Kobe, Japan EQ 1995

i

THE JANUARY 17, 1995 KOBE EARTHQUAKEAn EQE Summary Report

April 1995

ii

This report summarizes the effects of the Kobe Earthquake, one of the costliest natural disasters inhistory. Immediately after the main shock, an earthquake reconnaissance team from EQE went tothe affected region to evaluate the effects of the earthquake and to assess the extent and causes ofdamage to structures and infrastructure before critical evidence was removed.

The information presented in this summary report was collected over the two months followingthe earthquake. The statements and conclusions made are based strictly on our preliminaryfindings and assessments. Additional information and detailed investigation in the ensuingmonths may change some of our conclusions.

© 1995 by EQE InternationalALL RIGHTS RESERVEDNo part of this document may be reproduced or transmitted in any form or by any means,electronic or mechanical, including photocopying, recording, or by any information storage andretrieval system, without permission in writing from EQE International.

iii

CONTENTS

FOREWORD .............................................................................................................. v

EXECUTIVE SUMMARY ....................................................................................... vii

INTRODUCTION ...................................................................................................... 1

EARTH SCIENCE ASPECTS .................................................................................... 5

BUILDINGS.............................................................................................................. 13

INDUSTRIAL FACILITIES ..................................................................................... 35

TRANSPORTATION............................................................................................... 45

PORTS ...................................................................................................................... 61

OTHER LIFELINES ................................................................................................. 67

FIRE FOLLOWING EARTHQUAKE ..................................................................... 73

ECONOMIC IMPACT ............................................................................................. 77

SOCIETAL IMPACT................................................................................................ 85

CONCLUSIONS ...................................................................................................... 87

Kanji on the cover reads “The Great Hanshin Earthquake Disaster,” thepopular name for this earthquake in Japan. The official name for theearthquake as given by the Japan Meteorological Agency is “TheHyogo-ken Nambu Earthquake.”

iv

ACKNOWLEDGMENTS

Mr. Umeta of the Ashiya Fire Department; Dr. K. Ishida from the Central ResearchInstitute of Electric Power Industry (CRIEPI); Mr. K. Beppu and Mr. K. Koizumi of the DailyYomiuri Newspaper; Dr. Robert Kassawara from the Electrical Power Research Institute(EPRI); Mr. C.Y. Chang of Geomatrix Consultants; Felix Treibmann from Gerling-Konzern; Dr.Hitomi O. Murakami and Ms. Chiaki Watanabe, from the Department of ArchitecturalEngineering, Hokkaido University; Mr. Mikage and Mr. Takinami of the Japan Rail Corpora-tion; Mr. R. Norita, Mr. N. Fujii, Mr. M. Hirose, and Mr. Y. Mino from Kansai Electric’sheadquarters; Mr. T. Yamanaka from Kansai Electric’s New York Office; Mr. S. Ouchi at thePort Island Incinerator Plant, Kobe City Environment Bureau; Mr. S. Tanioka at Rokko IslandSlag Center, Kobe City Development Bureau; Mr. K. Tochio from the Port Island SewageDisposal Plant, Kobe City Sewage Department; Lieutenant H. Nakachi from the Kobe FireDepartment; Mr. S. Ohta from Kobe Newspaper; Professor S. Takada of Kobe University; Mr.Morita and Mr. M. Matsushita of the Kobe Water Department; Professor K. Toki, Professor H.Iemura, Professor Igarashi, Professor T. Sato, and Professor Sugito of Kyoto University; Dr.Haruo Hayashi and Dr. Hiroyuki Kameda, from the Disaster Prevention Research Institute,Kyoto University; Dr. K. Kawashima of the Public Works Research Institute; Dr. M. Izumi andDr. H. Katukura from Izumi Research Institute, Shimizu Corporation; Mr. M. Takeda from thePower/Energy Division, Shimizu Corporation; Professor T. Katayama and Professor F.Yamakazi of the University of Tokyo; Dr. M. Kimura from Tokyu Construction Co.; Dr. Y.Ogawa of the UN Centre for Regional Development; Professor M. Hamada of WasedaUniversity; Paul Sommerville from Woodward-Clyde Consultants; Professor S. Murikami ofYokohama University; Mr. T. Kumagi, Zurich International Company.

The “Earth Science Aspects” of this report was written in large part by GeomatrixConsultants, and we wish to extend a special thanks to them for their assistance.

EQE INVESTIGATION TEAM: John Alderman, John Beardall, Dr. Kenneth Campbell,Robert Campbell, Dr. Stephanie Chang, Craig Cole, Kent David, James Goltz, BillGordon, Ming Lee, David O'Sullivan, Thomas Roche, Dr. Charles Scawthorn, RichardTiong, Peter Yanev

RESEARCH & PUBLICATION: Linda Decker, Patrick Ellis, Kelly Fleming, AnnaGiaconia, Ronald Hamburger, David McCormick, Dr. Michael Rojansky, Scott Schleifer,John Smallwood-Garcia

v

ForewordAs individuals, many of us at EQE have worked intermittently in Japan since the 1970s,

and have had continuous contact with many of our Japanese peers over that time. As acompany, we have worked in Japan and for Japanese clients since 1983 and have investigatedseveral destructive earthquakes there.

At the time of the earthquake, four of our senior staff members were in Japan, includingDr. Charles Scawthorn, who received his Ph.D. from Kyoto University. Two more EQEengineers were en route to Japan to evaluate the industrial facilities of a U.S. multinationalcorporation. Once we realized the importance of what had happened in Kobe, we quicklydispatched seven more investigators. The team consisted of structural and mechanicalengineers, a fire protection engineer, a sociologist, an economist, and an insurance companyexecutive. Four members of the team are fluent in Japanese.

A number of our multinational clients were affected by the earthquake, so we were ableto gain a detailed understanding of the earthquake’s effects and its impact on their operations,in addition to studying buildings, industry, ports, the supporting infrastructure, and emer-gency response and recovery efforts. We are currently working extensively in Kobe andthroughout Japan, and continue to learn from the disaster. By the end of March 1995 we sentmore than 25 employees to Japan to evaluate damage, to work on projects, and study theeffects of the earthquake.

We wish to express our appreciation to our colleagues in Japan—Japanese, NorthAmericans, and Europeans—who assisted us in the midst of their own response to the disaster;to Geomatrix, Inc., for their contribution on the earth science aspects of this report; and to thededicated EQE employees who investigated this earthquake, researched and wrote thisreport, and produced this publication.

Finally, and most importantly, we wish to extend our sympathy to the victims of thisearthquake. To those of us in the earthquake engineering profession, the deaths and othertragic consequences of every earthquake are particularly hard to accept because we knowmany of the deaths and injuries could have been avoided. It is our sincere hope and belief thatby studying the causes of damage from earthquakes and publishing these reports futureearthquakes will result in fewer tragedies.

vi

vii

On the first anniversary of the momentmagnitude (MW) 6.7 1994 Northridge Earth-quake, Kobe, Japan was struck by an MW6.9earthquake. Both earthquakes struck in thepre-dawn hours, both ruptured beneathdensely populated areas, and both causedhorrible damage. Yet in Kobe there weremany more deaths, financial losses dwarfedthose in Northridge, and the amount of de-stroyed building stock and infrastructure wasfar worse in Kobe than in Northridge.

The reasons for these differences aremany, but it would be incorrect to issue ablanket condemnation of current Japaneseseismic engineering practice. While engi-neered structures did fail due to design flaws,they were predominantly older structuresbuilt before the current Japanese buildingcode became effective; or they frequentlyfailed due to problems revealed to be defi-ciencies in California design practices by theNorthridge Earthquake. Japanese seismicengineering expertise has justifiably beenconsidered among the best in the world, anda careful examination of the damage in Kobedoes not change that conclusion.

Despite differences in design and con-struction practices, the same general prin-ciples frequently came into play: highwaycollapses were often primarily due to insuffi-cient lateral ties in the concrete columns,nonductile concrete frame buildings did muchworse than ductile design, shear walls typi-cally helped to lessen catastrophic damage,and soft soils resulted in greater damage tothe structures constructed on them.

The most important lesson in both earth-quakes is that the knowledge to significantlyimprove structures to resist earthquake dam-age and thereby avoid most of the deaths andfinancial losses exists; what is lacking is aconsistent willingness to marshall the re-sources necessary to put that knowledge towork on the scale necessary to prevent disas-ters. It is an odd paradox, for time and timeagain it is demonstrated that it usually costsless to prepare for earthquakes in advancethan to repair the damage afterwards.

Differences in Kobe and NorthridgeWhile there are more similarities than

differences in structural performance in theKobe and Northridge earthquakes, there areimportant differences that explain why theKobe Earthquake was so much more damag-ing. Some of the lessons from these differ-ences apply only to Japan, others apply to allareas of the world at risk from earthquakes.

The vast majority of deaths in Kobe oc-curred in the collapse of housing built usingtraditional Japanese methods. TraditionalJapanese housing construction is based on apost-and-beam method with little lateral re-sistance. Exacerbating the problem is the prac-tice of using thick mud and heavy tile forroofing, resulting in a structure with a veryheavy roof and little resistance to the hori-zontal forces of earthquakes. U.S.-style framehousing with light-weight roofs is now com-ing into use in Japan and newer housingconstructed using these methods had little orno damage from the earthquake.

Another significant difference betweenthe Kobe area and the Northridge area is thequality of the soils. Because of a severe short-age of available land, much of modern urbanJapan, including Tokyo, is built on the worstsoil possible for earthquakes. Much of thenewer construction in Kobe, particularlylarger buildings, is built on very soft, recentalluvial soil and on recently constructed near-shore islands. Most of the serious damage tolarger commercial and industrial buildingsand infrastructure occurred in areas of softsoils and reclaimed land. The worst indus-trial damage occurred at or near the water-front due to ground failures-liquefaction,lateral spreading, and settlement.

The Port of Kobe was an extreme ex-ample of the problems associated with poorsoils in areas prone to earthquakes. The portis built almost entirely on fill. The engineer-ing profession has tried hard to developmethods for strengthening filled areas toresist failures during earthquakes, but mostof these methods have been put into practicewithout the benefit of being adequately tested

EXECUTIVE SUMMARY

viii

in strong earthquakes. The results were de-cidedly mixed, but the failures costly—mostretaining walls along the port failed, and therelated ground settlement pulled buildingsand other structures apart.

BuildingsThe large commercial and industrial

buildings in the Kobe area, particularly thosebuilt with steel or concrete framing, are simi-lar to buildings of the same vintage in Cali-fornia. The Japanese building code had amajor revision for concrete-frame buildingsand a more limited revision for steel-framebuildings in 1981. The Uniform Building Code,as used in California, had major changes in1973, 1975, and several times since then. Thecurrent Japanese code requires that build-ings in Japan be designed for somewhat higherforce levels than does the Uniform BuildingCode. Both areas require design for muchhigher forces than most other earthquakeregions of the world.

Typically, pre-1981 concrete-frame build-ings performed very poorly in Kobe, withmany collapses. Post-1981 buildings per-formed much better—some were extensivelydamaged, but most had light damage. Thebuildings that fared best, and those withoutsignificant damage, had extensive concreteshear walls.

As in other earthquakes, large commer-cial and industrial steel-frame buildings per-formed better than any other type. However,major damage and a few collapses were ob-served. Pre-1981 steel buildings had most ofthe serious known damage. Certain innova-tive types of steel buildings, including high-rises, had very serious damage, and collapsesmight have occurred if the duration of theearthquake had been a few seconds longer.

Building owners usually do not under-stand that the earthquake provisions of eventhe strictest building codes do not necessar-ily have reasonable performance criteria forlarger and stronger earthquakes. The currentregulations, including those for all of Califor-nia, are typically written with the expecta-tion that in a strong earthquake a building

will be severely damaged—in fact, it is as-sumed the building may need to be torndown, but it should not collapse. In Califor-nia, higher performance criteria are man-dated for certain types of structures—schools,hospitals, police and emergency responsebuildings, and certain power facilities. Aninformed building owner can choose to usethese higher criteria, and thus avoid havingtheir high-value, heavily occupied commer-cial building designed, in effect, to the sameearthquake performance level as a low-valuefarm building.

TransportationA number of major expressways, rail

lines, and bridges, some of very moderndesign, were severely damaged. There are nosignificant new lessons from the collapse anddamage of the older unretrofitted bridgesand elevated structures. The structural andfoundation details that typically caused dam-age to the expressways and rail lines havebeen observed in numerous earthquakes, andthe damage was predictable. Some of theupgrade details observed in older retrofittedstructures, such as steel column jacketing,are now widely used in California forstrengthening. The apparent good perfor-mance of these details in Kobe is important toongoing U.S. programs and needs to be stud-ied in detail.

Many bridges and bridge approacheswere severely damaged. The performance oflarge new bridges, including cable-tied arch,braced arch, and cable-stayed bridges, shouldbe studied extensively because this is thestrongest earthquake to affect such bridges.

The Port of Kobe, much of which wasnew, was devastated by widespread andsevere liquefaction and/or permanentground deformation, which destroyed morethan 90% of the port’s 187 berths and dam-aged or destroyed most large cranes. Ship-ping will be disrupted for many months, andsome shipping business will probably neverreturn to Kobe, resulting in significant lossesto the local economy.

ix

Other InfrastructureThe electrical and telecommunications

systems in Kobe and surrounding areas per-formed as expected based on experience fromprevious earthquakes. Long term power out-ages were isolated to the most heavily dam-aged areas. Facilities near the epicentersustained damage while resiliency of the sys-tems prevented widespread service inter-ruption. Most of the major transmission linesskirt the heavily damaged region of Kobe—the results may have been substantially dif-ferent if the epicenter was located closer tothe 500 kV transmission system. There weresubstantial financial losses to the electricalutilities, however, because expensive spe-cialized equipment must be replaced and thedistribution network must essentially be re-built within heavily damaged areas of Kobe.

During the earthquake, Kobe’s watersystem sustained approximately 2,000 breaks.Generally, ground or building failure wasthe cause of the severe damage to Kobe’swater systems. The resulting lack of watercontributed significantly to the fire problemand will be a major hardship on the popula-tion for several months. The gas system hadmajor damage, generally caused by groundor building failure, which also contributedsignificantly to the fire problem.

FireMore than 150 fires occurred in Kobe and

surrounding areas in the hours after the earth-quake. These resulted in several large fires,and fire fighters were for the most part un-able to combat them because of streets beingblocked by collapsed buildings and buildingdebris, traffic congestion, and severe watersystem damage. Calm wind conditions pre-vented conflagrations. The United States andJapan have both sustained the largest peace-time urban conflagrations in this century’shistory—because of earthquakes. Fire fol-lowing earthquake is a potential major agentof damage, and needs to be recognized assuch by planners.

ConclusionThe Kobe Earthquake dramatically illus-

trates the damage that can result when astrong earthquake strikes a modern industri-alized area. It should not, however, be takenas an indication of what should be consid-ered inevitable. While every major earth-quake teaches new lessons to the engineeringprofession, the lessons are now increasinglyrefinements to knowledge already put intopractice in many new structures. In Kobe andin Northridge, those structures with currentseismic design details survived strong earth-quakes with little damage and thereby vali-dated current seismic design philosophies.

Significant seismic engineering chal-lenges still need to be met, but the mostcritical challenge now is to society. Themeans to lessen the disastrous effects of strongearthquakes now exist and it is the responsi-bility of business and government leaders toput those means to work to save lives, andpreserve financial and economic prosperity.

x

EQE 1

On Tuesday, January 17, at 5:46 A.M. localtime, an earthquake of magnitude 7.2 (Mj)

1

struck the region of Kobe and Osaka in south-central Japan. This region is Japan’s second-most populated and industrialized area, afterTokyo, with a total population of about 10million. The shock occurred at a shallowdepth on a fault running from Awaji Islandthrough the city of Kobe, which in itself has apopulation of about 1.5 million. Strong groundshaking lasted for about 20 seconds andcaused severe damage over a large area.

Nearly 5,500 deaths have been confirmed,with the number of injured people reachingabout 35,000. Nearly 180,000 buildings werebadly damaged or destroyed, and officialsestimate that more than 300,000 people werehomeless on the night of the earthquake.

An entire city blockdestroyed by fire, ChuoWard.

The life loss caused by the earthquakewas the worst in Japan since the 1923 GreatKanto Earthquake, when about 140,000people were killed, mostly by the post-earthquake conflagration. The economic lossfrom the 1995 earthquake may be the largestever caused by a natural disaster in moderntimes. The direct damage caused by the shak-ing is estimated at over ¥13 trillion (aboutU.S.$147 billion). This does not include indi-rect economic effects from loss of life, busi-ness interruption, and loss of production.

Damage was recorded over a 100-kilometerradius from the epicenter, including the cities ofKobe, Osaka, and Kyoto, but Kobe and itsimmediate region were the areas most se-verely affected. Damage was particularlysevere in central Kobe, in an area roughly 5kilometers by 20 kilometers parallel to thePort of Kobe. This coastal area is composedprimarily of soft alluvial soils and artificialfills. Severe damage extended well northeastand east of Kobe into the outskirts of Osakaand its port.

1. Based on the Japan Meteorological Agency(JMA) magnitude scale, roughly equivalent to amoment magnitude (Mw) of 6.9.

INTRODUCTION

2 EQE

ASHIYA

ROKKO ISLAND

PORTISLAND

NISHINOMIYA

AWAJI-SHIMA

0 15

Kilometers

MJ7.2

KOBE CITY

OSAKA

SannomiyaMany houses collapsed,several conflagrations,

general damage

HanshinExpressway

Multiplecollapsed spans

over 20 km

Wangan (Harbor)Expressway

Varying damage tosteel spans, beam seats,

and major crossings

Extensive liquefaction,collapsed seawalls,

leaning cranes

ShinkansenCollapsed spans

east of tunnelentrance for 3 km

AKASHI

SUMA

TARUMI

NISHI

KITA

NADA

TAKARAZUKA

KAWANISHI

IKEDA MINOO

TOYONAKA

ITAMI

AMAGASAKIHIGASHINADA

NAGATAKOBE

CHUO

KAKO

MIKI

Hankyu, Hanshin,and JR rail linesExtensive roadbed,rolling stock, andstation damage

over 20 km

Heavy firedamage

Collapsedbuildings

Extensiveliquefaction

Denslypopulatedarea

Collapsedroads, trainderailment

Top: Map of the Kobe area.

Bottom: One of hundreds ofcollapsed buildingsthroughout central Kobe.

Opposite, top: Collapsedportion of the HanshinExpressway.

Opposite, bottom: Searchparty investigating acollapsed residential wood-frame building, Nada Ward.

EQE 3

Our experience with many past earth-quakes in developed, industrial areas is thatthe media, particularly television, can presentan exaggerated image of the damage by con-centrating on the few spectacular collapsesthat occurred. The actual damage in Kobeand the surrounding region, however, wasmuch worse than the media could convey,because it is very difficult to show more thanlocal damage at one time. For example, im-ages of the main, 550-meter-long collapsedsection of Kobe’s elevated Hanshin Express-way were ever-present throughout the me-dia, but that collapse was only a small fractionof the losses to the area’s highway system.

Central Kobe, according to many olderresidents and our investigators, presentedthe image of a war zone, with a large percent-age of both commercial and residential build-ings destroyed.

All of this happened in about 20 seconds.

4 EQE

Top left: Self-defense troops performing a search and rescue operation at a collapse site.

Top right: Many streets were blocked by collapsed buildings, hindering emergency response.

Bottom: This man is hauling water. Nine days after the earthquake, 367,000 households were still without water.

