takaji kokusho-earthquake geotechnical case histories for performance-based design-crc (2009)

8/16/2019 Takaji Kokusho-Earthquake Geotechnical Case Histories for Performance-based Design-CRC (2009)

http://slidepdf.com/reader/full/takaji-kokusho-earthquake-geotechnical-case-histories-for-performance-based 1/449

EARTHQUAKE GEOTECHNICAL CASE HISTORIES

FOR PERFORMANCE-BASED DESIGN

© 2009 Taylor & Francis Group, London, UK



Earthquake Geotechnical CaseHistories for Performance-Based

Design

Editor

Takaji Kokusho Department of Civil & Environmental Engineering,Chuo University, Tokyo, Japan




Cover photo: Huge landslide at Aratozawa during 2008 lwate-Miyagi Nairiku earthquake: M=7.2Courtesy of OYO CORPORATION, Japan.

Taylor & Francis is an imprint of the Taylor & Francis Group, an informa business


Typeset by Charon Tec Ltd (A Macmillan Company), Chennai, IndiaPrinted and bound in Great Britain by TJ International Ltd, Padstow, Cornwall

All rights reserved. No part of this publication or the information contained hereinmay be reproduced, stored in a retrieval system, or transmitted in any form or by anymeans, electronic, mechanical, by photocopying, recording or otherwise, withoutwritten prior permission from the publisher.

Although all care is taken to ensure integrity and the quality of this publication and the information herein, no responsibility is assumed by the publishers nor the author

for any damage to the property or persons as a result of operation or use of this publication and/or the information contained herein.

Published by: CRC Press/BalkemaP.O. Box 447, 2300 AK Leiden, The Netherlandse-mail: [email protected] – www.taylorandfrancis.co.uk – www.balkema.nl

ISBN: 978-0-415-80484-4 (Hbk)




Earthquake geotechnical case histories for performance-based design – Kokusho (ed) © 2009 Taylor & Francis Group, London, ISBN 978-0-415-80484-4

Table of Contents

Preface VII

About the Editor IX

Introduction XI

The 26 Case Histories XIII

Geotechnical performance of the Kashiwazaki-Kariwa Nuclear Power Stationcaused by the 2007 Niigataken Chuetsuoki earthquake 1

T. Sakai, T. Suehiro, T. Tani & H. Sato

2006 large-scale rockslide-debris avalanche in Leyte Island, Philippines 31 R.P. Orense & M.S. Gutierrez

Slope failures during the 2004 Niigataken Chuetsu earthquake in Japan 47T. Kokusho, T. Ishizawa & T. Hara

Slump failure of highway embankments during the 2004 Niigataken Chuetsu earthquake 71 K. Ohkubo, K. Fujioka & S. Yasuda

Fill slope failure of the Takamachi housing complex in the 2004 Niigataken Chuetsu earthquake 83S. Ohtsuka, K. Isobe & T. Takahara

Uplift of sewage man-holes during 1993 Kushiro-oki EQ., 2003 Tokachi-oki EQ.and 2004 Niigataken Chuetsu EQ 95S. Yasuda, T. Tanaka & H. Kiku

Fluidisation and subsidence of gently sloped farming fields reclaimed

with volcanic soils during 2003 Tokachi-oki earthquake in Japan 109Y. Tsukamoto, K. Ishihara, T. Kokusho, T. Hara & Y. Tsutsumi

Quay wall displacements and deformation of reclaimed land duringrecent large earthquakes in Japan 119T. Sugano

River dike failures during the 1993 Kushiro-oki earthquakeand the 2003 Tokachi-oki earthquake 131Y. Sasaki

Ground failures and their effects on structures in Midorigaokadistrict, Japan during recent successive earthquakes 159

K. Wakamatsu & N. Yoshida

Behaviour of SCP-improved levee during 2003 Miyagiken-Hokubu Earthquake 177 A. Takahashi & H. Sugita

V




Tsukidate failure compared with the other flow-type failure during2003 earthquakes in Northern Japan 185

M. Kazama, R. Uzuoka, N. Sento & T. Unno

Liquefaction and ground failures during the 2001 Bhuj earthquake, India 201 H. Hazarika & A. Boominathan

Las Colinas landslide caused by the 2001 El Salvador Earthquake 227 K. Konagai, R.P. Orense & J. Johansson

Seismic motions at Hualien LSST arrays during the 1999 Chi-Chi earthquake 245C.H. Chen & S.Y. Hsu

Chiufenerhshan landslide in Taiwan during 1999 Chi-Chi earthquake 259 M.L Lin, K.L. Wang & T.C. Chen

Tsaoling landslide in Taiwan during the 1999 Chi-Chi earthquake 273 M.L Lin, K.L. Wang & T.C. Chen

Liquefaction induced ground failures at Wu Feng caused by strongground motion during 1999 Chi-Chi earthquake 289W.F. Lee, B.L. Chu, C.C. Lin & C.H. Chen

Failures of soil structures during the 1999 Taiwan Chi-Chi earthquake 311C.C. Huang

Performance of buildings in Adapazari during the 1999 Kocaeli, Turkey earthquake 325 J.D. Bray & R.B. Sancio

Case histories of pile foundation subjected to ground displacementsin the 1995 Hyogoken-Nambu earthquake 341

K. Tokimatsu

Damage investigation on the foundations of the Hanshin Expressway Route5 caused by the 1995 Hyogoken-Nambu earthquake 357

N. Hamada, F. Yasuda, A. Nakahira & T. Tazoh

Damage to subway station during the 1995 Hyogoken-Nambu (Kobe) earthquake 373 N. Yoshida

Vertical array records during 1995 Hyogoken-Nambu earthquake by Kobe city,KEPCO and other organizations 391Y. Iwasaki

Observed seismic behavior of three Chilean large dams 409 R. Verdugo & G. Peters

Liquefaction-induced flow slide in the collapsible loess deposit in Tajik 431 K. Ishihara

VI





Preface

Performance-Based Design (PBD) is increasingly employed recently in structural design of build-ings and infrastructural facilities in many countries. However, PBD has not yet been established sufficiently in geotechnical engineering practice. Seismically induced ground deformation essen-tial to performance design is not easy to evaluate mainly because, in contrast to superstructures,the ground is a 3-dimensional continuum with tremendous spatial variability and its stress-strainrelationship is strongly nonlinear with dilatancy effect.

A rapid development and establishment of practical and reliable PBD is thus needed not only for foundation design but also for superstructures resting on incompetent soils. It is particularly true

under circumstances where seismic ground motions observed during recent destructive earthquakesare getting larger. Such large motions often lead to intolerable results of foundation ground and superstructures resting on it, if they are designed by the conventional limit design methodologies.Thus, we are urged to reconsider how to design new buildings and new civil engineering structures properly and also how to retrofit existing structures from the viewpoint of their performance under increasing seismic loads.

The first task toward this direction is to establish the performance criteria in the arena of earthquake geotechnical design to comply with the performance of buildings or civil engineer-ing structures. The next major challenge for the geotechnical engineering community is to shiftfrom the limit state design to the strain/deformation evaluation based on time/frequency-domaincalculations not only in research front but also in engineering practice as well.

More and more numerical analyses incorporating time-histories of input seismic motions and strong nonlinear response of soils are already in practice in this respect. However, in contrast to theconventional methods, uncertainties involved in the PBD become considerable in terms of seismicinput, large-strain soil properties, variability of soil properties, optional parameters in numericalanalyses, etc, which almost inevitably attracts designers’ attention from deterministic methods to

probabilistic approaches.What we need in choosing appropriate input parameters and judging the reliability of analyti-

cal results is a sort of benchmark case histories with well-documented geotechnical and seismicconditions. As one of the activities of TC4, ISSMGE in the 2005–2009 term, it has been planned to publish a case history volume accommodating well-instrumented geotechnical and earthquakedata of high qualities, so that international geotechnical researchers can refer it as a benchmark

for developing performance-based design methodologies. The title of the Volume is “EarthquakeGeotechnical Case Histories for Performance-Based Design”.

Items to be addressed in this volume are;

1) Dynamic ground response during strong earthquakes associated with soil nonlinearity, liq-uefaction, etc. by array systems including seismographs and/or piezometers, strain gauges,etc.

2) Post earthquake residual deformations or flow type failures of natural slopes or earth-structures,such as dams, levees, embankments, cut and fill residential lands, retaining walls, quay walls,etc. by showing 3-dimensional changes before and after earthquakes and also time histories of

the deformations if possible.3) Soil-foundation interactions in dynamic response or in lateral spreading ground for piles, raftfoundations, improved soils, etc.

The case histories included here are limited only for prototypes, excluding model tests, accompa-nied by earthquake records obtained nearby, geotechnical surveying data, structural data and other

pertinent parameters. Records, measurements and associated information relevant to case historystudies are accommodated as much as possible. Not only case histories of drastic failures but alsocomparable cases with smaller or no failures are dealt with if possible. Numerical studies associated

VII




with case histories are not included in this volume but listed in references if necessary. Each casehistory consists of site characterization, characterization of earthquake motions and descriptions of failure as quantitative as possible. Whenever available, digital data on earthquake records and other information associated with individual cases are stored in the attached CD-ROM as text files. Thevolume, intended exclusively for research purposes, has been peer-reviewed, edited and published

by TC4, ISSMGE.

I would like to acknowledge all the authors of the individual case histories in this volume for their great contributions despite their busy times. I am also grateful to Dr. Hemanta Hazarika, Secretary of TC4, Akita Prefectural University, Japan, for his great help in the review procedures. Dr. TomohiroIshizawa, Chuo University, who contributed in formatting data files in CD-ROM is also gratefullyacknowledged. The great cooperation from a number of reviewers who generously served their

precious time in reviewing and editing the manuscripts both technically and grammatically isgratefully appreciated.

Finally, it is my sincere hope that this case history volume will be serving as a sort of benchmark or common scale in developing design methodologies and numerical tools and judging their reliabilityfor Performance Based Design in Earthquake Geotechnical Engineering.

2009, June.Takaji Kokusho

Chairman, TC4 (Earthquake Geotechnical Engineering and Associated Problems), ISSMGE Professor of Civil Engineering Department, Chuo University, Tokyo, Japan.

VIII





About the Editor

KOKUSHO,Takaji

Professor, Faculty of Science and Engineering, Chuo University,Tokyo, Japan.Chairman, TC4, ISSMGE.

MS Degree from Tokyo University in 1969.MS Degree from Duke University USA in 1975.

PhD. (Engineering) from the University of Tokyo 1982.“Dynamic soil properties and nonlinear seismic response of ground”Research Engineer in Central Research Institute of ElectricPower Industry (CRIEPI) in Japan in 1969.Director of Siting Technology Division in CRIEPI in 1989Professor in Civil Engineering Department, Faculty of Science and Engineering, Chuo University since 1996

Research Field:Dynamic soil properties and their evaluations.

Seismic response of ground.Liquefaction of sandy/gravelly soils.Earthquake-induced slope failure.

Academic Society Activity:Vice-President of Japanese Geotechnical Society (2002–2003).Chairman of Asian Technical Committee No.3 (Geotechnology for Natural Hazards)of ISSMGE (1998–2005).Chairman of Technical Committee No.4 (Earthquake Geotechnical Engineering &Associated Problems), ISSMGE (since 2006)

IX


http://www.crcnetbase.com/action/showImage?doi=10.1201/NOE0415804844.fmatt&iName=master.img-000.jpg&w=91&h=116




Introduction

In this case history volume, 26 case histories are accommodated in a sequence from newer toolder seismic events (CH01∼CH26) as listed in The 26 Case Histories. As shown in the table, thesubjects of the papers include slope failures, liquefaction, lateral spreading, soil subsidence, failureof embankments, cut/bank residential lands, levees and dams, pile-soil interaction in liquefied ground, subsidence of building foundations, deformation of quay walls, failure of retaining walls,deformation and uplift of underground structures, soil-structure interactions and ground motionsin liquefied soil.

In most of the papers, photographs and diagrams are originally prepared in full color and theyare sometime not clear enough to understand well in the black & white printed version. The readersare therefore recommended to refer to the full color copies of the attached CD-ROM along withthe printed version.

Individual authors tried their best to make their case histories as quantitative as possible in termsof earthquake ground motion records, geotechnical investigation data, etc. Some of the papers areaccompanied by digital data stored in the CD-ROM as indicated in the table. They are formatted inthe same order as in the table of contents and the associated data files can easily be accessible byclicking the same table in the CD-ROM. However, in many cases, digital records were difficult to

be included in this volume due to copy right restrictions. To cope with this difficulty, appropriateliteratures, URLs and mailing addresses are introduced in individual papers whenever possible,from where the readers may be able to obtain the digitized data.

The attached CD-ROM is formatted in the following sequence;

1) Full-color electronic version of all the case history papers in the same order as in the table of content.

2) Data files associated with the papers designated in the table, which can be accessible by clickingthe rows of the table in the CD-ROM.

If the readers make use of these data, they are kindly requested to write the data source in theacknowledgments of their documents.

Takaji Kokusho

Chairman, TC4 (Earthquake Geotechnical Engineering and Associated Problems), ISSMGE Professor of Civil Engineering Department, Chuo University, Tokyo, Japan.

XI




The 26 Case Histories

Authors

Chapter title (Corresponding author) Major topics

CH01 Geotechnical performance at the Kashiwazaki- T. Sakai, T. Suehiro, Ground subsidence

Kariwa Nuclear Power Station caused by the T. Tani & H. Sato, Differential settlemen

2007 Niigataken Chuetsu earthquake

CH02 2006 large-scale rockslide-debris avalanche R.P. Orense & Slope failure

in Leyte Island, Philippines. M.S. Guttierez

CH03 Slope failures during the 2004 T. Kokusho, T. Ishizawa, & Slope failure

Niigataken Chuetsu earthquake in Japan. T. Hara

CH04 Slump failure of highway embankments during K. Ohkubo, K. Fujioka & Road embankment

the 2004 Niigataken-Chuetsu earthquake. S. Yasuda slump and slide failu

CH05 Fill slope failure of the Takamachi housing complex

in the 2004 Niigataken Chuetsu earthquake.

S. Ohtsuka, K. Isobe, &

T. Takahara

Failure of cut and fill

residential land

CH06 Uplift of sewage man-holes during 1993

Kushiro-oki EQ., 2003 Tokachi-oki EQ. and

2004 Niigataken-Chuetsu EQ.

S. Yasuda, T. Tanaka &

H. Kiku

Uplift of buried

structures due to

liquefaction

CH07 Fluidisation and subsidence of gently sloped

farming fields reclaimed with volcanic soils

during 2003 Tokachi-oki earthquake in Japan.

Y. Tsukamoto, K. Ishihara,

T. Kokusho, T. Hara &

Y. Tsutsumi

Slope failure due to

liquefaction

CH08 Quay wall displacements and deformation of T. Sugano Failure of quay walls

reclaimed land during recent large earthquakes

in Japan.

by liquefaction

Soil improvement

CH09 River dike failures during the 1993 Kushiro-oki

earthquake and the 2003 Tokachi-Oki earthquake

Y. Sasaki River dike failure

Liquefaction

CH10 Ground failures and their effects on structures K. Wakamatsu & Failure of cut and fill

in Midorigaoka district, Japan during recent

successive earthquakes.

N. Yoshida Seetlement and fissu

Retaining wall

CH11 Behaviour of SCP-improved levee during 2003

Miyagiken-Hokubu earthquake.

A. Takahashi & H. Sugita River dike failure

Soil improvement

CH12 Tsukidate failure compared with the other flow-type

failure during 2003 earthquakes in Northern Japan.

M. Kazama, R. Uzuoka,

N. Sento & T. Unno

Slope failure due to

liquefaction

Foundation

XI I I




CH13 Liquefaction and ground failures during the 2001

Bhuj earthquake, India.

H. Hazarika &

A. Boominathan

Liquefaction

Failure in port facilit

CH14 Las Colinas landslide caused by the 2001 El

Salvador earthquake.

K. Konagai, R.P. Orense

& J. Johansson

Slope failure

CH15 Seismic motions at Hualien LSST arrays during

the 1999 Chi-Chi earthquake.

C.H. Chen & S.Y. Hsu Soil-structure interac

of buildings

CH16 Chiufenerhshan landslide in Taiwan during1999 Chi-Chi earthquake.

M.L Lin, K.L. Wang& T.C. Chen

Slope failure

CH17 Tsaoling landslide in Taiwan during the 1999

Chi-Chi earthquake.

M.L Lin, K.L. Wang &

T.C. Chen

Slope failure

CH18 Liquefaction-induced ground failures at Wu Feng

caused by strong ground motion during 1999

Chi-Chi earthquake.

W.F. Lee, B.L. Chu,

C.C. Lin & C.H. Chen

Liquefaction and

lateral spread

Ground motion

Ground subsidence

CH19 Failures of soil structures during the 1999 Taiwan

Chi-Chi- earthquake.

C.C. Huang Reinforced soil wall

Retaining wall

CH20 Performance of buildings in Adapazari during

the 1999 Kocaeli, Turkey earthquake.

J.D. Bray & R.B. Sancio Building subsidence

due to liquefaction osilty soils

CH21 Case histories of pile foundation subjected to

ground displacements in the 1995

Hyogoken Nambu earthquake.

K. Tokimatsu Pile foundations of

buildings in

liquefied ground

CH22 Damage investigation on the foundations of the

Hanshin Expressway Route 5 caused by the

1995 Hyogoken-Nambu earthquake.

N. Hamada, F. Yasuda,

A. Nakahira & T. Tazoh

Pile foundations of

viaducts in

liquefied ground

CH23 Damage to subway station during the 1995

Hyogoken-Nambu (Kobe) earthquake.

N. Yoshida Soil-structure interac

of underground struc

CH24 Vertical array records during 1995 Hyogoken- Y. Iwasaki Ground response

Nambu earthquake by Kobe city, KEPCO and Soil nonlinearity other organizations. Liquefaction

CH25 Observed seismic behavior of three Chilean

large dams.

R. Verdugo & G. Peters Dam settlement

CH26 Liquefaction-induced flow slide in the collapsible

loess deposit in Tajik.

K. Ishihara Slope failure

XI V




Geotechnical performance of the Kashiwazaki-Kariwa Nuclear

Power Station caused by the 2007 Niigataken Chuetsu-okiearthquake

T. Sakai, T. Suehiro & T. Tani Niigataken Chuetsu-oki Earthquake Restoration Management Center, Tokyo Electric Power Company,Tokyo, Japan

H. Sato R & D Center, Tokyo Electric Power Company,Yokohama, Japan

ABSTRACT: This study outlines the geotechnical performance, such as ground subsidence,that occurred on the premises of the Kashiwazaki-Kariwa Nuclear Power Station as a result of the

Niigataken Chuetsu-oki earthquake on July 16, 2007 (magnitude of 6.8 on the JMA scale). Whereasthe plant buildings, buried ducts, and other major structures are supported by Tertiary beds, thesurrounding areas are backfilled with alluvial and Pleistocene sands of 10 to 25 m in thickness.Subsidence of a few tens of cm to more than 1 m (1.6 m maximum near the buildings) occurred widely in the backfill areas where the soil was dewatered and unsaturated, causing damage to nearbyequipments resting directly on the soil. The large subsidence of the backfill soil corresponding tocompressive strain up to 6% near the buildings may be explained by cyclic shearing of unsaturated

backfill soils reflecting the soil–structure interaction.

1 INTRODUCTION

Ground subsidence of a few tens of cm to 1 m occurred widely on the premises of the Kashiwazaki-Kariwa Nuclear Power Station (NPS) as a result of the Niigataken Chuetsu-oki earthquake onJuly 16, 2007. In particular, large subsidence of 1 m or more occurred around the reactor and turbine buildings, and damage to nearby equipment was caused by differential settlement between

buildings and equipment. Subsidence occurred mainly in unsaturated soil around buildings under

the influence of sub-drainage which will be mentioned later. The observed ground compressivestrain is as high as a few percent, which has been rarely observed in the past. Sand boils caused

by liquefaction were observed at many locations in the ocean side areas, where subsidence and cracking were also observed.

The cause of backfill soil subsidence, which led to equipment damage, is currently being studied now (as of April 2008). This paper summarizes the study conducted so far.

2 OUTLINE OF EARTHQUAKE AND OBSERVED SEISMIC DATA

2.1 Outline of earthquake

An earthquake with a magnitude of 6.8 on the Richter scale occurred at 10:13 a.m. on July 16,2007. The hypocenter was off the Jou-chuetsu area of Niigata Prefecture at a depth of 17 km. Thegreatest seismic intensity observed in Niigata and Nagano Prefectures was in the upper Six value inthe JMA scale. Many lifeline utilities, slopes, harbors, and houses were damaged. The earthquake,which is of the shallow crust type, is associated with a reverse fault having a northwest-southeastcompressive axis. The details of the earthquake mechanism are still being studied by governmentaland other organizations.

1




Figure 1. Locations of the Kashiwazaki-Kariwa NPS and the epicenter of the earthquake.

Figure 2. Locations of seismometers installed within the plant site.

Figure 1 shows the location of the Kashiwazaki-Kariwa NPS and the epicenter of the earthquake.The station, located in an area between Kashiwazaki City and Kariwa Village, has an area of 4,200,000 m2 facing the Sea of Japan. The station consists of seven boiling water reactors (BWR)

having a total generating capacity of 8212 MW, which is the world’s largest capacity held by a single power station. The station is situated 16 km and 23 km distant from the epicenter and hypocenter,respectively.

2.2 Seismic data recorded at station

Figure 2 shows the locations of seismometers installed within the premises of the station. Themaximum accelerations observed on the base mat of the reactor building of each unit are shown inTable 1, together with S2 design acceleration. As shown in Figure 2, Units 1 to 4 and Units 5 to 7 are

2


http://www.crcnetbase.com/action/showImage?doi=10.1201/NOE0415804844.ch1&iName=master.img-001.jpg&w=367&h=229




Table 1. Maximum acceleration, in Gal, observed on the base mats of reactor buildings.

Horizontal (north-south) Horizontal (east-west) Vertical

Unit Gal Gal Gal

1 311 (274) 680 (273) 408 (235)

2 304 (167) 606 (167) 282 (235)

3 308 (192) 384 (193) 311 (235)

4 310 (193) 492 (194) 337 (235)

5 277 (249) 442 (254) 205 (235)

6 271 (263) 322 (263) 488 (235)

7 267 (263) 356 (263) 355 (235)

Numbers in parentheses are design values.

Figure 3. Comparison of observed data with an existing attenuation relationship (modified from data on the

web site of the Kojiro Irikura Earthquake Motion Research Institute).

located in the southern site area (Arahama side area) and the northern site area (Ominato side area),respectively. The observed accelerations are greater on the Arahama side than the Ominato sideand are greater in the east-west direction than the north-south direction. The greatest accelerationof 680 Gal in the east-west horizontal direction was observed on the base mat of the Unit 1 reactor

building. The observed accelerations in all cases exceed the design values for the S2 event.Figure 3 shows observed data compared to existing attenuation relationships. Although the

details of the seismic fault model are not defined yet, the observed values are generally greater than values derived from the existing equation (curves denoted as “rock” in the figure) for both the

southeast-dipping and northwest-dipping fault models.Figure 4 shows observed acceleration time history waveforms as well as observed and designresponse spectra for Units 1 and 7. There is a distinct pulse in the time history waveform that is moredistinct in the southern units than the northern units. The differences in the observed seismic datamay be related to the location of hypocenter and the fold structure of the site (syncline and anticlinein the Arahama and Ominato side areas, respectively). The details are still under investigation.

Seismic data observed at the power station are presented on the web site of Tokyo ElectricPower Co. (http://www.tepco.co.jp/). Time history waveforms are available from the Associationfor Earthquake Disaster Prevention (http://www.aedp-jp.com/).

3


http://www.tepco.co.jp/


http://www.aedp-jp.com/









Figure 4. Observed acceleration spectra and design response spectra on the base mats of Unit 1 and 7 reactor

buildings (east-west horizontal).

3 OUTLINE OF DAMAGE TO POWER STATION

Four out of the seven Kashiwazaki-Kariwa units were in service when the earthquake occurred.The other three were out of service for periodic inspection. Upon detection of the earthquakemotion, (1) all control rods were inserted (shutdown), (2) reactor water level was maintained and coolant temperature and reactor pressure were decreased (cooling), and (3) pellet, fuel claddingtubes and the pressure vessels were contained without causing environment impact (containment),

although very minor radioactive leaks to the sea and the atmosphere occurred. As a result, safetywas maintained through designed plant behavior and appropriate operator performance.

Items of equipment damage identified up to now by inspection are summarized in Figure 5.Damage was not found in class-As and class-A equipment (reactor pressure vessels, primarycontainment vessels, reactor buildings, etc.), which are important to safety. Damage was restricted to class-B and class-C equipment (turbine facilities, transformers, fire protection piping, diversionchannels, etc.), which are less important to safety.

The soil condition of the power station is alluvial sand distributed widely around the area, and underlying Pleistocene sand. The reactor buildings and other important structures are based onthe Neogene Nishiyama stratum (mudstone), whereas less important structures are based directly

on backfill soils of the alluvial and Pleistocene sands. A lot of damage incidents were caused byrelative settlement between the different supporting beds.Figure 6 shows the location of the fire at the Unit 3 house transformer adjacent to the Unit

3 turbine building. The transformer founded on piles driven into the Nishiyama stratum did notsubside. However, the connection bus, based directly on the backfill soil, subsided by 20 to 25 cm.The relative settlement caused an oil leak from the transformer and eventually a fire.

Figure 7 shows ground subsidence around the Unit 1 light oil tank. Whereas the tank, whichis an important structure supported by the Nishiyama stratum, did not subside, the backfill soiladjacent to the tank subsided greatly so that a gap was generated in the tank foundation. As will

4





Figure 5. Damage of equipment.

5





Figure 6. Fire at Unit 3 house transformer.

be discussed in Section 4, the maximum subsidence is about 1.6 m, and the average compressivestrain calculated from the backfill thickness of 25 m exceeds 6%.

Figure 8 shows damage to fire protection piping buried in backfill soil around a building.

Subsidence of the backfill soil caused rupture of threaded and coupled joints.The above three damage incidents are all caused by the settlement of the surrounding soil relativeto the bedrock-supported house transformer, light oil tank, and building. The subsided backfillsoils around the building and the light oil tank are unsaturated because the groundwater levels werelowered by sub-drainage. The unsaturated soil subsided under cyclic shear during the earthquakeand the soil adjacent to the bedrock-supported structures subsided more than that distant from thestructures. The details of subsidence will be described in Section 4.

Figure 9 shows another type of structural damage; the base of a tank buckled and bolts securingthe base failed.

6





Figure 7. Subsidence around Unit 1 light oil tank.

Figure 8. Rupture of outdoor fire protection piping.

Figure 9. Buckled filtrate tank and failed bolt (Ominato side area).

7







4 GROUND DEFORMATION BY EARTHQUAKE

As described in Section 3, the earthquake caused subsidence in areas backfilled with alluvial and Pleistocene sand. The ground deformation is detailed in this section.

4.1 Ground deformation measurement methods

The ground deformation caused by the earthquake in the premises of the station was evaluated using the following survey data:

– Before the earthquake: aerial photogrammetric data (1/4000 scale; April 26, 2006) – After the earthquake: topographic survey (right after the earthquake) and aerial photographic

data (1/6000 scale; July 19, 2007)

On the basis of the above data, the following ground deformation maps were prepared:

– Sand boil and crack distribution maps (Figures 13 and 19)

– Subsidence contour maps (Figures 24 and 27)

Figures 13 and 19 were created by comparing the above aerial photographic data before and after the earthquake to identify sand boils and cracks that are considered to be caused by the earthquake.Figures 10 shows an example of creating a distribution map for the north side of the Unit 4 intakein the Arahama side area.

Figures 24 and 27 were created by comparing the topographic survey data obtained right after the earthquake with the aerial photogrammetric data obtained before the earthquake to evaluatethe amounts of subsidence. Although this is not a strict, quantitative evaluation because two dif-ferent types of survey are compared, the created maps should be suitable for broadly grasping thedistribution of subsidence.

4.2 Sand boils and cracks

The following observations are drawn from Figures 13 and 19.

(1) In both the Arahama and Ominato side areas, sand boils are observed only near the revetments(ground level 3 m aboveTokyo Peil, orTP 3 m) and on the inland side of the reactor buildings. Asnoted in Section 3, sand boils are not observed around the buildings, because the groundwater level is lowered by sub-drainage to mitigate uplift pressure acting on the building foundationsand to maintain horizontal shear resistance (see “Figure 32 Groundwater level distribution”).

Figure 10. Example of creating distribution map of sand boils and cracks (north side of Unit 4 intake).

8





(2) In the Arahama side area, sand boils are observed on the TP 3 m ground level along the Unit 1emergency intake channel, on the north side of the Unit 4 intake, and on the inland side of theUnit 3 and 4 reactor buildings. In the Ominato side area, sand boils are observed on the southside of the outlet and around the feed-water tank and parking lot. Representative sand boils areshown in Figure 11 (near the Unit 1 emergency intake channel) and Figure 12 (north side of the Unit 4 intake), with their locations shown in Figure 13 .

Figure 11. Sand boil near Unit 1 emergency intake

channel.

Figure 12. Sand boil on the north side of Unit 4

intake.

Figure 13. Distribution of sand boils and cracks (Units 1-4, Arahama side area).

9







Figure 14. Cracks on the ocean-side road of Unit 1. Figure 15. Cracks on slope between the outdoor

switchyard and the TP 5 m ground level.

Figure 16. Cracks in ground around incinerator building.

(3) There are many cracks on slopes and other areas in both the Arahama and Ominato side areas.

These cracks can be divided into five categories according to their assumed cause. The locationsof the categorized cracks are shown in Figures 13 and 19.

(a) Cracks caused by liquefactionThese cracks occur in the areas where sand boils are found; i.e. near the revetments and onthe inland side of the reactor buildings. Examples of this type in the Arahama and Ominatoside areas are shown in Figure 14 (ocean-side road of Unit 1) and Figure 20 (south side of Unit 5–7 outlet), respectively.

(b) Cracks caused by slope movementIn the Arahama side area, this type of cracks are observed on the slope between the TP 5 mground level, on which buildings and main facilities stand, and the TP 3 m ground level

(revetment side) as well as on the slope between the outdoor switchyard and the TP 5 mground level. In the Ominato side area, cracks are also observed on the slope between theTP 12 m ground level, on which buildings and main facilities stand, and the TP 3 m ground level (revetment side). Examples of this type in the Arahama and Ominato side areas areshown in Figure 15 (slope between the outdoor switchyard and the TP 5 m ground level) and Figure 21 (slope between the TP 12 m and TP 3 m ground levels), respectively.

(c) Cracks along building edges caused by ground subsidence around buildingsThese cracks occur around buildings because of differential settlement between the buildingsand the surrounding ground. Examples of this type of cracks in the Arahama and Ominato

10







Figure 17. Step in a buried duct between Units 2 and 3.

Figure 18. Cracks developed along the boundary of excavation line and backfill soil on the inland side of

Unit 4 reactor building.

side areas are shown in Figure 16 (around the incinerator building) and Figures 22 (around the Unit 5 seawater heat exchanger building), respectively.

(d) Cracks and steps along buried structures

These cracks and steps occur along the boundaries between buried structures supported by improved soil and their surrounding soil. Representative examples in the Arahama and Ominato side areas are shown in Figure 17 (buried duct between Units 2 and 3) and Figure23 (buried duct of the Unit 5 seawater heat exchanger building), respectively. In the caseof Figure 17, a step occurred between the buried duct, which did not subside significantly

because it was supported by low-strength concrete, and the backf ill soil that subsided.(e) Cracks along excavation lines formed during construction

These cracks occur along the boundaries of backfilled excavations. Figure 18 showsrepresentative cracks developed on the inland side of the Unit 4 reactor building.

4.3 Subsidence

The following observations are drawn from the subsidence contour maps of Figures 24 and 27.

(1) Subsidence is large (on the order of few meters) near the buildings and decreases as the distancefrom the buildings increases. There are also large-subsidence areas on the ocean and inland sides.

Accordingly, subsidence is described separately for three areas: near the buildings; slightlydistant from the buildings; and apart from the buildings on the ocean and inland sides.

11






Figure 22. Cracks in ground near Unit 5 seawater

heat exchanger building.

Figure 23. Cracks along a buried duct of Unit 5

seawater heat exchanger building.

Figure 24. Subsidence contour map (Units 1–4, Arahama side area).

Figure 25. Subsidence of ground near Unit 1 seawater heat exchanger building.

13








Figure 26. Subsidence of ground near Unit 2 reactor building.

Figure 27. Subsidence contour map (Units 5–7, Ominato side area).

Subsidence is large around the Units 1-2 turbine buildings and the Unit 1 seawater heatexchanger building; the maximum subsidence is 1.6 m with a backfill thickness of 25 m (Figure31). In the Ominato side area, the maximum subsidence (1.0 m with a backfill thickness of

17 m) is observed on the ocean side of the Unit 5 seawater heat exchanger building. In bothlocations, the average compressive strain calculated from backfill thickness exceeds 6%.Examples of this type of subsidence are shown in Figure 25 (Unit 1 seawater heat exchanger

building), Figure 26 (Unit 2 reactor building), Figure 28 (Unit 5 seawater heat exchanger building), and Figure 29 (pressure control room pool water surge tank).

(3) In the areas slightly distant from the buildings, subsidence is large around the Unit 1 turbine building and seawater heat exchanger building as well as the Unit 5 seawater heat exchanger building. The contour maps indicate a subsidence range of 10–50 cm in these areas. The averagecompressive strain calculated from backfill thickness (17–25 m) ranges between 1% and 2%.

14






Figure 28. Subsidence of ground near Unit 5 sea-

water heat exchanger building.

Figure 29. Subsidence of ground near pressure

control room pool water surge tank.

Figure 30. Subsidence of slope between the TP 12 m and 3 m ground levels.

The observed subsidence can be explained by volume contraction within both unsaturated soilabove the groundwater level and saturated soil below the groundwater level.

(4) In the areas apart from the buildings, subsidence is large on the ocean side of Units 1 and 2and the inland side of Unit 4 in the Arahama side area. In the Ominato side area, subsidenceis large on the ocean side of Units 5–7 and the inland side of Units 5 and 7. The distributionof this type of subsidence mostly agrees with the distribution of sand boils and cracks shownin Figures 13 and 19. It is considered that liquefaction and slope bulging of slope crest caused the observed subsidence.

Examples are shown in Figure 14 (ocean-side road of Unit 1), and Figure 30 (slope betweenthe TP 12 m and TP 3 m ground levels in the Ominato side area).

5 STUDY ON CAUSES OF GROUND DEFORMATION

Thickness, extent and quality of backfill and groundwater distribution were surveyed to investigatethe causes of ground subsidence.

5.1 Thickness and extent of backfill

Backfilling of buildings and other structures such as intake and diversion channels was conducted from 1981 to 1996. The horizontal and vertical distribution and type of backfill are summarized inFigure 31.

15







As described in Section 3, the reactor buildings, turbine buildings, and seawater heat exchanger buildings are directly supported by the Tertiary Nishiyama stratum. Man-made rock (MMR: soilcement using on-site rock to have identical mechanical properties to Tertiary stratum) is also used

partly as foundation rock. The stratum is overlain by backfill, which are on-site excavated material.Backfill thickness ranges between 15 and 30 m in the Arahama side area (Units 1–4) and between11 and 18 m in the Ominato side area (Units 5–7). As mentioned in Section 4, the reactor buildings

of Units 6 and 7 are surrounded by continuous underground walls and almost no backfill is used around the units.

The ordinary and emergency intake channels on the ocean side were constructed on Pleistocenesoils and backfilled with the alluvial sand as well as the Pleistocene sand. In some sections, soilcement and other man-made materials are used as foundation rock. Backfill sand thickness for thechannels is smaller than that for the buildings.

Most of the backfill used for the buildings, emergency intake channels and intake channelsconsists of the alluvial sand and the Pleistocene sand. Though different construction machinerywas used in backfilling according to the timing, the quality was controlled to achieve a standard

percent compaction (dry density/maximum dry density) of 95%. Subsidence and other ground

deformation occurred almost exclusively in the backfill during the earthquake.

5.2 Groundwater level

Additional groundwater observation wells were drilled throughout the areas where ground defor-mation occurred because the number of pre-earthquake wells was small. CPT boreholes, rock test

boreholes, and transformer oil leakage test boreholes were also used as observation wells.Figure 32 shows groundwater distribution observed by using the observation wells in October

and November 2007 (3 and 4 months after the earthquake). Figures 33 and 34 show cross sectionsof individual units along the direction perpendicular to the shoreline, including the backfill shownin Figure 31. Groundwater levels shown in the cross sections are based on the values in observationwells close to the cross section line.

The sub-drainage systems to reduce groundwater level consists of collecting pipes installed below and around the building base plane and pits at the corners of the building. The groundwater levels in the pits are also shown in Figures 32–34.

It was observed that:

(1) Groundwater levels around the buildings were lowered by sub-drainage.(2) The groundwater level around the buildings generally corresponds to the boundary between

the backfill and the bedrock Nishiyama stratum. In the Arahama side area, groundwater levelis notably low in Unit 1 and on the ocean side of Units 2–4 because the backfills are thick (25–30 m) as shown in Figures 31and 33.

(3) Groundwater level gradually rises to sea level from the building area toward the shoreline. Inthe Ominato side, the rise of groundwater is not significant because the site elevation (TP 12 m)is higher than the Arahama side and backfill thickness is small (about 12 m).

(4) Away from the buildings, the groundwater levels were higher. The distribution of highgroundwater levels broadly corresponds with the distribution of sand boils (Figures 13 and 19).

5.3 Physical properties of backfill soil

Boreholes were drilled to identify the basic physical properties of backfill soil in the areas whichdeformed around the buildings. Laboratory testing was conducted using disturbed specimens to

investigate the mechanism of ground subsidence.

5.3.1 Locations of exploratory boreholes and investigation itemsExploratory boreholes were drilled at nine locations near the reactor buildings Units 1 through 5reactors (a total of 18 boreholes were drilled, or two at each location) where relatively substantialsubsidence occurred. Boreholes were drilled at three locations at different distances from the

building of Unit 1 and 4 reactors. The locations are shown in Figure 35. The distances from the buildings to the boreholes are listed in Table 2.

17




Figure 32. Groundwater level distribution (October/November 2007).

18





Figure 33. Cross section of backfill and groundwater level observed in October/November 2007 (Arahama side are

1 9





Figure 34. Cross section of backfill and groundwater level observed in October/November 2007 (Ominato side are

2 0





Table 2. Locations of boreholes drilled for investigating backfill soil and distance from building.

Unit 1 Unit 4

Unit 2 Unit 3 Unit 5Items 1-a 1-b 1-c 2-a 3-a 4-a 4-b 4-c 5-a

Distance from building (m) 2.0 11.5 22.5 5.2 3.1 1.6 6.7 24.2 1.0

Backfill soil thickness (m) 25 15 15 15 12

Investigations included the logging of P and S wave velocities (downhole measurements), loggingof densities, measurement of borehole diameters with calipers and undisturbed soil samplings.Standard penetration tests were also conducted in boreholes adjacent to each location. The soilsamples were used in laboratory tests to measure grain size, water content, density and maximumand minimum densities. The measurements were made at vertical intervals of approximately 1 m.

5.3.2 Results of exploratory borehole drilling and physical testing

5.3.2.1 Physical properties of backfill soil near the buildingThe results of grain size test are shown in Figure 36. Soils are classified mainly as sand (S), sand with fines (S-F) or fine sand (SF) by Japanese Geotechnical Society Standard regardless of theUnits 1–5. The sand is relatively homogeneous with an average fines content of approximately 10%and the average grain size of 0.2 to 0.3 mm.

P and S wave velocities, SPT N-values, dry densities, wet densities, natural water contents,saturation ratios, fines contents and relative densities measured in boreholes near the building areshown in Figure 37. The P and S wave velocities at shallow depths are approximately 200 m/secand 100 m/sec for all the units, but vary depending on the units at greater depths. N-value isapproximately 5 to 15 at depths shallower than 12 m. No explicit variance was observed accordingto the units. Dry density varied from 1.5 to 1.7 g/cm3. No increase with depth was observed for allthe units. Natural water content varied from 10 to 20%. The water content is higher for Units 3 and 4 than for Units 1 and 2, especially high for Unit 4 (20% or higher), because of the groundwater distribution described earlier (Figure 32). The saturation ratio is correspondingly higher for Units3 and 4. Fines content varied from 5 to 15%, and is slightly higher for the Unit 3 (10 to 20%),contributing to a high relative density near the unit.

5.3.2.2 Distance from the building and physical properties of backfill soilAs described earlier, boreholes were drilled at three locations at different distances from the buildingof the Units 1 and 4. The results of investigations and testing of the same parameters as in 5.3.2.1above are shown in Figures 38 and 39.

In all of the boreholes near the Unit 1 building, N-values are high at depths of 10 to 15 m belowground level and at greater depths. The S wave velocity and N-value are lower in borehole 1-a near Unit 1 building than in the other two boreholes. Near the Unit 4, N-values at a depth of 9 m or greater are also lower in borehole 4-a than in the other two boreholes. The P and S wave velocitiesare also slightly lower in borehole 4-a than in the other two boreholes. No outstanding variance wasfound in the other physical test results with respect to the distance from the building. Soils mayhave been slightly loose near the buildings because compaction work might be difficult there.

5.3.3 Laboratory test itemsVarious laboratory tests were conducted using disturbedsamples with adjusted density to investigate

the mechanism of the backfill soil subsidence.Unsaturated soils were sampled for laboratory tests near the Unit 1 reactor building as shown inFigure 40 where the backfill was thick and the groundwater level was low. Specimens of saturated and unsaturated soils were sampled on the sea side of the Unit 1 emergency intake channel wherethe groundwater level was high and sand boils were observed in the vicinity.

Physical tests to identify soil particle density, water content, grain size and maximum/minimumdensities were carried out. Consolidated drained triaxial compression tests, cyclic undrained triaxialtests for obtaining liquefaction strength and torsional shear tests using hollow specimens were alsocarried out.

21




Figure 35. Locations of exploratory boreholes for measurement of backfill soil.

In the cyclic undrained triaxial tests, volumetric changes due to the drainage of pore water after the cyclic loading were measured to characterize the subsidence of saturated soil. Double-cell torsional shear test apparatus for hollow specimens was used to obtain volumetric changes of unsaturated soil by cyclic loading.

22





Figure 36. Grain size distribution of backfill soil around reactor building. (The bold lines indicate the areas

of grain size distribution. Extraneous curves indicating the mixing of gravels were discarded.)

23





Figure 37. Physical properties of backfill soil near the building.

24





Figure 38. Distance from the building and physical properties of backfill soil (Unit 1).

25





Figure 39. Distance from the building and physical properties of backfill soil (Unit 4).

26





Table 3. Results of physical tests and consolidated drained triaxial compression tests using disturbed

specimens of backfill soil.

Around Unit 1 Around Unit 1

Items reactor building emergency intake channel

(Physical tests)

Soil particle density (g/cm3) 2.69 2.72

Natural water content (%) 11.0 12.0

Grain size characteristics

Maximum grain size (mm) 4.75 2.00

Gravel content (%) 1.0 0

Fines content (%) 10.9 17.3

Soil class Sand with fine particles (S-F) Fine sand (SF)

Maximum dry density (g/cm3) 1.69 1.67

Minimum dry density (g/cm3) 1.32 1.28

(Consolidated drained triaxial compression tests)

State of specimen(saturated/unsaturated) Unsaturated, w=20% Saturated

Cohesion (kN/m2) 10.0 17.3

Internal angel of friction (degrees) 34 32

Figure 40. Locations where disturbed specimens of backfill soil were collected.

This paper mainly focuses on the physical properties. The characteristics of volumetric changesof saturated and unsaturated soil as well as the results of studies on subsidence mechanism will bereported separately.

5.3.4 Results of laboratory tests5.3.4.1 Results of physical tests and consolidated drained triaxial compression testsPhysical tests and consolidated drained triaxial compression tests were conducted using the dis-turbed samples. The results are listed in Table 3. Grain size distributions are shown in Figure 41.Fines content of the soil around the emergency intake channel is approximately 17%, and higher than 11% around the reactor building. However, physical properties and strength characteristics

(internal friction angle: 32 to 34 degrees) were nearly the same at the two locations.

5.3.4.2 Results of cyclic undrained triaxial tests (liquefaction tests)Saturated specimens were made using disturbed specimens sampled in the backfill soil around Unit 1 emergency intake channel. The saturated specimens were modified to have a dry densityof 1.6 t/m3 (relative density: 85%), which is equal to the average dry density of backfill soil. Theliquefaction tests were conducted under the confining pressure of 98 kN/m2 and four cyclic shear stress ratios. The relationship between the number of loading cycles and the shear stress ratio isshown in Figure 42. The shear stress ratio was 0.26 for 20 cycles of loading.

27





Figure 41. Backfill soil grain size distribution curves.

0. 00

0. 05

0. 10

0. 15

0. 20

0. 25

0. 30

0. 35

0. 40

0. 45

0. 50

RL20=0. 26

Number of cycles of loading Nc

Double amplitude of strain DA=1%

DA=2%

DA=5%

Excess water pressure ratio =95%

0. 00

0. 05

0. 10

0. 15

0. 20

0. 25

0. 30

0. 35

0. 40

0. 45

0. 50

RL20=0. 26


0. 001 10 100 1000

0. 05

0. 10

0. 15

0. 20

0. 25

0. 30

0. 35

0. 40

0. 45

0. 50

RL20=0. 26

C y

c l i c s h e a r s t r e s s r a t i o σ d / 2 σ 0

’



DA=2%

DA=5%



DA=2%

DA=5%


Figure 42. Liquefaction strength of saturated soil (backfill soil around Unit 1 emergency intake channel).

28





6 SUMMARY

The findings of a survey on the subsidence of backfill soil, caused by the earthquake at theKashiwazaki-Kariwa Nuclear Power Station are summarized as follows:

(1) Subsidence of soil increases as the distance from the building decreases, and reaches maximum

at the boundary between the soil and the building.(2) Subsidence of unsaturated soil at the boundary is as high as 1.6 m, which corresponds to about

6% in compressive strain (at Unit 1 turbine building, etc.).(3) Subsidence slightly distant from the building is 10–50 cm, which corresponds to about 1% to

2% in compressive strain. The observed subsidence can be explained by total subsidence of unsaturated soil above the groundwater level and saturated soil below that.

(4) In the areas apart from buildings, subsidence is large on both the ocean and inland sides of the buildings. The distribution of the subsidence correlates with the distribution of sand boilsand cracks. It is considered that the observed subsidence was caused by both liquefaction and slope bulging.

(5) Exploratory boreholes were drilled in the backfill around buildings where extensive defor-

mation was observed. Laboratory tests were conducted using disturbed samples. Basiccharacteristics of backfill soil were identified for investigating the mechanism of subsidence.

Model tests and analysis are now being conducted based on the results of tests on the backfillsoil. Efforts are being made toward quantitative evaluation of subsidence near the building due tothe interaction between the building and soil, subsidence in unsaturated soil around the buildingand subsidence of liquefied soil. The results of these studies will be reported at a later time.

29





2006 large-scale rockslide-debris avalanche in Leyte

Island, Philippines

R.P. Orense Department of Civil and Environmental Engineering, University of Auckland, New Zealand

M.S. Gutierrez Division of Engineering, Colorado School of Mines, USA

ABSTRACT: On February 17, 2006, a large-scale rockslide occurred in Southern Leyte Province,Philippines following days of heavy rainfall. This rockslide, considered to be one of the largest tohave occurred in the last few decades, buried almost the entire village of Ginsaugon and caused the death of more than 1300 people. The landslide, which occurred along the steep slope of Mt.Can-abag in the middle part of the province, mobilized large amount of rocks and debris withestimated volume of about 20–25 million m3 and a runout distance of almost 4 km. Heavy rainfall

before the landslide, including a 688 mm rainfall intensity during the 9-day period prior to the slidewhich is equivalent to 2.5 times the mean rainfall amount for the whole month of February, musthave played a major role on the instability of the slope. Minor earthquakes, the strongest of whichhad a surface-wave magnitude of M s =2.6 and whose epicenter was determined to be 23 km westof the landslide site, were reported to have occurred prior to or right after the slide. It is not clear

whether these earthquakes played a role in the triggering of the landslide or whether the landslidegenerated this ground tremors. This paper summarizes the main characteristics of the landslide,discusses its geological, tectonic and climatic setting and looks into possible mechanism and trigger of the landslide.

1 INTRODUCTION

On 17 February 2006, a large-scale landslide (see Fig. 1) occurred in the province of Southern Leyte,located in Leyte Island in the central part of the Philippines. The slide originated on the easternside of the steep rockslope of Mt. Can-abag and buried almost the entire village of Guinsaugon,St. Bernard town, resulting in the loss of life of 1328 people, including 248 school children. Thismakes it the most catastrophic single landslide event recorded in the Philippines and the first major landslide to occur in the 21st century.

In this paper, we summarize the main characteristics of the landslide, discuss its geologic,tectonic and climatic setting, and present possible mechanism and trigger of the landslide. It isworthy to mention that the cause of the Leyte rockslide is not fully understood yet in terms of

Figure 1. View of the large-scale landslide in Southern Leyte.

31





geological, geomechanical and hydrological processes, and no conclusive triggering mechanismshave so far been proposed.

2 GEOLOGIC, TOPOGRAPHIC AND TECTONIC SETTING

Leyte Island is located in the central part of the Philippine Islands. The archipelago is located in oneof the most active geologic settings in the world. The locations of major fault lines in the Philippinesare shown in Fig. 2a. The Philippine Fault Zone (PFZ), a 1,200 km-long active fault system whichtransects the whole archipelago from the Luzon Island in the north to Mindanao Island in the south,traverses through the Leyte Island, including the town of St. Bernard. The devastating earthquake( M s = 7.8) in Luzon Island in 1990 is attributed to the movement along the Digdig segment of the PFZ.

Figure 2. (a) Map of the Philippines showing the location of Leyte Island and the distribution of trenches

and faults (modified from Barrier et al., 1991 and adapted from Evans et al., 2007); (b) Traces of faults in

Southern Leyte (after Cardiel, 2006).

32





The geology of Leyte Island consists of a number of Pliocene-Quatrernary volcanic cones,generally andesitic in nature, tertiary sediments and thick successions of Tertiary volcanic and volcaniclastic rocks (Aurelio, 1992; Sajona et al., 1997). The dominant structure is the PhilippineFault, which bisects the island. The Leyte Segment is one of the most active portions of PFZwith an average left-lateral movement of 2.5- 3.5 cm/yr (Barrier et al. 1991; Besana and Ando,2005). Moderately large historical earthquakes, such as in 1875 with M s = 5.2, 1879 with M s = 6.9

(Bautista and Oike, 2000), 1991 with M s = 5.8 (Domasig, 1991) and 1994 with M s =6.2 (Lanuzaet al., 1994) have been attributed to the movements along the Leyte Segment. The cumulativemovement along this segment formed a steep ridge, which is susceptible to landslides. Alongthe base of the ridge are populated villages and agricultural lands that have been affected and continuously exposed to landslide hazards. The activity of PFZ in Southern Leyte is also manifested

by the frequency of shallow (focal depth< 10 km) small magnitude earthquakes, 87 of whichhave been recorded in the area from January 2000 to February 2006 (Presidential Inter-AgencyCommittee, 2006).

In addition to the PFZ, the area close to Guinsaugon is crisscrossed by a series of other minor faults as identified by surface lineaments (Cardiel, 2006). In addition to the faults that are parallel

to the PFZ, conjugated Reidel-type faults have also been identified, as shown in Fig. 2b. The fault,which forms part of the landslide scarp, is almost parallel to and is a splay of the PFZ. It has not been previously identified as it had no visible exposure on the surface which is overgrown withvegetation, banana and coconut trees.

Climate in Southern Leyte is characterized by the absence of dry season with a very pronounced maximum rain period occurring in the months of November to January. Based on the rainfallrecords obtained by PAGASA (Philippine Atmospheric Geophysical and Astronomical ServicesAdministration) from 1980–2005 in the Otikon rainfall station, located in Libagon town about 7 kmsouthwest of Guinsaugon and on the other side of the ridge, the average monthly rainfall duringthe typhoon season (November–January) is about 350 mm, while the average during the “driest”month (i.e. May) is 91 mm. For the month of February, the average rainfall is 275 mm. Overall, the

average annual precipitation in Otikon based on available data from 1980–2005 data is 2545 mm(Orense and Sapuay, 2006).

3 THE LANDSLIDE

3.1 Overview

The landslide (see Figs. 3a, 3b) occurred between 10:30 and 10:45 AM, local time, on 17 February2006. It involved the movement of an extremely large piece of rock on the eastern face of the 800 m

high Mt. Can-abag, part of the mountain chain that sits on geologic fault running north-souththroughout the province. The scarp created by the slide is about 600 m high, 200 m at its deepest part and about 600 m wide at its base.

From the data obtained by the space shuttle, it was estimated that the main scarp of the landslideis located at the peak of the mountain range (see Fig. 3c). The side slope of the range is rather steep, with an average inclination of about 47◦ on the eastern side where the landslide occurred and about 22◦ on the western side (GSI, 2006).

Initial media reports suggested that the landslide was a mudslide, debris flow or debris avalanche.However, based on f ield observations, the landslide is best classified as rockslide-debris avalancheas defined by Hungr et al. (2001). In a rockslide-debris avalanche, the landslide begins with afailure of a rock slope and proceeds to entrain large quantities of debris. It also involves extremely

rapid, massive, flow-like motion.

3.2 Precursor events

The rockslide followed extensive rainfall which fell on the area since 08 February 2006. Theamount of rain was much higher than normal, attributed to the appearance of La Niña phenomenon.Moreover, on the day of the massive landslide, the region was shocked by two low-magnitudeearthquakes.

33




Figure 3. (a) Aerial view of the Guinsaugon landslide (photo taken by M.D. Kennedy, U.S. Navy); (b) View

of the debris toward the source area of the rockslide-debris avalanche; (c) West to east cross-sectional profile of

the mountain indicating the location and region of flow and deposition of the landslide based on space shuttle

data (after GSI, 2006).

0

50

100

150

200

250

1/1 1/11 1/21 1/31 2/10 2/20

D a i l y R a i n f a l l ( m m )

0

300

600

900

1200

1500

C u m u l a t i v e R a i n f a l l ( m m )

Cumulative

RainfallDaily

Rainfall

Landslide

Figure 4. Daily rainfall from 01 January–28 February 2006 recorded near the landslide site (based on

PAGASA data).

