fill this need, a used to compare before and after • Advanced Technologies for number of earthquake researchers radar images of earthquake im- Loss Estimation, Develop- have developed software tools pacted areas, these methods have ment of Damage Functions based on conventional loss estima- been effective in identifying re- using Remote Sensing and tion methodologies that can gener- gions of widespread ground dis- Real-time Decision Support ate loss estimates from earthquake Systems, R.T. Eguchi, EQE placement. Figure 2 presents International and D.M. magnitude data (Eguchi et al., colored images of the co-seismic Tralli, MCEER Consultant 1997). However, in the last several displacements that were observed years, a new set of technologies after the 1994 Northridge, Califor- • Cost-Benefit Studies of based on remote sensing methods nia, and 1995 Kobe, Japan, earth- Rehabilitation Using Advanced Technologies, have found their way into disaster quakes. In the Northridge image, management. One of the first ex- R. Eguchi, EQE Interna- an interferometric method known tional, P. Kleindorfer and amples of a remote sensing appli- as repeat pass interferometry H. Kunreuther, University cation to earthquake hazards was (Gabriel and Goldstein, 1988) was of Pennsylvania, provided by Dr. Robert Crippen of used to quantify the amount of M. Shinozuka, University the Jet Propulsion Laboratory (JPL) relative displacement recorded af- of Southern California and in Pasadena (Crippen, 1992; ter the Northridge earthquake. K. Tierney, University of Delaware Crippen and Blom, 1993). Using Figure 2a shows that the highest SPOT1 satellite images acquired ap- rates, approximately 60 cm of rela- proximately one month after the tive displacement, occurred in the 1992 Landers earthquake, JPL cap- northwestern part of the San tured the spatial details of terrain Fernando Valley. movements along fault breaks associated with the earthquake that were virtu- ally undetectable by any other means. These changes, seen in Figure 1, allowed displays of fault location, patterns of drag and block rotation, and pull-apart zones to be revealed. Addi- tionally, separate applications of correlation analysis (i.e., image matching) on each side of the fault pro- Image produced by Dr. Robert Crippen, Jet Propulsion Laboratory, Pasadena. vided a comprehensive and ■ Figure 1. Landers Earthquake Ground Cracks, Emerson Fault, SW of Galway Lake, Mojave quantitative estimate of Desert, California. Before Satellite Image (Left) 27 July 1991; After Satellite Image (Right) 25 July 1992. Scale: 1.28 km across fault (upper left to lower right).

Improving Earthquake Loss Estimation15 In the Kobe earthquake, a similar models that allow a validation/cali- “Synthetic interferometric image was con- bration of SAR parameters. For ex- Aperature Radar is structed using pre- and post-earth- ample, GPS measurements can another promising quake SAR images. In Figure 2b, provide site or regional validations technology that each cycle of color corresponds to that surface displacement has oc- about 12 cm change in distance be- curred. Optical images created by extends the tween the satellite and the ground aerial photographs or satellite imag- applicability of surface. Both studies used images ing (e.g., SPOT) can provide impor- satellite-based or obtained by JERS2-1. SAR also has tant verification that damage has or airborne systems to the advantage of imaging through has not occurred. In addition, new post-earthquake cloud cover and during nighttime simulation models that replicate SAR analysis.” conditions, thus making it applicable images through analytical tech- in real-time. niques can be used to help quantify MCEER investigator Ronald building damage on a local level. In Eguchi is currently exploring the ap- combination, these technologies will plication of remote sensing methods provide a powerful tool that can for real-time post-impact damage as- quickly and reliably assess damage sessment (see A New Application for on a large regional scale. With such Remotely Sensed Data: Construc- systems in place, emergency re- tion of Building Inventories Using sponders can act more decisively Synthetic Aperature Radar Technol- after a major event, reducing overall ogy in this report). While the appli- response times, saving lives, and con- cations described above provide taining property losses. useful post-earthquake data, they fail Complementing this work, which to explain fully the changes in the is designed to detect damage at a post-earthquake images. Data are limited to differential measurements of elevation and scattering3 based on repeat passes (pre- and post-event) over the subject area. In general, the observer does not know whether these changes are due to surface displacement, building damage, or a combination of these two effects. MCEER’s research focuses on differentiating between these effects by introducing other independent data and Kobe Image from Ozawa et al., 1997

■ Figure 2. Interferograms Showing Relative Ground Displacements Measured After the 1994 Northridge and 1995 Kobe Earthquakes. Each color change reflects an increase or decrease in constant ground displacement. Note that the color scales are different in each figure. Northridge Earthquake image provided by Dr. Paul Rosen (JPL) 16 more macro-level, MCEER investiga- structures is an essen- tor Masanobu Shinozuka has been tial step in developing using SAR imaging to focus more macro-level or regional closely on specific structures. In his damage models, be- research, several buildings on the cause in the absence of University of Southern California’s SAR-derived empirical main campus are being modeled data, it is necessary to using Auto-CAD and MATLAB, with compile catalogues of possible future use of ARC/INFO’s images that accurately 3D Analyst. SAR approaches will be depict a range of pos- applied in simulation and completed sible damage states. for some buildings. Computations This new approach to are currently being carried out for a loss estimation is con- ■ Figure 3. Simulated SAR image of two boxes grouping of buildings, introducing sistent with two of in a slant plane (15 cm resolution) changes that resemble earthquake MCEER’s primary pro- damage. gram goals: its em- SAR operates by shooting a bundle phasis on advanced of many thousand rays at an object. technology and its com- The rays interact with the structure mitment to multidisci- at its boundaries, with attendant re- plinary research. For flections and refractions. Rays pen- the first time, a combi- etrate the material and may bounce nation of advanced several times before exiting an ob- technologies (satellite ject. Thus, structures and their edges imaging, Global Posi- are usually readily detectable using tioning Systems, real- this technology. Simulated SAR im- time ground motion ages can be projected on a slant mapping, advanced plane, ground plane, and on a plane simulations, and geo- vertical to the slant plane. The verti- graphic information cal image projection, which is used systems) is being em- for diagnostic purposes, reveals ployed to develop a height changes most directly. Due real-time damage as- to “layover,” the fact that higher ob- sessment system. This jects appear closer in a radar image research not only rep- but further away in a photograph, resents an integrated in- buildings are skewed in a predict- terdisciplinary effort, ■ Figure 4. Simulated SAR images of two able manner. Geometrical changes, but it also serves as an buildings on the University of Southern such as tilting, overturning, or pan- example of effective California Campus caking can be observed and mea- and coordinated tech- sured through the use of SAR, as nology transfer—in this case, trans- shown in Figure 3. The height and fer of NASA technology and straightness of buildings can be de- products. duced using these measures. Taller buildings cast longer shadows, and thicker horizontal contour edges at Direct Losses the front of structures are another Recent disasters such as the indication of building height, as Northridge and Kobe earthquakes indicated in Figure 4. This type of have demonstrated the importance analytic research on individual of evaluating not only the physical

Improving Earthquake Loss Estimation17 damage that future earthquakes may developing a “benchmark” dataset of cause, but also the losses to urban empirical loss data in recent disas- and regional economies caused by ters, and developing an improved that damage, particularly those re- methodology that can be applied in sulting from damage to urban life- the LADWP project (see Seismic Per- line systems. In its Los Angeles formance Analyses of Electric Department of Water and Power Power Systems in this report). These (LADWP) demonstration project, investigations focus in particular on MCEER will be developing an inno- the direct economic loss compo- vative loss estimation methodology nent of an integrated loss estimation for urban lifeline systems that pays methodology. In this research, the particular attention to assessing dis- term direct economic loss refers to ruption to the economies of affected the disruption to economic activity areas. This methodology will build that is caused by lifeline service out- on MCEER’s previous multidisci- age at the site of production. plinary work on Memphis Light, Gas MCEER investigator Stephanie and Water (MLGW) Division’s util- Chang has conducted a systematic ity systems (Shinozuka, Rose, and review of twelve methodologies that Eguchi, 1998). Initial efforts have fo- evaluate losses related to water life- cused on reviewing recent develop- line systems and that have made in- ments in earthquake loss estimation, novations in predicting water service disruption (“out- ■ Table 1. Lifeline Loss Estimation State-of-the-Art: Innovations and Gaps(a) age”) and/or the ensuing economic impacts. This yhtrowetoN aerAlacigolodohteM spaGgniniameR group of studies ranges from snoitavonnI )b( consulting projects for indi- seitiroirpdnaslaogecivreS seitiroirpdnaslaogecivreS noitamitsEssoLfosesU vidual utility systems to na- ;decudortni SIG detargetniylsuorogir tionally applicable, fully tsocriaperdezidradnatS tsoCriapeR software-implemented dohtem/atad methodologies. Table 1 sum- lluf;noitcaretnienilefiL degamadrofsisylanawolF marizes major noteworthy egatuOdnasisylanAsmetsyS -oicoshtiwnoitargetni metsys dnamedelbairav; sisylanacimonoce innovations and remaining methodological gaps identi- yticapac/dnamedriapeR dezidradnatS noitarotseR noitarotserdezimitpo;dohtem fied in this review, according atadyticapac/dnamed gnicneuqes to various technical areas yticapacgnithgif-eriF otknilytilibaecivresretaW within a comprehensive loss ssoLyradnoceS )yranimilerp( ekauqhtraegniwolloferif estimation methodology. lacitirc;detceffasnoitalupoP The specific methodologies stcapmIlaicoS stcapmicimonoceotkniL devresseitilicaf included in the review are dnatcerid;ssoleuneverytilitU listed in the footnote to the gnireenignehtiwnoitargetnI stcapmIcimonocE cimonocelanoigertceridni sisylanasmetsys table. Innovations made in ;ssol sulumitsnoitcurtsnocer the MCEER Memphis lifeline fonoitacifitnauQ fonoitalumisolraCetnoM loss estimation methodology ytniatrecnU tuohguorhtytniatrecnu egamad ygolodohtem are identified in italics.

Notes: (a) Based on review of the following 12 methodologies: HAZUS (NIBS, 1997; Whitman et al., The review found that 1997), CUREe model (Kiremidjian et al., 1997), PIPELINE-FIX (French and Jia, 1997), NCEER among current loss esti- model (Rose et al., 1997; Chang et al., 1996; Shinozuka, 1994; Koiwa et al., 1993), IRAS (RMS, 1995), EBMUD study (G&E, 1994), VRWater (Cassaro et al., 1993), ATC-25-1 (ATC, 1992), mation methodologies, the ASCE/TCLEE methodology (Taylor, 1991), K/J/C Everett study (Ballantyne and Heubach, 1991), approach taken in the K/J/C Seattle study (Ballantyne, 1990), and Boston study (URS/Blume, 1989); (b) Italics identify innovations made in the MCEER Memphis model.

18 MCEER Memphis study incorpo- allows not only damage, but also rated numerous innovations in economic loss, to be estimated terms of modeling water outage and on a probabilistic basis, produc- associated regional economic im- ing a “seamless” loss estimation pacts. That approach thus provides model. an excellent basis for further devel- • Incorporation of the spatial and opment in the LADWP demonstra- temporal dimensions of loss tion project. However, several through improved restoration significant improvements can be made by learning from other current methodologies, in particular by in- Scenario troducing explicit system service Earthquake(s) goals/priorities, adopting current models of post-disaster system res- Pre-event Post-event toration, and evaluating social im- Mitigation Loss Reduction • redundancy • restoration pacts and implications for fire • upgrade prioritization following earthquake. Furthermore, components • restor. speed many gaps remain, even in state-of- the-art approaches. One of the most MONTE CARLO SIMULATION critical economic loss modeling Damage shortfalls that will need to be ad- • Restoration dressed in MCEER’s LADWP project component fragility • system inventory is full integration of engineering sys- • repair rate tems analysis with economic analy- t = 0 • mutual aid • sis. While the Memphis study made Outage (t) restoration curve • important contributions toward spatial • system analysis sequencing such integration, improvements are • node supply v. still needed. For example, in that demand model, lifeline outage is modeled • GIS analysis t > 0 Updated Damage probabilistically, while economic (t) loss is modeled deterministically. As a first step in further refining Economic Loss (t) direct loss estimates, MCEER is con- • direct business ducting further analyses on the interruption loss Memphis lifeline system. This ap- • indirect loss proach takes advantage of existing data while allowing time for the development of databases for the Economic Hazard LADWP demonstration project. Fragility Curves • Probablility of Figure 5 outlines the refined meth- • Probablility of events (hazard loss given odology for estimating direct losses curve) hazard that is currently under development. The enhancements incorporated into this new approach to modeling direct losses include the following: Expected Annual Loss • Integration of economic loss modeling within the Monte Carlo simulation process. This ■ Figure 5. Flowchart of Refined Methodology for Estimating Direct Losses Due to Lifeline Disruption

Improving Earthquake Loss Estimation19 modeling. This approach makes Business-Level Losses “This new it possible to explore how post- approach to loss event strategies such as spatially Research to identify and quantify estimation is prioritizing restoration using GIS the factors that predict earthquake- methods can reduce total eco- related business losses can enhance consistent with two our understanding of the processes of MCEER’s primary nomic loss. • Development of “economic fra- leading to economic loss and can program goals: its gility curves.” As with compo- also point to ways of reducing losses through appropriate mitigation, re- emphasis on nent fragility curves, which sponse, and recovery measures. advanced technology indicate the probability of ex- Existing loss estimation methodolo- and its commitment ceeding a given damage state for gies have by and large not concen- various levels of ground motion, to multidisciplinary trated on investigating firm-level an economic fragility curve research.” earthquake impacts or on isolating would indicate the probability of the most significant contributors to exceeding a given level of loss business loss. Most efforts at esti- for various earthquake magni- mating losses focus on the aggregate tudes. Results can be integrated or regional level, rather than on with probabilistic hazard infor- populations of business firms. Re- mation to derive expected an- search in this area, which is being nual loss estimates, which are conducted at the University of necessary for cost/benefit analy- Delaware’s Disaster Research Cen- sis of loss reduction measures. ter (DRC), builds upon the Center’s This approach parallels the meth- earlier work on business vulnerability and earthquake-induced business odology developed by Shinozuka losses (Tierney, 1997; Dahlhamer and Eguchi (1998). The refined and Tierney, 1998; Tierney and model will be applied to simulat- Dahlhamer, 1998a, 1998b). ing and comparing the direct loss- DRC’s most recent analyses on risk reduction benefits of various factors for business loss have fo- mitigation strategies, ranging from cused on predicting dollar losses pre-disaster system upgrading due to both physical damage and through post-disaster mutual aid business interruption using data and optimized restoration. collected from large samples of busi- Finally, in this phase of MCEER’s nesses affected by the 1989 Loma work on loss estimation, efforts are Prieta and 1994 Northridge earth- also being made to develop a quakes. These analyses use several benchmark dataset of losses from types of predictor variables: busi- historic disasters that can be used ness-level characteristics, including in the validation and calibration of business size and economic sector; future loss estimation methodolo- measures of lifeline service disrup- gies. To date, disaster loss data have tion; peak ground acceleration been fragmented, inconsistently (PGA); and the age of the structure defined, and incomplete, and the housing the business. (Data on PGA goal of MCEER’s research is to col- and building age are currently avail- lect, reconcile, and compare empiri- able only for Los Angeles and Santa cal data on the regional economic Monica; those data will be incorpo- impacts of earthquakes, focusing rated into Santa Cruz County analy- primarily on the U.S. and Japan. ses when they become available.)

20 ■ Table 2. Regression (OLS) Results for Total Dollar Losses Santa Cruz Northridge (initial) Northridge (full) Unstd. Std. Unstd. Std. Unstd. Std. Variable Coeff. SEE Coeff. Coeff. SECoeff. Coeff. S Coeff. F*)nI(seeyolpmEemit-llu .4**72 .40 .*2 .3**51 .50 .*1 .3**61 .70 1.

W*liateRroelaseloh .3*14 .31 .*1 .3**75 .61 .*1 .3**45 .51 1.

Finance, Insurance, or .021 .22 .*0 .6**26 .31 .*1 .7*75 .21 1. Real Estate S*secivre .372 .91 .*0 .2*93 .21 .*1 .313 .01 1.

L*yticirtcelEfosso .904 .81 .*0 .2**88 .81 .*2 .3**58 .81 2.

L7senohpeleTfosso .31 .51 .*0 .272 .91 .90 .21 .61 0.

L*retaWfosso .1**75 .01 .*2 12**20. .71 .*2 .3**77 .01 2.

P-noitareleccAdnuorGkae ------*15**26. .22 2.

Y-)67-0691(tliuBrae ------1.11 .31 0. Y-)6791-tsoP(tliuBrae ------2.22 .71 0.

N5722 895 17 R2 .941 .32 3. F*eulav- 1***96.6 **02.84 ***55.53

100.

Improving Earthquake Loss Estimation21 twenty-four hours of structures in the area near the epi- 40000 electricity outage, center of the Northridge earthquake,

30000 but after that time which experienced the strongest losses began to esca- shaking, tended to be of relatively 20000 late. Similar patterns recent vintage. Parts of the impact

10000 were observed for region that had a greater concentra- telephone and water tion of older buildings experienced 0 Median Total Dollar Losses No outage <24 hours 24 hours 25-48 hours >48 hours loss following the less shaking in this particular event. Duration of Electricity Outage Northridge event, al- Studies that focus on risk factors for loss at the firm level contribute ■ Figure 6. Median Total Dollar Losses by Duration though the data also to loss estimation methodologies in of Electricity Outage (Northridge) suggest that any interruption of water several ways. First, they identify vari- service tends to be costly for busi- ables that need to be taken into ac- nesses. These findings point to the count in the development of importance of mitigation and rapid aggregate regional direct and indi- restoration measures for lifeline sys- rect loss models. Second, they pro- tems as strategies for containing eco- vide insights into the relative nomic losses. importance of different factors that Returning to Table 2, additional contribute to losses, such as ground analyses were conducted with the motion and lifeline disruption. And Northridge data, incorporating data relatedly, they provide data that can on PGA as a measure of earthquake be used to better calibrate the as- shaking, as well as on the time pe- sumptions made in regional loss riod in which the buildings hous- models. More generally, they serve ing businesses were constructed, to as a bridge between modeling ef- take into account structural vulner- forts that focus on direct physical abilities. PGA emerged as the sec- impacts and those that attempt to ond strongest predictor of total estimate indirect or induced eco- dollar loss. Not surprisingly, firms nomic losses, which are discussed located in areas experiencing more in the section that follows. intense shaking reported overall greater losses. While almost non- existent at low levels, losses sub- Indirect Economic stantially increase with higher Losses ground acceleration levels. Indirect losses, defined here as the While not reaching statistical difference between total business significance, the relationship be- interruption losses that propagate tween building age and total dollar losses is interesting and somewhat through the economy and the direct counterintuitive. Firms housed in losses stemming from physical dam- buildings constructed after 1976 age caused by ground shaking, are reported much greater losses than usually measured in terms of the in- those located in building con- terruption of flows in the produc- structed between 1960 and 1976 or tion of goods or services. Such losses prior to 1960. Contrary to what are typically distinguished from in- might be expected, businesses op- direct physical damage and ensuing erating in newer structures sus- business disruption, due, for ex- tained the highest losses among all ample, to fires in the aftermath of firms in the sample. This relation- an earthquake. While property dam- ship is likely due to the fact that age is generally immediate, business

22 interruption losses can last months Another modeling approach, and even years. Whether indirect econometric estimation, ranges losses proliferate depends consider- from studies of individual sectors, ably on the speed and extent of re- such as the real estate market (Ellson covery and reconstruction efforts. et al., 1984), to the entire economy Input-output (I-O) analysis is the (Guimares et al., 1993). Economet- most widely used approach to esti- ric models have much sounder sta- mating indirect losses resulting from tistical properties than other earthquakes and other hazards. In modeling approaches. However, its most basic form, I-O is a static, since these analyses are typically linear model of all purchases and based on time series data, they of- sales between sectors of an ten represent extrapolations of past economy, based on the technical behavior and thus are not especially relations of production (Rose and adept at modeling the disjointed Miernyk, 1989). I-O models are es- nature of hazard impacts. pecially adept at calculating multi- MCEER is currently investigating plier effects (Kawashima and Kanoh, another category of approaches, 1990; Gordon, et al., 1998), and em- computable general equilibrium pirical models are widely available (CGE) models. CGE analyses employ for any county or county grouping multi-market simulation models of the U.S. through the Impact Analy- based on the simultaneous optimiz- sis for Planning (IMPLAN) System, ing behavior of individual consum- developed by FEMA and several ers and firms, subject to economic other federal government agencies account balances and resource con- (see Minnesota IMPLAN Group, straints (see Shoven and Whalley, 1998). However, in its more basic 1992). Prior to research by MCEER forms, I-O is extremely rigid, inca- investigator Adam Rose, the only pable of incorporating the resiliency applications of CGE models to haz- often observed in the aftermath of ard analysis were pedagogical over- hazard events, and lacking in behav- views or pilot applications (see, e.g., ioral content. Boisvert, 1992; Brookshire and There are several alternatives to McKee, 1992). I–O analysis. One alternative, math- CGE models can incorporate the ematical programming models of an best features of the other modeling entire economy, adds to an I-O table approaches. They are typically based an objective function to be option an I-O table of detailed produc- mized, as well as various resource tion data and a social accounting ma- constraints (Rose, 1981; Cole, 1995). trix (SAM) extension of double entry This framework, which is able to accounts of institutions such as incorporate substitution possibilities households, corporations, and trade on both the supply and demand balances. CGE models have an opti- sides, has proved to be especially mizing feature, but it is based on the useful in analyses of how to mini- interaction of individual firms and mize indirect losses (see, e.g., Rose consumers. Moreover, the major et al., 1997). However, like I-O analy- parameters of CGE models can be sis, it fails to incorporate behavioral statistically estimated or can be considerations associated with de- based on engineering studies. cision making.

Improving Earthquake Loss Estimation23 Constructing and Applying a aftermath, while the latter pertains CGE Model to the period of reconstruction. A prototype CGE model has been In modeling the effects of natu- constructed for Shelby County, Ten- ral hazards, analysts must first set nessee, in order to simulate the im- the stage by identifying key char- pacts of a New Madrid earthquake acteristics, which provide the basis on the city of Memphis. The model for specifying assumptions and is patterned after a similar construct causal relationships of the analyti- for the Susquehanna River Basin cal model. The characteristics for developed by MCEER investigator two different contexts are enumer- Adam Rose to analyze the indirect ated in Table 3. For example, char- economic impacts of flooding (Rose acteristics 1 through 3 have et al., 1998) and structured to be implications for the extent and ag- comparable to input-output and lin- gregation of capital asset variables ear programming models previously in the model. They also identify the used by Rose to estimate the indi- conduit through which impacts rect impacts of a New Madrid earth- manifest themselves, or, more prac- quake (see Rose et al., 1997). The tically, indicate which variables are model consists of 22 production affected by the event. The capital sectors, with an emphasis on those asset variables also raise an impor- that are major users of electricity life- tant distinction. The terms short lines and those that are most cru- run and long run pertain to the cial to the functioning of the standard economic distinction be- regional economy. tween the period when some in- The model is currently being puts (usually capital) are fixed and applied to the simulation of direct the period when all inputs are and indirect economic impacts from variable. In relation to hazards, the a 7.5 magnitude earthquake in the former refers to the time of the New Madrid area, using data on elec- hazard event and its immediate tricity lifeline system vulnerability

■ Table 3. Key Considerations in Modeling the Economic Impacts of Earthquakes

SnnuRtroh uRgnoL

.yletaidemmidecudersikcotslatipaC.1 .emitsekatkcotslatipacfognidliubeR.1 .latipacedam-namnidetartnecnocsissoL.2 .serusaemnoitagitimsedulcninetfognidliubeR.2 .noitcuderyticapachguorhtflestistsefinamegamaD.3 .degnarraererasnoitutitsnirotliubersiyticapacnehwspotsegamaD.3 .evisavrepsimuirbiliuqesiD.4 .dehsilbatse-ersimuirbiliuqE.4 .detalosiyllanoigeryllausuerasessoL.5 .diapertonsidiafiesolsnoigerrehtO.5 .deliatrucsinoitcudorP.6 .derutpacerebnacnoitcudorpfollaroemoS.6 .esiryamsecirpemoS.7 .slevelsuoiverpotnruterotylekilerasecirP.7 .desuerasnoitanibmoctupniweN.8 .tsisrepyamsegnahcnoitanibmoctupniemoS.8 .pagpotsrojamaerastropmI.9 .nrettapdesiverromronretsasid-erpotnruterstropmI.9 .tsoobcimonocenasedivorpsgnivasfoesU.01 .niardnurgnolasinurtrohsnisgnivasfoesU.01 .tsoobcimonocenasedivorpdiayrevocert'voG.11 .tsoobcimonocenasedivorpdianoitcurtsnocert'voG.11 .tsoobaedivorpstnemyapecnarusnI.21 .esaercnisessolfienilcedlliwsnoitpoecnarusnI.21 .gninnalplartnecemoseriuqeryamyrevoceR.31 .)desabevitnecnignidulcni(gninnalpybdeplehsinoitcurtsnoceR.31 .stsixeytniatrecnutsedoM.41 .senilcedytniatrecnU.41 .sdeenetaidemmifoesuacebcipoymsignikam-noisiceD.51 .stnevefoycneuqerfnifoesuacebcipoymtahwemosgnikam-noisiceD.51

24 and direct economic losses Conclusion from previous MCEER research (Shinozuka, Rose, and Eguchi, 1998). MCEER’s research on earthquake Additionally, data from MCEER’s ear- loss estimation builds upon collabo- lier Loss Assessment of Memphis rations among engineers and social Buildings (LAMB) study (Chang and scientists that were initiated during Eguchi, 1997) are being used as the its first ten years as the National basis for estimates of broader busi- Center for Earthquake Engineering ness interruption and ensuing direct Research. This multidisciplinary effects. Sensitivity tests will be per- approach, which is demonstrated in formed related to assumptions on the 1998 MCEER monograph on the the main parameters of input and economic impacts of lifeline disrup- import substitution, factor mobility, tion in the central U.S. (Shinozuka, recovery timing, and insurance/aid Rose and Eguchi, 1998), begins with payment levels. an understanding of seismic hazards and continues with analyses that quantify the vulnerability of engi- Future Research Activities neered systems and the ways in Two new research activities are which the physical impacts of earth- being undertaken as part of this quakes on those systems subse- component of MCEER’s research quently affect economic activity, program. The first is a meta-analysis producing losses that ripple out- of factors influencing direct and in- ward through affected regional direct losses from natural hazards. economies. Consistent with its em- Meta-analysis is a statistical tech- phasis on the use of advanced technique that summarizes and synthe- nologies in earthquake loss sizes the results of individual studies. reduction, a second major theme in The literature is being reviewed to MCEER’s groundbreaking loss esti- identify causal factors, and additional mation research centers on the ways data are being collected. Multiple in which technologies originally regression estimates will yield prime developed for other purposes—in determinants, which in turn will be this case, remote-sensing technolo- used to specify important relation- gies—can be used to assess the vul- ships in future computable general nerability of the built environment equilibrium models. and to improve the speed and qual- In addition, a CGE model for Los ity of crisis decision making. Angeles will be constructed and will Future research will focus on col- be applied to analyzing the direct lecting additional data, systematizing and indirect economic impacts of what is known about earthquake- disruptions of utility lifeline services related losses, and further refining as part of MCEER’s Los Angeles life- and calibrating loss models. The line demonstration project. This methods developed by loss estima- multidisciplinary research will also tion researchers will be applied in incorporate mitigation consider- MCEER’s demonstration projects ations so as to fit into the overall and linked with other investigations cost-benefit analysis framework that that focus on assessing the costs and MCEER investigator Howard benefits of mitigation strategies for Kunreuther is developing (see critical facilities and lifelines. Kunreuther, 1999).

Improving Earthquake Loss Estimation25 Endnotes 1 SPOT is a company that provides satellite imagery data throughout the world. SPOT stands for Satellite Pour l’Observation de la Terra.

2 JERS is the satellite system operated in Japan.

3 Scattering describes the physical process that occurs when radar signals are reflected back from the earth’s surface to the sensor. The degree of scattering depends on the surface cover, e.g., type of vegetation, and the type of development.

4 While larger firms sustained greater overall financial losses, the impacts are more devastating to smaller businesses when losses are calculated on a per-employee basis. Standardized in this manner, small businesses report greater median losses than larger ones. Small Santa Cruz firms reported median per employee losses of $1,000, as compared with their larger counterparts, whose per capita losses were $352. The figures for Northridge for small and large firms were $851 and $31, respectively.

References Applied Technology Council (ATC), (1992), A Model Methodology for Assessment of Seismic Vulnerability and Impact of Disruption of Water Supply Systems, Report No. ATC-25-1, Redwood City, California.

Ballantyne, D.B., (1990), Earthquake Loss Estimation Modeling of the Seattle Water System, Report to the U. S. Geological Survey, Federal Way, Washington, Kennedy/Jenks/Chilton.

Ballantyne, D.B. and Heubach, W.F., (1991), Earthquake Loss Estimation for the City of Everett, Washington Lifelines, Federal Way, Washington, Kennedy/Jenks/Chilton. Boisvert, R., (1992), “Indirect Losses from a Catastrophic Earthquake and the Local, Regional, and National Interest,” Indirect Economic Consequences of a Catastrophic Earthquake, National Earthquake Hazards Reduction Program, Federal Emergency Management Agency, July, pp. 204-266. Brookshire, D. and McKee, M., (1992), “Other Indirect Costs and Losses from Earthquakes: Issues and Estimation,” Indirect Consequences of a Catastrophic Earthquake, National Earthquake Hazards Reduction Program, Federal Emergency Management Agency, July, pp. 267-325.

Cassarro, M.A., Cassaro, M.J. and Ragade, R.K., (1993), Facility Management and Risk-Based Decision Support for Water Delivery Utilities Subject to Natural Hazards, Center for Hazards Research and Policy Development, University of Louisville, Louisville, Kentucky.

Chang, S. and Eguchi, R., (1997), “Estimating Losses,” Loss Assessment of Memphis Buildings, D.P. Abrams and M. Shinozuka (Eds.), Technical Report NCEER-97-0018, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Chang, S.E., Seligson, H.A. and Eguchi, R.T., (1996), Estimation of the Economic Impact of Multiple Lifeline Disruption: Memphis Light, Gas, and Water Division Case Study, Technical Report NCEER-96-0011, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York. Cole, (1995), “Lifelines and Livelihood: A Social Accounting Matrix Approach to Calamity Preparedness,” Journal of Contingencies and Crisis Management, Vol. 3, pp. 1-11. Crippen, R.E., (1992), “Measurement of Subresolution Terrain Displacements using SPOT Panchromatic Imagery,” Episodes, Vol. 15, No. 1.