EQE 5

TokyoOsakaKobeMedian

Tectonic Line

Arima-Takatsuki

Tectonic Line

The causative plate action.In the Kobe area, cretaceous granites areoverlain by a relatively thick Plio-Pleistocenesedimentary unit called the Osaka group,which consists of alluvium interbedded withmarine clays. Relatively thin terrace depositsand recent alluvium overlie the Osaka group.Fill material has been placed along much ofthe waterfront and comprises human-madeislands, such as Port and Rokko islands.

Preliminary reports from the JapaneseEarthquake Research Institute indicate thatthe hypocenter of the Mj7.2 (equivalent toMw6.9) main shock occurred at a depth ofapproximately 15 to 20 kilometers. The mainshock’s focal mechanism indicates predomi-nantly strike-slip movement along a planethat dips 80° to 90° to the southwest. Theaftershock sequence (and, by inference, thefaulting below the surface) is approximately60 kilometers long, extending from the north-ern part of Awaji Island along the NojimaFault to northeast of Kobe along the RokkoFault zone.

Southwestern Japan is located on thesoutheastern margin of the Eurasian Plate,where the Philippine Sea Plate is being thrust(subducted) beneath the Eurasian Plate in anorthwest direction along the Nankai Trough.A portion of this relative plate motion istaken up by right-lateral strike-slip faultingalong a major east-northeast-trending faultknown as the Median Tectonic Line (MTL),located immediately south of Awaji Islandand Osaka Bay.

The main shock occurred along a north-west-trending branch of the MTL called theArima-Takatsuki Tectonic Line (ATTL). Thisfault system, like the MTL, has a predomi-nantly right-lateral strike-slip sense of dis-placement. Historically, this region has seensomewhat lesser seismicity than in the Tokyoarea and some other parts of Japan, but hashad magnitude 7 or greater events in histori-cal times (e.g., in 1596). In 1916, a magnitude6.1 earthquake occurred at almost the sameepicentral location as the 1995 event.

EARTH SCIENCE ASPECTS

6 EQE

130O 135O 140O 145O

200 km

42O

40O

38O

36O

34O

32O

30O

28O

MedianTectonic Line

KyotoKobe Osaka

Tokyo

Depth Scale (km)

0 100 200 300 400 500

5 6 7 8 9

Magnitude Scale

Tectonic Plate Boundaries

Japanese earthquakes, 1961-1994.

EQE 7

AVERAGE NO. OF SHALLOW EARTHQUAKES IN 90 YEARS

8.0 - 8.97.0 - 7.96.0 - 6.95.0 - 5.9

445

4504,500

RICHTERMAGNITUDE WORLD JAPAN

CALIFORNIAAND W. NEVADA

45775

7,10070,000

17

75730

RELATIVE SEISMICITY OF JAPAN

Relative Seismicity of JapanImmediately following the earthquake,

much speculation in the media attempted toplace in perspective the seismicity of Japan,particularly with respect to California. Vari-ous Japanese sources also professed surprisethat the earthquake occurred in the Kobearea, which had not been struck by a trulydevastating earthquake since 1596.

All of Japan lies in one of the mostseismically active regions of the world, withall heavily populated areas subject to strongearthquakes. The island nation is surroundedby major offshore faults and is crisscrossedby many active faults, one of which, theNojima Fault, ruptured in this event. The lastsimilar earthquake to cause severe damageand comparable life loss was the 1948 M7.0Fukui Earthquake, which was located nearthe city of Fukui, some 175 kilometers to thenortheast of Kobe. Since 1948, Japan has beenrelatively fortunate because none of the 12earthquakes greater than M6.9 during thisperiod were centered in a densely populatedarea or very close to a major urban area.

California, with a land area approxi-mately equal to that of Japan but with apopulation of approximately 25% of Japan’s(31 million versus 125 million), has recentlybeen more seismically active, with three re-cent earthquakes of magnitude 6.5 or greater

centered in heavily populated areas—theM7.1, 1989 Loma Prieta (San Francisco BayArea); the M6.7, 1994 Northridge (Los An-geles); and the M6.6, 1971 San Fernando(Los Angeles area) earthquakes. The recentNorthridge event was centered in the heavilypopulated, by California standards, SanFernando Valley.

In comparing the relative seismicity ofJapan with California and the rest of theworld, it is important to limit the types ofevents considered to those that are poten-tially damaging. Very deep events (i.e.,deeper than 60 kilometers) are typically non-damaging. Therefore, the comparisonshown below is based on shallow events,less than 60 kilometers deep, for correctedmagnitude (Ms) equal to or greater than 7.0.The data are for the period of 1900 to 1989 (90years) (Reference 1). For California (andwestern Nevada), the data are based onReference 2 and are directly comparable tothe world data for events with magnitudesequal to or greater than 6.0.

The number of shallow and potentiallydamaging earthquakes (M5 to M8.9) in Ja-pan is about a factor of 6 to 7 greater than inthe California region. Japanese earthquakesrepresent about 6% to 7% of the world’spotentially damaging shallow earthquakes.

8 EQE

displacements of 2.1 meters. Past surface-faulting events, which are probably similar tothe most recent event, were evidenced by the6- to 7-meter-high fault scarp along the fault.Given a long-term slip rate of 1 millimeter peryear for the ATTL, as listed in “Active Faultsin Japan: Sheet Maps and Inventory by theResearch Group of Active Faults,” and anaverage displacement of about 1 to 1.5 meters,as suggested from observed displacement onthe Nojima Fault, it appears that an earth-quake roughly the size of the Kobe shockoccurs on average once every 1,000 to 1,500years along this portion of the ATTL.

It is unknown whether the surface faultrupture extended to the northeast across theAkashi Strait and onland to connect with

0 20

Kilometers

0.06g

0.11g0.05g

0.19g

0.23g

0.49g

0.16g

0.32g

0.15g

0.20g

0.02g0.25g

0.11g

0.08g0.03g

0.24g

0.62g

0.83g0.57g

0.49g0.61g

0.80g

0.08g

0.20g 0.12g0.27g

0.84g

0.50g 0.20g

Adapted from Earthquake Research Institute, University of Tokyo

Epicenter (MA)

Active fault

Iso-accelerationcontours

Aftershocks

Recorded Accelerations0.53g

(inexact stationlocation)

(exact stationlocation)

0.53g

An approximately 9-kilometer-long sur-face fault rupture was identified along theNojima Fault, which is on the northwesterncoast of Awaji Island and southwest of Kobe.The fault strikes N40°W, dips steeply to thesoutheast, and has a predominantly right-lateral strike-slip sense of displacement con-sistent with the mechanism of the main shockand the trend of the aftershocks. GeomatrixConsultants (a geotechnical firm) measuredlocal displacements at two locations alongthe northern part of the fault from the recentearthquake: Vertical displacements were 1.2meters, and right-lateral displacements were1.5 meters. These displacements are in goodagreement with measurements by others,who reported maximum vertical displace-ments of about 1.2 meters and right-lateral

Ground motion map.

EQE 9

1 10 100Distance to Seismogenic Fault Rupture (km)

0.01

0.1

1P

ea

k H

ori

zon

tal A

cc

ele

rati

on

(g

)

RECORDEDACCELERATIONS

ON ALLUVIUM:

2nd percentile

16th percentile

50th percentile

84th percentile

98th percentile

Campbell & Bozorgnia (1994)

Kobe

Osaka

Kyoto

IX to XI

VII to VIII

VI to VII

V to VI

Less than V

MMI

0 50 100

Kilometers

Top: Comparison of Kobe Earthquake strong groundmotion data with predictions from Campbell andBozorgnia (1994) indicates that the Kobe strong motionwas typical.

Bottom: Generalized Modified Mercalli Intensity (MMI)map for the January 17 event.

10 EQE

faults in the Kobe-Nishinomiya area. Equivo-cal evidence of surface faulting has beendescribed in this area and apparently is con-sistent with the aftershock sequence, whichis approximately 60 kilometers long and ex-tends northeast of Kobe. Based on empiricaldata of earthquake magnitude versus sur-face fault length, a 9-kilometer-long surfacerupture should yield only an Mw6.2 earth-quake, whereas a 60-kilometer-long ruptureshould yield an Mw7.1 earthquake, which ismore consistent with the observed magni-tude for this earthquake.

A shaking intensity of up to 7 on the JMAintensity scale [equivalent to X to XI on theModified Mercalli Intensity (MMI) scale] has

Committee of Earthquake Observation andResearch in the Kansai Area. The maximumhorizontal accelerations are those reportedby several different agencies and representeither the maximum of the two peak horizon-tal accelerations or the vectoral combinationof the two horizontal components. A maxi-mum acceleration of 0.84g (g equals 981 cm/s/s) was reported in central Kobe, and sev-eral recordings in the range of 0.5g to 0.8gwere reported in the heavily damaged Kobe-Ashiya-Nishinomiya area.

A preliminary estimate of the 250 cm/s/s(0.25g) and 500 cm/s/s (0.51g) iso-accelerationcontours is overlain on the map on page 8.The contours show a distinct bulge towardthe northeast, indicating that ground mo-tions were higher northeast of the epicenterin the direction of rupture propagation prin-cipally because of source directivity (i.e., fo-cusing). The 250 cm/s/s contour does notextend as far as Osaka, which is consistentwith the lower intensity (JMA 4) reported forthis area. It is interesting to note that themaximum accelerations in the Kyoto area aresimilar to those in the Osaka area, even thoughthe former was reported to have a JMA inten-sity of 1 unit higher.

A comparison of the recorded maximumaccelerations with predictions for an Mw6.9strike-slip earthquake (page 9) indicates thatthe accelerations recorded during the earth-quake are generally consistent with, thoughpossibly slightly higher than, those recordedworldwide during other major strike-slipearthquakes of similar magnitude. The maxi-mum accelerations are also similar on aver-age to those recorded during the 1994 Mw6.7Northridge, California, Earthquake. Thiscomparison, along with other structural andgeotechnical information that is available,would seem to suggest that the greater dam-age and the larger numbers of deaths, casual-ties, and homeless sustained during the KobeEarthquake were likely caused by the aggre-gated effects of an extremely dense popula-tion, an older building stock, and thepredominance of poor soils in the stronglyshaken area.

been assigned to the coastal strip extendingfrom the Suma Ward to Nishi-nomiya and inthe Ichinomiya area on Awaji Island; JMA 5(MMI VII to VIII) to Iwakuni, Hikone, Kyoto,and Toyooka; and JMA 4 (MMI VI) to Nara,Okayama, Osaka, Takamatsu, Shikoku, andWakayama. The distribution of maximumhorizontal ground accelerations and veloci-ties recorded in the Kansai area is shown onpage 8. This figure was modified from a mapprovided by the Earthquake Research Insti-tute, University of Tokyo. The map has beenaugmented with additional accelerationand velocity recordings reported by the

Parking lot on reclaimedland near Ashiya. Sandcovering the lot is evidence oflarge-scale liquefaction andsand ejection.

EQE 11

Izumiotsu, Kishiwada, and other areasaround Osaka Bay. Massive liquefaction andlateral spreading took place in areas of re-claimed land and on the many artificial is-lands in the city of Kobe and Nishinomiya.Ejected sand from liquefaction covered muchof the islands and interfered with rescue andrecovery operations.

Similar effects were observed through-out the Kobe mainland along the coast, in-cluding parts of downtown. Typically, as indowntown Kobe, settlement and liquefac-tion of less than 50 centimeters were ob-served. That increased to as much as 3 metersalong the coastline. The settlement causedsevere damage to underground utilities, sev-ering all services (gas, water, sewage) tolarge parts of the mainland and to all re-claimed islands, including the largest is-lands—Rokko and Port. A month after theearthquake, these services had largely beenrestored to Rokko and Port islands.

The most obvious and destructive lique-faction and related lateral spreading of soils

Failed quay wall inNishinomiya. Lateralspreading and settlement offill material have pushed thewall to the right. Note thebackhoe for scale.

Liquefaction and Other GroundFailures

The earthquake caused extensive groundfailures, which affected buildings, under-ground infrastructure, the port, highways,all types of other facilities on soft or filledground, and recovery efforts.

Ground failures occurred primarily be-cause of liquefaction, the result of loose, water-saturated sand being shaken during anearthquake and assuming a semiliquid state.The areas affected by liquefaction were moreheavily developed than any other earthquake-stricken region to date. Therefore, the lessonsare valuable and will enhance our knowl-edge of liquefaction for both natural soils andreclaimed lands with high water tables.

The affected areas were located primar-ily along the coastline and the numerouswatercourses in the general area of Kobe andthe valleys between Kobe and Osaka. Wide-spread liquefaction, over many square kilo-meters, occurred around Kobe, Ashiya,Nishinomiya, Amagasaki, Osaka, Sakai,

12 EQE

and settlement occurred along the dozens ofkilometers of seawalls along the port. Lateralspreading on the order of 3 (or more) metersand vertical settlement of 2 to 3 meters wereobserved along the seawalls of numerousislands, including Port and Rokko islands,and throughout the Port of Kobe. The largestsettlements, and worst damage, seemed to beassociated with the older reclaimed lands,such as the older parts of the port. The newer,engineered fills performed somewhat betterthan did the old fills, but with less thanadequate results.

Numerous buildings on reclaimed landtilted because of ground settlement. Thesewere primarily older, heavy concrete, indus-trial buildings, probably on mat foundations.The majority of industrial and other build-ings on fill were supported on piles (most ofthese were lighter steel buildings). Most pile-supported buildings appeared to performwell; many multistory or large pile-supportedbuildings in areas where extensive liquefac-tion (and limited lateral spreading) occurredhad little or no damage. Typically, the side-walks of such buildings would settle 50 cen-timeters or more, but there would be noapparent damage to the buildings themselves.The same was generally true for newer

highway structures supported on piles. How-ever, the strong shaking may have exceededthe capacity of many pile foundations sup-porting elevated expressway and bridge piers,causing tilting or lateral movements (ob-served to be as much as 2 meters) of the piers.This often contributed to damage or collapseof the superstructures.

References

1. Pacheco, J. F., and L. R. Sykes. 1992.“Seismic Moment Catalog of LargeShallow Earthquakes, 1900 to 1989.”Bulletin of the Seismological Society ofAmerica, Vol. 82: 1306-1349.

2. Ellsworth, W. L. 1990. “EarthquakeHistory, 1769 - 1989.” In The San AndreasFault System, California. R. E. Wallace, ed.U.S. Geological Survey ProfessionalPaper 1515: 153-187.

Human-made island inNishinomiya showingevidence of large-scaleliquefaction, settlement, andlateral spreading.

EQE 13

The number of buildings destroyed bythe earthquake exceeds 100,000, or approxi-mately one in five buildings in the stronglyshaken area. An additional 80,000 buildingswere badly damaged. The large numbers ofdamaged traditional-style Japanese resi-dences and small, traditional commercialbuildings of three stories or less account fora great deal of the damage. In sections wherethese buildings were concentrated in the out-lying areas of Kobe, entire blocks of collapsedbuildings were common. Several thousandbuildings were also destroyed by the firesfollowing the earthquake.

Mid-rise commercial buildings, gener-ally 6 to 12 stories high, make up a substantialportion of the buildings in the Kobe businessdistrict. The highest concentration of dam-aged mid-rise buildings was observed in theSannomiya area of Kobe’s central business

district. In this area, most of the commercialbuildings had some structural damage, and alarge number of buildings collapsed on virtu-ally every block. Most collapses were towardthe north, which was evidently the result of along-period velocity pulse perpendicular tothe fault. This effect has also been observed inother earthquakes. Failures of major com-mercial and residential buildings were notedas far away as Ashiya, Nishinomiya, andTakarazuka. In general, many newer struc-tures performed quite well and withstoodthe earthquake with little or no damage.

In the heavily damaged central sectionsof downtown Kobe, approximately 60% ofthe buildings had significant structural dam-age, and about 20% completely or partiallycollapsed. One survey of a 120,000-square-meter area in downtown Kobe (the Sannomiyaarea) found that 21 out of 116 buildings, or

This collapsed concretebuilding in Kobe completelyblocked the street.

BUILDINGS

14 EQE

Top left: Badly damagedconcrete shear wall building.

Top right: Groundsettlement in central Kobe.

Bottom: Mid-heightcollapse of a mixed-usebuilding (built circa 1977)in Nishinomiya. This type ofcollapse was very commonin this earthquake.

EQE 15

foundations, but it did not appear to be thedominant problem for mid- and high-risestructures supported on piles that extendedinto dense soils or rock. Although hiddendamage may be discovered at a later date, theperformance of piles appeared to be good aslong as substantial lateral soil displacementdid not occur.

A survey of 24 commercial buildings be-ing demolished in the central Sannomiya areaof Kobe two months after the earthquakefound the following breakdown of buildingtypes: 70% were frame type, 20% were shearwall type, and 10% were braced frame type.The breakdown of the frame-type structuresincluded 50% nonductile concrete frame, 35%steel reinforced concrete (SRC) frame, 10%moment-resisting steel frame, and 5% steelframe with masonry infill. Of the shear wallbuildings being demolished, 75% were con-crete and one was unreinforced masonry.Several of the buildings being demolishedwere of multiple construction types.

Building CodeThe first building code in Japan was in-

troduced in 1926 after the 1923 Great KantoEarthquake and ensuing fire devastated

The ruins of the Ginza afterthe 1923 Great Kanto(Tokyo) Earthquake and fire.

18%, were visibly destroyed. Another reportindicated that 22% of office buildings in aportion of the Kobe city center were unus-able, while an additional 66% may need morethan six months for complete restoration.City inspectors declared approximately 50%of the multifamily dwellings in Kobe as un-safe to enter or unfit for habitation, leavingmore than 300,000 people homeless.

Age of construction, soil and foundationcondition, proximity to the fault, and type ofstructural system were major determiningfactors in the performance of structures.Damage was worst in the areas bordering theport or streams and rivers—where soils wereeither poorly consolidated alluvial depositsor fill—and tended to be relatively minor inthe foothills of Rokko Mountain, where ei-ther soils are very shallow or there are rockoutcroppings. Loose and soft soils amplifyground motions in comparison to bedrock,especially ground motions within a certainfrequency range. The duration of shakingalso tends to be longer on such soils.

Structural damage directly resultingfrom soil failures was observed forsmaller buildings without pile-supported

16 EQE

Buildings in central Kobe(Chuo Ward). In theforeground is the completecollapse of a two- or three-story traditional Japanesewood-frame building with aheavy tile roof. On the rightis a six- or seven-story officebuilding of 1960s' or 1970s'vintage. This reinforcedconcrete building is a typicalexample of a mid-heightstory collapse. The high riseto the left is a post-1981office building that has noapparent damage. Groundsettlement in the vicinity ofthese buildings was between30 and 60 centimeters.

EQE 17

Tokyo. The regulations have been reviewedand amended several times over the years asthe result of damage during subsequentstrong-motion earthquakes. Bridge codes andcodes for civil-engineering-type structures(e.g., quay walls) have undergone similarchanges over the years.

Since the 1926 code, Japan’s seismic codeshave typically been as advanced as any in theworld. Japanese engineers upgraded theirstandards after the 1968 Tokachi-oki Earth-quake in northern Japan and California’s1971 San Fernando Earthquake. In the early1980s, laws and orders concerning seismicdesign methods for buildings were exten-sively revised. The current Japanese seismicprovisions are specified in the Building Stan-dard Law Enforcement Order by the Minis-try of Construction (1981), and in theStandards for Seismic Civil Engineering Con-struction in Japan (1980). During the periodbetween 1971 and 1980, some lessons learnedin previous earthquakes were included in thedesign of major buildings, even though therequirements were not yet codified.

In the last several years, U.S. and Japa-nese professionals have been working to-gether to understand seismic performanceand to upgrade codes. Direct comparison ofthe codes for the two countries is difficultbecause of their different formats; however,comparative studies have suggested thatnewer Japanese mid- and high-rise buildingsare comparable to or somewhat stronger thantheir counterparts in the United States.