3.2.1 Rainfall records

Records of daily rainfall obtained at the Otikon rainfall station indicate that between 8 to16 Febru-ary 2006, the area was drenched by 687.8 mm of rainfall, with the peak daily rainfall of 171.0 mmoccurring on 12 February (see Fig. 4). Because of concerns regarding possible flooding and lands-liding, some residents of Guinsaugon left their homes and evacuated to safer places. After this, therainfall intensity somewhat subsided, with an intensity of 32.4 mm recorded the day prior to thelandslide. As a result, the people who sought refuge came back to their homes to resume normaldaily activities.

As mentioned earlier, the average rainfall in the area for the month of February is 275 mm,indicating that the nine-day rainfall (from 8 to16 February) prior to the landslide of 687.8 mm is

34





more than 2.5 times the monthly average. In fact, the rainfall intensity registered for the wholemonth of February 2006 was 970.8 mm, the highest monthly rainfall ever recorded since 1980,including the typhoon season. Note that severe rainfall in the area normally ran between November and January. Such unusual climatic changes are brought about by the appearance of La Niña

phenomenon, which is associated with above-normal sea surface temperatures in the West Pacificand stronger trade winds. This pattern had significantly enhanced rainfall across the West Pacific

region.It is worthy to mention that on the day of the slide, only 2.6 mm rain was recorded. By this time,

however, the ground was already saturated. Note also that the rainfall station in Libagon is located on the leeward side of the ridge, which receives lower amount of precipitation due to orographicevents (Catane et al., 2007). Thus, the amount of rainfall that fell on the failed slope adjacent toGuinsaugon must have been higher than those indicated in Fig. 4.

Due to the intense rainfall, several much smaller landslides occurred close to the site of the major landslide prior to 17 February 2006. The most damaging of these earlier landslides occurred on12 February 2006 in the town of Sogod, located about 30 km north of St. Bernard, wherein seven

people were killed.

3.2.2 Earthquake recordsThe Philippine Institute of Volcanology and Seismology (PHIVOLCS) reported that two smallearthquakes occurred in Southern Leyte on the day of the landslide. The first earthquake occurred at 6:07 AM with surface wave magnitude M s =2.3. This earthquake, with epicenter located 10 kmnorthwest of the slide area, is believed to be too far and too weak to have any influence on the slide.

At 10:36 A.M., another earthquake ( M s = 2.6, focal depth =6 km) occurred with epicenter ini-tially estimated by PHIVOLCS to be 23 km west of Guinsaugon (PHILVOCS, 2006). It registered an Intensity II based on PEIS (PHIVOLCS Earthquake Intensity Scale) in the nearby Sogod town(equivalent to Intensity II–III of the Rossi-Forel Scale). Similarly, the United States GeologicSurvey (USGS) recorded an earthquake (body wave magnitude mb =4.3) at the same time, with

epicenter about 4 km north of Guinsaugon and focal depth of 35 km (USGS, 2006). They may bethe same tremor, but because of the small magnitude of the earthquake and since the network of seismometers in the Philippines is not dense, there may have been problems in locating the exactepicenter.

The 10:36 A.M. ground tremor was detected by the two seismic stations positioned in LeyteIsland. The short-period waveforms recorded in PHIVOLCS unmanned satellite-telemetered seis-mic station located in Maasin, Southern Leyte (about 25 km southwest of Guinsaugon) are shownin Fig. 5. The station is equipped with a short-period velocity-type seismometer (Kinemetrics SS1)and Nanometrics Trident digitizer (Narag, 2007). Since the actual seismic data was recorded inSEISAN format, the waveforms, which are expressed in terms of digital counts, can be converted into velocity based on the instrument response. The maximum acceleration corresponding to this

record is in the order of 0.068 g (DPRI, 2006).The influence of the earthquakes in the triggering of the rockslide and the initiation of the debris

flow has not been ascertained. It is possible that the earthquakes, being of shallow in origin, alsooriginated from the hydraulic activation of PFZ or the nearby minor faults, similar to the activation of the fault on the slide scarp. Thus, it is possible that the opposite happened, i.e., the massive landslidegenerated the ground tremor instead of the earthquake being responsible for the triggering of theslide.

3.2.3 Slope instabilityA report by the Presidential Inter-Agency Committee (2006) indicates that previous events of

landslides in the area have occurred in the past, as extracted from topographic maps and recent photographs (see Fig. 6). The report claimed that the presence of screes, or loose rock debris onthe slopes of the mountain (now covered with trees), are indications of old landslide events. Old debris flow deposits were also visible at the base of the mountain. Three generations of witnessesinterviewed at the site claimed that they have not encountered landslides of such magnitude in the

past, indicating that the Guinsaugon landslide may have a return period of more than 100 years.Residents interviewed at the site mentioned that a small landslide occurred from the ridge top

on December 2005. On 13 February 2006, four days before the massive landslide, a small creepingsoil slide occurred at the base of the slope. Residents also reported muddy water flow of Aliho

35




0

5000

10000

0

5000

10000

0

5000

10000

103600H 103700H 103800H 103900H 104000H

N-S Component

E-W Component

U-D Component

Local Time

D i g i t a l C o u n t s



Maasin, Southern Leyte Station (MSLP)

Figure 5. Recorded short-period velocity waveforms at PHIVOLCS unmanned satellite-telemetered seismic

station located in Maasin, Southern Leyte (original data from PHILVOCS: see Appendix in the attached

CD-ROM for the digitized data).

Creek, which drains from the mountain, one day before the massive landslide (Lagmay et al.,2006). The occurrence of cracks on the slopes, minor landslides, and drying out of stream water were all noticed by the residents, but they did not recognize them as the precursory phenomena of a potentially significant hazard.

3.3 Chronology of events from eye witness accounts

Several eye witness accounts have been reported by Gutierrez (2006). One such interview wasmade with Mr. Virgilio Monghit who saw the slide as it occurred from a location only about 20 mfrom the left side of the scarp (viewed from below the slide). Mr. Monghit owns the land close tothe scarp, which he planted with banana and coconut, and he was working on his land on the dayof the slide. The witness said that a few minutes before the slide, he felt ground shaking from anearthquake. Then he heard a very loud rumbling noise similar to a jet engine. From his vantage

point, he saw the overhanging rock detach from the mountain. He was quite certain that the rock failed by sliding instead of by toppling or overturning.

Once the rock mass started to move, it cut through the two small hills located at the foot of themountain, creating a valley by which debris materials were transported. The witness also indicated that the rock which slid from the overhang and the debris materials from the two hills below themountain created three clusters of rock mass which moved downhill in a wave-like fashion beforedisintegrating and spreading further. Apparently, some large pieces of rocks and boulders wereobserved flying and hopping above the ground.

Another survivor of the slide confirmed this observation. Once the rock mass started to disinte-grate and spread, dust cloud covered and formed above Guinsaugon. This cloud of dust lingered for a few minutes and when the dust settled, the witness saw that the village has completely disappeared

36




Figure 6. (a) Escarpments interpreted to have been formed by old landslide events (after Presidential

Inter-Agency Committee, 2006); (b) Scree deposits on the flanks of the mountains probably as a result of

previous landslides (Photo by DENR-MGB).

below a wide area of soil and rock materials. The rockslide apparently released a huge amount of energy and created an air burst, which knocked down the witness to the ground. Once he fell, theground then heaved and jolted the witness a few inches above the ground surface.

An interview with another survivor was narrated by Biadog (2006). According to eye witnessTony Cabbang, he heard a rumbling noise coming from the mountain while he was weeding the rice paddies. He felt the ground trembling, and when he looked up, he heard a sound like an explosionand saw the top of the mountain came sliding down. There was a great cloud of dust and large wallof earth moving towards him, so he turned and ran away as fast as he could, looking back only onceto check if the wall was near.

3.4 Post-failure behavior

Distinct element modeling of the slide performed by Gutierrez (2006) indicated that the rockslideinitially occurred due to slip or activation of the fault, which is a splay of the PFZ, and the downward movement of an overhanging rock along the fault dip direction. Following slippage along the fault,a vertical shear failure plane was created causing the overhanging rock to be separated from the faceof the mountain. The falling rock then slid along the bedding plane at the base of the overhangingrock, and started to disintegrate to create a rock avalanche and debris flow. The overhanging block experienced almost no rotation, indicating that the block did not initially topple or overturn.

The detached overhanging block disintegrated after it slipped only a few meters along the fault.

The block tended to break more along the bedding and vertical planes, and failure and largeseparations along existing fractures appear to subdivide the falling block into several clusters beforesliding and disintegrating. Initially, the blocks at the top of the scarp moved mainly downwardsalong the dip direction of the fault while those at the bottom tended to move along the bedding

plane. With time, the blocks started to spread laterally although the main flow direction tended to be funneled and followed the small valley at the foot of the mountain.

The sliding materials spread out at the foot of the slope over an area of nearly 3 km2. Such wideexpanse covered by the moving debris can be attributed to the saturated rice field which served aslubricating layer for the portion of the rock avalanche that extended to the valley floor. Based on

37





Figure 7. Extent of the combined scarp and debris area from composite satellite imaging (UNOSAT, 2006).

simulation results, Evans et al. (2007) concluded that the presence of flooded paddy fields in the

valley bottom enhanced the travel distance through decreased frictional resistance at the base of thedebris sheet by undrained loading. Fig. 7 shows the extent of debris deposition based on satellitemapping (UNOSAT, 2006). The debris traveled as far as 3.9 km from the head of the scarp and spread laterally by as much as 1.5 km at the base.

Fig. 8 shows the profile of the center of the slope before and after the landslide, as well asthe thickness of the debris deposit based on LIDAR survey (DPRI, 2006). In Fig. 8a, the pre-slide profile was estimated from the 1:50,000 topographic map of NAMRIA (National Mappingand Resource Information Authority), while the post-slide profile was based on laser scanner measurement of National Research Institute of Fire and Disaster and total station measurement of the Disaster Prevention Research Institute. Based on the measurements, the debris thickness variesfrom 30 m at the foot of the mountain to a few meters close to the debris periphery. Using simple

calculation, the estimated landslide volume is about 20–25 million m3.As mentioned earlier, the horizontal distance from the top of the failure to the front of the debris

was about 3.9 km. To evaluate if such long runout distance is unusual, the fahrböschung (Heim,1932) is calculated. This is defined as the slope of a line connecting the crest of the source area withthe distal tip of the deposits measured on the straightened profile of the path, i.e., φave = arctan( H / L)(see box in Fig. 9). Heim (1932) noted that, should one analyze the motion of a sliding block withconstant frictional resistance from one end of the profile to the other, the frictional coefficientwould theoretically equal tan (φave). Fahrböschung is therefore sometimes called the “equivalentfrictional angle” of the slide. A plot showing the relation between the fahrböschung or equivalentfriction angle and the volume of landslide compiled for various subaerial and submarine slides is

shown in Fig. 9. Also plotted in the figure is the data range estimated for the Ginsaugon slide.It can be surmised that the data obtained for this slide is consistent with those observed in other subaerial slides, although the plot is somewhat in the lower limit.

Based on the accounts from residents of the area regarding the duration and the distance traveled by the landslide, the flow velocity was approximately 100–140 km/h (Lagmay et al., 2006; Evanset al., 2007). The houses along the debris path were believed to have been carried to distances asfar as 550–600 m from their original positions. One survivor interviewed mentioned that he wassleeping in his house when the landslide occurred. He claimed that his house was swept awaytogether with the debris, with himself and his house riding on top of the flowing mass of soils and

38





Figure 8. (a) Profile of the center of the slope before and after the landslide. (after DPRI, 2006); (b) thickness

of deposit based on LIDAR survey (DPRI, 2006).

rocks. Fortunately, he was rescued downslope. Only remnants and parts of the houses at the edge of the debris could still be seen. All of more than 300 houses and buildings in Guinsaugon, includingan elementary school and a church, were destroyed and almost all were completely buried under the debris.

3.5 Detailed features of the slope

Three major slide surfaces were identified on the scarp, which form a complicated wedge structure(see Fig. 10). One of the slide surfaces, and potentially the primary slide surface, is slickensided asevidenced by its smooth surface, which tended to shine like mirror and reflect the sunlight. Initially,it was thought that the slickensided surface was formed by the rockslide itself as evidenced byvertical streaks running from the top to the bottom of the surface. However, a closer investigationof the surface and its morphology revealed that the vertical streaks are formed by fine materials

39





Figure 9. Relation between fahrböschung (or slope angle φave) and volume of landslide (based on original

plot by Towhata, 2000).

Figure 10. Escarpment and interpreted failure planes.

sliding on the face of the slide, and that the surface is smoother along the horizontal direction and undulated along the vertical direction (see Fig. 11). Thus the surface was eventually identified as anexisting strike-slip fault and the undulated surface was deemed to have been formed by gouging dueto the lateral movement along the fault. The fault surface was very hard from calcite mineralizationas verified by Schmidt hammer rebound tests (Gutierrez, 2006).

Having recognized that one of the failure surfaces is an existing fault, it is very likely that theslide was initiated along the fault. There are two possible modes responsible for the failure of the overhanging rock: (1) overturning or toppling; and (2) sliding along the failure surfaces. As

indicated earlier from a witness account, it was concluded that the failure most probably occurred due to the sliding of the rock. This means that the rock mass slid along the existing fault and failed by shear on the vertical surface. One of the faces of the scarp is indicated in Fig. 11. Oncethe rock mass started to move, a third slide surface appears to have been created possibly alongthe bedding planes. However, the full extent and geometry of the failure surfaces and the scarp,and the size of the initial sliding block can not be completely ascertained because part of thescarp is partially filled with debris. Once the rock mass from the overhang started to move, it cutthrough two small hills at the foot of Mt. Can-abag creating a valley by which debris materials weretransported.

40






Figure 11. Close-up view of the surface of the fault forming part of the scarp.

During the f ield investigation in March 2006, the highest seepage was traced at 500 m.a.s.l. but presumably much higher during the rainy season and at the time of the landslide, the highest

seepage level as observed from aerial photos was about 650 m.a.s.l. It is worthy to note that AlihoCreek, which drains the slope that collapsed, originates from the upper reaches of Mt. Can-abagand flows down to the Guinsaugon village. Soil and regolith is thin (<2 m) in areas overlain byvolcaniclastic bedrock but relatively thicker in shear zones where intense fracturing and enhanced weathering have occurred. Secondary forest covers the upper slopes, while the basal slopes aregrown with coconut trees.

3.6 Debris characteristics

The materials deposited at the valley floor consisted of sandstones, conglomerates (sedimentaryrocks), mudstones, and breccias (produced by the movement of the Philippine fault). Some of

the rocks were huge, more than 4 m in diameter, and they were scattered throughout the fan of the deposit as they traveled considerable distance from the mountain source. One large boulder,measuring about 3.5 m in diameter, can be seen at the front end of the deposited debris (see Fig. 12).Apparently, this large boulder traveled a distance of almost 4 km, indicating the fluid-like behavior of the moving debris. A significant portion of the debris is made up of fine-grained soil matrix(including large amounts of silty sand, sandy silt and low plastic clay). This reflects the compositionof the sheared source rock mass and the colluvium entrained from the base of the rock slope as themass bulldozed the flooded paddy fields.

One unusual feature of the debris deposit is its hummocky nature, with several mounds of soilsobserved in many areas (see Fig. 13). These nearly conical mounds vary in height from 0.5 m to

as much as 3 m. Some of the mounds were formed by large boulders being covered by smaller debris materials. Others, however, appear to have been dumped or created from impact with theground. It appears that these mounds were formed from large blocks of conglomerate materials.The mounds were either formed when the conglomerates that were ejected from the slid materialsdisintegrated on impact, or the blocks of conglomerate rolled until they stopped and the subsequentrains disintegrated the rock.

The distal limit of the debris consisted of series of irregular lobes on the flat valley floor, whichare rimmed by a zone of dark mud. In some places, trees and buildings were destroyed by theejected mud beyond the limit of the coarser debris (Lagmay et al., 2006; Evans et al., 2007).

41





Figure 12. A large rock boulder at the front end of the debris flow.

Figure 13. A mound of debris materials possibly created from impact of deposited conglomerate with the

ground.

4 LANDSLIDE TRIGGER

As mentioned earlier, the triggering mechanism that caused the Leyte landslide is not fully under-stood. The intense rainfall could have triggered the slide by increasing ground water inflow,

particularly in faults and fractures, resulting in increased pore pressures in the fault which forms part of the scarp. In return, the increased pore pressure reduced the effective normal stress and reduced the frictional resistance of the fault, thereby causing the fault to slip and be activated. The

presence of several springs at the floor of Mt. Can-abag provides evidence of hydraulic pressur-

ization and connectivity of the faults close to Guinsaugon. In addition to causing potential faultactivation, water inflow from rainfall could have increased the unit weight making the overhangingrock heavier with increasing water saturation. Water could have also reduced the shear strength of the faults, fractures and the intact rock.

As for the role of the earthquake, it is not clear whether it was the final trigger for the landslide or it is the other way around – i.e., the landslide generated this ground tremor. For example residentsclaimed that they felt an earthquake before the massive landslide happened. On the other hand,some researchers (e.g., Suwa et al., 2006) argued that the conditional evidences suggest the higher

probability that tectonic earthquake did not occur, but the landslide generated that ground tremor.

42






A key parameter would be the actual time of occurrence of the landslide vis-à-vis the earthquake.For example, Yamanaka (2006) attempted to back-calculate the time of occurrence of the landslide

based on the recorded displacements from broadband seismograph (F-Net) stations which wereestablished and operated by the National Research Institute for Earth Science and Disaster Preven-tion (NIED) at various locations in Japan. He plotted the displacement time histories based on thedistance of each station from Leyte Island considering a band pass filter ranging from 20–50 sec.

The arrival of the Rayleigh wave is shown in Fig. 14, where it can be seen that the phases of thetime histories line up strikingly as indicated by the arrow. Yamanaka (2006) mentioned that theRayleigh wave caused by the massive landslide was observed in various locations in Japan, whichwere located by as much as 2000–4000 km. From back-calculation, he concluded that from theshape of the waves recorded, the landslide occurred at 10:37 AM, local time, indicating that thelandslide followed the earthquake.

However, the above analysis alone does not solve the problem. As pointed out by Suwa (2006),a minor earthquake with M s = 2.6 or mb = 4.3 is usually associated with fault rupture of at most1sec. Considering that the waveforms involved relative large periods of 20–50 sec, it is more

possible that the ground tremor is not tectonic in nature, but rather the massive landslide caused the

earthquake, which subsequently were recorded in at least three seismic stations in Leyte. It should be mentioned that the magnitudes of the recorded earthquakes are well below those associated withthe occurrence of earthquake-triggered landslides (e.g., Keefer, 1984).

5 CONCLUDING REMARKS

The large-scale landslide in Southern Leyte, which resulted in the death of more than 1,300 people, was caused by a confluence of geologic/tectonic and climatic factors. The rockslide-debrisavalanche originated in rocks within the damaged zone of the Philippine Fault, an active strike-slip

fault that shows a rate of movement in the order of 2.5 cm/yr. The sliding surface of the initialrockslide corresponds to an existing Reidel fault, which is a splay of the PFZ. Because of this,the source rock mass was weak and fragmented because of the shearing and brecciation associated with the fault movement. The failure mechanism also involved shear failure along a vertical planeand sliding along the bedding plane (no toppling or overturning). The presence of high steep slopesconsisting of rock masses that are subject to active tectonic shearing suggests a high potential for large-scale rock slope failure along the PFZ.

The intense rainfall, which increased the pore pressure in the fault, may be responsible for theinitial slip along the fault and the triggering of the landslide. Such heavy rains were associated withthe 2006 La Niña phenomenon. It is interesting to mention that the landslide occurred as a delayed response to a 5-day period of unusually heavy rains. This may have great implications in landslide

warning methodologies based on rainfall intensity.The sorting and distribution of the debris was controlled by topography and surface water con-

dition. The saturated rice paddy directly below the mountain had a major effect on the run-out of the rock avalanche. The distance traveled by the debris was magnified by the decreased frictionalresistance at the base of the debris sheet due to undrained loading, resulting in spreading and thinning of the rockslide-avalanche debris.

Finally, it can be said that the Southern Leyte rockslide is a possibly part of on-going land forming and mass wasting processes along the PFZ in Southern Leyte. The presence of older rockslide-debris avalanche deposits in the valley parallel to the fault indicates the possibility of future catastrophic landslide events in the area and may have potential impact on the people living

at the foot of Mt. Can-abag.

ACKNOWLEDGEMENTS

The first author would like to thank Dr. R. Solidum and Mr. I. Narag of PHILVOCS for providing theseismic waveforms recorded at Maasin station, Leyte. The second author would like to acknowledgefunding provided by the US National Science Foundation under Grant Number 0630474.

43




Figure 14. (a) Locations of F-net stations in Japan; and (b) plot of the waveforms obtained showing the

arrival of Rayleigh wave, as processed by Yamanaka (2006).

44





REFERENCES

Aurelio, M.A. 1992. Tectonique du segment central de al faille Philippine (etude structurale, cinematique et

evolution geodynamique), These de Doctorat, Universite Pierre et Marie Curie, Paris, France, 500pp.

Barrier, R., Huchon, P. and Aurelio, M. 1991. Philippine fault: a key for Philippine kinematics, Geology, 19(1),

32–35.

Bautista, M.L.P. and Oike, K. 2000. Estimation of the magnitudes and epicenters of Philippine historicalearthquakes, Tectonophysics, 317, 137–169.

Besana, G.M. and Ando, M. 2005. The central Philippine Fault Zone: Location of great earthquakes, slow

events and creep activity, Earth, Plants and Space, 57, 987–994.

Biadog, M. 2006. Eye witness account: A large wall of earth moving towards me, The Pacific Connection,

Spring 2006, 8–9.

Cardiel, G. 2006. Personal Communication.

Catane, S.G., Cabria, H.B., Tomarong, C.P., Saturay, R.M., Zarco, M.A.H. and Pioquinto, W.C. 2007.

Catastrophic rockslide-debris avalanche at St. Bernard, Southern Leyte, Philippines, Landslides, 4(1),

85–90.

Disaster Prevention Research Institute, DPRI 2006. Philippine Southern Leyte Landslide Preliminary Report ,

University of Kyoto, 30 pp.Domasig, W.F. 1991. Report on the ground investigation of reported landslides and ground features and other

earthquake-related damages in Cabalian-St. Bernard area in Southern Leyte, Memorandum Report, Mines

and Geosciences Development Service, 6pp.

Evans, S.G., Guthrie, R.H., Roberts, N.J. and Bishop, N.F. 2007. The disastrous 17 February 2006 rockslide-

debris avalanche on Leyte Island, Philippines: a catastrophic landslide in tropical mountain terrain, Natural

Hazards Earth System Science, 7, 89–101.

Geographical Survey Institute, GSI 2006. The position of the landslide in Leyte Island, Philippines as estimated

from Space Shuttle data, http://www.gsi.go.jp (in Japanese).

Gutierrez, M. 2006. DEM simulation of a massive rockslide, Special Lecture, International Symposium on the

Geomechanics and Geotechnics of Granular Media, Ube, Japan, 13pp.

Heim A. 1932. Landslides and Human Lives (Bergstürz und Menschen leben), N. Skerner (Translator), Bi-TechPublishers, Vancouver.

Hungr, O., Evans, S.G., Bovis, M.J. and Hutchinson, J.N. 2001. A review of the classification of landslides of

the flow type, Environmental and Engineering Geoscience, 7, 221–238.

Keefer, D.K. 1984. Landslides caused by earthquakes, Geological Society of America Bulletin, 95, 406–421.

Lagmay, A.M.A., Ong, J.B.T., Fernandez, D.F., Lapus, M.R., Rodolfo, R.S., Tengonciang, A.P., Soria, J.L.,

Baliatan, E.G., Quimba, Z.P., Uichanco, C.L., Paguican, E.R., Remedio, A.R.C., Lorenzo, G.R.H., Avila,

F.B. and Valdivia, W. 2006. Scientists investigate recent Philippine landslide, EOS, Transactions American

Geophysical Union, 87(12), 121–124.

Lanuza, A.G., Chu, A.V., Mangao, E.A., Soneja, D.S., Sanez, R. and Garcia, D.C. 1994. Aftershock observation

of 05 July 1994 earthquake in Cabalian area, Southern Leyte, Philippine Institute of Volcanology and

Seismology (PHILVOCS) Internal Report , 35pp.

Narag, I. 2007. Personal Communication.

Orense, R. and Sapuay, S. 2006. Preliminary Report on the 17 February 2006 Leyte, Philippines landslide,

Soils and Foundations, 46 (5), 685–693.

Philippine Institute of Volcanology and Seismology, PHILVOCS 2006. Earthquake Bulletin No. 2,

http://www.phivolcs. dost.gov.ph/Earthquake/LatestEQ/

Presidential Inter-Agency Committee 2006. The 17 February 2006 Bgy. Guinsaugon, Southern Leyte

Landslide, Unpublished Report , March 3, 2006, 30pp.

Sajona, F.G., Bellon, H., Maury, R.C., Pubellier, M., Quebral, R.D., Cotton, J., Bayon, F.E., Pagado, E. and

Pamatian, P. 1997.Tertiary and quaternary magmatism in Mindanao and Leyte (Philippines): geochronology,

geochemistry and tectonic setting, J. Asian Earth Science, 15, 121–153.

Suwa, H. 2006. Catastrophe caused by the 17 February 2006 Southern Leyte landslide in Philippine, J. Japan

Society for Natural Disaster S cience, 25–1, 83–97 (in Japanese).Towhata. I. 2000. Lecture Notes in Soil Dynamics, Department of Civil Engineering, University of Tokyo.

United States Geologic Survey, USGS 2006. Preliminary Earthquake Report: Magnitude 4.3 Leyte,

Philippines, http://neic.usgs.gov/neis/bulletin/neic_jgdn.html

UNOSAT 2006. Leyte Island – Village of Guinsaugon – Pre-Disaster Image (landslide included),

http://unosat.web.cern.ch/unosat/asp/prod_free.asp?id =46

Yamanaka, Y. 2006. The Landslide in Leyte Island Philippines on 17 February 2006, EIC Seismo-note Extra

Chapter , http://www.eri.u-tokyo.ac.jp/sanchu/Seismo_Note/2006/060217.html

45


http://www.gsi.go.jp/

http://www.phivolcs.dost.gov.ph/

http://neic.usgs.gov/

http://unosat.web.cern.ch/



http://www.eri.u-tokyo.ac.jp/













Slope failures during the 2004 Niigataken Chuetsu

earthquake in Japan

T. Kokusho & T. IshizawaCivil Engineering Department, Chuo University, Tokyo, Japan

T. HaraCivil Engineering Department, Wakayama College of Technology, Wakayama, Japan

ABSTRACT: The Niigata-ken Chuetsu earthquake ( M J=

6.8) caused more than 4000 slope fail-ures in the area about 200 km north of Tokyo. The slope displacements in the major failures have been quantified by using the DEM technique, which may be able to serve as valuable case historiesfor evaluating seismically induced slope failures from the viewpoint of the Performance-Based Design. In this paper, six representative slope failures of different failure modes are explained in detail. Three-dimensional changes of slopes due to the earthquake are delineated not only by

photographs, charts of plans and cross-sections but also by 3-dimensional digital image stored inthe CD-Rom together with geotechnical and seismic data.

1 INTRODUCTION

The Niigata-ken Chuetsu earthquake (MJ = 6.8) occurred on October 23, 2004, which caused more than 4000 slope failures in the area about 200 km north of Tokyo (Fig. 1) due to the mainshock and also several strong aftershocks. Out of the great number of slope failures during the2004 earthquake, 282 failures exceeded 104 m3 and 10 exceeded 105 m3 in terms of the affected areas. The failed soil mass flowed in a distance of several tens or hundreds of meters, with seriousconsequences to human lives, houses, roads, agricultural facilities and lifelines. It also blocked streams and made many natural reservoirs. During historical earthquakes in the past 200 years,similar slope disasters accompanying many landslides in the green-tuff soft rock areas spreadingover the northern to middle part of the main island of Japan had occurred once in every 25–30years on average (Japan Society for Civil Engineers. 2007).

In this paper, the soil movements in the major failures have been quantified by using the DEM(Digital Elevation Modeling) technique, which may be able to serve as valuable case histories

Figure 1. Area damaged during 2004 Niigataken Chuetsu earthquake.

47





for evaluating seismically induced slope failures from the viewpoint of the Performance-Based Design.

With that objective, general backgrounds of the slopes are first addressed to discuss variousfactors influencing failure modes, including geology and earthquake data. Then, six representativeslope failures of different failure modes are explained in detail. Three-dimensional changes of slopes before and after the earthquake are delineated not only by photographs, charts of plans and

cross-sections but also by 3-dimensional digital image stored in the accompanying CD-Rom. Inorder to facilitate better understanding the background of the failures, pertinent data concerninggeology, the seismic conditions, borehole logging and soil properties are also provided.

2 GENERAL BACKGROUNDS OF SLOPE FAILURES

2.1 Geological and geotechnical setting

The geological map of the area where the most of the slope failures occurred during the earthquake

is shown in Fig. 2. The major geology there consists of Neogene sedimented rocks, sand stonesand mudstones. It is known as a landslide-prone area of green-tuff, with geological structures of active folding which cover active faults underneath. Synclines and anticlines are running parallelin the north-south direction as indicated in Fig. 2, among which rivers are flowing almost in thesame direction from north to south. Mountains are about 500 m at the highest, and the slopesare composed of weak sedimented rock, alternating layers of strongly weathered sandstones and mudstones. Bedding planes or dip planes have a strong effect on the natural slopes here, causing

Figure 2. Geological map in the area of slope failures (Japan Geological Survey).

48





non-symmetric slope angles on the two sides of the mountain ridges; namely, slopes are gentle indipping directions while obviously much steeper on infacing directions.

In most of the slope failures, sand stones were highly responsible mainly because of their weakness due to strong weathering. Fig. 3 indicates unconfined compression test results on typicalsand stones compared with those of mudstones sampled from several locations in the area of slopefailures. The strengths of sandstones are much smaller than those of interbedded mud stones. Fig. 4

shows the grain-size curves of the disintegrated sand stones, which indicate that the sands are poorly graded and site-dependent difference in particle size distributions is almost ignorable. Itshould also be noted that the sandstones had higher permeability of the order of 10−3 cm/s than thatof mudstones of the order of 10−4

∼10−6 (JSCE Report 2005). It served as aquifer and may haveapplied excess pore-pressure, leading to the reduction of slope stability during the earthquake.

According to the pre-earthquake precipitation records in Nagaoka (a city about 16km northwestof the epicenter), rainfalls in three days prior to the quake were 120 mm, which may have had someinfluence on the seismic instabilities during the earthquake.

2.2 Earthquake data

Fig. 5 shows a map of the damaged area in which a great number of failed slopes are indicated withred spots. In the map, the epicenters of the main shock (No.1) and the aftershocks (No.2-5) with themagnitudes larger than M J = 6.0 are also shown. Many aftershocks of large magnitudes occurred not only immediately but also even a few days after the main shock; after the main shock of M J = 6.8at 17:56 Oct. 23, three aftershocks exceeding M J = 6.0 occurred in less than 1 hour and one with

M J = 6.1 occurred 4 days after the main shock. They might possibly cause some delayed additionalfailures. The specifications of the main shock and major aftershocks larger than M J = 6.0 are listed in Table 1.

The fault activated during the earthquake was not clearly identified at the ground surface,although it was a reverse thrust fault with the strike of NNE-SSW direction and westward diprunning at the east side of the damaged area.

There were several earthquake observation stations near the damaged area deployed by JMA(Japan Meteorological Agency) and NIED (National Research Institute for Earth Science and Disaster Prevention). Both systems consist of a single 3-dimensional accelerometer on the ground surface. The locations of the JMA and K-net stations are indicated in Fig. 5, and their longitudes,latitudes and elevations are listed in Table 2. Some of the accelerograms recorded during the mainshock at Yamakoshi, Ojiya and Kawaguchi by the JMA system are shown for the EW directionin Fig. 6. Among them, Yamakoshi site was located in the midst of the slope failure area. TheP/S-logging profile there is listed in Table 3.

Figure 3. Stress-strain curves of mudstone and

sandstone by unconfined compression tests.

Figure 4. Grain size distributions of sandstone from

different slopes.

49






Figure 5. Epicenters of the main shock and major aftershocks during the 2004 Niigataken Chuetsu earthquake

(Geographical Survey Institute).

Table 1. Specifications of the main shock and aftershocks larger than M J = 6.0 of 2004 Niigataken Chuetsu

earthquake.

Epicenter (degree)MJ: JMA Occurrence

Earthquake magnitude (y/m /d/time) Latitude (N) Longitude (E) Depth (D) (km)

Main shock 6.8 2004/10/23/17:56 37.29 138.87 13

After shock 6.3 2004/10/23/18:03 37.35 138.98 9

After shock 6.0 2004/10/23/18:12 37.25 138.83 12After shock 6.5 2004/10/23/18:34 37.31 138.93 14

After shock 6.1 2004/10/27/10:40 37.29 139.03 12

Table 2. JMA and K-net strong motion recording stations during 2004 Niigataken Chuetsu earthquake.

MS Epicenter Max. Acc

System Site (Code) Latitude (deg) Longitude (deg) distance (km) (EW) gal

JMA Yamakoshi 37.327 138.890 4.3 723

JMA Ojiya 37.310 138.795 7.0 898

JMA Kawaguchi 37.268 138.861 2.5 1676K-net Ojiya (NIG 019) 37.306 138.790 7.4 1308

The digitized acceleration time histories of the main shock recorded in the damaged areasincluding the sites in Table 2 are available at websites (JMA and K-net). For the K-net system,accelerograms of not only the main shock but also major aftershocks, the soil profiles and theP/S-wave logging results of the recording sites are available in the same website.

50





Figure 6. Acceleration time histories in EW direction recorded at 3 near-field stations.

Table 3. P/S logging profile at Yamakoshi site (after JSCE 2007).

Depth (m) Vs (m/s) Vp (m/s) Density (t/m3) Soil type

0–5 110 300 1.8 Fine sand

5–10 200 690 1.8 Sandy gravel10–24 260 1700 1.7 Weathered mudstone

24–33 310 1700 1.9 Slightly weathered mudstone

33–38 590 2100 1.9 Sandstone/mudstone

Figure 7. Schematic image of 3 types of slope

failures, A, B, and C, occurred during 2004

Niigataken Chuetsu Earthquake. Figure 8. Horizontal slope displacements versusslope inclination for 3 types of slope failures, A,

B, and C, occurred during 2004 Niigataken Chuetsu

Earthquake.

2.3 Classifications of slope failures

The slope failures due to this particular earthquake are classified into 3 types as schematicallyillustrated in Fig. 7;

Type-A: Deep slips parallel to sedimentation planes (parallel dip slip or daylight dip slip), in gentleslopes of around 20 degrees. In many cases, displaced soil mass had originally been destabilized

by river erosions or road constructions at the slope toes, and slipped almost as a rigid body alongthe slip plane. The displaced soil volumes were very large, translating ground surface with littledisturbance.

Type-B: Shallow slips not parallel to sedimentation planes (infacing dip slip) at steep slopes (>30degrees). This type far outnumbered Type-A, but the individual soil volume was not so large.Soils normally fell down in pieces, sometimes leaving trees with deep roots at original places.

51







Table 4. Locations of slopes studied here and their input earthquake energies during main shock and

aftershocks.

Incident energy at baserock EIP (KJ/m2)Latitude Longitude

North East Main shock Aftershock Aftershock Aftershock Aftershock

Slope Type (degree) (degree) MJ = 6.8 MJ = 6.3 MJ = 6.0 MJ = 6.5 MJ = 6.1

Higashi-Takezawa A 37.304 138.906 428.6 91.7 21.0 143.0 27.4

Yokowatashi A 37.330 138.827 382.4 50.8 22.4 103.1 14.8

Haguro-tunnel B 37.329 138.895 398.3 95.9 18.9 134.2 23.9

Entrance

Naranoki B 37.322 138.912 393.8 107.8 18.3 139.7 26.6

Kajigane C 37.309 138.899 430.8 89.9 21.2 141.0 25.2

Musikame C 37.349 138.886 356.1 91.3 16.7 121.1 20.7

in which M is the Richter’s Magnitude, although M J = 6.8 ( M J is Japanese Earthquake Magnitudealmost equivalent to the Richter’s Magnitude) is used in the computation. The comparison of the computed values with measured energies from vertical array records near epicentral areasfor the 1995 Kobe earthquake (Kokusho and Ishizawa 2007) and also for the 2004 Niigataken-Chuetsu earthquake (Kokusho et al. 2008) has already confirmed that these simple equations canapproximate the input seismic energy satisfactorily for engineering purposes. Table 4 indicates that,for all the slopes, the energy of the main shock is much larger than the largest aftershock, which isabout 1/3 of the main shock.

In all the slope failures, ground surface elevations before and after the earthquake are compared to quantify the 3-dimensional topographical changes. The post-earthquake elevations were obtained

by DEM data based on the air-borne laser survey carried out on 28 October, 2004, 5 days after the earthquake, when 4 major aftershocks with the magnitude larger than M J = 6.0 had alreadyoccurred. Due to the absence of similar data just before the earthquake, air-photographs takenin 1975 and 1976 were used (JSCE Report 2006). The maximum potential error involved in the

post-earthquake elevations is ±0.5 m, while that of pre-earthquake elevations is ±1.0 m.The topographical changes thus evaluated may reflect not only the effect of slope failures but

also two more influencing factors. The first is the tectonic movement due to the earthquake fault.However, it was not detected so clearly between the two elevation contours developed in the twotime sections for all of the failed slopes studied here, probably because the tectonically-induced elevation change was too small to be differentiated from the errors involved in the data analyses.

The second is topographical changes which may have occurred during the 28–29 years before theearthquake. It may well be estimated that the major changes in this time interval, much shorter than the geological time scale, are due to human activities such as construction of local roads and agricultural facilities, etc., being ignorable in most cases because of their scales far smaller thanthose of the 6 large slope failures.

The digitized data of pre/post-earthquake elevations for the 6 slopes are stored in the CD-ROMof this volume. The horizontal and vertical coordinates are taken as the Global Coordinate VIII. Thecoordinates (x: NS, y: EW, z: UD) before and after the earthquake are given in meter at all nodes of 1 m square meshes covering the areas including the 6 failed slopes. The coordinates of the referencenodes are given for each slope. Furthermore, in order to know the change of cross-sections of thefailed slopes, the coordinates (x, y, z) of surface points of every 0.5 m apart along the length of 21

equidistant lines parallel to the sliding directions are also given in the CD-ROM.

3.1 Higashi-Takezawa (HTZ) slide, (Type-A)

One of the typical Type-A failures during the earthquake was Higashi-Takezawa slide, which blocked a river and resulted in a large reservoir in the upstream. The geology there is interbedded sandstone and mudstone of the Shiroiwa Formation of Pliocene in late Neogene. Field observationindicated that a huge block of sandstone with horizontal dimension of about 300 m by 250 m slid on a smooth slip plane of mudstone. The rocks were so much weathered that the sandstone was

53




Figure 10. Photograph of Higashi-Takezawa slide

(Displaced soil block seen from top of scar stopped

the river on the left side).

Figure 11. Photograph of exposed mudstone slip

surface of about 20 degrees with some sand debris

left (water was running on the surface).

Figure 12. Contour map (a) and air-photograph (b) of Higashi-Takezawa slide before the earthquake (based

on photographs taken in 1975 and 1976) (JSCE 2007).

actually medium-dense sandy soil with minimal cementation. In the upper side of the present slide,older scarps could be recognized, indicating that this slope had experienced landslides repeatedlyin the past.

Photograph in Fig. 10 taken from the scarp shows the displaced brownish sandstone slid almostas a rigid body along the slip surface. On the top of the sliding body, cider trees had been standingupright and did not show any evidence of disturbance just after the failure. Terraced rice fieldsand koi-ponds in front of the brownish sliding body was pushed beyond the river and climbed

up the other side of the valley as indicated by the soil mass of whitish color in the photograph.Photograph in Fig. 11 shows the exposed slip surface of greenish mudstone of about 20 degrees,on which slickenside could be seen. Running water was actually witnessed on the mudstone slip

plane after the failure. No lubricating soft seam which could serve as a slip plane was observed on the mudstone surface. It is highly probable that the mudstone served as impervious layer and ground water submerged the slip plane during the failure.

Fig. 12(a) depicts a contour map of Higashi-Takezawa slope before the slide and Fig. 12(b)shows the corresponding airphotograph. The elevation step of each contour is 2 m. Fig. 13(a)depicts the contour map of the same slope after the failure (28 October, 2004) and Fig. 13(b) is

54







Figure 13. Contour map (a) and air-photograph (b) of Higashi-Takezawa slide after the earthquake (based

on photographs taken in 28, Oct. 2004) (JSCE 2007).

the corresponding photograph. In Figs. 12(a) and 13(a), the area affected by the slope failure issurrounded by a thick curve. The periphery of the affected area was determined from the two DEMdata so that the elevation difference between the two exceeds 1 m.

In Figs. 12 and 13, thin parallel lines are drawn toward the direction of the slope failure in order to evaluate cross-sectional change of the slope. There are 21 thin lines 15 m apart, among which 6

thick lines are 45 m apart to each other.The cross-sectional changes in the affected area are developed as illustrated in Fig. 14 from the

two DEM data before and after the earthquake along 6 thick parallel lines in Figs. 12(a) and 13(a).The slip surface shown with the dotted curve in Fig. 14, though difficult to locate very precisely,was determined from the exposed slip plane in the upslope side, the original location of the valleyand from the global changes of slope configuration.

Based on the cross-sectional changes, the horizontal area was calculated as 64800 m2 and 73100 m2, and the total volume of the sliding soil mass thus calculated 3-dimensionally was881000 m3 and 1190000 m3 before and after the failure, respectively. If the displaced soil massis idealized by a rectangular block as in Fig. 15, the thickness of the block is 13.6 m and 16.8 m

before and after the failure (Kokusho and Ishizawa 2008). The sliding movement of the center of

the block is 20.7 m vertically and 94.1 m horizontally.In the CD-Rom of this volume, the coordinates (x: NS, y: EW, z: UD) before and after the

earthquake are given in meters at all nodes of 1 m square meshes covering the rectangular areaincluding the failed slope. The coordinate of the starting point R1 (the bottom left corner of therectangle) is x= 35601 and y= 144501 and that of the end R2 (the top right corner) is x = 36299and y= 145199 by the Global Coordinate VIII. The coordinates of the grids are arrayed along eachhorizontal line from west to east and then the next line up to the north.

Furthermore, along 21 equidistant straight lines parallel to the sliding directions, the originalcoordinates (x, y, z) and the elevation change z due to the slope failure at surface points of every0.5 m apart in the sliding direction are also given. Presumably because of the errors involved and

some other reasons, even the nodes outside the failed slope show some non-zero elevation changes.The coordinates inside the periphery of the affected area is marked to differentiate them fromthe outside nodes. Thus, detailed topographical changes can be drawn from the digital data in theCD-ROM.

In Figs. 12(a) and 13(a), locations of 4 bore-holes, HB-9, 1, 2 and 5, are plotted along a line. The bore-hole survey was carried out by the Office of Yuzawa Sabo, a branch of the Ministry of Land,Infrastructure andTransport as a part of the post-earthquake restoration work. Fig. 16 illustrates thecross-section of the slope along the line of the bore-holes simplified from a detailed chart provided

by Yuzawa Sabo. Pre/Post-failure cross-sections of the slope surface and estimated slip surface are

55





Figure 14. Cross-sectional change of Higa

shi-Takezawa slope before and after the earthquake.

Figure 15. Idealization of sliding soil mass by a

rectangular block in Higashi-Takezawa.

Figure 16. Cross-section of Higashi-Takezawa slope along (simplified afterYuzawa-Sabo, Ministry of Land,

Infrastructure and Transport).

also illustrated. At the bottom of the Fig. 16, soil logging results at the 4 bore-holes obtained beforethe major restoration works are shown, in which soil types, water tables, estimated depths of theslip plane are indicated. A part of the prefailure surface in the down-slope side is not availablein this chart, and the post-failure surface reflects some initial restoration works. Nevertheless, itis readily understood that the 300-meters long soil block translated (along the slip plane of themudstone of 20 degrees on average) by 100 m horizontally and the front deformed and collided

56







Figure 17. Scarp of the slide consisting of weath-

ered sandstone (top) and thin-wall sampling of sand

(bottom).

Figure 18. Grain size distribution of weathered

sandstone for triaxial tests.

with the other side of the valley. It is interesting to note that the forest trees were inclined in thefront portion but stayed upright in the back portion where the almost straight slip plane of about

20 degrees underlay. The slip plane determined from the bore-hole survey is superposed in the 4thchart from the top in Fig. 14, which indicates a good agreement with that estimated from the DEManalysis.

Intact samples were recovered from remained weathered sandstone at the scarp after the failure by pushing thin wall tubes and curving the surrounding soil as shown by the photographs in Fig. 17.Undrained cyclic triaxial tests for the specimen size 5 cm in diameter and 10 cm in height werecarried out under the effective isotropic confining stress of 49 kPa and the backpressure 294 kPa(JSCE 2006). There were two types of sands; HTZ-A and HTZ-B with different fines content asindicated in Fig.18, though the grain size curves are basically the same as those in Fig. 4.

Fig. 19 shows the relationship between the number of loading cycles and the stress ratio for 5%double amplitude axial strain for the two types of sands under fully saturated or partially saturated

(natural water content) condition. Obviously, the liquefaction strength decreases considerably withthe increase of saturation to 100%. Fig. 20 depicts axial stress/porepressure versus axial straincurves obtained in undrained monotonic compression tests carried out after the specimen attained about 10% axial double amplitude strain during the cyclic loading tests. It is noted that saturated sand shows higher post-liquefaction residual stress because of the positive dilatancy at larger axialstrain than unsaturated sand.

3.2 Yokowatashi (YWS) slide, (Type-A)

Yokowatashi slide is another typical Type-A slope failure, which occurred in dip slope consisting

overwhelmingly of mudstone of the Shiroiwa formation of Pliocene. Fig. 21 shows the photographof mudstone slip surface of about 24 degrees along which overlying mudstone of about 2.5 m thick on average slid almost as a rigid block. The displaced soil mass which disintegrated into large

pieces covered the road and arrived at a ditch in front.Fig. 22 shows the zoom-up image of a vertical rock face (indicated by a square in Fig. 21)

near the slip plane exposed after the failure on the left side of the slide. A continuous seam of non-plastic silt of 1–2 cm thick was sandwiched between brownish weathered mudstone. Outsidethe weathered mudstone of about 20 cm thick, grayish fresh mudstone dominated. It is no doubtthat the seam served not only as a permeable layer in the low-permeable mudstone but also as a slip

57






Figure 19. Stress ratio R for 5% DA axial strainversus Number of cycles N for intact weathered

sandstones.

Figure 20. Post-liquefaction stress-strain curves of

intact weathered sandstones.

Figure 21. Yokowatashi slide of TypeA with smooth

mudstone slip plane of 24 degrees. The zoom-up of the slip plane (the small square) is shown in Fig. 22.

Figure 22. Non-plastic silt seam of 1–2 cm thick

continuously sandwiched by brownish weathered and grey fresh mudstones.

plane. The grain size curve for YKW included in Fig. 9(a) indicates the mean grain size of the seam D50 = 0.059 mm, the uniformity coefficient C u = 12 and the fines content F c = 58%. Soil particledensity of the seam ρs = 2.50 g/cm3, natural water content wn = 28%, liquid limit wLL = 40% and

plastic limit wPLis not measurable.Fig. 23(a) depicts a contour map of Yokowatashi slope before the slide and Fig. 23(b) shows

the corresponding airphotograph. The elevation step of the contours is 2 m. Fig. 24(a) depicts the

contour map of the same slope after the failure (28 October, 2004) and Fig. 24(b) is the corre-sponding photograph. In Figs. 23(a) and 24(a), the area affected by the slope failure is surrounded by a thick curve. The periphery of the affected area was determined from the two DEM data so thatthe elevation difference between the two exceeds 1 m.

In Figs. 23 and 24, thin parallel lines are drawn toward the direction of the slope failure in order to evaluate cross-sectional change of the slope. There are 21 thin lines 2 m apart, among which 4thick lines are 10 m apart to each other.

The cross-sectional changes in the affected area are developed as illustrated in Fig. 25 from thetwo DEM data before and after the earthquake along the 4 thick parallel lines. The dotted line is

58








Figure 23. Contour map (a) and air-photograph (b) of Yokowatashi slide after the earthquake (based on

photographs taken in 28, Oct. 2004) (JSCE 2007).

Figure 24. Contour map (a) and air-photograph (b) of Yokowatashi slide before the earthquake (based on

photographs taken in 1975 and 1976) (JSCE 2007).

the slip surface corresponding to the mudstone dip plane of 24 degrees. It was easy to locate theslip surface because of its smoothness with the constant slope angle. Fig. 25 indicates that the toeof the sliding block was open to air (daylighting slip) and this slope was not so stable even before

the earthquake. The thickness of the sliding block here was relatively thin and tectonic movementduring the seismic event might result in a measurable error in slope failure evaluation. However,in Fig. 25 the ground surface outside the affected area almost coincides to each other if compared

before and after the earthquake, indicating that the tectonic effect is almost negligible.The horizontal area was 3450 m2 and 3740 m2 and the total volume of the sliding soil mass

thus calculated from the 3D DEM image is 8600 m3 and 10500 m3 before and after the failure,respectively. If the displaced soil mass is idealized by a rectangular block as in Fig. 26, the thicknessof the block is 2.5 m and 2.8 m before and after the failure. The sliding movement of the center of the block is 21.7 m vertically and 61.8 m horizontally.

59






Figure 25. Cross-sectional change of Yokowatashi

slope before and after the earthquake.


rectangular block in Yokowatashi.

In the CD-Rom of this volume, the coordinates (x: NS, y: EW, z: UD) before and after the

earthquake are given in meters at all nodes of 1 m square meshes covering the rectangular areaincluding the failed slope. The coordinate of the starting point (the bottom left corner of therectangle) is x= 28801 and y= 147501 and that of the end (the top right corner) is x= 29099 and y= 147799 by the Global Coordinate VIII. The coordinates of all the grids are arrayed along eachhorizontal line from west to east and then the next line up to north.

Furthermore, along 21 equidistant straight lines parallel to the sliding directions, the originalcoordinates (x, y, z) and the elevation change z due to the slope failure at surface points of every 0.5 m apart in the sliding direction are also given. Because of the errors involved and someother reasons, even the nodes outside the failed slope show some non-zero elevation changes. Thecoordinates inside the periphery of the affected area is marked to differentiate them from the outside

nodes. Thus, detailed topographical changes due to the slope failure can be drawn from the digitaldata in the CD-ROM.

3.3 Haguro tunnel entrance (HGT) slide, (Type-B)

This slide of Type-B occurred at the west entrance of HaguroTunnel. Fig. 27 shows a distant view of the slide. The debris affected local roads, tunnel entrance facilities and houses in the downslope areathough it did not develop as high-speed mudflow. The geology there consists of Araya-formationof Miocene, dark gray mudstone interbedded with fine-grained sandstone. In contrast to Type-Aslide, the sliding soil mass disintegrated into pieces and slid down along the slip plane crossingthe dip plane (infacing slip). Soil particle density of sampled debris= 2.67 g/cm3, fines content

F c=

52%, natural water content wn=

60%, liquid limit wLL=

50% and the plastic limit wPL=

non-measurable indicating that it includes large sand fraction. Fig. 28 shows the zoom-up of the slideindicating that rock fragments slid down to the middle of the slope height leaving a clear scarp inthe top part.

Fig. 29(a) depicts a contour map before the slide and Fig. 29(b) shows the corresponding aerial photograph. The elevation step of the contours is 2 m. Fig. 30(a) depicts the contour map of thesame slope after the failure and Fig. 30(b) is the corresponding photograph. The thick curve inFigs. 29(a) and 30(a) indicates the periphery of the affected area, which was determined from thetwo DEM data so that the elevation difference between the two exceeds 1 m.

60






Figure 27. Haguro Tunnel entrance slide of Type-B

where debris of which affected local roads, tunnel

entrance facilities and houses in the downslope area.

Figure 28. Zoom up of sliding soil mass disinte-

grated into pieces and slid down leaving a clear scarp

in the top part.

Figure 29. Contour map (a) and air-photograph (b) of Haguro Tunnel entrance slope before the earthquake

(based on photographs taken in 1975 and 1976) (JSCE 2007).

In Figs. 29 and 30, thin parallel lines are drawn toward the direction of the slope failure in order to evaluate cross-sectional change of the slope. There are 21 thin lines 12 m apart, among which4 thick lines are 24 m apart to each other.

The cross-sectional changes in the affected area before and after the earthquake are developed asillustrated in Fig. 31 from the two DEM data along 4 thick parallel lines. The dotted line in Fig. 31is the slip surface estimated from the global changes of slope configuration, which may involvemore uncertainties than slope surface topography.

The horizontal area is 10200 m

2

and 21900 m

2

, and the total volume of the sliding soil massthus calculated from the 3D DEM image is 83500 m3 and 88000 m3 before and after the failure,respectively. If the displaced soil mass is idealized by a rectangular block as in Fig. 32, the thicknessof the block is 8.2 m and 4.5 m before and after the failure. The sliding movement of the center of the block is 50.7 m vertically and 111 m horizontally.

In the CD-ROM of this volume, the coordinates (x: NS, y: EW, z: UD) before and after theearthquake are given in meters at all nodes of 1 m square meshes covering the rectangular areaincluding the failed slope. The coordinate of the starting point R1 (the bottom left corner of therectangle) is x= 34701 and y= 147101 and that of the end R2 (the top right corner) is x = 35399

61







Figure 33. Global view of Naranoki slide of Type-B

where the run-out debris blocked the river and local

roads.

Figure 34. Zoom-up of the steep scarp of sedimen-

tation planes of mudstone and sandstone along which

disintegrated soil mass slid down.