26 Crippen, R.E. and Blom R.G., (1993), “Mapping of the Horizontal Strain Field of the 1992 Landers Earthquake by Imageodesy,” EOS Transactions of the American Geophysical Union, Vol. 73, No. 43, pp. 374. CUBE, (1992), CUBE Handbook – Third Draft, Caltech-UGSG- Broadcast of Earthquakes. Dahlhamer, J.M. and Tierney, K.J., (1998), “Rebounding from Disruptive Events: Business Recovery Following the Northridge Earthquake,” Sociological Spectrum, Vol. 18, pp. 121- 141. D’Ayala, D., Spence, R., Oliveira, C. and Pomonis A., (1997), “Earthquake Loss Estimation for Europe’s Historic Town Centers,” Earthquake Spectra, Vol. 13, No. 4, pp. 773-793. Eguchi, R.T., Goltz, J.D., Seligson, H.A., Blais, N.C., Heaton, T.H. and Burtugno. E., (1997), “Real-Time Loss Estimation as an Emergency Response Decision Support System: The Early Post-Earthquake Damage Assessment Tool (EPEDAT),” Earthquake Spectra, Vol. 13, No. 4, pp. 815-832. Ellson, R., Milliman, J. and Roberts, R., (1984), “Measuring the Regional Economic Effects of Earthquakes and Earthquake Prediction,” Journal of Regional Science, Vol. 24, pp. 559- 79. French, S.P. and Jia, X., (1997), “Estimating Societal Impacts of Infrastructure Damage with GIS,” URISA Journal, Vol. 9, No. 1, pp. 31-43. G&E Engineering Systems, Inc, (1994), East Bay Municipal Utility District Seismic Evaluation Program, Final Report, Oakland California. Gabriel, A.K. and Goldstein R.M., (1988), “Repeat Pass Interferometry,” International Journal of Remote Sensing, Vol. 9, pp. 857-872. Gordon, P., Richardson, H. and Davis, B., (1998), “Transport-Related Impacts of the Northridge Earthquake,” Journal of Transportation and Statistics, Vol. 1, pp. 22-36. Guimaraes, P., Hefner, F. and Woodward, D., (1993), “Wealth and Income Effects of Natural Disasters: An Econometric Analysis of Hurricane Hugo,” Review of Regional Studies, Vol. 23, pp. 97-114. Kawashima, K. and Kanoh, T., (1990), “Evaluation of Indirect Economic Effects Caused by the 1983 Nihonkai-Chubu, Japan, Earthquake,” Earthquake Spectra, Vol. 6, pp. 739-56. Kiremedjian, A.S., King. S.A., Basöz, N.I., Law, K.H., Vucetic, M., Doroudian, M., Iskandar, V., Olson, R., Eidinger, J., Koettel, K. and Horner, G., (1997), Methodologies for Evaluating the Socio-Economic Consequences of Large Earthquakes, John A. Blume Earthquake Engineering Center, Stanford University, Palo Alto, California. Koiwa, H., Tanaka, S., Shinozuka, M. and Hwang, H.H.M., (1993), “GIS Applications for a Water Delivery System Under Intact and Seismically Damaged Conditions,” Lifeline-W(I) User’s Guide, unpublished report, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Kunreuther, H., (1999), The Role of Benefit-Cost Analysis in Evaluating Protective Measures, The Wharton School, University of Pennsylvania, Philadelphia, Pennsylvania.

Minnesota IMPLAN Group, (1998), Impact Analysis for Planning System (IMPLAN), Stillwater, Minnesota.

National Institute of Building Sciences, (1997), HAZUS Technical Manual, National Institute of Building Sciences (HAZUS), Washington, D.C.

National Research Council, (1999), The Impacts of Natural Disasters: A Framework for Loss Estimation, National Academy Press, Washington, D.C. Ozawa, S., Murakami, M., Fujiwara, S. and Tobita, M., (1997), “Synthetic Aperture Radar Interferogram of the 1995 Kobe Earthquake and its Geodetic Inversion,” Geophysical Research Letters, Vol. 24, No. 18, pp. 2327-2330.

Improving Earthquake Loss Estimation27 Risk Management Solutions, Inc., (RMS), (1995), “What If the 1923 Earthquake Strikes Again?,” A Five-Prefecture Tokyo Region Scenario, Risk Management Solutions, Menlo Park, California. Rose, A., (1981), “Utility Lifelines and Economic Activity in the Context of Earthquakes,” Social and Economic Impact of Earthquakes on Utility Lifelines, J. Isenberg (Ed.), American Society of Civil Engineers, New York.

Rose, A. and Miernyk, W., (1989), “Input-Output Analysis: The First Fifty Years,” Economic Systems Research, Vol. 1, pp. 229-271. Rose, A., Carmichael, J., Oladosu, G. and Abler. D., (1998), Modeling the Economics of Natural Hazard Impacts and Policy Responses Using Computable General Equilibrium Analysis,” Department of Energy, Environmental, and Mineral Economics, Pennsylvania State University, University Park, Pennsylvania. Rose, A., Benavides, J., Chang, S., Szczesniak, P. and Lim, D., (1997), “The Regional Economic Impact of an Earthquake: Direct and Indirect Effects of Electricity Lifeline Disruptions,” Journal of Regional Science, Vol. 37, pp. 437-58. Shinozuka, M., (1994), “GIS Applications in Lifeline Earthquake Engineering,” Proceedings of the Second China-Japan-U.S. Trilateral Symposium on Lifeline Earthquake Engineering, Xi’Ian, China, April 19-23, pp. 2232-230 Shinozuka, M. and Eguchi, R., (1998), “Seismic Risk Analysis of Liquid Fuel Systems: A Conceptual and Procedural Framework for Guidelines Development,” Proceedings of the NEHRP Conference and Workshop on Research on the Northridge, California Earthquake of January 17, 1994, California Universities for Research in Earthquake Engineering, Richmond, California, pp. 789-796 .

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

Shoven, J. and Whalley, J., (1992), Applying General Equilibrium, Cambridge University Press, New York.

Taylor, C.E. (Ed.), (1991), Seismic Loss Estimates for a Hypothetical Water System, Monograph No. 2, American Society of Civil Engineers, Technical Council on Lifeline, Earthquake Engineering, New York.

Tierney, K., (1997), “Business Impacts of the Northridge Earthquake,” Journal of Contingencies and Crisis Management, Vol. 5, pp. 87-97. Tierney, K. and Dahlhamer, J.M., (1998a), “Earthquake Vulnerability and Emergency Preparedness Among Businesses.” Engineering and Socioeconomic Impacts of Earthquakes: An Analysis of Electricity Lifeline Disruptions in the New Madrid Area, M. Shinozuka, A. Rose and R. T. Eguchi (eds.), Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 53-72. Tierney, K. and Dahlhamer, J.M., (1998b), “Business Disruption, Preparedness, and Recovery: Lessons from the Northridge Earthquake,” Proceedings of the NEHRP Conference and Workshop on Research on the Northridge, California Earthquake of January 17, 1994, Universities for Research in Earthquake Engineering, Richmond, California, pp. 171-178.

URS Consultants, Inc./John A. Blume and Associates, (1989), Metropolitan Boston Area Earthquake Loss Study, URS/Blume, Report prepared for the Massachusetts Civil Defense Agency, San Francisco, California. Whitman, R.V., Anagnos, T., Kircher, C., Lagorio, H.J., Lawson, R.S. and Schneider, P., (1997), “Development of a National Earthquake Loss Estimation Methdology,” Earthquake Spectra, Vol. 13, pp. 643-662.

28 Benchmark Models for Experimental Calibration of Seismic Fragility of Buildings

by Andrei M. Reinhorn, Michael C. Constantinou and Dyah Kusumastuti, University at Buffalo, State University of New York

Research Objectives MCEER/NSF The seismic fragility of buildings is a performance measure, which is difficult to compute. Empirical evaluations require disasters, or field damage, for construction of such fragility, thus, analytical techniques were developed. However, the analytical tools need calibration for either the accuracy of prediction of response, or prediction of damage limit states in engineering terms, both components of fragility. The purpose of this research is to develop a benchmark model, which can • At this stage, collaboration be damaged for the above studies, but repairable using inexpensive with A.V. Rutenberg, means for further studies. Moreover, the model should be able to dis- Technion-Israel Institute of play damage due to irregularities, torsion, setbacks, or other types of Technology, has begun damage. The current research was dedicated to the design of the bench- regarding irregular mark model and preparations for construction. structures including experimentation and validation of analytical approaches. The research team in Israel is working The analytical project of MCEER (Tasks 1.5 and 2.5) explored several alter- with other Universities for natives to develop fragility information. The fragility is the probability that the European Community the expected response of a structure, or component, will exceed a limit (including the University of state during an expected level of ground shaking. A limit state usually Bristol, United Kingdom; represents, in the same terms as the response, a damage condition or a Universities of Bologna, limitation of usage or other special condition. This analytical project is Florence and Naples, Italy; University of Ljubljana, intended to develop a rational, simplified procedure to calculate fragility Slovenia; Technical curves based on the simplified spectral approach, and verify it with more University of Opole, Poland; rigorous alternatives. The development requires: University of Bucharest, • Calculation of the expected response after the onset of damage. Romania; and Joint Research Centre in Ispra, • Determining meaningful characteristics of “limit states” and their uncer- Italy). Summer visits for tainties, which are otherwise loosely defined by qualitative methods. common planning of • Redefine the meaningful levels of ground shaking and their representa- benchmark models is tion in fragility analysis. anticipated. Several procedures developed to evaluate seismic response and fragility use information at various degrees of detail regarding the structural model and ground motion. The level of detail of the information used determines the uncertainties involved in the resulting fragility curves. The fragility information is therefore developed along with the degrees of confidence

29 in such information. The target of the analytical studies is a simple method based on nonlinear spec- • Hospital Demonstration tral analysis using linearized inelas- Project tic spectrum of suites of ground motions and its statistical distribu- • Calibration of Fragility tion and a spectral inelastic capac- Estimation and Develop- ity model representation of ment of Fragility Informa- structure (obtained from static non- tion, E. Maragakis and R. Siddharthan, University linear analysis) in the presence of of Nevada - Reno uncertainties. The method can be further simplified to use nonlinear • Development of Fragility In- spectral analysis using code type formation, S. Billington, inelastic spectrum and its statisti- M. Grigoriu and cal distribution and simplified code A. Ingraffea, Cornell spectral inelastic capacity model University, M. Shinozuka, University of Southern representation of the structure (ob- ■ California, M.P. Singh, tained from static nonlinear analy- Figure 1. Benchmark Frame for Study of Virginia Polytechnic sis) in the presence of uncertainties. Dampers Institute and A.J. Aref, G.F. The new method of analytical the computational tools. The model Dargush and T.T. Soong, evaluation requires an experimen- is a single story structure with story University at Buffalo tal validation of the computational height of 6 feet 4 inches (approxi- tool. Moreover, the definition of • Experimental Facilities damage limit states in a structural mately) and is shown in Figure 1. Network, E. Maragakis, The horizontal span of the two iden- University of Nevada - Reno system comprising many components (beams, columns, braces, tical frames for each direction is 8 joints, connections, etc.) requires feet 4 inches. quantification based on observed The gravity load applied is given performance. by two concrete blocks with total The availability of a large shaking weight of 32 kips. The detailing of table at the University at Buffalo connections and additional devices made it possible to develop a series placed on the structure are given in of models which can be tested and the Appendix of Constantinou et retested in various damage condi- al., 1997 or at http://civil.eng. tions. One 1:2 scale model, which buffalo.edu/users_ntwk/index.htm. was prepared and tested for evalua- The drawings show the plan view tion of “toggle braces,” was also and elevations of the model, and tested without any braces to provide detailing for connections and addi- data for the analytical verification of tional devices.

All MCEER researchers dealing with the analytical evaluation of fragility will use this research. The model will allow the testing of integration of innovative devices and systems in structures to improve their behavior for retrofit or repair. Moreover, since a benchmark model will be used, developers of computational tools will be able to calibrate their software based on common experimental evidence.

30 A B C D E F Column (TYP.) 1 2 3 4 Gravity Columns (TYP.) Beam (TYP.) S3x5.7 For more S3x5.7 S3x5.7 Floor Plate (TYP.) PL 3.5”x7’x10’ information on Floor Plate (TYP.) All Columns PL 3.5”x7’x10’ S3x5.7 MCEER’s experimental/computational users network, see http://civil.eng. buffalo.edu/

Front Elevation Side Elevation users_ntwk/ Scale: 1”=4’-0” Scale: 1”=4’-0” index.htm.

φ1” A490 Bolt 3/16” Washer A B C D E F Connections of FrameType A Gravity Column (TYP.) Floor Plates (Detail B) 1 2 Section A - A Scale: 1”=8” 3 B

4 A A Floor Plates (TYP.) Detail A PL 3.5”x7’x10’ Tube PL 1/8”,φ 1-1/16”

1/4” Thick PL. φ3/8” A49D Bolts PL 5/8”x2-3/8”x10” 1/4” Thick PL. Plan B Scale: 1”=4’-0” Detail A Section B-B Scale: 1”=8” Scale: 1”-8”

■ Figure 2. An Alternative Benchmark Model Design

The main benchmark 1:3 scale • Use replaceable parts, which re- model was designed to have vari- quire minimum reinvestment; ous configurations of the lateral • Produce damaged conditions at loading system, while the floor low levels of shaking, which fit masses will remain rigid and un- the capability of the shaking damaged. The model is being pre- table; pared for construction. • Provide for accommodating new systems for retrofit or upgrading; • Provide for incorporating new Benchmark Model methodology of rocking col- The model was designed to be umns; built with three to five stories us- • Provide for instrumenting and ing mass simulation and has remov- monitoring structural character- able components, which can be istics beyond the onset of dam- replaced by advanced components age; of new materials or functions. The • Allow the model to be design of the model was based on reconfigured as an irregular the following principles: structure for studies of torsion, impact (pounding) and 3D be- • Have a series of relevant sacrifi- havior; cial structural systems which can • Design the model with detailed be “safely damaged” to collapse; • Include a secondary system to form materials and parts avail- carry gravity loads for control of able in the industry. stability;

Developing Benchmark Models for Experimental Testing31 Three alternative [Reinhorn, 1997 - based on configurations (see Krawinkler and Nasser] Figure 3) were de- Sd E  1  tailed and their con- Sd I =+−1 (R c 1) 4 m 1.5 m 2 m 1.5 m struction is expected Rc  to be completed in    SaEISd the summer of 1999. SaI =+−11␣    Figure 2 shows one R   u y  1.5 m 2 m 1.5 m alternative design of SaE W Sd E ■ the model. R= u = Figure 3. Alternative Designs for Side Frames Qg y R (type a-c clockwise from top left) The benchmark y model was designed a T b with gravity load carrying rocking c= 0 + + a T columns, which are not damaged in 1 T0 0 a seismic event (see Figure 4). The a =10 .;. b = 037 (␣ = 2 %) separation of the vertical and lateral load carrying system will be studied where the superscript E indicates as an alternative for new construc- elastic, versus I indicating inelastic,

tion or retrofit. The system can be Qy is the lateral yield resistance, W is realized in full scale without incur- the total weight and all other con- ring large costs. stants defined in the text or the The structure is designed with relation. The spectral capacity is replaceable side frames or other lat- characteristic for any structure and eral load carrying systems. The con- was obtained from the actual base nections are such that the load is shear, BS, and the top floor displace- transferred into the joint and does ment obtained from a nonlinear not affect the other components. static procedure (IDARC2D) accord- The system was designed to yield at ing to the following relations: low levels of excitation and to de- BS u velop near ultimate lateral capacity. Q**== ; u ⌫⌫2 g ␾ The model was evaluated using a 1 nonlinear time history analysis where ⌫ and ␾ are the modal char- based on IDARC2D acteristics of the dominant mode. (Valles et al, 1996) and an approximation φ7/8” Stud based on a nonlinear (φ 1” Hole) spectral capacity pro-

Half Moon cedure (Reinhorn, PL 8”x8” (R=10) 1997). The approxi- Section C - C mate evaluation con- Scale: 1”=8”

sists of using the D

composite nonlinear C C

response spectrum Half Moon Half Moon PL 8”x8” (R=10”) PL 8”x8” (R=10”) derived from the Elas- tic Design Spectra φ7/8” Stud φ7/8” Stud (DARS), S vs. , to ob- s Sd D tain the Inelastic De- sign Spectra (IDARS) Detail B Section D - D Scale: 1”=8” Scale: 1”=8”

■ Figure 4. Details of Gravity Column 32 ■ Table 1. Weight Sensitivity (Model 1 - 40 mtons, the inelastic deformation. Model 2 - 53 mtons) The sensitivity analysis shows that with the max- T )ces( R ␮ u /* H )%( Q* imum shaking table input, M11ledo 154. 237. 246. 58. 1.0 a global ductility of 3 is M12ledo 186. 25. 239. 64. 1.0 feasible. This ductility translates to larger local ■ Table 2. Sensitivity to Structural Changes for Single Bay ductilities and spread Frame plasticity effects. T )ces( R ␮ u /* H )%( Q* The benchmark model M11ledo 154. 126. 136. 36. 1.0 was designed here with M63ledo 151. 162. 182. 24. 2.0 steel frames, however, construction details were pre- The spectral characteristics Q* vs. pared to allow concrete frames to be used in a symmetric or non-sym- u* were then compared to the Sa metric way. Moreover, the model vs. Sd demand to estimate the response (Reinhorn, 1997). was designed to allow for mass A sensitivity analysis was per- eccentricity to permit testing of formed on the three types of mod- torsional effects and triaxial in- els to obtain an optimal construction teraction between individual com- and generate sufficient observable ponents. Computational models damage. Table 1 shows the influ- are currently being developed ence of changing the floor weight (Simeonov et al., 1999) which will from 40 to 53 mtons, i.e. 30%. require experimental The apparent response ductility, validation within the ■ Table 3. Sensitivity to Input Ground Motion ␮, changes by 10% when the global system. AGP R ␮ u /* H )%( Q* weight increases by 30%. This is Most importantly, an inefficient way to increase in- the benchmark model 052. 126. 136. 36. 1.0 elastic response or for other de- should be able to verify 02552. 1479. 1.0 signs to reduce the dynamic and validate the new 053. 254. 262. 47. 1.0 response. spectral capacity ap- 0553. 237. 246. 55. 71.0 The structural changes at the first proach and its adequacy 014. 33. 68. 1 1.0 floor using heavier steel shapes (S4 x 7.7 instead Model 1 (S3x5.7) Model 1 (S3x5.7) of S3 x 5.7) produce a re- (pga = 0.2 g) (pga = 0.25 g) 0.6 0.7 elastic elastic capacity 0.6 duction of 25% in the 0.5 capacity inelastic inelastic 0.5 0.4 ductility (see Table 2). Al- 0.4 0.3 Sa/g

Sa/g 0.3 though the strength 0.2 Model 1 (S3x5.7)0.2 (pga = 0.4 g) 0.1 increases substantially, 1.2 0.1 elastic 0.0 0.0 1.0 capacity 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 inelastic the deformation reduces (%) Sd/H 0.8 Sd/H (%) in smaller proportion. 0.6 Sa/g

Model 1 ( 0.4 1 (S3x5.7) The most sensitive = 0.35 g) (pga = 0 0.2 0.8 . elastic elastic parameter on the re- 0.7 0.0 capacity capacity 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 0.6 ne as c inelastic Sd/H (%) sponse and inelastic re- 0.5 0.6

0.4 0.5 Sa/g Sa/g sponse is ground motion 0.3 0.4 0.3 0.2 0.2 (see Table 3 and Figure 5) 0.1 0.1

0.0 0.0 An increase in the ground 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 Sd/H (%) Sd/H (%) motion produces almost proportional increases in ■ Figure 5. Spectral Response Evaluation of Model to Increasing Ground Motion

Developing Benchmark Models for Experimental Testing33 to new innovative protective with reuse of structural models systems, such as nonlinear dampers enabled them to design the model and active or semi-active systems. with the capability to accommo- The model details will accom- date modern protective systems, modate the addition of braces, while still producing inelastic non- removal of columns or other linear behavior. members, and so forth. The model will first be tested to obtain simple elastic and then in- Conclusion and Future elastic behavior to calibrate the analytical tools. The results will be Research made available via the web to the The benchmark model (under experimental/computational users construction) was also designed to network of MCEER (http:// reflect construction issues which civil.eng.buffalo.edu/users_ntwk/ are part of the MCEER hospital dem- index.htm - temporary address). onstration project. The model will The computational models will also be tested to determine the limit be presented for further reference. states associated with such con- The immediate future plans for struction and data interpretation this project are to construct and will enable development of fragil- instrument the model, including ity information. The issues already motion and force sensors, with mul- identified are the irregularities in tiple usage for immediate testing of construction and distribution of several steel frames. One particu- floor weights, and the influence of lar configuration, a candidate for the nonstructural and architectural first round of testing, is the multiple components. These can be simu- towers building (see Figure 3(c)). lated in the construction of the Preparations are currently being model for uncomplicated testing. made for this testing. The past experience of the authors

References Constantinou, M.C., Tsopelas, P. and Hammel, W., (1997), Testing and Modeling of an Improved Damper Configuration for Stiff Structural Systems, technical report submitted to the Center for Industrial Effectiveness and Taylor Devices, Buffalo, New York.

Reinhorn, A.M., (1997), “Inelastic Analysis Techniques in Seismic Evaluations,” Seismic Design Methodologies for the Next Generation of Codes, Krawinkler and Fajfar, eds., Proceedings of the Bled Slovenia Workshop, June 24-27, Balkema Publishers, pp. 277-287 Simeonov, V.K., Sivaselvan, M. and Reinhorn, A.M., (1999), “State Space Approach in Nonlinear Analyses of Frame Structures,” Engineering Structures, (Special Volume in honor of Professor Gear), (in review).

Valles, R.E., Reinhorn, A.M., Kunnath, S.K., Li, C. and Madan, A., (1996), IDARC2D Version 4.0: A Computer Program for the Inelastic Damage Analysis of Buildings, Technical Report NCEER-96-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

34 Development of a Semi-Active Structural Control System

by George C. Lee, Zhong Liang and Mai Tong, University at Buffalo, State University of New York

Research Objectives MCEER/NSF This project seeks to develop a cost-effective semi-active control Office of Naval Research system for use in buildings and other structures to protect them from through the Technology destructive earthquake ground motions. By applying a concept of Reinvestment Project of physical parameter modification, the semi-active system is different from the Defense Advanced active control systems in that its operation does not require a large Research Project Agency power source. Yet, the system is more effective in reducing structural Enidine Incorporated response than traditional passive protective systems. University at Buffalo, State The fundamental research to develop the concept and control University of New York principles for the semi-active control system reported in this paper National Science Foundation was funded by the National Science Foundation in 1993. Subsequently, the University at Buffalo provided seed money for the development of a small scale demonstration model that validated the principle. The current study is a continuation of the development towards commercialization funded through a Cooperative Agreement with the Office of Naval Research (ONR) as part of a DARPA (Defense Advanced Research Projects Agency), Technology Reinvestment Project (TRP) Enidine Incorporated grant, intended to develop advanced technologies for both defense and Hydro-line, Inc. civilian utilization. MCEER and the University at Buffalo are members Carderock Division, Naval of the ISMIS® Consortium, which includes Enidine Incorporated as the Surface Warfare Center lead organization and Hydro-line, Inc. The Carderock Division, Naval Naval Research Laboratory Surface Warfare Center collaborates with the ISMIS Consortium through a CRADA, Creative Research and Development Agreement. The semi- active system under development is based on the idea of real-time structural parameter modification (RSPM) of the system. The objectives are to develop and commercialize the RSPM technology (or semi-active control system), to improve the seismic performance of structures and to absorb shock in naval applications.

number of passive structural vibration reduction technologies (earth- A quake protective systems) have already been used in structural engineering practice (e.g., base isolation systems and to a lesser extent, fluid dampers). The next frontier of implementing structural control technologies in practice is expected to be the semi-active type. These devices offer the advantages of an active control system but without requiring a large external power source for operation.

35 This paper describes the recent Introduction progress of a special type of semi- active system (also known as a It has always been a major • Hospital Demonstration variable passive system) under challenge for the structural Project development at MCEER. This pro- engineering profession, both in research and practice, to design and • Rehabilitaion Strategies ject was initially funded by the National Science Foundation (NSF) construct buildings and other for Hospital Buildings and structures to resist forces of nature. Contents, A.J. Aref, through MCEER and the University M. Bruneau, M.C. at Buffalo for developing and Fifty years ago, most engineers Constantinou, G.C. Lee, proofing the concept. During the performed structural analysis and design based on principles of T.T. Soong, University at past two years, major progress has Buffalo, M.P. Singh, Virginia statics. For dynamic loading such been made through the ONR/ Polytechnic Institute, and as the forces generated by DARPA/TRP Project in partner- S. Billington and horizontal ground motions or wind ship with the Carderock Division, M. Grigoriu, Cornell gusts, a structure is designed with University Naval Surface Warfare Center, a stronger capacity (lateral stiffness) Naval Research Laboratory, Enidine increase in the direction of the Incorporated and Hydro-line, Inc. expected ground motions. This project illustrates how results Since the late 1940s, basic from a fundamental inquiry principles of structural dynamics supported by NSF was further and plasticity theories have been developed with major funding developed, and pseudo-dynamic from mission agencies and industry approaches (mostly based on sin- for commercialization purposes. gle degree-of-freedom dynamic When fully developed, these models) were introduced in devices will be used to protect earthquake engineering design. critical facilities such as hospitals Special emphasis has been given to and their contents from damage ductility requirements of structures due to earthquakes. in the lateral direction.

The current phase of study may be regarded as the development of enabling technology through a systems integrated approach. For defense applications, the product may be regarded as shock absorbers (an intelligent mechanical device) and the primary users would be designers and manufacturers of ships. For earthquake engineering, users would be planners, architects, structural engineers and contractors who are concerned with efficient and cost-effective methods to reduce the seismic responses of new and existing structures. The immediate next step of the project is to implement a full- scale set of the system in a full-scale, real world structure and to develop specific design guidelines for the technology. Architects and structural engineers who are challenged to design new structures or to retrofit existing structures to withstand seismic hazards have an increasing choice of devices to use. The supplemental energy dissipation device described in this paper offers another option to the designer.

36 In the last two decades, a new do this properly, one must consider concept in earthquake engineering the structure together with the design has been advanced: added device/system in the design performance-based engineering. At process. This may not be a simple • Current efforts have gone the same time, seismic vibration task, depending upon the type and beyond the fundamental reduction technologies have been configuration of the structure and research level and are in pursued by many researchers. This type of base isolation/energy the midst of developing “structural control technology” is dissipation systems to be installed. systems integrated part of a widespread advancement Because ground motions are enabling technologies by in intelligent mechanical and stochastic in nature and passive leveraging major funding from ONR through the material systems that are expected systems have only a limited range of Technology Reinvestment effectiveness, active control systems to improve the performance of Project (TRP) of the structures in a cost-effective fashion. are more efficient. However, except Defense Advanced Research Seismic response reduction tech- for protecting small or light weight Project Agency (DARPA). nologies are typically classified into objects, such as equipment, a major the following categories by earth- breakthrough on how to deliver • The RSPM technology will quake engineering researchers: large active counterforces is need- be applied to the hospital ed before widespread use can project as an approach to • Passive Systems occur. Active control and hybrid increase the performance - Base isolation systems control will remain at the research of the buildings and to protect the nonstructural - Tuned-mass damper systems level with respect to their - Energy dissipation systems components and application to civil engineering equipment of the hospital • Active and Hybrid Systems structures for some time. to ensure its acceptable • Semi-Active (Variable Passive) Semi-active systems include smart functionality during and Systems mechanical and material systems. after earthquakes. To date, a number of passive Through switching or on-off actions, systems have actually been the physical parameters of a implemented in buildings and other dynamic system can be modified in civil engineering structures. The real-time. Because semi-active con- most popular approach is to use a trol systems use passive forces, the base isolation system. The tuned- authors prefer to use the term mass damper approach has been variable passive control to contrast used for wind vibration reduction the active control that uses active in high-rise buildings, and various forces. The most comprehensive energy dissipation systems such as variable passive system modifies the bracing-type viscoelastic and all physical parameters simul- viscous (fluid) dampers have been taneously and is called real-time implemented in many buildings in structural parameter modification recent years. MCEER researchers (RSPM) (Lee et al., 1994). have been actively engaged in developing various passive and Input Output Input Output Input Output active control technologies since M, C, K M, C, K M, C, K 1986. An MCEER monograph Feedback summarizing passive energy dissipation systems has recently Passive System Active System RSPM been published (see Constantinou ■ Figure 1. Comparison of Passive, Active and Semi-Active (RSPM) Systems et al., 1998). Most passive seismic protective A simple conceptual comparison systems are based on the general of the three types of earthquake idea of increasing the damping of protective systems is given in Figure structures (Liang and Lee, 1991). To 1. It is useful to note that the power

Development of a Semi-Active Structural Control System37 supply needed by active F S1 F S 3 k control devices is to develop m counter forces, which con- sume the same type of K M mechanical energy as the C external forces. For the F S2 c RSPM (semi-active) system or any passive system, the power supply is not directly ■ Figure 2. Energy Removal Mechanisms by Using coupled with the mechanical Functional Switches (FS) energy of the structure. • Semi-Active Systems: Variable Passive Control System Mx” + C(t)x’ + Kx = F (1) variable damping, or The system under development is referred to as real-time structural Mx” + Cx’ + K(t)x = F (2) parameter modification (RSPM) variable stiffness technology. Although it is regarded • RSPM: as a semi-active system, it evolved [M(t) + ⌬M]x” + [C(t) + ⌬C]x’ from the traditionally defined active + [K(t) + ⌬K]x = F (3) control system as an improvement (for not requiring large power The RSPM system basically con- supply). In the following dis- sists of three components, the same cussion of research progress, the as those of an active control system. term RSPM control technology or They are: sensors, controllers and ISMIS control technology will be functional switches (actuators in the used. ISMIS®, an abbreviation for case of active control). Intelligent Shock Mitigation and The functional switches are Isolation System, is the registered essentially hydraulic devices with trademark for Enidine Incorp- the ability to deliver variable orated’s commercial system for stiffness and/or variable damping. naval and civilian applications. They are strategically located in a Conceptually, semi-active systems structure and individually con- may involve variable damping and/ trolled. Obviously, they can be or variable stiffness, as defined by connected/disconnected to mass Equations (1) and (2). The RSPM and to base isolators. control system is described by Three typical utilizations of the Equation (3). functional switches are illustrated ■ Table 1. Development of the ISMIS®Technology in Figure 2. In this figure, a functional switch is used to connect/ noitareneGts1 noitareneGdn2 noitareneGdr3 disconnect an auxiliary mass m (FS1), sretemoreleccA sretemoreleccA llecdaoL control an auxiliary damper (FS ), rosneS TDVL 2 srosnesyticoleV rosneserusserP llecdaoL and connect/disconnect auxiliary ,rellortnocgolanA ,rellortnocgolanA structural members to control the ,rellortnoclatigiD rellortnoC citengam-ortcelE citengam-ortcelE stiffness of the system (FS ). evlavovreS 3 evlavdionelos evlavlanoitroporp The MCEER project to develop lanoitcnuF ni6=htgneL ni83=htgneL ni83=htgneL RSPM technology began in 1993 hctiwS sbl4=thgieW sbl012=thgieW sbl07=thgieW Note: Length is determined by the application when the basic concept was con- ceived and the first generation

38 ■ Figure 3 shows the test Development of Control set up for the first Devices for RSPM generation system and Technology the corresponding functional switch. The first generation system was a small prototype control device. It was a hydraulic device consisting of a plung- er seated in a cylinder. An external fluid reservoir was connected to the internal chamber to back-fill it at push-back stage. The reservoir hardware system was manufactured. prevents the cavity in the internal To date, experimental studies of the chamber, but is only maintained at third generation system have been atmosphere pressure. Thus, the completed. This progression is device does not have any reserved illustrated in Table 1. stiffness from the pre-pressured For the first generation system, fluid. The valve control is realized the functional switch was through an electro-magnetic approximately 6 inches long and solenoid. Detailed test data and weighed about four pounds (see performance of the device are Figure 3). That system provided a reported in Liang et al. 1995; 1999. proof of the concept and a The second generation of the successful application for a U.S. control device was a servo-valve patent (No. 5,526,609) to develop hydraulic system. The unit operates the technology. The second and on high pressure hydraulic power. third generation systems were The device was built to realize high developed under the ISMIS/TRP performance with the option to also program sponsored by ONR and examine other continuous type DARPA. The second generation control schemes. The study of the system was developed and tested control device is during 1996-97 (see Figure 4). This reported in Lee et al., proof of concept functional switch used top of the line, commercial 1998a; 1998b. off the shelf components to provide the foundation for determining the best specifications for the next generation. These models weighed 200 lbs. The third generation system provided a more streamlined design, reducing the weight to 70 ■ lbs, and was considered to be Figure 4 shows the test set up for the second generation system and reasonably proportional to its the corresponding functional switch. length. This third generation The functional switch in this configu- system was manufactured and ration was 38 inches long. tested between 1997 and 1999. Results are summarized in a later section.