The current design philosophy in Japanis to keep seismic stresses within the elastic(non-damaging) range for earthquakes thatcan be expected to occur once or twice (mod-erate earthquakes) during a building’s lifespan, and to prevent collapse for larger, lessfrequent earthquakes. This means that for amoderate earthquake, the building is expectedto have little or no damage. A similar philoso-phy is used in the United States, although, ingeneral, more damage is considered accept-able for moderate-sized earthquakes.

Buildings are divided into four generaltypes in the current Japanese code. In gen-

eral, the guideline is: the larger the building,the more engineering and attention to qualityof the seismic-force-resisting system required.

■ Small buildings

- For small, one- or two-story woodbuildings and one-story buildings ofother construction types, prescriptiveconstruction requirements apply, and noexplicit design is required. A similarpractice is applied to wood-frame housesin the United States.

■ Buildings less than approximately 30meters high

- For buildings with a regular con-figuration, prescriptive requirements

Concrete-frame structurewith a mid-story collapse(Flower Road, Kobe).

18 EQE

apply. Additionally, a comparison ofcalculated and permissible stresses forthe loads associated with a moderateearthquake (0.2g peak ground acceleration)must be made.

- Irregularly shaped buildings arechecked using the same requirements asfor regular buildings. In addition,calculated drift (horizontal deflections)must be compared with allowable drifts,and the engineer must either (1) limitconfigurational irregularities, and meetminimum member size and d e t a i l i n grequirements that vary with constructionmaterial type, or (2) check the ultimatestrength at each floor level versus thedemands for a severe earthquake (1.0gpeak ground acceleration). The demandsfor a severe earthquake are amplifiedfor structures with large configurationalirregularities, and reductions in demandare made to account for the ductility ofthe construction type. Note: ManyJapanese buildings are quite irregular intheir configurations when compared toU.S. buildings, which makes them muchmore difficult to design for earthquakes.

Steel buildings less than 13 meters highcan be checked as regular buildings if the

Undamaged reinforcedconcrete school in theRokkomichi area. Thebuilding was used as arefuge center in the weeksfollowing the earthquake.

assumed moderate earthquake forces areamplified by 50%, and if the connections forthe braces and the frames are designed to bestronger than the braces, columns, and beams.

Concrete buildings less than 20 metershigh can be checked as regular buildings ifthey have a minimum combined shear walland column area at each story. (For largerbuildings, the minimum combined areas mustbe checked for each story in each direction.)

■ Buildings between the approximateheights of 30 and 60 meters

- These buildings are treated in the samemanner as are irregular buildings under30 meters high, except that an ultimatestrength check at each floor level for asevere earthquake is required.

■ Buildings more than approximately 60meters high

- These buildings require special permis-sion from the Ministry of Construction,and a dynamic (computer) analysis mustbe performed for the severe earthquakescenario. In practice, these buildings aresubjected to nonlinear analysis tech-niques. Peer review is also required.

EQE 19

It appears that, in general, buildings (otherthan smaller buildings) constructed using theabove provisions of the current code per-formed well in the earthquake and protectedlife safety. However, a number of newer build-ings, including high rises, were severely dam-aged and more damage may be uncovered asbuildings are carefully evaluated. Structuresthat did poorly included older houses andsmaller commercial buildings (both concreteand steel), and mid-rise concrete structuresdesigned and constructed prior to the early1980s using the same nonductile details thathad been employed in high-seismic U.S. re-gions up until the early 1970s.

Reinforced Concrete-frameBuildings

Many of the mid-rise structures in Kobewere reinforced concrete-frame buildings oftwo types: The older ones were of nonductileconcrete frame and the newer ones were ofSRC frame.

Dozens of reinforced concrete commer-cial buildings partially or completely col-lapsed at one or more floor levels. Typically,the buildings were 6 to 12 stories tall, and the

failure often occurred within the middle thirdof the building height. One possible contrib-uting factor was that the period of the strongground motion pulses may have been in arange that generally coincided with highervibration modes for these buildings. Thiswould have tended to amplify stresses in themiddle portion of the buildings.

Another possible factor was that therewere changes in building strength or stiffnessat these levels. For example, if shear walls orthe steel columns encased in concrete thatextend up from the foundation discontinueat a floor level, the strength and/or stiffnessof the structure above that floor may besignificantly less than at the floor below.

The pre-1981 code required that a concrete-frame building exceeding six stories in heighthave SRC construction for the lower six sto-ries as a minimum, although those buildingsfor which EQE engineers reviewed drawingsalways used SRC throughout the buildingheight. The older code also specified designlateral loading that is more uniform over theheight of the building, instead of having am-plified forces near the top and reduced forcesat the bottom, as is currently the practice in

Modern parking garage incentral Kobe (Chuo Ward).The building was undam-aged. The structural systemincludes steel moment-resisting frames and concreteshear walls.

20 EQE

Kobe City Hall, SannomiyaDistrict. In the foreground isthe old City Hall with amid-story collapse. Behind itis the new City Hall, whichexhibits signs of only minordamage.

EQE 21

Japan and the United States. The older code’spractice results in weaker upper stories.

Instances of concrete structures with col-lapses or failures in the bottom (ground) floorwere also fairly common. These failures typi-cally resulted from soft or weak stories cre-ated by the need for garages and the desire tohave numerous large open windows for store-fronts at the bottom floor. The high land costsand general congestion in Japan exacerbatethis problem. Very narrow multistory build-ings with open storefronts are very common.Irregular distribution of shear walls or con-crete frames resulted in substantial torsion,causing the structure to twist as well as swaydue to earthquake loading.

The damage mode most commonly ob-served was a brittle shear failure of concretecolumn elements, leading to a pancake col-lapse of the floor level above. The brittlefailures resulted from inadequate reinforc-ing details. In general, damaged columnswere observed to have lateral reinforcing(referred to as ties) with relatively large spac-ings. These ties typically had hooks at theirends that were bent only 90o. Consequently,when the earthquake struck and the concretecover outside the ties spalled or fell off, tiesopened up and could not provide the con-finement to the central concrete core. Com-plete failure quickly followed. Many of the

Top: Mid-height collapse of a concrete-frame building.

Bottom: Soft story collapse of a restaurant in Kobe.

22 EQE

Below: Severely damaged reinforcedconcrete building with shear walls atSannomiya Station, central Kobe. Thebuilding consists of a relatively simple(structurally and architecturally)upper portion on top of a complexlower portion. Inset: Detail of the shearwall damage at the setback level.

EQE 23

Nonductile Concrete

Few ties,stop at joint

Beam strongerthan column90o hooks 135o hooks

Bottom barsdiscontinuous

Ductile Concrete

Many ties,run through joint

Beam weakerthan column

Bottom barscontinuous

a way that they have less ductility than dotypical U.S. buildings in high seismic zones.

Hundreds of thousands of existing build-ings of similar nonductile construction arepresent in seismically active areas through-out the world. Unless these buildings areretrofitted, many lives will be needlessly lostin future major earthquakes.

Reinforced Concrete Shear WallBuildings

Many concrete shear wall buildings wereseverely damaged, and some had partialcollapses. Many of these were multifamilyresidential structures where the shearwalls had severe cracking, and horizontaldisplacements occurred at constructionjoints. One mid-rise concrete shear wall struc-ture overturned and fell into the street. Someof the damaged structures had concrete wallsin one direction only, and it appeared that the

damaged buildings in Kobe were also con-structed with undeformed reinforcing bars.

Similar nonductile concrete constructionhas been the source of building and elevatedhighway or overpass collapses in past earth-quakes, such as Southern California’s 1971San Fernando and 1994 Northridge earth-quakes. Current code requirements includecloser and larger ties of deformed steel, 135o

hooks that extend into the confined concrete,and cross-ties to supplement the rectangularties around the perimeter bars. In addition,ties must be closely spaced and extendthrough the joint created by the beams andcolumns. Buildings possessing these en-hanced detailing features are referred to asductile moment frames. “Ductile” refers to abuilding’s ability to dissipate energy anddeform without having brittle or suddenfailure. In general, designs produced usingthe Japanese code tend to result in strongercolumns and beams that are detailed in such

Diagram of typical detailingof ductile versus nonductilereinforced concrete columns.

24 EQE

Top: Mid-story collapse ofone wing of a Kobe hospital.

Bottom: Interior of a badlydamaged reinforced concretebuilding.

EQE 25

concrete frames had initially failed and al-lowed deformations, which caused damageto the shear walls in their weak or out-of-plane directions.

Failures of shear walls often led to per-manent offsets of one floor relative to thenext. This, in turn, led to damage of the framecolumns. It is not clear whether the walls inthese buildings were intended to function asthe primary lateral-load-resisting elements,or whether they were intended to share thisfunction with the reinforced concrete frames.

Again, the most severely damaged build-ings generally appeared to be of older con-struction, dating from about 1950 to 1980.Newer structures with configurations thatwere not too irregular and did not have softstories appeared to perform relatively well,generally ensuring the life safety of occupants.

Many severely damaged shear wall build-ings, including newer buildings, had unusualconfigurations by U.S. standards. These in-cluded dramatically varied architectural de-tails, such as many irregular wall openingsfor windows, in the lower floors. Such archi-tecture makes it much more difficult (andexpensive) to properly design the structuralsystem for earthquakes. Many severely dam-aged large commercial buildings had mixed-use occupancies—for example, stores in thelowest three floors and offices above. Typi-cally, the failures occurred in the lower sto-ries where the structural framing was moreirregular in order to accommodate large,clear spaces.

Reinforced Concrete-encasedSteel-frame Buildings

As previously mentioned, a popular con-struction type in Japan for the last 25 years isa structural steel-frame building encased inreinforced concrete, termed steel reinforcedconcrete (SRC). Older SRC buildings com-monly had solid structural steel elements inthe frame connections, but used trusses con-structed of smaller rolled steel shapes andplates in the center portions of the members.It appears that SRC construction gener-ally performed better than did the olderreinforced concrete-frame buildings; how-

ever, story collapses were noted in severalSRC buildings. Some of the collapsed build-ings thought to be concrete frame may actu-ally be partially SRC. This is due to therequirement in the old code that a buildingexceeding six stories in height must use SRCin the lower six stories, but can use reinforcedconcrete framing in the upper stories. Thatresults in a large stiffness and strength ir-regularity at the seventh floor. In newer con-struction of this type, the horizontal ties inthe concrete encasement around the steelshapes are generally spaced closer together,and the newer structures tended to per-form better.

Badly damaged olderreinforced concrete buildingin Sannomiya. Much of thedamage is concentrated atstructural discontinuities.

26 EQE

Top: Badly damagedreinforced concrete column.Note the heavy longitudinalreinforcement, with scantshear reinforcement.

Bottom left: The twobuildings in the foregroundlook similar, but theirdiverse performanceindicates that they areprobably structurallydissimilar. They may havebeen built at different times.

Bottom right: The collapseof this building was socomplete that it wasimpossible to deduce anobvious failure mode.

EQE 27

Steel-frame BuildingsGenerally, two types of steel-frame struc-

tures were observed, moment frames andconcentric braced frames. Many smaller steel-frame structures in the central business dis-trict had severe damage or collapsed. Ingeneral, such structures appeared to havebeen minimally engineered. In many cases,these damaged buildings contained relativelylight, flat-bar diagonal bracing memberswithin the side walls, which buckled or werefractured at connections. In some cases, lightsteel moment frames in the front of the build-ing were permanently distorted up to a fewmeters, causing the buildings to lean danger-ously. Fracture of welded connections wasobserved in several steel-frame buildings indowntown Kobe.

At the Ashiyama Seaside Town, 21 of 52mid- and high-rise condominium structuresbuilt between 1975 and 1979 had severe dam-age to the structural steel framing. This inno-vative and unconventional structural system

consisted of macro-steel moment frames inwhich the column and girder members werelarge steel trusses. Girders were typicallylocated at every fifth floor. Housing unitsconsisted of precast concrete assemblies thathad been brought to the site by barge. Dam-age observed included the brittle fracture ofsquare, tubular columns up to 50 centimeterswide with 5-centimeter-thick walls, and frac-turing of steel wide-flange diagonal bracingelements. Residual horizontal offsets in col-umn elements were observed to be as large as2 centimeters in some cases. In general, itappeared that the brittle fractures had oc-curred in framing elements subjected to highcombined tensile and shear stresses. In one ofthe units, six of the eight main steel columnsforming the lateral-load-resisting systemhad fractured.

Despite the serious damage to the steelframes, the other elements (including win-dows) of the buildings did not appear to havesignificant damage. The steel framing in these

A reinforced concretebuilding (shear walls in thetransverse direction) inNagata Ward, Kobe. Thebottom floor of this buildingcollapsed.

28 EQE

steel-frame buildings in the Los Angeles areaafter the 1994 Northridge Earthquake. Thismay have been the reason that several steel-frame buildings with no obvious major struc-tural steel damage were being demolishedtwo months after the Kobe Earthquake.

A common Japanese method of construct-ing steel moment-frame buildings incorpo-rates shop welding of beam stubs to thecolumns and field bolting of beam splices,away from zones of large strength demands.This practice has the advantage of allowingimproved quality control at critical locations.However, this does not eliminate all thevulnerabilities inherent in beam-columnconnections, and some fractures like thoseobserved following the Northridge Earth-quake were reported. Although this methodundoubtedly results in better-quality welds,it does not preclude the type of momentconnection damage observed after theNorthridge Earthquake.

In general, it appears that design phi-losophies and techniques used in steel con-struction in Japan result in structures withhigher degrees of redundancy than in theUnited States. In typical Japanese new steelconstruction, all of the steel frames in build-ings are included in the lateral-load-resistingsystem, whereas only a selected small num-ber of frames in many structures in the UnitedStates have been detailed to resist seismicloads. Similarly, many braced-frame struc-tures in Japan appear to have a large numberof smaller braces, whereas in the UnitedStates it is common to see a smaller numberof large braces. The redundancy provided bythe frames and braces results in more loca-tions where energy can be dissipated in a

modularly constructed buildings was locatedon the exterior of the building and was highlyvisible. In most high-rise steel structures inJapan, however, the framing is hidden byarchitectural elements and fireproofing. Con-sequently, there may be many other steel-frame structures where similar damage ispresent but hidden from view. That is whatwas observed with more than 140 modern

Top: Overview of theAshiyama Seaside Town,consisting of steel-framebuildings along the shorelineof Ashiya, across from RokkoIsland. The complex wasbuilt between 1975 and1979.

Bottom: Typical units at theAshiyama Seaside Town.Note the steel-truss elementsforming a moment frame forthe lateral-load-resistingstructural system.

EQE 29

Top: Detail of the truss elements at the Ashiyama Seaside Town. Note the minimaldamage to the concrete units. This photo shows the intersection of the vertical andhorizontal frames.

Bottom: Fractured web in a diagonal truss element, Ashiyama Seaside Town.

major earthquake. It is expected that suchredundancy provides added resiliency forthe buildings so constructed, and may havebeen a contributing factor to the relativelygood performance of modern steel struc-tures in the Kobe area.

Modern steel high-rise structures ap-peared to withstand the earthquake withlittle damage. With the lessons of the 1994Northridge, California, Earthquake and ofthe Ashiyama Seaside Town condominiumstructures, it will not be known how muchdamage steel buildings actually sustaineduntil more detailed investigations are per-formed on the structural connections.

Wood-frame BuildingsMost of the heavily damaged wood-

frame buildings were traditional one- or two-story residential or small commercialbuildings of Shinkabe or Okabe construction.These buildings normally have very heavymud and tile roofs (which are effective atpreventing typhoon damage), supported bypost-and-beam construction. Foundations areoften stone or concrete blocks, and the woodframing is not well attached to the founda-tions. The Shinkabe construction has mudwalls reinforced with a bamboo lattice. Okabeconstruction has thin-spaced wood sheath-ing that spans between the wood posts and isattached with limited nailing. The exteriorplaster is not reinforced with wire mesh orwell attached to the wood framing, so it fallsoff in sheets when cracked. In new (post-1981) construction, nominal diagonal brac-ing is required to resist lateral loads.

Traditional wood-frame constructionhad the most widespread damage through-out the region, resulting in the largest num-ber of casualties. Collapses led to the ruptureof many gas lines.

Failures in these buildings were typi-cally caused by large inertial loads from theheavy roofs that exceeded the lateral earth-quake load-resisting capacity of the support-ing walls. The relatively weak bottom storiescreated by the open fronts typically collapsed.

30 EQE

Top right: Severe racking of a steel building in Sannomiya.

Top left: Fractured building column at Ashiyama SeasideTown.

Bottom: Buckled diagonal brace in a parking garage.

EQE 31

Top: Severely damagedsteel-frame building inSannomiya. This buildingillustrates the practice ofbracing multiple bays.

Bottom: Steel buildings inthe Nagata Ward. While theframes appear to be in goodshape, the cladding wasshaken off the building,posing a severe life-safetyhazard.

Unlike most U.S. homes, Japanese homestypically have few if any substantive interiorpartitions to help resist the earthquake loads.In this respect, the bottom stories are similarto the U.S. homes that are supported onunbraced cripple walls.

In older homes, many framing membershad been weakened by wood rot. Soil fail-ures exacerbated the damage, because the

foundations have virtually no strength toresist settlement, and connections betweenthe residences and their foundations were weak.

The observed damage was reminiscentof that in the Marina District after the 1989Loma Prieta (San Francisco Bay Area) Earth-quake. In the Marina District, bottom storiesof old, multistory, wood-frame dwellingswere weakened by garage openings and a

32 EQE

Top: A steel building in theNagata Ward. The buildingwas probably very close tocollapsing in the earthquake.High flexibility of thebuilding probably contributedto the failure of the cladding.

Bottom: Collapsed housingwas responsible for most ofthe deaths in the earthquake.

EQE 33

structures was much better, although, in gen-eral, they tended to be located in the lessseverely shaken areas (on the hillsides). Newerhousing units, namely multistory concretebuildings and prefabricated single-familystructures, tended to perform better thandid their older wood-frame counterparts. Itwas common to find single newer structuresleft standing on otherwise destroyed blocks.

Top: These three-year-oldhouses collapsed fromexcessive loads imposed byheavy tile roofs. A similarhouse on the block with alightweight roof hadminimal damage (see bottomphoto).

lack of partitions on the ground floor. Entireblocks of homes swayed in one direction, andthe corner buildings or houses—which werethe weakest because they were open on twosides—were often pushed out into the street.

Impact between buildings occurred of-ten in Kobe’s residential areas. This interac-tion usually involved the lateral collapse of atraditional housing unit impinging upon aneighboring structure. The impact of theheavy roof from one collapsing house oftencaused the destruction of neighboring build-ings that probably would have otherwisesurvived the earthquake.

The poor performance of these struc-tures did not come as a surprise. Failures ofa large number of similar structures havebeen observed in past earthquakes, includ-ing the 1978 Miyagi-ken-oki Earthquake,during which 7,000 homes were destroyed. Itwas reported that some of the more modernand expensive house construction is similarto that in the United States and includesconcrete foundations, wood stud walls, andplywood sheathing. The performance of these

34 EQE

Unreinforced Masonry BuildingsFew unreinforced masonry buildings

were observed in the Kobe area. Those thatwere observed had extensive damage to themasonry walls and partial collapse of floorand roof systems. As could be expected fromthe performance of similar buildings in pastearthquakes, nonbearing gable walls wereespecially susceptible to damage. Had thesebuildings been occupied at the time of theearthquake, they would have posed an ex-treme life-safety hazard to the occupants andto passersby.