Figure 35. Contour map (a) and air-photograph (b) of Naranoki slide before the earthquake (based on


3.4 Naranoki (NRK) slide, (Type-B)

Naranoki slide of Type-B occurred at the steep slope of the right bank of Imo river. The geology wasthe Kawaguchi Formation of Pliocene consisting of sandy mudstone interbedded with sandstone.Fig. 33 shows the global view of the slide, the run-out debris of which blocked the river and local roads there. Fig. 34 shows the close-up view of the slide watching the steep scarp on whichsedimentation planes of mudstone and sandstone can be clearly seen. Thin surface layer slid downalong the steep slip plane of infacing dip and disintegrated into pieces. The grain size curves

investigated for the debris soils are shown in Fig. 9(a). The fines content of the interbedded sandstone was 10–14%, the mean grain size was 0.1–0.12 mm, and C u = 11. Soil particle densityof sampled debris= 2.62 g/cm3, fines content F c = 40%, natural water content wn = 38%, liquid limit wLL = 29% and plastic limit wPL = non-measurable.

Fig. 35(a) depicts the contour map before the slide and Fig. 35(b) shows the correspondingairphotograph. The elevation step of the contours is 2 m. Fig. 36(a) depicts the contour map of thesame slope after the failure (28 October, 2004) and Fig. 36(b) is the corresponding photograph.The thick curve in Figs. 35(a) and 36(a) indicates the periphery of the affected area, which wasdetermined from the two DEM data so that the elevation difference between the two exceeds 1 m.

63







Figure 36. Contour map (a) and air-photograph (b) of Naranoki slide after the earthquake (based on


In Figs. 29 and 30, thin parallel lines are drawn toward the direction of the slope failure in order to evaluate cross-sectional change of the slope. They are 21 thin lines 20 m apart, among which5 thick lines are 60 m apart to each other. The cross-sectional changes in the affected area beforeand after the earthquake are developed as illustrated in Fig. 35 from the two DEM data along the 5thick parallel lines. The dotted line in Fig. 35 is the slip surface estimated from the global changes

of slope configuration and also from field observation.The horizontal area is 42760 m2 and 33600 m2 and the total volume of the sliding soil mass

thus calculated from the 3D DEM image is 45050 m3 and 54300 m3 before and after the failure,respectively. If the displaced soil mass is idealized by a rectangular block as in Fig. 36, the thicknessof the block is 1.1 m and 1.6 m before and after the failure. The sliding movement of the center of the block is as high as 86.8 m vertically and 86.5 m horizontally.

In the CD-ROM of this volume, the coordinates (x: NS, y: EW, z: UD) before and after theearthquake are given in meters at all nodes of 1 m square meshes covering the rectangular areaincluding the failed slope. The coordinate of the starting point R1 (the bottom left corner of therectangle) is x= 36101 and y=146201 and that of the end R2 (the top right corner) is x = 36799and y=146899 by the Global Coordinate VIII. The coordinates of all the grids are arrayed along

each horizontal line from west to east and then the next line up to north.Furthermore, along 21 equidistant straight lines parallel to the sliding directions, the original

coordinates (x, y, z) and elevation change z due to the slope failure at surface points of every0.5 m apart are also given. Because of the errors involved and some other reasons, even the nodesoutside the failed slope show some non-zero elevation changes. In order to differentiate the out-side nodes, the coordinates inside the periphery of the affected area is marked. Thus, detailed topographical changes can be drawn from the digital data in the CD-ROM.

3.5 Kajigane (KJG) slide (Type-C)

Kajigane slide of Type-C occurred just next to the Kajigane syncline. Fig. 39 shows the upslopeview of the slide, where the slide was triggered by some kind of effect of a koi pond. The geologythere consists of Wanazu Formation of Pliocene sandstone, but highly weathered and cultivated as

paddy field and ponds. The disintegrated sliding soil mass slid down along the parallel dip planeof presumably mudstone. Fig. 40 shows the lower part of the slide indicating that muddy debriscrossed a road and spread down over a river valley, blocking the stream. The grain size curve shownin Fig. 9(a) indicates that the debris soil was well-graded with D50 = 0.3 mm, Cu > 500 and thefines content of about 28%. Soil particle density of sampled debris= 2.65 g/cm3, natural water content wn = 29%, liquid limit wLL = 47% and plastic limit wPL = 35% and I p = 12.

64





300 200 100 0100

200

30007–07

Horizontal distance (m)

100

200

300

100

200

300

11–11

A l t i t u d e ( m )

100

200

300

100

200

300

13–13

15'-15

Before After Slip plane

09–09

Figure 43. Cross-sectional change of Kajigane


18.3°

200.0

130.0

2.5627.1°

120.0°

90.0

5.33

∆h = 51.3

(δr)av = 164.3

17.4°


rectangular block in Kajigane slope.

Furthermore, along 21 equidistant straight lines parallel to the sliding directions, the originalcoordinates (x, y, z) and elevation change z due to the slope failure at surface points of every0.5 m apart are also given. Because of the errors involved and some other reasons, even the nodesoutside the failed slope show some non-zero elevation changes. In order to differentiate the out-side nodes, the coordinates inside the periphery of the affected area is marked. Thus, detailed topographical changes can be drawn from the digital data in the CD-ROM.

3.6 Musikame (MSK) slide, (Type-C)

Musikame slide of Type-C occurred just next to and in the west of the Higashiyama anticline,

where geology is Araya formation of Miocene with dark gray massive mudstone interbedded withfine-grained sandstone. Fig. 45 shows the airphotos of the slide just after the failure, indicatingthat a lot of koi ponds were in the area and one of them was the start point of the failure. Fig. 46shows the view from the top of the slide indicating that the muddy debris flowed like a liquid in along distance crossing over a road and reached a stream. The disintegrated muddy soil mass slid down along the parallel dip plane of presumably mudstone.

The grain size curve shown in Fig. 9(a) indicates that the debris consists of very fine soilwith D50 = 0.03 mm and the fines content of more than 90%. Soil particle density of sampled debris= 2.72 g/cm3, natural water content wn = 39%, liquid limit wLL = 47% and plastic limitwPL = 35% and I p = 12.

Fig. 47(a) depicts a contour map of Mushikame slope before the slide and Fig. 47(b) shows thecorresponding airphotograph. The elevation step of the contours is 2 m. Fig. 48(a) depicts the con-tour map of the same slope after the failure (28 October, 2004) and Fig. 48(b) is the corresponding

photograph. The thick curve in Figs. 47(a) and 48(a) indicates the periphery of the affected area,which was determined from the two DEM data so that the elevation difference between the twoexceeds 1 m.

In Figs. 47 and 48, thin parallel lines are drawn toward the direction of the slope failure in order to evaluate cross-sectional change of the slope. They are 21 thin lines 10 m apart, among which4 thick lines are 20 m apart to each other. The cross-sectional changes in the affected area are

67




Figure 45. Airphotos of Mushikame slide, indicat-ing that a koi pond was the start point of the failure

(after Asia Air Survey Co. Ltd.).

Figure 46. View from the top of the slide indicatingthat the muddy debris flowed on the mudstone dip

plane reaching a stream.

Figure 47. Contour map (a) and air-photograph (b) of Mushikame slide before the earthquake (based on


developed as illustrated in Fig. 49 from the two DEM data before and after the earthquake along4 thick parallel lines. The dotted line in Fig. 49 is the slip surface estimated from the global changesof slope configuration, which may include more uncertainties than slope surface topography. Thecross-sections outside the failed slope almost coincide before and after the earthquake, indicatingthat the tectonic effect is almost negligible here, too.

The horizontal area is 11900 m

2

and 22700 m

2

and the total volume of the sliding soil massthus calculated from the 3D DEM image is 162000 m3 and 181000 m3 before and after the failure,respectively. If the displaced soil mass is idealized by a rectangular block as in Fig. 50, the thicknessof the block is 13.6 m and 8.0 m before and after the failure. The sliding movement of the center of the block is 44 m vertically and 113 m horizontally.

In the CD-ROM of this volume, the coordinates (x: NS, y: EW, z: UD) before and after theearthquake are given in meters at all nodes of 1 m square meshes covering the rectangular areaincluding the failed slope. The coordinate of the starting point R1 (the bottom left corner of therectangle) is x= 33901 and y= 149301 and that of the end R2 (the top right corner) is x = 34599

68







Figure 48. Contour map (a) and air-photograph (b) of Mushikame slide after the earthquake (based on


350 300 250 200 150 100 50 0200

250

300

Horizontal distance (m)

08–08

A l t i t u d e ( m )

before after slip plane

200

250

300 10–10

200

250

30012–12

200

250

300 14–14

Figure 49. Cross-sectional change of Mushikame


15.1°

190.0

120.0

7.9829.9°

145.0

82.0

13.6

∆h = 44.0

(δr)av = 113.4°

21.2°


rectangular block in Mushikame slope.

and y= 149999 by the Global Coordinate VIII. The coordinates of all the grids are arrayed alongeach horizontal line from west to east and then the next line up to north.

Furthermore, along 21 equidistant straight lines parallel to the sliding directions, the originalcoordinates (x, y, z) and elevation changez due to the slope failure at surface points of every 0.5 mapart are also given. Because of the errors involved and some other reasons, even the nodes outsidethe failed slope show some non-zero elevation changes. In order to differentiate the outside nodes,

69





the coordinates inside the periphery of the affected area is marked. Thus, detailed topographicalchanges can be drawn from the digital data in the CD-ROM.

4 SUMMARY

The 2004 Niigata-ken Chuetsu earthquake ( M J = 6.8) caused more than 4000 slope failures inthe area about 200 km north of Tokyo. In most of the slope failures, sand stones were highlyresponsible mainly because of their small strength due to strong weathering. The slope failuresdue to this particular earthquake are classified into 3 types. Out of the great number of slopefailures, six representative slope failures of failure modes A, B, C are explained in detail. In order to understand the background of the failures better, pertinent data concerning geology, seismicconditions, borehole logging and soil properties are also provided.

Three-dimensional changes of slopes before and after the earthquake are delineated not only by photographs, charts of plans and cross-sections but also by 3-dimensional digital data stored inthe accompanying CD-Rom. Site specific geological or geotechnical conditions are also addressed

as much as possible. The digitized data may be able to serve as valuable case histories for eval-uating seismically induced slope failures including their run-out distances from the viewpoint of Performance-Based Design.

ACKNOWLEDGMENTS

Part of this research was supported by Special Coordination Funds for Promoting Science and Technology, “Earthquake damage in active-folding areas– Creation of a comprehensive data archiveand suggestions for its application to remedial measures for civil-infrastructure systems –” of Japan Science & Technology Agency. Strong motion records during the 2004 Niigata-ken Chuetsu

earthquake in websites of Japan Meteorological Agency (JMA) and NIED (National ResearchInstitute for Earth Science and Disaster Prevention) are gratefully referred. The Yuzawa SaboOffice, Ministry of Land, Infrastructure and Transport, Japan is gratefully acknowledged for their kind dissemination of geological survey results in damaged slopes.

REFERENCES

Geographical Survey Institute, http://www.gsj.go.jp/English/index.html

Geological Survey of Japan, Catalogue of geological maps, Quadrangle series 1/50,000. http://www.gsj.jp/Map/

index_e.html

Gutenberg, B. 1955. The energy of earthquakes. Quarterly Journal of the Geological Society of London .Vol.CXII. No.455, 1–14.

Japan Society for Civil Engineers (JSCE). 2007. Earthquake damage in active-folding areas – Creation of a

comprehensive data archive and suggestions for its application to remedial measures for civil-infrastructure

systems –. Report of JSCE by Special Coordination Funds for Promoting Science and Technology. Japan

Science & Technology Agency (in Japanese).

JMA. Japan Meteorological Agency. http://www.seisvol.kishou.go.jp/eq/kyoshin/jishin/index.html.

K-net. NIED (National Research Institute for Earth Science and Disaster Prevention. http://www.k-

net.bosai.go.jp/k-net/

Kokusho, T., Ishizawa, T. and Harada, T. 2004. Energy approach for earthquake induced slope failure evalu-

ation. Proc. 11th International Conf. on Soil Dynamics & Earthquake Engineering and 3rd International

Conference on Earthquake Geotechnical Engineering . Berkeley. Vol.2, 260–267.Kokusho, T. and Ishizawa, T. 2007. Energy approach to earthquake-induced slope failures and its implications.

Journal of Geotechnical and Geoenvironmental Engineering . ASCE. Vol.133. No. 7. 828–840.

Kokusho, T., Motoyama, R. and Motoyama, H. 2007. Wave energy in surface layers for energy-based damage

evaluation. Soil Dynamics & Earthquake Engineering 27. 354–366.

Kokusho, T. and Ishizawa, T. 2008. Energy-based evaluation of earthquake-induced slope failure and its

application. Geotechnical Special publication No.181, Geotechnical Earthquake Engineering and Soil

Dynamics, GEO Institute, ASCE, (CD-publication).

70


http://www.gsj.go.jp/

http://www.gsj.jp/

http://www.gsj.jp/

http://www.seisvol.kishou.go.jp/


http://www.knet.bosai.go.jp/


http://www.gsj.jp/

http://www.gsj.jp/









http://www.gsj.jp/

http://www.gsj.jp/




Slump failure of highway embankments during the 2004

Niigataken Chuetsu earthquake

K. Ohkubo & K. Fujioka Nippon Expressway Research Institute Co. Ltd

S. YasudaTokyo Denki University, Saitama, Japan

ABSTRACT: The 2004 Niigataken-chuetsu earthquake caused serious damage to highwayembankments of Kan-etsu Expressway in Japan. Severe slides of embankments occurred in ahill zone. Some excess pore water pressure had to be generated in the saturated part of the fillsand caused the slide. On the contrary, slump failure of embankments occurred in level grounds.The embankments settled several ten cm and caused differential settlement between embank-ments and culverts in gravelly grounds. In addition, large settlements of culverts and lateralspread of embankments occurred in clayey and sandy grounds. The settlement of the embankmentseemed to be induced due to the reduction of shear modulus of the filled materials and foundationgrounds.

1 INTRODUCTION

In Japan the first expressway was opened for traffic in 1963. Since then many expressways have been constructed. Total length of the constructed expressways reached 8,273 km in 2007. Severalearthquakes, such as the 1978 Miyagiken-oki, the 1987 Chibaken-toho-oki and the 1995 Kobeearthquakes hit the operated expressways. During the 1995 Kobe earthquake, many bridges and elevated bridges were seriously damaged. However, expressway embankments were not suffered severe damage. The 2004 Niigtaken-chuetu earthquake was the first event that earthquake caused severe damage to expressway embankments. Slope failures and large slump failures of embank-ments occurred at many sites of Kan-etsu Expressway. Then, detailed soil investigation was carried out to demonstrate the mechanism of the damages to embankments.

2 OUTLINE AND CLASSIFICATION OF THE DAMAGE TO KAN-ETSUEXPRESSWAY EMBANKMENTS

2.1 Outline of the damage to expressways

On October 23 in 2004, the Niigataken-chuetsu earthquake, of Magnitude 6.8, occurred and caused serious damage to many structures and slopes in Japan. Six expressways were closed due to theearthquake. Total length of the closed expressways was 580 km. Emergency treatments were applied to the damaged expressway embankments by f illing, placing and spreading. Then all expresswayswere able to open for emergency vehicles about 19 hours after the earthquake because no seri-ous damage induced for expressway bridges and tunnels. About 13 days after the earthquake allexpressways were opened for every vehicle.

Among the inflicted six expressways, the following two zones were severely damaged.

(1) Between Muikamachi IC and Nagaoka IC of Kan-etsu Expressway (57.6 km), and (2) Between Kashiwazaki IC and Sanjo-Tsubame IC of Hokuriku Expressway (50.3 km)

71




Figure 2. Kan-etsu Expressway constructed on the

level ground between Yamamotoyama Tunnel and

Yamaya PA.

24.5m

7 m

1 : 1. 8

Figure 3. A standard cross section of the embank-

ment on the level ground.

Embankment on level

ground 13.9 km (57%)

Widening of embankment

1.2 km (5%)

Half-bank and half-

cut 1.7 km (7%)

Others (Cut, Bridge,

Tunnel) 7.6 km (31%)

Between Koide IC and Ojiya IC, 24km

Figure 4. Percentage of length of embankments constructed by three methods.

0

2000

4000

6000

8000

10000

12000

Half-bank

and half-cut

T o t a l l e n g t h ( m )

Embankment

on level ground

Widening of

embankment

Not damaged

Damaged

Figure 5. Total length of damaged and not damaged embankments constructed by three methods.

seem to be two to three times compared with the length of intact embankments regardless of construction method. And, it must be noted many sites were damaged though the ground is flat.

In Japan, damage to road embankment is classified in three levels as shown in table 1. Thenthe damage to the embankments of Kan-etsu expressway were classified into three levels and showed in Figure 6. Serious damage occurred at half-bank and half-cut embankment only. In theembankments on level ground, medium or minor damages dominated.

According to the mechanism of failure, the damage of the Kan-etsu expressway embankmentsin the section between Koide IC and Ojiya IC, can be classified to three types as follows:

(1) Type 1: Serious slide of the embankment on the sloping ground as schematically shown inFigure 7 (a).

(2) Type 2: Settlement of the embankment on the level ground without the deformation of theground as schematically shown in Figure 7 (b)

(3) Type 3: Settlement of the embankment and the culvert on the level ground with the deformationof the ground as schematically shown in Figure 7 (c)

Locations where these types of failures occurred are shown on Figure 1.

73





Table 1. A classification method of the level of damage to raod embankments in Japan (JRA, 2007).

Level of damage Schematic diagram Definition of damage

Minor Surface slide of embankment at the top of slope only

Minor cracks on the surface of a road

Medium Deep slide of embankment or slump involving traffic lines

Medium cracks on the surface of a road and/or settlement

of embankment

Serious Serious slump of embankment

Serius slide of embankment

0%

20%

40%

60%

80%

100%

No damage

Minor damage

Medium damage

Serious damage

Settlement

No damage

Minor damage

Medium damage

Serious damage

Settlement

No damage

Minor damage

Medium damage

Serious damage

Settlement

P e r c e n t a g e o f d a m a g e l e v e l

Half-bank and half-cutEmbankmenton level ground Widening of embankment

Figure 6. Classification of damage level of embankments.

3 SOIL CONDITIONS AND ESTIMATED MECHANISM OF THE DAMAGEDEMBANKMENTS

3.1 Type 1: Serious slide of the embankment on the sloping ground

Serious slide of embankment occurred between Koiede IC and Kawaguchi IC as shown in Figure 1.The expressway was constructed on gentle slopes of hills. As the expressway crosses several smallvalleys, embankments were constructed by filling soils on the valleys. Severe slide occurred at thesesites as shown in Figure 8. Detailed soil investigation including triaxial tests and cyclic torsionalshear tests was conducted at 214.35 KP where the most severe damage occurred. Figure 9 showsthe cross section at the site. Embankment was constructed on a gentle slope of about 5 degrees. Athin gravelly clay layer is deposited on the surface of the slope. Dense gravelly soils are underlaid.Filled material is clayey gravel with 20 to 30% of fines as shown in Figure 10. SPT N -values of the fill were 2 to 10. According to the measurement of ground water level conducted about two

74




(a) Type 1: Serious slide of the embankment on the sloping ground

(b) Type 2: Settlement of the embankment on the level ground without the deformation of

the ground

(c) Type3: Settlement of the embankment and theculvert on the level ground with

the deformation of the ground

Settlement ofground

Lateral flow

Settlement of road surface

Soft ground

Separate at jointC-Box

Lateral flow

C-Box

Ground water tableround water table

Collapse of embankmentollapse of embankment

Flow of low of

collapsed soilollapsed soil

Dense groundense ground

C-Box Box

Settlement of road surfaceettlement of road surface

Figure 7. Classification of the damage to the embankment of Kan-etsu Expressway according to the

mechanism of failure.

Figure 8. Serious slope failure on the sloping ground at 214.35 KP.

month after the earthquake, water level was about 2 m higher than the bottom of the f ill. However,the measured water level is not accurate because retaining sheet piles had been constructed for emergency treatment before the measurement of water level.

In addition to the soil investigation, laboratory tests and analyses were carried out to demonstratethe mechanism of failure. As shown in Figure 7 (a), lower part of the fill was saturated. The filled soil contains not so much fines and comparatively easy to induce excess pore water pressure due

75





Figure 9. Cross section of the failed embankment as 214.35 KP.

0

10

20

30

40

50

60

70

80

90

100

0.0100 0.01 1001010.1

Grain size (mm)

P e r c e n t f i n e r b y w e i g h t ( %

)

Figure 10. Grain-size distribution curves of fill soils at Type 1 failure site.

to shaking. Therefore it is estimated that some excess pore water pressure was generated in thesaturated part of the fill and caused slide and subsequent flow towards downstream.

3.2 Type 2: Settlement of the embankment on the level ground without thedeformation of the ground

Figure 11 shows the soil cross section from Yamamotoyama Tunnel to Yamaya PA. Surface soil is

gravel near Yamamotoyama Tunnel, then, changes gradually to soft clayey soil or loose sandy soiltowards Yamaya PA. Type 2 and Type 3 failures occurred near Yamamotoyama Tunnel and around Ojiya IC, respectively.

Figure 12 shows C-Box Kawaguchi 11. Large differential settlement of 70 cm occurred betweenthe embankment and the culvert box as shown in Figure 13. The culvert box consists of two concrete

boxes. The culvert itself settled 10 cm only and joint of two concrete boxes opened 10 cm. Thereforeit can be said culvert box settled and stretched slightly as schematically shown in Figure 7(b).

Detailed soil investigation including triaxial tests and cyclic torsional shear tests was conducted.Figure 14 shows boring data and the estimated soil cross section of the embankment adjacent toC-box Kawaguchi 11. Subsurface soil of the foundation ground is dense gravel with SPT N -value of

more than 50. Height of the embankment is about 10 m. Filled materials are sandy silt with gravel,gravel with cobbles and clayey silt with gravel. Figure 15 shows grain-size distribution curves of these soils. Fines content of these soils was 50 to 60%. Measured water level was about 3 m higher than the bottom of the embankment. However, it is not clear whether the water was perched water or not. Based on the soil investigation and additional laboratory tests, it is estimated the settlementof the embankment occurred due to decrease of shear modulus of the filled materials.

The differential settlements between embankments and culverts, and the settlements of cul-verts themselves were measured between Yamamotoyama Tunnel and Yamaya PA, and plotted onFigure 16. The differential settlement was about 50 to 70 cm near Yamamotoyama Tunnel and the

76





Figure 11. Soil cross section between Yamamotoyama Tunnel and Yamaya PA.

Figure 12. C-Box Kawaguchi 11. Figure 13. Differential settlement between

embankment and culvert at.

Figure 14. Soil cross section adjacent to the culvert box at C-Box Kawaguchi 11

77








0

10

20

30

40

50

60

70

80

90

100

Grain size (mm)

P e r c e n t f i n e r b y w e i g h t ( % )

226-2

226-2-S1

226-2-S3

226-2-S4

0.001 0.01 0.1 1 10 100

Figure 15. Grain-size distribution curves of fill soils near C-Box Kawaguchi 11.

KP

0

20

40

60

80

100

226.00 226.50 227.00 227.50 228.00 228.50 229.00 229.50 230.00 230.50 231.00

226.00 226.50 227.00 227.50 228.00 228.50 229.00 229.50 230.00 230.50 231.000

20

40

60

80

100

Ojiya IC

Settlement ofculvert (cm)

Ojiya IC

C-Box

Kawaguchi 11

C-Box

Ojiya 2

C-Box

Kawaguchi 22

Differentialsettlement betweenembankment and

culvert (cm)

Separation of jointsof culvert boxes

KP

S 1

S 2

S S: Settlement of embankment

S1: Differential settlement

between embankment

and culvert

S2: Settlement of culvert

J: Separation of joint of

culvert boxes

J

before earthquake

after earthquake

Figure 16. Distribution of differential settlements between embankments and culverts, settlements of culverts

and separation of joints of culverts in level ground.

78




Figure 17. Differential settlement at C-Box

Kawaguchi 22.

Figure 18. Settlement of culvert at C-Box

Kawaguchi 22.

Figure 19. Separation of joint at C-Box Ojiya 2. Figure 20. Moved adjacent ground at C-Box

Ojiya 2.

settlements of culverts were about 10 to 20 cm. On the contrary, settlements of culverts were verylarger near Ojiya IC.

3.3 Type 3: Settlement of the embankment and the culvert on the level ground with thedeformation of the ground

Around Ojiya IC, differential settlement of several ten cm occurred between embankments and culverts as shown in Figure 17. Moreover culverts settled several ten cm and stretched as shown inFigure 18 and 19. Embankment soil fell down through the opened joints. Embankments, culvert

boxes and grounds deformed as schematically shown in Figure 7 (c). Both side of toes of theembankments spread in lateral direction and caused lateral displacement of adjacent grounds asshown in Figure 20.

Very detailed soil investigation including triaxial tests and cyclic torsional shear tests was con-ducted at C-Box Kawaguchi 22 and C-Box Ojiya 2 to demonstrate the mechanism of the Type 3

failure (Inagaki et al., 2005). Figure 21 shows locations of two sites, together with K-net OjiyaSite, where accelerograph is installed. The maximum surface acceleration recorded at K-net (NIED)Ojiya Site was 1314 cm/s2 in EW direction. Surface soil conditions at C-Box Kawaguchi 22 and C-Box Ojiya 2, investigated after the earthquake, are shown in Figures 22 and 23. Figure 24 and 25 show grain-size distribution curves of embankment soils and subsurface soils of the foundationground at C-Box Kawaguchi 22. Embankment soils at two sites are clayey soils with 70 to 80% of fines. SPT N -values of the embankment soils are 5 to 10. Heights of the embankments at two siteswere 5.6 to 6.8 m and 5.3 to 5.6 m, respectively. Water levels at two sites were about 3 m higher than the bottom of the embankments, though the embankments were constructed on level grounds.

79








Figure 21. Location of investigated sites (map: 1/25000 by GSJ).

Figure 22. Soil profile and tests results at C-Box Kawaguchi 22.

At C-Box Kawaguchi 22, thick soft silty layers, with about 5 of SPT N -values, are deposited fromoriginal ground surface to the depth of 16 m. A thin soft silt layer with 2 m thickness is deposited under the embankment at C-Box Ojiya 2. Then, silty sand, silt, sandy silt and silt layers, with 10 to20 of SPT N -values, were underlaid to the depth of 24 m. Figure 26 shows estimated cross sectionat C-Box Kawaguchi 22.

As mentioned before, large settlements of embankments and culverts occurred at these sites.Differential settlements between embankments and culverts were 20 cm and 70 cm at C-BoxKawaguchi 22 and C-Box Ojiya 2, respectively. Settlements of culverts were 48 cm and 30 cm,respectively. Therefore, total settlements of embankments were 68 cm and 100 cm, respectively.

80






Figure 23. Soil profile and tests results at C-Box Ojiya 2.

Figure 24. Grain-size distribution curves of f illed

materials.

Figure 25. Grain-size distribution curves of

subsurface soils of foundation grounds.

Opening of the joints of culvert boxes are plotted on Figure 16. The openings at C-Box Kawaguchi

22 and C-Box Ojiya 2 were 119.5 cm and 78 cm, respectively.Cyclic torsional shear tests were carried out to obtain cyclic shear strength and shear modulusafter cyclic loading. Then an analytical code “ALID (Yasuda et al., 2003)”, which is one of theresidual deformation methods, was applied to evaluate deformation of embankments and grounds.Analyzed settlements and horizontal displacements were fairly coincided with the measured values,then it was concluded that the settlement of the embankments and culverts occurred due to thereduction of shear modulus of the filled soils and foundation soils (Inagaki et al., 2005).

4 CONCLUSIONS

The 2004 Niigataken-chuetsu earthquake caused serious damage to highway embankments of Kan-etsu Expressway in Japan. The damage is divided into three types, 1) Type 1: slide of theembankment on the sloping ground, 2) Type 2: settlement of the embankment on the level ground without the deformation of the ground, and 3) Type 3: settlement of the embankment and the culverton the level ground with the deformation of the ground. In Type 1, some excess pore water pressurehad to be generated in the saturated part of the fills and caused slide and subsequent flow towardsdownstream. In Type 2 settlement of the embankment seemed to be induced due to the reduction of shear modulus of the filled materials. And the reduction of shear modulus of soils of the foundationgrounds also seemed to be influenced in Type 3.

81







Figure 26. Cross section of the embankment adjacent to C-Box kawaguchi 22.

REFERENCES

Inagaki, M., Itakiyo, K., Kasuda, K., Yamada S. and Yasuda, S. 2005. Deformation of embankments on clayey

grounds during the 2004 Niigataken-chuetsu earthquake (Part 2): Proc. of 4th Annual Conference on Japan

Association for Earthquake Engineering . (in Japanese)

Japan Road Association. 2007. Guideline for restoration work of road after earthquakes. (in Japanese)

National Research Institute for Earth Science and Disaster Prevention (NIED). K-NET WWW service

(http://www.k-net.bosai.go.jp/)

Yasuda, S., Ideno, T., Sakurai, Y., Yosida, N. and Kiku, H. 2003. Analyses of liquefaction-induced settlement

of river levees by ALID: Proceedings of the 12th Asian Regional Conference on SMGE , 347–350.

82


http://www.k-net.bosai.go.jp/








Fill slope failure of the Takamachi housing complex in the 2004

Niigataken Chuetsu earthquake

S. Ohtsuka & K. Isobe Department of Civil and Environment Engineering, Nagaoka University of Technology, Japan

T. TakaharaGraduate School of Natural Science and Technology, Kanazawa University, Japan

1 INTRODUCTION

In the 2004 Niigataken Chuetsu earthquake, a large number of houses were damaged by ground deformation. In particular, at the Takamachi housing complex in southeast Nagaoka City (as shownin Fig. 1), the ground deformed significantly in a fill area, and artificial fill slopes around thecomplex suffered considerable collapse. Damage to houses caused by ground deformation had beenrepeatedly observed in past earthquakes such as the 1978 Miyagiken-oki and the 1995 Hyogoken-nanbu earthquakes. Since flat land in Japan is limited, the cut and fill technique is widelly applied tothe development of housing complexes. Consequently, it is important to investigate the correlation

between damage to houses and the ground conditions of cut and fill areas. Artificial fill slopes invalleys and levees are also often reported to collapse in earthquakes. At the Takamachi housing

complex, artificial fill slopes suffered significant collapse at four sites in the area surrounding thecomplex.

This paper reports on the ground disaster involving damage to houses, ground deformationand slope failure at the Takamachi housing complex. The correlation between damage to housesand f ill distribution was investigated in detail. Fill distribution was analyzed based on a GIS(geographic information system) with site investigation results from a boring survey and surfacewave exploration. At the sites of fill collapse, static and dynamic mechanical properties of fill

Figure 1. The locations of the Takamachi housing complex and the epicenter (map by Google Earth).

83





Figure 2 (a)–(c). Measured acceleration at base (GL-104 m); (a) EW, (b) NS, (c)UD (National Research

Institute for Earth Science and Disaster Prevention).

material were investigated for intact and collapsed samples. Furthermore, dynamic properties of unsaturated soil were investigated in order to know the effect of saturation degree, as the rainstormthree days before the earthquake might have exacerbated the ground disaster.

2 EARTHQUAKE DAMAGE TO THE TAKAMACHI HOUSING COMPLEX

2.1 Outline of the Takamachi housing complex

The Takamachi housing complex is located at the western edge of the Higashiyama Hills as shownin Fig. 1. It is flanked to the west by the Echigo Plain and to the east by the Kagi River. Sincethe Chuetsu area of Niigata Prefecture is actively folded, the Higashiyama Hills form an anticlineaxis from north-northeast to south-southwest. The ground strata at the Takamachi housing complexconsist of the Oyama, the Uonuma and the Nishiyama layers downward from the ground surface.The Uonuma layer is made of sand, clay, gravel and andesite pyroclastic rock, while the Oyama layer is made of sand, clay and gravel. The Takamachi housing complex was developed on hilly terrain on

the east side of Nagaoka City in the early 1980s. In the development of the complex, it was planned that the border altitude for the cut and f ill was approximately 70 m elevation. The area is 1.2 kmin length and 0.3 km in width, and has a north-northeast to south-southwest orientation. It wasoriginally a hill with two low summits; the higher part was cut, and the ravines and circumferentialarea were filled with the cut soil. The fill was stabilized with concrete retaining walls of 5 m inheight and 1 m in width.

Fig. 2 indicates the acceleration time record of main shock measured at GL-104 m. This point iscalled NIGH01 and set by National Research Institute for Earth Science and Disaster Prevention,1.5 km northeast from the Takamachi housing complex as shown in Fig. 1. The f igures showthe acceleration records of east-west (EW), north-south (NS) and up-down (UD) directions. Themaximum magnitude of acceleration is registered as 412 gal in NS direction at base and 818 gal in

NS direction at surface.

2.2 Damage to the Takamachi housing complex

By making digital elevation maps with 2 m mesh from aerial photographs before and after thedevelopment of the Takamachi housing complex (by Japan Geographical Survey Institute in 1975and1989), a distributionmapof the cut andfill areas was compiled using a geographical informationsystem (GIS) as shown in Fig. 3. The ground was cut in the center area, and the cut soil was used to fill the periphery. The fill is mostly a widened embankment, and is partly located at a valley and levee section.

Fig. 4 indicates the fill area and the distribution of earthquake damage, including ground cracks,slope failures and damaged houses. The results of the emergent housing risk investigation areexpressed in the figure, with yellow marks (See the attached CD-Rom version) indicating a warninglevel (corresponding to a light level of damage) and red marks pinpointing a dangerous level(corresponding to a severe level of damage). From Fig. 4, it is clear that most of the damaged houses are distributed throughout the fill area. In the Takamachi housing complex, fill slopesunderwent significant deformation, and four slopes suffered considerable collapse. The four slopefailure sites are numbered 1 to 4 from the north. On the other hand, cut slopes did not deformextensively. Ground cracks in the fill area were widespread, but some were also found in the cut

84





Figure 3. Estimation of cut and fill distribution at the Takamachi housing complex.

area near the fill area. Fig. 5 shows an example of a damaged house. The residence itself was notdestroyed by the motion of the earthquake, but its foundations were severely damaged due to uneven

settlement and lateral displacement. Fig. 6 expresses the gap due to settlement of the periphery road in the fill area. Since the ground displacement at the shoulder of embankment was very large, thehouses on the shoulder were severely damaged without exception. Fig. 7 shows a house on the edgeof the main scarp at the slope failure site No. 4 in Fig. 4. Since the shoulder of the slope was utilized as a periphery road, the housing there just escaped from the slope failure. The site was a valley fillarea as seen in Fig. 4 with greater fill thickness. Collapsed soil moved a long distance downward along the valley to the border of the complex, possibly due to the presence of groundwater.

Conversely, the fill slope around the B-line in Fig. 4 did not collapse, even though thefill thickness was comparatively high. At the site, differential settlement of the ground was

85





Figure 4. House damage, ground cracks and slope failure distribution at the Takamachi housing complex.

actually observed before the earthquake, and anchor works had been conducted on the gravity-type retaining wall. This case may give an example that the seismic stabilization of the slope

by the anchor work was effective, though the design of the anchorage was only for settlement prevention.

86





Figure 5. Damage to houses as a result of ground

deformation (photo by Fukuda Co.).

Figure 6. Differential settlement of periphery road.

Figure 7. Slope failure of valley fill (photo by A. Shibata).

2.3 Damage to houses and ground conditions of cut and fill

In order to investigate the accuracy of the cut and fill distribution in Fig. 3, surface wave explorationwas conducted on the A-line and the B-line in Fig. 4. The obtained S-wave velocity distribution isshown in Fig. 8. From comparison with the bore-hole survey near theA-line, the border between the

cut and fill areas was assumed as an S-wave velocity of 160–170 m/s. In Fig. 8, the border betweencut and fill areas based on S-wave velocity and DEM data are shown with the black broken curveline and the red curve line, respectively. Both lines are almost coincident on the A-line and theB-line, and it is confirmed that both the DEM data and the surface wave exploration plus the boringdata (see Fig. 12, 14(c)) are effective in estimating the fill area accurately.

Based on the cut and fill distribution of Fig. 3, the correlation between fill thickness, ground cracks and damage to houses shown in Fig. 3 was analyzed. Fig. 9(a) expresses the relationship

between the numbers of damaged houses, ground cracks and the fill/cut thickness, where the nega-tive value indicates the cutting thickness. Fig. 9(b) indicates the damaged houses ratio according tothe fill/cut thickness. In addition, Fig. 9(c) shows the damaged houses ratio on the fill, the ground

cracks and both. Since the development of cracks depends on local filling thickness, cracks aredivided into segments of 5 m lengths and the fill thickness is measured at their centers for statisticalanalysis. Small cracks such as hair cracks and invisible cracks under the pavement and the housesare not counted in the analysis. From Fig. 9(a), it is apparent that the damaged houses and theground cracks are mostly distributed in the fill area or near the border, but some also exist in thecut area. It is found that the damaged houses ratio in the fill area and the cut area from −2.5mto 0.0 m in thickness is higher in Fig. 9(b). Fig. 9(c) shows that 43% of the houses in the f ill areawere damaged, 80% of the houses on the ground cracks were damaged. Since the percentage of the damaged houses both in the fill and on the ground cracks are almost the same as that on the

87







Vs

[m/s]

120

140160

180

200

220

0 10 20 30 40 50 60 70 80

edgeFill

FillA-line

D e p t h [ m ]

0

-5

-10

-15

-20

240

By surface wave investigation + boring data

By DEM data

Distance [m](a) A-line.

0

Distance [m]

10 20 50 60 7030 40

120

140

160

180

200

240

220

Fill Fill

B-line

0

-5

-10

-15

-20

D e p t h [ m ]

Vs

[m/s]

By surface wave investigation + boring data

By DEM data

(b) B-line.

Figure 8. Fill thickness estimation by surface wave exploration.

0

5

10

15

20

0

100

200

300

400

500

600

700

10 5 0 5 10

N u m b e r

o f d a m a g e d h o u s e s

N um b e r of c r a c k s

Fill and cut thickness [m]

Damaged houses

Ground Cracks

FillingCutting

Figure 9(a). Relationship between the fill/cut thick-

ness and the number of damaged houses/number of

cracks.

0

10

20

30

40

50

D a m a g e d

h o u s e s r a t i o [ % ]

2 . 5 ∼

0 . 0 m

0 . 0 ∼

2 . 5 m

2 . 5 m ∼

1 0 . 0 ∼

7 . 5 m

7 . 5 ∼

5 . 0 m

5 . 0 ∼

2 . 5 m

∼ 1 0 . 0 m

FillingCutting

(Danger+Warning)/All

Danger/All

Fill and cut thickness [m]

Figure 9(b). Relationship between the fill/cut thick-

ness and the damaged houses ratio.

ground cracks, a strong correlation is seen between the damage to houses and the ground cracks,indicating that many houses were damaged by differential settlement and lateral deformation onthe cracks as exemplified in the photograph of Fig. 5.

2.4 Characteristics of fill slope failure

Four large-scale slope failures occurred in the periphery of the land (see Fig. 4), three failures other than Site No. 1 were located in valley leachate collection areas. Site No. 4 (see Fig. 7) collapsed

88




0.0

20.0

40.0

60.0

80.0

100.0DangerWarningNo damage

D a m a g e d

h o u s e s r a t i o [ % ] 13 %

11 %

76 %

25 %

18 %

58 %

67 %

12 %

22 %

69 %

10 %

21 %

All Fill Crack Fill & Crack

Figure 9(c). Comparison of the damaged houses ratio on the fill, ground cracks and both.

Figure 10. Daily precipitation and maximum hourly precipitation measured in Nagaoka in October in 2004.

Figure 11. Slope failure at site No. 3.

as a result of the main shock; approximately 40 meters of the peripheral road was destroyed, and

the foundation ground under the houses in front of the road was subsided. The concrete retainingwalls were also displaced downward. Heavy rain (over 100 mm in a day, see Fig. 10) fell two days before the earthquake and the groundwater level was possibly high at the time of the quake.

At Site No. 2, a length of approximately 50 m of the peripheral road was ruined, and the concreteretaining wall supporting the f ill was translated downward. The slope failure was 40 m wide atthe top, 53 m long and 19 m wide at the bottom. Waterholes were observed in the center of thecollapsed soil, and groundwater seeped out indicating high degree of saturation in the fill. At Site

No. 3 shown in Fig. 11, the retaining wall was also displaced downward but remained standing asat Site No. 2. The slope failure was 35 m wide at the top, 50 m long and 24 m wide at the bottom.

89






Figure 12. Plane view of slope failure site No. 3.

Uonuma layer

Oyama layer

Bank

Collapsed sediment

8300 3000

300

8000

Surface after earthquake

Surface before earthquake

Slope deposit

15

10

5

0 20 40

205

0 20 40

B o r i n g I

B o r i n

g I I

Unit: m

Figure 13. Cross section of slope failure site No. 3.

Since the fill was on a slope, the collapsed soil reached as far as 50 m downward. A plane viewand cross section of Site No. 3 are shown in Figs. 12 and 13, respectively. As shown in Fig. 13, the

widened fill on the intact slope collapsed completely.

3 FIELD INVESTIGATION AND SOIL TESTS ON FILL SLOPE FAILURE

3.1 Physical and mechanical properties of fill material

Intact and collapsed soils were sampled near Site No. 3 as shown in Fig. 12. The void ratio wasobtained as 1.30 for the collapsed soil and 0.90 for the intact soil. The increase in void ratio of

90





Figure 14. Boring core logs at slope failure site No. 3; (a) boring I, (b) boring II, (c) boring III.

Table 1. Physical test results of disturbed and undisturbed soils.

e0 ρ s wn w L w P I P I L – [g/cm3] [%] [%] [%] – –

Intact 0.9 2.692 30.3 56.7 23.8 32.9 19.8

Collapsed 1.3 2.689 – 74.0 41.0 33.0 –

Table 2. Monotonic strength parameters of disturbed and undisturbed soils.

ccu φcu c φ

[kPa] [deg] [kPa] [deg]

Intact 20.9 20.6 0.0 36.9

Collapsed – – 0.0 36.9

(reconstituted)

collapsed soil indicates the dilation by water absorption during the collapse. The groundwater level of the ground was measured for 6 months beginning about a year after the earthquake at thesampling site, but it was lower than the border of the cut and fill, presumably because the ground water may have been lowered by the recovery work. The results of boring core logs in the site areshown in Fig. 14.

The physical properties of the collapsed and intact soils were obtained as listed in Table 1. Thetests for the collapsed soil were conducted after eliminating coarser grains than 2 mm by sieving.Although W L and W P are different between the collapsed and the intact soils, I P becomes coincident.

Takamachi fill soil is classified as an intermediate soil based on the JGS standard (1992) and thereference (Ito et al, 2001). The strength parameters for the intact and collapsed soils were obtained as shown in Table 2. In the tests, the specimens of the collapsed soils were reconstituted to havethe same void ratio as the intact soil.

Shear modulus degradation curves of the intact soil obtained by cyclic loading tests are shownin Fig. 15. In this figure, the relationship between normalized shear stiffness G /G o and shear strainγ is indicated, where G o is the initial shear stiffness. The relationship between damping ratio h and shear strain γ is also shown. In the tests, the loading frequency was set to 0.1 Hz, and dampingratio h was calculated using the 10th hysteresis loop in each step.

91





Figure 15. Dynamic deformation characteristics of Takamachi fill material.

Table 3. Test cases of cyclic tri-axial testing.

Case e Sr w σ d /2σ 0 Frequency p

No. – [%] [%] – [Hz] [kPa]

s1∼s41.3

100 – 0.195∼0.238

u1∼u4 51.7∼82.7 25.0∼40.0 0.2510.1 100

s5∼s70.9

100 – 0.500∼0.670

u5∼u7 59.5∼98.2 20.0∼33.0 0.510

0.00

0.20

0.40

0.60

0.80

0.1 Number of cycles Nc

0.225

0.600e 0.90

e 1.30

DA 5 % DA 2 % DA1 %

DA 5 % DA 2 %

DA 1 %

1 10 100 1000

Figure 16. Liquefaction strength of collapsed soil (reconstituted).

3.2 Cyclic shear characteristics of reconstituted Takamachi fill soil under saturated and unsaturated conditions

Since fill slopes are composed of both saturated and unsaturated soils, it is necessary to investigatethe effect of the saturation degree on the soil behavior during earthquakes. Hence, undrained cyclic

triaxial tests were conducted to ascertain the liquefaction strength of Takamachi fill soil for fullyand partially saturated conditions. Reconstituted specimens were used with void ratios identical tothe intact soil and the collapsed soil. The test cases are shown in Table 3.

The results for the saturated soils are shown in Fig. 16; the cyclic stress ratio (2σ d /σ

0) versusthe number of cycles for strain double amplitude DA= 1%, 2% and 5%. There are two groups of liquefaction curves for void ratios: e= 0.90 and 1.30. The liquefaction strength ( Nc= 20, DA= 5%)is obtained as 0.225 and 0.600 for e= 0.90 and 1.30, respectively.

Fig. 17 shows the relationship between the saturation degree and the number of cycles to yield the specific strain of DA= 5%. In the tests, the cyclic stress ratio (2σ d /σ

0) was set to the liquefaction

92





Figure 17. Influence of saturation degree on the liquefaction strength of collapsed soil (reconstituted).

strength under a saturated condition (0.510 for e = 0.90). Since the number of cycles to yield thespecific strain increases with the degree of saturation, the shear strength of soil increases witha decrease in the degree of saturation. From the figure, liquefaction cannot be seen for repeated liquefaction strength when the saturation degree is less than 70∼80%.

4 CONCLUSIONS

In the 2004 Niigataken Chuetsu earthquake, a large number of houses were damaged by ground deformation. Since residential properties are private and their recovery is generally difficult, there

is a greater need for disaster prevention measures when performing earth work in the develop-ment of land for housing. This paper reports on the ground disaster at the Takamachi housingcomplex. The correlation between damage to houses and fill distribution was investigated in detailthrough a GIS-based survey and surface wave exploration. At artificial fill collapse sites, staticand dynamic mechanical properties of fill material were investigated for intact and collapsed samples.

The conclusions reached in this paper are as follows:

(1) Damaged houses in the Takamachi housing complex were mainly located in the fill area, butsome were also in the cut area. This was because ground cracks occurred in the cut area too,causing damage to the properties. DEM data obtained from aerial photographs and surface

wave exploration were effective in estimating the fill area accurately.(2) Fill slopes in valley areas suffered significant collapse, and soil moved downward over a long

distance due to the presence of groundwater. The combination of the earthquake and heavyrainfall is identified as a factor that exacerbated the ground disaster.

(3) Fill slope failure at Site No. 3 was investigated through mechanical tests on both intact and collapsed soils (reconstituted). The liquefaction strength was obtained for reconstituted soilswith a void ratio equal to that of intact fill soil. With decreasing degree of saturation, theliquefaction strength tends to increase. Saturation of the soil at the bottom of the fill mighthave caused the fill slope to fail. The detail consideration regarding the conclusion (3) isexplained in the reference (see Ohtsuka et al. 2009, Konagai et al. 2007).

ACKNOWLEDGMENT

The present research was supported by “Earthquake Damage in Active-Folding Areas: Creationof a Comprehensive Data Archive and Suggestions for Its Application to Remedial Measures for Civil-Infrastructure Systems, Research and Development Program for Resolving Critical Issues,Special Coordination Funds for Promoting Science and Technology”. The authors would like toexpress their gratitude to Nagaoka City for offering valuable disaster data.

93





REFERENCES

Konagai, K. et al. 2007, Earthquake damage in active-folding areas: creation of a comprehensive data archive

and suggestions for its application to remedial measures for civil-infrastructure systems, Research and

Development Program for Resolving Critical Issues, Special Coordination Funds for Promoting Science

and Technology.

Ito, S., Hyodo, M., Fujii, T., Yamamoto, Y. and Taniguchi, T. 2001, Undrained monotonic and cyclic shear characteristics of sand, clay and intermediate soils, Journal of the Japan Society of Civil Engineers 680

(III-55): 233–243 (in Japanese).

Ohtsuka, S., Isobe, K. and Takahara, T. 2009, Consideration on fill slope failure in Takamachi developed

residential land in 2004 Niigata Chuetsu Earthquake, Proc. of International Conference on Performance-

Based Design in Earthquake Geotechnical Engineering – from case history to practice (IS-Tokyo 2009), in

printing.

The Japan Geotechnical Society 1992, Geotechnical note 2 Intermediate – sand or clay – (in Japanese).

94





Uplift of sewage man-holes during 1993 Kushiro-oki EQ., 2003

Tokachi-oki EQ. and 2004 Niigataken Chuetsu EQ

S. Yasuda & T. TanakaTokyo Denki University, Saitama, Japan

H. Kiku Kanto Gakuin University, Yokohama, Japan

ABSTRACT: Many sewage manholes and pipes were uplifted during the 1993 Kushiro-oki, the2003 Tokachi-oki and the 2004 Niigataken-chuetsu earthquakes in Japan. Before the restorationwork, detailed soil investigations were carried out to reveal the mechanism of the uplift. Based on the investigations, it was found that the uplift mainly occurred in clayey grounds. During theconstruction of buried pipes and manholes, the ground was excavated first, the pipes and manholeswere placed in the ditches, then the ditches were filled with sand. The research on the sand fill after the earthquake revealed that the sands were very loose and easy to liquefy. Then, it was concluded that the uplift of the manholes and pipes occurred due to the liquefaction of the sand f ill.

1 INTRODUCTION

Typical procedures for the construction of sewage pipes and manholes, in Japan, are as follows:

(1) Ground is excavated using sheet piles or other retaining walls.(2) Pipes or manholes are placed at the bottom of the ditch or holes.(3) The ditch or hole is filled with sand. If the excavated soil is sandy, it can be re-used as the

backfill soil. However, if the excavated soil is clayey, sand taken from other areas is used for the fill. This construction method is the same in sandy, clayey or gravelly grounds.

Sandy ground has liquefied frequently in Japan. However, only a few sewage manholes wereuplifted during the 1964 Niigata, the 1983 Nihonkai-chubu, and the 1995 Kobe earthquakes, eventhough liquefaction occurred in wide areas and many structures suffered severe damage. On the

other hands many sewage manholes wereuplifted during the1993 Kushiro-oki, the1993 Hokkaido-nansei-oki, the 1993 Hokkaido-toho-oki, the 2003 Tokachi-oki and the 2004 Niigataken-chuetsuearthquakes. In these cases, large uplift of sewage manholes and pipes was observed mainly inclayey grounds.

2 THE 1993 KUSHIRO-OKI EARTHQUAKE

In 1993, the Kushiro-oki earthquake of MJ = 7.8 occurred near Kushiro City and caused severedamage to sewage pipes, manholes, disposal plants and pump stations in and around Kushiro City.

In total, 7,744 km of sewage pipes were damaged in Kushiro City and 10.8km of sewage pipeswere damaged in Kushiro Town (JGS, 1994, Yasuda et al., 1994). The main types of pipe damagewere uplift, bends and joint failure. Figure 1 shows severely damaged sites in Kushiro City and Kushiro Town with the height of uplifted pipes. The northeast area is peaty ground. As shown inthe figure the damage was concentrated in the peaty ground. Strong motion records were obtained at two sites in Kushiro City. The maximum ground surface acceleration at Kushiro Port in NSand EW directions were 469.3 cm/s2 and 344.2cm/s2, respectively. At the Kushiro MeteorologicalObservatory which is located on a hill, the maximum ground surface acceleration in NS and EWdirections were 711.4 cm/s2 and 637.2cm/s2, respectively.

95




Figure 1. Locations of severely damaged sewage pipes during the 1993 Kushiro-oki earthquake.

In Kushiro Town, not only the sewage pipes but also sewage manholes were uplifted severely at20 sites, as shown in Figure 2. Figure 3 shows locations of the uplifted manholes. The maximumuplift was 1.3 m. After the earthquake, the Public Works Research Institute and Kushiro TownOfficials inspected the uplifted manholes by excavating them and carried out soil investigationsand laboratory tests to examine the uplift mechanism of manholes (Koseki et al., 1997). Figure4 shows the excavated trench. Uplift of a manhole and pipes can be seen. Figure 5 shows theestimated soil cross section of the ground along Nichii-route. An artificially filled layer with athickness of about 2 m and a peat layer with a thickness of 1 to 2 m were found immediately belowthe ground surface, which were in turn underlain by alluvial sand layers.The bottom of the uplifted manholes was almost 4 m below the ground surface. During the construction of the manholes and

pipes, the ground was excavated with a width of about 2 m. After placing the manholes and pipes,the excavated area was filled with sand as schematically shown in Figure 6. The alluvial sands and the replaced sand were silty fine sands with SPT N -values of less than 10 and clean sand with SPT

N -values of 0 to 11, respectively. Based on analyses of liquefaction, it is estimated that the sand fillliquefied and caused the upliftof the pipes and manholes. In Japan, ingeneral, diameterand depth of sewage manholesare 1.0 to1.5 m and 2 to5 m, respectively. Diameterof sewage pipes is15to50 cm.

In 1993, another earthquake, the Hokkaido-nansei-oki earthquake, hit southwest Hokkaido.Fifty-five manholes were uplifted in Oshamanbe Town due to the earthquake. A manhole in peatground was raised by 57 cm. Manholes in sandy ground where liquefaction occurred rose by about

96





Figure 2. A uplifted manhole in Kushiro Town.

Kiba

Katsuragi

Nichii side

Kushiro timberyard

S a t t s u r u B e n i y a s i d e

W

a t e r c h a n n e l

0 100 200 (m)

3 43 11 8 22 2

16 55 131 28 2

5

36

103

18

0 –3

–12

Number shows uplift of manhole (unit:cm)

: Manhole

N No.9

No.65

Figure 3. Height of uplifted manholes in Kushiro

Town.

Figure 4. Excavated trench to inspect the uplifted

manhole.

Figure 5. Soil cross section along the damaged pipe at Nichii side.

10 to 20 cm only. On the contrary, manholes in dense sandy or gravelly grounds were not uplifted (JGS, 1994). In 1994, the Hokkaido-toho-oki earthquake hit eastern Hokkaido and caused theuplift of sewage manholes in Nakashibetsu Town and Shibetsu Town. Damage occurred in peatyground also.

97







Figure 6. Schematic diagram of the cross section of a sewage manhole in Kushiro Town.

3 THE 2003 TOKACHI-OKI EARTHQUAKE

In Japan, after the Hokkaido-nansei-oki earthquakes, the uplift of sewage manholes did not occur until 2003, when the Tokachi-oki earthquake occurred in southern Hokkaido and inflicted muchdamage to river dikes, road embankments, port facilities and other structures. Sewage facilitieswere severely damaged in 14 towns from Kushiro City to Mukawa Town (JGS, 2004 and Yasudaet al., 2004). Figure 7 shows the maximum ground surface acceleration recorded by K-net (NIED).The maximum ground surface acceleration from Kushiro City to Mukawa Town along PacificCoast was about 200 to 800 cm/s2. Among the 14 towns, Onbetsu and Toyokoro towns located near Urahoro were most severely damaged.