Development of a Semi-Active Structural Control System39 The third generation of the control The second level of the hier- device was a compact unit with all archical control, L2, is also a local working parts built internally. The loop. It has been designed to deal control valve was changed to an with some major side effects electro-magnetic proportional valve associated with switching type of that operates on low voltage power. control. The typical scenarios are the The electronic control circuit was signal delay and overdrift due to simplified to further shorten the unbalanced force from a quick signal delay. Extensive tests of the change of physical parameters. The device on a four story steel structure delay issue involves many factors, have been carried out. which can become rather complicated. The overdrift often Hierarchical Control System occurs at the time when response frequency is lower than the The hierarchical control system dominant natural frequency at the is a special feature of the RSPM local area. This is primarily related technology. Four levels of loops to the phase differences between are included in the conceptual the input excitation and the local design. A detailed discussion of these hierarchical control loops can output response. Many algorithms be found in Liang et al., 1995; 1999. have been studied to modify the The following is a brief summary of primary control loop. Some detailed their logical connections. modeling on the signal delay is The first level of the hierarchical provided in Lee et al. 1998. Recent improvement of the technology has control, L1, is a local loop. This loop was realized by employing simple consisted of a combination of and robust actuation. At present, a passive damping in the control control algorithm for this level was devices to improve both side effects. created based on variation of The third level of the hierarchical

stiffness parameters. It was found control, L3, is a global loop. When that the physical parameters under the control object is a complicated control do not need to be changed very frequently. This led to the development of a switching type of control device, which can tolerate more hostile environments. A local or global sensing system can be used to feed the control unit with the proper control signal. While the switching type of control bears the advantages of being simple, reliable and cost efficient, its limitation is that once the control command has been issued, it ■ Figure 5. Test Setup for Third Generation System. The cannot reverse the effect. four story model is show at left; the top photo shows This problem is treated in the control unit; the bottom photo provides a detailed the second loop. view of the bracing type configuration.

40 structure, unevenly distributed dynamic working Semi-active Control Fail-safe 1: valve open) Case, range, the cost of White-noise input dynamic characteristics may result 60.00 1st Floor RelDis in reductions of the overall realizing the desired 50.00 2nd Floor RelDis specification may 3rd Floor RelDis performance of the control system. 40.00 4th Floor RelDis vary significantly. By employing a global optimization 30.00 scheme, the performance may be For instance, the 20.00 improved. The global algorithm is delay restriction for 10.00 still under development at present; a 2 Hz application is 0.00 the theoretical basis of the control much easier to solve 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 Semi-active Control Fail-safe 2: valve close) Case, than for 20 Hz. In White-noise input algorithm is minimization of con- 120.00 servative energy (Constantinou et order to test the 100.00 al., 1998, and Liang et al., 1995; 1999). RSPM technology 80.00 The third level loop will override the for practical seismic first and second loops. However, the applications, it was 60.00 global loop control often requires a necessary to include Magnitude 40.00 central processing unit, which adds some general struc- 20.00 tural characteristics 0.00 cost to the entire system. 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 in the test vehicle. Semi-active Control Case, White-noise input The fourth level of the hierarchical 45.00 Since the majority 40.00 control, L4, is a safety loop. The control criteria are established by of building struc- 35.00 tures requiring sup- 30.00 various safety concerns that are not 25.00 directly related to the improvement plemental energy 20.00 dissipation devices 15.00 of structural performance. Also, 10.00 when the control units fails, the are high-rises, a four 5.00 0.00 actuation device will set, by default, story moment re- 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 a fail-safe mode. sistant steel structure Freq. Hz was designed for ■ Figure 6. Floor Relative Displacement Frequency the shaking table Response Experimental tests. The structure Observations is 6 feet (w) x 6 feet (l) x 20 feet (h) with a natural Testing has been an important frequency of 2.0 Hz. The dead load part of the technology develop- is 50 lbs. per square foot, which is ment from the initiation of the similar to that in a real structure. project. Back in 1994, a small model The test structure was not structure (see Figure 3) was used intended to be a quarter scale to test first generation control structure in the strict sense. The devices and verify the system consideration to abandon the usual concept (see Liang et al., 1995; similitude approach in the design 1999). Since 1997, tests have been was due to the following factors: performed on a new model using the acceleration controlled shaking 1. The test results of the technology table, which provides more accurate are more relevant to practical calibration of the technology. application if the test dynamic Many technical specifications of working range is the same as the RSPM technology are related to its working range of full size devices. design dynamic working range. 2. The higher frequency problem Although the technology is robust is more demanding for the and capable of covering a wide response time and often provides

Development of a Semi-Active Structural Control System41 ■ Table 2. Kobe Earthquake Peak Response Comparison Test results obtained during 1998- 99 are summarized below. The (e)hcni marF Passive Damping High Stiffness RSPM

ts newest generation of the control 1 F2rool 0544. 0%91. 689.6 0%51. 742.3 0%31. 3.77 device has the ability to switch dn 2 F7rool 0899. 0%24. 618.7 0%63. 738.2 0%03. 2.77 between three states: damping, dr

Floor Disp. 3 F9rool 1814. 0%06. 639.7 0%56. 655.5 0%15. 8.27 stiffness and ISMIS control. Each 4 ht F2rool 1876. 0%17. 698.7 0%58. 645.1 0%66. 2.07 of the three states is compared under different input. worse scenarios than lower The test setup is illustrated in frequency problems. Therefore, Figure 5, where the RSPM control it was desirable to verify the devices are diagonally braced on the technology at the higher first and second floors of the four- frequency end. story structure. Absolute displace- 3. The higher frequency prototype ment sensors were placed at each of device and controller are more the floors. Accelerometers and strain expensive. gages were placed on the floors and 4. Regular low-rise buildings are at connection areas, respectively. not considered to be good can- Figure 6 shows the frequency didates for supplemental control response function of the three states devices for cost reasons. with the third generation system. It 5. Size limitations of the shaking was again verified that the switching table. scheme was effective when com- Although the test structure pared to high damping or high cannot provide all pertinent stiffness states. The corresponding information about the technology, damping ratios delivered by the in particular, information con- three states were 13%, 8% and 14%, cerning the multi-bay multi-story respectively. system, it has provided a basic Table 2 summarizes a set of understanding of the technology at displacement time history test re- its practical working range. sults. Under 16% Kobe earthquake

1st Floor Relative Displacement Time History (Kobe) 3rd Floor Relative Displacement Time History (Kobe) 0.600 2.000 Bare Frame100% Control 1.500 0.400 1.000 0.200 0.500

0.000 0.000 0 5 10 15 20 25 30 0 5 10 15 20 25 30 -0.500 -0.200 -1.000 -0.400 -1.500

-0.600 -2.000

2nd Floor Relative Displacement Time History (Kobe) 4th Floor Relative Displacement Time History (Kobe) 1.500 2.500 2.000 1.000 1.500

0.500 1.000 0.500 Relative Displacement Inches 0.000 0.000 0510 15 20 25 30 0 5 10 15 20 25 30 -0.500 -0.500 -1.000 -1.500 -1.000 -2.000 -1.500 -2.500 Time (sec) ■ Figure 7. Relative Displacement Time History With and Without Control

42 1st Floor Acceleration Time History (Kobe) 2nd Floor Acceleration Time History (Kobe) 0.40 0.80 Bare Frame 100% 0.30 Passive Damper 0.60 0.20 0.40

0.10 0.20 0.00 0.00 0 5 10 15 20 25 30 0 51015202530 -0.10 -0.20

-0.20 -0.40

-0.30 -0.60

-0.40 -0.80

3rd Floor Acceleration Time History (Kobe) 4th Floor Acceleration Time History (Kobe) 1.00 1.20 0.80 1.00

Acceleration g 0.60 0.80 0.60 0.40 0.40 0.20 0.20 0.00 0 5 10 15 20 25 30 0.00 0 51015202530 -0.20 -0.20 -0.40 -0.40 -0.60 -0.60 -0.80 -0.80 -1.00 -0.10 Time (sec) ■ Figure 8. Acceleration Time History With and Without Control excitation, the original responses of is potentially more cost effective the structure were compared with than other devices, especially if passive damping, braced stiffness combined with approaches such as and RSPM control states. structural bracing or base isolation Figure 7 provides the time (Ruan, 1997; and Ruan et al., 1997). histories of the displacement Currently, full scale imple- response with and without control. mentation in a building is being Figure 8 provides a comparison of reviewed. This will be one of the acceleration response of the four two major tasks for the project floors. Based on a wide range of during 1999-2000. A second major tests performed on the shaking task for 1999-2000 and beyond will table, which include white-noise, be the continued development of scaled earthquake records, and guidelines for optimal (or new modified earthquake records, it was optimal) RSPM placement in a given found that the RSPM technology is robust, and outperforms the two structure. This is a rather complex passive states in every test case. In problem but information is needed particular, the reduction effect is by the engineering profession more significant with the large dealing with structures (new or amplitude real earthquake records. retrofit) to be implemented with all types of earthquake protective systems. Conclusion and Future The RSPM technology can be Work applied to other energy dissipation and shock mitigation situations as a An extensive test program has mechanical device, with minor established that ISMIS/RSPM can modifications (see Lee et al., 1997, provide more effective control of and Rasmussen et al., 1997). This is story drift than many other passive beyond the scope of earthquake energy dissipation devices. In engineering. particular, testing showed that RSPM

Development of a Semi-Active Structural Control System43 References

Constantinou, M.C., Soong T.T. and Dargush, G.F., (1998), Passive Energy Dissipation System for Structural Design and Retrofit, Multidisciplinary Center for Earthquake Engineering Research Monograph Series, No. 1, 1998.

Lee, G.C., Liang, Z. and Tong, M., (1994), “Innervated Structures,” Proceedings of the First World Conference on Structural Control, Los Angeles, California, August. Lee, G.C., Tong, M., Liang, Z., Houston, B., Taylor, S., Clinard, L., Tomita, K. and Gottwald, W.D., (1997), “Structural Vibration Reduction with ISMIS Control Technology” Proceedings of the 68th Shock and Vibration Symposium, Baltimore, Maryland. Lee, G.C., Liang, Z., Tong, M., Niu, T., Song, J., Wu, Y., Qi, J. and Tomita, K., (1998a), “Seismic Test on a Model Structure with Variable Passive Control” Proceedings of the 69th Shock and Vibration Symposium. Lee, G.C., Liang, Z. and Tong, M., (1998b), “Development of a Variable Passive Control System for Structural Response Reduction” Proceedings of the Second World Conference on Structural Control, Kyoto, Japan. Liang, Z. and Lee, G.C., (1991), Damping of Structures: Part I – Theory of Complex Damping, Technical Report NCEER-91-0004, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Liang, Z., Tong, M. and Lee, G.C., (1995), Real-Time Structural Parameter Modification (RSPM): Development of Innervated Structures, Technical Report NCEER-95-0012, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York. Liang, Z., Tong, M. and Lee, G.C., (1999), “A Real-Time Parameter Modification (RSPM) Approach for Random Vibration Reduction. Part I: Principle and Part II: Experimental Verification,” Journal of Probabilistic Engineering Mechanics, June. Rasmussen, F.F., Gottwald III, W.D., Krasnicki, E. and Lee, G.C., (1997), “Intelligent Shock Mitigation and Isolation system Through Applied Semi-Active Vibration Control Technology,” Proceedings of the 68th Shock and Vibration Symposium, Baltimore, Maryland.

Ruan, S., (1997), Animated Control of Base Isolation Systems, Ph.D. Dissertation, State University of New York at Buffalo. Ruan, S., Liang, Z. and Lee, G.C., (1997), “Animated Control of Base Isolation System,” Proceedings of the 15th International Model Analysis Conference, Orlando, Florida.

44 GIS Characterization of the Los Angeles Water Supply, Earthquake Effects, and Pipeline Damage

by Thomas D. O’Rourke, Selcuk Toprak and Sang-Soo Jeon, Cornell University

Research Objectives The objectives of the research are to: 1) develop a comprehensive MCEER/NSF database of the size, composition, and geographic location of all Los Angeles Department of Water Power (LADWP) pipelines for system modeling and reliability analyses; 2) evaluate the spatial distribution of pipeline damage and its relationship with transient and permanent ground deformation parameters; and 3) use the resulting relationships to improve both loss estimation methodologies and the identification of seismic and geotechnical hazards in the Los Angeles area. There is a fourth objective that was also achieved, although its accomplishment • This research project has was not foreseen at the start of the project, but emerged as a true identified the components scientific discovery as the work progressed. This fourth objective is and facilities that most seriously affect the to improve GIS characterization by defining an explicit relationship earthquake performance of between the mesh size used to process point source geographic data water systems and thus are and the two-dimensional visualization of these data through mapping prime targets for improve- algorithms. ments using advanced technologies. This research is an integral part of the Lifelines Program, and is CEER-sponsored research focuses on improved loss estimation meth- also an important odologies and the application of advanced technologies to improve contributor to the M Emergency Response and water system performance. It is important therefore to evaluate how water Recovery Program through systems respond to real earthquake conditions, using information technol- its in-depth characteriza- ogy for comprehensive characterization of the spatially variable pipeline net- tion of critical water work, transient and permanent ground deformation patterns, and supplies and its develop- geotechnical, groundwater, and topographical features. Geographical Infor- ment of advanced GIS mation Systems (GIS) are ideally suited for this type of investigation, and applicable for post- were used to develop a detailed and extensive inventory of the Los Angeles earthquake damage Department of Water and Power (LADWP) water delivery system as well as a assessment and deploy- ment of emergency comprehensive assessment of system performance during the Northridge services and system earthquake. restoration resources. The 1994 Northridge earthquake resulted in the most extensive damage to a U.S. water supply system since the 1906 San Francisco earthquake. Los Angeles Department of Water and Power and Metropolitan Water District (MWD) trunk lines (nominal pipe diameter ≥ 600 mm) were damaged at 74 locations, and the LADWP distribution system required repairs at 1,013 locations. The widespread disruption provides a unique opportunity to evaluate 45 the geographic variability of the most seriously affected by the damage, the most vulnerable pipe- Northridge earthquake superim- lines, and the relationship among posed on the topography of Los • LADWP contributed damage, transient motion, and per- Angeles. The water supply system extensive data sets, maps, manent ground deformation. includes transmission lines, trunk and information about Cornell researchers collected in- lines and distribution lines. All large system facilities. The formation on nearly 1,100 water diameter pipelines upstream of the Department also contrib- pipeline repairs after the Northridge treatment plants are considered to uted substantial amounts earthquake, including location, pipe be transmission facilities. of engineering services to gather, evaluate, and diameter, and composition of the Figure 2 presents charts showing verify the data and GIS lines. In addition, they digitized ap- the relative lengths of LADWP and work performed. proximately 11,000 km of distribu- MWD trunk and distribution lines, tion mains (diameter < 600 mm) and according to pipe composition. It • Metropolitan Water 1000 km of trunk lines (diameter ≥ should be noted that the vertical District (MWD) of 600 mm) operated throughout the axis in Figure 2c is a logarithmic Southern California, city of Los Angeles. This information scale. The MWD trunk lines include California Department of Transportation was incorporated in a GIS using pipelines within the area of the (Caltrans), and Los ARCINFO software where it is com- LADWP system designated by MWD Angeles Bureau of bined with over 240 corrected as feeder lines. Total lengths of ap- Engineering (LABE), all strong motion records; vectors of proximately 700 km and 300 km for provided data and horizontal displacement, heave, and LADWP and MWD trunk lines, re- information pertaining to settlement determined by air photo- spectively, are included in the data- system characteristics, grammetry techniques; and Los An- base, and the pie charts in each fig- pre- and post- earthquake surveys, and geles street system, topography, sur- ure show the relative percentages Global Positioning System ficial soils, and depths to water table. of the combined LADWP and MWD (GPS) control points. Figure 1 shows that the portion of trunk lines associated with different the Los Angeles water supply system types of pipe material. • EQE International, Inc. and the California Office of Emergency Services provided the initial geo- The users of the research results include water utilities, such coded data about water as the Los Angeles Department of Water and Power (LADWP), supply system damage. East Bay Municipal Utility District, Memphis Light, Gas and Water, and many other companies operating systems in areas • The research work was vulnerable to earthquakes; governmental agencies, such as coordinated with MCEER- the Federal Emergency Management Agency (FEMA), which supported studies of the LADWP electric power promote the development and application of earthquake loss system supervised by estimation methodologies; and various private enterprises, M. Shinozuka, University such as electric power utilities, engineering firms, and insur- of Southern California ance carriers, that are interested in advanced applications of (USC), and with work GIS for civil infrastructure improvement and risk assessment. supported by the Pacific Research on advanced GIS technologies and risk assessment Earthquake Engineering of lifeline networks not only provides for substantial improve- Research (PEER) Center on ground failures triggered ments in seismic performance, but also establishes the plat- by the Northridge earth- form for better management irrespective of seismic hazards. quake, supervised by These improvements carry substantial societal benefits dur- J.P. Bardet, USC. ing normal operations through increased efficiency, safety and reliability, and through reduced maintenance and repair costs.

46 • Photogrammetric measurements with pre- and post- earthquake air photos were performed by the Hasshu Company through the supervision of M. Hamada, Waseda University, Tokyo, Japan.

• The evaluation of geotechnical and seismological characteristics of sites with concentrated lifeline damage was performed with input and cooperation with the U.S. Geological Survey, Menlo Park, California. O’Rourke and Toprak, 1997 ■ Figure 1. Map of Los Angeles Water Supply System Affected by the Northridge Earthquake

Figure 3 presents a map of distri- damage with the spatial distribu- bution pipeline repair locations and tion of various seismic parameters. repair rate contours for cast iron The maximum strong motion read- (CI) pipeline damage. The CI mains ings at the Tarzana-Cedar Hill Nurs- were shown to have the broadest ery were removed from the data- geographic coverage, and therefore base prior to GIS evaluation to to provide the most consistent ba- avoid distortions from possible to- sis for evaluation of seismic re- pographic influences. In addition, sponse throughout the entire system (O’Rourke and Toprak, Cast Iron Ductile 1997). The repair rate contours Concrete 11% Iron were developed by dividing the 18% 1% 10000 Trunk Lines : 1014 km map into 2 km x 2 km areas, deter- Distribution Lines : 10750 km mining the number of CI pipeline Concrete Riveted Steel Riveted repairs in each area, and dividing the Cast Iron Steel Steel Steel repairs by the distance of CI main 14% 56% Asbestos Cement Ductile Iron in that area. Contours then were a) Trunk Lines Steel 1000 drawn from the spatial distribution Asbestos 11% 9% of repair rates, each of which was Length (km) centered on its tributary area. The Ductile 2 km x 2 km grid was found to pro- Iron vide a good representation of dam- 4% 100 age patterns for the map scale of the Cast Iron LADWP LADWP MWD figure. 76% b) Distribution Lines The records from approximately c) Combined Lines 240 rock and soil stations were used O’Rourke and Toprak, 1997 to evaluate the patterns of pipeline ■ Figure 2. Composition Statistics of Water Trunk and Distribution Lines

GIS Characterization of the Los Angeles Water Supply47 records from stations at dam abutments were screened when • Seismic Reliability Analysis a station down- and Retrofit Method for stream of the dam Southern California Power was available, again Systems, T.C. Cheng, to minimize distor- University of Southern tion from topo- California graphic effects. • Rehabilitation Strategies for Figure 4 shows the Lifelines: LADWP Power CI pipeline repair Systems, S.T. Mau, New rate contours super- Jersey Institute of Technol- imposed on zones of ogy, T.C. Cheng and peak ground veloc- M. Shinozuka, University ity. By evaluating the O’Rourke and Toprak, 1997 of Southern California zones of ground ve- ■ Figure 3. Cast Iron Pipeline Repair Rate Contours for the locity with GIS, as il- Northridge Earthquake lustrated in Figure 4, it was possible to correlate the pipe- response. By correlating damage line repair rates in all the zones char- with various seismic parameters, re- acterized by a particular velocity with gressions were developed between the velocity pertaining to those zones. repair rate and measures of seismic As explained by O’Rourke (1998), intensity. similar evaluations were made of The most statistically significant pipeline damage relative to spatially correlations for both distribution distributed peak acceleration, spec- and trunk line repair rates were tral acceleration and velocity, Arias found for peak ground velocity. Fig- Intensity, Modified Mercalli Intensity ure 5a presents the linear regression (MMI), and others indices of seismic that was developed between CI pipeline repair rates and peak ground velocity on the basis of data from the Northridge and other U.S. earthquakes. Figures 5b and 5c show repair rate correlations for welded steel trunk lines and for cast iron, ductile iron, and asbestos cement distribution lines. Figure 5d compares the regressions developed in this research work with the default relationship used in HAZUS (NIBS, 1997), which is the computer program that implements the current earthquake loss estimation methodology sponsored by FEMA. The FEMA correlation does not distinguish between trunk or distribu- O’Rourke and Toprak, 1997 tion lines, nor does it allow for pre- ■ Figure 4. Pipeline Repair Rate Contours Relative to Northridge Earthquake dictions based on pipe composition. Peak Ground Velocity The data pertaining to the FEMA

48 4 8 7 3 Fit E q ua tion : 6 log(Y) = 1.62 * log(X)-3.64 5 F it E qu a tio n: R -squared = 0.86 4 log(Y) = 1.62 * log(X) - 3.64 2 R-squared = 0.86 3

2 0.10 9 8 7 0.10 9 6 8 5 7 6 4 5

3 4 1994 Northridge 3

2 1989 Loma P rieta 2 1987 W hittier Narrows Repair Rate (Number of Reparis/km) of (Number Rate Repair Repair Rate (Number of Reparis/km) of (Number Rate Repair 197 1 San Fernan do (so uth) 0.01 0.01

23456789 23456789 2 10 100 10 1 00 PGV (cm/sec) PGV (cm/sec)

a) Cast Iron Distribution Line b) Welded Steel Trunk Line

4 4 Fit Eq ua tion (C I): 3 3 HAZUS log(Y) = 1.62 * log(X) - 3.64 2 R-squared = 0.86 2 CI CI 0.1009 8 7 DI 6 5 AC 0.10 4 9 3 8 DI 7 2 6 5 4 0.0109 8 Fit Eq ua tion (D I): 7 3 AC 6 log(Y) = 1.84 * log(X) - 9.40 5 4 R-squared = 0.74 2 Trunk 3

2 Fit Equation (AC): log(Y) = 2.26 * log(X) - 11.01 Repair Rate (Number of Reparis/km) of (Number Rate Repair Repair Rate (Number Reparis/km)of R-squared = 0.72 0.001 0.01

23456789 23456789 2 10 100 10 1 00 PGV (cm/sec) PGV (cm/sec)

c) CI, DI, and AC Distribution Lines d) Distribution and Trunk Lines

■ Figure 5. Pipeline Repair Rate Correlation with PGV for CI, DI and AC Distribution and Welded Steel Trunk Lines

GIS Characterization of the Los Angeles Water Supply49 correlation were analyzed before the cur-

rent generation of GIS Surface Analysis technologies, and are Contour Interpolation strongly influenced by repair statistics for

the Mexico City water Ground Strains Ground Strain Contours supply after the 1985 Michoacon earthquake (Ayala and O’Rourke, 1989). In- Overlay spection of Figure 5d reveals that the cur- Length in Each Strain Range Pipeline System Repair Rate rent default relation- (Repairs/Length) ship in HAZUS is very vs. Ground Strain conservative, and results in predicted re- Overlay pair rates that exceed those provided by the Repairs in Each Strain Range Pipeline Repairs regressions devel- O’Rourke et al., 1998 oped in the MCEER- ■ Figure 6. Procedure for Calculating Repair Rate in Each Strain sponsored research Range by over an order of Complex were analyzed as part of magnitude for steel trunk lines and collaborative research between U.S. by a factor of two to three for dis- and Japanese engineers (Sano, 1998; tribution mains subjected to veloci- O’Rourke et al., 1998). The area near ties greater than 20 cm/s. the intersection of Balboa Blvd. and After the Northridge earthquake, Rinaldi St. has been identified as a pre- and post-earthquake air photo location of liquefaction (Holzer et measurements in the Van Norman al., 1996) where significant damage to gas transmission and water trunk lines was incurred. Ground strains were calculated in this area from the air photo measurements of horizontal displacement by superimposing regularly spaced grids with GIS software onto the maps of horizontal displacement and calculating the mean displacement for each grid. Grid dimensions of 100 m x 100 m were found to provide the best results (Sano, 1998). As illustrated in Figure 6, ground strain contours, pipeline network, and repair locations were combined using GIS, after which repair O’Rourke et al., 1998 rates corresponding to the areas ■ Figure 7. Distributions of CI Repair Rate and Ground Strain delineated by a particular contour 50 figure, repair 10 rates increase “The GIS- Airphoto linearly with 8 Ground Survey based research ground strain. A focused on the high r2 value 6 LADWP system 2 shows that a R = 0.8279 has resulted 4 large percentage of the data vari- in the largest 2 ability can be ex- U.S. database plained by the ever assembled Repair Rate (repairs/km) 0 regression line. of spatially 0 0.1 0.2 0.3 0.4 0.5 With GIS, it is distributed Ground Strain (%) very easy to di- transient and vide a spatially permanent O’Rourke et al., 1998 distributed data ■ ground Figure 8. Correlation between Ground Strain and CI Repair Rate set into arbi- deformation.” interval were calculated. Figure 7 trarily sized ar- shows the repair rate contours for eas. If the areas are delineated by a CI mains superimposed on the ar- framework of equally spaced, verti- eal distribution of ground strains, cal and horizontal lines, the result- identified by various shades and ing grid can be characterized by a tones. In the study area, there were single dimension representing one 34 repairs to CI water distribution side of each area, n, and the number, N, of areas comprising the total mains and two for steel water dis- 2 tribution pipelines. There were five area of the system, Nn . The choice water trunk line repairs in the area. of n can be regarded as a means of The repair rate contours were de- visually resolving the distribution of veloped by dividing the map into damage. 100 m x 100 m cells, determining Some practical questions emerge. the number of CI pipeline repairs Is there a useful relationship between in each cell, and dividing the repairs n and the visualization of zones with by the length of the distribution high damage? An additional question mains in that cell. The intervals of may be asked about what values of n strain and repair rate contours are represent the best choices for visual- 0.001 (0.1%) and 5 repairs/km, re- izing damage patterns? spectively. The zones of high ten- In this research project, a relation- sile and compressive strains coin- ship was discovered between the cide well with the locations of high area of the map covered by repair repair rate. rate contours and the grid size, n, In Figure 8, the relationship be- used to analyze the repair statistics. tween ground strains and repair If the contour interval is chosen as rates is presented graphically using the average repair rate for the en- linear regression. The repair rate in tire system or portion of the system each ground strain range, 0-0.1, 0.1- covered by the map, then the area 0.3, and 0.3-0.5%, was calculated as in the contours represents the explained previously. Ground strain zones of highest (greater than av- contours obtained both from the air erage) earthquake intensity as photo measurements and LABE reflected in pipeline damage. The survey were used. As shown in this area within the contour lines

GIS Characterization of the Los Angeles Water Supply51 divided by an area closely related of various dimensions from which to the total area of the map, , is the relationship was developed. AC referred to as the threshold area cov- This relationship can aid GIS us- erage, TAC (Toprak et al., 1999). Al- ers to get sufficiently refined, but ternate thresholds may also be easily visualized, maps of damage defined on the basis of the mean patterns. Because the relationship plus one or two standard deviations. is independent of size and will work A hyperbolic relationship was at the scale of the entire system or shown to exist between TAC and the any practical subset thereof, it can dimensionless grid size, defined as be used for damage pattern recogni- the square root of n2, the area of an tion, and for computer “zooming” individual cell, divided by the total from the largest to smallest scales to

map area, AT . This relationship is identify zones of concentrated dis- illustrated in Figure 9, for which a ruption. This relationship has great schematic of the parameters is pro- potential for data management to vided by the inset diaphragm. The support emergency response deci- relationship was found to be valid sions and planning for optimal post- over a wide range of different map earthquake recovery. scales spanning 1200 km2 for the entire Los Angeles water distribution system affected by the Northridge Conclusions earthquake to 1 km2 of the San Fran- The GIS-based research focused cisco water distribution system in on the LADWP system has resulted the Marina affected by the Loma in the largest U.S. database ever Prieta earthquake (Toprak et al., 1999). The data points refer to maps assembled of spatially distributed

0.6

0.5

Mean RR 0.4 A1

Fit Curve 0.3 All LA

All SFV A2

0.2 Section of SFV TAC = (A1 + A2)2 /AC 2 DGS = SQRT (N /AT) Sherman Oaks NNNNN N North of LA Basin

Threshold Area Coverage,TAC Area Threshold 0.1 San Francisco TAC = DGS / ( 1.68 * DGS + 0.11) 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Dimensionless Grid Size, DGS Toprak et al., 1999 ■ Figure 9. Hyperbolic Fit for Threshold Area Coverage and Dimensionless Grid Size

52 transient and permanent ground parameters, permanent ground de- deformation in conjunction with formation, geotechnical hazards, to- earthquake damage to water supply pography, and groundwater tables. and other lifeline systems. The re- This characterization will be used search has led to a better delineation in forthcoming systems analysis and of local geotechnical and seismologi- reliability assessments of the cal hazards that are shown by the LADWP water supply network. Be- zones of concentrated pipeline dam- cause of its unprecedented size, age after the Northridge earthquake. complexity, and accuracy, this GIS The research has resulted in correla- data set is currently being used in tions between repair rates for a vari- studies at the NSF-supported Insti- ety of trunk and distribution pipelines tute for Civil Infrastructure Systems and seismic parameters, such as peak to explore relationships between ground velocity. These correlations physical infrastructure systems and are statistically reliable and have im- the social, economic, and political proved predictive capabilities com- databases that coincide with them. pared with the default relationships An evaluation of the earthquake currently used in computer programs damage statistics of the LADWP sys- developed by FEMA for earthquake tem has disclosed critical compo- loss estimation. The research has led nents that are most susceptible to to the discovery of a relationship be- earthquake damage and have the tween the two dimensional represen- greatest impact on system perfor- tation of local damage and the grid mance. Based on the findings of size used in GIS to analyze the spatial this work, future research will fo- distribution of data. For practical pur- cus on the welded slip joints of poses, this relationship is independent steel trunk lines that are susceptible of scale and therefore ideally suited to compressive failure under tran- for damage pattern recognition and sient and permanent ground defor- computer “zooming” from largest to mation. Research will concentrate smallest scale to target areas for emer- on characterizing the load-deforma- gency response and recovery. tion behavior of these joints and the The research has resulted in a use of externally applied fiber rein- comprehensive GIS characteriza- forced composites (FRCs) to in- tion of the LADWP pipeline net- crease their load carrying capacity work, earthquake damage patterns, as part of either retrofit or new con- and spatial distributions of seismic struction activities.

References

Ayala, A.G., and O’Rourke, M.J., (1989), Effects of the 1985 Michoacan Earthquake on Water Systems in Mexico, Technical Report NCEER-89-0009, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York. Holzer, T.L., Bennett, M.J., Tinsley III, J.C., Ponti, D.J., and Sharp, R.V., (1996), “Causes of Ground Failure in Alluvium during the Northridge, California, Earthquake of January 17,1994,” Proceedings from the Sixth Japan-U.S. Workshop on Earthquake Resistant Design of Lifeline Facilities and Countermeasures Against Soil Liquefaction,” Waseda University, Tokyo, Japan, June 11-13, 1996, Technical Report NCEER-96-0012, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 345-360.

GIS Characterization of the Los Angeles Water Supply53 National Institute of Building Sciences, (1997), Earthquake Loss Estimation Methodology: HAZUS, NIBS Document No. 5201 prepared for Federal Emergency Management Agency, Technical Manual Vol. 1 and 2, Washington, D.C. O’Rourke, T.D., (1998), “Overview of Geotechnical and Lifeline Earthquake Engineering,” Geotechnical Special Publication No. 75, ASCE, Vol. II, P. Pakoulis, M. Yegian, and D. Holtz, Eds, Reston, Virginia, pp. 1392-1426. O’Rourke, T.D. and Toprak, S., (1997), “GIS Assessment of Water Supply Damage Caused by the Northridge Earthquake”, Proceedings of the Sixth U.S. National Conference on Earthquake Engineering, Seattle, Washington, EERI, May. O’Rourke, T.D., Toprak, S. and Sano, Y., (1998), “Factors Affecting Water Supply Damage Caused by the Northridge Earthquake,” Proceedings of the Sixth U.S. National Conference on Earthquake Engineering, Seattle, Washington, June, pp. 1-12. Sano, Y., (1998), GIS Evaluation of Northridge Eartrhquake Ground Deformation and Water Supply Damage, M.S. Thesis, Cornell University, Ithaca, New York. Toprak, S., O’Rourke, T.D., and Tutuncu, I., (1999), “GIS Characterization of Spatially Distributed Lifeline Damage,” Proceedings of the Fifth U.S. Conference on Lifeline Earthquake Engineering, Seattle, Washington, August, in press.