Base IsolationBase isolation is the name given to a tech-

nique that reduces the damage and vibra-tion of a structure subjected to earthquakemotions by isolating the building from theground motion through the use of mechani-cal bearings. These bearings are designed tolimit forces transferred from the foundationto the building. There was at least one suchbase-isolated structure in the Kobe area dur-ing the earthquake. This structure was lo-cated a significant distance (32 kilometerswest of Kobe) away from the area of greatest

destruction, and the intensity of shaking nearthe building’s location was not nearly assevere as the level of shaking in downtownKobe. Nonetheless, the reported data indi-cate that the base-isolation system workedwell, significantly reducing the level of mo-tion in the structure. The peak acceleration(0.10g) on the roof of the six-story isolatedstructure was reported to be only one-thirdthe level of acceleration of the ground (0.30g).This indicates that the isolation systemworked well in the earthquake.

Typical post-war housing—these houses are probably lessthan 20 years old. Thoseunits that did not collapsewere heavily damaged. Notethe very light framing withplaster or plasterboard overit. These buildings typicallyhave very little shearresistance in their walls andvery heavy roofs.

EQE 35

Somewhere between 3% and 5% ofJapan’s industry is located in the area ofstrong ground shaking in and around Kobe.This includes most types of industry—fromlight manufacturing to high-technology andheavy industry. As in most of Japan, andparticularly Tokyo, much of the industry isconcentrated near the port on landfill or veryrecent, soft soils. Due to strong ground mo-tion amplification on soft soils and the exten-sive ground failures (caused by settlementand liquefaction) in these areas, damage toindustry in the Kobe area was severe. Ob-served failures included extensive damageto large building foundations; all types ofindustrial buildings, equipment, and equip-ment systems; fire protection systems; racks;and inventory. The reduced ability to trans-port raw materials and finished goods to,from, and within the region will also greatlyimpact industry in the Kobe area. Industriesaffected include shipbuilding, steel plants,breweries, pharmaceutical firms, computercomponent manufacturing plants, and con-sumer goods production facilities.

Structural DamageAccess to industrial facilities in the re-

gion was very limited. The EQE team didhave access to the facilities of some U.S. andEuropean multinational companies, of whichthere is a large percentage in Kobe. In thosefacilities, structural damage was generallyminor. Other damage, however, was not.

Top: A heavily damagedsteel plant in the NishinomiyaPort area. The most apparentdamage was caused when thetop third of the concrete stacksheared off. The top portionof the stack plummeted intoa neighboring portion ofthe facility.

Bottom: Damage tounanchored laboratoryequipment.

INDUSTRIAL FACILITIES

36 EQE

The most severe structural damage, aswell as associated damage to exterior storageareas and tank farms, occurred to industrialstructures immediately adjacent to wharvesand other retaining structures at navigablewatercourses and other coastal facilities. Se-vere damage to industrial structures alongshorelines was observed from Nishinomiyato western Kobe. Numerous structures settledmore than 2 or 3 meters and were partially orfully submerged in water. Other structurespartially collapsed or tilted severely becauseof foundation failures. Most tilted structureswere probably buildings on mat founda-tions (usually pre-1980s’ vintage) withoutsupporting piles.

Wherever there was lateral spreading ofsoils and retaining structures along the shore-line, extensive damage was observed to tankfarms (tank tilting), silos (tilting and col-lapses), cranes, stacks, and other such struc-tures. Several tall, industrial, reinforcedconcrete stacks were leaning, and at least onecollapsed. The collapsed stack was observedat a steel facility along the waterfront inNishinomiya. The upper one-third of thestack broke off and embedded into theadjacent building.

Away from the shoreline, structural col-lapse of industrial facilities was relativelyrare compared to the collapses in the hous-ing, transportation, and commercial sectors.

EQE 37

Opposite, top: These saketanks appear to havesurvived the completecollapse of the traditionalwood-frame building thathoused them.

Opposite, middle: Theseunanchored tanks fell offtheir supports.

Opposite, bottom: Rockingand displacement of the quaywall was caused by lateralspreading of reclaimed land.This area, on theNishinomiya Port, was flatbefore the earthquake.

Left: A damaged port craneon Rokko Island. The damageoccurred when the quay wallmoved to the left from theoverall lateral spreading andsettlement of the island.There was differential lateralmovement of the two cranerails, pulling the crane legsapart. This phenomenoncontinued to occur for daysafter the earthquake.

38 EQE

Top: An example of a heavy concretestructure supported on a slab-on-grade.When the underlying soils liquefied andsettled, the building settled and rotated.Buildings on piles typically performedmuch better.

Bottom: Typical damage to Port of Kobefacilities. The large warehouse in the centerwas damaged when the interior slabsettled. The center of the roof wassupported by a column on the slab and waspulled in when the settlement occurred.Also note the severe displacement of thequay wall to the right.

Opposite: Severe damage to piers andwarehousing along damaged quay walls.The quay walls have rotated toward thewater, pulling the structures with them.The partially collapsed Hanshin Express-way can be seen in the background.

EQE 39

40 EQE

Top: Tanks in the port area.The ground shows signs ofmassive liquefaction andsettlement. The tanks appearto be on pile-supportedfoundations.

Bottom: Damage to facilitieson reclaimed land,Nishinomiya Port. The quaywalls have rotated anddisplaced, with the fullsurface dropping as much as3 meters in some areas. Inthe background is a badlydamaged bridge.

However, ground settlement caused exten-sive damage to the interiors of buildings, aswell as to the infrastructure that is routedinto the buildings. One case involved a pile-supported industrial building with a floatingfloor slab (or slab-on-grade). Although theperipheral piles successfully supported thestructure, the floor slab failed when the un-derlying soil settled; this failure pulled downthe columns that had supported the roof andcaused it to collapse. Loading platforms,roads, storage and parking areas, variousutilities, and other appurtenant structureswere often observed to be severely damaged.Such damage was particularly severe on thenumerous recently engineered islands inOsaka Bay. Many square kilometers of suchland were observed to be affected.

Steel manufacturers in the Kobe areawere severely affected by the earthquake.One steel company—Japan’s fifth largest—estimated that it would take months to re-sume full operations in its Kobe plant, whileanother steel company was unable a weekafter the earthquake to provide an estimateof restoration time. Both firms’ Kobe head-quarters buildings were declared unsafestructures and could not be occupied. It was

EQE 41

reported that four buildings at the first steelcompany collapsed, and the company wasconsidering closing its Kobe industrial facili-ties and shifting operations north to itsKakogawa plant.

Very large, multistory, reinforced con-crete shear wall warehouses on Rokko Islandand in central Kobe had very little damage.These buildings appeared to have superiorlateral strength and were evidently designedconsidering the contribution of heavy stor-age to the design earthquake loads. The ob-served damage to contents resulted primarilyfrom toppling of stacked goods or unanchoredstorage racks.

Small- and medium-sized manufactur-ing firms were heavily damaged. Structural,fire, or contents damage affected more than40% of the local knitted goods manufacturers

and more than 90% of the synthetic leathershoe manufacturing facilities. City officialsworry that production will now be moved tolow-wage countries like China.

Heavy damage to the numerous liquor(sake) production facilities also occurred inthe area stretching over Kobe’s Higashi Nadaand Nada wards and Nishinomiya. Aboutone-third of the country’s entire liquor pro-duction takes place in Kobe. Traditionalwooden plants and storehouses collapsed,and some reinforced concrete structures hadsevere damage. At many of the major facili-ties, modern reinforced concrete buildingsappeared to be undamaged. High-technol-ogy equipment housed in these structures,however, may have been severely dam-aged or destroyed, compounding busi-ness interruption losses.

A damaged concrete plant onreclaimed land east of RokkoIsland. The conveyors areseverely damaged, andseveral tanks have toppled.

42 EQE

Industrial SuccessFollowing a disaster of the Kobe magni-

tude, it is easy to focus on the negative—buildings and roadways collapsed, fire causeddestruction, industries were disabled, life-lines ruptured, and people were killed. It isimportant to understand, however, that asmany lessons can be learned from the suc-cesses of buildings and infrastructure duringthe earthquake. These successes indicate notonly that earthquake-resistant structures andfacilities can be built, but also that the prob-lems posed by intense earthquake motionson poor soil can be mitigated, even in an areaclose to the epicenter. The success stories ofbuildings and equipment that withstood anearthquake also provide valuable data onruggedness and survivability that are unob-tainable through other means. Design phi-losophies and calculation methods arevalidated when the fruits of these effortswithstand the effects of an earthquake.

In the Kobe area, there are many ex-amples of such successes. The relative lack ofdamage to modern pile-founded structuresin areas with tremendous soil liquefactionand settlement demonstrates the degree towhich the engineering of such foundationshas succeeded. Post-1981 buildings, in gen-eral, had a very good success rate. Whilethere are many examples of industrial facili-ties with considerable operational problemsand damage caused by inadequate anchor-age of equipment and fire protection sys-tems, there are also examples of industrialfacilities that had minimal damage becauseproper seismic equipment detailing had beenimplemented before the earthquake. Twosuch facilities were found on two of the re-claimed islands, which were severely hit ar-eas of Kobe.

Incinerator Facilities

One success story is an incinerator facil-ity on Rokko Island, which is used to burnsludge from a neighboring sewage treatmentfacility. The facility is a reinforced concreteshear wall structure supported on piles andwas completed in March of 1986.

The earthquake caused significantground settlement (up to 50 centimeters) atthe site. The facility’s equipment was wellsupported, anchored, and heavily braced,however, and there appeared to be no majordamage to either the equipment or the struc-ture. The incinerator was operable duringthe earthquake and for 30 minutes after theevent—at which time fuel for the backuppower source, a 1,000-kW gas turbine gen-erator, was exhausted.

The facility has had some trouble oper-ating because settlement of the roadway hasmade it difficult to weigh and unload trucksbringing waste to the facility. Loss of coolingwater has also caused some difficulty. Nor-mal operating procedures call for the use ofcooling water supplied from the HigashiNada sewage treatment facility, which wasrendered inoperable by severe damage dur-ing the earthquake. As an alternative to thiswater supply, the sludge center used sea-water for its cooling needs.

Although there was no instrumentationfor recording ground motion at the site, it isestimated that peak ground acceleration wasin the 0.6g range. This estimate is based onaccelerations recorded in the vicinity andthe level of damage to neighboring facilities.

There is a similar incinerator facility onPort Island. This facility is a six-story, re-inforced concrete, shear wall structurefounded on piles. Ground settlement aroundthe main structure was on the order of 50centimeters. The only structural damageobserved was partial failure of a walkwayconnecting the main structure and the stack.Like the Rokko Island facility, this plant wasbuilt to the post-1981 building code. Theonly known damage to the mechanical sys-tems was a leak in a small-diameter air linecontaining threaded connections. Similarequipment and details were found at theplant on Rokko Island.

EQE 43

Top: Silos and port facilities on reclaimed land next to Rokko Island. Tanks in the foreground are leaning. The silos in thebackground are severely damaged.

Bottom: Damage along the Port of Kobe shore. The severe rotation, lateral spreading, and settlement of the quay walls andfill material are typical of almost the entire developed border of the port. The building in the foreground has split into two parts.

44 EQE

Nonstructural DamageDifferential settlement and tilting of

ground-supported slabs within buildingsdamaged equipment. In one case, the slab-on-grade in a pile-supported structure settleddifferentially between the pile caps. Whilenot structurally significant, this resulted inextensive misalignment of manufacturingequipment. Re-leveling of the machinery wasexpected to take several weeks.

The shaking itself also caused damage tomore sensitive equipment and equipmentthat was not properly anchored. For someplants, short-term fixes to equipment thathad been affected by settlement involvedjacking up machinery as much as 30 centime-ters in order to achieve proper alignment.This procedure often caused significant de-lays in resuming production.

Breakage and leakage of fire sprinklerlines in manufacturing facilities were ob-served from Akashi to Osaka, resulting inextensive damage to manufactured goods,stock, and machinery. Virtually all of theleakage can be attributed to the failure ofunbraced or inadequately braced piping. For-tunately, there were no fires reported at thesefacilities. Had proper bracing been in place,considerable damage and business interrup-tion could have been avoided. It should benoted that it was not sufficient to simplyclean up the water damage to resume opera-tions. Repairs to the fire suppression systemsalso had to be completed.

One research facility located about 20kilometers northeast of central Kobe had onlyminor structural damage to most of its build-ings, and breakage of water and wastewaterlines caused by minor ground settlement.Extensive damage to the contents, however,was noted. Unanchored lab hoods shifted,bookshelves and cabinets toppled onto desks,and computer equipment fell to the floor.There was extensive breakage of glass jarscontaining a variety of chemicals. Most ofthis damage could have been easily preventedwith simple anchorage.

Other Causes of BusinessInterruption

Many of the industries affected by theearthquake are suppliers of parts for indus-tries outside the affected area. Since much ofJapanese industry relies on “just-in-time”delivery, damage to industry located in Kobeand the breakdown of the transportationsystem in the area are causing business inter-ruptions to a variety of industries not directlyaffected by the earthquake. Business inter-ruption insurance is typically not available inJapan, which will add significantly to theoverall industrial losses.

One report stated that by January 21, atleast four major electronics plants, six steel orheavy industrial plants, and three beverageplants had been shut down because of theearthquake. In some cases, facilities wereclosed because employees were unable to getto work, rather than because of severe physi-cal damage to the facility itself. By Monday,January 23, nearly one week after the earth-quake, many of these plants had reportedlyresumed at least partial production. Insome cases, the availability of water, gas,and power determined whether or not abusiness reopened.

A reduction in work force availability isan important factor in industrial operations.Personal tragedy, loss of housing, and thedebilitation of mass transit meant that manyemployees were unable to work right afterthe earthquake. This, in turn, means thatmany businesses will be unable to recoverfrom the disaster in a timely manner, whichmay bankrupt some industrial concerns.

Damage to major transpor-tation routes was a factor inbusiness interruption.

EQE 45

One of the most far-reaching and dis-turbing aspects of the earthquake was thesevere and extensive damage to the transpor-tation system. Kobe sits astride the principaltransportation corridor between the centraland southwestern parts of Japan’s main is-land, Honshu. The corridor is less than 5kilometers wide between Osaka Bay and themountainous terrain on the north side ofKobe. Earthquake damage to highways,bridges, and rail systems left Kobe’s citystreets as the only land access along thiscorridor, resulting in major congestion andgreatly impeded relief efforts. Many of thesesurface streets were also unusable, blockedby debris from collapsed structures and dam-aged by ground settlement. Use of alterna-tive road or rail lines added hours to normallyshort trips.

Damage to the transportation system hadthe potential to contribute greatly to the num-ber of fatalities. Had the earthquake occurredduring rush hour, there would have beenmany hundreds of fatalities on collapsedfreeways, and numerous crowded trainswould have derailed, in some cases plungingonto city streets.

Major Highways and BridgesTwo limited-access highways service the

Kobe-Osaka transportation corridor, theHanshin and Wangan expressways. Built inthe mid- to late 1960s, the Hanshin Express-way is the main through road and is almostentirely elevated for more than 40 kilome-ters. Much of the roadway is supported bysingle, large reinforced concrete piers spaced

A failed portion of theHanshin Expressway. Themost heavily damagedportions of the expresswayhad concrete road deck(background); the lessdamaged portions typicallyhad a steel superstructure(foreground). Where the twotypes joined, the heavierconcrete deck portion pulleddown the adjoining parts.

TRANSPORTATION

46 EQE

every 32 meters, many of which failed inshear or bending over a 20-kilometer length.Similar failures of the roadway occurred atmany locations, including complete top-pling of large reinforced concrete pillars

supporting a 500-meter section. It was ob-served that the road deck changed from steelto a heavier concrete section at the locationwhere this collapse occurred. These failureshave not only closed the Hanshin Expresswayfor an indefinite period, but have severelyimpeded traffic on Route 43, a street-levelhighway beneath the expressway.

Elevated highways in Japan typicallyconsist of single spans that have roller bear-ings at one end and are fixed at the other. Toconserve valuable space, single-column, can-tilever structures are common. Bearing widthson Kobe area expressways appeared to beinadequate in some instances. Column shear-ing revealed small reinforcing steel ties atrelatively large spacing. Failed welds at splicesof longitudinal bars were also observed.

The details that caused the failure andcollapse of the columns are similar to thosethat caused the failures of the older freewaysin the Los Angeles area in the 1994 NorthridgeEarthquake. These details received consider-able attention shortly after the 1989 LomaPrieta Earthquake, when the California De-partment of Transportation (Caltrans) initi-ated a massive program, amounting to severalbillion dollars, to strengthen all similar el-evated structures and bridges in California.The collapses in the Northridge Earthquakewere entirely of structures that had not yethad the post-Loma Prieta retrofits. Other

Collapsed sections ofexpressways.

EQE 47

CROSS SECTION OF HANSHIN EXPRESSWAY

Superstructure

Substructure

Pile

Footing

Pier

Bearing

Light pole

U.S. states, such as Illinois, Missouri, andWashington, have initiated similar programsin their earthquake-prone regions.

The Japanese design philosophy includesmuch larger columns, which result in stifferstructures that can be twice as strong as thosein the United States. Japanese-designed ex-pressways require 50% more steel and sit onsquat columns. However, it was assumedthat the column would have more strengththan required, so until recently, reinforcingsteel detailing has provided little ductility,resulting in more fragile structures. The con-sequence of unexpectedly large ground mo-tion can be catastrophic failures.

Japan has started five seismic retrofitprograms since 1971, but the major high-ways in the Kobe-Osaka area had not beenupgraded because the threat of earthquakeswas perceived to be low. So far, Japan hasretrofitted 25,000 out of 110,000 bridges. Cer-tain columns of the Hanshin Expresswaywere retrofitted with steel jackets that weresimilar to the jackets used by Caltrans. Thesedetails were found in areas that had beenstrengthened to accommodate the weight ofnew on- and off-ramps and did not appear tobe intended for additional seismic resistance.

Top: Diagram depicting a typical crosssection of the Hanshin Expressway.

Bottom: A failed hammerhead support forthe Hanshin Expressway.

48 EQE

Top right: Typical damage to a massive reinforced concrete column supporting the HanshinExpressway. This column had inadequate horizontal shear reinforcement.

Top left: Damaged vehicle on the remains of the Hanshin Expressway.

Bottom: Damage to columns supporting the Hanshin Expressway. The damage seen herewas probably caused by both horizontal shear and high axial loading. This is a result of boththe earthquake motion and the collapse of nearby portions of the freeway.

EQE 49

Top: Buckling failure (bending) of steel-encasedcolumns supporting the Hanshin Expressway.

Bottom: A steel-jacketed column (foreground)next to a non-jacketed column (far right).Although the steel-encased column shows signsof critical damage, it appears to have much morereserve capacity than the non-jacketed columndoes. It is thought that the jacketing was aretrofit for increased gravity loads—not forseismic resistance.

Typically, these details performed well, indi-cating that the Caltrans retrofit approachmay be correct.

The Hanshin Expressway is paralleled tothe south by the new Wangan (Harbor) Ex-pressway from Osaka to Rokko Island. TheWangan Expressway is a modern elevatedfreeway and includes several large bridges,some of which were under construction atthe time of the earthquake. The roadway issupported on a steel-frame deck. The deck issupported on bearing pads, which sit oneither steel or concrete piers. Long sections ofthe expressway have double decks. Thebridges along the expressway include longarch structures and a long cable-stayed sus-pension span. These, and many smaller spans,cross various navigable channels betweenthe mainland and numerous recently con-structed islands, which form parts of the Portof Kobe. All of the expressway is located onsoft native soils and engineered landfill.