Figure 8 shows locations of damaged and intact sewage manholes in Onbetsu Town. Manholesalong line B-B rose more than 60cm. Figure 9 shows a manhole along this line, that was raised by about 1.5m. The ground surface above sewage pipes subsided by about 50 cm. Boiled sand was

observed on the site of the subsidence. Figure 10 shows the grain-size distribution curve of boiled soil. As shown, soil fill seemed to be silty sand. On the contrary, manholes along line A-A werenot uplifted. Figures 11 and 12 show soil cross sections and depths of sewage pipes along line A-A

and line B-B, respectively. Along line B-B, peat and soft clay layers are deposited from the ground surface to the depth of the pipe. However, along line A-A’, soil to the depth of the pipe was mainlygravelly. Therefore, it is estimated that the soil fill in peaty or clayey ground liquefied, and thesoil fill in gravelly ground did not liquefy. Restoration work consisted of excavating the ground, placing new sewage pipes and filling the ditches again. As a countermeasure against liquefaction,cement was mixed in the soil f ill.

A similar type of uplift of sewage manholes occurred in Toyokoro Town. Figure 13 to 15 show

locationsof damagedandintact sewage manholesin threedistrictsofToyokoroTown.Themaximumuplift of manholes in Toyokoro district was 70 cm as shown in Figure 16. Peat layer is deposited at the depth of sewage pipe as shown in Figure 18. Depth of the damaged pipes was about 3 min average. In Ohtsu district, one manhole was raised 1.7 m in the Ohtsu Sewage Center as showin Figure 17. Depth of the manhole was 6.25 m. Uplift of other manholes was less than 30 cm.Figure 19 shows a soil cross section of the ground along B-B’ in Ohtsu district. Sand or clay layer is deposited at the depth of the pipe.

Figure 20 shows the relationship between depth of manholes and observed uplift amount. Itseems that the uplift of the manholes increased with the depth though the data are scattered.

98





Figure 9. A manhole in Onbetsu Town uplifted

during the 2003 Tokachi-oki earthquake.

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Grain size(mm)

P e r c e n t f i n e r b y w e i g h t ( % ) D60=0.346mm

D50=0.242mm D30=0.149mm D10=0.0136mmU c =25 r

s =2.589g/cm3

Figure 10. Grain-size distribution curve of boiled

soil for sewage pipes in Onbetsu Town.

: Fill : Sand : Silt : Clay : Peat : Volcanic soilvv vvv v

vv v

Sewage pipe

D=200mmSewage pipe

D=350mm

H-6 B-1

No.2

Z=3.53m

No.4

Z=2.90m

No.5

Z=2.56m

EX-No.2B-No.6

0m

-5m

5m

10m

No.3

Z=3.24m

No.10

Z=4.97m

No.9

Z=5.17m

No.8 Pump

station

No.7

Z=1.81m

No.6

Z=2.20m

No.1

Z=3.45m

0m50m

0 50

SPT-N valuSPT-N value

0 50

SPT-N value

500

A

A'

25m

Ground surface

Level

Z: depth of pipe

: Gravel

Figure 11. Soil cross section and depth of sewage pipes along line A-A in Onbetsu Town.

: Fill : Gravel : Sand : Silt : Clay : Volcanic soil

vv vvv v

vv v

50

B-13

0m

-5m

5m

10m

0m 50m

H-8 B-1

No.120

Z=5.35m

No.118

Z=4.57m

No.119

Z=4.99m

No.121

Z=5.06m

No.122

Z=5.21m

No.123

Z=5.10m

H-7 B-1

vv

vvvv

vvvv

vvvvv

Sewage pipe

D=200mm

SPT-N value

0 50

SPT-N value

050

SPT-N value

0

B B'

D-1

D-2 D-5

D-3 D-4

Ground surface

Level

Z: depth of pipe

: Peat

Figure 12. Soil cross section and depth of sewage pipes along line B-B in Onbetsu Town.

100





Figure 15. Locations of damaged and intact manholes in Ohtsu district of Toyokoro Town following the 2003

Tokachi-oki earthquake.

Figure 16. An uplifted manhole inToyokoro district

of Toyokoro Town.

Figure 17. An uplifted manhole in Ohtsu district

of Toyokoro Town.

4 THE 2004 NIIGATAKEN-CHUETSU EARTHQUAKE (PARTIALLY QUOTED FROMYASUDA AND KIKU, 2006)

Through these experiences, damage to sewage pipes and manholes due to liquefaction of sand f illwas already known before the 2004 Niigataken-chuetsu earthquake. However, the damage duringthe Niigataken-chuetsu earthquake was much severer than those observed in previous earthquakes.

On October 23, 2004, the Niigataken-chuetsu earthquake MJ = 6.8, occurred. Sewage facilitieswere damaged in 22 cities and towns, as shown in Table 1. The maximum distance from theepicenter to damaged towns was about 30km. The total loss of sewage facilities was valued at 20.6 billionYen. A length of 152.1 km of pipes was damaged. 1,453 manholes and many buried sewage pipes were uplifted. The maximum height of the uplifted manholes was about 1.5 m, as shown inFigure 21. Moreover, 6 sewage water treatment plants and 6 pumping stations were damaged.

102







V V V V

V V V V

V V V V

V V V V

15m

0m

10m

5m

100m0m 50m

No.15-1-3

Z=1.37m

H11 No.8

0

SPT-N value

H11 No.10

Sewage pipe

No.15-1-3

Z=1.37m

H11 No.8

0

f 150mm

No.15-1-2

Z=2.17m

No.15-1-2

Z=2.17m No.15-1-1

Z=2.00m

No.15-2

Z=2.64m

No.15-1

Z=2.90m

No.16-1

Z=2.87m

No.17-3

Z=3.17m

A A'Ground surface

H11 No.10

0

SPT-N value

Level

: Fill : Gravel : Sand : Silt : Clay : Peat : Volcanic soil

vvvvv v

vv v

50

50

Figure 18. Soil cross section and depth of sewage pipes along line A-A’ in Toyokoro district of Toyokoro

Town.

: Fill : Gravel : Sand : Silt : Clay : Peat : Volcanic soil

vv vvv v

vv v

Y Y Y

15m

0m

10m

5m

0m

5m

100m0m50m

No.20-1

Z=6.61m

No.20-3

Z=5.77m

No.17-1

Z=3.82m

f 200mm

Sewage pipe

'BB

No.20-2

Z=5.99m

No.17-2Z=4.29m

No.18-2

Z=4.61m

No.19-2

Z=5.02m

Ground surfaceSewage Center No.1

0

SPT-N value

No.19-1

Z=5.22m

No.20-4

Z=5.51m

No.18-1

Z=4.79m

H5 No.5

0

SPT-N value H3 No.4

0SPT-N value

Level

50

5050

Figure 19. Soil cross section and depth of sewage pipes along line B-B’ in Ohtsu district of Toyokoro Town.

A car collided with an uplifted manhole in Nagaoka City as shown in Figure 22 (TechnicalCommittee on the Sewer Earthquake Countermeasures, 2005). Roads subsided in at 5,908 sites.Figure 23 shows a road subsided in Nagaoka City. The surface of the road subsided by several tensof cm. Therefore, the damage to the sewage manholes and pipes not only prevented the disposal of waste water but also erected obstacles to traffic and restoration activities. Road cave-ins continued after the main shock of the earthquake. At some sites, road cave-ins were noticed a half year after the earthquake, when the snow melted.

The most severely damaged areas were in Ojiya City, Nagaoka City and Kawaguchi Town. Themaximum surface acceleration recorded in these cities and town was about 540 to 1,700 cm/s2.

103




1 2 3 4 5 6 7

50

100

150

200

0

Moiwa districtCentral districtToyokoro districtOhtsu district

Depth of manholes (m)

U

p l i f t ( c m )

Toyokoro Town

Figure 20. Relationship between depth of manholes and height of uplift.

Table 1. Cities and towns where sewage pipes or manholes were damaged during the2004 Niigataken-chuetsuearthquake (Partially quoted from Technical Committee on the Sewer Earthquake Countermeasures, 2005).

Length of sewage Length of damaged Number of Number of

Municipality pipe (km) pipes (km) uplifted manholes caved-in road

Niigata Prefecture 61.3 0.5 51 130

Nagaoka City 1258 62.9 436 3685

Kashiwazaki City 421.5 3.9 12 230

Ojiya City 182.8 31.1 400 349

Tochio City 135.1 2.5 9 20

Mitsuke City 195 0.2 64 315

Koshiji Town 83.7 4.7 93 157

Mishima Town 58.7 1.8 5 16

Y oita Town 56 5.1 88 187

Washima Village 37.7 6.1 36 114

Izumozaki Town 39.6 3.1 5 22

Oguni Town 61.3 9.6 158 107

Toukamachi City 198.1 2.9 10 110

Kawaguchi Town 43 9.3 24 93

Kawanishi Town 29.3 2.4 0 1

Horinouchi Town 75.1 37 93

Sumon Village 48.6 4.3 9 178

Koide Town 88.5 5 19Yahiko Village 100.2 0 0 3

Tsunan Town 59 1.5 0 20

Nakanoshima Town 33.3 0 0 19

Nishiyama Town 25.8 0.3 8 2

Others 3 38

Total 3291.4 152.1 1453 5908

Figure 25 shows the height of uplifted manholes measured by the authors in the Wakaba district of Ojiya City. The manhole shown in figure 24 was raised by 1.06 m at Site A. The road caved in by14cm. Boiled sand was observed on the cave-in. Other manholes were uplifted by 10 to 100 cm and the road caved in at these sites by 10 to 40 cm as indicated in Figure 25. Geomorphologically, thiszone is river terraces formed by the Shinano River. Surface soils in this zone are mainly silt, sand and gravel. Sand boil, which indicate the occurrence of liquefaction was not observed on naturalgrounds in this zone. However, boiled sands were observed above the damaged sewage pipes, asshown in Figure 24. Therefore, it was estimated that soil fill liquefied and caused the uplift of

104




Figure 21. A manhole in Ojiya City uplifted during

the 2004 Niigataken-chuetsu earthquake.

Figure 22. Car that collided with an uplifted man-

hole in Nagaoka City (Technical Committee on the

Sewer Earthquake Countermeasures, 2005).

Figure 23. Subsided road above sewage pipes in

Nagaoka City.

Figure 24. An uplifted manhole at A in Figure 25.

manholes, as in previous earthquakes. The damage in other districts of Ojiya City, Nagaka Cityand Kawaguchi Town was similar.

It is difficult to see the damage of sewage pipes just after an earthquake, because the sewage pipes are buried in the ground. So, the damage was investigated in detail one or two months after the earthquake using special inspection cameras. More damage was found during restoration work.Figure 26 shows a damaged pipe in Nagaoka City. The pipe was uplifted about 50 cm, bent and pulled off at a joint. The pipe was made of vinyl chloride and had a diameter of about 20cm. Thistype of pipe is typical in Nagaoka City.

As the damage to sewer facilities during the Niigataken-chuetsu earthquake was very serious, atechnical committee was organized by the Ministry of Land Infrastructure, Transport and Tourism,to investigate the mechanism of the damage and to select appropriate restoration work (TechnicalCommittee on the Sewer Earthquake Countermeasures, 2005). Detailed soil investigations werecarried out at sites of damaged and undamaged sites in Nagaoka City, Ojiya City and KawaguchiTown. Soil conditions at damaged and intact sites are compared in Table 2. Soil fill in the damaged sites were sand with fines to sand with gravel. Water levels in the soil f ill at the damaged sites wereextremely shallow, such as GL.- 0.2 m to GL.- 1.1 m.The density of the soil fill at the damaged siteswas very low, with degree of compaction DC of 74% to 81% or relative density Dr of 38 % to 41%.

105








Figure 25. Measured uplift and road cave-in in Wakaba district of Ojiya City

Therefore, it is certain that the soil fill liquefied and caused the uplift of manholes. In Nagaoka

City, water table at the undamaged site was deeper than the depth of pipes and manholes. This must be the reason why damage to manholes did not occur at this site. Water tables at the undamaged sites in Ojiya City and Kawaguchi Town were slightly deeper than those at the damaged sites.Thismay be the reason why manholes were not uplifted at these undamaged sites. In Table 2, one moreinteresting point is that the natural soils surrounding the replaced soil f ill were clayey soils, whichare hard to liquefy.

Next, Nagaoka City and Ojiya City were divided into grids of 250 m square and the relation-ships between the damage to sewage pipes and several factors were investigated. The followingrelationships were found:

(a) Damage increased with a decrease in the depth of the water table in clayey ground, as shownin Figure 27.

(b) Rate of damage to sewage pipes increases with the age of construction as shown in Figure 28.This implies the resistance to liquefaction of sand fill increased with age.

(c) The damage to pipes buried under sidewalks was greater than the damage to pipes buried under roadways. The soil fill under roadways must be denser due to cyclic application of traffic roadsthan the fill under sidewalks.

106





Figure 26. Uplifted pipe found during restoration

work in Nagaoka City.

–10

–5

0

0 50 100

Rate of damage to sewage pipes due to

the Niigataken-chuetsu earthquake (%)

D e p t h o f w a t e r t a b l e ( m )

Clayey ground

Figure 27. Relationship between depth of water

table and amount of damage to sewage pipes (Tech-

nical Committee on the Sewer Earthquake Counter-

measures, 2005).

Table 2. Detailed soil investigation conducted at sites of damage and undamaged sites in Nagaoka City, Ojiya

Cityand KawaguchiTown(Partiallyquoted fromTechnical Committeeon Sewer Earthquake Countermeasures,

2005).

Nakazawa in Nagaoka City Sakuramchi in Ojiya City Kawaguchi Town

Damaged Not damaged Damaged Not damaged Damaged Not damaged

site site site site site site

Surface Soil Silt, Silty Silt Sandy silt Sand with silt, Clayey Sandy soil,

natural type sand Sandy silt soil Clayey soil

deposits

up to the

depth of

pipes SPT- N 2 to 4 2 0 to 5 1 — —

valueReplaced Soil Sand Sandy Gravelly Gravelly Gravelly Sand with

soil type with gravel sand with sand with sand with fines

gravel fines fines fines

Water GL-0.65m Deeper GL-1.1m GL-1.38m GL-0.2m GL-0.9m

level than pipe

SPT 11∼ 14 — — — — —

N -value

Degree of — — 74% 78 to 82% 81% —

compaction

Relative 38 to 41% — — — — — density

Damage to Uplift of 40 cm No 8 to 20 cm No 24 cm No

manholes manhole

and road Road 30 cm No 20 cm No 23 cm No

cave-in

107





1985 1990 1995 2000

0

20

40

60

80

400

800

1200

0

1600

Year

R a t e o f d a m a g e t o s e

w a g e p i p e s d u e t o

t h e N i i g a t a k e n - c h u e t s u e a r t h q u a k e ( % )

L e n g t h o f c o n s t r u c t e d p i p e s p e r y e a r

Central area in Ojiya CityRate of damageConstructed length

(km)

Figure 28. Relationship between age of construction of pipe and amount of damage (Technical Committee

on the Sewer Earthquake Countermeasures, 2005).

5 CONCLUSION

Damage to sewage manholes and pipes during the 1993 Kushiro, the 2003 Tokachi and the 2004 Niigataken-chuetsu earthquakes in Japan were discussed. Many sewage manholes and pipes wereuplifted during the three earthquakes. The uplift of manholes and pipes occurred due to the liq-uefaction of sand f ill, a most severe uplift mainly occurred in clayey or peaty grounds. However,more study is necessary on the effect of the soil type of the ground on the uplift of manholes and pipes.

REFERENCES

Japanese Geotechnical Society 1994. Reconnaissance report on the 1993 Kushiro-oki earthquake and the 1993

Notohaoto-oki earthquake. (in Japanese)

Japanese GeotechnicalSociety 2004. Reconnaissance report on the2003Tokachi-oki earthquake. (inJapanese)

Koseki, J., Matsuo, O., Ninomiya,Y. and Yoshida, T. 1997. Uplift of sewer manholes during the 1993 Kushiro-

oki earthquake, Soils and Foundations, Vol.37, No.1: 109–121.

National Research Institute for Earth Science and Disaster Prevention (NIED). K-NET WWW service

(http://www.k-net.bosai.go.jp/)

Technical Committee on the Sewer Earthquake Countermeasures 2005. Report of the Technical Committee onthe Sewer Earthquake Countermeasures, Ministry of Land Infrastructure and Transport. (in Japanese)

Yasuda, S., Nagase, H., Itafuji, S., Sawada, H. And Mine, K. 1994. Shaking table tests on floatation of buried

pipes due to liquefaction of backfill sands, Proc. of from the 5th U.S.-Japan Workshop on Earthquake

Resistant Design of Lifeline Facilities and Countermeasures Against Soil Liquefaction: 666–677.

Yasuda, S., Morimoto, I., Kiku, H. and Tanaka, T. 2004. Reconnaissance report on the damage caused by

three Japanese earthquakes in 2003, Proc. of the 3rd International Conference on Earthquake Geotechnical

Engineering and 11th International Conference on Soil Dynamics & Earthquake Engineering , Keynote

Lecture, Vol.1: 14–21.

Yasuda, S. and Kiku, H. 2006. Uplift of sewage manholes and pipes during the 2004 Niigataken-chuetsu

earthquake, Soils and Foundations, Vol.46, No.6: 885–894.

108








Figure 1. Earthquake epicentre and location of site.

2 OUTCOME OF FIELD SURVEY

2.1 Location

Tanno area in Kitami city is located more than 200 km away from the epicentre, as shown in Fig. 1.It is found that Kitami city is located well within the range of influence of soil liquefaction, based on the empirical correlation between the earthquake magnitude, M, and the radial distance fromthe epicentre, R, within which soil liquefaction can occur, as proposed by Kuribayashi et al. (1974),

which is expressed as follows,

where R (km) is the limiting (largest possible) radial distance from the epicentre within whichsoil liquefaction can occur, and M is the earthquake magnitude determined by Japan Meteorolog-ical Agency. Therefore, the occurrence of soil liquefaction in Tanno area of Kitami city was notempirically beyond expectation.

The topographical map around Tanno area is shown in Fig. 2. The site of the fluidised farmingfield is indicated with the mark of “site investigated”. In the “site A” shown in Fig. 2, the significant

ground deformation and cracks of the sloping farming field were also found.

2.2 Rainfall

Based on the rainfall records at the observatory in Kitami city operated by Japan MeteorologicalAgency, there was precipitation of 17 mm per day on September 19. However, there was no pre-cipitation from September 20 to the time of the earthquake at 4:50 on September 26 in Kitami city.The influence of rainfall should therefore be negligible.

2.3 Seismic acceleration

One of the K-NET (Kyoshin Network) stations operated by National Research Institute for EarthScience and Disaster Prevention (NIED) in Japan was located in Kitami city and was the closestto the site of the fluidised farming field. The strong motion data records observed at this stationare shown in Fig. 3. The components of acceleration in the north-south (NS), east-west (EW) and up-down (UD) directions are plotted in Figs. 3(a), (b) and (c), respectively. The duration of themain shock was about 100 seconds. The maximum values of acceleration were just over 50 gals inthe NS and EW directions, and about 30 gals in the UD directions.

110





Figure 2. Topographical location of site.

Figure 3. Strong motion data records observed at K-NET station in Kitami.

111






Figure 4. Schematic illustration of ground subsidence.

Figure 5. Locations and directions of photographing.

2.4 Field observation

The reconnaissance field survey was carried out on October 18 and 19, 2003. The schematicillustration of the site is shown in Fig. 4. It was one of the gently sloping farming fields raisingwhite beets and green manure. The upstream portion of the site experienced subsidence up to 3.5metres, and the downstream portion was covered by the fluidised debris, which appeared to have

been erupted from two ejection holes located at the middle of the site. The ground water levelwas about 3 metres below the ground surface around this site. It was apparent that the ground subsidence occurred due to the fluidised subsurface soil expelled out from the two ejection holes.The great amount of the fluidised debris thus expelled out on to the ground surface flowed downthe water channel located further downstream, extending to about 1 km long and amounted up toabout 10,000 cubic metres. The site investigated can be visually seen in Figs. 6 to 12. The location

and direction of each photographing are indicated in Fig. 5.The ground deformation and cracks of the sloping farming field at the site A indicated in Fig. 2can also be visually seen in Figs. 13 and 14. The sand boils and cracks of the ground surfaceobserved at the site B indicated in Fig. 5 can also be visually seen in Figs. 15 and 16.

2.5 Soil conditions

The grain size distribution of the soil retrieved from the site is shown in Fig. 17. The fines content,Fc, less than 0.075 mm diameter was 33%, and the specific gravity, Gs, was 2.465, indicating that

112






Figure 6. Location 1. Figure 7. Location 2.



Figure 12. Location 7.

113











Figure 13. Ground deformation at site A. Figure 14. A series of cracks at site A.

Figure 15. Sand boil at site B. Figure 16. Cracks and subsidence at site B.

Figure 17. Grain size distribution of soil.

it contains some pumice aggregates. The values of the maximum and minimum void ratios wereemax = 1.872, and emin = 0.977, as determined by the method stipulated by JGS (2000). The soilwas found to be of local volcanic origin.

From the interview of local people, the subsided farming field corresponded to the area that had been used as a paddy field, however had been reclaimed with the deposits of local volcanic soil

about 30 years before. The subsurface layer was therefore found to be consisted of loosely dumped deposits of volcanic soil.

2.6 Field Swedish weight sounding tests

Swedish weight sounding test is relativelyeasy to handle andto carryout in thefield without any helpof machine. The details of the test equipment are shown in Fig. 18. There are two phases involved in the conduct of tests, i.e. the static penetration and rotational penetration. In typical tests, thescrew-shaped point attached to the tip of the steel rod weighing 49 N (5 kg) is statically penetrated

114









Figure 18. Equipment for Swedish weight sounding tests, (after Tsukamoto et al. 2004).

into the ground by putting several weights stepwise in increments until the total load becomes equalto 980 N (100 kg). At each stage of this load increment, the depth of penetration is measured. Theweight at each load application is denoted as Wsw. When the static penetration ceases, the rotational

phase of penetration is performed by invoking the rotation to the horizontal bar f ixed at the top of the rod. The number of half a turn necessary to penetrate the rod through 1 metre is denoted as Nsw

(ht/m).The values of penetration resistance thus measured by Swedish weight sounding tests can be

converted to SPT N-values by employing the empirical expressions. There are several empiricalcorrelations proposed in the past literature. Inada (1960) proposed the following correlations,

where Wsw is expressed in the unit of N. The most recently proposed empirical expression is found in Tsukamoto et al. (2004) as follows,

The above correlations take into consideration the effect of static penetration. In the correlation (4),the effect of static penetration is represented by the equivalent N sw-value of 40. It is also noticed

that this correlation depends upon the grain characteristics represented by the void ratio range,emax − emin. The value of N sw can also be directly converted to the relative density Dr of sandy soils,(Tsukamoto et al. 2004), as follows,

where σ

v is in kPa.

115





Figure 19. Positions of field tests.

Figure 20. Soil profile estimated from field tests at cross section A-A.

Figure 21. Soil profile estimated from field tests at cross section B-B.

Figure 22. Soil profile estimated from field tests at cross section C-C.

Multiple series of Swedish weight sounding tests were conducted to estimate the subsurface soil profiles at several cross sections. The positions at which the sounding tests were conducted areidentified in Fig. 19. The data of the sounding tests are listed in Tables 1 to 13 of APPENDIX. Thesoil profiles at five cross sections are estimated as shown in Figs. 20 to 24. The weak subsurfacesoil layer can still be detected. It would be reasonable to assume that the overlying unsaturated topstratum at the upstream portion prohibited the underlying liquefied layer from blowing directly

116








Figure 23. Soil profile estimated from field tests at cross section D-D.

Figure 24. Soil profile estimated from field tests at cross section E-E.

Figure 25. Progressive sequence of fluidisation and subsidence.

upwards, which was rather expelled from the weak portion of the top stratum corresponding to thetwo ejection holes located downstream, as indicated in Fig. 4.

2.7 Progressive sequence of fluidisation and subsidence

From the inspection on the non-uniform complexity of the subsided ground surface, it was apparentthat the fluidisation and ground subsidence occurred in a progressive manner. It would be reasonableto assume that the flow deformation of the fluidised subsurface soil layer occurred in a manner as shown in Fig. 25. The directions of flow deformation seem to have oscillated rightwards and leftwards as it propagated from downstream to upsteam portions. This might have been caused bythe valley-shaped bottom surface of the base layer underlying the fluidized deposits.

117







3 CONCLUSIONS

The reconnaissance field survey was carried out at the site where the fluidisation and subsidence of gently sloped farming fields took place during 2003 Tokachi-oki Earthquake in Japan. The site waslocated in Tanno area of Kitami city in Hokkaido. From the site inspection, it was found that thefarming field with 35 metres wide and 150 metres long experienced subsidence up to 3.5 metres,

and the water channel located downstream was filled with the fluidised debris for a length of about1 km. It was apparent that the subsurface deposits of loosely dumped volcanic soil liquefied and erupted onto the surface, which caused subsidence during the earthquake. Based on the series of Swedish weight sounding tests, the subsurface soil profiles at several cross sections were estimated.The progressive sequence of fluidisation and flow of subsurface deposits and associated ground subsidence was discussed.

ACKNOWLEDGEMENTS

The authors express sincere appreciation to Professors. M. Suzuki, S. Yamashita, Y. Ito of KitamiInstitute of Technology for their help in carrying out the reconnaissance survey. Thanks are alsoextended to Mr. R. Yamaguchi and the past students of Kitami Institute of Technology for their cooperation in carrying out the field tests described in the present study. The strong motion datarecords shown in Fig. 3 were retrieved from Kyoshin Network operated by National ResearchInstitute for Earth Science and Disaster Prevention (NIED).

REFERENCES

Inada, T. 1960. On the use of Swedish weight sounding test results. Domestic journal of Japanese Geotechnical

Society, Tsuchi-to-Kiso, Vol. 8, No. 1, 13–18, (in Japanese).Japanese Geotechnical Society. 2000. Testing methods and interpretations of geotechnical laboratory

experiments.

Kuribayashi, E., Tatsuoka, F. & Yoshida, S. 1974. History of soil liquefaction phenomena observed after Meiji

era in Japan. Report of Public Works Research Institute, No. 30.

Tsukamoto, Y., Ishihara, K. & Sawada, S. 2004. Correlation between penetration resistance of Swedish weight

sounding tests and SPT blow counts in sandy soils. Soils and Foundations, Vol. 44, No. 3, 13–24.

APPENDIX

The data of Swedish weight sounding tests conducted in the present study are listed inTables 1 to 13.

118




Figure 2. Plan of Kushiro Port (January, 1993).

2 DAMAGE TO PORT FACILITIES DURING 1993 KUSHIRO-OKI EARTHQUAKE

Kushiro-Oki Earthquake (MJ = 7.5: Japanese magnitude scale) occurred on January 15, 1993.The focus of this earthquake was about 10 km south from the Kushiro Port, Hokkaido Island,Japan (42.920◦ N, 144.353◦E) and 101 km deep. Kushiro Port was shaken with a peak horizontalacceleration of 0.47 g.

As shown in Figure 2, the Kushiro Port is located in the mouth of the Kushiro River, south eastern part of Hokkaido Island, and sand dune is formed by sand supplied from the Kushiro River. Quaywalls were constructed on relatively dense seabed deposits from the sand dune, and backfilled with dredged sand. The thick dotted line showed the estimated beach line. Most of the port has

been constructed step by step from the quay wall in the east end toward west direction along the beach line.

Details of damaged port facilities in this earthquake are available elsewhere (Iai, et. al. 1994).

2.1 Damage to sheet pile quay walls

The most serious damage to the quay walls in the Kushiro Port was found at the Fishery Pier inthe East Port as shown in Figure 2. The typical damage of the quay wall with the design seismiccoefficients of 0.2 is illustrated in Figure 3. The steel sheet pile was connected with battered steel

piles using anchor rods, and the backfill soil behind the pier consists of a loosely-deposited sand layer with a thickness of about 10 m. Due to the liquefaction of backfill sand, the steel sheet pilewas deformed as shown in Figure 3, and the apron exhibited maximum 1 m settlement. According

to the deformation, the sheet pile was suffered open cracks at about −

4 m below the low water level.On the other hand, there was no damage observed in steel sheet pile quay walls with the seismic

coefficients of 0.2 at Pier No.1 shown in Figure 4. The backfill soils were improved by graveldrain columns and sand compaction piles which are the most typical countermeasures againstliquefaction in Japan. The sand compaction piles have been widely used in order to densify, and increase liquefaction strength of the backfill ground, and the gravel drain method is usually used near structures such as sheet pile and rubble backfill because the compaction method may inducedeformation on the structures.

120





Figure 3. Cross section of a steel sheet pile quay wall at Fishery Pier.

Figure 4. Cross section of a steel sheet pile quay wall at Pier No.1 in West Port.

In comparison of damage conditions on these two steel sheet pile quay walls with and withoutground improvements, it could be verified these methods are effective countermeasures to controldeformation and damage level on the quay walls.

2.2 Damage to gravity type quay walls

In the Pier No.2 of West Port, 12 m water depth gravity type quay walls with the seismic coefficientsof 0.2 were damaged as shown in Figure 5 during the earthquake. The caisson was moved towardssea about 26 cm and settled about 40 cm due to seaward movement of the caisson. The relative leveldifference of caisson and apron pavement surface was about 40 cm. Some trace of liquefaction of

backfill soil was discovered around the apron. In order to ensure transportation by tracks, the leveldifference between the caisson and apron had to be 25 cm or less. Although the damage level issmall from a viewpoint of the stability of caisson type quay wall, it was not acceptable damagefrom a viewpoint of the function of the quay wall.

121






Figure 5. Cross section of a gravity type quay wall at Pier No.2 in West Port.

2.3 Restoration works

To make plan the restoration works, there are some key factors to be taken into account, such as;

1) Influences due to change of a quay wall face line when needed 2) Availability/restriction of construction area3) Time necessary for completing restoration work 4) Necessity of damaged quay wall operation during restoration work 5) Environmental considerations including effective usage of debris from damaged region6) Cost performance

Considering the above, the following restoration concepts of damaged quay walls were adopted in the Kushiro Port;

a) Soil improvement:To reduce earth pressure considering liquefaction by soil improvement method.

b) Weight increase:In case of gravity caisson type structure, to increase the resistance against sliding by usingheavier stuffing materials or extending structure size with additional caisson box or concrete.

c) Additional structure:To increase the resistance by adding new structure.

The quay wall shown in Figure 3 was completely destructed, the restoration design adopted newcaisson type quay wall installed in front of the damaged steel sheet pile quay wall as shown inFigure 6. The caisson was backfilled with stone as a remediation measure against liquefaction, and the backfill behind the damaged steel sheet pile quay wall was partially improved by gravel drainas shown in Figure 4.

3 DAMAGE TO PORT FACILITIES DURING 1995 KOBE EARTHQUAKE

The 1995 Hyogoken-Nambu Earthquake (MJ =7.2) occurred at 5:46 a.m. on January 17, 1995.The focus of this earthquake was about 20 km west-southwest from the Kobe Port and 16 km deep,

122





Figure 6. Restored cross section of damaged steel sheet pile quay wall at Fishery wharf.

Figure 7. Plan of Kobe Port.

very near from the Kobe Port. This earthquake generated very strong seismic motion; the peak ground accelerations recorded at the Kobe Port were 0.54 g and 0.45 g in the horizontal and verticaldirections, respectively. Most of the major facilities of the Kobe Port were significantly damaged,

and this destruction seriously affected operation of the logistic system through and around the port.Long-term suspension of the port operation gave serious blow to the regional economy, especiallyKobe City economy, for several years.

The Kobe Port covers an area about 6 km long by 12 km wide including two man-maid islandsas shown in Figure 7. The two man-made islands, Port Island and Rokko Island, were constructed from 1966 to 1981 in Phase I and from 1972 to 1990 in Phase II, respectively. Decomposed granite,called Masado, used for landfill, was excavated from Rokko Mountains, transported and placed by

bottom dump type barges in the sea with a water depth ranging from 10 to 15 m.More details are available in literatures (e.g. Iai, et. al. 1996, Inagaki et. al. 1996).

123






Figure 8. Deformation of quay wall at Rokko Island.

3.1 Typical damage to caisson type quay walls

Only the gravity caisson type quay wall was used for the Port and Rokko Islands project. Theseismic coefficients used for design were 0.1 and 0.15 depending on importance of the facilities.

One of the typical damaged quay walls in Rokko Island with the seismic coefficient of 0.1 is shownin Figure 8. The quay wall displaced toward sea about 4 to 5 m, settled about 2 m, and tilted about4 to 5 degrees on average. The backfill soil behind the wall subsided accordingly in the same order of magnitude as those in the horizontal wall movement. No collapse or overturning occurred duringthe earthquake. Another important fact of the damage was that the deformation of the caisson wallsoccurred uniformly maintaining almost straight face lines of the walls even with a few meters of the absolute deformation. The uniform deformation is considered to reflect the uniformity of thelandfill and the underlying replacing material.

There were three quay walls, PC13, PC14 and PC15, constructed along a straight face line side by side at Port Island Phase II. One was constructed on sand replacement foundation shown in

Figure 9 and two were constructed on a clay foundation which was improved with sand compaction piles (SCP) shown in Figure 10. The seismic coefficients used were 0.18 for PC13 and 0.15 for PC14 and 15. The horizontal displacements and elevations of these quay walls after the earthquakeare shown in Figure 11. In the figure, the initial elevations of the caissons were assumed the samedespite the difference in their construction time.

As shown in Figure 11, the displacements of the quay walls PC14 and 15 with SCP founda-tion are about 2.5 m and 0.3 m in horizontal and vertical directions, respectively. In contrast, theyare about 3.5 m and 1.5 m in PC13, much larger than in PC14 and 15, suggesting the effect of different foundation improvements. Since SCP improves the ground with vertical sand piles, theeffect is more remarkable in reduction of settlement than horizontal displacement. Both horizontaland vertical displacements are larger in PC13 with larger unevenness than in PC14 and 15. This

difference is considered to demonstrate the effectiveness of SCP for caisson wall foundation.

3.2 Typical restoration methods

As mentioned above, the most appropriate restoration methods were selected based on avail-ability/restriction of construction area, construction period, cost, construction materials and the extent of damage to individual facilities. Earthquake resistance was upgraded according to theimportance of facilities in the restoration. The seismic performance of restored port facilities wereevaluated by numerical or experimental simulations.

124





Figure 9. Cross section of a quay wall founded on sand replacement.

Figure 10. Cross section of a quay wall founded on subsoil improved with sand compaction piles.

The basic restoration techniques are summarized as follows:

1) Reduce the earth pressure acting on the structure2) Put additional structure to the damaged structure3) Replace the liquefiable soil

Various restoration methods were adopted in the Kobe Port as shown in Figures 12 through14 corresponding to structural and geotechnical conditions, importance and functions of the quaywalls. It is notable precast structures such as jackets, were effective to shorten the recovery time,as shown in Figure 14 because of the jacket main body is produced at factory, the site construction

period can be shortened.

4 DAMAGE TO PORT FACILITIES DURING 2003 TOKACHI-OKI EARTHQUAKE

Tokachi-Oki Earthquake (MJ = 8.0) occurred on September 26, 2003. The focus of this earth-quake was 41.778◦ N, 144.078◦E and about 45 km deep. The Kushiro Port was shaken with a peak

125






Figure 13. Detached structure type (caisson).

Figure 14. Pre-cast jacket type quay wall.

and had been completed in 2002. Dredged fine sand was used as a reclamation material and the backfill soil behind the wall was improved by cement solidification method. The backfillsoil was not liquefied but the quay wall was displaced toward sea about 0.2 m maximum duringthe earthquake. As shown in Figures 18 and 19, about 0.6 m–0.9 m differential settlement at the

boundary of caisson and apron was observed and deep vertical crack was located at toe of back f illrubble slope, too. The failure process of the gravity type quay wall is presumed as follows,

1) Due to the earthquake motion, the caisson body moved towards sea slightly,2) Confining pressure of backfill rubble suddenly decrease accordingly,3) Backfill rubble collapsed easily because of no confining condition,4) Cement treated backfill soil rotated and the deep vertical crack was generated, and 5) About 0.6 m–0.9 m differential settlement appeared, at the boundary of caisson and apron.

The restoration design considered with tow keywords as ‘confining pressure’, ‘ductility of soil’.The rotated cement treated backfill soil body was crushed and refilled sandy soil up o prescribed level.

127






Figure 15. Comparison of the acceleration time history of 1993 Kushiro-Oki and 2003 Tokachi-Oki

Earthquakes (at the Kushiro Port, ground surface).


The aforementioned quay wall damage suggests that:

1) Damage to port structures is often associated with significant deformation of a soft or lique-fiable soil deposit. Therefore, if any potential for liquefaction exits, implementing appropriateremediation measures may be an effective approach to attain significantly improved seismic

performance of port structures.2) Most of the structural failures result from excessive deformation rather than catastrophic col-

lapses. Hence design method based on displacements and ultimate stress states are desirable over conventional force based design methods for defining the comprehensive seismic performanceof port structures.

3) Most damage to port structures is the results of soil-structure interaction during the earthquakeshaking. Therefore, seismic analysis and design should also take into the account, both thegeotechnical and structural aspects of the port structures.

Based on the lessons from the damage during the 1995 Hyogoken-Nambu earthquake, a seismicPerformance-Based-Design methodology (Japan Port and Harbour Association, 1999) was intro-duced. In order to assess/evaluate the seismic performance, new simulation techniques need to beintroduced in the technical standards for port facilities in Japan. However, in practice, it is noteasy to incorporate simulation techniques such as dynamic analyses and model tests. To brush up

128





Figure 16. Comparison of the Fourier spectrum of 1993 Kushiro-Oki and 2003 Tokachi-Oki Earthquakes

(at the Kushiro Port, ground surface).

Table 1. Peak accelerations of 1993 Kushiro-Oki and 2003 Tokachi-Oki Earthquakes.

Peak acceleration (Gal)

EW NS UD

2003 Tokachi-Oki Eq. Surface 576 347 149

Downhole 202 154 66

1993 Kushiro-Oki Eq. Surface 343 450 362

Downhole 268 203 121

Figure 17. Plan of the Kushiro West Port (September, 2003).

129






Figure 18. Cross section of damaged caisson type quay wall at Pier No.4 (−14 m).

Figure 19. Damage of the caisson type quay wall at Pier No.4 (−14 m).

the seismic PBD methodology, it is still necessary to collect actual case history data as well asmodel test data and numerical simulation data, and feed them back to practice with appropriateinterpretation.

REFERENCES

Iai, S., Matsunaga,Y., Morita, T., Miyata, M., Sakurai, H., Oishi, H., Ogura, H., Ando,Y., Tanaka,Y. and Kato,M. 1994. Effects of remedial measures against liquefaction at 1993 Kushiro-Oki earthquake, Proc. 5th

US-Japan Workshop on Earthquake Resistant Design of Lifeline Facilities and Countermeasures against

Soil Liquefaction, NCEER-94-0026, National Center for Earthquake Engineering Research: 135–152.

Iai, S., Sugano, T., Ichii, K., Morita, T., Inagaki, H. and Inatomi, T. 1996 Performance of caisson type quay

walls. The 1995 Hyogoken-Nanbu Earthquake, -Investigation into Damage to Civil Engineering Structures-.

Japan Society of Civil Engineers: 181–207.

Inagaki, H., Iai, S., Sugano, T., Yamazaki, H. & Inatomi, T. 1996. Performance of caisson type quay walls at

Kobe port. Soils and Foundations. Special Issue on GeotechnicalAspects of the January 17 1995 Hyogoken-

Nambu Earth-quake: 119–136.

International Navigation Association, 2000. Seismic Design Guidelines for Port Structures:.8. Rotterdam:

BalkemaJapan Port and Harbour Association. 1999. Technical Standard for Port and Harbour Facilities and

Commentaries. (in Japanese)

130







River dike failures during the 1993 Kushiro-oki earthquake and the

2003 Tokachi-oki earthquake

Y. Sasaki Hiroshima University, Japan Japan Institute of Construction Engineering, Tokyo, Japan

ABSTRACT: This paper describes damages to river dikes during the 1993 Kushiro-oki Earth-quake and the 2003 Tokachi-oki Earthquake. During the Kushiro-oki Earthquake in 1993, it wasfound that dike failures were induced by liquefaction which occurred within a submerged part of fill.This condition was brought by consolidation of a peat layer beneath dikes. Dike failure in this caseshowed the importance of drainage to prevent the infiltrated rainwater from accumulating withina dike body. During the 2003 Tokachi-oki Earthquake an unusual type of damage was detected.Vertical cracks were found within a dike section at a location where no apparent abnormality wasvisible at the surface. Evidence suggests that the cracks were caused by expansion of the bottomwidth of the dike.

1 INTRODUCTION

Japan has suffered from recurrence of earthquake disasters since ancient days, and damageshave been frequently inflicted to houses, structures, river dikes and other facilities. This paper describes investigations of river dike failures during the 1993 Kushiro-oki and the 2003 Tokachi-okiEarthquakes in Hokkaido.

Hokkaido in Japan has suffered from repetitive seismic damages to flood protection dikes espe-cially from large scale earthquakes off its southeast coast which faces the subduction zone betweenthe Pacific and North American Plates (Table A-1 in the attached CD-ROM).

Very valuable data on dike failures have been accumulated in the last 15 years. They include thedata obtained at locations along the Shiribeshi-Toshibetsu River after the 1993 Hokkaido-nansei-oki Earthquake (Sasaki et al. 1997) which revealed the failure caused by stretching of the bottomof the dike. Lowering the water level within the dike body was effective in mitigating earthquakedamage during the 1994 Hokkaido-toho-oki Earthquake. Valuable lessons have been gained alsofrom damages during the Hyogo-ken Nanbu Earthquake (Sasaki & Shimada 1997, Matsuo 1996),Miyagi-ken Hokubu Earthquake (Nakayama et al. 2007) and some cases caused by the 2004

Niigata-ken Chuetsu (Oshiki & Sasaki 2006) and the 2007 Niigata-ken Chuetsu-oki Earthquakestill present (March, 2008).

This paper focuses on failures during the 1993 Kushiro-oki Earthquake and the 2003 Tokachi-oki

Earthquake because of their special features. In the 1993 earthquake, the first case of failure caused by liquefaction in the body of the dike was noted. The dike was built on a soft peat deposit. In the2003 earthquake, a very interesting type of failure was discovered. The bottom section of the dikewas stretched and cracks opened in the body of the dike at a section of the dike with no visibleabnormality on its surface.

Stretch type deformation is considered to adversely affect the water-cutoff performance of adike especially in raining condition, but its engineering features are not well studied. It is oneof the phenomena that should be elucidated in order to appropriately assess the flood protection

performances of a river dike in near future.

131




Table 1. Damages to river facilities during the 1993 Kushiro-oki Earthquake.

Dike failure Revetment damage

River No. of sections Length (m) No. of sections Length (m)

Tokachi River 20 9,168 – –

Kushiro River 18 10,124 10 1,080

Onbetsu River 1 2,002 1 90

Watenbetsu R. 11 1,459 1 230

Shibetsu River 1 3,553 – –

Total 51 26,306 12 1,400

2 RIVER DIKE FAILURE DURING THE 1993 KUSHIRO-OKI EARTHQUAKE

2.1 Summary of damages to river dikes

The 1993 Kushiro-oki Earthquake (M= 7.8) occurred at about 20:06 on January 15, 1993. Seismicintensity of 6 in Japan Meteorological Agency (JMA) scale was recorded at Kushiro City, and 5 in JMA scale was recorded at Obihiro, Hiro-o, Uraga and Hachinohe. Two persons were killed

by the earthquake, 967 were injured, and many dwelling houses and road sections were damaged (National Astronomy Observatory 2007).

The earthquake is reported to have occurred inside the Pacific Plate, which subducted under the North American Plate, and the hypocenter was as deep as 100 km. According to the seismometer

installed at the Kushiro Meteorological Observatory, which was 14 km from the epicenter, themaximum acceleration was 711 gal in N063E direction, 637 gal in N153E direction, and 363 galvertically (BRI 1998, see Figure A-1 in the CD-ROM). The duration of the intense motion wasabout 30 seconds (the duration of acceleration in N063E direction exceeding 50 gal was about55 seconds). The values were possibly amplified by the effects of the soil stratification and thetopography at which the observatory was located.

The epicenter of the 1993 Kushiro-oki Earthquake is shown in Figure 1 together with the peak ground accelerations (PGA) recorded by strong-motion seismographs (PWRI 1998, see Table A-2 in the CD-ROM). The river damaged by the 1993 Kushiro-oki Earthquake (KDCO 1994 and ODCO 1994) and the epicenter of the 2003 Tokachi-oki Earthquakes are also shown in this figure.At Hirosato Observatory (Kushiro EMB), which was located near the end of Kushiro Retarding

Basin, complete record was not available due to malfunction of the paper feeder (PWRI 1994),therefore the maximum amplitude recorded (320 gal) is shown in the figure.

The earthquake caused damages to two “first class rivers” and three “second class rivers” (des-ignated rivers), which are under the management of the Hokkaido Development Bureau (HDB), at51 dike sections in a total length of 26.306 km and 12 revetment sections over a total length of 1.4km (counted damages were as serious as they needed the special fund from the disaster restorationworks budget). The damage caused by this event is shown in Table 1 for each river.

It was known that 73% of the damaged length occurred along the Tokachi and Kushiro Rivers.Along the Tokachi River, damaged section were dispersed in a distance from the river mouth tothe 33.3 km point along the main stream, amounting to 7 km out of total length of 9 km which

failed along the Tokachi River dikes, and the length of 2 km was damaged along branch rivers(damaged 9 km corresponded to 27% of the total length of the main river course). Particularlyseriously deformed sections were observed from 31.77 to 33.26 km section on the right side bank (dike height of about 7 m).

Along the Kushiro River, dikes were damaged over a length of 10.124 km with sections of 1.08 km of revetment damage. Among the 10.124 km length of failed sections of dike, length of 2.4 km was caused at the lower reaches, which accounted for 31% of the total length of the dike inthis section. Among the 10.124 km length, 6.898 km was induced in the dikes of Kushiro RetardingBasin. 2.2 km failed on the left side bank of the Retarding Basin dike, accounting 74% of the total

132




Figure 1. PGA during the 1993 Kushiro-oki Earthquake and damaged rivers.

3 km on the left side bank, and 4.6 km on the right side bank of the Retarding Basin dike, whichaccounted for 57% of the total 8.1 km of the right side bank.

Locations and lengths of damaged section along those five rivers are shown in Table A-3 in theCD-ROM.

An example of the dike failure at 33.2 km point on the right side bank of the Tokachi River isshown in Figure 2. Damages at the left and right side banks along Kushiro Retarding Basin areshown in Figure 3. At those sections, the crests of the dikes subsided, and the crest surfaces inclined.Deep cracks were generated near the top of the slopes and berms.

A summary of the damage modes along the Tokachi and Kushiro Rivers is shown in Table 2.The table, which summarizes the damaged dike lengths along the Tokachi and Kushiro Rivers for

each damage mode, shows no traverse cracks (Mode 3). However, it should be noted that severaltraverse cracks like shown in Figure 4 were generated near the ends of longitudinal cracks but arenot classified as the representative mode of the sections. The length of failed sections that suffered especially serious damage and needed re-compaction over the entire width (Type 6) was 700 malong the Tokachi River and 1,601 m along the Kushiro River. The damage was more serious alongthe Kushiro River than along the Tokachi River, since it was closer to the epicenter.

An aerial photograph of the affected dike section on the left side bank in the Kushiro Marshis shown in Figure 5. The photograph clearly shows the damaged and the non-damaged sectionsappearingone after another. The intermittent appearance of damaged sections was likely attributable

133





Figure 2. Failure of the Tokachi River dike at Tohnai at its right side bank around Kp 33.2.

Figure 3. Typical view of the failed dike of the Kushiro River.

134






Table 2. Failure mode of Tokachi and Kushiro Rivers.

Tokachi River Kushiro River

Failure Length Ratio Length Ratio

mode (m) (%) (m) (%)

1 375 4.1 30 0.3

2 8,085 88.2 1,122 11.1

3 – – – –

4 8 0.1 – –

5 – – 7,371 72.8

6 700 7.6 1,601 15.8

Total 9,168 100 10,124 100

Figure 4. Transverse crack observed at the left side bank of the Kushiro Retarding Basin dike.

Figure 5. Aerial view of the damaged section at its left bank of the Kushiro River dike in the Kushiro Marsh.

135







to the three-dimensional dynamic response of the dike (Kano et al. 2007) because the conditionsof the dike and the ground were uniform throughout the range shown in the photograph.

The following sections describe the damage features and restoration works of dikes along KushiroRetarding Basin, which were typical failures of river dikes during this earthquake.

2.2 Dike damage along the Kushiro Retarding Basin

Figure 6 shows a plan of the dike from the river mouth to middle of the Kushiro Retarding Basin.The figure also shows the sections damaged during the 1993 Kushiro-oki Earthquake, the

schematic view of the cross sections showing the ranges of re-compaction during restoration fromits damage, and the sections damaged during the 1994 Hokkaido-toho-oki Earthquake. The sec-tions shown by solid line are the damaged area during the 1993 Kushiro-oki Earthquake, and thoseshown by dotted line are the area damaged again during the 1994 Hokkaido-toho-oki Earthquake(M= 8), which occurred a half year later after the completion of the restoration works.

2.2.1 Typical cross section of the dike and construction historyThe main stream of Kushiro River starts from Lake Kussharo-ko (water level elevation: 121 m),

merges with several branch rivers, and flows south through Kushiro City to the Pacific. The river isthe first class river that has a stream length of 154 km and a basin area of 2,510 km2. The KushiroRiver flows through a plane including Kushiro Marsh (elevation: 2 to 6 m) in its downstream for 67 km length (south from Shibecha town). The mean river bed inclination is 1/4,500 at the lower

Figure 6. Failure Sections of the Kushiro River Dike.

136





reaches, and is 1/1,200 at upstream of Shibecha town. Kushiro City, which is located near the river mouth, suffered serious disaster of flooding in October 1917 and August 1920, and flood controlworks was initiated. In order to control floods, a floodway was constructed from Iwabokki, and river dikes were constructed. In June 1931, the left side bank was completed, and the right side

bank was completed in 1934.Since floods occurred several times thereafter, the river improvement plan was revised in 1966

to change its design discharge and to use Kushiro Marsh as a retarding basin. In 1967, the KushiroRiver was designated as a first class river. In 1984, the basic plan for the implementation of construction, which was formulated in 1967, was revised. Typical cross sections of dikes along theRetarding Basin and their construction history since 1952 are shown in Figure 7 (KDCO 1994). Thedike had a crest elevation of 9.3 m (dike height was about 7 m), a crest width of 8 m, a berm widthof 3 m, slope inclination of 1:2 and a base width of about 45–50 m (fill materials are described later) at the time of the Earthquake. It is noted that there had been sections in the right side bank where sand mat had been placed at the landside bottom of the enlarged slope.

2.2.2 Geotechnical aspects of the Kushiro MarshKushiro Marsh is a national protected geologic formation and a wild life protection area designated

in 1967, which was later registered in the Ramsar Convention in 1980, and designated as a national park in 1987. As the site was in this location, special care was needed in maintaining river facilitiesso that nature is preserved.

The Marsh is covered by peat layer at its surface, and the thickness of the peat deposit variesfrom place to place. The thickness is over 4 m in the middle of the Retarding Basin and is about 3 min the other zones of the Retarding Basin. In general, the peat deposit is thick in the northern and northwestern zones and is thin in the south. An example of top layers at the Marsh is shown in Figure8 (Soil stratifications along dike axes in detail are shown in Figures A-2 and A-3 in the CD-ROM).

The thickness of the peat deposits was 2 to 3 m at the left bank side and 2 to 6 m at the right bank side. At the left side bank, the natural water content was 134 to 494%, the ignition loss was

26 to 77%, the soil particle density was 1.68 to 2.27 g/cm3

, the dry density was 0.18 to 0.23, and

Figure 7. Typical cross sections and their construction history of the Kushiro Dike

Figure 8. Stratification of surface soil.

137






the natural void ratio was 6.45 to 9.29. At the right side bank, the natural water content was 120 to1052%, the ignition loss was 20 to 95%, the soil particle density was 1.66 to 2.30 g/cm3, the drydensity was 0.12, and the natural void ratio was 12.93.

There are only a few technical papers on the dynamic properties of peat however, it was knownfrom a series of cyclic tests of peat (Noto 1993) that no strength loss like that caused by soilliquefaction in sand occurred in peat during cyclic loading.

The peat layer in the Kushiro Marsh is underlain by Holocene (alluvial) sand deposits and alluvialclay strata as shown in Figure 9. The thickness of the sandy layer beneath the peat varies from east towest, generally is thicker at left bank aide than at right bank side. The SPT blow count of this layer varies 3-50 over the whole area. The stiff layer (N > 50) appeared from around 30 m deep beneaththe ground surface at around the 9 km 850 point of the left side bank of the Retarding Basin dike.

Compression settlements of the peat layer due to embankment load were calculated using thee-logp curve given in the guideline for improving soft peat deposits (Noto 1991). Computed set-tlements were compared with the actual settlements identified from boring data. The range of consolidation settlement in the damaged section was about 0.7–1.5 m at the left side bank of theRetarding Basin dike and was about 0.5–2 m at the right side bank (HDB 1997).

Figure 9. Geological condition of the Kushiro River along Lines A-A’, B-B’ shown in Figure 6.

138





2.2.3 Distribution of the crest settlement The residual crest height after the 1993 Kushiro-oki Earthquake is shown in Figure 10. Prior to thisearthquake, the Tokachi-oki Earthquake in 1952 and the Nemuro-hanto-oki Earthquake in 1973 hitthe area. Dikes in Kushiro were damaged during the Tokachi-oki Earthquake in 1952 (M = 8.2).Figure 10 also shows the sections that subsided during the Tokachi-oki Earthquake in 1952.

The sections damaged during the Tokachi-oki Earthquake in 1952 and 1993 Kushiro-oki Earth-

quake differ. The dike had smaller cross sections in 1952 than in 1993 as shown in Figure 7.However, it is known that the maximum settlement during the earthquake in 1952 was as large asthe settlement induced by the earthquake in 1993, and the length of the damaged sections was alsolarge. The settlement in the Marsh was particularly more serious than in the lower reaches.