54 Axial Behavior Characteristics of Pipe Joints Under Static Loading

by Emmanuel Maragakis, Raj Siddharthan and Ronald Meis, University of Nevada - Reno

Research Objectives MCEER/NSF The objective of this study is to perform physical testing on pipe joint segments to determine the axial static load behavior characteristics of various types of pipe joints typically used in both above ground and buried piping systems. The pipe joints considered include cast iron, steel, and ductile iron bell and spigot joints as well as newer types of restrained joints, and other types of pipe materials such as PVC and polyethylene. This effort is part of an overall research project to determine the dynamic behavior characteristics of pipe joints and to develop fragility information and risk assessment data. • The results of this research are one of the critical inputs to the seismic vulnerability assessment of amage reports from past earthquakes have clearly revealed that bur- piping systems. Such Died and above surface pipelines are prone to severe damage in areas vulnerability studies are of strong shaking. Given the vital importance of this infrastructure necessary to undertake cost system, understanding how pipe and pipe joints respond during an earth- effective retrofit measures quake is a necessity. The proposed testing is the first phase of an overall based on engineering research project designed to determine the static and dynamic strength information. characteristics of pipe joints, and to develop fragility information and risk assessment data. Such information is critical to determine potential damage of piping systems when subjected to seismic motion. Among the many well-documented pipeline failures under earthquake loading, a few important studies have been selected and summarized below. O’Rourke (1996) reviewed the performance of and damage to pipelines following various earthquakes. In the 1989 Loma Prieta earthquake, the major damage was concentrated in areas of liquefaction such as the Marina district in San Francisco. San Francisco, Oakland, Berke- ley, and the Santa Cruz area had almost 600 water pipeline failures. In the 1994 Northridge earthquake, over 1,400 failures were reported including 100 failures to critical large diameter pipelines. In the 1995 Kobe earthquake, as many as 1,610 failures occurred in distribution water mains and 5,190 failures occurred in distribution gas mains.

55 Trifunac and Todorovska (1997) mechanisms that were observed. reported on a detailed investiga- They reported that the majority of tion of the amount of pipe breaks pipeline failures were at the joints, • Pipe manufacturers, that occurred in the Northridge and the predominate modes of fail- including U.S. Pipe, earthquake. They concluded that ure were slip-out of the joints and ACIPCO, EBBA Iron, and the “pipe breaks correlate well with the intrusion of the spigot into the Uni-Flange, donated in all the recorded amplitudes of strong bell. For this reason, the emphasis more than 15 pipe joint ground motion....”. They presented of this testing is on axial loading be- segments and appropriate havior (compression and tension). end flanges, shipping costs empirical equations which related and samples of joint the average number of water pipe A thorough survey of the litera- restraint mechanisms. breaks per km of pipe length with ture reveals that laboratory tests on the peak strain in the soil or inten- pipe joints are limited. Singhal • Manufacturer’s associa- sity of shaking at the site. (1984) performed a number of tions, including the Ductile O’Rourke and Palmer (1996) re- static experiments on bell and Iron Pipe Research viewed the historical performance spigot rubber gasketed joints to Association (DIPRA) and the American Water Works of gas pipelines, steel and plastic, determine their strength and stiff- Association (AWWA), in southern California over a 61 ness characteristics. The joints provided technical and year period. Statistics are provided were subjected to axial and bend- other research materials for 11 major earthquakes starting ing loading. In some tests, the joints and contact names. from the 1933 Long Beach earth- were encased in a “sand box” that allowed the soil-pipe interaction and • Several pipeline owners, quake up to the 1994 Northridge including East Bay earthquake. overburden pressures to be Municipal Utility District Iwamatu et al. (1998) and Kitaura included. The author provided fail- (Oakland-Berkeley area), and Miyajima (1996) documented ure criteria in terms of deformations Pacific Gas and Electric failures and the failure rate (per km) for various sizes of pipes. Wang and Company (Central and in the 1995 Kobe earthquake. Li (1994) conducted studies on the Northern California), Sierra These researchers provided a com- damping and stiffness characteris- Pacific Power Co. (Northern prehensive summary of pipeline tics of conventional ductile iron Nevada area), and Los pipe joints subjected to dynamic Angeles Dept. of Water and damage in terms of pipe material Power (Los Angeles Area), type, joint types, and the failure cyclic loading. have provided valuable input and encouragement The documentation and results of this project will be useful to this project. to several different groups. Other researchers involved with • Western Nevada Supply Co. physical testing of pipelines will be able to benefit from the and Intermountain Piping methodologies and procedures that have been developed and Systems have provided used for this project. Manufacturers of pipe and joint restraint items such as pipe systems will have information on the behavior characteristics segments for testing, of their product. It will also benefit those who intend to con- mechanical bolted joint sider using polyethylene pipe to replace conventional pipe restraints, and other miscellaneous supplies. material normally used for water supply. The project results will enable pipeline designers and manufacturers to effectively quantify the merits of relatively new products such as pipeline joint restraints and polyethylene pipe. Furthermore, pipeline owners can use fragility information and risk assessment data developed by this project to determine regions within their service area that may be vulnerable to damage from earthquakes. This will help them with upgrade and retrofitting plans of their piping systems.

56 consisted of apply- Connection Bolts Restraining Collar ing compression Ductile Iron Pipe and/or tension loading to pipe • Calibration of Fragility joint segments us- Estimation and Develop- ing a self-contained ment of Fragility Informa- loading frame that tion, A. Reinhorn, University at Buffalo Bolted Restraining Collar was designed and constructed spe- • Experimental Facilities cifically for this Network, A. Reinhorn and ■ Figure 1. Bolted Restraining Collar testing. More de- M.C. Constantinou, University at Buffalo This paper presents the initial tails on this phase of the testing pipe joint test results for a variety are presented subsequently. of pipe materials including ductile iron, welded steel, cast iron, PVC, Preliminary Compressive and polyethylene. For ductile iron Testing on Ductile Iron Pipe pipe, two different joint restraints were tested: Joints (Phase I) • reinforced gasket In this testing phase, the ends of • bolted restrained collars. the ductile iron bell and spigot pipe joint segments of different diameters Figure 1 shows a sketch of a were milled to achieve smooth end ductile iron pipe provided with surfaces. The specimens were bolted collar restraints. Details on placed in our SATEC compression the testing, configuration, loading testing machine and loaded until procedures, and test results are de- noticeable fracture occurred. The scribed with accompanying plots and graphs. pipe sizes tested were 4”, 6”, 8”, and 10” diameter. The load-displacement values were electronically recorded Overview of Testing and stored using a Megadac data acquisition system. The results of the The testing procedure used for testing are shown in Figure 2. It can this project consisted of developing a method of applying an axial load (compression and/or tension) Ductile Iron Pipe Compression Test to a pipe joint and recording the 400000 pipe barrel strains and the load- 350000 10" DIA displacement relationship. Initially, 300000 8" DIA a series of compressive loading 250000 6" DIA tests (Phase I) were conducted on 200000 4" DIA 150000 ductile iron pipe joint specimens Load (lbs) in order to establish the range of 100000 axial load capacity for various dia- 50000 ) 0 meters of pipe. This test series was 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 conducted using a SATEC compres- Disp (in) University of Nevada, Reno sion testing machine. The second ■ Figure 2. Results of Preliminary Tests on Ductile Iron Bell- phase (Phase II) of this testing Spigot Joints (Phase I)

Axial Behavior Characteristics of Pipe Joints57 be seen that, except in the case of 8 inch diameter pipe, there are measurable amounts of seating distance that had to be over- come before the joints exhibited any resistance to load. As expected, the strength of pipe joints is proportional to the pipe diameter. At the end ■ Figure 3. Failure Mechanism in Ductile Iron Pipe in Phase I of the tests, one Testing specimen was cut in half longitudinally to observe the without the use of reaction failure mechanism (Figure 3). The blocks. The loading and the an- failure mechanism can be de- choring setup were designed to scribed as a telescoping of the readily accept various diameters spigot end into the bell and a sub- of pipe specimens and to as- sequent buckling and fracture of semble them within a reasonable the spigot end. amount of time. Axial load, both in tension and compression, can be applied in incremental dis- Test Setup and Loading placement control by an MTS Configuration (Phase II) 450k hydraulic actuator. Table 1 A self-contained steel loading provides a list of the tests under- frame was designed and fabri- taken using this loading setup. cated (Figures 4 Several pipes of one single pipe and 5) that allows size of 8 inch diameter were tested. Actuator an actuator to Specimen Mounting Plate The information obtained from Flange apply axial com- WF Brace Test Specimen testing included load-displace- Flange pression and/or End Plate -B- tension load to a ment characteristics of the joint test specimen assembly and strains on the pipe WF Brace barrel. Typically, at some level of End Plate -A- loading, noticeable Pipe Test Loading Frame Exploded View fracture and buckling occurred, indicating ■ Figure 4. Specially Designed Self- pipe failure. However, contained Steel Loading Frame failure in a pipeline is governed by the leakage which possibly can occur at some point well before fracture. A generalized criterion for leakage failure may be defined as “substantial ■ Figure 5. Loading Frame Sitting on Laboratory Floor 58 ■ Table 1. Test Matrix Describing Various Pipe Material and Joint Types (Phase II)

lairetaM retemaiD epyTtnioJ stnemmoC “The project

norItsaC derruccoerutcarf;ylnodaolnoisserpmoC results will ni8 dekluacdael,togips-lleB )6erugiFees( llebnissertsidon;lerrabni enable pipeline norIelitcuD decrofnier,togips-lleB ;k521=daol.xam;ylnodaolnoisneT ni8 designers and )7erugiFees( teksag teksagnihteetlatemfoeruliafetamitlu manufacturers norIelitcuD detlob,togips-lleB ;k25=daol.xam;ylnodaolnoisneT ni8 )8erugiFees( rallocgniniartser selohwercsegdewralloctaerutcarf to effectively quantify the niderruccoerutcarf;daollanoitcerid-iB leetS ni8 dedlewpal,togips-lleB ereves;elitcudyrevtniojdlew;lerrab )9erugiFees( merits of llebtagnilkcub relatively new dedurtxetogips;ylnodaolnoisserpmoC CVP no-hsup,togips-lleB ni8 on;deniatniamlaesretaw;dnellebotni products such )01erugiFees( teksagrebbur k3=daol.xam;erutcarf as pipeline deniamertniojdesuf;daollanoitcerid-iB )EP(enelyhteyloP joint restraints ni8 desuf-ttuB erutcarf;epipfognilkcubereves;elitcud )11erugiFees( egnalfdnetaderrucco and polyethylene and continuous leakage.” To de- PVC, and polyethylene pipe can pipe.” tect such leakage failure, the test be interpreted from the plots as setup used completely sealed 460 k, 250 k, 85 k, 3 k, and 68 k, pipe joint specimens that con- respectively. Similarly, the tensile tained water under a small pres- capacity of ductile iron pipe with sure head (3-4 psi). A noticeable reinforced gasket, ductile iron drop in water pressure and an ob- pipe with bolted collars, welded servable amount of water leakage steel, and polyethylene pipes are indicated leakage failure. The wa- 125 k, 52 k, 125 k, and 52 k, re- ter pressure was monitored and spectively. correlated with the load and displacement values to determine load level at leakage. Conclusions/Future Research Load-Displacement Behavior This testing program establishes (Phase II) axial behavior characteristics and leakage failure levels due to axial Figures 6 through 11 provide (tensile and/or compression) the plots of load-displacement static loading for several different data recorded in Phase II testing. types of pipe material and pipe Tension loads were applied only joints. Static testing will continue to joints that were capable of re- for other diameters of pipe. In- sisting tension. These cases information gained, especially fail- cluded two ductile iron pipes ure loads, can be used in the with joint restraints (Figures 7 design of the next phase of our and 8), welded steel pipe (Figure testing which will use shake-table 9), and polyethylene pipe (Figure testing to simulate dynamic 11). The compressive capacity of loading. the cast iron, ductile iron, steel,

Axial Behavior Characteristics of Pipe Joints59 Cast Iron Pipe Ductile Iron Pipe with Reinforced Gasket

com pression tension

150000 500000 leakage 400000 100000 300000 leakage 200000 50000 Load (lbs)

Load (lbs) 100000 0 0 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 Disp (in) Disp (in) University of Nevada, Reno University of Nevada, Reno ■ Figure 6. Cast Iron Pipe with Lead Caulked Joint ■ Figure 7. Ductile Iron Pipe with Reinforced Gasket

Ductile Iron Pipe with Bolted Collar Steel Pipe

com pression tension 100000 60000 50000 40000 0 leakage 20000 no leakage at collar fracture -50000

Load (lbs) 0 -100000 Load (lbs) 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 -150000 tension -4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 Disp (in)

Disp (in) University of Nevada, Reno University of Nevada, Reno ■ Figure 8. Ductile Iron Pipe with Bolted Restraining Collars ■ Figure 9. Steel Pipe with Bell-spigot Lap-welded Joints

PVC Pipe Polyethylene Pipe

com pression com pression 100000 4000 no leakage no leakage 50000 3000 0 2000 -50000 Load (lbs)

Load (lbs) 1000 -100000 0 tension -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 0.00 0.50 1.00 1.50 2.00 2.50 3.00 Disp (in) Disp (in) University of Nevada, Reno University of Nevada, Reno ■ Figure 10. PVC with Push-on Rubber Gasketed Joints ■ Figure 11. Polyethylene (PE) Pipe with Butt-fused Joints

References

Iwamatu, J., et al., (1998), Proceedings of Workshop for Anti-Seismic Measures on Water Supply, IWSA, Tokyo, Japan. Kitaura, M. and Miyajima, M., (1996), “Damage to Water Supply Pipelines,” Special Issue of Soils and Foundation, Japanese Geotechnical Society, January, pp. 325-333.

O’Rourke, T., (1996), “Lessons Learned for Lifeline Engineering from Major Urban Earthquakes,” Proceedings of the Eleventh World Conference on Earthquake Engineering, Pergamon, Elsevier Science Ltd., Oxford, England, Disk 4, Paper No. 2172. O’Rourke, T. and Palmer, M., (1996), “Earthquake Performance of Gas Transmission Pipelines,” Earthquake Spectra, Vol. 12, No. 3, August, pp. 493-527.

Singhal, A., (1984), “Nonlinear Behavior of Ductile Iron Pipeline Joints,” Journal of Technical Topics in Civil Engineering, Vol. 110, No. 1, May, pp. 29-47. Trifunac, M. and Todorovska, M., (1997), “Northridge, California, Earthquake of 1994: Density of Pipe Breaks and Surface Strains,” Soil Dynamics and Earthquake Engineering, Vol. 16, pp. 193-207. Wang, L. and Li, H., (1994), “Experimental Study on the Damping and Resistant Characteristics of Conventional Pipe Joints,” Proceedings of the Fifth U.S. National Conference on Earthquake Engineering, Earthquake Engineering Research Institute, Oakland, California, Vol.II, pp. 837-846. 60 Seismic Performance Analysis of Electric Power Systems

by Masanobu Shinozuka and Tsen-Chung Cheng, University of Southern California; Maria Q. Feng, University of California, Irvine; and Sheng-Taur Mau, New Jersey Institute of Technology

Research Objectives MCEER/NSF The objectives of this research are to evaluate the seismic performance of the Los Angeles Department of Water and Power’s (LADWP’s) electric power system, particularly its substations, and recommend appropriate rehabilitation measures. More specifically, it will provide the analytical and empirical foundation to estimate direct economic losses, as well as indirect economic losses suffered by society at large due to seismically induced degradation of the LADWP’s power system. Indirect losses include commercial and industrial activities in the Los Angeles metropolitan area affected by service interruption of the electric power system. This research will use the results from the inventory survey and equip- Min-Hsiung Chien, formerly ment rehabilitation study being performed concurrently primarily by the of the Los Angeles Depart- ment of Water and Power members of this research team to examine the extent of mitigation en- Nobuo Murota, Bridgestone hancement such rehabilitation work can produce. The analysis requires Company, Japan a somewhat elaborate systems analysis of LADWP’s power system with Donald Penn, Los Angeles primary emphasis on the substation performance under damaging earth- Department of Water and quakes such as the 1971 San Fernando and 1994 Northridge earthquakes. Power While emergency repair and power supply was accomplished rapidly in the aftermath of the 1994 Northridge earthquake (one day) and the 1995 Kobe earthquake (three days), the costs of full restoration of their electric power systems was extremely high. Estimated direct costs were said to be approximately $500 million and $4 billion for the Northridge and the Kobe earthquakes, respectively. Since the “big one” appears to be imminent in California, a much longer and more costly interruption of electric power may have to be anticipated, which could have over- whelming socioeconomic impacts in the affected region. This research will help find rehabilitation measures to mitigate such impacts.

he MCEER research team on system performance evaluation has a Tunique capability of modeling lifeline systems and carrying out a seismic performance evaluation, given inventory data, system configuration and fragility information with or without rehabilitation. The evaluation requires the delicate coordination of various technologies involving interpretation and manipulation of sophisticated and voluminous inventory data, utilization of highly specialized computer codes for systems analysis, estimation of fragility enhancement resulting from the advanced rehabilitation 61 under usual operating conditions in each service area are • LADWP is an important shown in Figures 1 partner in this effort, and 2. The area not providing the MCEER colored is serviced research team with by the Southern inventory and operational California Edison. information of its electric Figure 3 is the North- power system and some ridge PGA map devel- statistics of the damage sustained by the system oped on the basis of from the past earthquakes. the contour map provided by David Wald, • Effort is being made to U.S. Geological Sur- entice Southern California vey and Figure 4 dem- Edison to participate in onstrates how the MCEER’s research as an industry partner. system deteriorates under the ground • A Memorandum of shaking shown in Understanding is being Figure 3 under the prepared between MCEER hypothesis that only and the National Center for the transformers are Research on Earthquake ■ Figure 1. Service Areas of LADWP Engineering (NCREE) in vulnerable to the Taipei, Taiwan to perform a technology and integration of all earthquake with the fragility curves joint experimental study on the above into a GIS platform for assumed in Figure 5. This is based transformers using the demonstration. This capability it- on the observation that the trans- shaking table in Taipei. self represents an advanced tech- former is one of the most critical nology and the purpose of this pieces of equipment for the func- • MCEER investigator research effort is to make use of M. Feng, University of tionality of the power network California, Irvine, is this technology on the Los Angeles system. The effect of other equip- collaborating with Department of Water and Power ment such as circuit breakers, N. Murota, on leave from (LADWP) electric power system. disconnect switches and buses on Bridgestone Company, in The Los Angeles Department of the system performance is currently the planning and prepara- Water and Power electric power being studied. These hypotheses tion of shake table testing. service areas and the power output are introduced to demonstrate the

Primary users of this research include both LADWP’s power division and water division because of potential interactive effects, as well as utility companies throughout the U.S. and else- where. The research results can further be used by government agencies such as the California State Office of Emergency Ser- vices for pre-event and post-event mitigation planning and by private-sector organizations including insurance and financial companies. Another important group of users include researchers in lifeline earthquake engineering, and providers of information for loss estimation databases and methodologies such as HAZUS.

62 proof of concept in relation to the transformers. Analytical simula- analytical simulation work used in tions were performed for a typi- this research. Figure 4 shows the cal transformer weighing 230,000 system deterioration by comput- lbs. subjected to the ground accel- ing the average ratio of power out- eration time histories observed at put relative to that associated with the Sylmar substation during the the system under undamaged con- 1994 Northridge earthquake. To ditions for each service area. The evaluate the effectiveness of the system analysis utilizes a Monte FPS for a wide range of earthquake Carlo simulation method under intensities, time histories were lin- the hypothetical fragility curves early scaled up to achieve higher (Cases 1, 2, and 3) provided in Fig- PGA values for the development ure 5. Fragility curves (Case 1) of fragility information. Figure 7 were used by Tanaka et al. (1996) shows that the degree of reduc- for substation equipment. The tion in the inertia exerted on the sample size is equal to 20 for each transformer depends on the time Monte Carlo simulation analysis. histories with differing levels of The increasingly improved system PGA (0.5 g, 1.0 g, and 1.5 g). The performance as fragility curves trend observed from Figure 7 is move to the right (Case 1 to Case that: (1) the FPS is more effective 2 and to Case 3) indicates the extent to which the rehabilitation or retrofit of transformers as represented by enhanced fragility curves contributes to improved system performance. This is conceptually not an unreasonable approach for evaluation of the effect of the rehabilitation or retrofit. In fact, M. Shinozuka (1998) gave an example (Figure 6) of such fragility curve enhancement involving a typical Memphis bridge retrofitted by a base isolator in which major damage was assumed to occur when the ductility demand at all the bridge columns exceeded 2.0. In the present research, reliable fragility curves for the transformers rehabilitated or not rehabilitated are still in the process of being developed. The FPS (Friction Pendulum System) was considered for enhancing the fragility of the LADWP’s ■ Figure 2. Electric Power Output for Service Areas Under Intact Condition

Seismic Performance Analysis of Electric Power Systems63 for earthquakes with larger PGA’s; LADWP’s Power (2) the reduction of acceleration exerted on the transformer is System more significant when FPS’ radius There are two electric power is larger at the expense of larger networks serving the Los Angeles displacements. Since most trans- region operated by different orga- formers are installed outside and nizations, Los Angeles Department have sufficient clearance with of Water and Power, and Southern neighboring equipment and build- California Edison. Basically, these ings, larger displacements may not networks are managed indepen- represent a serious obstacle in de- dently. However, for coping with ploying FPS devices; and (3) in the fluctuating power demand, general, for a reasonable size of ra- they cooperate with each other at dius (say 15 inches), the reduction several substations and operate ranges from 30% to 50% depend- the system from a regional point ing on the earthquake intensity be- of view. In addition, since the net- tween 0.5 g to 1.5 g in terms of works are a part of the very large PGA. This result was generally con- Western Systems Coordinating sistent with hypothetical fragility Council’s (WSCC) power transmis- curve enhancement introduced in sion network covering 14 western Figure 5. states, two Canadian provinces and northern Baja California, the analysis was performed by taking all the substations and transmission facilities covered by the WSCC network into account. In- deed, the fact that a black-out condition was observed over several states after the Northridge earthquake demonstrates the far-reaching impact of a local system failure throughout the network. In analyzing the functional reliability of each substation, the following modes of failure were taken into consideration: (1) loss of connectivity, (2) failure of the substation’s critical components, and (3) power system imbalance. It was noted that most of the transmission lines of the LADWP’s power system are aerial supported by transmission towers. While by no means this implies that the transmission lines are completely free from seismic vulnerability, it was assumed in this study that ■ Figure 3. PGA Under Northridge Earthquake

64 Monte Carlo Simulation Using the ARC/ • Seismic Reliability Analysis INFO GIS capabil- for Southern California ity, the electric Power Systems, T.C. Cheng, transmission net- University of Southern California work map was overlaid with the • Rehabilitation Strategies for PGA map (Figure Lifelines: LADWP Water 3) to identify the Systems, T.D. O’Rourke, PGA value associ- Cornell University ated with each • Seismic Retrofit Methods for substation under LADWP Power Systems, the Northridge S.T. Mau, New Jersey earthquake. The Institute of Technology and fragility curves as- M. Feng, University of sumed in Figure 5 California, Irvine were then used to • Socioeconomic Impacts of simulate the state Lifeline Systems, S. Chang, of damage involv- University of Washington ing the transformers at all the ■ Figure 4. Relative Average Power Output (Damaged/Intact) with substations of the Transformers Vulnerable LADWP’s power system. For each they were, primarily for the pur- systems analysis, the connectivity pose of analytical simplicity. Fig- and power flow were examined ure 8 shows an abbreviated system with the aid of IPFLOW, where flowchart for LADWP’s power sys- LADWP’s power system was tem with all the substations iden- treated as a part of WSCC’s overall tified together with the nodes, system. generators and transformers. Thick horizontal bars represent the nodes (buses with all other as- 1.0 sociated equipment) in substa- Case 1 (Xm=0.45g) 0.8 Case 2 (Xm=0.70g) tions as described by a model Case 3 (Xm=0.90g) shown in Figure 9. In the systems 0.6 analysis pursued here, however, substation data were taken from 0.4 the WSCC’s database and used for Probability of Failure 0.2 the systems analysis in conjunc- 0.0 tion with the computer code 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 IPFLOW, (version 5.0), licensed by the Electric Power Research Insti- PGA(g) tute (EPRI) to the University of ■ Southern California. Figure 5. Fragility Curves

Seismic Performance Analysis of Electric Power Systems65 Loss of connectivity occurs when the node of interest sur- Not Isolated (dd=2) “The research Isolated (dd=2) results can be vives the corresponding PGA, used by but is isolated from all the gen- government erators due to the malfunction agencies such as of at least one of the nodes on each and every possible path be- the California tween this node and any of the State Office of generators. Hence, the loss of Emergency connectivity can be confirmed ■ Figure 6. Fragility Curves for Bridges with and Services for pre- on each damage state by actually without Base Isolation and post-event verifying the loss of mitigation connectivity with re-

70 planning and spect to all the paths Number of Bearings, N = 12

by private-sector that would otherwise 60

organizations establish the desired 50 including connectivity. 40 insurance and As for abnormal power flow, it was 30 financial Input Acceleration 0.5g 1.0g noted that the elec- 20 companies.” Inertia Reduction (%) 1.5g tric power transmis- 10 sion system was

0 highly sensitive to the 5 101520253035 power balance and Radius (inch) ordinarily some crite- ■ Figure 7. Acceleration Reduction by FPS Bearings ria are used to judge whether or not the node continues to function imme- the operator of the electric system diately after internal and external must either reduce or increase the disturbances. Two kinds of crite- power generation to keep the bal- ria are employed at each node for ance of power. However, in some the abnormal power flow: power cases, the supply cannot catch up imbalance and abnormal voltage. with the demand because the gen- When the network is damaged due erating system is unable to re- to an earthquake, the total gener- spond quickly enough. In this case, ating power becomes greater or it was assumed that the power less than the total power demand. generation of each power plant Under normal conditions, the bal- cannot be increased or reduced by ance between power generation more than 20% of the current gen- and demand is within a certain erating power. When the power range of tolerance. Actually, the balance cannot be maintained total power generation must be even after increasing or reducing between 1.0 and 1.05 times the to- the generating power by 20%, the tal demand for normal operation system was assumed to be down even accounting for power trans- due to a power imbalance. In this mission loss. respect, the effect of the emer- In this study, it was assumed that gency management systems used if this condition was not satisfied, for power flow management will

66 be incorporated in the systems substations that will become analysis in the future study. inoperational. As to the abnormal voltage, volt- The simulation was repeated 20 age magnitude at each node can times on the network. Each simu- be obtained by power flow analy- lation provided a different dam- sis. Then, if the ratio of the volt- aged network condition. Figure 4 age of the damaged system to the shows the ratio of the average out- intact system is out of a tolerable put power of the damaged net- range (plus/minus 20% of the volt- work to that associated with the age in the intact system), it was as- intact network for each service sumed that a blackout will occur area. The average was taken over in the area served by the substa- the entire sample size equal to 20. tion. It was concluded from Figure 4 For the Monte Carlo simulation that the rehabilitation that lead to of system performance under the the fragility curve labeled as Case Northridge earthquake, each sub- 2 was good enough to protect the station was examined with re- transformers, and hence the entire spect to its possible malfunction power system, very well under the under these three modes of failure assumption that structures and for each simulated damage state. other equipment were not vulner- Thus, each simulation identifies the able to earthquake ground motion.

10004 10005 10006 20493 Sylmar 1R 20494 Sylmar 2R

10094 Sylmar LA

10010 Castaic Olive 10052 Sta. J 10086 Rinaldi 10061 Rinaldi 10062 Victorvl 10105 251 MW 132MW 10093 Sta. U 10055 Owens 26.4MVR 10115 Rinaldi 2 10003 Adelanto 50.2MVR 400MW 222MW 222MW 10087 Sta. K 10092 Sta. U 10103Valley 10079 Sta. E 38.6MVR 40.1MVR 40.1MVR 10102Valley 160MW 222MW 10088 Sta. KLD 10078 Sta. E 10066 Scatergd 32MVR 40.1MVR 10065 611MW 10067 Scatt 3G 10026 Scatergd 122.2MVR 10081 Sta. G 10025 Haynes 10027 10028 10077 Sta. E 10012 Glendal 10029 10030 10031 10089 Sta. N 10082 Sta. H 10068 Sta. A Scat 12G 190MW 400MW 163MW 166MW 193MW 10064 38MVR 38. 6MVR 10085 Sta. HDL 32.2MVR 10063 River 38.6MVR 33.2MVR 255MW 222MW 307MW 298MW 10083 Sta. H1 10084 Sta. H2 10015 GramerC2 51MVR 10080 Sta. F 40.1MVR 63.7MVR 68.2MVR 181MW 10076 Sta. D 10014 GramerC1 36.2MVR 10072 Sta. BLD 222MW 365 MW 10069 Sta. B 44.4MVR 10070 Sta. B1 73 MVR

Generator 10095TAP1 10016 Halldale 10096TAP2 10071 Sta. B2 10104Victorvl Node 41MW 8.2MVR 10073 Sta. C1 10075 Sta. CLD 10074 Sta. C2 Transformer 139MW 37.8MVR 10019 Harb

10091 Sta. Q

10090 Sta. Q 10023 Harb 5G 216MW 43.2MVR

■ Figure 8. LADWP’s System Flowchart

Seismic Performance Analysis of Electric Power Systems67 In cooperation with Professor Bus 1 Bus 2 C.H. Loh, Director of the National Center for Research on Earthquake Line A Line B Engineering (NCREE) in Taipei, Tai- Position 1 CB11 CB12 CB13 wan, the MCEER team with Bridgestone Company as an industrial partner, is participating in an Line C Line D experiment to verify the ef- Position 2 CB21 CB22 CB23 fectiveness of FPS and hybrid friction and elastometric base-isolators designed, manufactured and tested by Bridgestone. The experiment will be performed in July Equipment: Disconnect Switch 1999 on a transformer model in- Circuit Breaker stalled with a typical porcelain Bus bushing. For the experiment, NCREE’s 5 m x 5 m triaxial shak- ■ Figure 9. Typical Node Configuration Model ing table will be used. On-going Research Activities Future Research Now that all the analytical tools Figure 4 is based on the hypoth- are in place, in the year immedi- esis that transformers are the only ately following and beyond, the vulnerable equipment under the research will continue to proceed earthquake and that their fragility on three fronts. The first is to curves are given by the three refine the systems analysis meth- curves in Figure 5. Other equip- odology by incorporating all sig- ment such as circuit breakers, nificant substation equipment and buses, and disconnect switches are validating the results of the analy- currently being incorporated into sis with data from power interrup- the systems analysis. tion experiences caused by the In order to examine the ad- Northridge and Kobe earthquakes. equacy of the fragility assumptions Upon validation, other scenario for transformers and other equip- earthquakes will be considered for ment, a walk-down at LADWP’s the systems analysis to examine Sylmar substation is scheduled on the seismic performance of June 11, 1999 by S.T. Mau (and/or LADWP’s power system under a Professor M.A. Saadeghvaziri) of the New Jersey Institute of Tech- wide variety of earthquake mag- nology, M. Shinozuka (and/or Mr. nitudes, epicentral locations and X. Dong and Professor F. seismic source mechanisms. Inter- Nagashima) and T.C. Cheng (and/ action between LADWP’s power or Mr. X. Jin) of the University of and water systems will also be Southern California, and Professor considered. This requires, how- M. Feng (and/or Mr. N. Murota) of ever, additional effort for inventory the University of California, Irvine. and other database development.