The Wangan Expressway was severelydamaged along its entire length, fromNishinomiya Port to Rokko Island, a distance

50 EQE

of more than 8 kilometers. At many expan-sion joint locations, the bearings (hinges androllers) supporting the roadway structurewere damaged. In a number of cases, thebearings collapsed, allowing the road deckto drop from a few centimeters to more thana meter. This damage was caused by forcesthat exceeded those assumed in the designand very large longitudinal (along the road-way) and transverse (perpendicular to theroadway) displacements.

Typically, the columns (piers) support-ing the expressway appeared to be undam-aged, indicating that the current reinforcingdetails, in the case of concrete columns, areadequate, as opposed to the old details of theHanshin Expressway. The soil around thepile-supported columns typically settled froma few centimeters to more than 1.5 metersalong the length of the expressway. A fewcolumns rotated, almost dropping the high-way spans, because of ground failures—typi-cally lateral spreading.

The approach to a large, 252-meter-long,cable-tied, arch bridge in Nishinomiya col-lapsed, apparently the result of excessive

Top: Detail of the east end of thefailed approach span to a cable-tiedarch bridge on the WanganExpressway. This photographshows the failed mechanicalbearings, linkages, and other detailsof the roadway construction.

Bottom: Failed welds at splices oflongitudinal reinforcement in acolumn supporting the HanshinExpressway (at the 500-meter-longfailed section). These gas fusionwelds are 1960s’technology and arenot common today in the UnitedStates. All splices were at the samesection—another practice notcommon in the United States.However, these welds failed onlyafter the initial shear failureof the columns.

EQE 51

longitudinal displacements of the roadway.Due to the early hour, only two casualtiesresulted from the collapse. This bridge wasthree years old. The six-lane freeway will beunusable for an extended period while thisdamage is repaired. The bridge also hadsome distress: At least one of the cablessupporting the deck showed a large perma-nent lengthening.

At the other end of the Wangan Express-way, where it crosses to Rokko Island, asmaller (about 200 meters long) braced-archbridge failed, almost catastrophically. Thisbridge displaced laterally more than 30 cen-timeters and was in imminent danger offalling completely off its southern pier. Thedamage mode appears to be a bearing fail-ure. Marks on the pavement made by sliding,unanchored concrete crash barriers on theupper bridge deck indicate that the bridgedeck displaced laterally more than 1.2 meters.

Part of this displacement could have occurredbecause the bridge tilted when the southernbearing failed. Numerous approaches to theWangan Expressway, including the elevatedtoll plaza for the cable-stayed bridge, failedbecause of ground settlement bearing fail-ures, column rotations, and other inadequatesupport details.

The damage to this new expressway isquite disturbing. It indicates that the latestdesign criteria and practices for bridge struc-tures, at least in areas of soft soils, may beinadequate. In light of this damage, it isappropriate to reevaluate design criteria forboth new and retrofit designs of bridgesworldwide. This is the first real earthquaketest that large modern bridges have had, andthe results are rather disappointing.

The Akashi Ohashi Bridge, with a mainspan of 1,990 meters, will be the world’s

Pier failures beneath theGreat Nishinomiya Bridge.At least two of the bridge’ssix supporting piers wereseverely damaged. Damageappears to be caused by largesettlements at the bridge’sabutments, probably over-loading the first set of piers.

52 EQE

Top: The collapsed approachto the 252-meter-long mainspan of the NishinomiyakoBridge east of NishinomiyaPort on the new WanganExpressway. The failure wascaused by inadequatestrength and capacity of themechanical bearings to takevery large displacements inboth the transverse andlongitudinal directions.

Bottom right: Groundfailure at the base of anabutment of a bridge thatcrosses a shipping channelalong the Wangan Expressway.

Bottom left: In the sameabutment, the bent can beseen to have rotated, andthere is evidence oflongitudinal movement ofthe roadway and lateralspreading of the soil.

EQE 53

Top: The cable-stayed bridgelinking Rokko Island to themainland. Note the failed 46-centimeter water linedangling beneath the deck.All monorail lines to theisland failed in theearthquake; the other bridgeto the island is the steel-archbridge shown on page 54.

Bottom: Failed bearingswere typical along theWangan Expressway. Thisphotograph shows a 50-centimeter vertical drop atthe expansion joint from thebridge approach to the firstspan of a large cable-stayedbridge. The failures weretypically more dramatic atboth sides of major crossingsalong the Wangan Expressway.

54 EQE

RailwaysElevated railroad structures and railway

stations were particularly hard hit. Threemain lines (JR West, Hankyu, and Hanshin)run through the Kobe-Osaka transportationcorridor, generally on elevated structuresand embankments. All the lines had elevatedstructure and embankment failures, over-pass collapses, distorted rails, and other se-vere damage. A large number of cars weredamaged, and some fell onto city streets.Several stations and several kilometers ofreinforced concrete elevated structures weredestroyed, and numerous spans collapsed.

Top: The main span of asteel-arch bridge for theWangan Expressway. Notethe discontinuity in the roaddeck and the tilt of thebridge to the right.

Bottom: An overall view ofthe same bridge. The span ison the verge of falling off thepier at the south (closest)end of the bridge. Thebearings supporting this endof the bridge failed.

largest suspension bridge when it is com-pleted in 1998. It is currently under construc-tion and will span the strait between Honshuand Awaji islands. The fault that rupturedreportedly passes between the two towers ofthe bridge, and the two abutments havemoved apart more than 1 meter. At thistime, it is not known if significant damageoccurred to the structure.

Numerous other bridges, both new andold, were damaged. Other highways androads were also damaged. Most of the dam-age to roads was from ground settlement,typically caused by liquefaction.

EQE 55

The Rokkomichi Station (built in 1972) of theJR West line was virtually destroyed.

The Shinkansen (Bullet Train) was con-structed circa 1964. Most of its path in theKobe area is through two long tunnels underRokko Mountain. No information on the tun-nels’ performance was immediately avail-able. At the east portal of the tunnel, the lineis carried on an elevated viaduct built in1968. For a length of 3 kilometers, this via-duct was severely damaged, with a numberof the longer spans collapsing. In general,these collapses were caused by shear failureof the supporting concrete columns.

Included in this area is a multispan cross-ing of a river, consisting of approximately 40-meter spans carried on large, single, reinforcedconcrete piers, all of which had severe crack-ing and transverse reinforcement failure.

It was estimated that service on theHankyu line and the Shinkansen would bedisrupted for six months. The JR and Hanshinlines were expected to be interrupted forlesser periods. Within about a week after theearthquake, rail service was restored to Kobefrom Osaka via a major detour, which resultedin a 2-hour trip versus the normal half hour.

The damage to the elevated rail struc-tures and to the stations was predictable.Similar damage has occurred in numerouspast earthquakes, most significantly duringthe 1971 San Fernando Earthquake near LosAngeles. Japanese engineers believed thatthe columns were adequate, because the col-umns were typically larger in cross sectionand therefore stronger than similar struc-tures in California. The structural details thatresulted in damage to the Shinkansen and tothe other elevated structures, however, werevirtually identical to those that resulted inthe collapse of the nearby Hanshin Express-way, and were similar to those that resultedin earlier California highway damage in 1989

Top: Failure of the elevatedtrain railway nearRokkomichi Station.

Bottom: Heavily damagedviaduct of the Shinkansen(Bullet Train).

56 EQE

Top: Destroyed railway carstorage structure.

Bottom: Collapsed two-storyRokkomichi Train Station. Thestructure collapsed due toinadequate shear reinforcement inthe supporting columns.

Opposite, top: Failed columnsupporting Rokkomichi Station,built in 1972. Note the minimalamount of shear reinforcement andnonductile detailing of the steelreinforcement.

Opposite, bottom: The severelydamaged Sannomiya Station in thecenter of Kobe.

moving trains had the earthquake occurredlater, and the number of deaths could havebeen even higher.

In the port area, Rokko Island is servedby a dual monorail (people mover), which

and 1994. The damage to the Shinkansen isparticularly troublesome, because the line isso critical to Japan. Further, it was fortunatethat the earthquake struck about 14 minutesbefore trains in Kobe began to move. Thou-sands of commuters would have been in

EQE 57

58 EQE

Top: A section of theseverely damagedShinkansen (Bullet Train)elevated roadway. Thestructure is very stiff andvery strong, but hadinadequate shearreinforcement.

Bottom: Daikai SubwayStation, Kobe. Columnssupporting the roof atmidspan have failed in shear,resulting in the roof’scollapsing.

collapsed at several locations (downtownKobe, on Rokko Island, and on a major cable-stayed crossing to Rokko Island).

Kobe Subway System Damage to underground facilities, such

as mines, tunnels, or subways, is rare inearthquakes. An unusual example of severedamage to this type of facility occurred in theKobe subway system, a two-track line runningunder central Kobe, which was generallybuilt by cut-and-cover methods in the mid-1960s. The double track is typically carriedthrough a concrete tube 9 meters wide by 6.4meters high, which widens to 17 meters at the

stations. The tube typically has about 5 metersof overburden, which is supported by 0.4-meter-thick walls and roof slabs. The wallsand roof slab are supported midspan (be-tween the tracks) by a series of 5-meter-tall,1.0-meter-long by 0.4-meter-wide reinforcedconcrete columns.

At the Nagata and Daikai stations, and inthe tube section between, the between-trackcolumns failed catastrophically in shear, drop-ping the roof slab almost onto the tracks overabout a 90-meter length. More than 30 of the35 piers supporting the platform and theceiling at the Daikai Station were badly dam-aged, causing a 3-meter-deep cave-in on thestreet above, National Route 28. At theSannomiya Station, 20 piers were damaged.Because the failed section is central to theentire Kobe system, most of the subway is outof service, with repairs scheduled to be com-pleted in the fourth quarter of 1995.

Failure of the columns was caused byexcessive deflection of the roof slab dia-phragm combined with very light transverse(shear) reinforcement, relative to the main(bending) reinforcement. Excessive deflec-tion of the roof slab would normally be re-sisted by (1) diaphragm action of the slab,

EQE 59

Massive depression in theroadway caused by partialcollapse of the DaikaiSubway Station below.

60 EQE

lize passive earth pressures. In effect, thetube behaved almost as a freestanding struc-ture with little or no extra support frompassive earth pressure.

AirportsKansai International Airport was only

recently completed (1994) on a human-madeisland some 30 kilometers to the southeast ofthe epicenter. Itami is the former interna-tional airport for Osaka and now handlesmuch of the domestic traffic. It lies about 10kilometers east of the heavily damaged area.Neither airport appears to have been signifi-cantly damaged, but neither was located inan area that had severe ground motion.

supported by the end walls of the station, and(2) passive earth pressure of the surroundingsoils, mobilized as the tube racks. Diaphragmaction was less than anticipated, however,due to the length of the station.

The lack of passive earth pressure isextremely interesting, because it highlights acommon cause of failure—the differencebetween design assumptions andconstructibility. That is, designers in thissituation often assume that passive earthpressure of the surrounding soil will providepartial support for the tube under lateralloading. Because of the method of construc-tion (cut-and-cover, involving sheathed ex-cavation with narrow clearance between thesheathing and the tube wall), compaction ofbackfill would have been difficult to impos-sible, resulting in the tube’s inability to mobi-

Cou

rtes

y Ja

pan

Rai

l Cor

pora

tion

Cross sections of Daikai SubwayStation after the earthquake,showing the roof collapse.

EQE 61

The Port of Kobe, one of the largest con-tainer facilities in the world, sustained majordamage during the earthquake. In effect, theport was practically destroyed. The total di-rect damage to the port will easily exceedU.S.$11 billion. Reconstruction costs, if infact the port is restored to its pre-earthquakecondition and capacity, will probably exceedthe cost of the damage. The port complex,constructed on three human-made islands—Maya Container Terminal, Port Island (withan area of 10 square kilometers), and RokkoIsland (with an area of 6 square kilometers)—accounts for approximately 30% (2.7 millioncontainers per year) of Japan’s container ship-ping. At the time of the earthquake, the threefacilities included 27 active container berthsand various other wharves, ferry terminals,roll-on facilities, and warehousing. In addi-tion, the older parts of the port contain nu-merous other facilities, such as an extensiveshipyard. Also, at the time of the earthquake,several new islands were under develop-ment, and new berths were under construc-tion to the east of Rokko Island.

The port is owned and managed by thePort and Harbor Bureau of the Kobe City

Government, and the various berths areleased to more than 50 major internationalshipping lines. The container port was con-structed in phases commencing with theMaya Container Terminal, followed by PortIsland and then Rokko Island, where con-struction started in 1972. A second develop-ment phase on Port Island, including fivecontainer terminals and other berths, wasunder construction and nearly completewhen the earthquake struck.

Top: Water behind a failedquay wall. Caissons havetipped and slid toward thewater, allowing thesupported deck to collapsebehind them.

Bottom: Failed structures atthe edge of the port acrossfrom Rokko Island.

PORTS

62 EQE

Top: Damage to crane legs at a port facility. Damagewas the result of the relative movement between thecrane rails on top of the quay wall and those supportedon fill.

Middle: Lateral spreading, settlement, and liquefac-tion along the shore of the Port of Kobe.

Bottom: Detail of a failed gantry crane, Rokko Island.

Opposite: Typical failure of a caisson wall; the wallhas moved to the right (lateral spreading), allowingthe fill behind to settle. This island is newly con-structed. An unused caisson can be seen to the left.

Sixty-seven gantry cranes operated inthe various berths. Fifty-five were wide-gage(30.5 meters between the rails) and 12 werenarrow-gage (16 meters between the rails).The cranes were fabricated by different manu-facturers including Pacheco, Sumitomo,Kawasaki, Mitsubishi, Mitsui, and IHI. Thesecranes have lift capacities ranging from 30 to40 tons, and their loading outreach rangesfrom 36 to 40 meters. At the time of theearthquake, most cranes were in their stowedposition with the pins engaged.

Land reclamation at all three mainhuman-made islands and at numerousother parts of the port was done by means

EQE 63

64 EQE

of gravity-founded, concrete caisson quaywalls enclosing the perimeters of the islands.Prior to installing the caissons, some of thenative soils beneath were dredged and re-placed by a screeded sand and gravel base.The caissons are rectangular in plan and haveinternal bulkhead walls, creating a cellularstructure, which is designed to withstand theimposed hydrostatic pressure while afloat.The caissons were lowered on the preparedseabed by filling the caisson cells with sandballast. Caisson dimensions vary, but in gen-eral, they were designed to accommodatewater depths in the range of 13 to 14 meters.

Once the quay walls along the islands’perimeters were nearly complete, granularfill, quarried from the nearby Rokko Moun-tain, was transported and placed using bot-tom dump barges. In general, only the upperfew meters of fill above the mean sea levelwere compacted in an engineered manner. Aconcrete cap beam was placed over the cais-sons to support the waterside crane rails andto provide some continuity among the quaywall caissons. Narrow-gage landside cranerails were mounted on top of pile-supportedgrade beams. Wide-gage landside crane railswere mounted on grade beams supported bythe fill. Storage yards were generally toppedwith asphalt or concrete pavement.

The predominant damage to the portfacilities resulted from soil liquefaction andlateral spreading. A large number of thegravity-founded quay wall caissons rotated

and slid outward. Soil settlements immedi-ately behind the caissons were as much as 3meters and generally decreased toward thecenter of the islands. Pile-supported struc-tures remained at their original elevations,while the surrounding ground settled sub-stantially. Significant quantities of sandwere ejected because of liquefaction andcovered large portions of the pavements.Most gantry cranes were damaged, and onecollapsed because the quay wall caissonswere displaced.

Damage to the gantry cranes was in theform of leg and cross-beam buckling, as wellas rupture at the wheels. The extent of buck-ling varied, depending primarily upon therelative horizontal displacement resultingfrom the movement of the caissons. Rela-tively few cranes jumped off the tracks, whichcan be attributed to most of the cranes beingin the stowed position with their pins en-gaged at the time of the earthquake. Numer-ous other cranes throughout the port weredamaged because of foundation damage.Several cranes collapsed; some collapsedbecause of structural damage caused by iner-tial forces generated by the earthquake.

Cranes damaged by lateralspreading, Rokko Island.

EQE 65

Quay wall caisson displacements, whichundoubtedly propagated the major damagein the port, may be attributed to severalphenomena. Earthquake accelerations ap-plied to the massive sand-filled caissons re-sulted in large horizontal forces, which mayhave exceeded the sliding resistance offered

by the base. This can be further aggravatedby the rocking motion of the caissons, whichmay result in excessive bearing pressures atthe toe of the base. The latter coupled with thepossibility of liquefaction may explain theobserved tilting of many caissons. Since theislands’ fill placed below water was dumped

Sandballast

Gravel

Prepared base

Engineeredfill

Sand reclamationbarged & bottom

dumped

Piles

31.85 m January 17, 1995

30.50 m January 16, 1995

January 16, 1995

January 17, 1995

Settled grade beam

Displacedposition

Top: Partially submergedpier and buildings at theolder section of the Portof Kobe.

Bottom: Typical quayconstruction, Rokko Island.

66 EQE

from barges, it was relatively uncompacted.Hence, soil settlements resulted from lateralspreading as well as compaction. Such settle-ments continued during the first few daysafter the earthquake and are likely to con-tinue for some time, especially in the event offurther aftershocks.

Severe damage to other types of piersand their quay walls was observed through-out the port. In the older parts of the port,particularly to the south of central Kobe, suchas Hyogo Pier, large parts of piers weresubmerged because of massive soil settle-ment. Settlements in excess of 2 to 3 meterswere observed. Numerous warehouses andother facilities were also submerged and/orseverely distorted and damaged because ofground settlement. Severely damaged, par-tially submerged, and collapsed cranes wereobserved throughout the older parts of theport, the shipyards, and other facilities.

Container and other operations at thePort of Kobe were essentially halted by theearthquake. Other ports at nearby Osaka,Nagoya, and Yokohama are being used toprocess container and other traffic. Theseports had little damage.

Container pier on RokkoIsland. Severe damage to thepier and cranes was causedby lateral spreading. Notethe damage to the legs of thecranes.

Ferry service was gradually restored inthe days following the earthquake. While thedirect damage to the port is severe, the loss ofport service to the local economy is devastating.

Much of the damage to the containeroperations was caused by the failure of thecaisson quay walls. All of the walls examinedby the EQE team were severely damaged.Many of these walls were new or were underconstruction and presumably designed tothe latest standards. This generic failure indi-cates that the current design standard forsuch caisson walls is inadequate. In addition,the generic failure indicates that other com-monly used systems for quay walls—such asthose used on the West Coast of NorthAmerica, and which have not yet been testedby very strong earthquakes—may also havegeneric weaknesses. Fortunately, these de-signs rely on pile-supported systems and arequite different from the caisson walls thatfailed in Kobe. Considering the devastationin the Port of Kobe, it would be prudent forU.S. and other ports in earthquake regions toevaluate carefully the designs that are com-monly used. Detailed and independent riskanalyses are one method that may be used toassess their strengths and weaknesses.

EQE 67

Lifeline performance varied in this event,with electric power and telecommunicationsquickly restoring most system functionality,and water, wastewater, and gas basicallylosing service to most of Kobe. All of theselifelines, however, were extensively damaged.

Electric PowerEQE, with support from the Electric

Power Research Institute (EPRI), dispatcheda team of engineers to investigate the perfor-mance of the electric power system duringand after the earthquake. The following arepreliminary observations based on field in-vestigations and technical information pro-vided by Kansai Electric Power CompanyInc. (Kansai Electric).