Sections of serious damage occurred during the 1993 Kushiro-oki Earthquake concentrated inthe middle to upper reaches in the Marsh unlike damages during the earthquake in 1953. Thedifference was possibly attributed to the differences in earthquake motion, but the mechanism of the effects has not been studied well.

Several meandering former river courses were found in the Marsh and cross the left side bank at some points. However, the damage during the 1993 earthquake was relatively small at places on

the former river courses, so it was concluded that their effects were negligible (KDCO 1994).

2.2.4 Damage morphologyAs shown in Figures 2 and 3, dikes were damaged with several lines of longitudinal cracks. Thecracks opened up significantly at some points. The maximum open width was 6 m on the right side

bank from section 9 km 360 to 9 km 500, Section Number 12 (shown in Table A-3 in the CD-ROM).Level differences were observed between both sides of the cracks.

Deformation at the toe of the left side bank is shown in Figure 11. At the lowest end of the failed slope, soil mass of about 1 m in length of the surface around the toe turned almost vertically asshown by the poles, showing that the soil bellow slope was pushed horizontally towards outside.But there was neither heaving nor any trace of ground movement on surrounding field.

This deformation indicated that a large deformation occurred not in the natural ground outsidethe dike width but either within the dike body or the ground directly under the dike.

Cross sectional survey was conducted at 40 sections of the left side bank from 5 km 800 to 10 km500 and at 36 sections of the right side bank for a length of 9 km 720 m (from 5 km 680 to 15 km400). Consequently cross section profiles about 100 m interval were obtained. Utilizing the resultsof the survey (Table A-4 in the CD-ROM), the elevations at the crest center and the lowest points

Figure 10. Settlement of dike crest of the Kushiro Dike.

139





Figure 11. Feature of riverside toe of failed dike.

Figure 12. Results of survey.

in the crest were shown in Figure 10. The elevations at the crest center and the lowest point werenot always equal.

Settlement of crest means a partial decrease in the sectional area of the dike, and the bulging of

the slope like shown in Figure 11 means a partial increase in sectional area. A simple accumulationof these amounts (Figure 12(b)) gives changes in sectional area. The net changes of cross sectionalarea are plotted against the maximum settlement of the crest in Figure 12(a).

Although precise data on cross sectional areas before the earthquake were not available and Figure 12(a) only shows the change of the area at a surveyed cross section, which can not representthe change of entire volume at damaged sections, however the following tendencies can be noted in Figure 12(a). On the right side bank, the reduction in the cross sectional area by settlementwas larger than the increase in the area caused by expansion, which resulted in a reduction of cross sectional area. On the other hand, the left side bank showed large bulging, even though largesettlement of the crest was also observed, which resulted in an increase in cross sectional area. Thissuggested that the damage morphologies differed between the right and left side banks. The actualcauses of the difference are not clear but are possibly attributable to the differences in the thicknessof the peat layer and the fact that the right side bank was once used as a road, and a sand mat was

placed under the widened bank section on the landward side of the slope.

2.2.5 Cracks within dike bodyFigure 13 shows an example of an internal view of a cross section of the dike observed at 9 km850 point of the left side bank during an open-cut examination. Dike sections including this sitewere entirely re-filled during restoration works. Cracks, deformation and soil compositions were

140






Figure 13. Open-cut Investigation of failed Dike (at 9km850 on Left bank).

Figure 14. Distribution of Cracks in Dike.

investigated and soil specimens were sampled from the cut surface utilizing the opportunities of removing the damaged portion of the fill. Opened cracks running vertically are shown in Figure13. It was also recognized that traces of liquefied sand intruded into some of these cracks (SeeFigure A-4 in the CD-ROM).

Distributions of cracks like those shown in Figure 13 are summarized in Figure 14 for different

cross section (in height and width) of the dikes. Among the sections, sketches of Type “a” and “e” section were gathered by executing entire open-cut investigation during restoration work. Thesketches of other types result from a synthesis of the observation records at excavating pits, whereinvestigations were conducted at some selected points to confirm the cracks. Each type showed thefollowing characteristics (HDB 1997).

– At Type “a” section in the lower reaches, cracks were generated near the top or shoulder and they reached to the bottom of the dike. Based on the results of crack detection, the method of the restoration work of the left side bank was changed to re-fill the entire cross section.

141






– At Type “b” section, no cracks were found in the middle of the crest but only found near the topof the slopes. More cracks were found on the top of the riverside slope than that of the landsideslope in both right and left side banks.

– Type “c” sections were wide at both the crest and base. Cracks were generated from the top of the riverside slope in both right and left side banks.

– Type “d” section was only observed in the right side bank. At sections classified as Failure Mode

6, cracks were generated not only on the riverside slope but also at the top of the landside slope. – At sections classified as Failure Mode 6, cracks were generated at the entire body of the dike

except at the lower part of the landside slope. The depth of the cracks reached the bottom of thedike.

Compared to the crack depths estimated from the appearance of the ground surface immediatelyafter the earthquake, the actual depths of the cracks detected by excavating pits were deeper, and most cracks reached the design elevation of main channel and/or the foundation. It should be noted that cracks are not limited to the ones that can be identified on the surface but there are some cracksthat occur only inside of the dikes.

No correlation was found between the depth of a crack and its aperture at the surface. As shown

in Figure 15, the higher the dike, the deeper the cracks were. Crack depths were at least 20% of thedike height, and in this case, most of the crack depths exceeded 50% of the dike height.

2.3 Soil properties of the dike and foundation ground at 9 km 850 section on left side bank

2.3.1 Deflection of the bottom surface of the dike and groundwater level in the dikeA cross section of the dike including the foundation at 9 km 850 (the cross section of the largestdeformation) of the left side bank is shown in Figure 16. It should be noted that the bottom surfaceof the dike subsided for about 2 m at most from the surrounding ground due to consolidationsettlement, which was induced mainly by consolidation in peat layer of the foundation ground.

Figure 15. Crack depths vs. dike height.

Figure 16. Cross section of the failed dike at 9 km 850 on left side bank.

142






Figure 18. N-value distribution in dikes.

Figure 19. Monitoring of groundwater level.

2.3.3 Records showing rises in pore water pressure in the alluvial sand layer Wells had been installed for monitoring groundwater level near Kushiro Marsh for the environment

protection purposes (see Figure 6 for their locations). Though the majority of the water levelrecorders fell down during the earthquake shaking and few data were available, water level recordersnear the damaged section of left side bank did keep a record as is shown in Figure 19. The record shows that pore water pressure rose even in the sand layer underlying the peat deposit, but the statesof the peat deposits detected from the open-cut examination and boring tests suggested that theliquefaction of the sand layer had almost no effect on the deformation of the dike.

2.3.4 Failure mechanism of dikes during the 1993 Kushiro-oki EarthquakeObserved facts described so far strongly implies the damaging process of the dike during the 1993Kushiro-oki Earthquake as illustrated in Figure 20.

Construction of a dike (about 6–7 m high using sand fill) directly resting on highly compress-

ible peat deposit resulted in consolidation settlement (2–3 m) under the weight of the dike. Thissettlement had two important effects. First it brought the lower part of the sand fill below the water table that existed at the time of the earthquake. This created the potential for liquefaction in the fill.Secondly, the large settlement caused a redistribution of stresses in the lower part of the dike. Thestretching and arching of the dike reduced the confining stresses in the saturated region of the fill.

Simplified method of evaluation of liquefaction susceptibility (Specifications for HighwayBridges) showed that FL= 0.27 for saturated dike bottom and FL = 0.72 for the alluvial sand layer under the peat deposits were gained against the maximum acceleration monitored at Hirosato(320 gal).

It is considered that the saturated bottom part of the dike was liquefied and this triggered the

dike failure.Supposed failure mechanism in Figure 20 is supported by the observed fact shown in Figure 21that the amounts of subsidence at the dike crest during the earthquake increased with the amountsof consolidation of the peat deposits.

Dynamic analysis on the dike section at the 9 km 850 site was conducted. Figure 22 shows thetime history of porewater pressure analyzed against the input motion of the Kushiro-oki Earthquakewhich was reproduced from the records obtained at Kushiro Meteorological Observatory. Resultsof this analysis revealed that the pore water pressure in the saturated sandy soil in the dike was builtup showing complete liquefaction in ten seconds from the start of the shaking during the selected

144






Figure 20. Suspected failure process.

Figure 21. Crest settlement vs. bottom bending.

145






Figure 22. Porewater Pressure in the dike at 9 km 850 section of left bank of the Kushiro Dike.

Figure 23. Restoration of right side bank.

25 seconds segments of input motion. The analysis results also showed the rise up of pore water pressure ratio in the alluvial sand layer beneath the peat layer up to about 40 %. The figure alsoshows the resulted deformation caused by the complete liquefaction in the bottom part of the dike.Details of this analysis were presented elsewhere (Finn et al. 1997).

This case history illustrates the special design requirements that are necessary when constructingdikes on highly compressible ground. Toe drains should be incorporated in the fill to allow infiltrat-ing rainwater to drain from the fill, preventing ponding. And the consequences of the consolidationsettlements of the dike on the internal stress distribution of the dike should be investigated.

2.3.5 Restoration worksBased on the investigations on causes of damages described above, it was decided to restore thedamaged dike sections by re-compaction in order to restore the flood protection performances of the dike and prevent dike damage against future earthquakes. It was decided to limit re-compactionfor the damaged area in a cross section for sections with Failure Modes 1 to 5 and to re-compactthe entire cross section for sections with Failure Mode 6. At sections of Failure Mode 6, there

was also concern that the loose, subsided sandy soil might remain within the ground and be prone to liquefaction during the next earthquakes, therefore the ground was improved by the sand compaction pile (SCP) method.

Coffering works were made at the sections to be entirely or mostly re-filled, using doublesteel sheet piles to prevent flooding during the restoration. Cross sections of replacement and re-compaction are shown in Figures 23 and 24. As shown in the figures, coffering works weremade at the inland side to preserve the natural environment of the marsh in the Retarding Basin. Aspecial species of salamander in the Kushiro Marsh living near waterfront were removed to their disturbance during the restoration works and were released to the site after the work was completed.

146






Figure 24. Restoration of left side bank.

Figure 25. Soil improvement.

Table 3. PGA at Hirosato during the 1994 Hokkaido

Toho-oki Earthquake.

Dike Ground

Direction (gal) (gal)

LG 157 303

TR 189 406

UD 43 187

A cross section of the improved ground is shown in Figure 25. A longitudinal transition areawas used to mitigate the effects of sharp changes in the ground conditions between improved and non-improved zones in order to keep smooth continuity of seismic response.

As shown in Figure 24, drains were also installed at the toe of the slope on the landward side toquickly discharge rain water after the water seeps into the dike.

2.3.6 Hokkaido-toho-oki EarthquakeSix months after the completion of the restoration works for the sections damaged by the1993 Kushiro-oki Earthquake, another damaging earthquake named the 1994 Hokkaido-toho-oki

Earthquake (M=

8.2) hit the eastern area of Hokkaido at around 22:22 on October 4, 1994.Accelerations of 314 gal in N063E direction, 392 gal in N153E direction and 189 gal verticalwere observed at the Kushiro Meteorological Station, which corresponded to about 45–60 % of thePGA at the same station during the 1993 Kushiro-oki Earthquake. At Hirosato Observatory, a PGAof 300–400 gal was recorded as shown in Table 3, and it was known from the recorded ground motion at Hirosato that the ground motion in the Kushiro Retarding Basin was about 70 % of theground motion during the 1993 Earthquake.

The earthquake caused damage to the Shibetsu and the Kushiro Rivers. Along the Kushiro River,three dike sections of 960 m length in total were damaged again and four revetment sections of

147






Table 4. PGA along the Tokachi River dike during the 2003 Tokachi-oki Earthquake.

Ground Dike

LG TR UD LG TR UD

Name of observatory (gal) (gal) (gal) (gal) (gal) (gal)

Ohtsu 439 NS 528EW 191 – – –

Tabikorai 483 349 237 496 328 204

Toitokki 580 604 280 382 586 277

Reisakubetsu 626 632 533 665 783 401

Horo-oka 593 591 217 431 250 317

Ushisyubetsu 295 349 195 321 583 255

840 m in total length were damaged. Two of the dike failures along the Kushiro river were in theRetarding Basin for a total length of 220 m, and one occurred along a branch river named the

Osobetsu River for a length of 740 m. The failure mode of both sections of dike in the KushiroRetarding Basin was longitudinal cracks at the crest. One of the damaged sections was found at asection where no damage was observed during the 1993 Kushiro-oki Earthquake, and the other waslocated at a section where partial re-compaction had been conducted after the 1993 Earthquake.

The locations of these damaged sections are shown in Figure 6 by dotted line. However the lengthof the new 1994 failures was extremely short and the severity of damage was minor compared tothat during the 1993 Earthquake.

This finding suggested that the restoration works conducted after the 1993 Kushiro-oki Earth-quake were appropriate. It should be noted that, restoration works which lowered the groundwater table in the Kushiro dike were effective in preventing significant damage from such an intenseseismic motion.

3 DAMAGE TO DIKES ALONG THE TOKACHI RIVER DURING THE 2003TOKACHI-OKI EARTHQUAKE

3.1 Damages induced to river facilities

The 2003 Tokachi-oki Earthquake (M8.0) occurred on September 26, 2003 at around 4:50 in themorning. The location of the hypocenter was at about 80 km offshore Erimo Point (144◦ 4.7E,41◦ 46.7N) at a depth of 45 km. Maximum intensity of 6 minus in the JMA scale was measured at9 municipalities in Hokkaido, and a tsunami with a maximum height of about 4 m was observed.

Two persons were declared missing, 849 were injured, 116 houses totally collapsed, and 368 houses partly collapsed (National Astronomy Observatory 2007).

The earthquake caused damages to five rivers under the jurisdiction of the HDB, namely Ishikari,Kushiro, Shibetsu, Tokachi and Abashiri Rivers including their branch streams mainly in easternHokkaido. It should be noted that sections of a dike founded on thick peat layer using so-called Pile-Net method along the Kiyomappu River failed along a length of about 290 m on the right side

bank and for 190 m along the left side bank during the earthquake. These sites were located on a branch stream of the Ishikari River, at a distance of 250 km from the epicenter. Except for this site,the remaining four rivers were located in the eastern part of Hokkaido, closer to the epicenter.

Most of the damages to river facilities were dike failures, and they were concentrated mainly

along the Tokachi River and its branches. The PGAs in the affected area were 300–600 gal as shownin Figure 26 and Table 4, and it was reported that a PGA of 118 gal was observed at the Kiyomappusite (CERI 2003).

Damaged sections of dikes along the Tokachi River are shown in Figure 26.The scale of the damage during the 2003 Tokachi-oki Earthquake was large in the lower

reaches, which were close to the epicenter, in terms of both length and degree of deformation.Table 5 shows a comparison of dike damage caused to the Tokachi River and its branches during

the 2003 Tokachi-oki andthe 1993 Kushiro-oki Earthquakes.The lengthof damaged sections duringthe 2003 Tokachi-oki Earthquake was 1.8-times of that during the 1993 Kushiro-oki Earthquake.

148




Figure 26. Damaged sections of the Tokachi River during the 2003 Tokachi-oki Earthquake.

Table 5. Comparison of failed amount between 1993 and 2003 Earthquakes.

2003 Tokachi-oki 1993 Kushiro-oki

Length Length

Name of river Number of sections (m) Number of sections (m)

Tokachi R. 8 10,511 13 7,105

Ushishubetsu R. 3 1,909 3 1,496

Reisakubetsu R. 1 700 – –

Toshibetsu R. 1 450 2 183

Urahoro-Tokachi R. 2 185 – – Urahoro R. 3 445 1 284

Shitakorobe R. 7 2,165 1 100

Total 25 16,365 2 384

The dike section in Horo-oka, which had been restored after the 1993 earthquake by partialre-compaction and had its cross section area enlarged later, had large longitudinal cracks at thecrest (Figure 27). However, the section at Tohnai, which had been improved using the SCP method

after the 1993 Earthquake, was not damaged by the 2003 Tokachi-oki Earthquake.Figures 27 and 28 show examples of the dike damage. Around the mouth of the Tokachi River, between sections 2 km 967 to 5 km 020 in Ohtsu town, three large failures took place in the landward slope of the right side dike (Figure 28). There were a few traces of sand boils on the riverside ground surface, and no abnormality was found on the river side slope including crest of the dike.

During the 1993 Kushiro-oki Earthquake, the damages to the dike at the Ohtsu section weremuch less than those at the other sections. This was attributed to the lowered groundwater elevationwithin the fill because of gabions installed at the toe of the dike and the effects of drainage ditchesdug in the surrounding inland area (Ikeda River Work Office 2003).

149





Figure 27. Damage of Horo-oka section of the Tokachi Dike.

Figure 28. Dike failures at land-side slope of the Ohtsu Section of Tokachi River Dike.

Figure 29. Cross sectional view of the failed Ohtsu dike.

There was no dike near Ohtsu City until the late 1950s. Dike construction started in 1958.A temporary low bank was completed in 1963, which was 0.7 m lower than the design high water level. The dike was raised from 1967 to 1977, and was enlarged to the designed cross section from1987 to 2001.

The dike at this section was now about 7 m high above the surrounding ground, with a crestwidth of 9 m, a bottom width of about 75 m and side slopes of 1:5.

150







Figure 30. Residual deformation of flexible joint.

As shown in Figure 28, the top of the slides started from the middle of the slope between the berm and the shoulder of the dike, and the slide ended around the toe of the dike. Several horseshoe-shaped major gaps were to be seen in the slide mass. Those major gaps accompanied by severalminor cracks. The lower part of the slid mass bulged horizontally, and no heaving was detected onsurrounding ground.

No deformation was detected on the riverside slope and the dike crest. It was apparent that thefailure of this case was limited to a part of the dike, that is, around the berm on the land ward slope,which served as an inspection path. This location of the failed part in the cross section coincided

with the location of the old dike beneath the enlarged portion of the present dike.The cause of the failure was considered to be the liquefaction occurring at the bottom part of

the enlarged portion of the dike. This type of failure has been already covered in previous sections.A special type of failure where hidden cracks were found inside the dike (Kawai et al. 2006) will be discussed below.

3.2 Dike at the section of the Ohtsu-shigai Sluice

The Ohtsu-shigai Sluice was located near the dike section mentioned above (Figure 28) with thehidden Cracks.

It was an 80.85 m-long sluiceway constructed in 1995, consisted of six segments of concreteculvert boxes that have an inner section of 1.5 m×1.5 m connected by flexible rubber joints, and they were supported by pre-stressed high-strength concrete piles having a diameter of 35 cm and a length of 33 m as shown in Figure 33.

During the urgent inspections immediately after the earthquake, partial failure on the landward berm was found on the dike surface at the section which includes the sluice. The damage at thissection was minor and the water-cutoff function of dike was not lost.

However, serious inter-displacements at joints were found by a subsequent inspection of theinside of the culvert box, shown in Figure 30. And it was found that the river side slope just abovethe sluiceway was lifted up by about 40 cm relative to the adjacent slope as shown in Figure 31.

Cracks at the toe and traces of sand boiling around the wing wall were also found.As a leveling survey results, Figure 32 was gained. A water flow test was conducted and it wasfound that the culverts retained water sealing ability although the water-cutoff ability of the cutoff walls became marginal. From a successive inspection by digging pits to observe the pile heads, itwas found that the inspected two pile heads had been badly damaged as shown in Figure 34, and from a series of integrity tests on selected 14 piles, it was found that most of piles located under the river side slope had been broken as shown in Figure 33.

It was found from the past aerial photographs that the dike section at Ohtsu had been enlarged towards riverside by reclaiming the former natural river course as its foundation. From the identified

151





Figure 31. River side slope of the dike on the Ohtsu-shigai sluice.

Figure 32. Surveyed elevation of the culvert floor.

Table 6. Distance between each segments (joint width).

Horizontal (cm) Vertical (cm)

Joint A 30= (75− 45) 0.4− 0.5

Joint B 43= (88− 45) 7.7− 8.3

Joint C 6= (51− 45) 0.4− 0.5

Joint D 8= (53− 45) 0.9− 1.0

Joint E −1= (44− 45) 1.2− 1.7

* Initial horizontal distance is around 45 cm.

location of the waterfront in the channel and the boring data, the stratification of the foundationground was compiled as illustrated in Figure 33 and Table 7 (ODCO 2007).

As shown in Figure 33 and Table 7, the foundation at the land side consists of a peat depositlayer of a thickness of about 1.5 m at surface underlain by alternative layers of silty sand and silt.To the contrary, the foundation at the riverside was covered by reclaimed soil about 4 m thick at itssurface.

It was considered that the cause of the pile damage was the lateral movement of the foundationdue to the liquefaction in sandy soil layer beneath the dike. The liquefaction potential was evaluated

using the simplified method of liquefaction evaluation and the soil profile in Figures 32 and Table 7.The evaluated results are shown in Figure 32. The sluiceway was reconstructed. The re-constructionof the sluice provided an opportunity to examine an entire cross section of the dike fill by open-cutinvestigation. Results of the investigation are summarized in Figures 35 and 36.

Several diagonal cracks running in the uppermost part of the reclaimed soil beneath the middleof the river side slope were found at the open-cut surface about 20 m downstream from the sluiceas shown in Figure 35. Although those cracks ended at the boundary between foundation and dike

bottom and fissures could not be traced in fills on their extensions, it is thought that this trace of shear deformation of the ground showed that a lateral thrust from the dike bottom acted towards

152






Figure 33. Schematic view of the damaged Ohtsu-shigai Sluice.

Table 7. Soil Properties of the foundation ground.

γ t D50 Fc Average N

Stratum (kN/m3) (mm) (%) value

New bank 19.0 0.259 19.3 13 (8–16)

Old bank

Reclaimed soil (1989) 17.0 0.985 4.8 10

Reclaimed soil (1960’s-1970’s) 17.0–19.0 0.025–0.407 9.0–76.3 7(2–13)

Alluvial sand (upper) 17.0 0.07–0.406 9.0–51.8 9(5–14)

Alluvial silt (upper) 16.0 0.004–0.008 97.0–98.4 4(2–6)

Alluvial sand (lower) 18.0 0.107–0.149 18.7–35.5 12(4–21)Alluvial silt (lower) 16.0 0.004–0.008 97.0–98.7 5(3–5)

Figure 34. Damaged pile head of the Ohtsu-shigai Sluice.

the river on top of the reclaimed soil, though its amount was not known. It was considered thatthe width of the dike bottom was elongated by this deformation of the ground. Consequently theelongation had caused a stretch type of deformation to the dike, though no visible trace was left inthe dike or on the dike surface.

The open-cut observation at the cross section of the sluice shown in Figure 36 showed clearly thetrace of a stretch type deformation of the dike through generation of vertical cracks. In this crosssection, it was found that various kinds of soil had been used as the fill materials, and vertical crackswere found in the organic soil layer about 0.5 m thick in the dike at shallow part of the riversideslope, and those cracks could be traced in the sandy soil layer beneath the organic soil layer. The

153






Figure 35. Sketch of the internal view of the dike cut surface at a location of 20 m downstream of the Sluice

(apparent trace of shearing deformation in the reclaimed soil layer was seen).

Figure 36. Sketch of the internal view of the dike at the location of the Sluice (vertical cracks were seen

within the riverside slope region).

location of these cracks was consistent with the location of the segment between sluiceway joints

A and B in Figure 33. It was found that collars at those two joints had vertical cracks inclined asshown in Figure 37.

Those observed facts implied transverse elongation of the dike bottom during the earthquake,and the length of the elongation must be around 86 cm judging from the accumulated amount of the expansions between segments at joints. Considering that cause of damage to the sluice was aliquefaction at the reclaimed soil layer in the ground on the river side, the averaged tensile strainin the riverside half of the dike near bottom was estimated to be about 2% deduced from 86 cm.

Despite of such an amount of suspected tensile strain in the dike, the reason why apparentdeformation such as cracks or slide of soil mass in the dike slope had not appeared on the dikesurface is not yet clear. Further, the reason why the apparent uplift of the slope surface on the

sluice was induced and was found couple of days later and not at the time of immediate inspectionafter the earthquake are not quantitatively clarified though it was thought to be brought by theconsolidation of the liquefied layer.

From this experience it should be noted that although no apparent deformation such as visiblecracks or slide of soil mass were found at the surface of the embankment, there may be sites wherea certain amount of deformation of fill was seismically induced. As the degree of degradation inwater-cutoff ability of a dike by such deformation is a big concern, it is considered important tostudy about the deficiency of water-cutoff ability brought by such a seismically induced looseningof dike so that an appropriate method to evaluate the deficiency is developed.

154






Figure 37. Inclination of joint collars.

4 CONCLUSIONS

Cases of seismically induced river dike failures that occurred in 1993 and 2003 were described and critically assessed in this paper.

From the cases inflicted during the 1993 Kushiro-oki Earthquake, it was revealed that there

is a type of seismic failure of dike induced by liquefaction within fill materials but not in thefoundation. This type of failure had been caused in the past before the 1993 Earthquake without

being noticed about its real cause and failure process. An example of such failure can be known inthe excellent report by Kohno & Sasaki (1969) involving a Tokachi River dike failure during the1968 Tokachi-oki Earthquake.

It was concluded from the case in 1993 that highly compressible foundation condition made the bottom part of dike saturated and caused liquefaction within fill. This conclusion means that rapid discharge of infiltrated rainwater from dikes can ensure that it remains stable.

It was found that the transverse distribution of groundwater height was not symmetric althoughthe consolidation settlement was symmetric about the dike center. This implies the need for moredetailed study of the water retention characteristics of fill material.

From the damage morphology in the Kushiro Retarding Basin dike, it was found that actualdepths of cracks within dike were sometimes deeper than that detected from dike surface, and itwas found at some sites, that cracks were not limited to the ones that could be identified on thesurface but there were some cracks that occurred only inside of the dikes.

In the case of dike at the location of Ohtsu Shigai Sluice, residual vertical cracks were detected within the dike, although no apparent traces of deformation such as cracks was evident on dikesurface. At this site, it was deduced from the pull out of sluice segments at the bottom of the dikewas stretched transversely. It teaches us that there occurs a possible deformation mode that may

be easy to miss but very serious from the viewpoint of the water-cut-off ability. Therefore it isconcluded that we should pay careful attention to the possibility of cracks developing within dike

body so as not misjudge the residual water-cut-off ability of seismically deformed dikes.River dikes have long history of continuous construction in Japan, and because they are longspread linear structures neither fill materials used nor soil conditions in their foundation ground are well documented. External forces of seismic shaking and the initial condition of fill materialsat an occasion of their failure are also hardly known exactly. Further, as damaging earthquakesdo not take place often at the same place in reasonably short periods, sufficient well documented records of dike failures during earthquake have not been accumulated.

In order to improve the design method for river dikes against seismic effects, it is considered essential to establish better indices for evaluating the seismic performance of dikes. Not only the

155





residual crest height, but the quality of the residual embankment should be properly documented.The author considers that the best progress will come through the detailed critical study of casehistories of past performance as described in this paper.

ACKNOWLEDGEMENT

The author is grateful to the Hokkaido Development Bureau for permission to use all the dateconcerning the case histories described in this paper.

REFERENCES

Advanced Construction Technology Center (ACTEC) 1997. Study report to Hokkaido Development Bureau

on the seismic stability of river dikes.

Building Research Institute (BRI), MOC 1996. Verification of recorded strong ground motion and destruc-

tiveness during Kushiro-oki, January 15th of 1993, Earthquake, Report of the Building Research Institute.

(134): 1–127.

Finn, L.W.D., SasakiY. and Wu G. 1997. Simulation of response of the Kushiro River dike to the 1993 Kushiro-

Oki and 1994 Hokkaido Toho-Oki earthquakes. Proc., 14th International Conference on Soil Mechanics

and Foundation Engineering. (1): 99–102.

Hokkaido Development Bureau (HDB) 1997. Report of research committee on damage caused by the Kushiro-

oki and the Hokkaido-nansei-oki earthquakes.

IAI Civil Engineering Research Institute (CERI), HDB 2003. Prompt report of the investigation on the damages

due to the 2003 Tokachi-oki Earthquake. Monthly Report of CERI.

Ikeda River Work Office, Obihiro Development and Construction Office (ODCO), HDB 2003. History of the

Moiwa and Ikeda River Office, History of the Tokachi River.

Kano, S., Sasaki, Y. and Hata, Y. 2007. Local failures of embankments during earthquakes. Soils and

Foundation, 47(6): 1003–1015.

Kawai M., Takebe T., Sato K., Minobe N., Kakubari S., Shiwa M., Sasaki Y. 2006. Report on the sluice damage

caused by the 2003 Tokachi-oki Earthquake. Proc. Japan-Taiwan Workshop.

Kohno, F. and Sasaki, H. 1968. Damages to River Dikes, Report of the Investigations of Damages Caused

by the “Tokachi-Oki Earthquake in 1968”. Report of the civil engineering research institute. (49): 9–24.

CERI, HDB.

Kushiro Development and Construction Office (KDCO), HDB 1994. Report on the rehabilitation works for

the damaged dikes during the Kushiro-oki earthquake.

Matsuo, O. 1996. Damage to river dikes. Special issue of Soils and Foundations.: 235–240.

National Astronomy Observatory (ed.). 2007. 2008 Chronological Scientific Tables.: Maruzen Co. Ltd.

Noto, S. (1991). Peat Engineering Handbook . : CERI, HDB. Noto private correspondence concerning a paper by Noto & Kumagaya.

Nakayama O., Sasaki Y., Sekizawa M., Hiratsuka T., and Suzuki Y. 2007. Deformation of a river dike due

to the Miyagi-ken Hokubu Earthquake. Proc. 4th International Conference on Earthquake Geotechnical

Engineering: Paper ID 1240

Oshiki H. and Sasaki Y. 2006. Damage of the Shinano River dike due to the Niigata-ken Chuetsu Earthquake.

Proc. Japan-Taiwan Workshop.

Obihiro Development and Construction Office (ODCO), HDB 1994. Report on the rehabilitation works for

the damaged Tokachi River dikes during the Kushiro-oki earthquake.

Obihiro Development and Construction Office (ODCO), HDB 2007. Interim report of the investigation results

on the damage of the Ohtsu-shigai-sluice during the 2001 Tokachi-oki Earthquake.

Public Works Research Institute 1994. Report of the disaster caused by the Kushiro-oki Earthquake of 1993. Report of PWRI .: 193.

Public Works Research Institute 1998. Strong-Motion Acceleration Records from Public Works in Japan

(No.22). Technical Note of PWRI.: 65.

Sasaki, Y. 1994. River dike failure due to the Kushiro-oki Earthquake of January15, 1993. Proc. 4th US-Japan

Workshop on Soil Liquefaction. Tsukuba, Japan.

Sasaki, Y., Oshiki, H. Nishikawa, J. 1994. Embankment failure caused by the Kushiro-oki Earthquake of

January 15, 1993. Performance of ground and soil structures during earthquakes, 13th ICSMFE. New

Delhi.: 61–68.

156




Sasaki,Y., Moriwaki, T. and Ohbayashi, J. 1997. Deformation process of an embankment resting on a liquefiable

soil layer. Deformation and progressive failure in geomechanics, Proc. IS-Nagoya’97 : 553–558.

Sasaki, Y. & Shimada, K. 1997. Yodo River dike damage by the Hyogoken-nanbu earthquake. Seismic

Behavior of Ground and Geotechnical Structures, Proc. of Discussion Special Technical Session on Earth-

quake Geotechnical Engineering during 14th International Conference on Soil Mechanics and Foundation

Engineering . Hamburg/Germany: A. A. Balkema,: 307–316.

APPENDICES

Table A-1 List of damaging earthquakes in Hokkaido

Table A-2 PGA during the 1993 Kushiro-oki Earthquake

Table A-3 List of sections damaged during the 1993 Kushiro-oki Earthquake

Table A-4 Table of sectional area changes based on survey results

Table A-5 List of sections damaged during the 2003 Tokachi-oki Earthquake

Figure A-1 Time history of the Kushiro-oki Earthquake

Figure A-2 Soil stratification in the Kushiro Marsh along the left side bank

Figure A-3 Soil stratification in the Kushiro Marsh along the right side bank

Figure A-4 Open-cut sketch at 9 km 850 of left side bank and at 11km 650 on right side bank

Figure A-5 Settlement of dike due to consolidation

Figure A-6 Detailed borehole logs at the cross section of 9km 850 of left side bank

Figure A-7 Reproduced waveform of seismic motion at Hirosato during the Kushiro-oki and the

Hokkaido-toho-oki Earthquakes (Digital Data in Excel)

157





Ground failures and their effects on structures in Midorigaoka

district, Japan during recent successive earthquakes

K. Wakamatsu Department of Civil & Environmental Engineering, Kanto-Gakuin University, Yokohama, Japan

N. YoshidaTohoku Gakuin University, Tagajo, Miyagi, Japan

ABSTRACT: Midorigaoka, Kushiro City, northeast Japan, suffered liquefaction-induced ground failures during four successive earthquakes in the past thirty years. This paper presents the ground failures and their effects to structures observed in Midorigaoka during the earthquakes, and exam-ines the relationships between recurrent liquefaction-induced damage and subsurface conditions.As a result, thick liquefiable fill, slope of the ground surface, and subsurface water conditions,which resulted primarily from filling a marshy valley, are found to be responsible on the damage.

1 INTRODUCTION

Midorigaoka is a residential area situated in Kushiro City, Hokkaido, Japan, approximately 900 kmnortheast of Tokyo. It was developed for housing in the 1960s by cutting and filling a Pleistoceneterrace with a maximum elevation of approximately 30 meters above sea level. The district has

been frequently affected by large earthquakes and suffered various kinds of ground deformationand damage to buildings and underground pipelines. The earthquakes include the 1973 Nemuro-hanto-oki M7.4 quake, the 1993 Kushiro-oki M7.5 quake, the 1994 Hokkaido-toho-oki M8.2 quake,and the 2003 Tokachi-oki M8.0 quake. We conducted field survey in these earthquakes and found that the damage was repeatedly concentrated in particular areas in Midorigaoka.

In order to investigate the causes of concentration of the damage on such local areas, the historyof land reclamation and construction projects in Midorigaoka was investigated and geotechnicaldata were collected to determine the subsurface conditions in the area. Old topographic maps were

collected and photogrammetric analyses were conducted to establish the profiles of the land forms before and after the residential development. Furthermore, the effects of changes in land forms,variation in fill thickness and density, subsurface water conditions and intensity of ground motionin concentrating damage in the area are evaluated.

2 GEOLOGY, HISTORIC DEVELOPMENT OF MIDORIGAOKA

Midorigaoka is located on a Pleistocene terrace in the eastern part of Kushiro City that faces thePacific Ocean on the southern end, Hokkaido, Japan. The terrace is composed of, from upper tolower, volcanic ash known as the Kussharo Pumice Flow Deposit, volcanic sandy soil known as

the Otanoshike Formation, and sand, gravel, silt or clay known as the Kushiro Group on a base of Paleogene rock. Figure 1 is an old topographic map surveyed in 1958. It should be noted that theterrace is dissected by a deep valley in the eastern part of Midorigaoka. The bottom of this valleyis marsh and extends towards Kushiro Wetland.

The development of the terrace started from the north-western part of the area around the estuaryof Old Kushiro River in the early 1900s and gradually extended toward the north-east. Midorigaoka

began to be developed in the 1960s and completed in 1972–73. The area was developed by cuttingand leveling the terrace and then filling the valleys with soils from the terrace. Therefore thereare two general types of subsurface conditions in Midorigaoka: 1) natural soils associated with

159




Figure 1. Topographic map surveyed in 1953 before development.

the volcanic pumice flow deposit overlying the original or excavated surface of the terrace and 2) fills consisting mainly with the volcanic pumice flow deposit placed on the valleys and hollow

parts in the terrace. Typical soil profiles and elastic wave velocities in the terrace and in the valleyare shown in Figure 2. Areas underlain by these different profiles behaved differently during the

earthquakes, as discussed later.

3 GROUND FAILURES AND THEIR EFFECTS ON STRUCTURES DURING THE 1993KUSHIRO-OKI AND THE 1973 NEMURO-HANTO-OKI EARTHQUAKES

3.1 Outline of the 1993 Kushiro-oki earthquake

The Kushiro-oki earthquake occurred at 8:06 PM local time, on January 15, 1993, and registered 7.5on the Japan Meteorological Agency (JMA) magnitude scale. The focus was located at, longitude42◦55.2 N and latitude 144◦21.2 E, with depth of 101 km (JMA, 2007). Kushiro city is located

only 10 km from the epicenter. The event is considered to be an intra-plate earthquake within thePacific Plate. Two persons were killed, 69 persons were seriously injured and 778 persons werereported to have minor injury. Sixty-one residential houses were completely collapsed, 348 houseswere half collapsed and 7,095 houses were partially collapsed. The total economic loss due to theearthquake was estimated 55 billion yen (approximately $550 million) within Hokkaido (HokkaidoGovernment, 1993).

Strong motions were recorded at the Kushiro Meteorological Observatory which is located about2.5 km southwest of Midorigaoka (see Figure 1 for location). The peak acceleration and velocity are711 cm/s2 and 33.5 cm/s, respectively. A record with peak acceleration 919 cm/s2 is also obtained at the ground floor of a building in the same site at a distance of about 20 m from the previousstation. This site is located on the Pleistocene terrace, which is considered to amplify the earthquake

shaking in short period range resulting high peak value. Earthquake records at other stations inKushiro City did not show such large peak values. The peak acceleration at Kushiro Port in areclaimed land was 469 cm/s2, at the Kushirof River Dike, 320 cm/s2 and at Otanashike Bridge,456 cm/s2 (NIEDSTA, 1993).

3.2 Ground failures caused by the 1993 Kushiro-oki earthquake

Figure 3 shows locations of major ground failures and damage to structures in the eastern partof Kushiro City caused by the 1993 Kushiro-oki earthquake. The ground failures include slope

160





Figure 2. Typical soil profiles and elastic wave velocities in terrace and valley (Wakamatsu and Yoshida,

1995).

failures, collapses of retaining walls, ground cracks, ground settlements and sand boiling. Thelocations of damaged structures coincided with those of ground failures, which imply that most of the damage was caused by the ground failure. The damage was heavily concentrated in the zoneinvestigated in this paper.

Figure 4 shows locations of ground failures and uplifted manholes in Midorigaoka. A slope witha height of 11 m collapsed at site A in the figure. A wooden house slipped down the slope and twohouses were crushed by the collapsed ground as shown in Figure 6. Liquefaction effects such assand boils and uplift of manholes were observed accompanied by ground cracks. In many sites,manholes uplifted by 5 to 20 cm. Sand boils were observed at very few locations compared withthose during the 1994 Hokkaido-toho-oki earthquake, which is discussed at a later section, even

though the ground shaking is stronger in this earthquake. At the time of the earthquake, the ground surface was frozen with 50–100 cm of thickness, which may have prevented the ejection of sand boils to the ground surface.

Underground pipelines and foundations of wooden houses settled differentially and were displaced laterally where the ground cracks were abundant. This implies that a large amount of permanent ground deformation occurred in these areas. The magnitudes of the vertical and hor-izontal ground displacements were estimated to be in the range of approximately 30 to 50 cmon the basis of widths of ground cracks and displacements of pipelines and foundations of thehouses.

161





Figure 4. Map of Midorigaoka showing locations of ground failures and uplifted manholes during the 1993

Kushiro-oki earthquake and area of liquefaction during the 1973 Nemuro-hanto-oki earthquake (modified from Wakamatsu and Yoshida, 1995).

Figure 8 shows the locations, types, diameters and lengths of damaged pipes. At the damagelocations, the average bury depth of the pipes was 3.4 m and they were supported by timber pilesor macadam. Damage was primarily in the form of upliftment and settlement of the pipelines dueto the displacement of the joints, round cracks of pipes and manholes, pull-out or break of joints,and uplift of manholes as shown in Figure 9 and 10. The amount of both uplift and settlement of the pipelines were ranged from several cm to several ten cm.

3.5 Damage to gas distribution line

The gas distribution and local service pipelines in Midorigaoka are composed mostly of steel pipesand some polyethylene pipes, which are buried at cover depths of about 1.0–1.2 m. Small diameter

pipes of 25, 30 and 50 mm were built with threaded joints, whereas large diameter pipes of 100,150 and 200 mm were built with welded and A-II type mechanical joints. Figure 11 shows thelocations of damage to mains and branches of high and low pressure gas distribution lines. All of the damage occurred at threaded and mechanical joints. No breaks or other damage to pipes withwelded joints were reported.

163





Figure 5. Cracks of retaining wall

caused by the 1993 Kushiro-oki

earthquake.

Figure 6. Sliping down of a house due to slope failure and crashed

houses by collapsed soil caused by the 1993 Kushiro-oki earthquake

(Courtesy of Asahi Shimbun Co.).

Figure 7. Map of Midorigaoka showing locations of damage to water distribution and service pipelines during

the 1993 Kushiro-oki earthquake (Wakamatsu and Yoshida, 1995).

164







Figure 9. Deformation of a sewage

pipeline during the 1993 Kushiro-oki

earthquake (Courtesy of Kushiro City

Government).

Figure 10. Rupture and separation of connection between

manhole and a sewage pipe during the 1993 Kushiro-oki

earthquake (Courtesy of Kushiro City Government).

Figure 11. Map of Midorigaoka showing locations of damage to high and low pressure gas distribution

systems during the 1993 Kushiro-oki earthquake (Wakamatsu and Yoshida, 1995).

166







Figure 12. Map of Midorigaoka showing locations of collapsed houses during the 1993 Kushiro-oki

earthquake (Wakamatsu and Yoshida, 1995).

Thirty-two persons were seriously injured and 404 persons were minor injured. Building damagewas observed over a large area, but it was localized and relatively limited: sixty-one residen-tial houses were completely collapsed, 348 houses were half collapsed and 7,095 houses were

partially collapsed. The total economic loss due to the earthquake estimated was 57 billion yen(approximately $570 million) within the Hokkaido region (Hokkaido Government, 1995).

The quake also caused ground failures such as land slides, slope failures, and soil liquefactionin a widespread area in Hokkaido area. Highway, railways, buildings, houses, ports and fisheryfacilities, and underground utilities were affected by these ground failures.

The epicenter of the earthquake was located approximately 270 km east-northeast from Kushiro

city. A peak horizontal acceleration of 473 cm/s2 was recorded at Kushiro Meteorological Observa-tory and 269 cm/s2 was recorded at Kushiro Port 5 km west-northwest of Midorigaoka (NIEDSTA,1994).

4.2 Ground failures and damage to buildings caused by the 1994 Hokkaido-toho-oki earthquake

Figure 13 shows locations of ground failures, collapsed houses, and damaged pipelines in Midori-gaoka. Although the epicenter was much more distant (or was quite far away) and the peak

167





Figure 13. Map of Midorigaoka showing locations of ground failures, damaged houses and lifelines during

the 1994 Hokkaido-toho-oki earthquake.

acceleration was smaller by half than that in the 1993 Kushiro-oki earthquake, the similar ground failures were also observed; sand boiling occurred at the sites where liquefaction-induced damagewas wide-spread in the 1993 earthquake. Despite less damage to structures than that due to the

previous 1993 quake, sand boiling was observed at more locations. This implies that the frozenlayer of ground at the surface prevented the ejection of sand boils at the time of the 1993 Kushirookiearthquake that occurred in midwinter.

Figures 14 to 19 show typical damage in the area. A major prevention work for slope failure was performed at Site A in Figure 13 where the slope failure occurred at the time of the 1993 Kushiro-oki earthquake (Figure 6). Although no slope failure occurred during the 1994 Hokkaido-toho-okiearthquake, an exterior wall of a house (Address: 6-12 Midorigaoka), next to the house slipped down due to the slope failure in 1993, broke away as shown in Figure 15 and the ground fissureobserved at the boundary between fill and natural ground.

At Site B in Figure 13, walls tilted significantly, sand boiling and ground cracks occurred at thesame site as at the time of the 1993 earthquake. In addition uplift of manholes were observed insouth of the area. At Site C in Figure 13, several sand boils were observed as shown in Figure 19.

168





Figure 14. Damaged house at 6-12 Midorigaoka

during the 1993 Kushiro-oki earthquake.

Figure 15. Damaged house at 6-12 Midorigaoka

during the 1994 Hokkaido-toho-oki earthquake.

Figure 16. Tilting of walls due to liquefactionduring the 1994 Hokkaido-toho-oki earthquake.

Figure 17. Ground cracks and tilting of wall duringthe 1994 Hokkaido-toho-oki earthquake.

Figure 18. Uplift of manholes during the 1994

Hokkaido-toho-oki earthquake.

Figure 19. Sand boils in a house yard during the

1994 Hokkaido-toho-oki earthquake.

At Site D in Figure 13, the most serious liquefaction induced effect such as sand boiling, ground cracking and damage to houses occurred. According to the resident of that area, a small stream

existed in the area prior to its development and thus, the damaged zone in this area is thought to bea filled area.

4.3 Damage to lifeline facilities

The main water distribution system was damaged at one location near Site B in Figure 13. The pipe was poorly earthquake-resistant ductile iron pipe with a diameter of 150 mm and broken by pull-out of the joint.

169










Figure 20. Map of Midorigaoka showing locations of ground failures, damaged houses and water distribution

pipes during the 2003 Tokachi-oki earthquake.

Figure 21. Wall settled differentially due to liq-

uefaction during the 1993 Kushiro-oki earthquake

and tilted and settled much larger due to the 2003

Tokachi-oki earthquake (looking side).

Figure 22. Wall settled differentially due to liq-

uefaction during the 1993 Kushiro-oki earthquake

and tilted and settled much larger due to the 2003

Tokachi-oki earthquake (looking front).

171







Figure 23. House settled and pulled apart due to

liquefaction-induced ground displacement during the

2003 Tokachi-oki earthquake.

Figure 24. Superstructure of a house was sepa-

rated from foundation during the 2003 Tokachi-oki

earthquake.

Figure 25. Map of Midorigaoka showing locations of cross-sections and contour lines in 1963 before

development.

6 RELATION BETWEEN SUBSURFACE SOIL CONDITIONS AND DAMAGE

6.1 Subsurface soil conditions

Figure 25 shows contour lines in 1963 before development in the Midorigaoka. It can be seen fromthe figure that the areas of the heaviest concentration of the damage during the four earthquakes

172







Figure 26. Soil profile along cross-section A-A in Midorigaoka (see Figure 25 for location) (Modified from

Wakamatsu and Yoshida, 1995).

coincide with the valley bottom at an altitude about 5 meters and valley wall at altitudes between5 and 25 meters above sea level.

Several bore hole investigations and standard penetration tests had been conducted to clarifyfoundation conditions of sewage system in Midorigaoka. On the basis of the site investigations, across-section in the north-south direction was developed as shown in Figure 26. Loose fill with SPT

N-values less than 5 extends along the section with a thickness ranging from 1 to 9 m. Underlyingthe fills is very soft peat, which was originally at the bottom of the marshy valley. The depth of the water table is several meters deep in the terrace from the ground surface, whereas it is 0.3 to2 m deep in the valley, which indicates that the water table has risen up into the f ill. The water hascome from neighboring area because the valley forms water catchment.

The slope failure at the time of the 1993 Kushiro-oki earthquake described before (A in Figure 4)occurred where the fill is thick and the slope is steep, as shown at the left end of the Figure 26.

Sand boils were observed in the gentle valley slope where 1 to 2 m thick fill overlies the water table.Major cracks and settlements occurred in the upper part of the slope with sand boils. This patternof ground deformation implies that lateral movement toward the center of the valley occurred dueto the liquefaction of loose fill.

Figure 27 is a comparison of the grain size distribution curves of sand boils ejected during the1993 Kushiro-oki earthquake at different locations in Midorigaoka and natural volcanic ash whichis generally used as fill material in this area. Both soils have similar grain size characteristics,which supports that the fill liquefied during the earthquakes.

6.2 Landform changes due to development

To examine further, the relationships between ground deformation and subsurface conditions, land form changes are investigated by analyzing a pair of aerial photographs taken before and after thedevelopment. A total of 13 sections were selected for analysis; 12 sections through Midorigaokaand 1 outside Midorigaoka where a few effects of the earthquakes were observed during the 1993Kushiro-oki earthquake. The accuracy of the measurements is about ±1 m in the vertical direction.

Typical examples of the photogrammetric analysis were shown in Figure 28, in which the loca-tions of streams (terrace runoff) before the development and the ground failures and the damage tostructures during the 1993 Kushiro-oki earthquake are plotted. In the figure, the damaged houses

173





Figure 27. Grain size distribution curves for volcanic ash used as fill material and sands collected from sand

boils due to the 1993 Kushiro-oki earthquake in Midorigaoka (Wakamatsu and Yoshida, 1995).

Figure 28. Cross-sections showing land form changes due to development and damage caused by the 1993

Kushiro-oki earthquake (see Figure 25 for locations) (Wakamatsu and Yoshida, 1995).

174






previously shown in Figure 12 are re-classified according to damage mode based on the record bythe city government into three types: 1) damage in foundations; 2) damage other than foundationsuch as collapsed chimney and cracked exterior wall; and 3) portion of damage not reported. Themajority of the damage in Midorigaoka was categorized as the type-1).

It can be seen from the figure that the heaviest concentration of the damage was observed in thefilled area where slope failures, ground cracks and sand boils were concentrated. The slope failure

occurred at the boundary between the terrace and the valley where fill is thick and the slope of ground surface is steep as shown in section D-D , whereas liquefaction effects such as sand boilsand uplift of manholes occurred in the valley where fill is thick, slope is gentle and streams flowed

before the development. Ground cracks also occurred on both steep and gentle slopes in the filled area. Damage to pipelines and foundations of houses in general, was observed in the locationswhere ground failures occurred, whereas structural damage was absent in the valley bottom wherethe fill is flat and not so thick, as shown in section D-D. These data imply that the permanent ground deformation associated with ground failures was the principal cause of the damage to structures.In contrast, damage such as collapse of chimney and cracks of exterior walls not associated withfoundations did occur in both cut and fill areas, which implies that primary cause of the damage

was the strong ground shaking.


There was substantial damage to underground pipelines and buildings in Midorigaoka, KushiroCity during the successive earthquakes of the 1973 Nemuro-hanto-oki, 1993 Kushiro-oki, 1994Hokkaido-toho-oki and 2003 Tokachi-oki. The damage was caused primarily by ground defor-mations associated with soil liquefaction and liquefaction-induced flow, and partially by ground shaking and slope failure.

In summary, Midorigaoka had several geotechnical conditions such as thickness and density

of fills, slope of the ground surface, and subsurface water conditions, which led to various typesof ground failures. The conditions resulted primarily from filling marshy valleys. They stronglyaffected not only the locations and pattern of the ground failures but also the resulting damagemodes of structures. Although the level of damage was different, it is notable that locations and

patterns of ground failures were very similar in the all earthquakes.

ACKNOWLEDGEMENTS

The authors wish to thank several individuals who provided us useful materials and informationconcerning the earthquake damage in Midorigaoka. Special thanks are extended to Mr. Iwamoto of Kubota Corporation, Dr. Shimizu and Mr. Nakayama of Tokyo Gas Co., Mr. Honma and Mr. Taisyuof the Kushiro Water Department, Mr. Okabe and Kainuma of the Kushiro Sewerage Division,Mr. Masuoka of the Kushiro Urban Development Department, Mr. Sugimura of the Kushiro GeneralAffairs Department, and Mr. Kobayashi of Kushiro Gas Co.

REFERENCES

Fire and Disaster Management Agency (FDMA). 2004. Report on the 2003 Tokachi-oki Earthquake (Final).

Hokkaido Government. 1993. Overall Damage in Hokkaido caused by the 1993 Kushiro-oki Earthquake

(Final).Hokkaido Government. 1995. Overall Damage in Hokkaido caused by the 1994 Hokkaido-toho-oki Earthquake

(Final).

Japan Meteorological Agency (JMA). 2007. The Annual Seismological Bulletin of Japan for 2006 : Japan

Meteorological Business Support Center.

Japan Gas Association. 1994. The 1994 Kushiro-oki and Hokkaido Nansei-oki Earthquakes and Urban Gas

Facilities (in Japanese).

Kushiro City Government. 1993. Document of Earthquake Hazard of the 1993 Kushiro-oki Earthquake (in

Japanese).

175





Behaviour of SCP-improved levee during 2003 Miyagiken-Hokubu

Earthquake

A. TakahashiTokyo Institute of Technology, Meguro, Japan (Formerly Public Works Research Institute, Japan)

H. Sugita Public Works Research Institute, Tsukuba, Japan

ABSTRACT: Records of excess pore water pressure changes as well as ground accelerationswere obtained at Nakashimo, located near the mouth of the Naruse River, Miyagi Prefecture,during the earthquakes on 26 July 2003. Comparisons of records of two arrays at the site provide aunique opportunity to study the effectiveness of densificationof sandy soil by theSand CompactionPile (SCP)methodas liquefactionremediation. In themain shock, thepeak ground accelerationwasabout 400 gal and the excess pore water pressure increased both in the improved and unimproved ground. The maximum excess pore water pressure ratio in the former was less than 0.4 while itwas around 0.8 in the latter. These records suggest that marked increase of liquefaction resistanceof soil can be expected for the densified ground with the low replacement ratio SCPs, even if theincrease of SPT N -value by piling is insignificant.

1 INTRODUCTION

On 26 July2003, a series of major inland earthquakes occurred northof the Miyagi Prefecture, withthe main shock having a JMA magnitude (M j) of 6.2 at 7:13 a.m. The main shock was preceded by foreshocks, the largest one with M j= 5.5 occurred about seven hours before the main shock.The main shock was also followed by several aftershocks, with the largest having a magnitudeof 5.3 about 10 hours after the main shock. All of these three events had shallow hypocen-tres, about 12 km from the surface (JMA). Epicentre locations of these earthquakes are plotted in Fig. 1.