68 The second is to further study tem facilities and resulting pos- seismic vulnerability of the equip- sible system interruption. This ment and develop their fragility endeavor expands the MCEER curves. In this regard, the results team’s capability in this area dem- from the research carried out by onstrated by the study of the seis- the MCEER investigators on fragil- mic vulnerability of the Memphis ity information will be used as area’s electricity lifelines (see they become available. Rehabili- Shinozuka et al., 1998). To assist tation measures can then be the MCEER investigators in loss es- expressed as fragility curve en- timation, the Monte Carlo simula- hancements, which can in turn be tion will be performed in such a directly reflected on the systems way that direct and indirect loss analysis with the aid of Monte estimation will be pursued by re- Carlo techniques. In this connec- cording a specific inventory of tion, rehabilitation measures other equipment damage observed for than those by base isolation will each realization of system damage. be explored. Possibilities include As detailed in Shinozuka and use of advanced semi-active damp- Eguchi (1997), this allows statistics ers. The shaking table tests for trans- on direct and indirect losses based formers will be completed and the on individual states of damage as- continued MCEER-NCREE collabo- sociated with corresponding simu- ration will lead to additional tests lation to be obtained, rather than involving other equipment to de- based on the average of the power termine their fragility characteris- output taken over the entire tics and enhancement measures. sample of simulation. The third area of future endeavor involves direct and indirect economic loss estimation arising from physical damage to the sys-

References Shinozuka, M. and Eguchi, R., (1997), “Seismic Risk Analysis of Liquid Fuel Systems A Conceptual and Procedural Framework for Guidelines Development,” Proceedings of The Northridge Earthquake Research Conference, Los Angeles, California, August 20-22. Shinozuka, M., (1998), “Development of Bridge Fragility Curves,” Proceedings of the U.S.- Italy Workshop on Protective System for Bridges, New York, April 26-28. Shinozuka, M., Rose, A. and Eguchi, R.T., (eds.), (1998), Engineering and Socioeconomic Impacts of Earthquakes: An Analysis of Electricity Lifeline Disruptions in the New Madrid Area, Monograph No. 2, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York. Tanaka, S., Shinozuka, M., Schiff, A. and Kawata, Y., (1997), “Lifeline Seismic Performance of Electric Power Systems during the Northridge Earthquake,” Proceedings of The Northridge Earthquake Research Conference, Los Angeles, California, August 20-22.

Seismic Performance Analysis of Electric Power Systems69

National Representation of Seismic Hazard and Ground Motion for Highway Facilities

by Maurice S. Power and Shyh-Jeng Chiou, Geomatrix Consultants, Inc., and Ronald L. Mayes, Dynamic Isolation Systems, Inc. for the Applied Technology Council

Research Objectives MCEER Highway Project The objective of this research is to develop recommendations for National Cooperative Highway the national representation of seismic hazard and ground motion for Research Program the seismic design of highway bridges and other highway structures. Transportation Research This includes: (1) the selection and utilization of national ground Board motion maps; (2) the representation of site response effects; and (3) the possible incorporation of other parameters and effects, including energy or duration of ground motions, vertical ground motions, near- source horizontal ground motions, and spatial variations of ground motions.

Applied Technology Council CEER, through the Federal Highway Administration (FHWA) spon- Msored Highway Project, is conducting research to develop improved criteria and procedures for the seismic design of new highway facilities and the seismic evaluation and retrofit of existing highway facilities. An important area of research is new directions for representing seismic hazard and ground motions in nationally applicable guidelines and specifications, including the AASHTO seismic design specifications for bridges and the FHWA/ MCEER seismic evaluation and retrofitting manual for bridges and other highway structures. Some of the key seismic ground motion representation issues are: (1) • This project is reviewing selection and utilization of national ground motion maps; (2) representa- and assessing the results tion of site response effects; and (3) incorporation of other parameters from research currently in and effects, including ground motion energy or duration, vertical ground progress or recently motions, near-source horizontal ground motions, and spatial variations of completed in other MCEER ground motions. These issues were discussed in some detail at the MCEER projects, U.S. Geological Workshop on National Representation of Seismic Ground Motion for Survey, Caltrans, the Federal New and Existing Highway Facilities, May 29 and 30, 1997, in San Fran- Highway Administration, and others for its possible cisco (Friedland et al., 1997). incorporation into the In addition, in August 1998, the ATC/MCEER Joint Venture, a partnership recommendations that are of the Applied Technology Council (ATC) and MCEER, was awarded a con- being developed. tract by the AASHTO-sponsored National Cooperative Highway Research Program (NCHRP) of the Transportation Research Board, National Acad- emy of Sciences, to develop a comprehensive new specification for the seismic design of bridges (NCHRP Project 12-49). The research and 71 development effort being under- vibration of 0.2, 0.3, and 1.0 second taken on this project is the devel- and peak ground acceleration for opment of new Load and Resistance probabilities of exceedance of 10%, • Compile and Evaluate Maps Factor Design (LRFD) Specifications 5%, and 2% in 50 years (correspond- and Other Representations and Commentary for the seismic ing approximately to return periods and Recommend Alterna- design of bridges. Issues being ad- of 500, 1000, and 2500 years, re- tive Strategies for Portray- dressed include: (1) design philoso- spectively). A key question is ing the National Hazard phy and performance criteria; (2) Exposure of Highway whether the new USGS maps pro- Systems, M. Power, seismic loads and site effects; (3) vide a substantially improved scien- Geomatrix Consultants, analysis and modeling; and (4) de- tific national representation of Inc. sign and detailing requirements. ground motion. The MCEER High- The new specification must be na- way Project and the May 1997 • Development of a Unified tionally applicable with provisions Model of Spatial Variation, MCEER Workshop concluded based for all seismic zones. The results of M. Shinozuka, University on an evaluation of (1) the process research currently in progress or re- of Southern California, and by which the maps were prepared, cently completed by MCEER, G. Deodatis, Princeton (2) characterizations of seismic Caltrans, and the FHWA are the prin- University sources and ground motion attenu- cipal resources for this project. • Northridge SR14/I5 Case ation used in the mapping, and (3) Study: Structural Implica- comparison of the map values with tions, A. Reinhorn, Selecting & Using results from detailed site-specific University at Buffalo National Ground studies, that the new USGS maps • Two-Dimensional provided a major improvement in Nonlinear Site Response Motion Maps the representation of seismic Analysis, G. Martin, One of the most important issues ground motion at a national scale University of Southern and therefore should provide the California addressed by the MCEER Highway Project is whether new (1996) na- basis for a new national seismic • Evaluation of Site tional ground motion maps devel- hazard portrayal for highway facili- Coefficients Based on oped by the U.S. Geological Survey ties. The NCHRP project intends to Northridge and Kobe Data, (USGS) (Frankel et al., 1996) should use these maps as the basis for de- R. Dobry, Rensselaer fining seismic loads for the new Polytechnic Institute be recommended to provide a national representation of seismic seismic design provisions for • Vertical Seismic Ground ground motion for bridges and high- bridges. Similarly, the maps will be Motions, M. Power, ways. The new set of maps contour used as a basis for defining seismic Geomatrix Consultants, elastic 5%-damped response spec- loads for the FHWA/MCEER seismic Inc. tral accelerations for periods of retrofitting manual (see Seismic • Site Response Effects, R. Dobry, Rensselaer Polytechnic Institute The recommendations will be incorporated in a new set of seismic design provisions that will be considered for adoption by the • Effects of Vertical Accelera- American Association of State Highway Transportation Officials tion on Structural (AASHTO). If and when they are adopted, they will be used by all Response, M. Button, Dynamic Isolation Systems Federal, State, and local transportation agencies and departments, and private organizations involved in the design and operation of new highway bridges. The recommendations will also be incorporated in a set of retrofitting manuals for existing bridges and other highway structures being developed by MCEER (see Seismic Ret- rofitting Manual for Highway Systems in this report). Users of these manuals will be similar to that of the AASHTO seismic design specifications for new bridges.

72 0.2 sec Spectral Accel. (%g) with 2% Probability of Exceedance in 50 Years of 50% during the site: NEHRP B-C boundary 50û -120û design life of a -70û 50û -110û -80û -100û -90û bridge, taken as

20 16 12 20 75 years; and (2) 8

40 40û 40 120 20 40û design for col- 12 20 120 16 lapse prevention 16 16 12 16 16 160 20

40 for rare or maxi- 40 12

160 160 40 40 120 8 30û 40 mum-earthquake 20 30û

120 20 40 16 ground motions, Nov. 1996 12 12 20 8 defined as ground 12 -120û 16 4 -70û motions having a -110û -80û -100û -90û probability of U.S. Geological Survey National Seismic Hazard Mapping Project exceedance of 3% ■ Figure 1. U.S. Geological Survey 0.2 second 2,500 Year Spectral in 75 years. These Acceleration Map probability levels correspond, re- Retrofitting Manual for Highway spectively, to return periods of ap- Systems in this report). proximately 100 years and 2,500 A key issue in the utilization of years. No recommendations regard- the new USGS maps is the choice ing probability levels or return peri- of an appropriate probability of ods have yet been developed for exceedance or return period for seismic retrofit design for existing seismic design provisions i.e., whether to continue with the 10% probability of exceedance in 50 10 10 years in existing AASHTO provi- Site 82 CAL, 37.5 -122.3 Site 49 WUS, 45.0 -112.5 sions or recommend another probability level. The May 1997 MCEER 1 1

Workshop recommended that for ( ( Sa (g) a prevention-of-collapse perfor- Sa (g) 0.1 California, 0.1 WUS, mance criterion, probability of High Seismicity Moderate Seismicity exceedance levels lower than 10% 2% PE in 50 Yr 2% PE in 50 Yr 10% PE in 50 Yr 10% PE in 50 Yr in 50 years should be considered 0.01 0.01 in design. This recommended di- 0.01 0.1 1 10 0.01 0.1 1 10 rection is consistent with revisions Period (Sec) Period (Sec) to the 1997 National Earthquake Hazards Reduction Program Site 27 CE US, 37.5 -90.0 Site 17 CE US, 42.5 -85.0 (NEHRP) Provisions for new build- 1 1 ings (BSSC, 1997a) in which the new USGS maps for a probability (

0.1 ( 0.1 Sa (g) of exceedance of 2% in 50 years Sa (g) have been selected as a collapse CE US, CE US, Moderate Seismicity Low Seismicty prevention design basis for build- 0.01 2% PE in 50 Yr 0.01 2% PE in 50 Yr ings (see Figure 1). The NCHRP 12- 10% PE in 50 Yr 10% PE in 50 Yr 49 project recommended for the 0.01 0.1 1 10 0.01 0.1 1 10 seismic design of new bridges: (1) Period (Sec) Period (Sec) elastic design for expected ground ■ Figure 2. Comparison of Simplified Response Spectrum Construction motions, defined as ground motions Procedure in 1997 NEHRP Provisions with Spectral Accelerations having a probability of exceedance Computed by USGS for National Ground Motion Mapping

National Representation of Seismic Hazard and Ground Motion73 bridges. If an approach 3 “Site response were taken similar to that in the NEHRP guidelines Soft Sites effects for code NEHRP 1994 2 provisions are for existing buildings Stiff Sites

(BSSC, 1997b), then, for a NEHRP best F planned retrofits, a bridge 1994 characterized 1 owner could select from Stiff and Soft Sites at present by AASHTO 1996 among a range of ground 0 the site motion probability levels Short-Period Site Coefficient 0 0.1 0.2 0.3 0.4 0.5 categories and and retrofitting perfor- Rock Shaking, Aa or A (g) site factors mance objectives, de- developed at pending on economic 4 the 1992 Site factors and the owner’s Soft Sites objectives. NEHRP 1994 Effects Work- 3 In the 1997 NEHRP Stiff Sites shop and NEHRP Soft Deep Sites provisions for buildings, a 1994 AASHTO 1996

subsequently v simple procedure was F 2 adopted into Soft Sites proposed for construct- the 1994 and AASHTO ing a response spectrum 1 1996 1997 NEHRP using spectral acceler- Stiff Sites

Long-Period Site Coefficient AASHTO 1996 Provisions and ations from national 0 the 1997 ground motion maps (see 0 0.1 0.2 0.3 0.4 0.5 Uniform Figure 2). The procedure Rock Shaking, Av or A (g) Building Code.” consists of intersecting a ■ Figure 3. Comparison of Site Coefficients Contained in 1997 constant-spectral-accel- NEHRP, 1997 Uniform Building Code and 1996 AASHTO eration plateau defined Specifications for Highway Bridges by the mapped spectral acceleration at 0.2-second period spectrum is generally conservative. with a long period branch that de- Overall, these spectral comparisons clines as 1/T (T = period) and passes indicate that the simplified re- through the mapped spectral accel- sponse spectrum construction pro- eration at 1.0-second period. This cedure is adequate. The decline of procedure has been evaluated by spectral accelerations as 1/T in the comparing the resulting spectra long-period range is a better char- with spectral accelerations com- acterization of long-period ground puted by the USGS at a larger num- motions than the 1/T2/3 decline in- ber of periods (0.1, 0.2, 0.3, 0.5, 1.0, corporated in the current AASHTO 2.0, and 4.0 seconds) for different provisions. tectonic environments and geographic locations throughout the Representing Site United States. The proposed NEHRP simplified response spectrum con- Response Effects struction procedure generally is in The current AASHTO provisions good agreement with the computed characterize site response effects spectral accelerations in the period on ground motion using the S1 range of 0.2 to 2 seconds. At short through S3 site categories and cor- periods (0.1 second) and longer responding site factors that were periods (4 seconds), the simplified originally developed in the ATC-3

74 Project (ATC, 1978), and the S4 cat- should be considered as further re- egory which was added after the search on site effects is completed. 1985 Mexico City earthquake. The The Workshop therefore recom- May 1997 MCEER Workshop con- mended that these site factors be sidered whether these site factors proposed as part of a new national and site categories should be re- representation of seismic ground placed by the site categories and motion for highway facilities de- site factors developed at a Site Ef- sign. It was also concluded that fects Workshop sponsored by these site factors should not be MCEER, SEAOC, and BSSC and held coupled with the conservative in 1992 at the University of South- long-period response spectral char- ern California (USC) (Martin and acterization in the current AASHTO Dobry, 1994), and subsequently provisions (spectral values decay- adopted in the 1997 Uniform Build- ing as 1/T2/3). Rather, long-period ing Code and 1994 and 1997 ground motions should be permit- NEHRP Provisions. ted to decay in a more natural fash- Since the 1992 Workshop, two ion (approximately as 1/T), as significant earthquakes occurred, discussed above (see Figure 4). the 1994 Northridge and the 1995 Kobe earthquakes. These earthquakes provided substantial addi- Incorporating Energy tional data for evaluating site and Duration of effects on ground motions, and research using these data has been Ground Motions conducted. Dobry et al. (1997, At present, the energy or duration 1998) and Borcherdt (1996, 1997) of ground motions is not explicitly presented results indicating that recognized in the design process the new site factors developed at for bridges or buildings, yet many the USC Workshop are in fairly engineers are of the opinion that good agreement with data from the Northridge earthquake for stiff soil sites, and Borcherdt (1996, To = 0.2Ts 1997) found similar agreement for Fa x Sar(0.2) Fv x Sar(1.0) Ts = Note: Sar is spectral soft soil site data from the Kobe Fa x Sar(0.2) Sar(0.2) acceleration on earthquake. Borcherdt also found rock Type B that the effects of soil nonlinearity (reducing site amplifications with increasing levels of excitation) Fv x Sar(1.0) might be somewhat smaller than T those incorporated in the new Sar(1.0) Soil response site factors for stiff soil sites, based spectrum Sa=0.4Fa x Sar(0.2) Sar(1.0) on preliminary analyses of North- T ridge earthquake data (see Figure Acceleration, Sa Spectral 3). The May 1997 MCEER Workshop Rock (Type B) concluded that, overall, the post- response spectrum Northridge and post-Kobe earth- 0 To 0.2 Ts 1 234 quake research conducted to date Period, T (sec) supported the new site factors, although revisions to these factors ■ Figure 4. Proposed Rock and Soil Response Spectra Definition in 1997 NEHRP Provisions

National Representation of Seismic Hazard and Ground Motion75 the performance of a structure may (e.g., Silva, 1997). Our current be importantly affected by these pa- understanding and ability to rameters in addition to the response characterize near-source vertical spectral characteristics of ground ground motions is good, especially motion. On the basis of papers and for the Western United States. presentations at the May 1997 Furthermore, analyses by Foutch MCEER Workshop by Cornell (1997) and Gloyd (1997) have (1997), Krawinkler (1997), and demonstrated that high vertical Mander and Dutta (1997), it was accelerations, as may be expected concluded that some measure of in the near-source region, can the energy of ground motions is significantly impact bridge re- important to the response of a sponse and design requirements in bridge but, currently, there is no some cases. On the basis of these accepted design procedure to ac- findings, the May 1997 MCEER count for energy. Research in this Workshop concluded that vertical area should be continued in order ground motions should be con- to develop energy-based design sidered in the design of some types methods that can supplement cur- of bridges in the near-source region rent elastic-response-spectrum- and that specific design criteria and based design methods. The MCEER procedures should be formulated. Workshop also concluded that en- The NCHRP 12-49 project is ergy, rather than duration, of ground addressing this issue based on more motions is the fundamental param- recent MCEER-sponsored work eter affecting structural behavior. (Button et al., 1999).

Incorporating Vertical Representing Near- Ground Motion Source Horizontal At present, the AASHTO Ground Motions specifications for bridges do not It is well established that, in ad- contain explicit requirements to dition to the increasing amplitude design for vertical accelerations. of ground motions (in terms of Ground motion data from many peak ground acceleration, spectral earthquakes in the past 20 years acceleration, etc.) with decreasing have shown that, in the near-source distance to earthquake sources, region, very high short-period near-source ground motions have vertical response spectral accel- certain characteristics that are not erations can occur. For near-source found at greater distances. For moderate- to large-magnitude horizontal ground motions, the earthquakes, the rule-of-thumb ratio most significant characteristic ap- of two-thirds between vertical and pears to be a large pulse of inter- horizontal spectra is a poor mediate- to long-period ground descriptor of vertical ground motions when an earthquake rup- motions; at short periods, the ture propagates toward a site. Fur- vertical-to-horizontal spectral ratios thermore, this pulse is larger in the can substantially exceed unity, direction perpendicular to the whereas at long periods, a ratio of strike of the fault than in the di- two-thirds may be conservative rection parallel to the strike (e.g.,

76 Somerville et al., 1997, Somerville, account in relatively sophisticated 1997). This characteristic of near- site-specific analyses (e.g., Power et “The MCEER source ground motions has been al., 1993). However, questions re- May 1997 observed in many earthquakes, in- main as to the classes of bridges Workshop (e.g., related to bridge span length, cluding most recently in the concluded that Northridge and Kobe earthquakes. overall bridge length, and other the bridge Preliminary analyses of bridge re- characteristics) for which spatial sponse by Mayes and Shaw (1997) variations of ground motion may categories and indicate that near-source ground safely be neglected in design or the conditions for motions may impose unusually effects of these variations incorpo- which spatial large displacement demands on rated using simplified code-type variations of bridges. design procedures. ground At the May 1997 MCEER Workshop, Limited studies on effects of spa- motions can be it was concluded that traditional tial variations of ground motions on neglected is ground motion characterizations bridge response (Shinozuka and not yet well Deodatis, 1997; Simeonov et al., (i.e., response spectra) may not be defined.” adequate in the describing near- 1997; and Der Kiureghian and source ground motions, because Keshishian, 1997) indicate that, in the pulsive character of these mo- general, in the absence of strong dif- tions may be more damaging to ferential site response effects, the bridges than indicated by the re- response of “ordinary” highway sponse spectra of the motions. It bridges is not greatly affected by was recommended that additional spatial variations of ground motions. research be carried out to evaluate However, the MCEER May 1997 more fully the effects of near-source Workshop concluded that the ground motions on bridge response bridge categories and conditions for and to incorporate these effects in which spatial variations of ground motions can be neglected is not yet code design procedures. Until ad- well defined, even for the case of equate procedures are developed, relatively uniform soil conditions consideration should be given to along a bridge. In a more recent evaluating bridge response using study conducted under the MCEER site-specific analyses with represen- Highway Project (Shinozuka et al., tative near-source acceleration time 1999), it is concluded that, for histories. bridges that are more than 1,000 to 1,500 feet (300 to 450 meters) in Spatial Variations of overall length or for bridges of any length with foundations supported Ground Motion on different local soil conditions, Spatial variations of ground mo- there can be significant increases tions along an extended structure in peak relative forces and displace- such as a bridge include spatial ments. The study therefore recom- incoherency in ground motions, mends that time history dynamic wave passage effects, attenuation analyses be performed for the de- effects, and differential site re- sign of bridges under such condi- sponse. For major long-span tions. However, further research is bridges, procedures are available needed to broaden these conclu- and have been employed in some sions and recommendations to a cases for taking these effects into wider variety of bridge types and

National Representation of Seismic Hazard and Ground Motion77 construction, and to develop sim- effects for code provisions are best plified procedures for incorporating characterized at present by the site the effects of these variations in categories and site factors devel- design. oped at the 1992 Site Effects Work- shop and subsequently adopted into the 1994 and 1997 NEHRP Pro- Conclusions visions and the 1997 Uniform Build- The principle current conclu- ing Code; (5) code provisions sions regarding the national repre- should be formulated for design of sentation of seismic hazard and some types of bridges for vertical ground motions for highway facili- ground motions in the near-source ties design are the following: (1) regions; (6) response spectra may new (1996) USGS national ground not adequately characterize the motion maps provide a substantially pulsive character of near-source more accurate representation of horizontal ground motions and fur- seismic ground motion than earlier ther work is needed to more fully maps and are recommended to evaluate the effects on bridge re- provide a basis for a new national sponse and develop appropriate seismic hazard portrayal for high- analytical procedures; (7) energy- way facilities design; (2) for col- based design methods appear to be lapse prevention design, probability promising for supplementing elastic- levels lower than the 10% probabil- response spectrum-based design but ity of exceedance in 50 years (i.e., further work is needed to develop return periods longer than approxi- accepted formulations; and (8) in the mately 500 years) currently in absence of strong-differential site AASHTO should be considered for response, the effect of spatial varia- highway facilities design; (3) simpli- tions of ground motion on “ordinary” fied NEHRP procedures for re- (i.e., not long-span) bridges appears sponse spectrum construction to be small, but the matrix of bridge appear to be reasonable for per- categories and conditions for which iods of vibration equal to or less these effects can be safely neglected than 4 seconds; (4) site response is not sufficiently defined.

References Borcherdt, R.D., (1996), “Preliminary Amplification Estimates Inferred from Strong Ground Motion Recordings of the Northridge Earthquake of January 17, 1995,” Proceedings of the International Workshop on Site Response I, Vol. 2, Yokosuka, Japan, Port and Harbor Research Institute, Japan. Borcherdt, R.D., (1997), “Estimates of Site-Dependent Response Spectra for New and Existing Highway Facilities (Methodology and Justification),” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97- 0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 171-201.

78 BSSC, (1997a), NEHRP Recommended Provisions for Seismic Regulations for New Buildings: Part 1 Provisions; Part II Commentary, Report FEMA 222A and 223A, Building Seismic Safety Council, Washington, D.C.

BSSC, (1997b), NEHRP Guidelines for the Seismic Rehabilitation of Buildings, Report FEMA-273, and NEHRP Commentary for the Seismic Rehabilitation of Buildings, Report FEMA-274, Building Seismic Safety Council, Washington, D.C.

Button, M., Cronin, C. and Mayes, R., (1999), Effect of Vertical Ground Motions on the Structural Response of Highway Bridges, Technical Report MCEER-99-0007, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Cornell, C.A., (1997), “Does Duration Really Matter?,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97- 0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 125-133. DerKiureghian, A. and Keshishian, P., (1997), “Effects of Incoherence, Wave Passage, and Spatially Varying Site Conditions on Bridge Response,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97- 0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 393-407. Dobry, R., Ramos, R. and Power, M.S., (1997), “Site Factors and Site Categories in Seismic Codes: A Perspective,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 137-169.

Dobry, R., Ramos, R. and Power, M.S., (1999), Site Factors and Site Categories in Seismic Codes, Technical Report MCEER-99-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York. Foutch, D.A., (1997), “A Note on the Occurrence and Effects of Vertical Earthquake Ground Motion,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 253-275. Frankel, A., Mueller, C., Barnhard, T., Perkins, D., Leyendecker, E.V., Dickman, N., Hanson, S. and Hopper, M., (1996), National Seismic Hazard Maps Documentation, June 1996, U.S. Geological Survey Open-File Report 96-532.

Friedland, I.M., Power, M.S. and Mayes, R.L., eds., (1997), Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97- 0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Gloyd, S., (1997), “Design of Ordinary Bridges for Vertical Seismic Acceleration,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 277-290.

Krawinkler, H., (1997), “Impact of Duration/Energy in Design,” Proceedings of the FHWA/ NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 115-123.

National Representation of Seismic Hazard and Ground Motion79 Mander, J.B. and Dutta, A., (1997), “How Can Energy Based Seismic Design be Accommodated in Seismic Mapping,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 95-114. Martin, G.R. and Dobry, R., (1994), “Earthquake Site Response and Seismic Code Provisions,” NCEER Bulletin, Vol. 8, No. 4, pp. 1-6; and Research Accomplishments: 1986-1994, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 121-120. Mayes, R.L. and Shaw, A., (1997), “The Effects of Near Fault Ground Motions on Bridge Columns,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 319-340. Power, M.S., Youngs, R.R., Chang, C.Y., Sadigh, R., Coppersmith, K.J., Taylor, C.L., Penzien, J., Wen, S.T., Abrahamson, N. and Gates, J.H., (1993), “Development of Seismic Ground Motions for San Francisco Bay Bridges.” Proceedings of the First U.S. Seminar, Seismic Evaluation, and Retrofit of Steel Bridges,” San Francisco, October 18, University of California at Berkeley, and California Department of Transportation, Division of Structures. Shinozuka, M. and Deodatis, G., (1997), “Effect of Spatial Variation of Ground Motion on Seismic Response of Bridges,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 343-358.

Shinozuka, M., Saxena, V. and Deodatis, G., (1999), Effect of Spatial Variation of Ground Motion on Highway Structures, Highway Project draft final task report, Task 106-E-2.2/ 2.5, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, April. Silva, W., (1997), “Characteristics of Vertical Strong Ground Motions for Applications to Engineering Design,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 205-252. Simeonov, V.K., Mylonakis, G., Reinhorn, A.M. and Buckle, I.G., (1997), “Implications of Spatial Variation of Ground Motion on the Seismic Response of Bridges: Case Study,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 359-391. Somerville, P.G., (1997), “The Characteristics and Quantification of Near Fault Ground Motion,” Proceedings of the FHWA/NCEER Workshop on National Representation of Seismic Ground Motion for New and Existing Highway Facilities, San Francisco, California, May 29-30, Technical Report NCEER-97-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, pp. 293-318. Somerville, P.G., Smith, N.F., Graves, R.W. and Abrahamson, N.A., (1997), “Modification of Empirical Strong Ground Motion Attenuation Relations to Include the Amplitude and Duration Effects of Rupture Directivity,” Seismological Research Letters, Vol. 68, pp 199- 222.

80 Updating Assessment Procedures and Developing a Screening Guide for Liquefaction

by T. Leslie Youd, Brigham Young University

Research Objectives The main objectives of this research program are to provide consensus MCEER Highway Project updates to standard procedures and prepare guidance documents for assess- National Science Foundation ing liquefaction hazards for highway bridge sites. The scope of these studies includes evaluation of liquefaction resistance of soils with standard and cone penetration tests, shear wave velocity measurements, and Becker penetration tests. Additional issues, such as updated magnitude scaling factors, are addressed. The research findings are incorporated in unified, well-established guidelines for use by practicing engineers.

• Effects of Liquefaction on iquefaction-induced ground and foundation displacements have Vulnerability Assessment, L been major causes of bridge damage during past earthquakes. The Great G. Martin, University of Alaskan earthquake of March 27, 1964 marked the commencement of stud- Southern California and ies to understand and mitigate liquefaction hazard (McCulloch and Bonilla, T.L. Youd, Brigham Young 1970; Kachadoorian, 1968; Youd, 1993). Over the past 30 years, a proce- University dure, termed the “simplified procedure,” has evolved for evaluating the • Liquefaction Remediation seismic liquefaction resistance of soils. This procedure has become the Techniques for Bridge standard practice in North America and throughout much of the world. Foundations, G. Martin, Seed and Idriss (1971) developed and published the basic “simplified pro- University of Southern cedure.” The procedure has been corrected and augmented periodically California and J. Mitchell, since that time with landmark studies by Seed (1979), Seed and Idriss (1982), Virginia Polytechnic and Seed et al. (1985). Institute In 1985, the Committee on Earthquake Engineering of the National • Synthesis Report: Lique- Research Council (NRC) organized a workshop with experts from the pro- faction Vulnerability fession and observers who thoroughly reviewed the state-of-the-art for Assessment, G. Martin, assessing liquefaction hazard in order to evaluate and update the proce- University of Southern dure. The workshop produced a report (NRC, 1985) that has become a California and T.L. Youd, widely used reference. Another workshop, held in 1996 and sponsored by Brigham Young University MCEER, was convened to review developments and gain consensus for further augmentations to the procedure. The scope of the workshop was limited to evaluation of liquefaction resistance. The workshop proceedings provide futher updates to the simplified procedure (see Youd and Idriss, 1997) and various recommendations were made on the following topics: 1. Use of the standard and cone penetration tests for evaluation of liquefaction resistance 81 2. Use of shear wave velocity Evaluating Liquefaction measurements for evaluation of liquefaction resistance Resistance of Soils Workshop Organizers 3. Use of Becker penetration test In general, soil liquefaction is a for gravelly soils major concern for structures con- T.L. Youd, Brigham Young 4. Magnitude scaling factors structed on saturated sandy soils. University 5. Correction factors for large I.M. Idriss, University of Major earthquakes, such as the 1906 California, Davis overburden pressures San Francisco, 1964 Alaska, 1964 In addition, a “screening guide” Niigata, Japan, 1989 Loma Prieta, and Workshop Participants for assessing liquefaction hazard 1995 Kobe, Japan, produced exten- was developed by Youd (1998). The sive damage as a consequence of liq- R.D. Andrus, National guide presents procedures for the uefaction and illustrate the need for Institute of Standards and engineering procedures to assess Technology systematic evaluation of liquefac- I. Arango, Bechtel Corp. tion resistance and damage poten- and mitigate the hazard. Since 1964, G. Castro, GEI Consultants, tial for bridge sites and guidance for experimental and analytical studies Inc. the prioritization of sites for further have been carried out to better un- J.T. Christian, Consultant, investigation and possible remed- derstand this phenomenon. Much Boston iation. The screening guide proce- of the early work was based on labo- R. Dobry, Rensselaer dures are generic, and can be used ratory testing of reconstituted Polytechnic Institute samples subjected to cyclic loading W.D. Liam Finn, University to determine liquefaction hazard of British Columbia for a variety of other types of struc- by means of cyclic triaxial, cyclic L.F. Harder, California tures. simple shear, or cyclic torsional Department of Water Resources M.E. Hynes, U.S. Army This research developed unified procedures for the Engineers Waterways assessment of seismic hazard to highway systems as a Experiment Station consequence of liquefaction, and produced two significant K. Ishihara, Science University of Tokyo publications. Proceedings of the NCEER Workshop on J. Koester, U.S. Army Evaluation of Liquefaction Resistance of Soils, NCEER-97-0022, Engineers Waterways provides consensus updates to standard procedures that can Experiment Station be used for evaluations of liquefaction susceptibility for a wide S. S.C. Liao, Parsons array of applications. The consensus approach to liquefaction Brinckerhoff evaluation is being referenced in many new documents for W. Marcuson III, U.S. Army geotechnical engineers. The approach has recently been Engineers Waterways recommended by the city and county of Los Angeles, thus it is Experiment Station expected that geotechnical consultants in the Los Angeles area G. Martin, University of Southern California will refer to the updated “simplified procedure” described in J. Mitchell, Virginia Polytech- these proceedings. The Screening Guide for Rapid Assessment nic Institute of Liquefaction Hazard at Highway Bridge Sites, MCEER-98- Y. Moriwaki, Woodward Clyde 0005, is intended for use by highway engineers with Consultants experience in geotechnical engineering practice, but not M. Power, Geomatrix necessarily specialized in seismic hazard evaluation. It is based Consultants, Inc. on well-established experimental and analytical procedures P. Robinson, University of developed in the U.S. and Japan and structured for Alberta, Edmonton R. Seed, University of implementation by practicing engineers. The simplified California, Berkeley procedures and the screening guide are not highway specific, K. Stokoe, University of Texas, i.e., they can be used for generic liquefaction evaluation Austin purposes for a wide variety of structures.