Kansai Electric’s service area covers28,663 square kilometers, which includes thecities of Osaka, Kyoto, and Kobe, three ofJapan’s major economic and cultural centers.The service area covers 8% of Japan’s totalland area and 16% of the nation’s electricityconsumption. Kansai Electric operates aninstalled generating capacity of 35,035 MWto service the highly urbanized and industri-alized region: 18,581 MW nuclear; 9,768 MWfossil fuel; and 6,686 MW hydroelectric. ThePower System Engineering Department man-ages 62 control centers; 868 substations; 15switching stations; 10,819 kilometers of over-head transmission lines; and 1,740 kilome-ters of underground transmission lines.

The Kansai Electric transmission systemincludes a 500-kV system that loops aroundthe region, connecting to nuclear generationplants in the north and neighboring utilitiesto the east and west, and looping aroundOsaka to the south. The earthquake’s epicen-ter was located in one of the few areas of theKansai Electric service area that would notsignificantly affect the 500-kV transmissionsystem. The earthquake’s location also spareda cluster of fossil fuel plants to the southeastfrom the high ground motion and did notaffect the nuclear power plants located morethan 100 kilometers to the north. The greatestdamage occurred at 187- and 275-kV substa-tions, a few of the fossil plants, and a gas

turbine plant. Details on distribution systemdamage are not currently available.

Widespread blackouts such as those dur-ing the Northridge Earthquake did not oc-cur, in part because of the earthquake’slocation relative to the 500-kV loop. All of the500-kV substations continued normal opera-tion. Nevertheless, 1 million customers werewithout power for a few hours following theevent. More than 4,700 restoration crewsfrom Kansai Electric, contractors, and sixother electric utilities successfully restoredthe system within a few days. System resto-ration in terms of the number of customers

Undamaged, well-anchoredash filter at a waste disposalplant, Port Island.

OTHER LIFELINES

68 EQE

without power following the earthquake issummarized in the table above.

Power-generating station damage dur-ing the earthquake was limited to 10 of KansaiElectric’s 63 fossil-powered units, with 12out of 36 units that were on-line before theearthquake tripping off-line. None of the 141hydroelectric power stations or three nuclearplants (11 units) were damaged.

Ten units with a total capacity of 1,631MW were damaged. Five units had a fewbroken boiler tubes each. Amagasaki-HigashiUnits 1 and 2; Amagasaki No. 3, Units 1, 2,and 3; and Higashi Nada Gas Turbine Units1 and 2 had substantial ground settlementand liquefaction. Himeji No. 2, Unit 2, had asafety valve seat leak that was most likely notearthquake related.

The EPRI/EQE reconnaissance team vis-ited Amagasaki No. 3, Higashi Nada, and theItami Substation. The two generating sta-tions had the highest ground motion and themost serious damage.

Amagasaki No. 3 is an oil-fired steamplant that includes three units with a capacityof 156 MW each. Horizontal peak groundaccelerations at the site were about 0.3g, with

1. Does not include more than 70,000 customers whose buildings or houses were destroyed bythe earthquake or the fire thereafter.

TIMEDATE CUSTOMERS WITHOUT POWER

KANSAI ELECTRIC POWER RESTORATION

(Data not available)5:46 A.M.

7:30 A.M.

8:00 P.M.

5:00 P.M.

7:00 P.M.

1:00 P.M.

9:00 A.M.

9:00 A.M.

3:00 P.M.

1,000,000

500,000

200,000

110,000

70,000

17,000

2,000

01

January 17

January 17

January 17

January 18

January 19

January 20

January 22

January 23

January 23

Failure of seismic ties for theboiler at the AmagasakiNo. 3 plant.

EQE 69

significant ground settlement observed. Themost significant damage was the failure ofseismic ties between the boiler and its sup-port structure. Preliminary observations in-dicate that the ties were not loaded linearly,resulting in dramatic failure of the ties nearthe top of the structure. U.S. fossil plants withsuspended boilers have undergone similarground motion but have seismic stops ratherthan ties. The stops absorb energy by deform-ing during an earthquake, while ties are stron-ger but less ductile. Both the seismic tiesobserved at Amagasaki No. 3 and the stopsfound at U.S. stations have had substantialdamage from moderate ground motion, al-though stops do appear to prevent damage tothe boiler.

Amagasaki No. 3 also had a small pipefailure caused by steam drum displacementrelative to the structure and miscellaneous

failures from ground settlement. Pipe andelectrical raceway supports adjacent to thebuilding foundation were damaged byground settlement. The piping and racewaysremained functional.

Higashi Nada includes two gas turbineunits that were constructed in 1974 on re-claimed land. The soil conditions at the siteconsist of 7 to 8 meters of very soft clayoverlying a layer of sediment more than 10meters thick. The plant underwent a peakground acceleration of about 0.6g during theearthquake, but it was not operating at thetime because the plant is used only duringpeak demand periods, typically in thesummer months.

Major buildings, equipment, and tanksat Higashi Nada are founded on piles. Thereare four major flat-bottom tanks, including

Pipe failure caused by steamdrum displacement at theAmagasaki No. 3 plant.

70 EQE

two fuel tanks with a weight of about 4,000kilograms, one raw water tank (about 1,000kilograms), and one purified water tank(about 500 kilograms). All the tanks are sup-ported on concrete foundations without an-chorage, and the foundations are supportedon 30-meter-long precast concrete piles. Theground settled near the tanks by as much as70 centimeters. As a result of the settlement,the tops of the piles could easily be seenbeneath the foundations. The foundationswere observed to have tilted slightly, with nodamaging effects to the tanks.

The buildings and equipment at the plantwithstood the earthquake without directdamage. However, differential settlement offoundation slabs did result in misalignmentof equipment on adjacent foundations. Atseveral locations, ground settlement exceededa meter relative to the pile-supported foun-dations. Piping systems had substantial de-formation, but the only failure was associatedwith a clamped mechanical coupling. In sev-

eral cases, the ground settled to such anextent that several pipe supports and theirconcrete foundation blocks were left dan-gling in the air, supported by the pipes towhich they were attached. Virtually all dam-age was related to ground settlement andrelative displacement between foundations.

Nine 275-kV substations were damaged,including bus disconnect switch failure, trans-former oil leaks, transformer anchorage fail-ure, transformer bushing failure, and othermiscellaneous damage. Liquefaction andground settlement were evident. The exten-sive use of dead-tank gas-insulated and oil-filled circuit breakers resulted in positivecircuit breaker performance.

TelecommunicationsTelecommunications systems did very

well in the earthquake, with very few serviceinterruptions. Telephone service was avail-able on a limited basis in the most heavilyBroken arresters at the

Itami Substation.

EQE 71

These expansion joints atpump discharges at a PortIsland wastewater disposalplant had no damage.

As a result of extensive ground settle-ment and other failures, underground waterpipelines were severely damaged in the earth-quake, with approximately 2,000 breaks re-sulting in general lack of service in Kobe,Ashiya, and Nishinomiya. The massive dam-age to the water transmission lines causedthe tanks without automatic shutoff valves todrain in the first 1 to 8 hours after the earth-quake. By the time the fires had started,much of the unreserved water had alreadydrained from the system. With the transmis-sion lines destroyed, the reserve water wasalso unavailable for fire fighting.

Nine days after the earthquake, 367,000households in Kobe still had their water sup-ply cut, and 98% of Ashiya households and85% of Nishinomiya households were alsolacking service. The population in the heavilyimpacted areas was notified to plan for nowater service for about two months.

damaged areas on January 18, although aweek after the earthquake, more than 25,000telephone lines were still disconnected in the18 municipalities hardest hit. More than 2,000telephones were installed for public use atshelters and public offices.

WaterThe Kobe area had a water system de-

signed to be operable after earthquakes. Thereare approximately 30 reservoirs supplyingwater to the Kobe area through a gravity-fedsystem. Of these, 22 reservoirs had automaticemergency shutdown valves and multiplestorage tanks. In the event of an earthquake,these valves are designed to automaticallyshut off water flow out of half of the reservoirtanks. All 22 valves tripped and workedcorrectly. This enabled 30,000 cubic metersof water (8 million gallons) to be stored inreserve in the reservoirs equipped with auto-matic shutdown valves.

72 EQE

homes have automatic gas shutoff systems,but many failed to work because of buildingcollapses, other building damage, and bro-ken pipes. The population in the heavilyimpacted areas was also notified to expect nogas service for about two months.

Information on other gas system facili-ties was not available at the time of thiswriting, although a large gas holder near thePort of Kobe did not have any obvious struc-tural damage. There are a number of petro-leum and other at-grade fuel tanks in the portarea, the largest being perhaps 25 meters indiameter and 15 meters high. Only a fewwere observed to have any damage, and onlyone was observed to have collapsed. Many ofthese tanks were at-grade and freestanding,while some were bolted to their foundations.Most appeared to have fixed roofs.

Several liquefied petroleum gas (LPG)tanks exist in the port area, and one wasreported to have cracked, resulting in thetemporary evacuation of 70,000 people. Twogroups of three large spherical tanks wereseen along the waterfront in Kobe. They werewell braced with heavy diagonal pipe brac-ing between column supports and appearedto have no damage. There were no reports ofliquid fuel pipe breaks, with the exception ofone line at Kansai International Airport.

Workers repairing damage toa water treatment plantacross from Rokko Island.

Emergency water distribution was verylimited in the days after the earthquake, withcitizens resorting to buckets, tubs, and othermakeshift containers in order to haul limitedquantities of water to their homes from arelatively few tank trucks. In areas near theharbor, private construction and shippingcompanies brought in tugboats and othersmaller vessels and set up water distributioncenters for nearby neighborhoods, suppliedby condensers on the ships.

The condition of the sewer system wassimilar to that of the water supply system.This damage was of less consequence sincewater was not available for flushing toilets.Public rest rooms overflowed, and sanitationwas a concern.

In summary, the city of Kobe had madea sincere effort to provide an earthquake-resistant water and fire protection system(see the following section, “Fire FollowingEarthquake”). However, the system failedand was not scheduled to be fully restoredfor many months.

GasThe gas system had at least 1,400 breaks

in its underground distribution system, pri-marily at service lines, with general curtail-ment of service by Osaka Gas Company to834,000 households. Japanese buildings and

EQE 73

Large fires following strong earthquakeshave long been considered to be capable ofproducing losses comparable to those result-ing from the shaking.

The risks are particularly high in Japanbecause of high population densities; verynarrow streets and alleys, which cannot actas fire breaks; numerous old wood-framesmaller commercial and residential build-ings mixed in the commercial zones of towns;unanchored or unprotected gas storage tanksor heaters; and a mix of collapse-prone oldbuildings in all built-up areas. These riskswere most recently exhibited in the large fire thatdestroyed much of the town of Aonae on theIsland of Okushiri during the Mw7.8, July 12,1993, Hokkaido Nansei-oki Earthquake.

Many Japanese municipalities, and par-ticularly Tokyo, have long considered

earthquake-generated fires to be very highrisks, and various risk management programshave been started in Japan. Kobe, for ex-ample, had specially constructed under-ground cisterns for fighting fires if parts or allof the distribution water lines failed. How-ever, whatever measures had been taken inKobe were overwhelmed following the Janu-ary 17 earthquake.

The Kobe Fire Department (KFD) is amodern, well-trained fire response agency,organized into Prevention, Suppression, andGeneral Affairs sections, and a Fire Acad-emy. The city is divided into 11 wards for fireprotection purposes. KFD maintains 11 firestations and 15 branch stations, served by1,298 uniformed personnel. Equipment in-cludes two helicopters, two fireboats, and196 vehicles. Other equipment includes 72portable pumps. Fire engines carry predomi-

FIRE FOLLOWING EARTHQUAKE

Fire in central Kobe.

UP

I/K

yodo

74 EQE

Top: The state of the fires at4:30 P.M. on January 17.

Bottom: Burned area in theNagata Ward.

EQE 75

nantly 50- and 65-millimeter hose; larger hoseis not available except for drafting purposes.

KFD has a civil disaster prevention pro-gram as well as a cadre of volunteer fire corpswith about 4,000 members. This corps pro-vides the first on-scene engagement of thefire, performing functions such as giving di-rections to arriving emergency vehicles andhelping to guide people to safety.

Fire water is primarily from the city wa-ter system, served by gravity from 30 reser-voirs. Of these, 22 have dual tanks, with onetank having a seismic shutoff valve so that, inthe event of an earthquake, one tank’s con-tents is conserved for fire fighting. In thisevent, all 22 valves functioned properly, con-serving 30,000 cubic meters of water, which,however, could not be delivered because ofapproximately 2,000 breaks in the under-ground piping. Kobe has approximately23,500 fire hydrants, typically flush-mounted(i.e., under a steel plate in the sidewalk orstreet) with one 150-millimeter-diameter hoseconnection. The city has provided under-ground storage of water for disaster firefighting in 968 cisterns, generally of 40,000-liter capacity, sufficient for about a 10-minutesupply of a pumper. All engines carry hardsuction, so that additional water can bedrafted from Osaka Bay or the several streamsrunning through Kobe.

KFD had minimal staffing on duty at thetime of the earthquake, possibly because theprevious day had been a holiday. Initial ac-tions included recalling off-duty personneland responding to fire calls. Approximately100 fires broke out within minutes, primarilyin densely built-up, low-rise areas of thecentral city, which comprise mixed residen-tial-commercial occupancies, predominantlyof wood construction. Within 1 to 2 hours,several large conflagrations had developed.There were a total of 142 fires reported inKobe on January 17, the majority being in thewards of Higashi Nada (24), Nada (24), Hyogo(37), Nagata (19), and Suma (18). Modes offire reporting were unclear as of this writing,and fire response was hampered by extremetraffic congestion, and collapsed houses,buildings, and rubble in the streets. Because

A typical street scene in aresidential and lightcommercial area of Kobe. Ahouse has collapsed, blockingthe very narrow street andpreventing access to the areaby the fire department.Thousands of narrow streetswere blocked like this. Inother areas, many of thecollapses were traditionalbuildings with ground-floorstores or shops and second-floor residential units.

of the numerous collapses, many areas wereinaccessible to vehicles.

Water for fire-fighting purposes wasavailable for 2 to 3 hours, including the use ofunderground cisterns. Subsequently, waterwas available only from tanker trucks. KFDattempted to supply water with a fireboatand relay system, but this was unsuccessfuldue to the relatively small hose used by KFD.An EQE engineer overflew the area at about5:00 P.M. on January 17 and was able to ob-serve all of the larger fires (about eight in all)

76 EQE

from an altitude of less than 300 meters. Nofire streams were observed, and all fires wereburning freely—several with flames 6 metersor more in height. No fire apparatus wereobserved in the vicinity of the large fires,although fire apparatus could be seen atother locations (their activities were unclearfrom the air). Some residents formed bucketbrigades (with sewer water) to try to controlthe flames.

Fire spread was via radiant heat andflame impingement, building to building inthe densely built-up areas. The wind wascalm, and fire advance was relatively slow.In a number of cases, fires were observed tohave stopped at relatively narrow fire breaks(e.g., 10 meters) or, in at least one case, at ahigh-rise apartment building, probably as aresult of active fire fighting. The final burnedarea in Kobe was estimated at 1 million squaremeters, with 50% of this in the Nagata Ward.

The Ashiya Fire Department reported 11fires on January 17; nine of them were before7:30 A.M. Distribution of the fires was alongan east-west line about 1 kilometer widecentered on National Route 2. The totalburned area for the 11 fires was about 4,400square meters.

This truck carrying kerosenecaught fire when theoverhead roadway collapsedonto it.

EQE 77

While it is still too early to estimate thetotal economic impact of the Kobe Earth-quake, it is already clear that the extent of theregional economic disruption exceeds theexperience of any modern urban area in anatural disaster. The direct and indirect busi-ness interruption losses will most likely out-weigh repair costs, particularly with respectto the consequences of the transportationand other lifeline failures. The impact on thenational economy will likely be minimal,although uncertainty and speculation causedthe Nikkei Index to drop as much as 5.6% ofits value in one day. The overall economicimpact and long-term effects of this disasterwill be influenced to a large extent by thespeed with which physical infrastructure canbe repaired and business activity resumed.

Repair CostsCurrent estimates of the repair costs in

this earthquake have been reported in therange of U.S.$95 billion to U.S.$147 billion,many times the damage inflicted by the 1994Northridge Earthquake. These figures do notinclude the loss to building contents such asequipment and inventory, which will also besubstantial. Repair cost estimates made bythe Hyogo Prefectural Government, the Na-tional Land Agency, and the Ministry ofTransportation three weeks to a month afterthe disaster are shown in the table on thenext page.

Some of the cost of repair and recon-struction will be financed through a vari-ety of government programs. The national

Residents walked the streetsof Kobe after the earthquakeexamining the damage.

ECONOMIC IMPACT

78 EQE

government is considering setting up anemergency budget of ¥900 billion (U.S.$10billion) to deal with the impact of the earth-quake, especially to repair roads, water andsewer systems, harbor facilities, and schools.It has announced preliminary plans to subsi-dize up to 90% of the cost of repairing publicfacilities. Earthquake victims are eligible forrelief grants, low-cost loans, and tax breaks.To cover the cost of reconstruction, officialsare planning to issue some ¥700 billion (U.S.$8

billion) in construction bonds and ¥600 bil-lion (U.S.$7 billion) in deficit-covering bondsin anticipation of tax revenue drops. Takinginto account earthquake-related reductionsin corporate and personal income taxes, theFinance Ministry is anticipating a drop of¥500 billion to ¥600 billion (U.S.$6 billion toU.S.$7 billion) in tax revenues for the fiscalyear. Officials are also considering raisingtaxes, and the consumption tax increasescheduled for 1997 may be adopted sooner.

People carrying supplies andproperty into and out ofKobe on foot—the onlyavailable means of transpor-tation for much of thepopulation for weeksfollowing the earthquake.

HyogoPrefecturalGovernment

NationalLandAgency

Ministry ofTransportation

_ _

6.8

_ _

_ _

5.7

_ _

REPAIR COST ESTIMATES (U.S.$ BILLIONS)

AGENCY BUILDINGS

65.7

71.3

_ _

ROADSRAIL-

ROADS

4.6

_ _

4.7

UTILITIES OTHERHARBORS

11.8

_ _

8.4

6.8

24.9

_ _

EQE 79

Business Interruption and RecoveryIn addition to the cost of repairing physi-

cal damage, the regional economy is alsobeing severely affected by temporary busi-ness interruption and the loss of import/export capabilities. After the earthquake, alleconomic activity in Kobe virtually halteddue to earthquake damage to buildings andcontents, loss of water and other utilities,difficulties in obtaining supplies, and em-ployee absenteeism. Hyogo Prefecture, whichincludes Kobe and the other most heavilyimpacted cities, produces about 4% of thenational output but accounts for 19% of thecountry’s production of leather goods, 10%of its rubber manufacturing, and 9% of thecountry’s steel. One-third of the prefecturaleconomy is engaged in manufacturing. Theregion shaken by the earthquake accounts foralmost one-fifth of the Japanese economy.

Among Kobe’s principal industries, steelproduction, shipbuilding, chemicals, andfood processing were heavily damaged bythe earthquake. Two of the three largest

employers in Kobe closed their plants forseveral days. The third sustained severestructural damage to its facilities, and afterone week was able to resume production atonly one of its two area plants at 40% ofnormal production levels. These three com-panies together employ more than 14,000people in Kobe.

The manufacturing sector had losses inboth large and small businesses. Many majorplant facilities were shut down for severaldays or more, including at least four largeelectronics plants, six steel or heavy industryplants, and three beverage production facili-ties. Small- and medium-sized manufactur-ing firms were reportedly heavily damagedin the earthquake, including structural, fire,or contents damage to more than 40% of localknitted goods manufacturers, more than 90%of synthetic leather shoe manufacturing fa-cilities, and many Japanese wine (sake) manu-facturers, which are heavily concentrated inthis region of Japan.