The main shock of the earthquake caused damage to levees of the Naruse River. Locations of serious damage are indicated by the ‘x’ marks in Fig. 1 (JSCE & JGS, 2004; NILIM & PWRI,2003; NILIM, PWRI & BRI, 2003). Most of these locations were distant from the river mouth(about 8 to 17 km from the mouth,) even though the epicentre of the main shock was close tothe mouth. They were on old river channels (pre-improved channel) or marshes, i.e., the leveeswere constructed on loose deposits, resulting in large deformation of the levees. The most seriousdamage was found at Kimazuka (right-bank, 13 km away from the mouth,) whose photos are shownin Fig. 2. At this site, large deformation of the levee was observed on the landward side, whilethe slope facing the river channel (left-hand side in the pictures) remained almost intact since theface was armed by revetment and sheet piles were placed at the toe of the levee (KKRO, 2005;

Sekizawa et al ., 2005).After the 1995 Kobe Earthquake, Ministry of Construction (now, Ministry of Land, Infrastruc-ture and Transport,) Japan, installed liquefaction arrays as well as seismometer arrays in variouslocations. At Nakashimo, located near the mouth of the Naruse River, the liquefaction arrays were placed in the levee with and without liquefaction remediation in 1997 and 1998. The levee equipped with the liquefaction arrays was almost undamaged during the earthquake. However the liquefac-tion arrays functioned well and provide a unique opportunity to study effectiveness of densificationof sandy soil as liquefactionremediation bycomparingtheamountof theexcess pore water pressurerise in the two arrays (Matsuo et al ., 2004).

177




Kimazuka (13kp)

Naruse River

Nakashimo (0.8kp)Main shock7:1326/07/03

Largest

foreshock

0:1326/07/03

Largest

aftershock

16:5626/07/03

10 km

Ishinomaki Bay

Figure 1. Locations of epicentre and seriously damaged levees.

Figure 2. Levee damage at Kimazuka (right-bank, 13km away from river mouth).

2 DESCRIPTION OF THE SITE

Location of the liquefaction arrays is shown in Fig. 1. Distance from the mouth of the Naruse River to the site is 0.8 km and the arrays were installed in the right-bank. The levee faces to the river atan angle of 70 degrees to the north.

Cross section of the levee at Nakashimo is shown in Fig. 3 together with soil profiles withSPT- N value distribution and locations of accelerometer and pore water pressure cell. The levee isunderlain by silty and sandy soil layers. Soil properties of these layers are summarised in Table 1.Among them, the sandy soil layers (Acs & As) sandwiched by the silty soil layers (Ac1 & Ac2) arevulnerable to liquefaction. Although large difference can be seen in the SPT- N values at Acs and As, no marked difference could be seen in the particle size distributions, i.e., the f ine contents. For the sensors, the maximum measurable acceleration is 2 G and the maximum measurable pore water pressure is 100 kPa. At the berm, all the sensors were installed in between the Sand CompactionPiles (SCPs.)

178





Figure 3. Cross section of levee at Nakashimo (right-bank of Naruse River, 0.8 kp.)

Table 1. Soil properties.

Grading (%)∗1

Specific Natural water Plasticity qu

Soil type gravity∗1 content

∗1∗2 Gravel Sand Fines index∗4 (kPa)

∗5 R∗6

L20

Bk Fill 2.69 20%∗3 0–2 60–82 16–40

Ac1 Sandy silt 2.56 51% 0–1 12–38 62–87 29–48 60

46%

Acs Sand with fines 2.65 36% 8–10 70–88 4–22 0.19

As 27% 0.23

Ac2 Sandy silt 2.63 72% 0–1 20–31 59–89 48–67 12057%

To Soft rock – – – – – – – –

(siltstone or arenite)

∗1 All data were obtained from boreholes for instrumentation.∗2 Except Bk, the upper value is for the boreholes near the crest, while the lower is for those at the berm.∗3

γ t = 16.6 kN/m3

∗4 At H7-2-1, H7-2-1-A and H7-2-1-C∗5 At H7-2-1∗6 At H7-2-1-A (0.19) and H7-2-1-C (0.23) using tube samples

P- and S-wave velocity profiles measured by down-hole tests at the berm using the boreholefor the deepest accelerometer installation are shown in Fig. 4. Clear change in the P-wave velocitycan be seen around the water table (+0.9 m in elevation); while the marked change in the S-wavevelocity cannot be observed down to the bottom of the soft soil layers.

To mitigate the levee damage induced by liquefaction, a remedial measure for liquefaction wasmade by densifying sandy soils with the SCP method in 1996 at the berm of the levee as shown

179





Figure 4. P- and S-wave velocity profiles measured at berm.

Figure 5. Excess pore water pressure changes for main shock.

in Fig. 3. Location of the densified zone is enclosed by the chain line in the figure. Length of the improved zone is 350 m (from 0.7 to 1.05 kp) and the liquefaction array is situated around themiddle of it (0.8 kp). Diameter of the casing for SCP was 700 mm and four rows (in the directionnormal to the levee axis) of piles were installed. The spacing between the piles in the levee axisdirection is 2.2 m, while that in the direction normal to the levee axis is 1.7 m.Thus, the replacementratio is 0.10. Unfortunately, there is no SPT- N value profile before installation of SCPs at the berm.However, the following facts may support that the sandy layers were improved in some measure;

180






Figure 6. Acceleration time histories for main shock around and beneath shoulder.

(1) the natural water content, i.e., void ratio, of the sandy soil layers (Acs & As) at the berm (withSCPs) is smaller, and (2) SPT- N values in the sandy soil layers at the berm (Borehole B-5) areslightly larger than those at nearby Boreholes H7-2-1 and H7-2-1-A.

181





Figure 7. Acceleration time histories for main shock around and beneath berm (array within densified zone

with SCPs).

3 THE RECORDS OBTAINED

At the site, records for a series of major inland earthquakes on 26 July 2003 were obtained. Asmentioned in the above section, the levees distant from the river mouth (about 8 to 17 km from the

182





mouth) were severely damaged during the main shock, while noticeable damage was not observed at the site even after the several large aftershocks. In many cases, an even settlement of dozens of centimetres is not noticeable after earthquakes by visual inspection.Thus, the settlement at the sitewas probably at that level or smaller.

Excess pore water pressure changes during the main shock are plotted in Fig. 5. Unfortunately,as the maximum measurable pore water pressure was set at 100 kPa including hydrostatic pressure,

the maximum pore water pressure in the ground without SCPs could not be captured. However, itcan be said that the maximum excess pore water pressure ratio in the ground without SCPs (beneaththe crest) was around 0.8, while it was less than 0.4 in the ground with SCP-improvement (beneaththe berm). These records suggest that marked increase of liquefaction resistance of soil can beexpected for the densified ground with the low replacement ratio SCPs, even if the increase of SPT

N -value by piling is insignificant.Figures 6 and 7 plot all the acceleration records obtained at the site, but the records at GL-5 m

for the array within the densified zone using SCP method, during the main shock. The recordsfor the array in the ground without SCPs were obtained around and beneath the shoulder, whilethose in the ground with SCPs, i.e., the densified ground, were measured around and beneath the

berm as shown in Fig. 3. In both arrays, the attenuation of wave amplitude due to liquefaction in a broad sense, i.e., softening of the ground due to excess pore water pressure increase, was observed and dominant period of the waves seem to have got longer at the ground surface. However, asthe acceleration responses at the surface may have been affected by the geometry of the levee,it is very difficult to say something on effectiveness of densification as liquefaction remediationonly by comparing obtained acceleration records without help of numerical analysis, such asmulti-dimensional dynamic f inite element analysis.

4 SUMMARY

Records of excess pore water pressure changes as well as ground accelerations were obtained at Nakashimo, located near the mouth of the Naruse River, Miyagi Prefecture, during the earthquakeson 26 July 2003. Comparisons of records of two arrays at the site provide a unique opportunityto study effectiveness of densification of sandy soil by the Sand Compaction Pile (SCP) method as liquefaction remediation. Only from the acceleration records, marked effects of densificationon the dynamic ground responses cannot be seen. However, the pore water pressure records showlarge difference and suggest that marked increase of liquefaction resistance of soil can be expected for the densified ground with the low replacement ratio SCPs, even if the increase of SPT N -value by piling is insignificant.

Enquiry regarding the records presented in this paper should go to Earthquake Disaster Pre-vention Division, Research Centre for Disaster Risk Management, the National Institute for

Land and Infrastructure Management, the Ministry of Land, Infrastructure and Transport, Japan( [email protected]).

ACKNOWLEDGEMENT

All the records presented in this paper are provided by the National Institute for Land and Infras-tructure Management, the Ministry of Land, Infrastructure and Transport, Japan. Photos shown inFigure 2 are provided by the Tohoku Regional Bureau, the Ministry of Land, Infrastructure and Transport, Japan.

REFERENCES

Japan Metrological Agency, http://www.seisvol.kishou.go.jp/eq/2003_07_26_miyagi/ (in Japanese).

Japan Society of Civil Engineers and Japanese Geotechnical Society. 2004. Reconnaissance Report on the

July 26, 2003 Northern Miyagi Earthquake, 33pp (in Japanese).

Kitakamigawa-Karyu (Downstream region of Kitakami River) River Office for Downstream of the Kitakami

River, Ministry of Land, Infrastructure and Transport. 2005. Technical Report on failure mechanism of

levees of Naruse River in the 2003 Miyagiken-Hokubu Earthquake, 130pp (in Japanese).

183


mailto:[email protected]








Matsuo, O, Kusakabe, T., Uehara, H., Sekizawa, S. & Sato, S. 2004. Acceleration and pore water pressure

responses of SCP-improved levee during the 2003 Miyagiken-Hokubu Earthquake, Proc. 59th Annual

Conference of JSCE , I-775, 1547–1548 (in Japanese).

National Institute for Land and Infrastructure Management and Public Works Research Institute. 2003. Pre-

liminary Report of the 2003 Miyagiken-Hokubu Earthquake, Civil Engineering J., Vol.45, No.9, 10–15 (in

Japanese).

National Institute for Land and Infrastructure Management, Public Works Research Institute and BuildingResearch Institute. 2005. Report of the 2003 Miyagiken-Hokubu Earthquake, 77pp (in Japanese).

Sekizawa, S., Sato, S., Okada, S. & Hiratsuka,T. 2005. Damege simulationof Narusegawa river dike during the

2003 Miyagiken-Hokubu Earthquake, Proc. 40th Japan National Conference on Goetechnical Engineering ,

2007–2008 (in Japanese).

184





Tsukidate failure compared with the other flow-type failure during

2003 earthquakes in Northern Japan

M. Kazama & R. UzuokaTohoku University, Sendai, Japan

N. Sento & T. Unno Nihon University, Kouriyama, Japan

ABSTRACT: Flow-type failures occurred at two sites during 2003 earthquakes in northern Japan.Two landslides are of similar magnitude and configurations, but of different material and failure process. One slope was a fill with pyroclastic sediments.The failure occurred during or immediatelyafter the principal motion of the earthquake. It is considered that the unsaturated fill lost the initialshear strength and fluidized by earthquake induced cyclic shearing. The other failure occurred a few minutes after the principal motion of the main shock. Rainfall and multiple large shocks wereimportant features that exacerbated the landslide. The upper portion of the fill lost shear strength,descended along the slope and spread on the rice field. The authors conducted site investigationsand examined physical and mechanical soil properties of the failures, and discussed the causes of failures.

1 INTRODUCTION

1.1 Earthquake events

Two disastrous earthquakes hit the northeastern area of Honshu Island of Japan – the Tohoku area.The epicenters of the two earthquakes are shown in Figure 1. The first earthquake occurred in thePacific Plate off the coast of Miyagi Prefecture at 18:24, local time on 26 May, 2003. We will callthe earthquake “526 Eq.” hereafter. The National Research Institute for Earth Science and Disaster Prevention (NIED) assigned a moment magnitude (Mw) of 7.0 to this earthquake, with focal depth

of about 70 km. The second earthquake occurred at an inland north area of Miyagi Prefecture at7:13, local time, on 26 July, 2003, just two months after the 526 Eq. We will call the earthquake“726 Eq.” hereafter. The NIED assigned Mw of 6.1 to this earthquake, with about 12.9-km focaldepth. In addition, a large foreshock and aftershock with slightly smaller magnitudes than the mainshock occurred at 0:13 and 16:56, respectively, on the same day. The respective epicenters of thesethree shocks are located around the Asahiyama Hill. The NIED assigned Mw of 5.5 and focaldepth of 12.8 km to the foreshock, and Mw of 5.3 and focal depth of 13.3 km to the aftershock. Itwas remarkable that three large earthquakes hit this limited area within a day. Many strong motionrecords for these events were observed by NIED, JMA (Japan Meteorological Agency), MLIT(Ministry of Land, Infrastructure and Transport), and other organizations. Peak acceleration of over 1 G was recorded at some observation sites. Table 1 shows the summary of the earthquake

events.

1.2 Outline of the damages

These earthquakes caused severe damage to infrastructure and buildings. They injured severalhundred people. The 526 Eq. caused extensive damage over a wide area of Miyagi and IwatePrefectures. More than 2,000 housesin theprefectures werepartiallycollapsed. Reinforced concrete

piers of the Tohoku express railway over 100 km apart from the epicenter were partially damaged.

185




Figure 1. Locations of major landslides and epicenters of the 526 Eq. and 726 Eq. (epicenters by Japan

Meteorological Agency).

Table 1. Summary of the earthquake events.

Earthquake events Date Time Mw Depth (km) Seismic intensity Seismic records

5 26 Eq. 2003/5/26 18:26 7.0 70.0 5+ (Tsukidate) Fukumoto et al. (2007)

726 Eq. Foreshock 2003/7/26 0:13 5.5 12.8 5− (Kanan) National Inst. (2003)

mainshock 7:13 6.1 12.9 6− (Kanan)

aftershock 16:56 5.3 13.3 6− (Kanan)

The 726 Eq., on the other hand, caused intensive damage in a relatively small area of MiyagiPrefecture. More than 1,000 houses had totally collapsed; more than 10,000 houses had partiallycollapsed. Some reinforced buildings, hospitals and schools, were severely damaged. Many road embankments and river dikes deformed and settled by over 1 m at some locations.

Geotechnical events such as liquefaction, rock falls and landslides including small slope failureswere observed during the earthquakes. Sites at which we found traces of sand boil after the 526 Eq.spread along the coast. Traces of sand boil at some sites after the 726 Eq. were observed at thesame places that were liquefied during the 526 Eq. Rock falls during the 526 Eq. were observed extensively around the mountain area. Some public roadways were closed temporarily. The largest

landslide induced by the 526 Eq. was the Dateshita landslide in Tsukidate-cho in northern Miyagi.The muddy collapsed soil flowed from a very gentle slope. The site was situated about 60 km awayfrom the epicenter, as shown in Figure 1.

More than 180 landslides, slope failures and rock failures occurred around the Asahiyama Hillduring the 726 Eq. (Irasawa, 2003). A landslide with similar magnitude and configuration to theDateshita landslide occurred after the main shock at Nishisaruta, Kanan-cho. The Nishisarutalandslide, on the other hand, was very close to the source area of the earthquake as shown inFigure 1. Although the angle of the original slope was steeper than that of Dateshita landslide, theaverage slope angle was less than 30◦.

186





Figure 2. Topography and collapsed configuration of Dateshita landslide. (courtesy of Miyagi Prefectural

Government).

the lower part of the photo is observable, even though it had not rained for 4 days prior to theearthquake. Therefore, the collapsed soil possibly contained much water originally.

Traces of mud splashes were observed at some places. Figure 4b) shows traces of mud splashesremaining on the wall of the house at the western side of the slide. Other traces of mud splashes

with height of about 2 m were observed on the trees along the landslide and on the bamboo on arice field as shown in Figure 4d). The traces evidenced that the flow speed of the slide was notslow. In fact, a witness who lived in a house beside the slide said: “After the shaking intensity

became stronger, I watched the slope movement through a window. The mass of collapsed soilsnaked along the slope down to my house. My wife, who left the house immediately after theshaking, was buried to the waist.” Another witness who lived in the house that was broken by theslide also said: “After the shaking intensified, a propane tank that was set outside the house wallshot into the house, breaking the wall with mud; a supporting column of the house sounded asthough it had broken. Then, I got out of the house after shutting off the electricity. It took about60–90 seconds from the beginning of the shaking. An electric pole, located on the north side of

the house before the earthquake, had moved south, a distance of about 5 m. The collapsed soil had spread on a rice field. Bamboo stands, located on the north side of the road at the north side of the house, had moved to the rice field.” These accounts clarify that the slide traveled a distanceof about 120–180 m over about 60–90 seconds. Therefore, the averaged velocity is estimated to

be about 1.3–3.0 m/s. Konagai (2003) also suggested that the velocity at the center of the streamline was about 6–7 m/s based on kinematic analyses of traces of mud splashes. Moreover, theseevidences indicated that the landslide occurred during or immediately after the principal motion of the earthquake. Furthermore, the first evidence that the slide snaked down was supported by theobservation of traces of mud splashes on the trees along both sides of the slide.

188





Figure 3. Soil profiles and the results of sounding tests (courtesy of Miyagi Prefectural Governm

1 8 9





Figure 5. Configuration of collapsed soil and locations of sampling site, and aerial view of Landslides, after

the earthquakes. (courtesy of Kokusai Kogyo Co, Ltd.)

Figure 6. Fluidized collapsed soil by traveling of heavy construction machines on May 28.

The specific gravity of soil particles was 2.31–2.48, smaller than that of ordinary sand. Thenatural water contents of 26–56%werevery high, as shown inTable 2. These features are attributableto the porous microstructures of pumice, as shown in Figure 8 taken by scanning electron microscope(SEM) with 2,000 times magnification. The soil particle has potentially high water-absorbingcapacity. In fact, after a dry sample was misted with water, the water content of the soil becamegreater 40%. In the tests, the soil sample was dried outside until the initial water content wasabout 1%. The dry density of the sample was 1.1 g/cm3, which corresponded to that of undisturbed

191






Table 2. Summary of physical properties for samples at Dateshita landslide (Refer to Figure 5 for the locations

of the sampling sites).

Sampling Site A B C D E F G H I*** J***

Sampling Method** D U D D U D D D D B

wn(%) 26.1 28.0 30.6 55.7 38.6 31.8 40.9 42.6 39.1 29.8

Gs 2.478 – 2.40 2.313 – 2.438 2.428 2.442 2.354 2.447

Gravel Content (%) 17 – 18 15 – 17 18 – 20 14

Sand Content (%) 53 – 50 54 – 53 50 – 48 53

Silt Content (%) 22 – 23 23 – 20 20 – 24 25

Clay Content (%) 8 – 9 8 – 10 12 – 8 8

Void Ratio – 1.175* – – 0.909* – – 1.159 1.098 –

γ t (kN/m3) – 14.30 – – 17.36 – – 15.82 15.3 –

S r (%) – 59.0* – – 103.6* – – 89.8 83.8 –

*using ρ s at the site., **D; Disturbed, U; Undisturbed, B; Block, ***after vibration of the sample.

Figure 7. Grain size distributions of collapsed soil at Dateshita landslide and Toyoura sand. (Refer to Figure

5 for the locations of the sampling sites).

Figure 8. Microstructure of pumice by SEM (×2,000 magnifications).

samples. In addition, it is noteworthy that the degree of saturation of fluidized soil (sampling at thesite of H and I as shown in Table 2) was about 90%; the fluidized soil was not completely saturated.

2.5 Soil water characteristic curve of the filled soil

Clarifying moisture characteristics of soil consisting of porous particles like pumice is an importantstep in considering the unsaturated strength of collapsed soil. Figure 9 shows the relationship

between suction and water content for the collapsed soil. The results for Toyoura sand is alsoshown for comparison. The Toyoura sand is fine sand which is representative fine sand used for

192






was 0.19 under the conditions of DA= 5% and N= 20. The residual deviator strength after cyclichistories was 10 kPa, which was about 20% of undrained peak strength without cyclic loading.Therefore, it was considered that the trigger of the failure would be the loss of shear strength at thesliding surface.

In addition to this, the present authors researched cyclic shear behavior of unsaturated samples, because the unsaturated soil mass behave like liquid entirely. In the study, both pore water and

pore air pressures were measured during the cyclic shear. As a result, it was found that volcanicsandy soils like Tsukidate soil is easy to liquefy even if the degree of saturation is about 75%. Theother papers written by the authors (Kazama et al. 2006, 2007, Unno et al. 2008) are available for detailed discussion about the liquefaction of unsaturated soils.

2.7 Summary and possible mechanism of the failure

Based on the field investigation and laboratory tests, the features of the Dateshita landslide aresummarized as follows:

(1) The subsurface soil of the gentle slope with the angle of about 7◦ was a fill with pyroclastic

sediments, pumice tuff.(2) It had not rained for a week before the earthquake. Therefore, the upper part of the collapsed

layer was unsaturated to a few meters’ depth. The lower part was possibly saturated before theearthquake.

(3) The soil structure of the fill was very loose, but the unsaturated soil remained stable with highsuction.

(4) The landslide occurred during or immediately after the principal earthquake motion.(5) The slide mass behaved as a mudflow with a small residual strength, because the collapsed soil

was easily fluidized with cyclic shear.

It is likely that the saturated fill liquefied during the earthquake. No strong motion was recorded

close to the Dateshita landslide; however, “5 upper” of the JMA seismic intensity was recorded in Tsukidate town including the landslide area. Moreover, the maximum horizontal accelerationof over 300 gal was recorded at other observation site of the same epicenter distance. The ground motion of 300 gal could easily induce liquefaction of the loose saturated fill (Fukumoto et al. 2007).Moreover, even unsaturated fill was presumably fluidized during shaking, losing the initial shear strength on the rice field. If the unsaturated fill had retained its initial strength, the collapsed soilwould not have spread on the rice field.

Volcanic soils, including pumice, generally crush their particles under the shear process (Miuraet al. 2003). Particle crushing in the shear zone might cause high fluidity of the collapsed soil (Sassa,1996). However, it remains unclear whether particle crushing of Dateshita soil occurs under thelow overburden pressure with the depth of a few meters.

Earthquake-induced landslides of pyroclastic sediments similar to the Dateshita landslide wereoften recorded in past earthquakes as follows: embankment failures in Hachinohe during the 1968Tokachioki earthquake (Mishima & Kimura 1970), slope failure of residential filled ground in Shi-raishi during the 1978 Miyagiken-oki earthquake (Kawakami et al. 1978), large scale debris flow atMt. Ontake during the 1984 Naganoken-seibu earthquake (Kawakami et al. 1985) and the Las Col-inas landslide during the 2001 El Salvador earthquake (Konagai et al. 2002). Further investigationis required along with more strategic laboratory tests, to determine the landslide mechanism.

3 LANDSLIDE AT NISHISARUTA, DURING THE JULY 26 2003 EARTHQUAKE

3.1 Slide overview

During the 726 Eq., a landslide with similar magnitude and configuration to Dateshita landslideduring 526 Eq. occurred after the main shock at Nishisaruta, Kanan-cho, as shown in Figure 11.The Nishisaruta landslide was very close to the source area of the earthquake, as shown in Figure 1.Although the original slope angle was steeper than that of Dateshita landslide, the average slopeangle was less than 30◦. The collapsed portion was about 30 m wide and 30–50 m long, with adepth of about 5 m. The soil debris deposited about 20–30 m wide and 60 m long at the downstream

194




Figure 11. Topography and collapsed configuration of Nishisaruta landslide (courtesy of Tohoku branchOYO-corporation) and aerial view of Landslides, after the earthquakes. (courtesy of Kokusai Kogyo Co, Ltd.)

side. The volume of the collapsed soil was estimated to be about 3,000 m3. The uppermost scarpof the collapsed slope was about 20 m high and 150 m far from the lowermost deposits at a ricefield. The flow of collapsed soil stopped upon touching a narrow road over the rice field. The slidefortunately destroyed no structures and caused no casualties.

3.2 Site geology and topography

The surface soil of a hill behind the slide consists mainly of a inter-bedded medium to coarsesandstone or conglomerate of Miocene. Flattening the top of the hill, farmers made a rice field onthe hill in the 1960s. The collapsed area is located on a fill where the excavated hilltop sedimentswere pushed into a valley at the original topography. The geotechnical condition of the slope closelyresembles the Dateshita landslide.

Figure 11 shows a sketch of the site after the slope failure. The longitudinal cross section at thecenterline of the slide is also shown in Figure 12. The configuration of the original topographyestimated with the results of sounding tests (Irasawa et al. 2003) is also depicted in Figure 12. Themain failure is estimated to be a rotational slip at the upper part, the elevation higher than 10 m.The slope angle at the upper collapsed part gradually changed from about 30 ◦ to 20◦. Then, themass of soil detached from the upper portion and moved down along the original slope. It spread

with high water content on the rice field. The debris carried some trees, originally on the toe of theoriginal slope, near the narrow road with a traveling distance of about 30 m, as shown in Figure 11and Figure 13a). Although the large slide occurred after the main shock, a small slide occurred during the aftershock at 16:56 on the same day around the right side of the scarp in Figure 11.

3.3 Detailed feature and possible motion of the slide

Rainfall was one important feature at the Nishisaruta landslide, in contrast to the Dateshita land slide, Tsukidate, where no rainfall had been observed for a week before the failure. It had rained

195





Figure 12. Cross section and soil profiles of Nishisaruta landslide.

Figure 13. Photographs of Nishisaruta landslide on July 28. (courtesy of Techno Hase Co., Ltd.)

for a week since July 23 in northern Miyagi. The accumulated precipitations of 114 mm for 3 days before the earthquake were measured. The 24-hour precipitation was estimated as 27 mm, whichwas about one-tenth of the maximum records at the nearest observation site, Kashimadai. Therainfall infiltration made the collapsed fill wet, whereas little water was in the rice field on the

hill at the time of the earthquake. Some springs were actually observed on the scarp, as shown inFigure 11. The ground water height estimated by the spring locations is also depicted in Figure 12.In addition, the continuous rainfall helped the collapsed soil deposit on the slope and lower ricefield to retain high water content after the failure. In fact, the deposit was soft with the high water content, as shown in Figure 13b).

Some extensive cracks on the upper rice field were behind the scarp, as shown in Figure 13c).These cracks were running parallel to the edge of the hill. If these cracks were opened during theforeshock, 7 hours before the main shock, rainwater could easily infiltrated through the cracks intothe ground and raise the ground water table in the fill.

196






Table 3. Summary of physical properties for samples at Nishisaruta landslide (Refer to Figure 12 for the

locations of the sampling sites).

Sampling point A B C D E

Sampling method D:Disturbed, U:undisturbed D U D D D

Specific gravity Gs 2.700 2.710 2.709 2.699 2.752

Gravel contents (%) 0 13 12 4 0

Sand contents (%) 76 51 64 49 81

Silt contents (%) 8 21 12 22 8

Clay contents (%) 16 15 12 25 11

Maximum particle size Dmax(mm) 4.75 – – 19 4.75

Liquid limit wL(%) – – 38.5 – –

Plastic limit wP(%) – – 24.7 – –

Plasticity index I p(%) – – 13.8 – –

Natural water content Wn (%) – – 28.8 – –

Figure 14. Grain size distributions of collapsed soil at Nishisaruta landslide (Refer to Figure 11 for the

locations of the sampling sites).

It is noteworthy that the Nishisaruta landslide occurred a few minutes after the principal motion of the main shock, whereas the Dateshita landslide occurred during or immediately after the principalmotion. A witness who lived in the house behind the debris in Figure 13b) described the experienceas: “I felt large vertical vibration during the earthquake in the morning. Holding my baby, I got outmy house with my grandfather. I intended to go to the car in my garage (next to the house). Whenwe were getting in the car, the hill suddenly failed. It took a few minutes after the large vibration had

passed. The slide reached the stone walls of the house in 60–120 seconds. Trees at the toe of the slopethat had moved with the landslide remained standing. After witnessing the landslide, we returned to our house. After the earthquake in the evening (the largest aftershock), the right side of the scarpfailed.” This story clarifies that the slide occurred a few minutes after the principal motion of themain shock. Furthermore, the slide traveleda distance of about 100–150 m in about 60–120 seconds.Therefore, the average velocity was slightly slower than that of the Dateshita landslide.

3.4 Physical and mechanical properties of the filled soil

Some disturbed and undisturbed samples were taken from the points as shown in Figure 11. Table 3

summarizes the results of physical tests for samples. The grain size distribution at B is presented in Figure 14 Similar distributions were obtained at the fill sites of B, C, and D: gravel – about10%; sand – about 55%; silt – about 15%; and clay – about 20%. The results suggest that thesamples originated from the same soil, weathered sandstone, or conglomerate and categorized asfine-graded sand.

Table 4 summarizes the results of physical and mechanical tests for undisturbed samples. Undis-turbed samples were taken by a block sampling method at the uppermost scarp, the point B inFigure 11, where the fill was considered to remain stable. Parameters in Table 4 were obtained

by the following laboratory tests with the undisturbed samples, density tests, permeability tests,

197





Table 4. Summary of physical and mechanical properties for undisturbed samples at Nishisaruta landslide.

Dry density ρd (g/cm3) 1.35

Wet density ρ sat (g/cm3) 1.78

Water content w(%) 32.1

Specific gravity Gs 2.71

Void ratio e 1.01

Coefficient of permeability k (g/cm3) 9.2×10−2

Compression index λ 1.42×10−2

Swelling index κ 6.27×10−3

Initial shear modulus G 0 (kPa ) 7.65×104

Internal friction angle φ f (degree) 43.8

Phase transformation angle φ p (degree) 26.0

Cyclic shear stress ratio (N= 20, DA= 5%) CSR 0.34

Figure 15. Cyclic behavior of collapsed soil at Nishisaruta landslide.

isotropic consolidation tests, CD tests, stress-controlled undrained cyclic shear tests with constantstress amplitudes, and strain-controlled undrained cyclic shear tests with gradually increasing strainamplitude. Shear tests were performed with a conventional triaxial test apparatus.

Next we address cyclic undrained deformation behavior. Figure 15 shows the cyclic stress-strainrelation and the effective stress path obtained from stress-controlled cyclic shear tests with shear stress ratio of 0.3. Test samples were consolidated isotropically: the initial effective stress was50 kPa, and the B value was 0.84. We could not achieve the sufficient B value of 0.95 with standard method because the soil sample contained much fines. However, we adopted the B value of 0.84

in this study because the liquefaction strength with the B value of about 0.8 was almost sameas that with the B value of 0.95 (Tsukamoto et al., 2002). The mean effective stress graduallydecreased with the increase of the cycles and finally liquefied. After the mean effective stress

became almost zero, cyclic mobility, recovering the shear stiffness, was observed. These behaviorsreflected normal cyclic behavior of fine-graded sand.

Kokusho et al. (2004) conducted undrained triaxial tests with undisturbed samples under satu-rated conditions. The initial void ratio and fine contents of the material were about 1.0 and 20%,respectively. The initial void ratio was almost the same value as that shown in Table 4. However,the fine contents were slightly less than that of the sample in Figure 15 in this study. It is reported that the cyclic shear strength ratio was 0.22, smaller than 0.34 in Table 4, under the condition: theconfining pressure of 49 kPa, DA= 5% and N= 20. The discrepancy may possible be attributable

to the different fine content.

3.5 Summary and possible mechanism of the Nishisaruta slide

Based on field investigations, laboratory tests andnumerical simulations, features of the Nishisarutalandslide are summarized as follows:

(1) The subsurface soil of the slope was a fill with fine-graded sand, which originated fromsandstone on the hill.

198





(2) The main failure of the slide occurred at the upper part of the slope. The slope angle at thecollapsed part gradually changed from about 30◦ to 20◦.

(3) The mass of soil detached from the upper portion moved down along the original slope, and spread with high water contents on the lower rice field.

(4) It is possible that the precipitations of 114 mm for three days before the earthquake made thecollapsed fill wet.

(5) The landslide occurred a few minutes after the principal motion of the main shock.

The upper portion of the fill lost strength during the down slope movement and spread on thelower rice field. The residual strength, after undrained cyclic shear loading, of the collapsed soilwas much larger than that of the collapsed soil at Dateshita (Kokusho et al. 2004). Therefore, thesteeper slope angle here possibly allowed the detached soil to spread widely.

The numerical simulation conducted by the present authors is available for understanding thefailure time sequence (Uzuoka et al, 2005). The simulations suggested that the saturated fill lique-fied during the main shock. In addition, the residual excess pore pressure induced by the foreshock affected the slope stability.

4 CONCLUSIONS

We conducted site investigation for two major landslides. In addition, physical and mechanicalsoil tests and preliminary numerical simulations were introduced. We summarized the results asfollows.

The Dateshita landslide was a large landslide induced by the 526 Eq. The subsurface soil of the gentle slope with the angle of about 7◦ was a fill with pyroclastic sediments of pumice tuff.The fill was very loose, but the unsaturated soil maintained stability with high suction. The land-slide occurred during or immediately after the principal motion of the earthquake. The mass of

the slide behaved like a mudflow, and the collapsed soil easily fluidized with cyclic shear. Satu-rated fill liquefy during the earthquake because the maximum horizontal acceleration at the surfacewas estimated to about 300 gal. Moreover, even unsaturated fill was presumably fluidized duringshaking, losing the initial shear strength and spread on the rice field.

During the 726 Eq., a landslide with similar magnitude and configuration to the Dateshitalandslide occurred after the main shock at Nishisaruta, Kanan-cho. The Nishisaruta landslide wasalso one of the largest landslides induced by the 726 Eq. Some features were different from theDateshita landslide. The Nishisaruta landslide was very close to a source area of the earthquake. For this reason, a larger seismic load affected the slope. The angle of the original slope (with the angleof 20◦ –30◦) was steeper than that of the Dateshita landslide. Rainfall was an important featureof the Nishisaruta landslide, whereas no rainfall was observed for a week before the Dateshita

landslide. Moreover, the Nishisaruta landslide occurred a few minutes after the principal motion of the main shock, whereas the Dateshita landslide occurred during or immediately after the principalmotion.

The subsurface soil of the slope of the Nishisaruta landslide was a fill with fine-graded sand that originated from sandstone on the hill. The upper portion of the slope that lost its shear strength

because of liquefaction moved down along the slope, and spread with high water contents onthe lower rice field. Rainfall with precipitations of 114 mm for three days before the earthquake,

possibly moistened the collapsed fill.

ACKNOWLEDGEMENTS

The Japanese Geotechnical Society supported a part of the investigation. The members of thereconnaissance team on the earthquakes in the Japanese Geotechnical Society provided valuableinformation to the authors. In addition, Miyagi Prefectural Government and Techno Hase Co.,Ltd. provided the plane and sectional topographic maps at the Dateshita site. Tohoku branch of OYO Corp. provided the plane topographic map at the Nishisaruta site. Kokusai Kogyo Co., Ltd.

provided aerial photos of the sites. Dr. Motoyuki Ushiyama of former lecturer of Tohoku University provided valuable photos and suggestions regarding landslides. The National Institute for Land

199




and Infrastructure Management, Ministry of Land, Infrastructure and Transport, provided strongmotion records in the July 26 earthquake. Graduate and undergraduate students in the GeotechnicalLaboratory, Tohoku University, were very helpful for the site investigation and laboratory tests. Theauthors wish to express their deep gratitude to these persons and organizations for their assistance.

REFERENCES

Irasawa, M. Ushiyama, M. Matsumura, K. Kawabe, H. Hiramatsu, S. and Higaki, D. 2003. Sediment-related

disasters caused by earthquake in the offing the northern part of Miyagi prefecture in July, 2003 (prompt

report). Journal of the Japan Society of Erosion Control Engineering , 56(3): 44–54 (in Japanese).

Fukumoto, S. Unno, T. Sento, N. Uzuoka, R. and Kazama, M. 2007. Estimation of Strong Ground Motions at

Tsukidate Landslide Site during the 2003 Sanriku-Minami Earthquake –Realization of GroundMotion wave-

forms using the data of Strong Motion Seismometers and Seismic Intensity and Its Fluidization Mechanism

using Laboratory Testing–, Journal of Japan Association for Earthquake Engineering , 7(2): 160–179.

Kawakami, F. Asada, A. and Yanagisawa, E. 1978. Damage to embankments and earth structures due to

Miyagiken-oki earthquake of 1978, Tuchi-to-Kiso, JGS , 26(12): 25–32 (in Japanese).

Kawakami, H. Konishi, J. and Saitoh, Y. 1985. Mechanism of slope failures by the Naganoken-seibu earthquake1984 and the characteristics of pumice, Tuchi-to-Kiso, JGS , 33(11): 53–58 (in Japanese).

Kazama, M. Takamura, H. Unno, T. Sento, N. and Uzuoka, R. 2006. Liquefaction Mechanism of Unsaturated

Volcanic Sandy Soils, Journal of Geotechnical Engineering C, JSCE, 62(2): 546–561 (in Japanese).

Kazama, M. & Unno, T. 2007. Earthquake-induced mudflow mechanism from a viewpoint of unsaturated

soil dynamics, Experimental Unsaturated Soil Mechanics, 112 Springer proceedings in Physics, T. Schanz

(ed.): 437–444.

Konagai, K. Johannson, J. Mayorca, P. Yamamoto, T. Miyajima, M. Uzuoka, R. Pulido, E.N. Duran, F.C.

Sassa, K. and Fukuoka, H. 2002. Las Colinas landslide caused by the January 13, 2001 off the coast of El

Salvador earthquake, Journal of Japan Association for Earthquake Engineering , 2(1): 1–15.

Konagai, K. 2003. Slope failure at Tsukidate (Topography and configuration). Reconnaissance Report on the

May 26, 2003, MIYAGIKEN NO OKI EARTHQUAKE, Joint Delegation Team with Japan Society of Civil Engineers and Japan Geotechnical Society: 9–10 (in Japanese).

Kokusho, K. Hara, T. Tsutsumi, Y. and Hoshino, K. 2004. Mechanical soil properties in slope failure under

seismic loading in Tsukidate-cho and Kanan-cho in Miyagi prefecture. Proc. of the 39th Japan National

Conference on Geotechnical Engineering , 2085–2086 (in Japanese).

Mishima, S. & Kimura, H. 1970. Characteristics of landslides and embankments failures during theTokachioki

earthquake, Soils and Foundations, 10(2): 39–51.

Miura, S. Yagi, K. and Asonuma, T. 2003. Deformation-strength evaluation of crushable volcanic soils by

laboratory and in-situ testing, Soils and Foundations, 43(4): 47–58.

National Institute for Land and Infrastructure Management. 2003. Ministry of Land, Infrastructure and

Transport, Strong motion records at Kanan site during July 26, 2003.

Sassa, K. 1996. Prediction of earthquake induced landslides, Special Lecture of 7th International Symposium

on Landslides. Rotterdam Balkema. 1: 115–132.

Sassa, K. Fukuoka, H., Scarascia-Mugnozza, G. and Evans, S. 1996. Earthquake-induced-landslids: dis-

tribution, motion and mechanics, Special Issue on Geotechnical Aspects of the January 17, 1995

Hyogoken-Nambu Earthquake. Soils and Foundations, 53–64.

Tsukamoto, Y. Ishihara, K. Nakazawa, H. Kamada, K. and Huang, Y. 2002. Resistance of partly saturated

sand to liquefaction with reference to longitudinal and shear wave velocities, Soils and Foundations, 42(6):

93–104.

Unno, T. Kazama, M. Uzuoka, R. and Sento, N. 2006. Change of Moisture and Suction Properties of Volcanic

Sand Induced by Shaking Disturbance. Soils and Foundations, 46(4): 519–528.

Unno, T. Kazama, M. Uzuoka, R. and Sento, N. 2008. Liquefaction of unsaturated sand considering the pore

air pressure and volume compressibility of the soil particle skeleton. Soils and Foundations, 48(1): 87–99.

Uzuoka, R. Sento, N. Kazama, M. and Unno, T. 2005. Landslide during the earthquakes on May 26 and July

26, 2003 in Miyagi, Japan, Soils and Foundations, 45(4): 149–163.

200





Liquefaction and ground failures during the 2001 Bhuj earthquake,

India

H. Hazarika Department of Architecture and Environment System, Akita Prefectural University, Akita, Japan

A. Boominathan Department of Civil Engineering, Indian Institute of Technology Madras, Chennai, India

ABSTRACT: A massive earthquake of magnitude MW=

7.7 struck the Kutch region, GujaratState, India. The earthquake, known as the Bhuj Earthquake, caused wide spread destruction and casualties. Earthquake-induced ground failures, including liquefaction and lateral spreading wereobserved in many areas causing damage to dams, embankments, pipelines and ports, etc. This paper is an attempt to document and identify the available geological, geotechnical and earthquake field data related to this earthquake and the liquefaction induced damage for possible use as reference inthe future earthquake resistant performance-based design. Namely, this report summarizes lique-faction and related ground failures, as well as the resulting damage to major port facilities duringthe 2001 Bhuj earthquake. The relationship between the subsurface ground conditions and theground failures is analyzed. Subsurface marshy land containing tertiary and quaternary sedimentsmay be responsible for the observed widespread liquefaction.

1 INTRODUCTION

On January 26, 2001 a massive earthquake of magnitude MW = 7.7 struck the Kutch (also spelled “Kachchh”) region, Gujarat State, India causing vast destruction and casualties. The earthquakeoccurred at 8:46 am, local time, during a national holiday in honor of India’s Republic Day. Hence,many businesses, schools and government offices were fortunately closed. The earthquake, knownas the Bhuj Earthquake, is one of the largest seismic events of its kind in the last fifty years inIndia. The event lasted for more than 30 seconds. According to the report by the United StatesGeological Survey (USGS), the epicenter of the main shock was located at 23.36 N and 70.34 E,

50 km northeast of the town of Bhuj at a depth of 22 km. Damage was spread over a radius of 400 kilometers from the earthquake epicenter. An important feature of this earthquake was thewidespread liquefaction extending over an area of more than 15,000 sq. km, including parts of theBanni plain, Great Rann, Little Rann and the gulf of Kutch. The Rann of Kutch (the area that coversthe great Rann and the little Rann surrounding the gulf of Kutch) and Banni plain are low-lyingareas (Fig. 1) underlain by a Holocene sedimentary sequence of sand, silt and clay related to deltaicand estuarine deposition into an arm of the Arabian Sea (Malik et al., 1999). Satellite imagerysuggests that liquefaction may have occurred along the coast about 180 km west of the epicenter.The 2001 Bhuj earthquake caused large-scale liquefaction that induced damage and even led tofailure of many multistoried buildings, earth dams, bridges, embankments and port structures in the

Kutch region. Widespread liquefaction resulted in ground subsidence and lateral flows particularlyalong the seashore, riverbeds, ponds and marshlands, and salt playas. It is noteworthy that, therewere reports of liquefaction as far away as from Ahmedabad city, which is located nearly 240 kmeast of the epicenter (Bhuj town).

This paper focuses upon the liquefaction ground failures and the resulting damage to portfacilities during the Bhuj earthquake. The main objective is to document and identify the avail-able geological, geotechnical and earthquake field data related to this earthquake, as well asthe liquefaction induced damage for possible use as reference in the future earthquake resistant

performance-based design.

201




Figure 1. Geologic map and cross-section location ( After Hengesh and Lettis, 2002).

2 THE 2001 BHUJ EARTHQUAKE

2.1 Geologic and tectonic setting

The general nature of the tectonic plate of the Himalayan region, which is associated to the overallnorthward movement (at a rate of 53∼63 mm/year) of the Indian subcontinent and its collision withthe Eurasian plate has been fairly understood. However, the actual local tectonic settings resultingin the individual events have never been studied in detail, primarily due to lack of proper recordingsystem. Consequently, the tectonic setting of the Kutch region is not well understood to date.

India has faced a number of major earthquakes in the past; e.g., the 1897 Assam earthquake(Magnitude 8.7; ∼1,500 deaths), the 1905 Kangra earthquake (Magnitude 8.6; ∼19,000 deaths),the 1934 Bihar-Nepal earthquake (Magnitude 8.4; ∼11,000 deaths), the 1935 Quetta earthquake

(Magnitude 7.6; ∼30,000 deaths; in Baluchistan, now in Pakistan), and the 1950 Assam-Tibetearthquake (Magnitude 8.7; ∼4,000 deaths). The area affected by the Bhuj earthquake has alsoexperienced a large earthquake in 1819 (Magnitude 8.0 Kutch earthquake; ∼1,500 deaths), and a moderate earthquake in 1956 (Anjar earthquake of Magnitude 7.0; ∼115 deaths). There havealso been several moderate earthquakes in India in the last 20 years (e.g., the 1988 Bihar-Nepal:Magnitude 6.6, ∼1,004 deaths; the 1991 Uttarkashi: Magnitude 6.6, ∼768 deaths; the 1993 Latur:Magnitude 6.4, ∼8,000 deaths; the 1997 Jabalpur: Magnitude 6.0, ∼38 deaths; and the 1999Chamoli: Magnitude 6.5,∼100 deaths). However, these earthquakes occurred mostly in rural areas.The Bhuj earthquake is the first major earthquake to hit an urban area of India in the last 50 years.An analysis of the historical seismicity in the Kutch region shows a recurrence of approximately

50 years for large magnitude events. The seismic zoning map promulgated in the national buildingcode of India (IS 1893, 2002), is shown in Fig. 2. It consists of f ive zones denoted as I to V, withV depicting the highest hazard risk. The Kutch region (Bhuj) belongs to zone V. Sufficient prior knowledge, therefore, existed about the seismicity of the region as reflected by the relatively highrates of historical seismicity in the Kutch region, compared to the peninsular India.

The earthquake caused extensive liquefaction over an area of tens of thousands of square kilo-meters, although it produced no primary surface fault rupture. Minor ground cracks were observed on the ground surface, but these features are attributed primarily to liquefaction-induced lateralspreading and/or strong ground shaking, but not to primary slip on the main fault plane. Fig. 3

202





Figure 3. Topographic relief map showing MSK intensity level and general distribution of liquefaction

( Modified from Tuttle and Hengesh, 2002).

ground motions. Therefore, the level of shaking at Ahmedabad may be related to basin and soil

amplification.Strong motion data of three aftershocks, recorded at Bhuj observatory maintained by the IMD

were available which are shown in Fig. 6. An empirical approach using empirical Green functionhas been presented by Iyengar and Raghu Kanth (2006) in which the attenuation of the peak ground acceleration (PGA) during the earthquake was obtained using 5% damped response spectra based on available near field velocity data and spectral response recorder (SRR) data. Results of analysis

by Iyengar and Raghu Kanth (2006) is presented in Fig. 7, which indicates that very close to theepicenter the PGA would have been 0.6 g and at Bhuj city PGA it would have been 0.31g∼0.37 g.This figure also reveals that the PGA atAhmedabad city (at a distance of 240 km from the epicenter)is about 0.1 g, which is close to that obtained from the strong motion records of Fig. 4. Therefore, itcan be presumed that the chart presented by Iyengar and Raghu Kanth (2006) is reasonably reliable.

3 LIQUEFACTION AND LIQUEFACTION-INDUCED FAILURES

Bhuj earthquake generated a variety of liquefaction-related ground failure phenomena, includinglateral spreading, sand blows and waterspouts. Sand blows occurred over a wide area in Gujaratand were even reported from adjoining areas of neighboring Pakistan. The most distant appearanceof liquefaction was reported at Bharuch and Jambusar in south-eastern Gujarat. Liquefaction inthe Great Rann and the Little Rann was extensive covering an area of 10,000 sq. km. Fig. 8shows the liquefaction features with sand blow locations and thicknesses (heights) as well as other

types of deformation structures in the areas covering the Banni plain and the Rann of Kutch.Liquefaction caused damage to several bridges, the Ports of Kandla, Navlakhi and Adani, and numerous embankments in the epicentral regions. In this section, the features of the liquefactionand liquefaction induced failure of structures are discussed in detail.

3.1 Extent of liquefied area

Both aerial and field reconnaissance surveys have reported widespread liquefaction following theearthquake. The topographic relief map of Fig. 3 (in Section 2) shows that the liquefaction and

204





Figure 4. Time histories of the ground accelerations at Ahmedabad.

lateral spreading area covers a large portion of the epicentral area that includes the Great Rann,Little Rann, Banni plains, Kandla River and the gulf of Kutch. Liquefaction was manifest at thesurface as sand blows, such as the one near Ranbir (Figure 9) which continued to spout water for 3 weeks after the earthquake.

Figure 10 shows the landsat telemetric (TM) image of the epicentral area clearly showing thedistribution of liquefaction. Fig. 10(a) shows the image two weeks before the earthquake, and

205





Figure 5. Predominant frequencies of the horizontal ground motion recorded at Ahmedabad.

Fig. 10(b) shows the same image taken two weeks after the earthquake. The image of Fig. 10(b)shows accumulation of surface water produced during widespread liquefaction and subsequent

consolidation. Subsidence of the ground surface may also have occurred due to consolidation and tectonic down-warping.

3.1.1 Geotechnical setting around Bhuj regionThe Rann of Kutch is a complete marshy land with 100% saturated soils. In Bhuj, Gandhidam and Anjar area, loose deposits extending to greater depths resulted in liquefaction, causing full or partialsubsidence of structures in the area. The index properties of soils collected from a site very close tothe epicenter of the earthquake is shown in Table 1. This table suggests that the soil is susceptibleto liquefaction as it contains a large percentage of medium and fine sand with appreciable amounts

of non-plastic fines.Figure 11 shows a typical soil profile of the Port of Kandla (details are discussed in the subsequentsection), which is worst affected by the liquefaction induced damage. The figure indicates a layered soil consisting of soft clay and silty sand deposits. The soft clay in this coastal area is moderatelysensitive, with plastic limit of 20%, liquid limit of 50∼70%, in-situ moisture content very close tothe liquid limit and a very high value of compression index (Cc = 0.8∼1.0).The moderate sensitivityof such marine clay may lead to quick–clay under vibration. This behavior may have contributed to the differential settlement of many structures in the area surrounding the Port of Kandla. A

particular example of differential settlement of structure (The Port and Customs Office Tower) is

206





Figure 6. Aftershock acceleration time histories at Bhuj observatory (23.25◦ N, 69.65◦E).

Figure 7. Attenuation of PGA during the main event ( After Iyengar and Raghu Kanth, 2006).

207






Figure 9. A view of sand blow produced by liquefaction.

Figure 10. Landsat TM image showing the distribution of liquefaction ( After EERI, 2001).

SHAKE analysis, the deconvoluted motion was used as base motion at the bottom of soil profile(as shown in Fig. 12) at which SPT ‘N’ value is 50. Their f inding indicates that the high degree of damage in that area may be due to the amplification of shear waves by the thick sandy soil deposit.Ground response study by Kumar (2006) also found a significant amplification in the upper 4 m

depth.

3.2 Features resulting from liquefaction

3.2.1 Sand blows and sand-blow cratersSand blows were the most common liquefaction features observed in the 2001 Bhuj earthquake.Sand blows resulting from liquefaction during the earthquake range from tens of centimeters to tensof meters in length and up to tens of centimeters in thickness (see Fig. 8). Satellite images obtained after the earthquake suggest that a few sand blows larger than those documented during various field

209






Table 1. Index properties of sand from Bhuj.

Properties Values

Specific gravity 2.66

Percentage of coarse sand 1.00

Percentage of medium sand 39.42

Percentage of fine sand 44.44

Percentage of silt 13.00

Percentage of clay 2.14

Maximum void ratio (emax) 0.71

Minimum void ratio (emin) 0.37

Figure 11. Typical soil profiles at Kandla port.

Figure 12. Typical soil profile of a site in Ahmed-

abad and the assumption in the SHAKE analysis.

survey reports may have occurred north to the epicenter, along the southern margin of the GreatRann. In the areas experiencing lateral flows, sand blows typically formed along fissures oriented

parallel to head-scarp grabens and roughly perpendicular to the down slope direction. There weresome instances (sand blow height= 0–4.99 cm), however, where no particular orientations wereobserved as indicated in Fig. 8.

Unlike sand blows, only a few sand-blow craters were documented in the Bhuj earthquake. Thesand cones are similar in size to those of the sand blows. Their central craters range from about1.5 m to 10 m across. The largest documented sand-blow crater was located near Umedpur, about45 km north-west of the epicenter (Fig. 13). The central crater, which was filled with water evenafter a month of the earthquake, was 10 m by 5 m in plan view. As shown in Fig. 13, the channelsincised in vented sand filled with late phase deposition of silty clay. The vented sand deposit was33 m long, 32 m wide and 26 cm thick.

3.2.2 Lateral spreading Lateral spreading was commonly observed on gentle slopes (1 to 20) in the epicentral area and along rivers and bays at greater distances. In fact, an east-northeast trending, 16 km long and 500 m(see Fig. 14) wide zone of ground cracks, bulges and associated sand blows near the earthquakeepicenter have been attributed to lateral spreading.

In the epicentral area, lateral spreading was responsible for damage to many water wells and pipelines. One such example is an area near the village of Budharmora (see Fig. 8 for location).Local residents described generation of waterspouts of about 1 m high issuing from the ground

210






Table 2. Index properties of soils near Sabarmati river.

Grain size analysis

Natural S.P.T

G S M C moisture content value

Sl. No. Depth(m) % % % % (%) (N)

1 0.00 0 86 11 3 8.51 –

2 1.00 1 84 13 2 8.56 7

3 1.50 0 86 10 4 8.62 –

4 2.00 1 86 10 3 8.74 –

5 2.25 0 87 11 2 8.87 10

6 3.00 0 87 10 3 8.88 –

7 3.75 0 87 10 3 8.95 11

8 4.00 3 80 13 4 8.95 –

9 4.50 0 83 13 4 9.01 –

10 5.00 0 86 11 3 9.04 –

11 6.00 0 85 13 2 9.08 –

12 6.75 0 86 10 4 9.18 1713 7.50 0 87 10 3 9.27 –

14 9.00 3 84 11 2 9.32 –

15 9.75 0 87 10 3 9.35 22

16 10.00 4 83 10 3 9.44 –

17 10.50 0 83 13 4 9.53 –

18 11.25 0 83 13 4 9.65 26

19 12.00 0 86 11 3 9.68 –

20 12.75 2 83 13 2 9.70 32

21 13.50 0 86 10 4 9.83 –

22 14.00 0 87 10 3 9.95 –

23 15.00 2 85 11 2 10.08 –

G= gravel. S= sand M= silt C= clay.

Figure 13. Large sand-blow crater near Umedpur.

fissures following the earthquake (Fig. 15). A water pipe, crossing that area and oriented parallelto the down slope direction, was broken and displaced in several locations. In the upslope portionof the area, ground deformations were characterized by grabens 0.35 m and 1 m wide, tensionscracks and backward-rotated blocks, and sand blows up to 3.5 m long, 50 cm wide and 4 cm thick.Here, the water pipe was broken and separated laterally by 1.35 m across two grabens. In contrast,several uplifted linear features and related ground cracks occur in the down slope portion of thearea. In this area, the water pipe was broken as well.

3.3 Liquefaction related damage to port facilities

Gujarat has 41 ports, all under the control of Gujarat Maritime Board. These ports include onemajor port (Kandla), 11 intermediate ports (including Navlakhi and Adani) and 29 minor ports

211





Figure 14. Map of the Kutch region showing the locations of documented liquefaction features ( After Tuttle

and Hengesh, 2002).