82 tests. The outcome of these studies may be used to estimate average val- generally confirmed the fact that re- ues of rd for use in Equation 1: sistance to cyclic loading is influ- 1.. 0−≤ 0 00765z z 9 . 2m  enced primarily by the state of the  −<≤ Co-authors of Papers r = 1.. 174 0 0267z 9 . 2 z 23m soil, the intensity and duration of the d 0.. 744−<≤ 0 008z 23 z 30m  Presented 050. z > 30m  cyclic loading, and the grain char- (2) R. Boulanger, University of acteristics of the soil. However, the California, Davis results also showed that the distur- where z is depth below ground sur- R. Kayen, U.S. Geological bance induced by sampling and test Survey, Menlo Park face in meters. Average values of rd preparation procedures so greatly estimated from these equations are S. Noble, DOWL Engineers, affected the test results that labora- plotted on Figure 1. Anchorage tory procedures were abandoned R. Olson, U.S. Army Several procedures have been ap- Engineers Waterways for routine engineering practice. At plied to determine CRR. As noted Experiment Station that point, the laboratory procedure above, field tests have become the C. Wride, University of was replaced by a procedure based state-of-the-practice for routine in- Alberta, Edmonton on cheaper and generally more reli- vestigations to avoid the difficulties able field tests, such as standard associated with sampling and test- Reviewers of the cone penetration tests, for evalua- ing. Accordingly, as part of the gen- Workshop Proceedings tion of liquefaction resistance. eral consensus recommendations The calculation or estimation of from the 1996 workshop (see Youd D. Anderson, CH2M-Hill T. Blake, Fugro-West two primary seismic variables is re- and Idriss, 1997), four field tests quired to evaluate liquefaction were recommended for general use resistance. These variables are the in evaluating liquefaction resist- seismic demand placed on a soil ance for engineering practice. layer, expressed in terms of cyclic These are: (1) standard penetration stress ratio (CSR), and the capacity test (SPT), (2) cone penetration test of a soil layer to resist liquefaction, (CPT), (3) measurement of shear- wave velocity (V ), and (4) Becker expressed in terms of cyclic resis- S tance ratio (CRR). Seed and Idriss (1971) formulated ( max) d rd = the following equation for calculat- ( max) r ing CSR: 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0

3 (10) CSR =(␶␴′␯␱ ) = av Average values ( )(␴␴′ ) (1) 6 (20) 065 . amax g␯␱ ␯␱ r d Mean values of rd calculated 9 (30) from Equation 2 where is the peak horizontal 12 (40) Range for different amax soil profiles acceleration during an earthquake, 15 (50) g is the gravitational acceleration, ␴ Simplified procedure

␯␱ Depth, m (ft) 18 (60) not verified with and ␴′␯␱ are total and effective over- 21 (70) case history data burden stress, respectively; and r is in this d region a stress reduction factor. Curves 24 (80) showing the range and average val- 27 (90) 30 (100) ues of rd are plotted in Figure 1. For noncritical projects such as hazard ■ screening, the following equations Figure 1. rd versus Depth Curves Developed by Seed and Idriss (1971) with Added Mean Value Lines from Equation 2

Liquefaction: Assessment and Screening Guide83 penetration test (BPT) for gravelly penetration criterion for a simpli- sites. The advantages and disadvan- fied procedure and is referred to as tages of each test are listed in Table the “simplified base curve.” A rec- 1. A conscientious attempt was made ommended adjustment to this plot to correlate liquefaction resistance was to modify the trajectory of the

criteria from various tests to provide simplified base curve at low (N1)60 generally consistent results, no mat- to a projected CRR intercept of ter which test is employed and inde- about 0.05 as shown in Figure 2. pendent of the testing conditions. This adjustment reshapes the base Some recommendations and consid- curve to achieve consistency with erations for each test are briefly dis- CRR curves developed from cone cussed in the following. penetration test (CPT) data and probabilistic analysis by Liao et al. (1988) and Youd and Noble (1997). Standard Penetration Test (SPT)

Criteria for evaluating liquefaction Cone Penetration Test (CPT) resistance based on SPT blow counts are largely embodied in the Although not as commonly used

CSR versus (N1)60 plot as shown in as the SPT, the CPT is becoming a Figure 2. Conservatively drawn CRR major tool for delineating soil curves separate data indicative of stratigraphy and for conducting pre- liquefaction from data indicative of liminary evaluations of liquefaction nonliquefaction for various fines resistance. Criteria have been de- contents. The CRR curve for mag- veloped for calculating liquefaction nitude 7.5 earthquakes and for fines resistance (CRR) directly from CPT contents less than 5% is the basic data (see Robertson and Wride in Youd and Idriss, 1997). These cri- ■ Table 1. Comparison of Advantages and Disadvantages of Various Field Tests teria may be applied in practice— provided adequate samples are Test Type erutaeF retrieved, preferably by the SPT pro-

CTTP PS VS TPB cedure—to verify the soil types and tsetforebmuN liquefaction resistance assigned. tastnemerusaem Attnadnub Adnadnub Leetimi srapS Figure 3 shows the primary chart setisnoitcafeuqil used for determining liquefaction -ssertsfoepyT yllaitraP egral,deniarD ,deniardyllaitraP resistance from CPT data for clean roivahebniarts S,niartsllam deniard niarts niartsegral tsetgnicneulfni niartsegral sands. The chart shows CSR plotted against corrected and normal- lortnocytilauQ Vddoogyre Pdoogotroo Groo ooP ized CPT resistance, qc1N, from sites ytilibataeperdna where liquefaction was or was not

fonoitceteD observed following past earth- liosfoytilibairav Vddoogyre Groo Fria iaF quakes. Similarly, a CRR curve de- stisoped fines the boundary between lique- nisepytlioS yliramirP faction and nonlique-faction. This sitsethcihw Nllevarg-no Nlevarg-no lA levarg chart is valid for magnitude 7.5 dednemmocer earthquakes and clean, sandy soil. sedivorptseT Nso Yoe NoN liosfoelpmas The figure also shows that cyclic shear strain and ground deforma- serusaemtseT roxedni gnireenignE tion potential at liquefiable sites Ixxedn ednI xedn gnireenigne ytreporp I decrease as penetration resistance ytreporp increases (dashed curves).

84 Because the CPT equipment and 0.6 37 procedures are less variable than 29 25 those for the SPT, fewer corrections Percent Fines = 35 15 < 5 are required. Nevertheless, correc- 0.5 tions are still required for overburden pressure and grain characteris- 10 tics. These corrections are discussed 0.4 vo' 20

in detail in papers by Robertson and / CRR curves for 5,15, and av Wride, and Olsen, in Youd and Idriss 31 35 percent fines, respectively

20 (1997). 0.3 12 50+

Finally, theoretical as well as labo- 17 27 18 80 50+ ratory studies indicated that cone 60 20 11 10 20 10 10 40 10 0.2 50 48 resistance is influenced by softer or 92 10 20 26 12 Cyclic Stress Ratio, Cyclic Stress 10 20 10 18 80 25 12 stiffer soil layers above or below the 30 20 12 22 13 75 12 FINES CONTENT > 5% cone tip. It was observed that the 27 67 30 75 50+ Modified Chinese Code Proposal (clay content = 5%) 0.1 60 1010 20 CPT did not usually measure the full 10 13 30 27 Marginal No Liquefaction Liquefaction Liquefaction penetration resistance in thin sand 31 Adjustment Pan - American data Recommended Japanese data layers sandwiched between layers By Workshop Chinese data 0 of softer soils. Based on an elastic 0 10 20 30 40 50 solution, Vreugdenhil et al. (1994) Corrected Blow Count, (N ) developed a procedure for estimat- 1 60 ing full cone penetration resistance ■ Figure 2. Simplified Base Curve Recommended for Calculating CRR of thin, stiff layers contained within from SPT Data Along with Empirical Liquefaction Data (after Youd and softer strata. Robertson and Fear Idriss, 1997) (1995) further suggested a correc- Shear Wave Velocity, VS tion factor for cone resistance, KH , as a function of layer thickness as Several simplified procedures shown in Figure 4. have been proposed for the use of field measurements of small-strain shear wave 0.6 M=7.5 0.25 < D50 (mm) < 2.0 velocity, VS , to assess liq- FC (%) < 5 uefaction resistance of Y ≈20%≈10%≈3% 0.5 e granular soils. The advan- CRR Curve tages of using VS are that 0.4 (1) it can be accurately Liquefaction measured in-situ using a 0.3 No Liquefaction number of techniques such as crosshole and 0.2 downhole seismic tests, the seismic cone penetra-

Cyclic Stress Ratio (CSR) Cyclic Stress 0.1 tion test, or spectral analy- Field Performance Liq. No Liq. NCEER (1996) Stark & Olson (1995) sis of surface waves, (2) Workshop Suzuki et al. (1995b) 0 measurements are pos- 0 50 100 150 200 250 300 sible in soils that are diffi-

Corrected CPT Tip Resistance, qc1N cult to penetrate with CPT and SPT, (3) measure- ■ Figure 3. Recommendations for Calculating CRR and CPT ments can be performed Data Along with Empirical Liquefaction Data (after Youd in small laboratory and Idriss, 1997)

Liquefaction: Assessment and Screening Guide85 specimens allowing direct compari- test (BPT) has become an effective son between laboratory and field be- tool using large-diameter penetrom- havior, and (4) it is directly related eters. The BPT consists of a 3 m long to small-strain shear modulus. Two double-walled casing driven into the ground with a double-acting diesel- significant limitations of using VS in liquefaction hazard evaluations are driven pile hammer. The BPT resis- that seismic wave velocity measure- tance is defined as the number of ments are made at small strains blows required to drive the casing whereas liquefaction is a large strain through an increment of 300 mm. phenomenon, and seismic testing The BPT is not correlated directly does not provide samples for classi- with liquefaction resistance, but is fication of soils and identification of used to estimate equivalent SPT blow nonliquefiable soft clay-rich soils. A counts through empirical correla- paper by Andrus and Stokoe (Youd tions. The equivalent SPT blow count and Idriss, 1997) reviews current is then used to estimate liquefaction simplified procedures for evaluat- resistance. However, studies have ing the liquefaction resistance of shown that SPT blow counts can granular soil deposits using small- only be roughly estimated from BPT strain shear wave velocity. measurement due to deviations in hammer energy for which Harder and Seed (1986) developed an en- Becker Penetration Tests (BPT) ergy correction procedure based on Liquefaction resistance of non- measured bounce-chamber pressure, gravelly soils has been evaluated pri- and friction along the driven casing marily through CPT, SPT and occa- and its influence on the penetration resistance. sionally with VS measurements. However, CPT and SPT are not generally reliable in gravelly soils as large Workshop Conclusions gravel particles may interfere with the normal deformation of soil ma- In addition to discussing the vari- terials around the penetrometer, in- ous tests described above, workshop creasing penetration resistance. participants examined magnitude Therefore, the Becker penetration scaling factors; corrections for high

H/dc (a) 0 20 40 60 80 qc* 2 (b)

H qcB Deposit D

H qcA Layer A 1.5

10 = qcA /qcA qc*

5 qcB Recommended 2 Deposit B

Thin Layer Correction Factor, K Factor, Correction Layer Thin Relationship qc*=KH xqcA 1 Z 0 1000 2000 3000 2 Layer Thickness, H (mm) for a 10 cm Cone (i.e. dc = 35.7 mm)

■ Figure 4. Thin-layer Correction Factor, KH, for Determination of Equivalent Thickness-layer CPT Resistance (modified from Robertson and Fear, 1995)

86 overburden pressures, static shear environments. Taking advantage of stresses and age of deposit; seismic relationships between these envi- “The simplified factors, such as magnitude and peak ronments and liquefaction suscep- procedures and acceleration; and energy-based crite- tibility, a screening guide was de- the screening ria and probabilistic analyses. Gen- veloped which guides geotechnical guide are not eral consensus recommendations engineers in conducting rapid as- highway included the following (see Youd and sessments of liquefaction hazard. Idriss, 1997): The guide presents a systematic specific, i.e., they can be used • Consenesus criteria for evaluat- application of standard criteria for ing liquefaction resistance were assessing liquefaction susceptibility, for generic developed for SPT, CPT, shear evaluating ground displacement liquefaction wave velocity and BPT tests. potential, and assessing the vulner- evaluation • Two or more test procedures ability of bridges to liquefaction-in- purposes for a should be applied at each site to duced damage. The screening pro- wide variety of assure both adequate definition ceeds from least complex, time-con- structures.” of soil stratigraphy and consis- suming and data-intensive evalua- tent evaluation of liquefaction re- tions to the more complex, time- sistance is attained. consuming, and rigorous analyses. • New sets of magnitude scaling Thus, many bridge sites can be factors are recommended for en- evaluated and classified as low haz- gineering practice. These factors ard with very little time and effort. are greater than those used pre- Only bridge sites with significant viously for earthquakes with hazard need to be evaluated with magnitude less than 7.5. The the more sophisticated and time- new factors yield safe but less consuming procedures. conservative estimates of lique- The screening guide is conserva- faction resistance. tive—that is, at each juncture in the • Evaluating liquefaction resis- screening process, uncertainty is tance beneath sloping ground or weighed on the side that liquefac- embankments is not well under- tion and ground failure could oc- stood at this time. cur. Thus, a conclusion that liquefaction and detrimental ground dis- • Moment magnitude, Mw , should be used as an estimate of earth- placement are very unlikely is a quake size for liquefaction resis- much more certain conclusion than tance calculations. the converse outcome—that lique- • The preferred procedure for esti- faction and detrimental ground dis- mating peak acceleration is to ap- placements are possible. This con- ply attenuation relationships servatism leads to the corollary con- consistant with soil conditions at clusion that additional investigation a given site. is more likely to reduce the estimated liquefaction hazard than increase it. Developing a The principal steps and logic path Screening Guide for the screening procedure are listed in Figure 5. In assessing liq- Liquefaction does not occur ran- uefaction hazard, the recom- domly in natural deposits but is lim- mended procedure is to start at the ited to a rather narrow range of seis- top of the logic path, perform the mic, geologic, hydrologic, and soil required analyses for each step, and

Liquefaction: Assessment and Screening Guide87 Soil Classification Analysis SCREENING EVALUATION FOR LIQUEFACTION HAZARD AT BRIDGE SITES

Soil Classification Analysis

All Soils Non-liquefiable by Review Prior Evaluations Criterian Screening Guide, p.31 of Liquefaction Hazard or All soils are USCS Soil Types FS > 1.3 for current estimates No CL, CH, OL or OH, of seismicity mapped as or Liquefaction Suscueptibility All Soils are AASHTO is very low Soil Types A-2-5, A-2-6, A-2-7, A-5, A-6, A-7-5, or A-7-6 Yes No Previous Evaluation No or Unknown

Geologic Evaluation of Liquefaction Susceptibility Penetration Analysis susceptibility is very low FS > 1.3 Insufficient Information prioritize for further investigation

Possible liquefaction hazard; No

No or Yes Unknown low priority for

further investigation Analysis of Slope Stability Low liquefaction hazard; with Liquefied Layers FS > 1.1 Seismic Hazard Evaluation

amax for given M is less than limits in Screening Guide p. 15 Yes

Analysis of Slope No or Deformation Unknown Estimated Displacements less than 100 mm low priority for

Water Table Evaluation further investigation Water Table Depth is Yes Low liquefaction hazard; Persistantly Deeper than 15 m No or Unknown further investigation hazard; prioritize for

Probable high liquefaction Analysis of Lateral Spread Displacement DH < 100 mm No or Unknown

Yes

Evaluation for Extra Sensitive Clay Analysis of Ground Yes (N ) _< 5 or CPT q < IMP , LL< 40% 1 60 CIN a Settlement MC > 0.9LL and LI > 0.6, and USCS Soil types CL or ML or DH < 100 mm AASHTO Types A-4, A-2-4, A-6 or A-2-6 hazard; high priority Possible sensitive soil for further investigation Deposits of sensitive clay or depositional conditions for sensitive clay confirmed in area Yes

Analysis of Bearing Capacity No

No Adequate Capacity with Yes Liquefied Layers

■ Figure 5. Flow Diagram Showing Steps and Criteria for Screening of Liquefaction Hazard for Highway Bridges

88 proceed downward until the bridge the U.S. The updated “simplified pro- is classified into one of four catego- cedure” has been recommended in “The updated ries: both the city and county of Los An- ‘simplified 1. Confirmed high liquefaction and geles as the preferred approach to procedure’ has use to assess liquefaction potential ground failure hazard—very high been priority for further investigation at a given site. In a companion effort, liquefaction hazard maps have recommended and possible mitigation; in both the city 2. Confirmed liquefaction suscep- been produced for southern Califor- tibility but unknown ground fail- nia and the California Division of and county of ure hazard—high priority for fur- Mines and Geology will produce Los Angeles as ther investigation; similar maps for northern California. the preferred 3. Insufficient in-formation to as- Taken together, the updated maps approach to use sess liquefaction susceptibility— and the updated “simplified proce- to assess prioritized for further investiga- dure” will greatly enhance the accu- liquefaction tion; racy of liquefaction hazard assess- potential at a ments. Accurate assessments will 4. Low liquefaction hazard—low given site.” priority for further investigation. allow retrofit projects to be prioritized according to potential impact If there is clear evidence that liq- and new projects to be designed to uefaction or damaging ground dis- accommodate potential hazards. placements are very unlikely, the site Finally, there are issues that should is classed as “low liquefaction haz- be further investigated and ad- ard and low priority for further in- dressed in the liquefaction evalua- vestigation,” and the evaluation is tion procedures. The evaluation of complete for that site. If the avail- liquefaction resistance beneath slop- able information indicates a likely ing ground or embankments (slopes hazard, or if the data are inadequate greater than 6%) is not well under- or incomplete, the site is classed as stood. Hence, such evaluations are having possible liquefaction hazard, beyond the applicability of the sim- and the screening proceeds to the plified procedure, and further stud- next step. If the available site infor- ies are required to develop proce- mation is insufficient to complete a dures for the evaluation of liquefac- liquefaction hazard analysis, then tion resistance beneath sloping simplified seismic, topographic, geo- ground. Moreover, it is known that logic, and hydrologic criteria are liquefaction resistance increases used to prioritize the site for further with soil plasticity. However, more investigation. The complete details research is needed in order to quan- of the procedure are given by tify this relationship. Recently, Youd (1998). probabilistic methods have been used in some risk analyses, but are still outside the mainstream of stan- Conclusions and dard practice. Similarly, seismic energy passing through a liquefiable Recommendations for layer can be potentially adopted as Future Research a liquefaction resistance criteria. This concept is relatively new and The consensus approach to lique- also requires further research. faction evaluation is being referenced in many new documents for geotechnical engineers throughout

Liquefaction: Assessment and Screening Guide89 References Kachadoorian, R., (1968), Effects of the Earthquake of March 27, 1964, on the Alaska Highway System, U.S. Geological Survey, Professional Paper 545-C. Liao, S.S.C., Veneziano, D., Whitman, R.V., 1988, “Regression Models for Evaluating Liquefac- tion Probability,” Journal of Geotechnical Engineering, Vol. 114, No. 4, pp. 389-411. McCulloch, D.S. and Bonilla, M.G., (1970), Effects of the Earthquake of March, 1964, on the Alaska Railroad, U.S. Geological Survey, Professional Paper 545-D. National Research Council (NRC), (1985), Liquefaction of Soils During Earthquakes, Na- tional Academy Press, Washington, D.C. Robertson, P.K. and Fear, C.E., (1995), “Liquefaction of Sands and Its Evaluation,” Proceed- ings of the First International Conference on Earthquake Geotechnical Engineering, Keynote Lecture, Tokyo, Japan. Seed, H.B., (1979), “Soil Liquefaction and Cyclic Mobility Evaluation for Level Ground Dur- ing Earthquakes,” Journal of Geotechnical Engineering, ASCE, Vol. 105, No. GT2, pp. 201- 255. Seed, H.B. and Idriss, I.M., (1971), “Simplified Procedure for Evaluation Soil Liquefaction Potential,” Journal of the Soil Mechanics and Foundation Division, ASCE, Vol. 97, No. SM9, pp. 1249-1273. Seed, H.B. and Idriss, I.M., (1982), Ground Motions and Soil Liquefaction During Earth- quakes, Earthquake Engineering Research Institute Monograph. Seed, H.B., Tokimatsu, K., Harder, L.F. and Chung, R.M., (1985), “The Influence of SPT Proce- dures in Soil Liquefaction Resistance Evaluations,” Journal of Geotechnical Engineering, ASCE, Vol.111, No. 12, pp. 1425-1445. Vreugdenhil, R., Davis, R. and Berrill, J., (1994), “Interpretation of Cone Penetration Results in Multilayered Soils,” International Journal for Numerical Methods in Geomechanics, Vol. 18, pp. 585-599. Youd, T.L., (1993), Liquefaction-Induced Damage to Bridges, Transportation Research Record, No. 1411, pp. 35-41. Youd, T.L., (1998), Screening Guide for Rapid Assessment of Liquefaction Hazard at High- way Bridge Sites, Technical Report MCEER-98-0005, Multidisciplinary Center for Earth- quake Engineering Research, Buffalo New York. Youd, T.L. and Idriss, I.M., eds, (1997), Proceedings of the NCEER Workshop on Evaluation of Liquefaction Resistance of Soils, Technical Report NCEER-97-0022, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York. Youd, T.L. and Noble, S.K., (1997), “Liquefaction Criteria Based on Statistical and Probabilis- tic Analyses,” Proceedings of the NCEER Workshop on Evaluation of Liquefaction Resis- tance of Soils, Technical Report NCEER-97-0022, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

90 Fragility Curve Development for Assessing the Seismic Vulnerability of Highway Bridges

by John B. Mander, University at Buffalo, State University of New York

Research Objectives MCEER Highway Project The principal objective of this research is to develop, from the National Institute of fundamentals of mechanics and dynamics, the theoretical basis of Building Sciences (NIBS) establishing fragility curves for highway bridges through the use of rapid analysis procedures. In contrast to other methods that have been used in the past, such as either empirical/experiential fragility curves or individualized fragility curves based on extensive computational simulations, the present approach seeks to establish dependable fragility curves based on the limited data readily obtainable from the National Bridge Inventory (NBI).

Retrofitting Manual Ian Buckle, University of he purpose of this research is to set forth the basis for developing Auckland Tfragility curves that can be used in various ways as part of a seismic Anindya Dutta, EQE vulnerability analysis methodology for highway bridges. International, formerly of the University at Buffalo To develop a set of fragility curves for various damage states for a specific Gokhan Pekcan, University bridge, the notion of adjusting “Standard Bridge” fragility curves for site- at Buffalo specific effects is adopted. Only three sources of data are needed for this Andrei Reinhorn, University analysis: (1) National Bridge Inventory (NBI) records that contain the bridge at Buffalo attributes and geographical location; (2) ground motion data (this is best Stuart Werner, Seismic obtained from the USGS web site); and (3) geological maps from which Systems and Engineering soil types and hence S-factors can be inferred. Full details of this approach Consultants are given in Basöz and Mander (1999). HAZUS Developments Nesrin Basöz, K2 Technolo- gies Michael O’Rourke, Rensselaer Polytechnic Institute

91 The Approach Using of characterizing the probabilistic nature of the phenomena con- “Standard Bridge” cerned is through the use of so- Retrofitting Manual for Fragility Curves called fragility curves. Figure 1 Highway Systems shows how the inherent uncer- • Volume I is concerned with tainty and randomness of bridge network analysis and Fragility Curve Theory capacity versus ground motion de- requires the use of bridge For a given bridge, it is possible mand can be used to establish fra- specific fragility curves to gility curves. Figure 1(a) shows an assess the direct and to predict, deterministically, the acceleration-displacement spectra indirect losses as a result level of ground shaking necessary of earthquake-induced to achieve a target level of response for the ground motion. Superim- damage to bridges (Werner and/or damage state. In addition to posed with this curve is the push- et al., 1998; 1999). It is assuming material properties and over capacity of a bridge. In a de- important that a rapid certain other structural attributes terministic analysis, the intersection analysis procedure is that affect the overall of a of the two curves gives the ex- available to minimize the capacity pected level of performance. How- computational effort and bridge, such a deterministic assess- data collection require- ment requires that certain assump- ever, probability distributions are ments. tions be made about the ground drawn over both the capacity and motion and site conditions—both demand curves to indicate the as- • Volume II is concerned with the seismic evaluation factors that affect seismic demand. sociated uncertainty and random- and retrofitting measures Naturally, values of these param- ness of performance. From this fig- for bridge structures. In eters are not exact—they invariably ure it is evident there is a wide range order to assess the seismic have a measure of both randomness of possible performance out- vulnerability of a suite of and uncertainty associated with comes—there is not a unique or bridges, “rapid screening” them. An increasingly popular way exact answer. techniques are used. As an alternative to using indexing methods, similar It is anticipated that planning engineers in State and Federal to the multiplicity of existing approaches used government may use network analysis software that has been currently by various states developed as part of Volume I of a three-volume set of Retrofit- and the present retrofitting ting Manuals under the MCEER Highway Project (see Seismic Ret- manual, the proposed rofitting Manual for Highway Systems in this report). This soft- fragility curve based ware will contain fragility curves for bridges developed as part method for the rapid of this research. screening of highway Bridge engineers in State departments of transportation use bridges is consistent with the network analysis used “rapid screening” techniques for assessing the seismic vulner- in Volume I and the de- ability of bridges as a planning/scheduling tool. The new ap- tailed analysis of Volume II. proach is expected to use the fragility curve methodology developed herein. • The underlying theory that forms the basis of the Engineers, social scientists and planners use the FEMA-NIBS fragility curve development software HAZUS, which is to contain a new fragility curve proce- for these applications is dure developed as part of this research. also the same as that used State departments of transportation and their engineering con- in the detailed analysis of sultants are expected to use detailed seismic evaluation proce- bridge structures and the dures as part of their retrofitting design studies. The underlying seismic performance of theory that forms the basis of the fragility curve development is individual components also the same as that used in the detailed analysis of bridge struc- and members. These detailed procedures are tures and the seismic performance of individual components and described in Volume II. members.

92 If structural capacity and seismic Based on these investigations it is ␤ demand are random variables that recommended that c = 0.6. roughly conform to either a normal Median values of the peak ground or log-normal distribution then, fol- acceleration for five different dam- HAZUS Developments lowing the central limit theorem, it age states are assessed using an al- • The HAZUS project, which is can be shown that the composite gorithm that is based on the so- sponsored by FEMA through performance outcome will be log- called capacity-spectrum method, a contract with NIBS, is using the results of this normally distributed. Therefore, the as indicated in Figure 1(a). This dis- research to develop probabilistic distribution is ex- placement-based nonlinear static software for the second- pressed in the form of a so-called analysis procedure assumes a stan- generation of fragility fragility curve given by a log-normal dard AASHTO-like earthquake curves for highway bridges. cumulative probability density func- response spectrum shape, which This is part of the first tion. Fortunately, only two param- can be adjusted later to account for major revision of HAZUS and is included in the eters are needed to define such a site-specific spectral ordinates and/ HAZUS98 software (refer curve—a median (the 50th percen- or soil types. The five damage states also to HAZUS, 1997). tile) and a normalized logarithmic and their associated performance standard deviation. Figure 1(b) pre- outcomes are listed in Table 1. sents the form of a normalized fragility curve for bridges. The cumulative probability function is given by:

Median Demand  1  S  FS( ) = ⌽ ln a  a ␤   (1)  c Ai  where ⌽ is the standard log-normal Median Capacity cumulative distribution function; Sa is the spectral acceleration ampli- Spectral Acceleration tude (for a period of = 1 sec.); T Ai Spectral Displacement is the median (or expected value) spectral acceleration necessary to (a) Capacity-Demand Acceleration-Displacement Spectra Showing Randomness and/or Uncer- th cause the i damage state to occur; tainty in Structural Behavior and Ground ␤ and c is the normalized composite Motion Response log-normal standard deviation which incorporates aspects of un- 1 certainty and randomness for both capacity and demand. The latter parameter is sometimes loosely re- 0.5 ferred to as either the coefficient of variation or the coefficient of dis- persion. The parameter has been Cumulative Probability calibrated by Pekcan (1998), Dutta 0 0123 and Mander (1998) and Dutta Spectral Acceleration Ratio S /A (1999) from a theoretical perspec- a i tive, and validated by Basöz and (b) Normalized fragility curve that accounts for Mander (1999) against experiential uncertainty and randomness in both demand ␤ fragility curves obtained from data and capacity. Note c = 0.6. gathered from the 1994 Northridge and 1989 Loma Prieta earthquakes ■ Figure 1. Probabilistic Definition of Uncertainty/Randomness in by Basöz and Kiremidjian (1998). Establishing Fragility Curves for a Seismic Vulnerability Analysis

Fragility Curve Development93 ■ Table 1. Definition of Damage States and Performance Outcomes

egamaD rofrotpircseD ekauqhtrae-tsoP egatuOfoemiT deriuqeRsriapeR etatS egamaDfoeergeD erutcurtSfoytilitU detcepxE

1)Nldleiy-erp(eno Neamro N-no - 2tMehgilS/roni SIggamadthgil nsnihctap,tsujda,tceps yad3< 3eMetaredo Rsgamadelbariape Rstnenopmocriape keew3< 4eMevisnetxE/roja Isgamadelbariaperr Rstnenopmocdliube htnom3< 5eCespalloC/etelpmo Iegamadelbariaperr Rsrutcurtsdliube htnom3>

■ Table 2. Fields in NBI used in Determining Bridge Fragility

ataDIBN DKnoitinife K esUrehtO metI weks D3 1eS*tat gisededocfoepytrefnioT n 8rebmunerutcurtS rebmunnoitacifitnedilareneG lanoitnevnocrocimsiesrehtehwsrefnI 2t7 Y*liubrae ngised 3w4 S*ek 4e2 pytecivreS segdirbyawhgihtcelesoT 4e3 S*3pyterutcurt elbaTmorfevrucytiligarfesabrefnioT 4t5 N*ninuniamnisnapsforebmu apselpitumroelgnisrehtehwrefnioT 4s6 napshcaorppaforebmuN egdirbrojamasiegdirbrehtehwrefnioT egdirbrojamasiegdirbrehtehwrefnioT 4n8 L*apsmumixamfohtgne fiegdirbrojam( L )m051> etupmocot;htgnelnapsegarevarefnioT 4h9 S*tgnelerutcurt eulavtnemecalper 5h2 tdiwkceD eulavtnemecalperetupmocoT

1.0 1.0 0.9 0.9 ] ] 0.8 0.8 0.7 0.7 PGA PGA

| |

0.6

0.6 i i 0.5 0.5 = ds 0.4

> 0.4

D D > = ds [ 0.3 [ 0.3 P P 0.2 0.2 0.1 0.1 0 PGA (g) 0 PGA (g) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4

USGS-moderate moderate - analytical USGS major major analytical WCFS major WCFS-moderate Loma Prieta minor & moderate

(a) Moderate damage, DS3 (b) Major damage, DS4 ■ Figure 2. Comparison of Analytical and Empirical Fragility Curves for Discontinuous Multiple Span Bridges with Single Column Bents and Non-monolithic Abutments

94 Validation of Theory with A “standard bridge” is assumed to Empirical Evidence be a “long” structure with no appreciable three-dimensional (3D) Analytically predicted fragility effects present. For several types • Structure Fragility - curves for various different bridge of “standard bridges,” median PGA Examination and types were validated against fragil- values have been derived for each Upgrading of ATC-13, A. ity curves that were empirically Cakmak, Princeton of the damage states (a2, a3, a4, a5). derived from data gathered for high- The results of the “standard bridge” University way bridges damaged in the 1994 are then modified by factors • Lateral Strength Analysis Northridge and 1989 Loma Prieta accounting for skew (K ) and 3D Methods, R. Imbsen, earthquakes. A sample of the vali- skew effects (K3D) using the NBI database Imbsen Associates, Inc. dation for one common class of described in Table 2. This table California bridge is given in Figure shows which fields of the NBI are • Risk Assessment of 2. In spite of the large degree of used and for what purpose. Highway Systems, uncertainty in defining both bridge S. Werner, Seismic damage and the spatial distribution Systems and Engineering Consultants of ground motion in the field, it will “Standard Bridge” Fragility be noted that both the median dam- Curves • Seismic Risk Analysis of age values (50 percentile) and the The “standard bridge” fragility Bridges in Memphis, shape (for ␤ = 0.6) of the fragility H. Hwang, Center for c curves are presented in Table 3 curves agree rather well. Earthquake Research and where median fragility parameters Information, University of are listed for conventionally (non- Memphis Implementation and Database seismically) designed and seis- • Evaluation of Bridge Requirements mically designed bridges. Implicit in the median fragility curve values Damage Data from Recent The National Bridge Inventory are assumptions which are consid- Earthquakes, (NBI) is a database maintained by the A. Kiremidjian, Stanford ered to be in keeping with typical University Federal Highway Administration that construction practice throughout contains information on every high- the U.S. • SRA Validation and way bridge in the U.S. The database Fragility Curves, has 116 fields that are used to de- M. Shinozuka, University scribe structural and operational Scaling the “Standard of Southern California and characteristics of a bridge. Because G. Deodatis, Princeton Bridge” Fragility Curves University all the nation’s highway bridges are to Account for Skew required to be inspected on a bien- • Major Bridge Damage State nial basis, the NBI database is kept and 3D Effects Modeling, W.D. Liu, up to date. Information in the NBI In order to convert the “standard Imbsen Associates, Inc. provides functional and operational bridge” fragility curves to a bridge– characteristics, but there is insuffi- specific value for a given spectral cient detail to permit a detailed analy- acceleration, the parameters K sis to be performed when deriving skew and K3D are used as scaling fragility curves. Therefore, the basis relations. The effect of bridge skew of obtaining a bridge specific fragil- on changing fragility curves can be ity curve is to take the results of a accounted for by applying the angle “standard bridge” fragility curve and ␣ of skew given in the NBI ( skew) to scale those results using selected using: data from the NBI. = ␣ K skew sin skew (2)

Fragility Curve Development95 ■ Table 3. U.S. Highway Bridge Fragility Curve Median Values of PGA Median PGA Damage Classification NBI Class Cdyllanoitnevno Dsengise egdirB yllacimsieS State dengiseD NaainrofilaC-no inrofilaC 2 62.0 33.0 54.0 601-101 ,stnebnmuloc-itluM 3 53.0 64.0 67.0 603-103 detroppus-ylpmis 4 44.0 65.0 50.1 605-105 5 56.0 38.0 35.1 2 53.0 45.0 ,stnebnmulocelgniS 602-502 3 24.0 88.0 xob ,sredrig elbacilppaton 606-506 4 05.0 22.1 suounitnocsid 5 47.0 54.1

2 *06.0 *06.0 *19.0 602-102 3 97.0 97.0 19.0 etercnocsuounitnoC 706-106 4 50.1 50.1 50.1 5 83.1 83.1 83.1

2 *67.0 *67.0 *19.0 3 67.0 67.0 19.0 C0leetssuounitno 14-204 4 67.0 67.0 50.1 5 40.1 40.1 83.1

2 *8.0 *8.0 *8.0 3 9.0 9.0 9.0 Slnapselgni lA 4 1.1 1.1 1.1 5 6.1 6.1 6.1 2 4.0 4.0 6.0 3 5.0 5.0 8.0 segdirbrojaM 4 6.0 6.0 0.1 5 8.0 8.0 6.1

* Short period portion of spectra applies, therefore evaluate Kshape.