Lines of trucks bringingsupplies to the populace, aswell as equipment to startdemolition and removal ofdamaged buildings.

80 EQE

The transportation and utilities sectorswere also severely affected by this disaster.With systems restoration time frames typi-cally on the order of several months, revenuelosses during the interim will be significant.For example, Hanshin Electric Railway hasestimated that in addition to ¥79 billion(U.S.$895 million) in damage to its rail facili-ties, it will also lose ¥4.5 billion (U.S.$51million) in revenue. Osaka Gas, Japan’s sec-ond-largest gas utility, estimates that in addi-tion to ¥15 billion (U.S.$170 million) in piperepairs, it may lose ¥6 billion (U.S.$68 mil-lion) in revenues because of the disruption.

During the first week after the earth-quake, virtually no retail or service establish-ments were open for business. In addition tobuilding and inventory damage, as well asemployee absenteeism, these businesses were

critically affected by the disruption of lifelineservices. For the most part, those businessesthat were able to open did not depend heavilyupon water or natural gas, which were widelyunavailable; for example, some gas stations,electronic repair stores, shoe stores, and con-venience stores operating out of the store-front were in business.

Indirect business disruption was also feltby producers that had little damage to theirown facilities but had difficulties in obtainingsupplies and other input to production or inselling products. For example, one week afterthe earthquake, water for industrial consump-tion was still unavailable to 190 companies.The pearl industry in the region, which in-cludes more than half of Japan’s pearl proces-sors, may not return to normal operationsuntil June, because the earthquake has halted

Relief workers unloadingboxes of supplies at a refugecenter in the Nada Ward.

EQE 81

activity in the peak auction season, and grow-ers are unable to sell their cultured pearls.

Just-in-time production methods em-ployed by large manufacturing firms appearto be vulnerable to the widespread transpor-tation disruption. Four auto manufacturersand a motorcycle manufacturer reported pro-duction cutbacks and partial or temporaryshutdown of operations in plants outside theshaken region (as far away as Tokyo) becauseparts could not be obtained from or trans-ported through the affected area. Japan’slargest automobile manufacturer cut backproduction by 20,000 cars and closed plantsthroughout Japan for several shifts becauseof the difficulties in obtaining supplies.

Transportation disruption has also raisedtransportation costs for producers movinggoods in the region. For example, the closureof the Port of Kobe has forced many busi-nesses to divert cargo shipment to other fa-cilities around Japan or East Asia. However,with the cost of shipping a container fromKobe to Tokyo/Yokohama approaching thatof moving it across the Pacific to the UnitedStates, these alternatives will add substan-tially to producers’ transportation costs. Theimpact of Kobe’s port closing on regionalindustries will be greatest in manufacturing,because Japanese exports through Kobe tothe United States, for example, consistedprimarily of auto parts, tires and tubes, chemi-cals, engines, motors, and hardware. On theother hand, imports arriving in Kobe fromthe United States had included goods such asanimal feed, chemicals, paper, cotton, veg-etables, and meat.

The indirect effects of the earthquakewill extend beyond the areas of greatestphysical damage. The Port of Kobe servednot only as a port for goods going in and outof Kobe, but also as a transshipment hub fortransportation to and from the Far East.With its central location, modern shippingfacilities, and infrastructure tailored to themodern shipping industry, the Port of Kobewas ideally suited to the transfer of smallerlocal ships’ goods to and from large trans-ocean ships.

Immediately following the earthquake,other port facilities in the Far East, includingports in Kaohsiung, Singapore, and HongKong, temporarily took over this functionfrom the Port of Kobe. It is feared that thedevelopment of these other transshipmenthubs will result in the long-term decline inthe use of Kobe as a transshipment hub. Inaddition to Osaka, which has been overbur-dened by the increased traffic, Maizuru Portin Kyoto Prefecture, and ports in Yokohamaand Nagoya have been serving as alter-nate outlets for goods from the centralregion of Japan.

Indirect effects are also being felt in theregional real estate market: In Osaka, for

Completely destroyed pier atthe Port of Kobe.

82 EQE

Japanese Versus California Residential Earthquake InsuranceWhile comparisons are very difficult, it is of interest to contrast the situation of two typical

houses, one in Tokyo and the other in Los Angeles (this comparison is not based on particularstructures, but rather on the general knowledge of several experts):

Several observations emerge from this comparison: U.S. homeowners receive more fortheir premium, not even considering the fact that the U.S. homeowner is already insured forfire following earthquake via the basic fire policy; and the bulk of Japanese residential valuesare in the land rather than in structures. Thus, even in the case of a total loss, the Japanesehomeowner, while sustaining a monetary loss similar to the U.S. homeowner’s, still maintainssubstantial equity (and borrowing power) via the land value. This is further reinforced by thefact that Japanese homeowners are typically not as heavily mortgaged (expressed in percent-age terms) as are U.S. homeowners. In addition, another channel of relief for some homeownersin Kobe has emerged, reflecting the underlying nature of Japanese society. Recognizing theplight of their employees, some companies are offering low-interest mortgages, or are evenrebuilding employees’ houses in a few cases, while smaller businesses impacted by theearthquake are generally supported by their “parent” companies in the keiretsu (or Japaneseconglomerate system).

APPROXIMATE COMPARISON, TOKYO AND LOS ANGELESRESIDENTIAL INSURANCE COSTS1

TOKYO LOS ANGELES

Suburban neighborhood Tachikawa Northridge

Per capita income $20,000 $22,000

Commute minutes to central 90 60business district

Typical house lot, square meters 80 500

Typical house floor area, square meters 140 180

Typical sales price (1995, estimated) $750,000 $250,000

Typical replacement value (structure) $250,000 $180,000

Ratio, structure replacement to total price 0.33 0.72

Total sum insured (i.e., Cov. A $100,000 $180,000 or structure only)

Earthquake premium (annual) $500 $3602

Notes: 1. All sums in 1995 U.S.$2. Northridge premium is prior to 1994 Northridge Earthquake

EQE 83

example, the flood of households and busi-nesses seeking temporary relocation hasdriven up the price of apartment and com-mercial office space by about 10% in manyareas, despite the fact that there had been adepressed market for commercial space be-fore the earthquake. Consumer spending, onthe other hand, has reportedly dropped by20% to 30% in Osaka after the disaster, largelydue to the feeling among consumers that“self-restraint” in spending was appropriatewhen people in nearby Kobe were still suffer-ing. Kyoto, which had very little damage,saw a sudden drop in tourism.

Recovery of local businesses will takemonths or even years, with some industriesrecovering more quickly than others. A gov-ernment survey found that six days after theearthquake, two-thirds of the gas stations inthe impacted region were back in operation.About a week after the earthquake, bankswere widely in operation, with some chainsreporting close to 90% of their branch officesopen in the heavily impacted area. One weekafter the earthquake, many other businessesreported partial resumption of business op-erations. For example, in the case of Japan’slargest supermarket chain, 25 of its 49 storesin Hyogo Prefecture were closed by the earth-quake; after two weeks, 12 of these hadreopened. At least one major departmentstore, however, announced that it will per-manently close its downtown Sannomiyastore because of structural damage, whileanother expected to take at least a month toopen its downtown store.

While the recovery will take some time,as reconstruction gets underway, the rebuild-ing efforts will provide a boost to the re-gional economy. Already, reports from onedistrict in Osaka indicate that the number ofconstruction day jobs has doubled from be-fore the earthquake because of the need toclear rubble. The Hyogo Prefectural Govern-ment is currently considering designatingKobe as a duty-free zone to encourage re-building and investment in the region.

Insurance AspectsWhile the estimates of the Kobe Earth-

quake’s U.S.$95 billion to U.S.$147 billion inproperty damage are unprecedented, theimpact of this event on the insurance indus-try will be significantly less than a number ofother recent events, including TyphoonMireille in Japan (which had total insuranceclaims of U.S.$5.7 billion), Hurricane An-drew (U.S.$16 billion), and the NorthridgeEarthquake (U.S.$12 billion). Total insurancepayments arising from the Kobe Earthquakeare presently estimated at about U.S.$6 bil-lion, although this could rise because of twofactors: (1) additional newly discovered dam-age, as buildings are inspected (similar towhat occurred after the Northridge event),and (2) as-yet unreported claims against off-shore insurers, either against multinationalcorporate policies, or for time element claimsresulting from the need for shipping to bediverted around the Port of Kobe.

When compared with previous events inJapan or the United States, there appears tobe a major disparity between property lossand the portion borne by the insurance in-dustry in Japan. This difference exists be-cause the government and the Japaneseinsurance industry, which consists of onlyabout a dozen very large insurance compa-nies, recognized the difficulties in insuringfor earthquake in Japan. That is, they real-ized that earthquakes in Japan may be“uninsurable,” in the sense that Japan has thepotential for large earthquakes almost any-where on the archipelago, and that for themost part Japan also comprises a relativelysmall number of major cities. In effect, Japanin its entirety represents a unique sort ofadverse selection (particularly Tokyo), forwhich the government and the insuranceindustry have been unable to identify anadequate insurance solution.

The current scheme of residential earth-quake insurance was introduced in 1966 fol-lowing the 1964 Niigata Earthquake. Thisscheme established Japan Earthquake

84 EQE

Reinsurance Company Ltd., which in turn isreinsured by the central government. Basi-cally, the scheme offers a limited earthquakeendorsement to the basic fire policy (notethat, in contrast to the United States, firefollowing earthquake in Japan is not coveredunder the basic fire policy but rather requiresthe earthquake endorsement). The indem-nity under this policy is typically limited toabout 30% to 50% of the structure’s replace-ment value, capped at about U.S.$100,000.

In the claims process, the structure isdetermined to fall into one of three catego-ries: “total loss,” “half loss,” or “significantlyless than half loss.” If the structure is deter-mined to be a total loss, then payment is forthe total sum insured. If it is a “half loss,” theindemnity is prorated at 50% of the total suminsured (i.e., 15% to 25%), with the contentsfurther severely limited unless they are to-tally destroyed. Losses that fall under “sig-nificantly less than half loss” are notreimbursed. A minor allowance is made forincidental expenses. Earthquake premiumsfor this coverage are based on pure premiumplus loading, with no profit (for the earth-quake cover). The pure premium is based onan estimation of annualized loss determinedfrom a 500-year record of earthquakes (i.e.,since 1494). For example, typical premiumsin the Tokyo area are about 0.5% of the TotalSum Insured (TSI), versus 0.2% in California.

Under this scheme, homeowners maypurchase the earthquake endorsement.Nationally, about 7% of Japanese homeown-ers purchase this endorsement (versus about

25% in California), although this percentagevaries significantly (about 3% in Kobe and16% in Tokyo, versus perhaps 40% in the SanFrancisco and Los Angeles areas).

Total liability of the residential insur-ance scheme is limited to about U.S.$18 bil-lion, with about U.S.$1 billion borne by theJapan Earthquake Reinsurance Company,about U.S.$2 billion by the direct writers, andthe remainder (approximately U.S.$15 bil-lion, or 85%) by the government.

Commercial lines insurance risk is notreinsured by the government, but capacity islimited by governmental intervention. A par-ticularly large exposure that absorbs a sig-nificant portion of this capacity is thenumerous oil refineries in and around TokyoBay. It is of interest to note that petrochemi-cal facilities, in general, and tanks in particu-lar, were not heavily damaged in the KobeEarthquake, which is unusual. Because of thevery limited commercial capacity in Japan,some risk is placed offshore, so thatnonadmitted (overseas) insurers will bearmore loss, due to both offshore primary in-surance and offshore reinsurance.

In summary, while the Kobe Earthquakeis perhaps the world’s most damaging singleevent (in property terms), the insurance im-pact is relatively small. This small impactresults from the specifics of the Japaneseinsurance industry, which reflect the uniqueexposure of Japan to earthquakes. This situ-ation may change very rapidly, however,since the Japanese insurance industry will belargely deregulated within several years.Demand for earthquake insurance followingthe Kobe disaster, combined with increasedcompetition among Japanese insurers andthe entrance of foreign insurers interested inthe commercial market, may result in a ma-jor shift offshore of the earthquake risk (asdiscussed above, presently borne largely byproperty owners). This could rapidly createa major increase in the global insuranceindustry’s exposure to a catastrophic earth-quake in Tokyo.

The temporary office of aKobe newspaper on January18, after the regular officewas closed by damage.

EQE 85

SOCIETAL IMPACT

The most significant societal impact ofthe earthquake was the tremendous loss ofhuman life. In addition, for more than 300,000survivors in the heavily impacted cities ofKobe, Ashiya, and Nishinomiya who weredisplaced from their homes, there were thehardships of finding shelter; securing foodand water; locating friends and family mem-bers; and acquiring warm clothing for thecold, damp winter weather.

Although some of the displaced peoplewere taken in by relatives and friends, andothers possessed the means to relocate tohotels, those requiring emergency shelterreached a peak of 235,443 on the evening ofJanuary 17. Many camped in public parks orassembled makeshift shelters from materialssalvaged from the wreckage of their homes.The 1,100 shelters included community cen-ters, schools, and other available and un-damaged public buildings. Facilities weretoo few to avoid severe crowding in some

shelters, however, causing sanitation prob-lems and increased risk of communicabledisease. Indeed, two weeks after the earth-quake, reports of influenza and pneumoniawere common.

Food, water for drinking and sanitation,blankets, and warm clothing were in shortsupply for at least the first few days after theearthquake, and many people from thehardest-hit wards made the long walk to theNishinomiya Railway Station, journeyed toOsaka for necessities, then returned via railwith whatever they were able to transport byhand.

By Friday, January 20, both official andvolunteer efforts to supply the basic needs ofthe impacted area were becoming increas-ingly evident. Corporations and other non-governmental organizations donated goods,and transportation was provided by bothbusiness and government vehicles. In some

Earthquake victims campedout at Kobe City Hall.

86 EQE

cases, normal production schedules and pro-cesses were modified to assist in the reliefeffort. Kirin Breweries, for example, filledliter-sized beer bottles with drinking waterand shipped thousands of cases into theKobe area.

Amid the overwhelming need for safeshelter, some residents chose to remain indamaged residential buildings despite un-certainty regarding structural integrity. Therewas little evidence during the first week thataccess and egress of even the most severely

damaged homes and apartment buildingswere being monitored, and cordoned areaswere few and unenforced.

Although temporary housing was be-ing constructed within two weeks after theearthquake, and rent-free rooms were be-ing offered by apartment owners, the de-mand for longer-term housing still exceededavailability by a factor of 10. Those displacedby the earthquake are likely to compete foravailable housing with construction work-ers, technicians, and engineers convergingon the area to begin reconstruction.

Disproportionately, it was the poor andthe elderly who lost their homes, jobs, andlives—high rises were scarcely affected, whiletwo-story homes were at greatest risk.

Post-earthquake reliefscenes.

EQE 87

The Kobe Earthquake dramatically illus-trates the damage that can be expected tomodern industrialized society from earth-quakes. Most of what happened could havebeen predicted, and much of the damagewas preventable. Hopefully, the disasterwill spur building owners to continue—and to increase where needed—their ef-forts to improve the earthquake resistanceof their properties.

Estimates of the potential effects of largeearthquakes in the United States, Japan, andother countries have been developed by en-gineers and scientists over the past 20 years.The Kobe Earthquake now provides answersto many “what if” questions that those sce-nario studies raised. An event that lastedabout 20 seconds caused 5,500 deaths and aneconomic loss that is greater than the grossnational product of many countries. Much ofthe infrastructure and building stock of amodern city, which many considered to beprepared to withstand a strong earthquake,was destroyed.

While the true dimensions of what hap-pened in Kobe and the surrounding regionwill not be fully understood for some time,the nature and magnitude of the known lossespermit some conclusions to be drawn. Whilesome of our conclusions are applicable pri-marily to Japan, California, and other parts ofthe United States, most are applicable toseismically active regions worldwide.

Are We Surprised by WhatHappened?

Little of the damage observed in Koberepresents new lessons. Most of the causes ofthe building damage have been observedrepeatedly in past earthquakes—particularlysince the 1964 Anchorage, Alaska, earthquake.For example, the design details that causedthe collapse of elevated structures of theHanshin Expressway and the Shinkansen(Bullet Train) were similar to those that causedfreeway structures to collapse or sustain dam-age in Oakland and San Francisco in 1989,and in Santa Monica and Los Angeles in 1994.The types of damage observed at the Port of

Kobe were also observed (to a lesser extentbut because of the same engineering deficien-cies) in numerous past earthquakes—severalof which were in Japan, starting with theNiigata Earthquake of 1964 and the Tokachi-oki Earthquake of 1968.

A few important new lessons have beenlearned, however. Some types of new build-ings and other structures performed verywell, with minimal damage. These includedboth concrete and steel structures. Unfortu-nately, other innovative building designsfailed dramatically, even though they wereexpected by some engineers to perform well.This inadequate performance demonstratesthe need to incorporate adequate factors ofsafety when implementing innovative tech-nology for which actual earthquake perfor-mance data have not been obtained.Although mathematical methods and small-scale tests are excellent tools for developingnew technologies, additional conservatism isnecessary until the new technology has beenfield proven.

Can It Happen Elsewhere?Japan, in general, is earthquake country,

and the Kansai region has a long history ofmajor earthquakes, although the largest eventin this century within the immediate vicinityof Kobe was magnitude 6.1 (in 1916). Perhapsbecause of the recent lack of seismicity in theimmediate region, comments such as “wedidn’t think such a large earthquake couldhappen here” were frequently heard from thegeneral public, as well as from city officialsand emergency responders.

We constantly hear similar statementsregarding the seismicity of areas in Califor-nia, for example, which have not had strongearthquakes in this century. San Diego andSacramento are two such areas. Like Japan,all of California and the entire West Coast ofNorth America are earthquake regions. It istrue, to the best of our knowledge, that theprobability of a major earthquake near Port-land, Oregon, or Vancouver, British Colum-bia, is lower than near Los Angeles. It was alsotrue for Kobe, when compared to Tokyo, but

CONCLUSIONS

88 EQE

the earthquake happened in Kobe. That is thenature of estimates based on probabilities.

Lack of recent seismicity may only be asign of accumulating seismic stresses. Areasof the United States and Canada, with clearsigns of past large earthquakes—such as thePacific Northwest; the Wasatch Front in Utah;the New Madrid region in the U.S. Midwest;Charleston, South Carolina; and similar

Similarities Between Kobe and the San FranciscoBay Area

The similarities between the geological, seismological,and land-use characteristics of the Kobe area and the East Bayof the San Francisco Bay Area are both striking and importantin their implications for seismic safety along California’sHayward Fault. Although there are significant differencesbetween the situation in Kobe and in the East Bay, the lessonslearned from the disaster in Kobe should be carefully consid-ered for earthquake disaster planning and damage-mitigationplans for the San Francisco Bay Area.

In both areas, cities lie nestled along a narrow strip of landbetween mountains and a bay. These strips of land consistmostly of soft to very soft alluvial deposits. The seismicsimilarities are striking—major faults run directly under themost densely populated section of both areas, along the toe ofthe mountain ranges. The earthquake in Kobe was a momentmagnitude (Mw) 6.9 event. The Hayward Fault is thoughtcapable of releasing enough energy to generate an Mw7.5 event.

The land use in Kobe and the East Bay is also quite similar.Both areas have high densities of residential, commercial, andindustrial buildings that are a wide range of structural typesand ages. For both areas, the faults run near important trans-portation and lifeline corridors, including elevated highways,buried subways, long-span bridges, and important railways.The port facilities of Oakland, the largest city in the East Bay,are built on soft soils that may be susceptible to damagesimilar to that which occurred in Kobe.