Figure 15. Sand blows along tension cracks and grabens in upslope portion of lateral spread at Budharmora.

(Fig. 16). The Port of Kandla, founded in 1952, is situated toward the eastern end of the Gulf of Kutch and on the western bank of Kandla Creek. Extensive ground and structural failures wereobserved at the Port of Kandla, which is located about 50 km from the earthquake epicenter.

The Port of Kandla covers an area of 1.163 sq. km and extends over 7.5 km of coastline. Itssouthern section contains eight dry cargo berths in a straight line (total length of 2 km), two

passenger jetties, one steel floating dry dock, and one maintenance jetty. Four additional dry cargo berths are planned, which will extend the existing quay walls by 800 m. The southern section of the port has 23 warehouses, while the northern section of the Port of Kandla has six oil jetties. Presentlayout of the Port of Kandla is shown in Fig. 17.

212






Figure 18. Ejection of sand through ground crack. Figure 19. View of the natural ground on which the

Port of Kandla was built.

Figure 20. View of sand boils in tidal near Kandla

creek estuary.

Figure 21. View of cracks below pile cap.

3.3.2 Damage to berthsAs shown in Fig. 17, there were eight (I–VIII) dry cargo berths in service at the Port of Kandla atthe time of the earthquake. More than 2,300 piles in Berths I-V had suffered damage. In contrastto Berths I-V, which had cracks in hollow RC piles (Figs. 21, 22 & 23), Berths VI & VIII had nonoticeable damage. On the contrary, as depicted from Fig. 24, piles in Berth VII were damaged due to slope failure. All the piles in Berths I-V are hollow, except for the first two rows of piles inthe quay walls, which are filled with concrete (Fig. 25). This was one of the reasons for damageapart from the other design related factors, which is discussed elsewhere. Table 3 lists the maincharacteristics of these berths, which are labeled I to IX. Berth IX was under planning during theearthquake. The average pile length is about 18 m.

The piles in Berths I-V were designed in the 1950s by the German engineering firm HeinrichBülzer. As illustrated in Fig. 26, the design of the quay wall panels considered three types of lateralloads: 1) bollard load (FB), which pulls the individual panel towards the sea; 2) ship impact (FA);and 3) earthquake inertial load (FE1 & FE2). Table 4 shows the calculation of lateral earthquake loadsin the quay wall panels of Berths I –V and VII. As evident from Table 4, a very low earthquakecoefficient, α (<0.12) was used in the design, which may have been one factor responsible for the damage. Considering the fact that the Bhuj area belongs to the Zone V of the seismicityzone stipulated in the Indian design code (Fig. 2), use of such low design coefficient is prettymuch questionable. Also, Berths I-V and VII were designed solely on static equilibrium of forces,

214








Figure 22. View of crack between pile and pile cap. Figure 23. Crack 30 cm below pile cap.

Figure 24. View of damaged pile in Berth VII.

Figure 25. Reinforced concrete hollow pile used for foundation of Berths I-V (Kandla Port Trust, 2002).

without any dynamic analysis to account for the effects of transient and permanent displacementsinduced by earthquake ground shaking.

3.3.3 Damage to port buildingsMost of the damage to buildings at the Port of Kandla was concentrated in a 250 ×60 m2 arealocated in the central section of the port. The six-floor tower of the Port and Customs Officeleaned and separated from its adjacent building. Fig. 27 shows the location of the tower buildingand its damaged pattern. In the narrow alley between the tower and waterfront, ground cracksappeared resulting from permanent lateral ground deformation toward Kandla Creek. Ground cracks observed near the Port and Customs office is shown in Fig. 28. The six-story tower, whichwas founded on short cast-in-place concrete piles, tilted and settled differentially (by about 30 cm)

215








Table 3. Main characteristics of dry cargo berth quay walls in the Port of Kandla.

Berths Berth Berth Berth Berth

Characteristics I-V VI VII VIII IX

Total length (m) 1046 274 343 205 203

Number of panel 51 10 14 9 5

Panel length (m) 22.9 27.4 24.7 22.9 40.5

Panel width (m) 23.2 67.3 65.7 55 55

Maximum vessel size (ton) 45000 35000 55000 55000 55000

Draft (m) 9.8 9.1 10.5 10.5 10.5

Live load (kN/m2) 32.3 49 49 49 49

Year of construction Prior to 1955 1985 1992 2000 Under planning

Number of pile rows per panel 7 7 4 9 9

Number of piles in a row 9 12 13 9 9

Total number of piles 2268 756 416 458 435

Average pile length (m) 17.7 38 38 38 38

Pile diameter (m) 0.51 1-1.2 0.75-1 1-1.2 1-1.2

Type of piles Hollow RC RC drilled shaft with steel casing

Figure 26. Horizontal forces used in the design of the quay wall panels of Berths I-V.

as soils beneath it settled and shifted laterally. Fig. 29 shows the states of the building before and after the earthquake. Some tile floors settled uniformly and were littered with fine sand deposits,indicating liquefaction of the soils under the building during the earthquake. Other adjacent tilefloors settled unevenly, but neither sand boils nor water mark were observed. The northern wing of the Port and Customs Office settled more than its southern wing. The ground close to the northernwing shifted laterally toward the waterfront as indicated by about 76 mm of separation at a joint inthe concrete walkway. Due to building debris, this structural collapse was not investigated in termsof geotechnical origins. The two-story building of the North Gate Office, located approximately150 m from the tower, settled more than 30 cm. The shallow foundation of the building settled and tilted as the ground spread laterally toward the Kandla Creek. To the east of the North Gate Office,the asphalt paving at the entrance of a passenger jetty settled more than 30 cm.

3.3.4 Damage to passenger and berthing JettiesIn addition to Berths I-V, two passenger and berthing jetties located in the central section of the

port were also damaged. The berthing jetty for port craft is located about 100 m from the Port and Customs Office tower. This jetty permanently moved toward Kandla Creek due to liquefaction. The

jetty deck is composed of RC beams supported by RC drilled shafts with steel casing, stiffened laterally with a row of rectangular RC columns tied to the deck beams. As a result of the earthquake,the entire jetty deck was permanently displaced toward Kandla Creek, with more than 250 mm

216





Table 4. Calculation of lateral earthquake loads in quay wall panels of Berths I-V and VII.

Design parameters Berths I-V Berth VII

Width of berth panel (m) 23.2 24.7

Length of berth panel (m) 22.9 65.7

Live loading (kN/m2) 32.3 49

Total live load, LL (MN) 17.1 79.5

Dead loading (kN/m2) 23.5 8.1

Total dead load, DL (MN) 12.5 13.2

Earthquake coefficient, α 0.1 0.12

Earthquake force, FE1 and FE2 (MN) 2.1 6.4

Figure 27. The Port and Customs Office tower which sideways and separated 300 mm from the adjacent

building.

Figure 28. The cracking pattern around the area.

Figure 29. Schematic diagram showing the damage

of the Port and Customs Tower building.

217







Figure 30. Rail misalignment due to lateral

displacement.

Figure 31. Failure at the column-deck connection.

Figure 32. Earthquake damage to one of the passenger jetties.

separation between deck sections. The jetty deck deformed, and its crane rails lost their alignment(Fig. 30). This lateral displacement caused failure of the rigid, rectangular RC columns at the deck connection (Fig. 31). In contrast, the RC drilled shafts with steel casings were more ductile, and,therefore were not damaged.

The earthquake also damaged one of the two passenger jetties to the south of the berthing jettyfor port craft. The damage was concentrated in the connections, which allow the floating deck to slide vertically with the tide (Fig. 32). Another passenger jetty was damaged, but its distressoriginated from previous storms, not from the earthquake.

3.3.5 Damage to oil jettiesAs indicated in Fig. 17, the Port of Kandla has six jetties (1-6) for liquid cargo of oil and chemicals.Jetties 1 and 4 were seriously damaged by the earthquake, while Jetty 2 was partially damaged.The steel shell of the shafts has been corroded severely in the 18 years since construction, withexposed corroded rebar observed in the cast-in-place concrete cap. As a result of the earthquake,vertical cracks and shear zones developed at the top of most shafts, and horizontal cracking wasreported below the water line. There was differential vertical movement of the shafts across thedeck, causing cracking of the main beam (Fig. 33).

4 SUB-SURFACE SOIL CONDITIONS AND THEIR RELATIONSHIPS TO DAMAGEDSTRUCTURES AT THE PORT OF KANDLA

4.1 Port structures

Table 5 summarizes the main properties of seven soils types identified at the Port of Kandla. Two-dimensional soil profiles were constructed from the borehole data near berths I-V and berth VII

218







Figure 33. Differential vertical movement of the shafts across the deck causing crack in the main beam.

Table 5. Classification and main characteristics of soils at the Port of Kandla ( After Mahindra acres, 2000).

Soil type 1 2 3 4 5 6 7

Soil Name Grey Grey Grey Grey Yellowish- Reddish- Yellowish-

Very Soft Firm Stiff Brown Brown Brown

Soft Silty Silty Silty Medium Hard Dense to

Silty Clay Clay Clay Dense Silty Very

Clay Clayey Clay Dense

Sand with Clayey

Gypsum Sand Engineering CH CH CH CH SP, SP- CL, CH SP, SP-

Classification SC,SM- SC,SM-

SC & SC & GM-

GM-GC GC, GC

Gravel fraction (%) 0 0 0–4 0 0–12 0–24 0–34

Sand fraction (%) 4–17 7–17 7–11 6–11 67–98 0–49 36–96

Silt fraction (%) 43–51 44–48 45–49 46–51 1–27 27–59 4–32

Clay fraction (%) 40–47 39–46 40–46 40–48 16–47

Liquid limit (%) 55–76 62–68 59–66 54–77 54–77

Plastic Limit (%) 25–29 26–28 25–27 22–29 39–64

Water content (%) 38–70 42–47 34–45 37–64 12.5 18–27 10

obtained from the Port of Kandla (Kandla Port Trust, 2000), and they are shown in Figs. 34, 35 &36. The north south two-dimensional soil profile of Fig. 36 shows that the soil deposits are rather uniformly stratified under Berths I-VIII, except for local variations due to erosion channels in tidalflats.

The superposition of two-dimensional soil profiles and panel cross-sections show that the rigid and heavy box structures of Berths I-IV were embedded in soft soils, whereas the deeper parts of the

piles rested on stiffer soils. During the earthquake, the heavy piles moved laterally as rigid blocks

due to their own inertia and the amplified transient motion of upper soft soil layers. Due to abruptchange in stiffness at the connection between the box structure and piles, the bending momentsin the hollow 500 mm diameter RC piles were concentrated beneath the pile cap, resulting in theobserved flexural cracking of a very large number of piles. Cracks were observed in all the 2,300

piles of the quay wall panels. Similar cracking damage may have extended to the RC battered pilessupporting the transit shed panels because battered piles are less ductile than vertical piles whensubjected to lateral loads. These battered piles have apparently not been inspected, since such aninspection would require excavating and exposing foundations beneath the transit sheds. In contrastto Berths I-V & Berth VII, Berth VI & Berth VIIII performed well during the earthquake. That was

219





Figure 34. East-west soil profiles across the transit shed and parking place of Berth III.

Figure 35. East-west soil profiles at a cross section near shoreline across Berth VII.

Figure 36. North-south soil profiles parallel to shoreline under Berths I-VIII ( Note: horizontal and verticalscales are different).

because of their lighter decks, use of more ductile piles, and the absence of a stiffness discontinuityat the pile caps.

Two typical borehole (BH 2 & BH 9) profiles (close to the proposed Berth IX) are shown inFig. 37. The grain size distribution curves obtained from these boreholes are shown in Fig. 38.The upper soil layer consists of 5 to 10 m thick deposits of soft silty clay underlain by sand and hard clayey materials. The lower soil layers consist of yellowish-brown medium dense clayey sand,

reddish brown hard silty clay, and yellowish brown dense to very dense clayey sand. As seen fromTable 5, the upper soils layers have liquid and plastic limit representative of highly plastic clays,and have in-situ water content close to the liquid limit. The undrained shear strength at berths VIand VII and the SPT (Standard Penetration Test) blow counts (N-value) at Berth V are shown inFig. 39. It shows that the undrained shear strength measured from the field vane shear tests is aslow as 20 kPa near the surface (within the 10 m depth). The strength shows a linear variation withincreasing depth.

The grain size distribution curves (Fig. 38) of the soils under berth IX of the Port of Kandlareveals that the soils have grain size corresponding to fine sands with fine contents less than

220







Figure 38. Grain size distribution of the soils under Berth IX of the Port of Kandla.

Figure 39. Profiles of undrained shear strength (vane shear test) of soils under Berths VI-VII and profile of

SPT blow count under Berth V.

222






Figure 40. Details of the Kandla Port and Customs Office tower.

motions were accompanied by the permanent ground deformation toward the Kandla creek. Therewas no evidence of permanent deformation at the berths, but the berths could have masked it. Onthe contrary, there was ground cracking in areas north of the berths.

4.2 Building structures

The Port and Customs office tower (referred hereafter as building) was founded on 32 short cast-in-place concrete piles. Major structural details of the building are shown in Fig. 40. Each pilewas 18 m long and 0.4 m in diameter. The piles were passing through 10 m of clayey crust and terminated in a sandy soil layer. The loads from the superstructure are transferred to the piles

through the foundation mat, which eventually acts like a pile cap. The service load of the buildingis estimated to be 10749 kN.

Figure 41 shows the borehole profile of the natural ground adjacent to the building. As the soil profile comprises of unconsolidated deposits of interbedded clays, silts and sands, it is quite naturalthat the clayey soils will exhibit stiffness degradation and sandy soils will undergo liquefactionduring strong earthquake. Seismic response analyses of the foundation soils were carried out byDash et al. (2008) based on the procedures recommended by Idriss and Boulanger (2004) and Boulanger and Idriss (2005). The factors of safety (F.O.S) against liquefaction potential and cyclicfailure with depth are shown in the same figure. It is evident from this figure that most part of the claylayer except for the top 2 m undergoes cyclic failure that may have resulted in ground deformation

and cracking. Furthermore, the entire sandy stratum between 10 m and 22 m is likely to haveexperienced liquefaction, which may have resulted in ground settlement and flow failure. A detailed numerical analysis and report on this particular case study can be found in Dash et al. (2009).


The 2001 Bhuj earthquake was a devastating event affecting a large area of the state of Gujarat,India. A scientific and objective understanding of the event could be possible only in terms of

223





Rajendran, K., Rajendran, C.P., Thakkar, M., and Tuttle, M.P. 2001. The 2001 Kutch (Bhuj) Earthquake:

coseismic surface features and their significance, Current Science, 80(11): 1397–1405.

Saikia, C.K. (2002). Geology and Seismology (Seismicity), Bhuj, India Earthquake of January 26,

2001reconnaissance report , EERI, California, USA: 33–43.

Sitharam, T.G. and Govindaraju, L. 2004. Geotechnical aspects and ground response studies in Bhuj

Earthquake, India, Geotechnical and Geological Engineering , 22: 439–455.

Tuttle, M.P., and Hengesh, J.V. 2002. Ground failures and geotechnical effects (Liquefaction), Bhuj, India Earthquake of January 26, 2001reconnaissance report , EERI, California, USA: 79–100.

United States Geologic Survey, USGS 2001. Preliminary Report on the 2001 Bhuj earthquake, India

http://neic.usgs.gov/neis/bulletin/mag7.html

226








Las Colinas landslide caused by the 2001 El Salvador Earthquake

K. Konagai Institute of Industrial Science, University of Tokyo, Japan

R.P. Orense Department of Civil and Environmental Engineering, University of Auckland, New Zealand

J. Johansson Norwegian Geotechnical Institute, Oslo, Norway

ABSTRACT: An earthquake of magnitude M w = 7.7 occurred in the southeast coast of El Sal-vador on 13 January 2001, causing widespread damage to buildings and several kinds of civilengineering structures due to ground shaking and earthquake-induced ground failures, includingseveral large-scale landslides. The most tragic among these landslides occurred on the steep north-ern flank of the Bálsamo Ridge, where an estimated 200,000 m3 of soil slid. Once mobilized, thelandslide material behaved as semi-fluid mass and traveled northward an abnormally long distanceof about 700 m into the Las Colinas neighborhood of Santa Tecla and covered many houses, buryingmore than 500 people. The slide materials involved pyroclastic deposits, i.e., silty sands and sandysilts. This paper focuses on the features of the landslide, the properties of the soils involved, and the possible mechanism involved in the failure.

1 INTRODUCTION

On 13 January 2001, an earthquake of magnitude M w = 7.7 occurred in the Republic of El Sal-vador in Central America. This was the first major earthquake of the third millennium and thefifth destructive earthquake to affect the small Central American republic in 50 years. This wasfollowed exactly one month later by a second event, of different tectonic origin, on 13 February,which magnified the destruction. These two earthquakes claimed almost 1200 lives and rendered approximately 100,000 people homeless, as a result of direct shaking and secondary failures such

as landslides. Close to 1.4 million people, nearly a quarter of El Salvador’s total population, wasaffected by the earthquake. Economic losses were set at US$1.6 billion, which is equivalent to 12%of the country’s GDP of the previous year (Bommer et al., 2002).

The 13 January earthquake triggered more than 500 landslides across El Salvador and a further 70 occurred as a result of the 13 February earthquake. The most damaging of these landslidesoccurred off the steep northern flank of Bálsamo Ridge, about 2 km west of San Salvador (thecountry’s capital). This landslide originated at an elevation of about 1080 m and traveled northward a distance of approximately 700 m into the Las Colinas neighborhood of Santa Tecla. The verticaldrop from the ridge to the terminus was about 160 m. The slide materials, estimated at about200,000 m3, buried several hundreds of residential houses at the foot of the slope, resulting in

death of more than 500 people. Considering the ratio of the slope height to run-out length of the slide, this indicates very low coefficient of internal friction. Such long run-out relative to thevolume of slide materials involved is rather unusual.

This paper focuses on the characteristics of the earthquake and strong motion records, as wellas the features of the Las Colinas slope, the characteristics of landsliding event, and its impacton the environment. The type of soil involved and the conditions that led to in the long run-outdistance are also discussed. The primary objective of this paper is to document available geological,geotechnical and earthquake field data related to this earthquake-induced landslide for possibleuse as reference in the context of performance-based design.

227




2 GEOLOGIC, TOPOGRAPHIC AND TECTONIC SETTING

El Salvador is one of the smallest countries in Central America, with an area of just over 20,000 km2.It is located in the western border of the Caribbean Plate, under which the Cocos Plate is subductingat an estimated rate of 7 cm/year. Most of the large earthquakes that have occurred in El Salvador are generated in this subduction zone. The other source of Salvadorian earthquakes is the activities

of the Quaternary volcanoes along the volcanic belt running across the country from west to east,and forming part of a chain extending throughout the isthmus from Guatemala to Panama. Dueto their shallow foci and coincidence with large population centers, these earthquakes have beenresponsible for more destruction in El Salvador than the larger earthquakes in the subduction zone(White and Harlow, 1993).

The Las Colinas landslide occurred in the northern flank of the Bálsamo Ridge, the northern boundary of the Cordillera de Bálsamo. The ridge lies west-southwest of San Salvador, with eleva-tions up to 1100 m above sea level. The landscape is rugged, with the ridges rising by 200∼300mfrom valleys. The widths of the crest of these ridges are generally narrow, often in the range of around 50 m. The hillsides are generally steep, typically greater than 30 degrees.

The Cordillera de Bálsamo is underlain by the Bálsamo Formation, which consists of maficvolcanic breccias, lavas, welded ignimbrites, and other well-indurated volcanic rocks. In the area,the top of the Bálsamo Formation is marked by weathered soil layer that contains enough clayand fine-grained material that it poses a barrier to downward-draining surface layer that percolatesthrough the porous overlying, young volcanic deposits (Jibson and Crone, 2001).

The surface soil in the Bálsamo Ridge consists mainly of volcanic tuff sediments, commonly poorly consolidated young pyroclastic and epiclastic deposits (Schmidt-Thomé, 1975). The brownand yellowish medium to fine grained layers are referred to locally as Tobas color café, while thewhite deposit, the Tierra blanca, consists of light-gray to white, fine-grained pumice ash. Withinthis ash are numerous large blocks of pumice which are commonly concentrated in layers, althoughthey are almost unsorted. These deposits are underlain by volcanic epiclastic and pyroclastic rocks

with intercalation of andesitic lava flows.Although these volcanic ash deposits can form almost vertical slopes in incised ravines and

in road cuts, they are susceptible to sudden and catastrophic failure under sustained or intenserainfall and under earthquake shaking. Schmidt-Thomé (1975) pointed out that these thick, poorlyconsolidated volcanic deposits are especially sensitive to erosion by surface water, and the erosionis particularly intense if the protecting overgrowth and, along with it, the soil cover were originallylacking or have been removed.

Based on a review of topography, lithology, rainfall, seismic hazard and historical cases of earthquake-induced landslides conducted by Rymer and White (1989), Bommer et al. (2002)confirmed that the landslide hazard in Bálsamo Ridge is high. Earthquake-induced landslides arecommon, where soil and rock slides on volcanic slopes and soil falls and slides in steep slopes of

pumiceous ash dominate.Two main geological fault trends exist near the crest of the ridge, one in NW-SE direction and the

other at ENE-WSW direction (Baxter, 2001). The latter fault systems can be traced great distancesalong the northern scarp of the ridge and they pass through the crest of the slope in Las Colinas.

Regarding climatic condition, the average annual precipitations reported by the MeteorologyDepartment of the Salvadorian Ministry of Agriculture (MAG) indicate that rainfalls for the year 2000 were slightly low in many parts of the country, at least compared to the previous two years (seeTable 1), although it should be noted that 1998 was an exceptional year because of Hurricane Mitch(Bommer et al., 2002). It is worth mentioning that the 13 January earthquake occurred during thedry season, when slopes are generally in a stable state.

Table 1. Annual average precipitations at rainfall stations near San Salvador

(after Bommer et al., 2002).

Year Ilopango Puente Cuscutlan

1998 1958 mm 2037 mm

1999 1504 mm 1303 mm

2000 1454 mm 1637 mm

228




3 THE 13 JANUARY 2001 EARTHQUAKE

3.1 Overview

On 13 January 2001, an earthquake occurred off the coast of El Salvador, about 110 km south-southeast of the capital, San Salvador (see Fig. 1). The source parameters for this earthquake, as

obtained by different agencies, are given inTable 2. It canbe seen that they are remarkablyconsistentin terms of size and depth. The epicenter of this earthquake, considered as among the strongest tohit El Salvador in recent years, was located within the Caribbean plate above the subducting CocosPlate.

Figure 1. USGS focal mechanism solutions and CIG aftershocks distribution of the 13 January 2001 earth-quake (dark gray) and 13 February 2001 earthquake (light gray). The approximate rupture area corresponding

to USGS fault plane solution is shown by a rectangle. The epicenter determined by CIG and USGS are shown

by a star. The faults within El Salvador volcanic range are shown (after JSCE, 2001).

Table 2. Source parameters for the 13 January 2001 earthquake.

Epicenter

Time (UTC) N◦ W◦ Depth (km) Magnitudes Agency

17:33:32 13.049 88.660 60 M w = 7.7, M s = 7.8, mb = 6.4 NEIC17:33:46 12.97 89.13 56 M w = 7.7, M s = 7.8, mb = 6.4 HRV

17:33:30 12.868 88.767 60 M w = 7.7 CASC

17:33:31 12.8 88.8 60 M w = 7.6 ERI

NEIC – National Earthquake Information Center.

HRV – Harvard Geophysical Laboratory.

CASC – Central America Seismological Consortium.

ERI – Earthquake Research Institute, University of Tokyo.

229





Figure 2. Location of strong-motion recording stations in El Salvador (after JSCE, 2001).

The main shock was also felt in neighboring countries, such as in Guatemala, Honduras, Nicaragua, Costa Rica, and as far as Mexico City. The intensity of shaking based on the Mod-ified Mercalli Scale was VII-VIII in El Salvador and VI∼VII in south of Guatemala and southeastof Honduras.

The earthquake claimed more than 800 lives and approximately 100,000 people became home-less, as a result of direct shaking and secondary failures such as landslides. Close to 1.4 million

people, nearly a quarter of El Salvador’s total population, was affected by the quake.

3.2 Strong motion recordsThe 13 January 2001 earthquake was well-recorded by three accelerograph networks in operation inEl Salvador: (1) a network of SMA-1 analogue instruments operated by the CIG (Centro de Inves-tigaciones Geotecnicas); (2) a network of digital and analogue instruments operated at geothermaland hydroelectric plants by GeSal; and (3) the TALULIN network of digital SSA-2 instrumentsoperated by the Universidad Centroamericana (UCA) Jose Simeon Canas. The locations of theinstruments are shown in Fig. 2. Records were also obtained from the network of INETER in

Nicaragua. The records from the CIG network were digitized and processed by the USGS.Table 3 lists the main characteristics of the strong motion records obtained during the 13 January

earthquake. In the table, the distance, d f , refers to that measured from the assumed fault rupture, as

proposed by Youngs et al. (1997) for subduction zone earthquakes. Except for the CIG stations inSan Salvador for which investigations were carried out as part of a microzonation study followingthe 1986 earthquake (Faccioli et al., 1988), there was no information about soil profiles is availableat the recording sites (Bommer et al., 2002).

The available strong ground motion recordings from seismic stations operated by UCA showlarger amplitudes in the western part of the country compared with the eastern part. The strongestmotion was recorded in the alluvial deposit in La Libertad near the southern coast, with readingsof 1.113 g and 0.575 g in N-S and E-W directions, respectively; however, since no damage wasobserved in this area, questions arose regarding the validity of these recordings.

230





Table 3. Strong motion records of 13 January 2001 earthquake (after Bommer et al., 2002).

Location PGA ( g )

Network Station N◦ W◦ d f (km) N-S E-W U-D

GESAL Berlin Geoth. 13.50 88.53 54 0.459 0.370 0.235

UCA Armenia 13.744 89.501 93 0.601 0.454 0.223UCA La Libertad 13.468 89.327 60 1.113 0.575 0.617

UCA Panchimalco 13.614 89.179 75 0.177 0.154 0.089

UCA San Bartolo 13.705 89.106 85 0.157 0.199 0.166

UCA S Pedro Nonualco 13.602 88.927 50 0.580 0.488 0.439

UCA San Salvador ESJ 13.707 89.201 85 0.301 0.278 0.154

UCA Santa Tecla 13.671 89.279 83 0.496 0.487 0.243

UCA Tonacatepeque 13.778 89.114 93 0.234 0.205 0.263

UCA Zacatecoluca 13.517 88.869 47 0.260 0.314 0.253

CIG Ahuachapan 13.925 89.805 123 0.146 0.214 0.124

CIG Acajutla 13.567 89.833 95 0.098 0.108 0.050

CIG Cutuco 13.333 87.817 125 0.078 0.079 0.063CIG Presa 15 de Sept 13.616 88.550 66 0.152 0.187 0.122

CIG San Salvador DB 13.733 89.150 84 0.225 0.250 0.160

CIG San Salvador RE 13.692 89.250 83 0.304 0.323 0.329

CIG San Miguel 13.475 88.183 107 0.136 0.120 0.089

CIG Sensuntepeque 13.867 88.663 81 0.082 0.061 0.058

INETER Boaco 12.473 85.658 336 0.004 0.003 0.002

INETER Chinandega 12.632 87.133 175 0.090 0.070 0.042

INETER DEC 12.124 86.267 276 0.045 0.044 0.028

INETER Esteli 13.092 86.355 263 0.014 0.011 0.009

INETER Granada 11.937 85.976 312 0.009 0.009 0.006

INETER Jinotega 13.086 85.995 302 0.006 0.005 0.004

INETER Juigalpa 12.107 85.372 371 0.003 0.003 0.002

INETER Leon 12.117 86.266 276 0.040 0.037 0.026

INETER Managua (ESSO) 12.144 86.320 270 0.057 0.045 0.022

INETER Managua (INET) 12.149 86.248 277 0.034 0.041 0.014

For the Las Colinas landslide, the site of interest is Santa Tecla, located about 83 km from theepicenter and about 1 km away from the landslide site. The station is situated in soil deposit adjacentto a large hospital warehouse. The time histories of the recorded motions along N-S, E-W and U-D

directions are shown in Fig. 3(a), while the acceleration response spectra for 5% damping areillustrated in Fig. 3(b). It can be observed that the duration of strong shaking is approximately10 sec with a predominant period of about 0.26 and 0.50 sec for the N-S and E-W directions,respectively, suggesting higher energy contents at higher frequencies.

3.3 General features of damage

One of the most spectacular aspects of the January 13 earthquake was the damage inflicted bylandslides. The data compiled by the National Emergency Committee (COEN, 2001) indicate thatabout 500 landslides and other shallow slope failures have occurred in various parts of the country

as a result of the earthquake.Most of these landslides occurred in the central volcanic ridge running east to west of the country.Of particular interest are the large-scale landslides and shallow slope failures that occurred in theCordillera del Bálsamo. Notable among these are the landslides which occurred in Las Colinas and in Comasagua town. Other landslides were reported in the region east and north of Lago Ilopangoand some near Lago Coatepeque (east of Volcan Santa Ana), in San Vicente, and in an area east of Usulután (NLIC, 2001; OCHA, 2001). Except for the Las Colinas landslide, most of the landslidesgenerally occurred in sparsely populated areas, and they have disrupted transportation routes, suchas the Pan American Highway.

231




Figure 3. (a) Acceleration records at Santa Tecla (original record from UCA, 2001: The digitized records are

in the attached CD-ROM); (b) Acceleration response spectra for 5% damping. Note: 1 g = 980 gal.

The numerous slope failures that have been observed along the slopes of the Bálsamo Ridge can be classified into three categories: (1) shallow slope failures in the hillsides, typically involvingmovement of surface soils; (2) deep rotational slides in the slopes; and (c) rockfalls. Most of the

slope failures observed were shallow, typically less than 1m thick. They involved either the surfacesoil or the weathered zone of the rock formation and movement is generally translational. It is likelythat the surface soil or the thickly weathered portion at the hilltop was stripped off by tensile forcesgenerated by the high-intensity seismic shaking, and the soils slid down the hill slope as debris.In some cases, deep rotational slides, such as those observed in Las Colinas and in Comasagua,caused the movement of large volume of soil. Rockfalls were also prevalent, especially in slopesof road cuts.

In addition, it was observed that the slope failures, at least those that occurred in Bálsamo Ridge,occurred mostly on the northern face of the slopes. This can be attributed to directivity effects of the seismic motion. In some sites visited, clear stratification of the tuffs and pumice was distinctlyobserved; however, in other areas, a much more heterogeneous cross-section was noted. Some soil

layers in the slope were moist, but not saturated. The geographic distribution of landslide sitesroughly corresponds to locations of young volcanic soils in valleys.

In addition, soil liquefaction occurred in the alluvial plains near the coast, specifically near Rio Lempa (Lempa River). Sand boils and ground cracks were observed adjacent to the river, and liquefaction-associated lateral spreading was also noted near the riverbanks. One of the spans of an old railway bridge collapsed due to the movement of the riverbank induced by soil liquefaction(Orense et al., 2002).

4 THE LAS COLINAS LANDSLIDE

The Las Colinas landslide, which occurred on the northern flank of the Bálsamo Ridge, caused thegreatest loss of life in a single location from the 13 January earthquake. In this section, the featuresof the landslide are discussed in detail.

4.1 Features of the Las Colinas landslide

4.1.1 General damageAs mentioned earlier, the most disastrous landslide associated with this earthquake occurred in LasColinas, located in Nueva San Salvador (Santa Tecla), southwest of San Salvador, where the sliding

232





Figure 4. A bird’s eye view of the Las Colinas landslide.

materials covered several hundreds of residential structures. The slide, shown in Fig. 4, occurred off the northern flank of Bálsamo Ridge. The failure is generally rotational in the upper part of theslope and translational in the middle and lower portions. The section involved in the slide is about80 m wide, with a depth of about 20 m at the rotational part. The mass of soil detached from theupper portion of the slope appears to slid down and traveled a distance of about 700 m, buryinghundreds of residential houses along the way. Cars hit by the moving mass of soil were completelydemolished. It is estimated that the volume of materials involved in the flow is about 200,000 m3.Extensive networks of cracks were observed on top of the crest, indicating potential instability.

4.1.2 Configuration of the slopeThe Las Colinas sliding surface was surveyed by the JSCE Reconnaissance Team using a laser- based theodolite connected to a portable computer. The surveyed slope was then plotted using amap software. The following description of the survey was taken from the JSCE Report (2001).

Based on field measurements, the distances from the top end of the scar to the toe of the slopeand to the terminus of the slid material were 480 m and 700 m, respectively (JSCE, 2001). Theelevation of the head-scarp and the toe of the slide are 1080 m and 920 m, respectively, indicatingaverage gradient of 160 m over a distance of 700 m in the damaged area.

As shown in Figs. 4 and 5, the failure surface can be roughly divided into three zones. Theuppermost zone (Zone 1) is a hollow of about 100 m in diameter, which was caved in some10–20 m from the original ground surface. North beyond the hollow, there appears a steepest zone

(Zone 2) which becomes gradually gentle as it comes close to the toe (Zone 3). The inclination of the slope is about 26 degrees above the elevation of 960 m in Zone 2, and is about 9 degrees belowthe level in Zone 3.

In Zone 2, two ravines coming down from both sides of the slope curve forward and meet at themiddle of the slope width. These ravines are shown also in the original topography. Therefore, thesurface configuration in Zone 2 after the earthquake roughly remained as it was.

To estimate the total volume of the sliding soil mass, the original slope data provided by theMinisterio de Obras Publicas (1970) was digitized and compared with the slope obtained from field measurements. Based on this procedure, the total slide volume was estimated to be in the order of 200,000 m3.

4.1.3 Detailed features of the slope4.1.3.1 Uppermost main scarp (Zone 1)Figs. 6 show two photographs of the same soil layers appearing on the west half of the uppermostscar. The left photo was taken on 14 January, while the right photo was taken on 4 February. Dark and blight colored stripes appearing on the scarp shows a stratified soil profile with the top lapillituff layer of about 2 m thickness overlying a pair of two differently colored (ocher and white) pumicelayers of about 12 m thickness. Some bedding planes separate the lapilli tuff into some sub-layersof varying thickness (10–25 cm). Dark brown color of soil indicates that the soil is wet, and the pair

233





Figure 5. Details of the Las Colinas slope as measured in the field. The green marker delineates the location

of the failed slope (JSCE, 2001).

Figure 6. Photos showing the west slope of the main scarp taken (a) on January 14 (photo by Mr. JoseAntonio

Rivas); and (b) on February 4.

of photos shows that the dark colored stripes were steadily thinning, i.e., the exposed soils were

drying. This fact may be an evidence that the intact soils along the slip surface were wet beforethe event. Beneath the dark and wet soils, there appear white and/or ocher colored pumice soils.They also appear wet and weakly cemented. The pumice is easily broken into small pieces or into powder just by rubbing together with fingers. Broken fragments of this pumice have sub-angular or angular shapes. Closer investigation revealed that intact pumice also showed cracks and joints.

To map the scar of the landslide, a total of 43 points were marked along the perimeter using aGPS receiver. The coordinates of the scar are plotted in Fig. 7. Also plotted in the figure are thelocations of cracks, which were observed on top of the ridge running in almost west-east direction.The crack openings were measured along line A-A, and based on the amount of crack width, it was

234






Figure 8. View of the steep slope (Zone 2).

Figure 9. View of the toe of the slope (Zone 3).

found on a wall of a house on the east perimeter of the slid soil mass. The parabola had a peak of about 4.5 m high, and then drops downward and reaches the ground after about 5 m horizontal run.Simple calculation indicates that the moving soil mass was traveling at about 5 m/s (JSCE, 2001).The main stream of the soil mass flow might have moved faster after running through dwellingsstanding close together.

236






Figure 10. Location of soil sounding (C1-C3) and soil sampling (S1).

An interesting feature of the Las Colinas landslide is the long runout distance relative to the smallvolume involved. Despite the low water content and relatively low potential energy, the sliding soil

mass traveled 700 m. The equivalent coefficient of internal friction was about 13◦

, inferred fromthe ratio of the slope height to run-out length of the slide. This indicates a significant drop infrictional strength, which may be as high as 39◦, as discussed later. Based on laboratory tests, the

pumice sand deposits can be easily reduced in grain size during motion by fragmentation, graincrushing and fast shearing, which allowed rapid and progressive decrease in available strength

both within the mass and along the topographic surface. Furthermore, simulation studies (Crostaet al., 2005) showed that the flat runout zone with the buildings and roads aligned with the flowdirection allowed the slide to maintain confined conditions during the entire flow duration, therebyincreasing the potential mobility of the landslide mass.

4.2 Soil properties4.2.1 In-situ testsIn order to estimate the strength of the soils in the Bálsamo Ridge, portable cone penetration testswere performed at three points, C1-C3 as indicated in Fig. 10. C1 was located slightly west off the main scarp, while C2 and C3 were positioned farther west of the slide area. The number of

blows obtained from the portable cone penetration test, N d , was converted into equivalent number of blows, N s, from standard penetration tests using an empirical equation (JSCE, 2001).

The variation with depth of the equivalent N s values at the three points is shown in Fig. 11. Ascompared with the other points, all equivalent N s values at Point C1 are considerably low in general.Specifically, extremely low values reaching zero were found at about 1.2 m and 2.5 m depths. Thesesmall N s values suggest the presence of weak layers which, judging from their depths, can be either

pumice and/or fragmented volcanic products.

4.2.2 Laboratory testsThe Pumice sand samples obtained at a point slightly below the west main scarp (S1 in Fig. 10)were examined to obtain their physical properties, as well as static and dynamic strengths usingtriaxial and ring shear apparatus.

Table 4 shows the physical properties of the pumice sand. Its density is comparably low whencompared to natural sand, while its maximum and minimum void ratios, obtained using Japanese

237





Figure 12. Results of soil tests: (a) dynamic strength curves for pumice soil; and (b) Stress paths and failureline for the pumice soil based on ring shear tests (after JSCE, 2001).

Figure 13. Optical micrographs showing the pumice grains (a) before the test; and (b) after the test (from

tests by Yamamoto as reported in JSCE, 2001).

that due to technical and practical limitations of cable length, the measurements at the top and bottom were not performed simultaneously. The Fourier spectra of the three components at the top

of the ridge were divided by the corresponding components at the toe. The results are shown inFig. 14. The three graphs on top depict the average of the Fourier spectra (obtained at four separatewindows of 40.96 sec opening) at the top (red) and at the toe (blue) of the ridge, respectively, whilethe three graphs at the bottom show the ratios of the spectra (i.e., top of the ridge/toe of the ridge).From the plots, it can be observed that the spectral ratios of all NS, EW and even UD componentsshow clear peaks equally at around 1Hz, and the peak is the highest in NS direction, the directionnormal to the rim of the ridge, probably reflecting the topographical effect.

4.3 Cross-sectional profile

Based on several in situ sounding and laboratory testing, Lotti & Associati-Enel. Hydro (2001)characterized the failed area in Las Colinas and its surroundings along Cordillera del Bálsamo. Thecross-section of the slope is shown in Fig. 15. There are four main geological strata in the slopesection: (1) pyroclasts, consisting of the ash layers known as Tierra Blanca, consolidated lapili and

pumice layers; (2) brown ashes, consisting of basaltic fall and epiclastic deposits; (3) Paleo-soils,with 1.0–1.5 m thickness; and (4) consolidated tuffs and ignimbrites. The geotechnical propertiesobtained for each unit are summarized in Table 5. It can be noticed that the weak paleo-soil and

brown ash layers are dipping toward the slope direction, and this may have contributed to the slopeinstability and long run-out distance.

239






Figure 14. Upper three graphs are the average Fourier spectra at the top (red) and at the toe (blue) of the

ridge, respectively. Lower three graphs show the ratios of the spectra (top of the ridge/toe of the ridge).

Figure 15. Cross-section of the Las Colinas slope showing the geological stratum (after Lotti & Associati,

2001).

4.4 Possible causes of the landslide

Based on the field survey, three factors can be speculated as possible triggering mechanism for theLas Colinas landslide: site amplification due to topographic effect, occurrence of liquefaction, and residential development on the slope (Orense et al., 2002).

4.4.1 Site amplificationThere is a strong possibility that the magnitude of shaking during the earthquake may have beenmagnified at the crest because of the topography of the site. Based on the microtremor mea-surements, the lateral components of the ground motions are amplified at the top, and a 1 Hz

240






Table 5. Properties of generalized geotechnical units in Las Colinas area (after Lotti & Associati-Enel.Hydro,

2001 as reported by Sigaran-Loria, 2003).

Pyroclasts Brown ashes Paleosoils Tuffs

Average thickness (m) 15–25 20–80 1.5 –

SPT ( N ) 10-40 10-40 8-10 Refusal

Density, ρ (kg/m3) 1.50 E+03 1.53 E+03 1.76 E+03 1.90 E+03

Dry Unit Weight, γ (kN/m3) 11 11 18

Shear wave velocity, V s (m/s) 90–145 390–750 660–1100

Poisson ratio, ν 0.425 0.327 0.262

Young’s modulus, E (MPa) 60 360 3,780

Shear Modulus, G (MPa) 20 150 1,670

Max. Shear modulus, G max (MPa) 12–32 233–860 268–990 828–2,300

Cohesion, c (kPa) 60–80 30–40 5–10 200

Angle of intenal friction, φ (◦) 30–40 26–34 20–24 35–38

component of the shaking could have been amplified considerably. Roughly estimating from theacceleration response spectra for the seismic record at Santa Tecla (Fig. 3b), peak accelerationlarger than 0.4 g may have occurred at the top of the slope. Ground response analyses performed

by Lotti & Associati-Enel. Hydro, as reported by Sigaran-Loria (2003), indicated an amplificationfactor between 1.3–1.4 on top of the ridge. This observation is also supported by the collapse of one- and two-story reinforced concrete buildings and adobe structures built on top of the ridge.

4.4.2 Possibility of liquefaction

It is also conceivable that liquefaction may have been a factor in the landslide in the Las Colinaslandslide, as proposed by several authors (e.g., Devoli et al., 2001; Jibson and Crone, 2001;Mendoza et al., 2001). The type of movement observed in Las Colinas is similar in many waysto the characteristics of liquefaction-induced flow failures, particularly the high mobility and longtravel distance, such as those observed in Mochikoshi Tailings Dam during the 1978 Izu Oshima-Kinkai Earthquake (Ishihara, 1984) and in Nikawa, Nishinomiya City (Japan) during the 1995Hyogoken-Nambu Earthquake (Sassa, 1996), to name a few.

Field investigations on Zone 1 of the landslide showed that the exposed soils were graduallydrying, and that the intact slip surface was somewhat wet before the event. The steep slope in Zone2 was covered by thin film of mud, including porous fragments of pumice. Moreover, the bottomsurface of the slid soil mass was wet. Ring shear tests performed by Sassa and reported by Konagai

et al. (2002) on pumice from Zone 1 showed that grain crushing during shear of the sample withonly 81% degree of saturation can cause high pore water pressure to build up.

Based on these observations, two conceivable phenomena arise regarding the role of water in theLas Colinas landslide. Firstly, liquefaction may have occurred in the saturated volcanic materials(pumice) overlying the less permeable loamified soil layer, triggering the flow-like failure of the slope. Secondly, water may have been released anywhere within the failure plane during theearthquake, causing the decrease in the shear strength of the materials. If the materials werecollapsible soils with some degree of cementation in unsaturated condition, water infiltration could have easily destroyed the cementation and resulted in collapse of the soil structure.

4.4.3 Residential development A topographic map of Las Colinas circa 1970 is shown in Fig. 10. The top portion of the ridge isin the lower part of the figure. The contour lines, with 25 m interval, as well as the approximateextent of debris flow during the 2001 earthquake, are indicated in the figure. It can be seen that30 years prior to the earthquake, residential development was confined to an area several hundredsof meters from the toe of the slope. In comparison, the extent of development at the foot of theslope can be seen in Fig. 4. Thus, when the earthquake occurred, housing development has not onlycrept at the base of the ridge, but construction was evident on the hillside as well. Although not thedirect cause of the landslide, the subsequent grading and excavation for such development, as well

241




Ministerio de Obras Públicas 1970. Topographic map 1:5,000, Instituto Geográfico Nacional.

National Emergency Committee, COEN 2001. Statistics of the January 13, 2001 Earthquake,

http://www.coen.gob.sv (in Spanish).

National Landslide Information Center 2001. Damaging Landslides in Central America - Earthquake of

January 13, 2001, http://landslides.usgs.gov/html _files/centrala.html.

Office for the Coordination of Humanitarian Affairs, OCHA 2001. Central America - Earthquake OCHA

Situation Report No. 1∼7 , January 13–17, http://www.reliefweb.int.Orense, R.P., Vargas-Monge, W. and Cepeda, R. 2002. Geotechnical aspects of the January 13, 2001 El

Salvador Earthquake, Soils and Foundations, 42(4), 57–68.

Rymer, M.J. and White, R.A. 1989. Hazards in El Salvador from earthquake-induced landslides, Landslides:

extent and economic significance, Rotterdam, Balkema, 105–109.

Sassa, K. 2000. Mechanism of flows in granular soils, Proc. GeoEng2000, Melbourne, 1, 1671–1702.

Sassa, K., Fukuoka, H., Scarascia-Mugnozza, G. and Evans, S. 1996. Earthquake-induced landslides: distri-

bution, motion and mechanisms, Special Issue of Soils and Foundations, Japanese Geotechnical Society,

53–64.

Schmidt-Thomé, M. 1975: The Geology in the San Salvador Area (El Salvador, Central America), A Basis for

City Development and Planning, Geol. JB, B13, 207–228, Hannover, Germany.

Sigaran-Loria, C. 2003. Numerical assessment of the influence of earthquakes on irregular morphologies :analysis of Colombia, 1999 and El Salvador, 2001 earthquakes, MSc Thesis, International Institute for

Geo-Information Science and Earth Observation, Technical University of Delft, The Netherlands, 141pp.

Suzuki, M., Umezaki, T. And Kawakami, H. 1997. Relation between residual strength and shear displacement

of clay in ring shear test, Journal of Geotechnical Engineering, JSCE , 575/III-40, 141–158 (in Japanese).

Universidad CentroAmericana 2001. Strong-Motion Data from the January-February 2001 Earthquakes in El

Salvador , http://www.uca.edu.sv/investigacion/terremoto/20010113.htm.

White, R.A. and Harlow, D.H. 1993. Destructive upper-crustal earthquakes of Central America since 1990,

Bulletin of Seismological Society of America, 83, 1115–42.

Youngs, R.R., Chiou, S.J., Silva, W.J., Humphrey, J.R. 1997. Strong ground motion attenuation relationships

for subduction zone earthquakes, Seismological Research Letters, 68(1), 58–73.

243


http://www.coen.gob.sv/

http://landslides.usgs.gov/

http://www.reliefweb.int/


http://www.uca.edu.sv/













Seismic motions at Hualien LSST arrays during the 1999 Chi-Chi

earthquake

C.H. Chen Department of Civil Engineering, National Taiwan University, Taiwan

S.Y. Hsu National Center for Research on Earthquake Engineering, Taiwan

ABSTRACT: The Large Scale Seismic Test (LSST) at Hualien is an international cooperativeresearch project to study the effects of soil-structure interaction. A 1/4 scale containment model and accelerometer arrays were installed at the seismic active area of Hualien. During the Sept. 21, 1999,Chi-Chi earthquake, Taiwan, entire seismic motions at the Hualien LSST site were successfullyrecorded in the deployed LSST arrays. Acceleration time-histories and associated response spectrafrom surface and subsurface arrays and from containment responses are reported herein to providea field data base for further applications.

1 INTRODUCTION

The effect of soil-structure interaction (SSI) is important for the seismic design of massive struc-tures. Although a variety of SSI analysis techniques and associated computer codes have beendeveloped, when used for prediction analysis, their results may vary significantly. In order toquantify the uncertainties involved, there is a need to have real earthquake data for verification

purposes.Recognizing this need, a Large Scale Seismic Test (LSST) was conducted in 1980’s at Lotung,

Taiwan, under the cooperation of Electric Power Research Institute (EPRI) and Nuclear RegulatoryCommission (NRC) from USA, and the Taipower Company (TPC) of Taiwan (Tang, 1987). Twoscaled (1/4 and 1/12 scales) reinforced concrete containment models were constructed at Lotung,Ilan, a seismic active area in northeastern Taiwan, for seismic tests. Both the models and their sur-rounding soils (at grade and below grade) were fully instrumented to monitor earthquake responses.

Since the completion of the test facility in 1985, more than thirty earthquakes with Richter magni-tudes 4.5∼7.0 have been recorded. These data have been applied to evaluate some current availableSSI methods (EPRI, 1991a, b; Chen et al., 1990). The research results were evaluated to developguidelines for soil-structure interaction analysis (Hadjian et al., 1991).

An international consortium was organized in 1990 to perform a second phase seismic test atHualien, named the Hualien LSST project (Tang et al., 1991). Since the Lotung test was installed ina soft site, a stiff soil site at Hualien was selected this time to construct a quarter scale containmentmodel and a tank model for field tests. The containment model and surrounding areas are denselyinstrumented to monitor earthquake responses. This paper aims at reporting the Hualien projectand the earthquake data recorded during the September 21, 1999, Chi-Chi earthquake in Taiwan.

2 LARGE SCALE SEISMIC TEST AT HUALIEN

The second phase Large Scale Seismic Test (LSST) was carried out at Hualien, Taiwan, under thecooperation of five nations. The participants included: NRC and EPRI (USA),Tokyo Electric Power Company (TEPCO) and Central Research Institute of Electric Power Industry (CRIEPI) (Japan),Commissariat A L’Energie Atominque (CEA), Electricite de France (EdF) and FRAMATOME(France), Korea Institute of Nuclear Safety (KINS), Korea Electric Power Corporation (KEPCO)

245




Figure 2. Surface and downhole strong motion arrays at the Hualien LSST site.

addition, on the location of the A15 vertical array, a station named ST07 that belongs to theSMART-2 array was installed at a depth of 100 m below the ground surface. The purpose of thisstation was to provide earthquake information at deeper elevations. Since A15 and A25 verticalarrays are located at a distance of 52.6 m from the center of the containment model, these arrays can

be regarded as free-field arrays. The A21 vertical array is located just beside the containment modeland will be regarded as the near-field array. For each station in the surface and vertical arrays, theaccelerometers installed are in vertical (V), east (EW) and north (NS) directions, respectively.

2.1.2 Containment model The testing containment model is a hollow cylindrical concrete structure with a total height of 16.13 m (Fig. 3(a)). The roof slab is a rigid large mass with a diameter of 13.28 m and a thicknessof 1.5 m. The roof slab is supported by a 0.3 m-thick cylindrical wall. The basemat has a diameter of 10.82 m and a thickness of 3.0 m. The mass density of the containment concrete is 2.4 ton/m3.The containment model was built on top of a gravelly layer at G.L. −5 m after removing a surfacesandy soil layer. Then the excavated pit was backfilled with compacted crushed stone around thecontainment model. Inside the containment model, a total of 15 accelerometers were installed atfour different heights to monitor the response of the structure during earthquakes. (Fig. 3(b)).

2.2 Local geology

The Hualien LSST site is located on the terrace of Meilun, north of Hualien city. The geologicmap of the Hualien site and its vicinity is shown in Fig. 1 (Honsho, 1990). The terrace deposit of Meilun was formed by upheaval of a Pleistocene marine ground deposit. The Meilun Formation,also known as Meilun Conglomerate Formation, is a gravelly soil layer distributed across the coastalarea near the test site. The thickness of the Meilun Formation is reported to be more than 350 m (Ho,1988). Since no fossil has been found in the Meilun Formation, an accurate age of the formation

247





Figure 3. Containment model of Hualien LSST.

is unknown. However, according to recent research, the Meilun Formation is thought to belong tothe Pleistocene of the Quaternary period because of the geological development features.

Near the LSST site, a magnitude 7.1 earthquake occurred in 1951 as a result from the ruptureof the Meilun Fault. The Meilun Fault has a length of 5 km, and a strike direction of N20◦∼55◦E

as shown in Fig. 1. It is a left reverse oblique slip fault with a maximum horizontal slip of 2.0 m;the southeast side moved 1.2 m upward.

2.3 Soil conditions

Subsurface conditions of the test site were extensively investigated by borings, large penetrationtests, velocity loggings, frozen soil samples and laboratory tests. Boring logs show that the top5 meters are loose sands. Underlying the sand layer, the gravelly formation is the Meilun Formation,a massive unconsolidated conglomerate composed of pebbles varying in diameter from 10 cm to20 cm. Before the construction of the containment model, a downhole velocity measurement at

the site (near the location of the A15 station) showed that the shear wave velocities increase from133 m/s at the ground surface to about 550 m/s at a depth of 50 m. Summarizing all field and laboratory test results, ground profile and soil properties are listed in Table 1 (CRIEPI, 1993).

Backfill materials were also extensively investigated by several methods, such as the LargePenetration Tests, shear wave velocity measurements and laboratory tests on frozen soil samples(CRIEPI, 1993). Based on these investigations, a unified soil profile model for seismic site responseanalysis was deduced and suggested by CRIEPI as shown in Fig. 4. From the cross-hole shear wavevelocity measurements, results show that Backfill 1 (above ground water level), Backfill 2 (belowground water level) and Gravel 1 (underneath the containment model) have shear wave velocities

248





Table 1. Soil properties of Hualien LSST site.

Shear wave Density Poisson’s ratio

Depth Thickness velocity Vs γ t n

Soil m m m/s t/m3

Sand 1 0∼2 2.0 133 1.69 0.38

Sand 2 2∼5 3.0 231 1.93 0.48

Gravel 1 5∼6.5 1.5 333 2.42 0.47

Gravel 2 6.5∼12 5.5 333 2.42 0.47

Gravel 3 12∼20 8.0 476 2.42 0.47

Gravel 4 20∼46 26.0 476 2.42 0.47

Gravel 4 46∼70.6 24.6 550 2.42 0.47

Gravel 4 70.6∼100 29.4 510 2.42 0.47

Figure 4. Unified soil profile and parameters for seismic site response analysis (CRIEPI, 1993).

of 400, 400 and 333 m/sec, respectively. Based on the results of laboratory tests on frozen soilsamples, the shear modulus and damping curves are deduced as shown in Fig. 5. (CRIEPI, 1993).