■ Table 4. Modification Rules Used to Model 3D Effects K K D3 Type ssalCIBN D3 0991>raeYngiseDcimsieS ngiseDlanoitnevnoC )ACni5791>(

C6etercno 1/01-10 52.0+1 np /52.0+1 np etercnoC 2/602-10 33.0+1 n /33.0+1 n suounitnoC p

/90.0+1 n ; L ≥ 20 m S0leet 13-103 p /52.0+1 n < p /02.0+1 np ; L 20 m /50.0+1 n ; L ≥ 20 m S0suounitnoCleet 14-204 /33.0+1 n /01.0+1 n ; L < 20 m p

P6etercnoCdessertser 5/05-10 52.0+1 np /52.0+1 np etercnoCdessertserP 6/706-10 33.0+1 n /33.0+1 n suounitnoC p

n = number of spans in bridge; np = n - 1 = number of piers

96 Table 4 presents ■ Table 5. Modified Repair Cost Ratio for All Bridges modification rules to etatSegamaD naeMtseB foegnaR account for 3D effects oitaRtsoCriapeR soitaRtsoCriapeR via the parameter K3D. 10)dleiy-erp(egamadoN: 0 This parameter con- 23egamadthgilS: 030. 0.0ot10.0 verts a long “standard 38egamadetaredoM: 050. 1.0ot20.0 bridge” structure to a 45egamadevisnetxE: 002. 4.0ot01.0 specific (straight/right) 56etelpmoC: S0noitauqEee .1ot03.0 bridge with a finite number of spans. “standard bridge” fragility curves were derived. This equation is Scaling Relations for Damage necessary to ensure all fragility States 3, 4, and 5 curves possess a common format— either PGA or Sa at T = 1.0 second. The modified median fragility Note that where the PGA level curve parameter is given by given in Table 3 is identified with an *, the short period motion A = K K a /S (3) i skew 3D i ≤ governs, K shape 1, and the soil amplification factor for “short” where ai is the median spectral acceleration (for T = 1.0 second period structures (provided by spectral ordinate) for the ith damage NEHRP) is used. Otherwise, state listed in Table 3; S is the soil Kshape = 1. amplification factor for the long Note that in Equation (4) there is period range, that is the 1.0 second no modification assumed for skew period amplification factor, Fv, g iven and 3D effects. The structural dis- by NEHRP (note S = 1 for rock sites placements that occur for this dam- was assumed in deriving the “stan- age state are assumed to be small dard bridge” fragility curves). (generally less than 50 mm), thus the 3D arching effect is not engaged Scaling Relations for Damage since the deck joint gaps do not State 2 close. For slight damage, the median fragility curve parameter is given by Direct Economic the following equation: Losses

A2 = Kshape a2/S (4) Based on the work of Mander and Basöz (1999), the damage ratios where a is the PGA level given in 2 listed in Table 5 can be used to esti- table 3; S is the soil amplification mate the extent of damage ex- factor; and K is defined by the shape pected as a result of an earthquake. following equation: The best mean repair cost ratio for

K = 2.5C /C (5) “complete” damage—that is RCRi=5 shape v a for damage state 5—is defined as a In Equation (5), the factor 2.5 is function of number of spans as the ratio between the spectral given below: amplitude at 1.0 second (Ca) and 0.3 =≤ seconds (Cv) for the standard code- RCRi=5 210 n; . (6) based spectral shape for which the

Fragility Curve Development97 where n is the number of spans in bridge, the expected direct the main portion of the bridge. In monetary dollar loss can be this equation, it is assumed that the assessed. Work on this aspect is most common failure mechanism ongoing. will result in unseating of, at most, two spans simultaneously. The total repair cost ratio—that Illustrative Example is the expected proportion of the Problem total replacement cost of the entire bridge resulting from earthquake Consider a three-span simply damage, or the direct loss supported prestressed concrete probability—is defined as follows: bridge located in the Memphis area on very dense soil and soft rock 5 =⋅ < (site class C). Table 6 lists the data RCRTii∑( RCR P[] DS|. Sa) 10 i=2 (7) for this bridge obtained from NBI. The following ground motion data is assumed for a scenario earth- where PDS[]ia| S is the proba- quake: S (T = 0.3 sec) = 1.4 g, (S = bility of being in damage state DSi a 1.0); S (T = 1.0 sec) = 0.28 g, (S = for a given spectral acceleration, Sa, a for a structural period of T = 1.0 1.52); and A = 0.35 g. seconds; and ■ Table 6. Bridge Data Necessary for the Example Analysis RCR is the i Solution dleiFIBN ataD skrameR repair cost Since the bridge was constructed 287 1t69 liubraeY ratio for the ith damage in 1968 and is located outside of 384 5weksfoelgnA mode. If this California, the first and the fifth 413 5n0 apselpmis,etercnocdessertserP total repair rows of Table 3 apply, respectively,

435 napsforebmuN s cost ratio is thus for type 501: a2 = 0.26 g; a3 = 0.35 g; a = 0.44 g; and a = 0.65 g. 438 2)m(htgnelnapsmumixaM multiplied 4 5 by the re- Note that no * is used for a2; this 469 5)m(htgnelegdirblatoT placement implies “long periods” always gov- 502 1)m(htdiwegdirB ern, therefore = 1. cost of the Kshape

025..025 K =+1 =+1 = 1.; 125 1 3D n − 1 31− 0.9 ==␣ = K skew sin sin58 0 . 92 0.8

] 0.7 (8) a S

i 0.6 minor-analytical From Equation (4) A = 0.17, and moderate-analytical 2 0.5 major-analytical Equation (3) Ai = 0.681ai, thus: A3 = 0.4 collapse-analytical

D > = ds 0.24 g; A = 0.30 g; and A = 0.44 g.

[ 4 5 P 0.3 As shown in Figure 3, the prob- 0.2 ability of being in a given damage 0.1 state, when Sa (T = 1 sec) = 0.28␣ g, 0 is given in Table 7. This table also 0 0.2 0.4 0.6 0.8 1 1.2 1.4 presents the repair cost ratios from Spectral Acceleration (g) which the expected loss ratios are ■ Figure 3. Fragility Curves for the Example Prestressed Concrete Bridge determined in terms of the total with Simply Supported Girders

98 ■ Table 7. Analysis of Direct Loss Probability

P[D> SD |S ] P[ SD |S ] i i a i a RCR i tcudorP )1( )2( )3( )4( )3( X )4(

1 1 302.0 00.0 00000.0 2 797.0 691.0 30.0 88500.0

3 106.0 741.0 80.0 67110.0

4 454.0 822.0 52.0 00750.0 5 622.0 622.0 76.0 24151.0

latoT :seitilibaborp 000.1 RCR T = 622.0 replacement cost for the entire mechanics. However, where the bridge. From Table 7, the total loss rapid screening and detailed approaches differ is in the extent of ratio is RCRT␣ =␣ 0.226. data gathering, and time and effort necessary to perform a seismic vul- Conclusions and nerability analysis. The rapid Future Work screening approach operates on limited data as it is intended for Previous editions of the seismic evaluating a suite of bridges and retrofitting manual have used an ranking them in order of seismic indexing method as part of the vulnerability. On the other hand, a screening approach. Indeed there detailed analysis is intended to be are many such methods available a more exacting assessment of in- that have been used by various dividual bridges and the vulnerabil- state/owner agencies. The method ity of individual components. presented herein, however, is a new In the future, it is intended to ex- development. It is the same method tend the loss estimation ratios to that is used in Volume I of the new include direct and indirect losses in Manual that is concerned with the dollar terms. Direct losses arise post-earthquake integrity of an en- from damage to the bridge struc- tire highway system. The method ture itself, whereas indirect losses has also been adopted as the future may arise as a result of collapse re- approach for defining fragility sulting in the loss of life and limb. curves in HAZUS. The fragility These parameters can be used as a curve-based rapid screening basis for sorting and assigning method is consistent with the de- retrofit/repair/rehabilitation priori- tailed seismic evaluation approach ties. The choice of sorting strategy adopted as it is derived from the can be left to the value-system that same theoretical basis that is is adopted by the owning agency founded upon the fundamentals of and/or underwriting authority.

Fragility Curve Development99 References

Basöz, N. and Kiremidjian, A.S., (1998), Evaluation of Bridge Damage Data from the Loma Prieta and Northridge, CA Earthquakes, Technical Report MCEER-98-0004, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Basöz, N. and Mander, J.B., (1999), Enhancement of the Highway Transportation Lifeline Module in HAZUS, Final Pre-Publication Draft (#7) prepared for the National Institute of Building Sciences, March 31, 1999.

Dutta, A., (1999), On Energy-Based Seismic Analysis and Design of Highway Bridges, Ph.D. Dissertation, Science and Engineering Library, University at Buffalo, State University of New York, Buffalo, New York.

Dutta, A. and Mander, J.B., (1998), “Seismic Fragility Analysis of Highway Bridges,” INCEDE- MCEER Center-to-Center Workshop on Earthquake Engineering Frontiers in Transportation Systems, Tokyo, Japan, June. HAZUS, (1997), Earthquake Loss Estimation Methodology, Technical Manual, National Institute of Building Sciences for the Federal Emergency Management Agency.

Pekcan, G., (1998), Design of Seismic Energy Dissipation Systems for Concrete and Steel Structures, Ph.D. Dissertation, University at Buffalo, State University of New York, Buffalo, New York. Werner, S.D., Taylor, C.E., Moore, J.E., Jr., Mander, J.B., Jernigan, J.B. and Hwang, H.M., (1998), “New Developments in Seismic Risk Analysis Highway Systems,” 30th UJNR Panel Meeting on Wind and Seismic Effects, Gaithersburg, Maryland. Werner, S.D., Taylor, C.E., Moore, J.E. and Walton, J.S., (1999), Seismic Retrofitting Manuals for Highway Systems, Volume I, Seismic Risk Analysis of Highway Systems, and Technical Report for Volume I, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

100 Changes in the New AASHTO Guide Specifications for Seismic Isolation Design

by Michael C. Constantinou, University at Buffalo, State University of New York

Research Objectives MCEER Highway Project Two projects at the University at Buffalo under sponsorship of MCEER/NSF MCEER concentrated, respectively, on establishing new values of response modification factors for substructures of seismically isolated bridges, and on the study of the longevity and reliability of seismic isolation hardware. The latter culminated in the development of the concept of system property modification factors. This concept and the new values of response modification factors have been implemented in the new AASHTO Guide Specifications for Seismic Isola- AASHTO T-3 Task Group tion Design. James E. Roberts, Chairman, California Department of Transportation Roberto Lacalle and n 1993, a project began at the University at Buffalo under the title Li-Hong Sheng, Working I“Longevity and Reliability of Sliding Seismic Isolation Systems” with the Chairman/Co-chair, support of MCEER. The objectives of the project were then defined as the California Department of collection of laboratory and field data on the behavior of sliding bearings Transportation and the qualitative prediction of the long-term frictional properties of these Ralph E. Anderson, Illinois bearings. In 1995, the author of this paper became involved in the devel- Department of Transporta- opment of the new AASHTO Guide Specifications for Seismic Isolation tion Paul Bradford, R.J. Watson, as a member of a task group of the T-3 Seismic Design Technical Design Inc. Committee of the AASHTO Bridge Committee. Specific challenges for the Michael C. Constantinou, T-3 task group were the proposal of new response modification factors for University at Buffalo, State bridge substructures and the justification thereof, and the development of University of New York a rational procedure for determining bounding values of isolator proper- Hamid Ghasemi, Federal ties for analysis and design. Highway Administration Based on the needs of the T-3 task group, the objectives of the research James M. Kelly and Ian project were modified to include the development of a procedure for es- Aiken, University of California, Berkeley tablishing bounding values of isolator properties. Moreover, a new project Myint M. Lwin, Washington began in 1996 at the University at Buffalo with the support of MCEER to Department of Transporta- develop appropriate response modification factors for the substructures of tion seismically isolated bridges. These efforts culminated in the establishment Ronald L. Mayes, Dynamic of the concept of System Property Modification Factors, the development Isolation Systems of revised values for response modification factors, and the inclusion of both John F. Stanton, University in the new AASHTO Guide Specifications, which were published in 1999 of Washington Victor A. Zayas, Earthquake (American Association of Highway and Transportation Officials, 1999). Protection Systems

101 Changes in the New to an increase in the isolation system displacement. These phe- AASHTO Guide nomena have been convincingly • Development of Earthquake Specifications for demonstrated in NCEER-funded Protective Systems for Seismic Isolation research (Constantinou et al., Bridge Retrofit: Stability of 1993, Tsopelas et al., 1994). Elastomeric Bearings, I.G. Design Buckle, University of • The requirements for sufficient Auckland, Energy Dissipa- The new specifications were tion Systems, M. developed by the T-3 task group lateral restoring force have been Shinozuka, University of during the period of 1995 to 1997 by changed so that the use of isola- Southern California, considering the then current state-of- tion systems with very low Comparative Assessment, practice and the results of completed restoring force is disallowed in M. Feng, University of and ongoing research efforts. A order to prevent the accumula- California, Irvine number of changes in the new tion of large permanent displace- • Evaluation of Mechanical specifications over the predecessor ments and to reduce the Property Changes for HDNR specifications of 1991 are signif- sensitivity of the displacement Bearings Over Time, icant, either because they drastically response to the details of the J. Kelly,University of change the analysis and design seismic input. Experimental re- California, Berkeley procedures or because they impose sults from another NCEER- funded project (Tsopelas and • Field Testing of a constraints that limit the appli- Constantinou, 1994) was the Seismically Isolated Bridge, cation of some isolation systems. impetus for the implementation S. Chen and J. Mander, Some of the changes are: University at Buffalo, and of this change. B. Douglas, University of • The methods of analysis have Nevada-Reno been modified to include the • The response modification fac- effect of the flexibility of the sub- tors (R-factors) for the substruc- structure. The substructure in- ture of isolated bridges has been creases the flexibility of the reduced so that, effectively, the structural system and results in a substructure remains elastic. damping ratio that is less than The T-3 task group endorsed a that of the isolation system (pro- proposal by the author and vided that there is no inelastic reduced the R-factor on the ba- action in the substructure). The sis of a small number of analyti- result is a net increase in the cal results and engineering displacement of the structural judgement. Research conducted system, which usually is related in the meantime established the

Results of this research have been included in the AASHTO Guide Specifications for Seismic Isolation Design, published in 1999, where they represent the two major changes over the predecessor 1991 Specifications. The concept of system property modification factors is considered an innovation in the design of seismically isolated bridges and has been proposed for inclusion in the Structural Engineers Association of Cali- fornia Blue Book and the NEHRP Recommended Provisions, which apply for buildings. It is expected that this design concept will be both mandated and regularly used in the design of seismically isolated building and bridge structures.

102 necessity for lower R-factors on where R␮ is the ductility-based por-

the basis of a comprehensive tion and R␱ is the overstrength fac- analysis, and verified the appro- tor. The ductility-based portion is priateness of the selected values the result of inelastic action in the (Constantinou and Quarshie, system. The overstrength factor is 1998). the result of reserve strength that exists between the design force and • A procedure for determining the actual yield strength. Single bounding values of the isolator column substructures of bridges properties for analysis and have no overstrength (that is, R␱ = design has been included. This 1.0), whereas multiple column bent procedure is based on the de- substructures have overstrength termination of system prop- which typically is assumed to cor- erty modification factors, or respond to R = 1.67. ␭ ␱ -factors, which account for The ductility-based portion of the the effects of aging, environ- R-factor has been presumed to be ment, contamination, history related to the ability of the substruc- of loading and other conditions ture to undergo inelastic action. on the mechanical properties of Accordingly, the original 1991 isolators. The concept repre- AASHTO Guide Specifications for sents a drastic departure from Seismic Isolation Design specified previous practice and it is a R-factors that were identical to bold procedure for consid- those specified for the substruc- ering the long-term behavior of tures of conventional, non-isolated the isolators rather than just their short-term performance bridges. The assumption was thus in the laboratory. The concept, made that the inelastic demand in together with an extensive col- the substructures of seismically iso- lection of data to support it, has lated and non-isolated bridges been the result of a long- would be the same if the two were term MCEER-funded project designed for the same R-factor. (Constantinou et al., 1999). This presumption was incorrect. The demand in the substructure of seismically isolated bridges is Response strongly dependent on the relation Modification Factor between the strength of the isolation system and the strength of the Response modification factors substructure. It is apparent that the (R-factors) are used to calculate the strength of the substructure should design forces in the substructures be higher than that of the isolation of bridges from the elastic force system, or otherwise the isolation demand. That is, the demand is system becomes totally ineffective. calculated on the assumption of This principle has been convinc- elastic substructure behavior and ingly demonstrated in a series of subsequently the design forces are simple examples by Constantinou established by dividing the elastic and Quarshie (1998), who also per- force demand by the R-factor. formed a systematic study for estab- The R-factor consists of two lishing appropriate values for the components. That is, R-factor.

R = R␮ · R␱ (1)

Changes to the AASHTO Guide Specifications103 Results of Analysis on • For a particular combination of parameters characterizing the R-factors for system, analysis was performed Seismically Isolated assuming elastic substructure Bridges behavior and utilizing the simplified analysis procedures of Figure 1 shows a simple deck-iso- AASHTO. lation system-substructure model used in the study of Constantinou • The strength of the substructure and Quarshie (1998). A variety of was then established as the cal- behaviors for the isolation system culated elastic force demand and substructure are shown in Fig- divided by the R-factor. The lat- ure 1, and a range of parameters ter is now just the ductility-based were considered in the dynamic portion since the system lacks analysis of this system. Analysis was redundancy. performed as follows: • The system was then analyzed and its nonlinear response his-

u tory was calculated for 20 earth- Tributary quake motions which were Deck W appropriately scaled to repre- Weight d Isolation sent the applicable response System Effective W spectrum. Pier Weight p u b • The analysis results were used to calculate, among other quan- Pier tities, the average displacement ductility ratio in the substructure which provided the most useful up information in establishing ap-

ISOLATION SYSTEM propriate values of the R-factor. Table 1 presents a summary of F Fb F b b C u K K b b selected results from these analyses. Qb b Qb b Kb To appreciate the level of inelastic ~ Ki = 10Kb action in the substructure of the u u Yb ub b b isolated bridges, analyses were also (a) Bilinear Hysteretic (b) Sliding (c) Linear Elastic- performed for non-isolated bridges. Linear Viscous Table 2 presents a sample of results PIER obtained from such analyses, which F F p Fyp p Fyp is appropriate to compare with the ␣K ␣K p p sample of results in Table 1. Such a K Kp p comparison reveals that the ductil- u Y 2as u ity ratio in the substructure of iso- Yp p p p lated bridges is more, actually much (d) Perfect Bilinear (e) Pinched with Slip Hysteretic more, than that of non-isolated bridges when both are designed for ■ Figure 1. Analyzed System and Illustration of Utilized Force-Displacement the same R-factor. Relations for Isolation System and Pier

104 ■ Table 1. Average Substructure Displacement Ductility Ratio of Isolated Bridges

metsyS R␮ 0.1= R␮ 5.1= ,IIepyTlioS,4.0=A,reiPciteretsyHraeniliB 8.1-2.1 7.4-4.2 metsySnoitalosIciteretsyHraeniliB ␦ 60.0= ,IIepyTlioS,4.0=A,reiPciteretsyHraeniliB 1.2-3.1 5.6-8.2 metsySnoitalosIciteretsyHraeniliB ␦ 01.0=

,IIepyTlioS,4.0=A,reiPciteretsyHraeniliB 6.1-4.1 0.4-0.3 metsySnoitalosIsuocsiV/citsalEraeniL ␰ 2.0= ,IIepyTlioS,4.0=A,reiPciteretsyHraeniliB 5.1-4.1 2.4-0.3 metsySnoitalosIsuocsiV/citsalEraeniL ␰ 3.0= ,IIIepyTlioS,4.0=A,reiPciteretsyHraeniliB 4.1-9.0 3.4-2.2 metsySnoitalosIciteretsyHraeniliB ␦ 60.0= ,IIIepyTlioS,4.0=A,reiPciteretsyHraeniliB 5.1-9.0 9.4-9.1 metsySnoitalosIciteretsyHraeniliB ␦ 01.0= ,IIIepyTlioS,4.0=A,reiPciteretsyHraeniliB 6.1-9.0 9.3-7.1 metsySnoitalosIsuocsiV/citsalEraeniL ␰ 2.0= ,IIIepyTlioS,4.0=A,reiPciteretsyHraeniliB 3.1-9.0 0.4-2.2 metsySnoitalosIsuocsiV/citsalEraeniL ␰ 3.0= ,IIepyTlioS,4.0=A,reiPciteretsyHdehcniP 1.2-5.1 4.5-8.2 metsySnoitalosIciteretsyHraeniliB ␦ 60.0= ,IIepyTlioS,4.0=A,reiPciteretsyHdehcniP 4.2-5.1 7.6-1.3 metsySnoitalosIciteretsyHraeniliB ␦ 01.0=

■ Table 2. Average Substructure Displacement Ductility Ratio of Non-Isolated Bridges

metsyS R␮ 0.1= R␮ 0.2= R␮ 0.3=

IIepyTlioS,4.0=AreiPciteretsyHraeniliB 9.0 9.1 5.3

*IIIepyTlioS,4.0=AreiPciteretsyHraeniliB 4.1 1.3 2.5

IIepyTlioS,4.0=AreiPciteretsyHdehcniP 9.0 9.2 6.4

seulavevitavresnoC*

Changes to the AASHTO Guide Specifications105 On the basis of ■ Table 3. Proposed Values of R-factor for Substructures of “It is expected such comparisons, Isolated Bridges and additional re- that the system erutcurtsbuS R␮ R R sults on the sensitiv- o property gnortS(reiPepyT-llaW ity of the substruc- 0.1 76.1 76.1 modification )noitceriD ture inelastic re- ,noitceriDkaeW(reiPepyT-llaW factors concept 5.1 0.1 5.1 sponse of isolated )nmuloCasadengiseD will be both bridges, it was con- snmuloCelgniS 5.1 0.1 5.1 mandated and cluded that the duc- tneBnmuloCelpitluM 5.1 76.1 5.2 regularly used tility-based portion in the design of the substructures could be established on the basis of seismically of isolated bridges should be less of statistical analysis of the variabil- isolated than or equal to 1.5. Moreover, ity of the properties and the likeli- building and analyses of the overstrength in iso- hood of occurrence of relevant bridge lated bridges have shown that, in events, including that of the consid- structures.” general, the overstrength is slightly ered seismic excitation. This is an higher than in non-isolated bridges. admittedly very difficult problem. Finally, values of the R-factor for iso- However, it is relatively easier to lated bridges have been established assess the effect of a particular and are presented in Table 3. Nearly phenomenon on the properties of identical values (1.5 instead of 1.67) a selected type of bearing, either by have been included in the new testing (e.g., effect of temperature AASHTO Guide Specifications for on friction coefficient in sliding Seismic Isolation Design. bearings) or by a combination of testing, rational analysis and engi- System Property neering judgement (e.g., effect of aging). This leads to the establish- Modification Factors ment of system property modifica- The properties of seismic isolation factors, that is, factors which tion bearings vary due to the effects quantify the effect of a particular of wear, aging, temperature, history phenomenon on the nominal prop- of loading, and so on. The exact erties of an isolation bearing, or system in general. state of the bearings at the time of Consider that a nominal value of seismic excitation cannot be a property of an isolation system is known. However, it is possible to known. It could be that this value establish maximum and minimum is assumed (on the basis of experi- probable values of important prop- ence from previous testing) during erties (i.e., characteristic strength the analysis and design phase of the and post-yielding stiffness) within project or it is determined in the the lifetime of the structure. The prototype bearing testing. Typically, analysis can then be conducted this nominal value applies for spe- twice using the bounding values of cific conditions, such as fresh bear- properties. In general, the maximum ing conditions, temperature of 20ЊC force and displacement responses and the relevant conditions of ver- will be obtained in these analyses. tical load, frequency or velocity and In principle, the probable maxi- strain or displacement. Let this mum and minimum property values value be Pn.

106 The minimum and maximum val- the tested temperature to the co-

ues of this property, Pmax and Pmin efficient of friction at the reference respectively, are defined as the temperature (say 20ЊC). Factor product of the nominal value and a ␭ min,␣ t will be based on the data for series of System Property Modifica- the highest temperature (50ЊC), tion Factors, or ␭-factors as follows: ␭ whereas max,␣ t will be based on the data for the lowest temperature ␭ ⅐ Њ Pmax = max Pn (2) (-10 C). ␭ ⅐ As another example, consider Pmin = min Pn (3) the effect of wear on the friction where coefficient. On the basis of the geometric characteristics of the ␭ = ␭ ⅐ ␭ ⅐ ␭ ⅐ ⅐ ⅐ (4) bridge (span, girder depth, etc.), max max, 1 max, 2 max, 3 average vehicle crossing rate and ␭ = ␭ ⅐ ␭ ⅐ ␭ ⅐ ⅐ ⅐ (5) min min, 1 min, 2 min, 3 lifetime of the structure, the cumulative travel is determined. Test Each of the ␭ , i = 1,2 ⅐⅐⅐ max,i data are then utilized to establish factors is larger than or equal to the ␭-factors for wear (or travel). unity, whereas each of the ␭ , min, i Typically, ␭ is the ratio of the ⅐⅐⅐ max,␣ tr i = 1,2␣ is less or equal to unity. coefficients of friction determined ␭ Moreover, each of the -factors is in high velocity testing following associated with a different as- to and prior to a sustained test at pect of the isolation system, such the appropriate velocity (~1mm/s) as wear, contamination, aging, for a total movement equal to the history of loading, temperature, calculated cumulative travel. The and so on. ␭ min,␣ tr is determined in a similar As an example, consider the effect manner but for a total movement of temperature on the friction co- less than the calculated cumulative efficient of a sliding bearing. travel for which the coefficient of The range of temperature over the friction attains its least value. lifetime of the structure is first es- The system property modifica- tablished for the particular site or tion factors are associated with general geographic area of the different aspects of the isolation project. This range need not be system and combined on the basis one of the extreme (lowest and of (4) and (5). While each one of highest) temperatures. Rather, it these factors describes the range could be a representative range de- of effect of a particular aspect, termined by the responsible pro- their multiplication results in a fessional (more appropriately, this combined factor of which the range could be included in the ap- value may be very conservative. plicable specifications). Say this That is, the probability that several range of temperature is -10ЊC to events (such as lowest tempera- 50ЊC. Testing is then performed at ture, maximum travel, maximum the two temperatures and the corrosion, etc.) occur simulta- ␭-factors are established as the ra- neously with the design-basis tio of the coefficient of friction at earthquake is very small.