Although the residential stock in California may not be asvulnerable to earthquake damage as the residential buildingsthat were destroyed in Kobe, there is a much higher concen-tration of vulnerable unreinforced masonry structures in theSan Francisco Bay Area than there was in Kobe. Also, just asin Kobe, most of the industrial buildings in the East Bay arebuilt near the bay, where amplification of ground motionthrough soft soil deposits can be expected. As demonstratedin Kobe, and in San Francisco in the 1989 Loma Prieta Earth-quake, these areas are susceptible to large-scale soil failures.

regions—need to heed the lesson of Kobe.Stronger efforts are required in the areas ofpublic education, and in strengthening pub-lic and private structures. Research to betteridentify hazards in the above regions shouldbe increased.

Ground MotionFor Japan, near-record ground motions

were recorded in and near Kobe. The strengthof the shaking, however, was similar to thatobserved in several recent events in and out-side of Japan. For example, the strength of therecorded shaking in the San Fernando Valleyof Los Angeles in 1994 was about the same asin Kobe. The shaking is typical of what can beexpected in a moderate- to large-magnitudeevent. What was different in Kobe, and themain reason for the relatively severe dam-age, was that the strong shaking occurred inthe middle of a metropolitan area, under-neath tens of thousands of structures.

Current building codes are based on theobserved strength of shaking—which, untilrecently (mid-1980s), was believed by engi-neers to be on the order of 50% of that ob-served in numerous recent earthquakes. Thisbelief was based on a limited data set ob-tained with relatively few strong-motion in-struments. We now have many moreinstruments deployed, and the resultingmany more near-field records being obtainedare now telling us that near-field shaking ismuch stronger (2 times or more) than previ-ously believed. We need to review buildingcode design force levels, given these new data.

Building PerformanceThe large commercial and industrial

buildings in the Kobe area, particularly thosebuilt with steel or concrete framing, are simi-lar to buildings of the same vintage in Cali-fornia. Typically, those buildings were muchstronger than buildings of similar vintage inother areas, such as the Pacific Northwestand almost all other countries. The Japanesebuilding code had a major revision forconcrete-frame buildings and a more limitedrevision for steel-frame buildings in 1981.The Uniform Building Code, as used in Califor-

EQE 89

Near-field Effects of EarthquakesBuildings are much like the strings on a violin or guitar. Each building has a unique set of natural

frequencies at which it will vibrate when disturbed by a transient load such as an earthquake or windstorm.This is particularly important in earthquake engineering, since the energy delivered by an earthquake to abuilding is strongly related to the natural frequency of the building. To characterize these effects, engineersuse tools called acceleration response spectra. These are plots of the response of structures with different naturalperiods to specific earthquake ground motions.

Since the mid-1970s, building codes have incorporated standard response spectra as a basis for design.In effect, these response spectra set the minimum strength for which a building must be designed. In severalrecent earthquakes, including the 1992 Landers and Big Bear, 1994 Northridge, and 1995 Kobe events,seismographic instruments located within about 5 kilometers of the fault rupture have recorded groundmotions with unusually large response spectra. Originally believed to be “freak” recordings, seismologistshave now demonstrated that such large motions are to be expected within a few kilometers of a major event.The figure below compares the spectra recorded by instruments in the near field of the Kobe and Northridgeearthquakes with one of several standard spectra contained in the Uniform Building Code and used as the basisfor building design. These spectra clearly show that when a building is very close to the source of a large-magnitude earthquake, the forces produced in the building can be much larger than anticipated by thebuilding code. The message is clear—buildings designed to the minimum provisions of the building code maynot have adequate strength if they are very close to the source of a major earthquake. Seismologists andengineers are currently studying this problem to determine whether building codes should be modified toaccount for these “near-field” effects.

Acceleration Response Spectra, 5% Damped

Period (sec)

Acce

lera

tion

(g’s

)

0.00

0.50

1.00

1.50

2.00

2.50

3.00

0.10 1.00 10.00

Kobe E-WKobe N-SSylmar E-WSylmar N-S0.4g UBC, Soil Type 3

90 EQE

nia, had major changes in 1973, 1975, andseveral times since then, for various types ofbuildings. The changes in both Californiaand Japan were first triggered by the lessonsof the 1971 San Fernando Earthquake. Ineffect, the current Japanese code requiresthat buildings in Japan be designed for some-what higher force levels than does the Uni-form Building Code. These forces are muchhigher than those required in most otherearthquake regions of the world.

Reinforced Concrete Buildings

Typically, pre-1981 concrete-framebuildings performed very poorly in Kobe,with many collapses. Post-1981 buildings per-formed much better—some were extensivelydamaged, but most had light damage. Thebuildings that fared best, and those withoutsignificant damage, had extensive concreteshear walls.

Many concrete buildings, similar to thosethat collapsed, have not been strengthened inCalifornia and other earthquake-prone re-gions and are expected to collapse in futureseismic events. There is essentially nothingnew in the lessons from Kobe for such build-ings—similar damage and collapses havebeen studied in many recent earthquakes.The most dangerous buildings are the oldernonductile concrete buildings and moremodern tilt-up and precast concrete build-ings in California and elsewhere, and evennew concrete-frame buildings without mas-sive shear walls elsewhere in the UnitedStates and in Canada. This includes manybuildings that are more than 10 stories high.It is obvious, from the performance of thesebuildings, that a major strengthening pro-gram also needs to be initiated in Japan.

Steel Buildings

As in other earthquakes, large commer-cial and industrial steel-frame buildings per-formed better than any other type. However,major damage and a few collapses wereobserved by the investigators.

Pre-1981 steel buildings had most of theserious known damage. Certain innovativetypes of steel buildings, including high rises,

had very serious damage, and collapses couldhave occurred if the duration of the earth-quake had been a few seconds longer. Sev-eral types of different innovative buildingsystems are popular in Japan, Seattle (Wash-ington), and other parts of the United States,for example. The lesson here is that whileinnovation is necessary, the engineeringanalysis and designs are often inadequate;therefore, detailed third-party reviews shouldbe conducted and detailed full-scale testingis absolutely necessary. Often, only earth-quakes can provide such tests, so certaininnovations, especially when developed toreduce costs, are potentially dangerous. Fur-ther, the typical codes in Japan, the UnitedStates, and elsewhere are prescriptive and donot deal well with innovation or with ideasand practices that fall outside of ordinarydesign practices.

One of the primary lessons from the 1994Northridge Earthquake was that most of thedamage to steel buildings cannot be uncov-ered immediately because it is often hiddenunder building finishes that are undamaged.This is proving to be the case in Kobe, andmore time is needed to evaluate the overallperformance of modern steel buildings inJapan. However, it is obvious from these twoearthquakes that certain types of steel build-ings, including high rises, may be collapsehazards in strong earthquakes with longerduration of shaking.

Building Performance Criteria

Building owners usually do not under-stand that the earthquake provisions of theUniform Building Code do not have reason-able performance criteria for larger and stron-ger earthquakes. The current regulations,including those for all of California, are in-tended to protect a new building from col-lapse in an earthquake like the one that struckKobe. A building is expected to be severelydamaged—in fact, it may need to be torndown—but it should not collapse. This isunderstood by engineers, but they are usu-ally not in a position during the design pro-cess to communicate this information to theowner. Further, some do not attempt to warnan owner that if “reasonable” or light dam-

EQE 91

age is the owner’s expectation, higher designstandards should and can easily be applied,at some increase in cost. In California, higherperformance criteria are actually mandatedfor certain types of structures—schools, hos-pitals, police and emergency response build-ings, and certain power facilities. An ownermay specify that these standards be used forother types of structures. Otherwise, a high-value, heavily occupied commercial build-ing is, in effect, designed to the sameperformance level as a low-value farm build-ing, if the basic code is merely followed.

In Kobe, many commercial buildings hadspectacular and extensive damage to theirfinishes and certain structural elements, los-ing more than 50% of their value, yet theywere considered by structural engineers tohave performed well because they did nothave fatal structural damage. Such buildingsmet the requirements of the code, but did notmeet the expectations of their owners. Thestructural engineering profession needs tocommunicate better the limitations of theirdesigns to the owners. In this respect, theJapanese engineers in Kobe appear to havebeen somewhat successful. It was apparentthat many new buildings were designed tohigher standards than required by code, andthey performed much better.

Buildings on Soft Soil and Fill

Much of the newer construction in Kobe,particularly larger buildings, is built on verysoft, recent alluvial soil and on recently con-structed near-shore islands. Most of the seri-ous damage to larger commercial andindustrial buildings and infrastructure oc-curred in areas of soft soils and reclaimedland. The worst industrial damage occurredat or near the waterfront because of severeground failures—liquefaction, lateral spread-ing, and settlement.

Such lessons have been observed in nu-merous other earthquakes, and no great sur-prises occurred in this event. Because of asevere shortage of constructible land, muchof modern Japan, including Tokyo and mostother large cities, is built on the worst groundpossible for earthquakes. For Japan the

lessons are critical. The engineering profes-sion has tried hard to develop methods forstrengthening existing and filled soft or weaksoils to resist failures during earthquakes.Much has been built based on innovativetechniques and theoretical analyses that hadnever been fully and adequately tested inreal, very strong earthquakes. The results aredecidedly mixed, but the failures are verycostly. Most new buildings on piles appearedto perform well, often with no significantstructural or even architectural damage.Older buildings not on piles often performedvery poorly; many were tilted severely be-cause of settlement. The infrastructure,particularly underground piping, also hadsevere damage because of settlement andlateral spreading. This caused extensive busi-ness interruption because of a lack of waterand gas.

Industry in the port areas was severelyaffected because of lateral ground spreadingand the failures of retaining walls and fills.Most retaining walls along the port failed,and the related ground settlement pulledbuildings and other structures apart. Thistype of damage should be expected else-where in Japan to facilities near the water-front, because the construction in Kobe istypical of Japan. The same observations, tovarious degrees, apply worldwide.

PortsThe Port of Kobe, much of which was

new, was devastated by widespread andsevere liquefaction and/or permanentground deformation, which destroyed morethan 90% of the port’s 187 berths and dam-aged or destroyed most large cranes. Dam-age is estimated at more than U.S.$11 billion.Shipping will be disrupted for many months,with major losses to the local economy and astrain on alternative transportation modes.

Ports all along the western United Statesare particularly susceptible to effects likethose seen in the Kobe Earthquake. The portsof Los Angeles and Long Beach, serving thesecond-largest urban area of the United States,are on or near the Newport-Inglewood andPalos Verdes faults, while the Port of Oak-

92 EQE

land (significantly damaged in the 1989 LomaPrieta Earthquake) is only a few kilometersfrom the Hayward Fault. New data on faultsin the Seattle, Washington, and Vancouver,British Columbia, port areas are only nowemerging. Significant sections of these portswere built before modern geotechnical prac-tice permitted mitigation of liquefaction risk.The real possibility exists that a large portionof the West Coast’s shipping capacity couldbe crippled by a major earthquake.

TransportationBridges and Expressways

The Hanshin Expressway, built in the1960s and primarily of nonductile, reinforcedconcrete construction, was virtually de-stroyed over more than 20 kilometers. Thealmost completed Wangan Expressway,which is largely composed of steel super-structures, had many spans lose their bearingconnections, damaging the superstructuresand closing the route indefinitely. A numberof major bridges of very modern design wereseverely damaged, resulting in some comingclose to collapse.

There are no significant new lessonsfrom the collapse and damage of the olderunretrofitted bridges and elevated structures.Some of the upgrade details observed inolder retrofitted structures, such as steel col-umn jacketing, are now widely used in Cali-fornia for strengthening. The apparent goodperformance of these details in Kobe is im-portant to ongoing U.S. programs and needsto be studied in detail.

The damage to the new Wangan Ex-pressway is disturbing and needs to bestudied in detail. Possibly cumulative dis-placements greater than 2 meters were ob-served on the deck of the expressway. Thesedisplacements and the resulting forces prob-ably exceed existing criteria for new struc-tures in the United States (and othercountries). The failures of many bearings,and other details, were dramatic and alsorequire detailed investigation.

Further, the performance of the largenew bridges, including cable-tied arch, bracedarch, and cable-stayed bridges, should bestudied extensively because this is by far thestrongest earthquake to affect such bridges.The lessons are invaluable, for both newdesigns and the retrofit of existing largebridges worldwide. Much of the damageobserved in these structures has not beenseen before. This earthquake also dramati-cally illustrates the need to speed up theongoing strengthening program of bridgessuch as the Golden Gate and San Francisco-Oakland Bay bridges, before a strong eventcauses even more costly damage and largereconomic disruptions than those observed inthe Kobe region.

Rail Systems

The narrow Kobe transportation corri-dor is almost the only rail link between cen-tral and southern Japan. It was entirelysevered by the earthquake. A number ofstations, elevated rail structures, and bridgesfailed. Several kilometers of elevated struc-tures, including the main north-to-south Bul-let Train line, were severely damaged. Thestructural and foundation details that causedthe damage have been observed in numer-ous prior earthquakes, and the damagewas predictable.

While some differences do exist, theserail structures are similar in many respects toU.S. construction. Japanese commuters relymuch more on rail lines than do U.S. com-muters, but the United States is increasinglyemphasizing mass transit. The San FranciscoBay Area relies heavily on the Bay AreaRapid Transit (BART) rail line, and the LosAngeles region is in the midst of a majorcommuter rail construction program. Thedamage to elevated structures and the largedisplacements observed in soft soils, as dis-cussed above, suggest that a review of cur-rent design standards, as well as areevaluation of existing public rail sys-tems such as San Francisco’s BART sys-tem, should be conducted.

EQE 93

Other InfrastructureGas

The gas system in Kobe had major dam-age, generally caused by ground or buildingfailure, which contributed significantly tothe fire problem. Additionally, loss of thesystem for several months has been a majorhardship on the population and has caused along period of business interruption for manycompanies. The earthquake once again pointsout that in high-seismic-intensity areas, gassystem damage can contribute significantlyto the overall level of damage and long-termsocioeconomic impact. Further efforts arerequired by suppliers to review gas systemsin seismic areas for potential damage and toinvestigate high-technology methods forautomatic system control at the meter andupstream as part of their emergency pre-paredness planning.

Electric Power and Telecommunications

The electrical and telecommunicationssystems in Kobe and surrounding areas per-formed as expected based on experience fromprevious earthquakes. Facilities near theepicenter were damaged but the systems’resiliency prevented widespread service in-terruption. Long term power outages werelimited to the most heavily damaged areas.

Electrical generation facilities (powerplants) had limited damage, with the excep-tion of failures related to displacement ofsuspended boilers and soil failures. The elec-trical transmission system performed well,with most of the major transmission linesskirting the heavily damaged region of Kobe.The results may have been substantially dif-ferent had the epicenter been located closerto the 500-kV transmission system.

Although the power systems performedwell operationally during the earthquake,there were substantial financial losses. Foun-dation repairs or replacement, and boilerrepairs at generation facilities will take manymonths. Expensive, extra high-voltage sub-station equipment must be replaced, and thedistribution network must be essentially re-built within heavily damaged areas of Kobe.

The transmission and distribution compo-nents have been shown in past earthquakesto be one of the more earthquake-vulnerableportions of the electrical power industry, afinding further emphasized in the aftermathof the Kobe Earthquake.

Efforts are required to review electricand telecommunications systems in seismicareas for potential damage and emergencypreparedness, and to investigate methodsfor reducing damage.

Water

Generally, ground or building failure wasthe cause of the severe damage to Kobe’swater system. The resulting lack of watercontributed significantly to the fire problemand will be a major hardship on the popula-tion for several months.

Earthquake improvements had been de-veloped for the system, including the use ofseismic shutoff valves at reservoirs. How-ever, three major deficiencies were observed.

First, while seismic shutoffs existed atthe reservoirs, none existed within the distri-bution system, so that even a few breakscould severely lower pressure and impair thesystem. During the earthquake Kobe’s sys-tem sustained perhaps 2,000 breaks. The sys-tem needed a series of seismic shutoff valvesat key locations, placed so as to isolate areasof likely failure and thus maintain a “back-bone” system. These valves needed to beremotely operable.

Second, cisterns for backup fire watersupply were plentiful in Kobe—nearly a thou-sand; however, these typically provided onlya 10-minute fire-fighting supply. In the UnitedStates, comparable cisterns provide a 1-hoursupply, making them much more useful.

Third, the emergency response capabili-ties of the water department were very lim-ited, and the fire department’s 65-millimeterhose proved insufficient for relaying adequateamounts of water. Kobe needed a large-diameter hose, portable water supplysystem, such as that employed by the San

94 EQE

Francisco Fire Department in the 1989 earth-quake. Only a few cities in the United States,and none in Japan, have such a system.

Fire Following EarthquakeMore than 150 fires occurred in Kobe and

surrounding areas in the hours after the earth-quake. These resulted in several large fires,and fire fighters were for the most part un-able to combat them because of streets beingblocked by collapsed buildings and buildingdebris, traffic congestion, and severe watersystem damage. Fortuitous calm wind con-ditions prevented conflagrations.

The United States and Japan have bothsustained the largest peacetime urban con-flagrations in this century’s history—becauseof earthquakes. Fire following earthquake isa potential major agent of damage, of possi-bly holocaust proportions, for both the UnitedStates and Japan. This is verified by majornon-earthquake conflagrations in Southernand Northern California in recent years.Should a major earthquake occur in the UnitedStates or Japan under unfavorable meteoro-logical conditions, loss of major parts of a cityis quite likely.

Further efforts are required to increaseefforts to analyze post-earthquake conflagra-tion risk; increase awareness for the fire ser-vice and the public; and begin research anddevelopment for improved post-earthquakefire-fighting response, such as increased train-ing of citizen fire brigades, alternative watersupplies, and “smart” control of gas andwater systems.

Preparedness and ResponseWhile data are not yet complete, prelimi-

nary observations suggest that preparednessand emergency response efforts in Kobe wereless than satisfactory. The immediate urbansearch and rescue effort was inadequate forthe thousands of buildings destroyed in thisevent. The problem would have been furthercompounded had the earthquake occurredduring the day, when thousands more peoplewould have been trapped in major derail-ments and office building collapses.

Preparedness and emergency responseare often the most affordable, if not the onlypossible, mitigation techniques available tomany regions. U.S. preparedness and re-sponse have often been satisfactory, but lapseshave occurred (e.g., the immediate responsefollowing Hurricane Andrew). No matterwhat structural retrofitting may precede anevent, it can never entirely mitigate the prob-lem, so that a large, prepared emergencyresponse capability will always be required.

Efforts are needed to (1) continue andincrease support for emergency prepared-ness and response, at all levels, public andprivate; (2) encourage development of inno-vative techniques for improved response suchas automated, rapid post-event damage as-sessment and decision-making using geo-graphic information system-based tools;and (3) investigate enhanced responsethrough development of citizen cadres fordisaster assistance.

SummaryThe Kobe Earthquake is a terribly strik-

ing example of what earthquakes can do to amodern industrialized society. The loss ofnearly 5,500 lives and the hardships of hun-dreds of thousands of Kobe’s residents aretragic. Similar or larger earthquakes are go-ing to occur in Tokyo, Los Angeles, SanFrancisco, Wellington (New Zealand), andother major cities. We have been “lucky” inthe 1989 San Francisco, 1994 Los Angeles,and even the Kobe earthquakes when weconsider conditions of time-of-day, wind,and other factors that influence mortality,conflagrations, and the other earthquake con-sequences that we seek to reduce. The cata-strophic loss of more than 140,000 lives inTokyo in 1923 is an example of what can occur.

We have made the point that there arerelatively few new lessons to be learned fromthe Kobe Earthquake from an engineeringviewpoint. The real lesson is that we mustmotivate our societies to act—to replace orstrengthen deficient structures and systems,and improve our planning and preparedness.