Changes to the AASHTO Guide Specifications107 It is necessary that some adjustment factors have been proposed ment of the system property modi- by the author and included in the fication factors is applied to reflect new AASHTO Guide Specifica- the de-sired degree of conserva- tions: tism. This adjustment should be 1 for critical bridges based on a statistical analysis of the 0.75 for essential bridges property variations with time, the 0.66 for all other bridges probability of occurrence of joint events and the significance of the These values are based on engi- structure. It is also desirable to neering judgement and a desire to apply this adjustment with the sim- employ the most conservative de- plest possible procedure. sign approach for critical bridges. Such a procedure is based on sys- It is expected that as experience tem property adjustment factors, develops over the years of obser- a, such that the adjusted value of vation of the performance of the ␭-factor is given by seismically isolated bridges and other structures, and more data are adjusted collected on the variations of prop- ␭ ␭ ⅐ max = 1 + ( max - 1) a (6) erties, more refined values of sys- adjusted tem property adjustment factors ␭ = 1 + (1 - ␭ ) ⅐ a (7) could be established. min min Values of ␭-factors have been es- That is, the property adjustment tablished for sliding and elasto- factor is multiplied by the amount meric isolation systems on the basis by which the ␭-factor differs from of a long-term study which included unity and the result is added to a comprehensive review and analy- unity to yield the adjusted ␭-factor. sis of available data, extensive test- It is evident that the adjustment ing, application of principles of factor can take values in the range solid mechanics and use of engi- of 0 to 1. The value a = 0 results neering judgement (Constantinou in an adjusted ␭-factor of unity et al., 1999). There are too many (that is, variations in properties are disre- Unfilled PTFE garded – least con- Breakaway Sliding 0.25 servative approach). Pressure = 20.7 MPa The value a = 1 re- 0.20 sults in no adjust-

ment (that is, the 0.15 maximum variations

are considered to 0.10 -40 C -30 C occur simultaneously -20 C -10 C 0 C – most conservative Coefficient of Friction 0.05 +10 C +20 C approach). +50 C 0.00 The following sys- -0 1 234 50 50 100 150 200 250 300 350 tem property adjust- Velocity (mm/sec)

■ Figure 2. Friction of Unfilled PTFE-Polished Stainless Steel Interfaces at Various Temperatures as Function of Sliding Velocity

108 ■ Table 4. System Property Modification Factor for Effects of Tempera- the increased ve- ␭ ture ( max, t) on the Coefficient of Friction of Sliding Bearings locity, is apparent. On the basis of erutarepmeT detacirbulnU detacirbuL cillatemiB (Њ )C EFTP EFTP secafretnI these and other results, which were 200 10. 1A. /N 0113. 1A. /N generated in very -201 15. 1A. /N time-consuming -503 10. 3A. /N experiments, the -704 1A. NA/ /N ␭-factors of Table 4 -005 2A. NA/ /N were developed (Constantinou et factors to describe each one and al., 1999) and incorporated in the the physical phenomena respon- new AASHTO Guide Specifications. sible for the effects. It is sufficient To assess the effect of frictional to present herein some represen- heating, an analytic solution was tative results on one of the effects derived to predict the temperature and the related ␭-factors. rise at the sliding interface and at Low temperature causes an in- some depth below. Carefully crease in the friction of PTFE- planned experiments (including stainless steel interfaces used in the use of extremely fine thermo- sliding bearings and in the stiff- couple wires) were also conducted ness and characteristic strength of to obtain reliable measurements elastomeric bearings. The effect in of histories of temperature rise the case of elastomers is time- dependent, that is, the increase in stiffness is greater with increasing 50 45 Recorded time of exposure at a particular Predicted low temperature. For sliding inter- 40 35 faces, the effect of low temperature 30 is highly dependent on the speed 25 Predicted Peak f = 0.13 Hz Њ u = 96.5 mm of sliding motion since frictional 20 Surface Temperature = 54 C s 0 102030405060 heating can cause substantial in- 60 creases in temperature following very small travel. 50 While testing of seismic isolation 40 bearings at low temperature is a 30 Predicted Peak f = 0.26 Hz relatively straight forward exercise Surface Temperature = 65.7Њ C u = 96.5 mm 20 s (albeit not an easy one), the inter- Temperature ( Њ C) 5 10152025303540 pretation of the results and the es- 80 ␭ Plate Uplift,Thermocouple Sliding tablishment of -factors requires an 70 understanding of the frictional 60 heating problem. Figure 2 presents 50 a sample of experimental results on 40 30 Predicted Peak f = 0.53 Hz the frictional properties of unfilled Surface Temperature = 87Њ C us = 96.5 mm 20 PTFE-highly polished stainless steel 81012 14 16 18 20 22 interfaces for a range of velocities Time (sec) of sliding, and temperature at the start of the experiment. The sub- ■ Figure 3. Recorded and Predicted Histories of Temperature at stantial effect of frictional heating, Depth of 1.5 mm Below the Sliding Interface in Large Amplitude as made evident in the figure with Tests

Changes to the AASHTO Guide Specifications109 for verification of the theory. Fig- Isolation Design, which were pub- ure 3 presents a comparison of mea- lished in 1999. sured and predicted histories of Of particular interest in this work temperature during testing of a slid- is the development of the concept ing bearing. of system property modification factors and the substantial multi- year effort to establish values for Conclusions these factors on the basis of testing, rational analysis and engineer- Research supported by MCEER ing judgement. Despite this effort, resulted in the establishment of there are a significant number of new response modification factors problems that could not be ad- for the substructures of seismically equately addressed in this research. isolated bridges and in the devel- Most important of these problems opment of a new concept in the is that of the prediction of the analysis seismically isolated struc- aging characteristics of seismic iso- tures. lation hardware, which requires a Both developments have been substantial, multidisciplinary ba- incorporated in the new AASHTO sic research effort to adequately Guide Specifications for Seismic address.

References

American Association of State Highway and Transportation Officials (AASHTO), (1999), Guide Specifications for Seismic Isolation Design, Washington, D.C. Constantinou, M.C., Tsopelas, P., Kim, Y-S and Okamoto, S., (1993), NCEER-Taisei Corporation Research Program on Sliding Seismic Isolation Systems for Bridges: Experimental and Analytical Study of a Friction Pendulum System (FPS), Technical Report NCEER- 93-0020, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Constantinou, M.C. and Quarshie, J.K., (1998), Response Modification Factors for Seismically Isolated Bridges, Technical Report MCEER-98-0014, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Constantinou, M.C., Tsopelas, P., Kasalanati, A. and Wolff, E., (1999), Property Modification Factors for Seismic Isolation Bearings, Technical Report MCEER-99-0012, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Tsopelas, P. and Constantinou, M.C., (1994), NCEER-Taisei Corporation Research Program on Sliding Seismic Isolation Systems for Bridges: Experimental and Analytical Study of a System Consisting of Lubricated PTFE Sliding Bearings and Mild Steel Dampers, Technical Report NCEER-94-0022, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Tsopelas, P., Okamoto, S., Constantinou, M.C., Ozaki, D. and Fujii, S., (1994), NCEER-Taisei Corporation Research Program on Sliding Seismic Isolation Systems for Bridges: Experimental and Analytical Study of Systems Consisting of Sliding Bearings, Rubber Restoring Force Devices and Fluid Dampers, Technical Report NCEER-94-0002, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

110 Seismic Retrofitting Manuals for Highway Systems

by Ian M. Friedland, Multidisciplinary Center for Earthquake Engineering Research and Ian G. Buckle, University of Auckland

Research Objectives MCEER Highway Project The MCEER Highway Project is developing a set of guidelines for conducting seismic hazard evaluations and retrofitting of highway systems and system components. These guidelines will provide assis- tance in performing seismic vulnerability screening, evaluation and retrofitting of highway system components including bridges, retaining structures, slopes, tunnels, culverts, and pavements, and in determining the impacts of scenario earthquakes on highway system performance, including traffic flows, economic consequences and other factors. The guidelines will be contained in the three-volume Seismic Retrofitting Manuals for Highway Systems. Volume I will contain procedures for conducting a seismic risk assessment of highway systems Ian G. Buckle, University of and regional networks, Volume II will contain procedures for seis- Auckland mic screening, evaluation and retrofitting of highway bridges, and Ian M. Friedland, MCEER Volume III will contain similar procedures for retaining structures, Maurice S. Power, slopes, tunnels, culverts and pavements. Geomatrix Consultants, Inc. Stuart D. Werner, Seismic Systems and Engineering n the fall of 1992, MCEER initiated work on a comprehensive research Consultants I program, sponsored by the Federal Highway Administration, to develop and improve tools to evaluate the seismic vulnerability of the national highway system in the United States. The research program includes a series of tasks intended to improve understanding of the seismic hazard in the eastern and central U.S.; the behavior and response of soils, foundations and highway structures during earthquakes; and the overall impacts on highway systems resulting from earthquake damage. In addition, the program is developing and improving appropriate retrofit technologies for those highway system components deemed vulnerable to earthquake damage. An important end-product of this program is the development of seismic evaluation and retrofitting guidelines for existing highway systems which will have national applicability. These guidelines are being prepared in three volumes, where: • Volume I contains the methodologies, procedures, and examples for conducting a seismic risk assessment of highway networks and systems;

111 • Volume II contains guidance on also prepared by MCEER, served as seismic vulnerability screening, the “strawman” for Volume II. Work analysis, and retrofitting of high- on the final versions of each volume Volume 1 Contributors way bridges; and commenced in late 1997; drafts of Stuart D. Werner, Coordina- • Volume III contains guidance on each have been completed and tor, Seismic Systems and screening, analysis, and retrofit- distributed to a select group of Engineering Consultants ting of other major highway researchers and practitioners for Craig E. Taylor, Natural system components, including re- Hazards Management, Inc. detailed technical reviews. Follow- James E. Moore II, Univer- taining structures, slopes, tunnels, ing final revisions and edits, it is sity of Southern California culverts, and pavements. anticipated that each of these three Masanobu Shinozuka, volumes will be published and avail- University of Southern As a first step towards the completion of each of these volumes, a able for distribution by the FHWA California in the year 2000. Jon Walton, City of San Jose “strawman” was prepared in late 1996 for Volumes I (system risk This paper provides an overview Volume II Contributors assessment) and III (evaluation and on the intent and coverage of each Ian G. Buckle, Coordinator, retrofitting of components other volume of these retrofitting guide- University of Auckland lines. Ian M. Friedland, MCEER than bridges). Each strawman: John B. Mander, University 1. Summarized the current state-of- at Buffalo practice in highway system Seismic Risk Geoffrey R. Martin, University of Southern assessment, component evalua- Assessment of California tion, and retrofitting. Richard V. Nutt, Consultant 2. Incorporated current research Highway Systems Maurice S. Power, results. Past experience has shown that Geomatrix Consultants, 3. Identified important gaps in the direct impacts of earthquake Inc. knowledge which required reso- damage to highway structures (e.g., Volume III Contributors lution prior to completion of bridges, retaining structures, and Maurice S. Power, Coordina- each volume. tunnels) is both a life-safety issue tor, Geomatrix Consultants, and a repair and replacement cost The 1995 FHWA Seismic Retro- Inc. issue. Furthermore, indirect im- Geoffrey R. Martin, Co- fitting Manual for Highway pacts from closed or restricted coordinator, University of Bridges (FHWA, 1995), which was Southern California Rowland Richards, University at Buffalo Kenneth Fishman, Univer- The Seismic Retrofitting Manuals for Highway Systems will sity at Buffalo provide engineers with necessary knowledge concerning cur- Faiz Makdisi, Geomatrix rent state of the art regarding the performance of bridge and Consultants, Inc. other highway structures during earthquakes, methods to Dario Rosidi, Geomatrix strengthen them, and tools to determine how and when to Consultants, Inc. Jon Kaneshiro, Parsons retrofit. These manuals are intended to be used with the Engineering Science, Inc. seismic design requirements for new highway bridges con- Samuel C. Musser, Utah tained in the American Association of State Highway and Department of Transporta- Transportation Officials (AASHTO) Specifications. Private, tion local, state and federal bridge and highway engineers T. Leslie Youd, Brigham throughout the country will find these manuals useful in de- Young University veloping retrofit strategies. Emergency managers and planning professionals can use these manuals to determine appropriate levels of lifeline route planning and redundancy.

112 routes, disrupted access and de- screening and ranking criteria ap- layed post-earthquake emergency plied specifically to the inventory “This is the response, repair, and reconstruction of structures of the same type. For first known operations can be as costly or example, formal screening and effort to capture greater than the actual direct struc- ranking guidelines for highway the important tural damage and repair costs. bridges have existed since the early aspects of The extent of these impacts 1980s (FHWA, 1983). However, depends not only on the seismic such procedures treat each indi- screening, performance characteristics of the vidual component independently, evaluation, and individual components, but also on without regard to how the extent retrofitting of the characteristics of the highway of its damage may affect overall non-bridge system that contains these com- highway performance. highway system ponents. For example, studies of Consideration of each com- structural highway systems in the San Francisco ponent’s importance to system components and Bay area after the 1989 Loma Prieta performance can provide a much to present earthquake have shown that post- more rational basis for: results and earthquake traffic flows strongly depended on the following factors • establishing seismic strengthen- recommendations (Hobeika et al., 1991; Wakabayashi ing priorities; in a formal, and Kameda, 1992): • defining seismic design and procedural strengthening criteria; manner.” • the configuration of the highway • effecting emergency lifeline system; route planning; and • the locations of the individual • estimating economic impacts components within the overall due to component damage. system and within specific links and subsystems; and It is important to recognize that • the locations, redundancy, and system performance issues are im- traffic capacities and volumes of portant to all regions that are at risk the links between key origins and due to earthquakes. As a result, sys- destinations within the system. tem issues are now being incorpo- When one considers these system rated into newly developed methods characteristics, it is evident that for prioritizing bridges for seismic earthquake damage to certain com- retrofit (Basöz and Kiremidjian, ponents (e.g., those along im- 1995; Moore et al., 1995). portant and nonredundant links Volume I of the MCEER Highway within the system) will have a Project’s Seismic Retrofitting greater impact on overall system Manuals for Highway Systems performance than will other contains procedures for conduct- components. Currently, such system ing a seismic risk assessment (SRA) issues are not considered when of highway networks and systems specifying seismic performance (Werner et al., 1999). The proce- requirements and design or retro- dures contained in this volume fitting criteria for new or existing provide a basis for addressing components, due primarily to the these seismic performance issues lack of adequate systems-based and incorporate data and method- evaluation tools. ologies pertaining to engineering Instead, each component type is issues (structural, geotechnical, usually evaluated individually, with and transportation), repair and

Seismic Retrofitting Manuals for Highway Systems113 reconstruction, system network • a socioeconomic module for and risk analysis, and socioeco- characterizing economic, emer- nomic considerations for impacts gency response, and societal im- from system damage. They also pacts of earthquake damage to provide a mechanism to estimate highway systems system-wide direct losses (i.e., costs for repair of damaged com- Key to the SRA procedures are ponents) and indirect losses due four GIS-based modules that act as to reduced traffic flows and/or in- pre-processors to the procedure in creased travel times (economic order to model the system, hazards, impacts). components, and socioeconomics for the system. This modeling com- Specifically, Volume I provides: prises the bulk of the effort in the • a detailed framework for carry- application of the risk analysis pro- ing out deterministic and proba- cedure. bilistic evaluations of seismic The volume provides the back- risks to highway systems; ground and an overview of the • a discussion on the types of data methodology and procedures, and details the four principal modules needed to characterize the comprising the SRA: system, hazards, and components, together with the form of • the system module, which con- structural, geotechnical, and tains system and inventory data, transportation engineering traffic management measures, analysis results needed for these and system analysis procedures; characterizations; • the hazards module, which con- • a procedure for rapid analysis of tains the earthquake ground post-earth- motions, geologic hazard evalu- quake traffic ation, liquefaction, landslides, flows based and surface fault rupture infor- on artificial mation; intelligence • the component module, which concepts and contains the overall model the current development including loss and state of knowl- functionality models, seismic re- edge for traffic sponse evaluation, and repair and flow model- reconstruction procedures; and ing; and • the socioeconomic module, ■ Figure 1. Procedures for con- which contains ducting a Seismic Risk Assessment the models and (SRA) of highway networks and local or regional systems is the focus of Volume I. demographic and The SRA provides a detailed frame- economic data work for modeling the system, needed to esti- hazards, components and socio- mate the socioeco- economic factors for a given region. nomic impacts Shown above are the Memphis/ due to reductions Shelby County highway system and in traffic flows the entire transportation system for a given post-earthquake state.

114 resulting from earthquake dam- in the state-of-the-art and the state- age. of-practice are the result of an ag- gressive research program which In addition, the volume contains was started by the California Depart- an example application based on ment of Transportation (Caltrans) the highway system in and around following the 1989 Loma Prieta Memphis, Tennessee which dem- earthquake. Since then, seismic onstrates the application and inter- screening, evaluation, and retrofit- pretation of the results of the SRA ting procedures for highway bridges procedure. have been widely implemented in parts of North America, Asia, and Screening, Evaluation Europe. and Retrofitting of In order to capture these ad- Highway Bridges vances in seismic In the late 1970s, the Applied retrofitting and Technology Council developed a to make the cur- set of guidelines for the seismic ret- rent state-of-the- rofitting of highway bridges under art available to FHWA sponsorship. These guide- bridge owners lines were published in 1983 by the and engineers FHWA as the Seismic Retrofitting across the U.S., Guidelines for Highway Bridges the FHWA initi- (FHWA, 1983). The guidelines rep- ated a project to resented what was then the state- update the 1983 of-the-art for screening, evaluating, and retrofitting of seismically deficient bridges.

At the time the guidelines ...... Steel were issued, experience ...... Plate with highway bridge retro- .. . . fitting in the U.S. was limited Detail A SectionA-A and many of the proposed Starter Bar techniques had not actually Main Bar been implemented in field Hexagonal Coupler Hexagonal Lock Nut applications. In the 15 or so Couplers years since, there has been ■ Figure 2. Innovative techniques for significant progress in under- Fuse Bar Details seismic retrofitting strategies are an standing the seismic re- integral part of MCEER’s highway sponse of bridges and the Original Rebar project research, the results of which development of new and im- are included in Volume II of the retrofit Steel Plate proved retrofitting technolo- Couplers manuals. A Control and Repairability gies for bridge columns and of Damage (CARD) column design Lock Nut footings, methods to stabilize philosophy for new structures that uses replaceable fuse bars in the plastic soils to prevent liquefaction, Original Rebars and to ensure adequate con- Cut at Midlength hinge zone is being tested for application in retrofit situations. It is nectivity between the bridge Fuse Bar anticipated that this type of retrofit will superstructure and substruc- DetailA permit rapid and cost-effective repairs ture. Many of these advances following a damaging earthquake.

Seismic Retrofitting Manuals for Highway Systems115 guidelines as part of the MCEER of additional research and develop- Highway Project research program. ment conducted under the FHWA- This effort resulted in the 1995 Seis- sponsored research program at • Risk Assessment of Highway mic Retrofitting Manual for High- MCEER, Caltrans, and others. Vol- Systems, S. Werner, way Bridges, which was published ume II will be issued in two parts, Seismic Systems and by the FHWA (FHWA, 1995). including: Engineering Consultants The 1995 FHWA manual offers • the main volume containing the • State of the Art Review: procedures for evaluating and up- screening, evaluation, and retro- Foundations and Retaining grading the seismic resistance of Structures, I.P. Lam, Earth fitting procedures for highway Mechanics, Inc.; Slopes, existing highway bridges. Specifi- cally, it contains: bridges, and a series of case-stud- Retaining Structures and ies demonstrating the applica- Pavements, G. Santana, University of Costa Rica; • a preliminary screening process tion of key parts of various Culverts and Soils and to identify and prioritize bridges procedures Liquefaction, T.L. Youd, that need to be evaluated for • a supplementary volume contain- Brigham Young University seismic retrofitting; ing engineering drawings and • Liquefaction Remediation • alternative methodologies for details of typical retrofits, and Techniques for Bridge quantitatively evaluating the vendor details and applications of Foundations, G. Martin, seismic capacity of an existing seismic protective systems. University of Southern bridge and determining the over- Among the major changes that California and J. Mitchell, all effectiveness of alternative Virginia Polytechnic are included in the new bridge seismic retrofitting measures by Institute evaluation and retrofitting manual either the component-based are: • Improving the Lateral capacity/demand (C/D) approach Capacity of Pile Founda- tions and Pile Foundation or the lateral strength (“push- 1. significantly expanded coverage Retrofitting Synthesis, I.P. over”) methodology; and of seismic hazards. This includes Lam, Earth Mechanics, Inc. • suggested retrofit measures and methods for characterizing the seismic hazard, selecting appro- • Pile-to-pile Cap Connec- design requirements for increas- tions, J. Mander, University ing the seismic resistance of priate return periods and param- at Buffalo, G. Martin and existing bridges. eters, consideration of local site Y. Xiao, University of factors and effects, procedures Southern California The manual does not prescribe and rules for constructing elas- requirements dictating when and • Seismic Retrofit of Shear tic spectral demand, guidance on Critical Bridge Columns, how bridges are to be retrofitted – developing time histories, and J. Mander, University at the decision to retrofit is left to the consideration of vertical ground Buffalo engineer and depends on a num- motions and near-field effects; ber of factors. These include, but • Evaluation of Seismic 2. expanded coverage of potential Retrofit Methods for are not limited to, the availability geotechnical hazards and evalua- Reinforced Concrete Bridge of funding, and political, social, and tion of geotechnical components. Columns, F. Klaiber, Iowa other economic considerations. This includes characterization of State University The primary focus of the manual is geotechnical hazards, identifica- • Design Procedures for directed towards providing guid- tion of liquefaction potential and Longitudinal Restrainers: ance on the engineering factors for quantification of liquefaction ef- Multi-Span Simply seismic retrofitting. fects in the free-field, settlement Supported Bridges; The 1995 FHWA manual is being of approach slopes due to ground Multiple-Frame Bridges, updated and significantly expanded, shaking, and concerns with ac- M. Saiidi, University of Nevada-Reno and and will be reissued as Volume II of tive faults and fault rupturing un- G. Fenves, University of the Seismic Retrofitting Manuals derneath the structure. In California, Berkeley for Highway Systems on the basis addition, the new bridge manual

116 contains guidance on foundation retrofit approaches, including the modeling for soil-structure inter- use of energy dissipation systems, action, and stiffness and capacity seismic isolation, restrainers, and (cont.)(cont.) evaluation of abutments, footings, combinations of these tech- • Longitudinal Restrainer pile groups, drilled (pier) shafts, nologies. Experiments, E. Maragakis and caissons, along with methods Recommendations concerning and M. Saiidi, University of Nevada-Reno to determine liquefaction-based post-earthquake emergency assess- displacement demands; ment and repair are also being • Development of Earthquake 3. the incorporation of additional added to the new manual. A new Protective Systems for Bridge Retrofit: Stability of methods for bridge system evalu- chapter has been added which Elastomeric Bearings, ation, including the capacity/ describes the functions, makeup, I.G. Buckle, University of demand spectral evaluation (lin- and pre-event training and planning Auckland; Energy Dissipa- earized elastic and inelastic for both an emergency damage tion Systems, R-factor) method, lateral strength assessment team and a structural M. Shinozuka, University of Southern California; (pushover) analysis, and detailed performance investigation team. Sliding Isolation Systems, computation methods including The emergency damage assessment M. Constantinou, nonlinear time history analysis; team is responsible for immediate University at Buffalo; and on-site evaluations of damaged struc- Comparative Assessment, 4. expanded coverage of widely tures to determine if they are ca- M. Feng, University of employed and new seismic pable of safely carrying traffic, while California, Irvine retrofit strategies. Among these the structural performance investi- • Evaluation of Mechanical are strategies related to strength- gation team performs on-site evalu- Property Changes for HDNR ening, displacement enhance- ations to gather information and data Bearings Over Time, ments, and ground remediation, J. Kelly, University of to determine or theorize and docu- California, Berkeley and a discussion on when the ment the causes of failure or dam- engineer should consider either age, primary and secondary failure • R-Factors for Isolated the full-replacement or “do- modes and sequences, and the va- Bridges, M. Constantinou, nothing” option. The engineering lidity of current design practice and University at Buffalo and economic evaluation con- codes, relative to observed damage. • Field Testing of a siderations for selection of This chapter also provides guidance Seismically Isolated Bridge, appropriate strategies are fully in identifying the significance and S. Chen and J. Mander, described in the manual. extent of damage to bridge compo- University at Buffalo and A large number of new and nents commonly put at risk during B. Douglas, University of Nevada-Reno modified retrofit techniques are de- earthquakes, and in providing sug- scribed in the manual, including gested temporary shoring and repair • Abutments and Retaining many intended for low-to-moderate strategies. Structures, and Develop seismic zones where cost-ef- Analysis and Design fectiveness and simplicity may be Screening, Evaluating Procedures for Retaining as important as moderate increases Structures, R. Richards in seismic performance. Among and Retrofitting of and K. Fishman, Univer- sity at Buffalo these new retrofits are hinge shifting Retaining Structures, and fusible hinge techniques for substandard columns, and methods Slopes, Tunnels, that provide improved performance Culverts and Pavements for existing steel roller and rocker Current national highway seismic bearings. In addition, the new standards in the U.S. are primarily volume includes expanded coverage limited to design and retrofitting of the use of protective systems as provisions for highway bridges. part of the arsenal of potential

Seismic Retrofitting Manuals for Highway Systems117 However, a typical highway system these peripheral elements are is composed of a number of major widespread, the potential impact structural and geotechnical com- on traffic flow from their failure ponents, which include retaining during an earthquake is expected structures, engineered slopes (cuts to be limited. However, traffic flow and fills), tunnels, culverts, and impacts from the failure of the pavements. In addition, a typical other major structural and geo- highway system also contains other technical components during an functional and peripheral elements earthquake could be as severe as such as sound walls, sign and light that historically demonstrated by structures (towers and sign bridges), the failure of highway bridges. This and motorist service facilities. While was evident during the 1995

(b) (c) (a)

(d) (e) ■ Figure 3. Volume III focuses on the development of screening, evaluation and retrofitting of non-bridge components. Experience from past earthquakes has shown that damage to these components can be just as severe and/or destructive as that to the actual bridge structure. Shown above, from left, are: (a) a road severed by a large landslide and (b) pavement damage caused by a failed embankment and flow slide following the Hokkaido Nansei-oki, Japan earthquake; (c) the San Fernanado earthquake caused the headwall to fracture and deformed the inlet of this culvert; (d) cracks at the street surface were caused by earthquake damage to tunnels near the Daikai Station in Kobe; and (e) a retaining wall in Redondo Beach was damaged following the Northridge earthquake.

118 Hanshin-Awaji earthquake in Japan, • seismic vulnerability screening; which resulted in the failure of • detailed structural evaluation; numerous retaining structures, rail and tunnels, and roadway beds. • recommended retrofitting con- As a result, Volume III of the cepts. Seismic Retrofitting Manuals for Highway Systems focuses on the The volume discusses factors development of screening, eval- related to performance criteria and uation, and retrofitting methods the expected level of service or and technologies for these other acceptable damage for each com- major highway system structural ponent, based on the anticipated components. Over time, individual level of seismic shaking. Due to structures have been evaluated and limited previous work on some of retrofitted on a case-by-case basis. these highway system components, Much of this experience, however, the overall philosophy of this is fragmented and not well- volume is intended to be somewhat documented. This volume is the first conservative and anticipates that known effort to capture the additional work may be necessary important aspects of screening, to fully characterize and understand evaluation, and retrofitting of these the behavior of some of these non-bridge highway system struc- highway system components. tural components and to present results and recommendations in a Conclusion formal, procedural manner. Unlike highway bridges, there is The development of the Seismic no precedent for a manual for Retrofitting Manuals for Highway screening, evaluation, and retrofit- Systems is a major component of the ting of highway retaining structures, MCEER highway project research slopes, tunnels, culverts, and pave- program being conducted for the ments. The first step in this devel- Federal Highway Administration opment, therefore, was the (FHWA) and will significantly impact preparation of the strawman docu- highway engineering practice in the ment, which summarized the cur- U.S. following their publication. rent state of knowledge and Drafts of the three volumes have practice for these highway system been completed and the final ver- components, incorporated current sions are expected to be available research results, and identified im- from the FHWA in mid-2000 follow- portant gaps in knowledge where ing a detailed period of technical additional research was necessary review and revision of each volume to complete the volume (NCEER, draft. These volumes will provide 1996). the basis for guidance to agencies Volume III (MCEER, 1999) is com- embarking on a program to evalu- posed of five sections (one for each ate and reduce the seismic vulner- highway component covered in the ability of highway systems. document). Topics covered within each section cover: • classification of the structural component;

Seismic Retrofitting Manuals for Highway Systems119 References Basöz, N. and Kiremidjian, A., (1995), Prioritization of Bridges for Seismic Retrofitting, Technical Report NCEER-95-0007, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

Federal Highway Administration (FHWA), (1983), Seismic Retrofitting Guidelines for Highway Bridges, Report No. FHWA/RD-83/007, U.S. Department of Transportation, Federal Highway Administration, McLean, Virginia.

Federal Highway Administration (FHWA), (1995), Seismic Retrofitting Manual for Highway Bridges, Publication No. FHWA-RD-94-052, U.S. Department of Transportation, Federal Highway Administration, McLean, Virginia.

Hobeika, A.G., Manchikalapudi, L.N., and Kim, S., (1991), Transportation Problems Faced After Big Earthquakes: A Case Study of the Loma Prieta Earthquake, Center for Transportation Research, Virginia Polytechnic Institute and State University, Blacksburg, Virginia.

Moore, J.E., Kim, G., Xu, R.X., and Cho, S., (1995), Evaluating System ATMIS Technologies via Rapid Estimation of Network Flows, MOU-120 Draft Interim Report, California PATH Program, University of California Research Report Nos. VKC-351 and MC-0042, University of Southern California, Los Angeles, California.

Multidisciplinary Center for Earthquake Engineering Research (MCEER), (1999), Seismic Retrofitting Manuals for Highway Systems, Volume III, Screening, Evaluation and Retrofitting of Retaining Structures, Slopes, Tunnels, Culverts, and Pavements, Sections contributed by M.S. Power, K. Fishman, R. Richards, F. Makdisi, S. Musser and T.L. Youd, MCEER Highway Project Task 106-G-3.2, May 1999 Draft, MCEER, Buffalo, New York.

National Center for Earthquake Engineering Research (NCEER), (1996), Seismic Retrofitting Manual for Highway Systems: Volume III Strawman – Screening, Evaluation, and Retrofitting of Retaining Structures, Slopes, Tunnels, Culverts, and Pavements, sections contributed by R. Richards, K. Fishman, F. Makdisi, M. Power, D. Rosidi, J. Kaneshiro, S. Musser, and T.L. Youd, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York. Wakabayashi, H. and Kameda, H., (1992), “Network Performance of Highway Systems Under Earthquake Effects: A Case Study of the 1989 Loma Prieta Earthquake,” Proceedings of the U.S.-Japan Workshop on Earthquake Disaster Prevention for Lifeline Systems, Public Works Research Institute, Tsukuba Science City, Japan.

Werner, S.D., Taylor, C.E., Moore, J.E. and Walton, J.S., (1999), Seismic Retrofitting Manuals for Highway Systems, Volume I, Seismic Risk Analysis of Highway Systems, and Technical Report for Volume I, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.

120 Acknowledgements This report was prepared by the Multidisciplinary Center for Earthquake Engineering Research through grants from the National Science Foundation, a contract from the Federal Highway Administration and funding from New York State and other sponsors. The material herein is based upon work supported in whole or in part by the National Science Foundation, Federal Highway Administration, New York State and other sponsors. Opinions, findings, conclusions or recommendations expressed in this publication do not necessarily reflect the views of these sponsors or the Research Foundation of the State University of New York.

Contributors Ian G. Buckle, Professor of Civil Engineering and Vice-Chancellor for Research, University of Auckland, New Zealand Stephanie E. Chang, Assistant Professor of Geography, University of Washington and Project Engineering Economist, EQE International, Inc. Tsen-Chung Cheng, Lloyd F. Hunt Professor of Electrical Engineering, University of Southern California Shyh-Jeng Chiou, Senior Seismologist, Geomatrix Consultants, Inc. Michael C. Constantinou, Professor and Chair, Department of Civil, Structural and Environmental Engineering, University at Buffalo, State University of New York Ronald T. Eguchi, Vice President, Center for Advanced Planning and Research, EQE International, Inc. Maria Q. Feng, Associate Professor of Civil Engineering, University of California, Irvine Ian M. Friedland, Assistant Director for Transportation Research, Multidisciplinary Center for Earthquake Engineering Research Bijan Houshmand, MCEER Consultant Charles K. Huyck, GIS Specialist, EQE International, Inc. Sang-Soo Jeon, Graduate Research Assistant, Cornell University Dyah Kusumastuti, Graduate Research Assistant, University at Buffalo, State University of New York George C. Lee, Samuel P. Capen Professor of Engineering, University at Buffalo, State University of New York and Director, Multidisciplinary Center for Earthquake Engineering Research Zhong Liang, Research Associate Professor, University at Buffalo, State University of New York John B. Mander, Associate Professor of Civil Engineering, University at Buffalo, State University of New York Emmanuel Maragakis, Professor and Chair, Department of Civil Engineering, University of Nevada, Reno Sheng-Taur Mau, Distinguished Professor of Civil Engineering, New Jersey Institute of Technology Ronald L. Mayes, President, Dynamic Isolation Systems, Inc. Ronald Meis, Graduate Researcher, University of Nevada, Reno Thomas D. O’Rourke, Thomas R. Briggs Professor of Civil Engineering, Cornell University Maurice S. Power, Vice President, Geomatrix Consultants, Inc. Andrei M. Reinhorn, Professor of Civil 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 Masanobu Shinozuka, Fred Champion Chair in Civil Engineering, University of Southern California Raj Siddharthan, Professor of Civil Engineering, University of Nevada, Reno Kathleen J. Tierney, Co-Director, Disaster Research Center, University of Delaware Mai Tong, Research Scientist, Multidisciplinary Center for Earthquake Engineering Research Selcuk Toprak, Research Civil Engineer, U.S. Geological Survey David M. Tralli, MCEER Consultant T. Leslie Youd, Professor and Chair of Civil and Environmental Engineering, Brigham Young University

Production Staff Jane E. Stoyle, Editor Jenna L. Tyson, Layout and Design Hector A. Velasco, Illustration