2.4 Forced vibration tests

To characterize the dynamic characteristics of the model structure, two phases of Forced VibrationTest (FVT) were performed byTEPCO. The first phase FVT-1 was performed on the model structure before backfilled (Fig. 3), and the second phase FVT-2 was performed on the model structure after backfilled (Fig. 4). Results were distributed to all consortium members for blind prediction and associated correlation studies. Based on correlation studies, it was found that the Hualien site has

249




Table 2. Peak accelerations recorded at each station during the Chi-Chi earthquake.

Vertical array station Surface array station Containment station

Peak Acc. Value (gal) Peak Acc. Value (gal) Peak Acc. Value (gal)

Name V EW NS Name V EW NS Name V EW NS

D11 30.77 97.09 67.81 A11 29.99 109.44 73.59 BAE 32.04 100.46 75.01

D12 30.02 75.87 60.73 A12 32.83 104.35 69.09 BAS 32.44 97.63 78.74

D13 27.95 59.9 47.18 A13 33.02 107.64 65.66 BAW 30.51 98.3 78.33

D14 21.92 40.55 46.07 A14 32.12 98.1 65.77 BAN 34.09 102.84 78.1

D21 30.56 88.07 71.17 A15 31.89 86.03 73.92 WLE 34.07 97.32 79.4

D22 28.79 72.32 55.75 A21 31.93 96.88 76.13 WLN 35.21 107.43 80.78

D23 27 52.94 45.45 A22 29.75 97.93 72.76 WHE 74.28 145.9 119.79

D24 23.65 46.07 44.85 A23 33.22 108.76 77.67 WHS 35.62 105.81 101.74

D25 27.01 114.86 66.24 A24 32.66 117.54 87.08 WHW 34.05 106.65 101.96

D26 24.57 81.81 57.43 A25 28.18 118.93 84.48 WHN 39.21 108.96 101.59

D28 21.03 52.26 46.82 A31 31.41 89.26 80.66 RFE 35.29 119.18 126.42A32 25.82 97.8 69.62 RFS 38.88 118.42 129.35

A33 27.49 99.6 62.44 RFW 36.01 111.34 126.3

A34 26.85 105.76 63.96 RFN 34.94 112.94 126.54

A35 27.66 105.19 62.97 RFC 31.58 112.5 127.68

Figure 6. Variations of peak ground accelerations recorded by surface arrays.

Figure 7. Five percent damping ratio response spectra along ARM1.

in the NS direction. The maximum responses in the EW direction occurred at a frequencyof 2 Hz, but in the NS direction at a frequency of 3 Hz. Besides, it can be observed that themaximum responses in all three ARMs increased a little with respect to the distance from thecontainment.

251






Figure 8. Five percent damping ratio response spectra along ARM2.

Figure 9. Acceleration time histories recorded by the A15 vertical array.

3.2 Ground motions on vertical arrays

At the Hualien LSST site, ground motions at different depths were recorded by the three vertical

arrays A15, A25 andA21. In each vertical array, four downhole accelerometers were installed at thedepths of 5.3 m, 15.8 m, 26.3 m and 52.6 m, respectively. All accelerometers functioned properlyduring the Chi-Chi earthquake, except the D27 station in A25 vertical array.

Ground motions recorded by the three vertical arrays are summarized and compared as follows:

(1) Acceleration time histories recorded by the A15, A21 and A25 vertical arrays are shown inFigs. 9, 10 and 11, respectively. From the depth of 52.6 m upward to the ground surface, thewaveforms are similar, but gradually amplified. The PGA’s recorded by the three vertical arraysare summarized in Table 2 and plotted in Fig. 12. This figure shows the upward amplification

252






Figure 10. Acceleration time histories of recorded by the A21 vertical array.

Figure 11. Acceleration time histories recorded by the A25 vertical array.

253






Figure 12. Variations of PGA on vertical arrays.

Figure 13. Five percent response spectra along A15 vertical array.


of PGA to the ground surface. It is important to note that the major amplification occurred at

shallow soil layers, within 30 m from the ground surface approximately. More importantly, itcan be seen that the PGA at the depth of 52.6 m are similar in both EW and NS directions inthe three arrays, but the amplification with decreasing depth is much pronounced in the EWcomponents than that in the NS components. The same phenomenon has also been identified for the Jan. 20, 1994 earthquake (Chen & Chiu, 1998).

(2) The five percent damping acceleration response spectra along the A15, A25 and A21 verticalarrays are shown in Figs. 13, 14 and 15, respectively. In comparison, it can be seen thatthe acceleration responses are significantly amplified from the depth of 52.6 m upward tothe ground surface. The primary amplification occurs at the frequency of 2 Hz in the EW

254








components and at the frequency of 3 Hz in the NS components. More specifically, it can beidentified that the major amplification occurred at shallow soil layers, within 15 m from the

ground surface approximately.

4 RESPONSES OF CONTAINMENT MODEL

For the containment model, a dense array of accelerometers was installed inside the structure asshown in Fig. 3(b). Four accelerometers are located on the basement (BAE, BAS, BAW and BAN),two accelerometers on the lower level of the wall (WLE and WLN), four accelerometers on theupper level of the wall (WHE, WHS, WHW and WHN), and five accelerometers on the top surfaceof the roof (RFE, RFS, RFW, RFN and RFC). At each station, three accelerometers are installed

in the V, EW and NS directions, respectively. All accelerometers are used to record the dynamicresponses of the containment model during earthquakes.The responses of the containment model during the 1999 Chi-Chi earthquake can be summarized

as follows:

(1) The EW and NS components of acceleration time history recorded on the north side stationsand the roof-center station (i.e., RFC, BRFN, BWHN, BWLN, BBAN shown in Fig. 3(b))are shown in Fig. 16. Significant amplification of structural response can be clearly identified from this figure.

(2) Five percent damping response spectra at stations BAN, WLN, WHN and RFN are calculated asshown in Fig. 18. It can be seen that the maximum EW response of the model structure occurred

at a frequency of 2 Hz. However, in the NS direction, the maximum response occurred at afrequency of 3 Hz. Besides, maximum responses also occurred at frequencies of 4 and 5 Hz atthe roof stations in the NS directions.

(3) The vibrational characteristics of the containment model structure can be observed from thespectral ratio of the roof response to that of the basemat. By dividing the Fourier amplitudesof the RFC records by the BAN records, the spectral ratios are obtained and shown in Fig. 17.From this figure, it can be seen that the maximum value of amplification is located at 5.6 Hzin the EW direction and at 6.2 Hz in the NS direction. These frequencies can be regarded asthe predominant frequencies of the response of the system during the Chi-Chi earthquake. Themode of vibration is contributed primarily by the rocking vibration of the model structure.


Surface and subsurface ground motions as well as responses of the containment model structure atthe Hualien LSST project during the Chi-Chi earthquake were recorded and reported herein. Thescope of this paper is limited to present the earthquake data recorded and some direct interpretationsof the earthquake responses. The data recorded provides very valuable information for further studies, in particular for ground response analysesandseismic soil-structure interaction studies. Thecontribution of the consortium of the Hualien LSST project is enormous and gratefully appreciated.

255





Figure 16. Acceleration time histories of north side stations on containment model.

Figure 17. Five percent damping ratio response spectra of north side stations on containment model.

Figure 18. Spectral ratios between the station roof and base.

256







ACKNOWLEDGEMENT

The authors are grateful to the Taipower Company for providing the earthquake data of HualienLSST Project.

REFERENCES

Chen, C.H., Lee, Y.J., Jean, W.Y., Katayama, I. and Penzien, J. 1990. “Correlation of Predicted Seismic

Response Using Hybrid Modelling with EPRI/TPC Lotung Experimental Data,”, Earthquake Engineering

and Structural Dynamics, Vol.19, No.7, 993–1024

Chen, C.H. and Chiu, H.J. 1998. “Anisotropic Seismic Ground Responses Identified from the Hualien Vertical

Array,” Soil Dynamics and Earthquake Engineering , Vol. 17, No. 6, 371–395.

Chiu, H.C., Yeh, Y.T., Ni, S.D., Lee, L., Liu, W.S., Wen, C.F. and Liu, C.C. 1994. A new strong-motion array

in Taiwan: SMART-2, TAO, 5, 463–475.

CRIEPI. 1993. “Soil Investigation Report for Hualien Project,” Report, Central Research institute of Electric

Power industry, Tokyo, Japan.

EPRI. 1991a. “Post-Earthquake Analysis and Data Correlation for the 1/4-Scale Containment model of theLotung Experiment,” Report, EPRI NP7305-M, Electric Power Research Institute, Palo Alto, California,

USA.

EPRI. 1991b. “A Synthesis of Predictions and Correlation Studies of the Lotung Soil-Structure Interaction

Experiment,” Report, EPRI NP7307-M, Electric Power Research Institute, Palo Alto, California, USA.

Hadjian, A.H. et al. 1991. “The Learning from the Large Scale Lotung Soil-Structure Interaction Experiment,”

Proceedings, 2nd International Conference on Recent Advances in Geotechnical Earthquake Engineering

and Soil Dynamics, March 11–15, 1991, St. Louis, Missouri, USA.

Ho, C.S. 1988. An Introduction to the Geology of Taiwan, Explanatory Text of the Geologic Map of Taiwan ,

Central Geological Survey, Ministry of Economic Affairs, Taiwan, pp.151–152.

Kobayashi, T., Kan, S., Yamaya, H. and Kitamura E. 1997. “System Identification of the Hualien LSST Model

Structure,” International Journal of Earthquake Engineering and Structural Dynamics, Vol.26, 1157–1167.Honsho, S. 1990. “Geological Structure of Hualien Region in Taiwan,” Central Research Institute of Electric

Power Industry, Tokyo, Japan.

Tang, H.T. 1987. “Large-Scale Soil-Structure Interaction,” Report No. NP-5513-SR, Electric Power Research

Institute, Palo Alto, California, USA.

Tang, H.T. et al. 1991. “The Hualien Large-Scale Seismic Test for Soil-Structure Interaction Research,”

Transactions of the 11th SMiRT , Tokyo, Japan, K04/4.

TEPCO. 1993. “Hualien LSST Project, Status Report of the Forced Vibration Test Results, (Before Backfill

and After Backf ill),” Report, Tokyo Electric Power Company, Tokyo, Japan.

257




Figure 4. The topography of the Chiufenerhshan dip slope landslide.

Figure 5. The dammed-up lake of the She-Tsu-Ken river formed by the debris dam of the Chiufenerhshan

landslide (by M.L. Lin).

262






-900

-450

0

450

900

-900

-4500

450900

-900

-450

0

450

900

A c ( g a l )

-900

-450

0

450

900

TCU071

TCU074

TCU076

TCU089

-900

-450

0

450

900

-900

-450

0

450

900

A c ( g a l )

-900

-450

0

450900

20 25 30 35 40 45 50 55

Time Elapse (Sec.)

TCU078

TCU084

TCU079

Figure 8. The horizontal ground acceleration records in the slope dip direction of the nearest free field stations

on the hanging wall side.

range from 28∼96 kPa and the residual cohesion range from 3∼ 6 kPa for the sandstone. The peak friction angle was about 36 degree, and the residual friction angle was about 33 degree. Therefore,the difference in shearing strength of the specimens is attributed to the difference in cohesionrather than that in the friction angle. Noted that the peak cohesion of the fractured sandstone layer is approximately zero and the friction angle is also much lower compared to other rock layers.Chen (2001) used the samples from the sliding surface to perform slake durability test, uniaxialcompression test, point load test, Brazilian test, and direct shear test. The slake durability test of tested material ranges from medium to high according to Gamble (1971). The result of uniaxialcompression test result ranges from 1339 to 3734 kPa with the average of 2911 kPa. The resultof point load test ranges from 7 to 37 kPa with the average of 17 kPa. The result of Brazilian test

ranges from 258 to 488 kPa with the average of 375 kPa. The base friction angle ranges from 22 to32 degree with the average of 28 degree. The peak friction angle ranges from 28 to 50 degree withthe average of 39 degree (Chen, 2001).

5 CAUSES OF THE LANDSLIDE

The weather condition before the Chi-Chi earthquake was fairly dry without much precipitationin the nearby area of Chiufenerhshan landslide. Among the rain fall gauge stations installed by

265




-900

-450

0

450

900

-900

-450

0

450900

-900

-4500

450

900

A c ( g a l )

-900

-4500

450

900

TCU071

TCU074

TCU076

TCU089

-900-450

0450

900

-900-450

0

450

900

A

c ( g a l )

-900

-450

0

450

900

20 25 30 35 40 45 50 55

Time Elapse (Sec.)

TCU084

TCU079

TCU078

Figure 9. The vertical ground acceleration records of the nearest free field stations on the hanging wall side.

the Central Weather Bureau, the Sun Moon Lake station is the closest station to the epicenter of Chi-Chi earthquake and Chiufenerhshan landslide area. The location of the Sun Moon Lakestation is at 120◦ 3225 east longitude 23◦ 3148 north latitude, and the precipitation record of September, 1999 is as shown in Figure 16. The daily precipitation was smaller than 8 mmsince September 17th, and thus it was unlikely to have significant effects of ground water pressureduring the earthquake. However, the sandstone with thin shale layer of Changhukeng formationis not so well cemented. With the cracks in the brittle sandstone, water can easily filtrate into thesandstone and pond on the shale, which will lead to weathering and weakening of the shale. TheGiou-Tsei-Hou river joined the She-Tsu-Ken river and flowed from west to east passing the toe of

the dip slope as shown in Figure 12. The erosion action of the riverbed at the toe of the slope caused the upper part of the Changhukeng formation to daylight, which provided the necessary conditionfor the dip-slope landslide. Although with such geomorphollogical and geological conditions, no

prior dip slope landslide record was found, which might be contributed by the mild slope anglecomparing to the strength of the sandstone/shale. A back analysis performed by Lin et al. (2000a)using pseudo-static analysis for the stability of the dip slope, taking the profile A-A as shown inFigure 4. With the slope profile as shown in Figure 17, and assigning cohesion ranging from 220to 330 kPa and friction angle ranging 22◦ to 28◦ based on Chen (2001), the critical acceleration of 180 gal in vertical direction, and 280 gal in horizontal dip direction was determined while the safety

266




Figure 10. The horizontal peak ground acceleration in the slope dip direction at the Chiufenershan landslide

by interpolation of the data from the nearest free f ield stations on the hanging wall side. (SML represents Sun

Moon Lake).

Figure 11. The vertical peak ground acceleration at the Chiufenershan landslide by interpolation of the data

from the nearest free f ield stations on the hanging wall side. (SML represents Sun Moon Lake).

267






Figure 12. The geological map of Chiufenerhshan landslide area (plot from the geological map of Taiwan

1:50,000, Puli, the Central Geological Survey, 2000).

268





Figure 13. The geological profiles in the direction of NW-SE. Geological formations and structures are

defined as follows: TL=Tanliaoti Shale; SM=Shihmen Formation; CHb, CHm, CHt=bottom,middle and top

Changhukeng Formation; KC=Kueichulin Formation; SF=Shuilikeng fault; TS=Taanshan syncline. (Chang

et al., 2005).

Figure 14. The location of bore holes located along the west part of landslide scarp. (revised from Tseng,

2004).

factor equaled to 1. The critical acceleration thus determined were significantly smaller than the peak ground acceleration determined in the previous section as 270 gal in vertical and 450 gal inhorizontal, respectively. Due to no prior record of landslide records caused by the rainfall, and the

possible weakening of the shale formation due to the perched precipitation, it is likely that the mostimportant factor for causing the Chiufenerhshan landslide is the strong ground motion induced bythe Chi-Chi earthquake, which causes the sliding to occur along the bedding of weaken formation.

269






0

2

4

6

8

10

12

14

16

18

20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

R a i n f a l l ( m m )

Day

EQ

Figure 16. The precipitation record of September, 1999 of the Sun Moon Lake rain gauge station, the Central

Weather Bureau.

Figure 17. The slope profile used for back analysis of critical acceleration by Lin, et al. (2000a).

by the earthquake are also discussed. No prior dip slope landslide record was found, which might be contributed by the mild slope angle comparing to the strength of the sandstone/shale. There wasalso not much of rainfall prior to the earthquake, which implied not much influence of the ground water pressure. Based on the back analysis performed by Lin et al. (2000a), the critical accelerationthus determined were significantly smaller than the calculated peak ground acceleration. Thus it

271





is suggested that the most important factor for causing the Chiufenerhshan landslide is the strongground motion induced by the earthquake.

REFERENCES

Central Geological Survey, 2000, The geological map of Taiwan 1:50,000, PuliChang, K.-J., Taboada, A. and Chan, Y.-C., 2005. Geological and morphological study of the Jiufengershan

landslide triggered by the Chi-Chi Taiwan earthquake. Geomorphology, 71: 293–309.

Chen, J.H., 2001, The study of the engineering geological characteristics of rock material in Juo-Feng-Err-Shan

of the Nantoi aream Master thesis, National Taiwan University

Gamble, J.C., 1971, Durability-Plasticity: Classification of Shales and Other Argillaceous Rocks, Ph.D. Thesis,

University of Illinois, Urbana, Illinois.

Lin, M.L., Chen, T.C., and Wang, K.L., 2000a, The characteristic and preliminary study of landslides induced

by Chi-Chi earthquake, Report of National Center for Research in Earthquake Engineering

Lin, M. L., Wang, K. L., & Chen, T. C., 2000b. Characteristics of the Slope Failure Caused by Chi-Chi

Earthquake, Proceedings of International Workshop on Annual Commemoration of Chi-Chi Earthquake,

III-Geotechnical Aspect, 199–209. National Center for Research on Earthquake Engineering, office for NAtional science and technology Program

for Hazard Mitigation, and Taiwan Geotechnical Society, 1999. Reconnaissance report of the geotechnical

hazard caused by Chi-Chi earthquake, National Research Center on Earthquake Engineering, Taiwan, 111p

Tseng, C.W., 2004, Study on the Observation of slope stability and Characteristics of Groundwater flow at

Joe-Fen-Er-Shan Landslide Area, Master thesis, National Taiwan University

Wang, K.L., Lin, M.L., and Dowman, I., 2007, The observation of landslide coupling uplift of earthquake with

Interferometric Synthetic Aperture Radar – the case study of Chi-Chi earthquake and Ju-Fen-Err mountain

area, EGU General Assembly, Vienna, Austria

Wu, C.Y., 2008, The structural and geomorphic characteristics before and after the Chiufenerhshan landslide

and possible mechanisms of the slope failure, Master thesis, National Taiwan University

272





Tsaoling landslide in Taiwan during the 1999 Chi-Chi earthquake

M.L. Lin & K.L. Wang Department of Civil Engineering, National Taiwan University, Taiwan

T.C. Chen Department of Soil and Water Conservation, National Pingtung University S&T, Taiwan

ABSTRACT: During the 1999 Chi-Chi earthquake, extensive slope failures were triggered bythe earthquake in central Taiwan. Among those, a large-scale dip-slope slide in Tsaoling occurred,

which involved mass movement of about 120 million cubic meters. Four catastrophic dip-sloperockslide events in history in Tsaoling are presented, and the geological and geomorphologicconditions and possible causes of theTsaoling landslide are discussed in this paper. A pseudo-staticstability analysis of the Tsaoling landslide caused by the 1999 Chi-Chi earthquake was conducted to examine the effects of the vertical ground acceleration.

1 INTRODUCTION

On September 21, 1999, an earthquake with the local magnitude of 7.3 struck central Taiwan and

caused catastrophic losses of lives and properties. The earthquake known as the 1999 Chi-Chiearthquake caused extensive slope failures in central Taiwan, landslides occurred from Juolan,Miao-Li County to Alishan, Chia-Yi County. The ground-based reconnaissance report coordinated

by National Center for Research on Earthquake Engineering (NCREE, 2000) documented 436items of landslides, while the Soil and Water Conservation Bureau (SWCB) reported more than20000 items of ground variation based on the SPOT satellite photos taken before and after theearthquake. The distributions of the documented slope failure and items of variations are as shownin Figure 1; the rupture caused by the Che-Lung-Pu fault is also plotted in the figure. Amongthe reported landslides caused by the earthquake, a large scale dip-slope landslide covering anarea of about 400 hectare in the Yun-Lin County occurred involving more than 120 million cubic

meters of soil and rock mass movement. The dip-slope landslide known as the Tsaoling landslidecaused severe casualties and losses of properties. The location of the Tsaoling landslide is shownin Figure 1 to the south of the epicenter of the Chi-Chi earthquake. In this paper, the case historyand causes of the Tsaoling dip-slope landslide are presented.

2 THE DIP-SLOPE LANDSLIDE INDUCED BY THE CHI-CHI EARTHQUAKE

During the Chi-Chi earthquake, in the Tsaoling area of Yun-Lin County, a large scale dip-slopelandslide of the Kou-Long Mountain occurred, which was known as the Tsaoling landslide shownin Figure 2. The location of the dip-slope landslide was situated at 120◦4024 east longitude,

23

◦

34

44

north latitude in the TsaolingVillage, Kou-Kern Township, ofYun-Lin County in CentralTaiwan. The soil-rock mass slid from the northeast near the crest toward the southwest directionduring the earthquake. The landslide was located at a distance of 35 km to the south of epicenter and 6.6 km to the southern tip of Che-Lung-Pu fault as shown in Figure 1. The topography of thelandslide area is as shown in Figure 3, which ranged from elevation of 400 m to 1234 m abovesea level, covering an area of 440 hectare. The average width of the landslide was approximately3800 m, and the length was 2800 m. The large-scale dip-slope landslide in Tsaoling involved movement of rock mass as much as 120 million cubic meters. However, only 25-million cubicmeters (approximately 20%) of the sliding mass dropped into the valley of the Ching-Shui River

273




Figure 1. The locations of Tsaoling landslide, Che-Lung-Pu fault, epicenter, and documented variation by

SWCB in 2000 based on the SPOT satellite images (Lin et al., 2000c).

Figure 2. The wreckage of houses (lower-left), the van (center right), and 39 people were behind the crest of

the dip slope slid with the landslide mass over the Ching-Shui River, and landed on top of the remaining part

of the old landslide dam. (Photo by J.J. Hung).

274






Figure 3. The topographical map of the Tsaoling landslide area (scarp profile is redrawn from Sinotech

Engineering Consultants, Ltd., 2000).

at the toe of the slope. The remaining sliding mass of about 100 million cubic meters slid passingthe Ching-Shui River, and landed on the remains of old landslide dam (called “Dao-Giao-Shan” bylocal people). Residents and their houses sitting on the sliding mass near the crest of the slope flewwith the sliding mass and landed on top of the landslide dam through a horizontal distance of 3100 maway and 800 m downward, and a total of 32 people were killed and 7 survived after the “sliding-landing” process shown in Figure 2. The vegetation on a few hill-slopes downstream of the landslidedam was stripped, most likely by the air-blast released from the impact of the sliding mass. TheChing-Shui River was dammed by the large amount of debris and the dammed-up lake held a total

of 45 million cubic meters of water. The length of the landslide dam measured 4 kilometers fromupstream heel to downstream toe. The plugged length of the Ching-Shui River channel was about5 kilometers. The water depth of the blocked Ching-Shui River valley was 50 meters, which ismuch lower than the height of new Dao-Giao-Shan. Figure 4 is a photo showing a part of the Dao-Giao-Shan from left to right, the dammed-up lake of Ching-Shui River, and the lower section of the remaining dip slope, looking from the southeast after the Chi-Chi earthquake, 1999. In order tomitigate the hazard caused by possible dam breach, the emergency spillway was constructed throughthe plugged section of Ching-Shui River valley. Overflow of the impounded water commenced onDecember 22nd, 1999, without causing damage to the dam. Check dams were also constructed downstream of Ching-Shui River for protection against possible debris flow due to sudden dam

breach. Figure 5 is a bird’s eye view of Tsaoling area taken on June 29th, 2000, in which the overalllandslide area and debris dam can be observed clearly.From the field investigation, it was found that the remaining slope consisted of 4 steps of scarps

each with the height ranging from 20 to 30 m, and all day-lighted as shown in the lower half of Figure 3. The dip angles measured on the remained slope ranged from 12 to 14 degree and the strikeof the slope was N35◦W dipping into south direction. The topographical variation after the landslideis illustrated in Figure 6, based on the changes in digital elevation before and after the landslide.By taking the cross-section A-A in Figure 6, the slope profile and possible sliding surface wasdetermined as shown in Figure 7. The main sliding occurred in the Cholan formation. The total

275





Figure 4. The dammed-up lake and landslide dam of Tsaoling, Oct., 1999. (Photo by T.C. Chen).

Figure 5. Bird’s eye view of dam-up lake, Tsaoling village, and the landslide area, looking to the north-westdirection. (Photo taken on 29th June 2000, courtesy: Mr. Lien, Yung-Wang).

Table 1. The calculated debris volume of Tsaoling landslide from different time frame of

digital terrain model.

Landslide volume (m3) Deposition volume (m3) Time frame

149.9× 106 166.4× 106 1980 and 1999 DEM

116.1× 106 162.0× 106 1980 and 2003 DEM

amount of soil-rock mass involved was about 150 million cubic meters based on the digital terrainvariations. The estimations of the associated soil-rock mass computed using different time frameof digital terrain models are listed in Table 1. Field observation and aerial photographs indicated the existence of tension cracks and graben near the crest of the remaining slope as illustrated inFigure 7.

276






Figure 6. The variation of topography before and after the Tsaoling landslide caused by the Chi-Chi

earthquake.

300

400

500

600

700

800

900

1000

1100

1200

1300

E l e v a t i o n , m

300

400

500

600

700

800

900

1000

1100

1200

1300

T a - W u o S a

n d s t o n e m

e m b e r

C h o - L a n f o

r m a t i o n

1999 landslidedebris

1941 to1942profile surface 1979 profile surface

slip surface after1999

before 1999CHY080

C h i n - S h u

i s h a l e f o

r m a t i o n

Figure 7. The slope profile and sliding surface of the Tsaoling landslide caused by the Chi-Chi earthquake,

1999. (Chen et al., 2004).

3 PREVIOUS LANDSLIDE HISTORY OF TSAOLING

The Ching-Shui River flows pass the toe of the Kou-Long Mountain, which causes the bedding planeto daylight due to the erosion process. Catastrophic dip-slope landslides occurred in 1862, 1941,

1942, 1979, and 1999 (triggered by the Chi-Chi earthquake). Debris of the landslides dammed up the Ching-Shui River, and breaching of the landslide dam of the previous events took place in1898, 1951, and 1979, respectively. A summary of the dip-plane landslide events of the Tsaolingarea is listed in Table 2. A large number of papers and reports on the case history of Tsaoling have

been published (Taipei Observatory, 1942; Chang, 1951; Hsu, 1951; Hsu and Leung, 1977; Hung,1980, 1999; Hung, et al., 1994, 2000; Huang, et al., 1983; Chang and Lee, 1989; Lee, et al., 1993,1994; NCREE, NAPHM, and Taiwan Geotechnical Society, 1999; Lin et al., 2000b; Yeng, 2000).Three major landslide events prior to the landslide triggered by the Chi-Chi earthquake are brieflyreviewed in the follows.

277





Table 2. Summary of the dip-plane landslide events of the Tsaoling area (revised from Wu, et al., 2001)

Landslide Dam break Dam Debris Dam Block

height volume capacity length

Date Time Factor Date Time Factor (m) (m3) (m3) (m)

(1) 1862 – Earthquake – – – – – –

– – – 1875 – – – – – – – – – 1898 – – – – –

(2) 1941.12.17 04:17 AM Earthquake – – – 150× 106

1942.08.10 12:00 PM Rainfall – – – 200 200× 106 120× 106 2,000

1942.08.10 12:10 PM Rainfall – – –

– – – 1951.05.18 – Rain – – – –

(3) 1979.08.15 04:00 AM Rainfall – – – 90 5× 106 40× 106 2,000

– – – 1979.08.24 11:30 AM Rain – – – –

(4) 1999.09.21 01:47 AM Earthquake – – – 50 120× 106 46× 106 4,815

Figure 8. Tsaoling landslide area after 1941 rockslide caused by Chia-Yi earthquake on 17th December 1941,

M= 7.1 (Taipei Observatory, 1942).

Figure 9. Reconstructed ground profile of 1941 Tsaoling landslide event. (The height refer to the spot height

of the landslide dam; shadow areas are deposits after events) (Lee, et al., 1993).

3.1 Landslide event in 1862

The first reported event of dip-slope landslide and subsequent formation of a landslide dam occurred on June 6th, 1862, and was caused by an earthquake with magnitude between 6 and 7. Breachingof the landslide dam occurred later in 1898. (Taipei Observatory, 1942)

3.2 Landslide events in 1941 and 1942

On December 17th, 1941, a rockslide involving a mass movement of more than 100-million cubicmeters occurred and formed the southwest flank of Tsaoling. A strong earthquake known as Chia-Yi earthquake with magnitude of 7.1 hit central Taiwan and triggered the landslide; the photo of the rockslide area in 1941 is shown in Figure 8. On August 10th, 1942, heavy rain caused another rockslide on the remaining slope. More than 150 million cubic meters of the rock mass slid downthe dip slope, and the Ching-Shui River was dammed with debris. The profiles of 1941 and 1942rockslide events were reconstructed as shown in Figure 9, along the same profile line A-A inFigure 6.

278






Figure 10. Reconstructed ground surface profile of 1979 Tsaoling landslide event. (shadow areas are deposits

after events; Lee, et al., 1993).

Following 5 days rainfall with cumulative precipitation of 776 mm, the landslide dam with heightof 140 m to 200 m and width of 4800 m at base was overtopped on May 18th, 1951, and 120 millioncubic meters of water was released. A total number of 137 army engineers, who were installing the

spillway on top of the landslide dam, lost their lives. The resulting flood inundated 3000 hectaresof land and destroyed 1200 houses downstream.

3.3 Landslide event in 1979

On August 15th, 1979, heavy rainfall caused failure of the lower part of remaining slope sincethe previous event with volume of 5 million cubic meters. The sliding debris mass collided with theremaining landslide dam and the Ching-Shui River was once again dammed. Following 2 days of rainfall with cumulative precipitation of 624 mm, the landslide dam with a height of 90 meterswas overtopped on August 24th, 1979. Two bridges downstream were destroyed, and the Tung-Toucommunity in the Chu-Shan Township was flooded. Fortunately, there was no casualty due to field monitoring and a timely warning. The reconstructed profile of the 1979 rockslide event is shown inFigure 10. The lower part of the sliding plane of the 1979 rockslide was in Chinshui shale formation.

4 CHARACTERISTICS OF GROUND MOTION

The Chi-Chi earthquake was caused by the thrusting of the hanging wall side of the Che-Lung-Pufault toward the northwest direction. The ground motion at theTsaoling landslide caused by the Chi-Chi earthquake could be determined using records of the free field strong motion stations installed

by the Central Weather Bureau. (http:// http://www.cwb.gov.tw/V5e/seismic/chichi.htm) The strong

ground motion records of station CHY080 of the Central Weather Bureau located just north to thecrest of Tsaoling as shown in Figures 7 and 11, and thus can be used to represent the ground motion at the landslide site. The peak horizontal ground accelerations were 841.5 gal in North-South direction and 792.4 gal in East-West direction, and the peak vertical ground acceleration was715.8 gal, respectively. The peak values for all three components of ground acceleration were of similar magnitudes. In contrary to records from most other stations, the peak horizontal accelerationin the N-S direction was higher than that of the E-W direction, and the vertical ground accelerationwas about the same magnitude as the horizontal components, and being very significant. This is

probably because three separate triggered events were identified just after the main shock by Shin,1999, and among them the first event occurred with the epicenter close to Tsaoling, as shown in

Figure 12, which had a major effect for inducing Tsaoling landslide.In order to consider properly the effects of ground motion on the dip slope sliding, the horizontalground motion records in both N-S and E-W directions were combined and resolved into thecomponent in dip direction of the slope, with the positive acceleration indicating ground motionoutward of slope. The transformed horizontal ground acceleration in the dip-slope direction isshown in Fig. 11 (d) with a peak acceleration of 646.6 gal. The peak ground acceleration resolved into the dip direction is lower than those in N-S and E-W directions, and the acting duration is about41.95 seconds with acceleration larger than 50 gal. The seismic parameters representing ground motion characteristics are listed in Table 3.

279


http://www.cwb.gov.tw/






10

5

0

5

10

10

5

0

5

10

10

5

0

5

10

A c c e l e r a t i o n

( * 1 0 0

g a l )

Vertical, ZMax:715.8 gal

Horizontal, N-SMAX:841.5 gal

Horizontal, E-WMax:792.4 gal

Station:CHY080

20 25 30 35 40 45 50

Time Elapse (Seconds)

10

5

0

5

10Horizontal, DipMax:646.4 gal

(a)

(b)

(c)

(d)

Figure 11. Ground acceleration records of strong motion station CHY 080, Central Weather Bureau, of

Chi-Chi earthquake (download record form http://www.cwb.gov.tw/V5e/seismic/chichi.htm).

5 GEOLOGICAL FORMATION AND MATERIAL PROPERTIES

Tsaoling is located in northeastYun-Lin County in the foothill region of central Taiwan.The geologyis dominated by sedimentary rocks that dip in the same direction as the slope of the landslide. Thegeological condition (Central Geological Survey, 1999) of the Tsaoling area is shown in Figure 12.The landslide was located between the pair of Tsaoling anticline and Chiuhsiungping synclinerunning from NNE to SSW direction. The landslide area is located on the west side of the Tsaoling

anticline, and the landslide slope oriented NNE–SSE with low angle submergence into Ching-ShuiRiver valley. The site geology is composed of Tawo sandstone (Pliocene epoch), Chinshui shaleand Cholan formation (Pliocene to Pleistocene epoch) from the bottom to the top as illustrated inthe profile map in Figures 7, 9, and 10. The orientation of all strata is N20◦W∼ N40◦W and dippinginto the direction of 14◦ SW, which is approximately the same as the strike and dipping directionof the slope. The Chingshui River passes through the toe of dip slope and cuts into the toe of slope, which causes the bedding planes to daylight. Two sets of orthogonal open joint (N25◦E∼ N40◦E/SW, N48◦W∼ N88◦W/N) are observed on the remaining dip slope. Surface water can thusinfiltrate into deeper formation and soften the rock. Observing the profiles of previous landslidehistory (Figures 7, 9, and 10), the dip-slope sliding could have occurred in the Cholan formation,

at the interface of Cholan formation and Chinshui shale, and in the Chinshui shale. The Cholanformation is mainly composed of thick layer of weakly cemented sandstone which is highly porousand permeable, and with thin interlayer of siltstone and mudstone. While the Chingshui shale is aweak material with low permeability, and easily softened by submergence of water.

Material properties of both the intact rock and debris material were reported by many researchers(Hung et al., 1981, 1982, Lee, et al., 1993, Chen, 2000, Lin et al., 2000c, Yeng, 2000, SinotechCo., 2000). The laboratory testing results including unit weight and both peak and residual strength

parameters were summarized in Table 4 for Cholan formation and Chinshui shale. Noted that theresidual friction angle of the Chinshui shale (13◦) is about the same as the inclination of the bedding

280







Figure 12. The geological map of Tsaoling landslide area (modified from the geological map of Taiwan

1:50,000, Puli, the Central Geological Survey, 2000) The epicenter of the first triggered event is shown near

the top of the map.

Table 3. Ground motion characteristics of station CHY080 of Chi-Chi earthquake.

Vertical Horizontal Horizontal Horizontal

Parameter \Direction (Z) (N-S) (E-W) (Dip direction)

Peak ground acceleration (gal) 715.8 841.5 792.4 646.4

Bracketed duration, sec (as Ac >±50 gal) 12.94 33.72 41.95 41.95

Arias intensity (m/sec) 0.020 0.072 0.096 0.123

RMS acceleration (m/sec2) 0.934 1.127 1.172 1.323

plane ranging from about 12

◦

to 14

◦

. The testing results of Cholan formation shown in Table 4 wereobtained from specimens with higher shale contents.For both formations, material strength reduced significantly from peak values to residual, par-

ticularly when subjected to submergence. Based on field investigation before Chi-Chi earthquake,it was found that the calcium carbonate cementation of the Cholan formation has been dissolved by

percolating water, and the process led to increased permeability and reduced strength of the sand-stone (Hung, et al., 1982). The water permeated into the sandstone or entered through the open jointseasily, and perched near the contact face of the Cholan formation and Chinshui shale and softened the rocks. Figure 13 is a photograph taken on March 23rd, 1996, before the Chi-Chi earthquake,

281





Table 4. Material properties of Chinshui shale and Cholan formation.

Item Chinshui shale Cholan formation Reference source

Total unit weight γ t (kN/m3) 25.8 24.5 Lee, 2001

Peak friction angle, φ p (◦) 36.8 38.5 Lee, 2001

Peak cohesion, c p (kPa) 664 980 Lee, 2001

Residual friction angle, φr (◦) 13.4 18.9 Lee, 2001

Residual cohesion, cr (kPa) 0 0 Lee, 2001

Submerged friction angle (◦) @ w = 2% 19 – Yeng, 2000

Submerged friction angle (◦) @ w = 4% 14 – Yeng, 2000

Friction angle (◦) (long-term submergence) 2.8 – Yeng, 2000

Cohesion (kPa) (long-term submergence) 0 – Yeng, 2000

Figure 13. Streaks of recrystallized calcium carbonate hanging on the bushes and the surface of rock near

the toe of the cliff, photo taken before Chi-Chi earthquake. Note the water seeping out from the lower corner

of the cliff. (Photo by J.J. Hung).

in which it can be found that the calcium carbonate in the Cholan formation was dissolved and recrystallized on the surface of rock and hanging on bushes and cliff wall. The Cholan formation isa shallow marginal sea deposition formation, with seashell fossils, which also provided the sourcesof calcium carbonate. The water seeped down-slope along the interface of the two formations and out from the lower right corner of the cliff in Figure 13. Figure 14 is a photograph taken on 17th

November 2002 showing water seeping down slope along the interface of the two formations. Theweathering process of the rock material proceeds rather rapidly as observed in the field. Based onthe slake durability test conducted by Lee, et al. (1993) on Cholan formation material, the durabilityindex of the first cycle was 46.2% and that of second cycle was 4.8%, and the rock was classified as very low durability according to Gamble (1971). It indicated that the weathering process would

precede rapidly accompanying material strength degradation.

6 CAUSES OF THE LANDSLIDE

The weather condition before the Chi-Chi earthquake was fairly dry without much precipitationin the nearby area of Tsaoling landslide. Location of three rainfall gauge stations installed by theCentral Weather Bureau near the Tsaoling landslide, the Sun Moon Lake station, the A-Li-Shan

282





Figure 14. The photograph taken on 17th November 2002 showing water seeping down slope along the

interface of the two formations. (Photo by T.C.Chen).

Figure 15. The locations of the three weather stations nearby the Tsaoling landslide.

station, and the Chia-Yi station are shown in Figure 15. The precipitation records of September,1999 of the three stations are shown in Figures 16. Observing the rainfall records of the Sun MoonLake and Chia-Yi stations, there were no rainfall since September 18th, and no rainfall at the A-Li-Shan station since September 19th before the earthquake. Since the A-Li-Shan station is located inthe high mountain area, and typically with much higher precipitation due to the topography, it is

283






Figure 16. The precipitation records of September, 1999 of the Sun Moon Lake, Chia-Yi and A-Li-Shan rain

gauge station, the Central Weather Bureau.

suggested that the prior precipitation is unlikely to have significant effects of ground water pressureon the Tsaoling landslide during the earthquake.

However, the sandstone with thin shale layer of Cholan formation is weakly cemented and highly permeable. With the cracks and open joints in the brittle sandstone, water can easily infiltrate intothe sandstone and pond on the shale and Chinshui formation, which will lead to weathering and weakening of the shale. Observing the previous landslide history listed in Table 2, it was found that three out of four events were triggered by earthquakes and only one event was triggered byheavy rainfall. It is likely that the geomorphology and precipitation conditions provide the necessary

condition of dip-slope landslide; however, the slope would remain stable due to the mild slope angle.The landslide occurs majorly due to the excessive ground motion introduced by the earthquake.

For the Chi-Chi earthquake, Shin (1999) reported that three triggered events with magnitudegreater than 6 were identified. The ground motion records and approximate locations of the epi-centers are shown in Figure 17. The epicenter of the first triggered event was located approximatelyat 120.66 E 23.61 N just to the north of the Tsaoling (details shown in Figure 12) and occurred on01:47:36 with two distinctive sets of P- and S- waves identified (Shin, 1999). The station CHY080was located at about 32 km south from the epicenter of the Chi-Chi main shock, and the second set of wave arrivals were recorded at about 20 seconds after the first set of wave arrivals withmuch higher amplitudes of ground motion. Both the peak horizontal ground acceleration and the

peak vertical ground acceleration occurred after the arrival of the second set of motion waves. In

addition, the peak vertical ground acceleration is very high, which is a typical characteristic for locations near epicenter.

In order to study the effects of earthquake ground motions in causing slope failure, and to assessthe importance of the degradation of material properties on the stability of the slope, a pseudo-static analysis using Janbu’s method (1968) was performed (Chen et al. 2004). The slope profileand failure surface were reconstructed based on the digital terrain model data of 40 m by 40 m

before and after the earthquake as shown in Figure 18. The ground motion used was obtained byresolving the ground motion records from strong motion station CHY080 into the dip direction of the slope as shown in Figure 11. The input ground acceleration was defined as the peak ground acceleration times a reduction factor, R f , to provide a representative ground motion. The reduction

factors of: 0, 1/6, 1/3, 1/2, 2/3, and 5/6 for both vertical and horizontal ground motion wereselected for the analysis. The peak strength parameters of the Cholan formation listed in Table 4were used, and then gradually reduced to the residual strength parameters. Results of the analysisare shown in Figure 19. It was found that the slope is unlikely to fail with the peak strength and novertical acceleration. The factor of safety against the landslide to occur decreased significantly withincreasing vertical ground acceleration and decreasing shearing strength. Thus the landslide could

be triggered with shear strength smaller than peak strength and acceleration significantly smaller than the peak ground acceleration when the vertical ground motion was considered. Observing theground acceleration data presented in Figure 11 and Table 3 of the Chi-chi earthquake, the vertical

284





Figure 17. The isoseismal map and the acceleration records of main shock and the first triggered event

(redrawn from Shin, 1999).

300

500

700

900

1100

1300

0 500 1000 1500 2000 2500 3000 3500 4000

Distance (m)

E l e v a t i o n ( m )

slip surface of 1999Cho-Lan formation

Chin-Shui shale formation

Ta-Wuo sandstone member

CHY080

Figure 18. The slope profile used for back analysis of the Tsaoling landslide triggered by Chi-Chi earthquake

by Chen et al. (2004).

peak ground acceleration is about the same magnitude as the horizontal peak ground acceleration,and both peak ground accelerations are quite large. Such large ground motion was resulted from thefirst triggered event as illustrated in the lower right corner of Figure 17. Thus it is suggested that

the dip-slope landslide of Tsaoling in the Chi-Chi earthquake was likely triggered by the firsttriggered event nearby and the vertical ground acceleration played an important role in triggeringthe landslide.

7 CONCLUSIONS

On September 21, 1999, the Chi-Chi earthquake with the local magnitude of 7.3 struck centralTaiwan and caused extensive slope failures. A large scale dip-slope landslide known as theTsaoling

285





Figure 19. Contours of safety factor for various strength degradation and ground acceleration reduction

factors. (Chen, et al. 2004).

landslide occurred, which caused severe casualties and losses of properties. In this paper, the casehistory, geological and geomorphologic properties, and possible causes of the Tsaoling landslideare presented. The ground motions caused by the earthquake are also discussed. Observing the

previous landslide history, the landslides were most likely triggered by the excessive ground motioncaused by earthquakes despite the rather mild slope angle compared to the weakened strengthof the sandstone/shale formation. In all three landslide cases triggered by the earthquakes, thelandslide bodies slid passing the river channel and landed the other side of the Ching-Shui River to form the debris dam. For the case of the 1999 Chi-Chi earthquake, not much rainfall prior to theearthquake was recorded, implying minimal influence of the ground water pressure. The high peak vertical ground acceleration as well as peak horizontal ground acceleration recorded were likely

induced by the first triggered event just after the main shock of Chi-Chi earthquake. Based onthe back analysis performed by Chen et al. (2004), the landslide could be triggered with strengthsmaller than peak strength and acceleration significantly smaller than the peak ground accelerationswhen the vertical ground motion was considered. Thus it is suggested that the dip-slope landslide of Tsaoling in the Chi-Chi earthquake was likely triggered by the larger ground motion caused by thefirst triggered event, and the vertical ground acceleration played an important role in triggering thelandslide.

REFERENCES

Central Geological Survey, 1999. Geological Map of Taiwan 1:50,000.Central Weather Bureau, 2009. the 1999 921 Chi-chi earthquake report, http://scman.cwb.gov.tw/eqv5/special/

special-index.htm.

Chang, L.S., 1951. Topographic features and geology in the vicinity of the nature reservoir near Tsao-Ling.

(in Chinese), Taiwan Reconstruction Monthly, vol. 1(6), 22–27.

Chang, S.Y. and Lee, C.T., 1989. The geology and landslides in Tsao-Ling area (in Chinese). Sinotech

Engineering , vol. 19, pp. 27–43.

Chen, H.Y., 2000. Engineering geological characteristics of Taiwan landslides. Sino-Geotechnics, vol. 79,

pp. 59–70.

286


http://scman.cwb.gov.tw/









Chen, T.C., Lin, M.L., and Hung, J.J., 2004, “Pseudostatic Analysis of Tsao-Ling Rockslide Caused by Chi-Chi

Earthquake”, Engineering Geology, Vol. 71, pp. 31–47.

Gamble, J.C., 1971. Durability-Plasticity classification of shale and other argillaceous rock. Ph.D thesis,

University of Illinois.

Hsu, S.T., 1951. The Tsao-Ling dammed up lake (in Chinese). National Taiwan University Civil Engineering,

vol. 1, pp. 3–4.

Hsu, T.L. and Leung, H.P., 1977. Mass movement in the Tsao-Ling area, Yunlin-Hsien, Taiwan (in Chinese). Proc. of the Geological Society of China, vol. 20, pp. 114–118.

Huang, C.S., HO, H.C., and Liu, H.C., 1983. The geology and landslide of Tsao-Ling area, Yun-Lin, Taiwan

(in Chinese). Bulletin of the Central Geological Survey, vol. 2, pp. 95–112.

Hung, J.J., 1980. A study on Tsao-Ling rockslides, Taiwan (in Chinese). Journal of Engineering Environment ,

vol. 1, pp. 29–39.

Hung, J.J. and Lin, M.L., 1981. Preliminary study on Tsao-Ling landslide: Material testing and stability

analysis (in Chinese). Report for Council of Agriculture Development.

Hung, J.J., Lin, M.L., Liu, M.L., Lee, C. D., Jan, K. H, and Hsieh, P. C., 1982. Engineering geological

investigation on Tsao-Ling landslide: Rock material testing report (in Chinese). Council of Agriculture

Development.

Hung, J.J., Lin, M.L., and Lee, C.T., 1994. A stability analysis of the Tsao-Ling landslide area (in Chinese).‘94 Rock Engineering Symposium in Taiwan, Chung-Li, December, 15th–16th, pp. 459–467.

Hung, J.J., 1999. Historical photographs of Tsao-Ling rockslides (in Chinese). Sino-Geotechnics, vol. 76,

pp. 113–124.

Hung, J.J., Lee, C.T., Lin, M.L., Lin M.L., Jeng, F.S., and Chen, C.H., 2000. A flying mountain and dam-up

lake (Tsao-Ling rockslides) (in Chinese). Sino-Geotechnics, vol. 77, pp. 5–18.

Janbu, N., 1968. Slope stability computation. Soil Mechanics and Foundation Engineering Report, The

Technical University of Norway, Trondheim, Norway.

Lee, C.N., 2001. Preliminary study on theTsao-Ling landslide area under earthquake. Master’s Thesis, Institute

of Civil Engineering, National Taiwan University.

Lee C.T., Hung, J.J., Lin, M.L., and Tsai, L. L.Y., 1993. Engineering geology investigations and stability

assessments on Tsao-Ling landslide area, (in Chinese). A Special Report Prepared for Sinotech Engineering

Consultants, pp. 224.

Lee C.T., Lin, M.L., Wu, L.H., and Cheng, J.S., 1994. Geological investigations and the determination of

sliding planes of rockslide events in the Tsao-Ling landslide area (in Chinese). ‘ Proceedings of 94 Rock

Engineering Symposium in Taiwan, Chung-Li, December, 15th–16th, pp. 459–467.

Lin, M.L., Chen, T.C., and Wang, K.L., 2000a, The characteristic and preliminary study of landslides induced

by Chi-Chi earthquake, NCREE

Lin, M.L., Liao, H.J., and Ueng, Z.S., 2000b. The geotechnical hazard caused by Chi-Chi earthquake.

Proceedings of International Workshop on the September 21, 1999 Chi-Chi Earthquake, Taichung, Taiwan,

June 30th, pp. 113–123.

Lin, M. L., Wang, K. L., and Chen, T. C., 2000c. Characteristics of the Slope Failure Caused by Chi-Chi

Earthquake, Proceedings of International Workshop on Annual Commemoration of Chi-Chi Earthquake,

III-Geotechnical Aspect , 199–209.

National Center for Research on Earthquake Engineering, office for NAtional science and technology Program

for Hazard Mitigation, and Taiwan Geotechnical Society, 1999. Reconnaissance report of the geotechnical

hazard caused by Chi-Chi earthquake, National Research Center on Earthquake Engineering, Taiwan, 111p

Shin. T. C., 1999. Chi-Chi Earthquake – Seismology. Proceedings of International Workshop on the

September 21, 1999 Chi-Chi Earthquake, Taichung, Taiwan, pp. 1–14.

Sinotech Engineering Consultants, Ltd., 2000. Assessment for the treatment of Tsao-Ling slides (I and II)

(in Chinese). Report for the Water Conservacy Agency, Ministry of Economic Affairs.

Tai-Pei Observatory, 1942. Report on Chia-Yi Earthquake on 17th December 1941 (in Japanese). Taiwan

Governors Office, 227p.

Wu, Y.S., Lai, J.S., and Lin, C.H., 2001, Summary of the emergency repairing of the Tsao-Ling dam-up lake,

Quarterly Journal of the Water Resource Management , Vol. 3, No. 1, pp 18–23. (in Chinese)Yeng, K.T., 2000. The residual strength of Chin-Shui shale in relation to the slope stability of Tsao-Ling.

Master’s Thesis, Institute of Civil Engineering, National Taiwan University.

287





Liquefaction induced ground failures at Wu Feng caused by strong

ground motion during 1999 Chi-Chi earthquake

W.F. Lee National Taiwan University of Science and Technology, Taipei, Taiwan

B.L. Chu & C.C. Lin National Chung Hsin University, Taichung, Taiwan

C.H. Chen National Taiwan University, Taipei, Taiwan

ABSTRACT: During the 1999 Chi-Chi Earthquake, a site named Wu-Feng very close to thefault rupture in central Taiwan has suffered serious damages caused by soil liquefaction induced ground failures. Detailed case study of Wu-Feng is reported in this paper in an effort to investigate

both failure causes and damage types. In this paper, the authors first present soil condition and characterization of the near fault strong ground motion of Wu-Feng. Secondly, both reconnaissanceand surveyed results of liquefaction induced ground failures are presented. Lastly, conclusions onfactors that possibly contribute to such extensive ground failures and future research suggestionsare provided. Progress of presented study is hopefully to improve our understanding on liquefaction

induced ground failures caused by strong ground motions.

1 INTRODUCTION

On 21 September 1999, a disastrous earthquake with magnitude 7.3 (Mw = 7.6) hit Taiwan and caused devastating casualties and infrastructure damages (Figure 1). This earthquake, named after the first located epicenter, is recognized as the 1999 Chi-Chi earthquake. Because of its shallow

Figure 1. Building and infrastructure damages in Chi-Chi earthquake.

289





Figure 2. Location of Wu Feng.

focal depth (8 km) and near 80 km surface fault rupture (the Che-Lung-Pu fault); the Chi Chiearthquake possessed very unique strong ground motion characteristics. Structure performance

and geotechnical hazards such as soil liquefaction under such strong ground motion are of greatinterests in geotechnical earthquake engineering research.

During the Chi-Chi earthquake, a site named Wu-Feng has suffered serious ground failures.Wu-Feng is a small town located in central Taiwan, Taichung County (Figure 2). It was within30 km range from the epienter of the Chi-Chi earthquake. Surface fault rupture of Che-Lung-Pu fault has just stroke along the foothill line which is the geological boundary separating thealluvium plain and hills of Wu-Feng Township. Because of the close distance from the epicenter and exposed surface fault rupture, strong ground motion ranked as level II earthquake occurred during the main shock of Chi-Chi earthquake at Wu-Feng. Recorded peak ground acceleration atWu-Feng was as high as 774 gal in east-west direction and duration was as long as 44 sec. Suchstrong ground motion has caused serious ground failures and facilities damage in Wu-Feng area.

In additional to the strong ground motion characteristics, geological condition of Wu-Feng also isalso very distinctive comparing to other sites that have been studied before for earthquake-caused ground failures. Hillside area of Wu-Feng is mainly thick laterite (gravel with silty infill) formationlying above the sandstone and shale interlayers. This hillside area mainly suffered from destructiveground deformation caused by the fault rupture. Instead of thick sand deposit like adjacent plainsites in central Taiwan, formation of plain area of Wu-Feng is laterite deposit interlayered withhigh f ine content silty sand deposit. Surprisingly liquefaction induced ground failures includingsubsidence and lateral spreading have occurred extensively over most plain area of Wu-Feng duringChi-Chi earthquake despite the laterite layers and high fine content of the silty sand deposit.

Both ground motion characteristics and unique geological condition have made Wu-Feng a site

of great research interests. However, the damage was not well studied due to serious fatalities thatneeded special attentions as well as limited research resources when the f irst phase reconnaissancework was conducted. Detailed survey and analysis was not done until 2005 via photos and availabledocumentations by the authors (Lin, 2006). In this paper, the authors will first present soil conditionand characterization of the near fault strong ground motion of Wu-Feng during the Chi-Chi earth-quake. Secondly, both reconnaissance and surveyed results of liquefaction induced ground failuresare presented. Lastly, conclusions on factors that possibly contribute to such extensive ground failures and suggested future research are provided. Results of this study are to hopefully improveour understanding on liquefaction induced ground failures caused by strong ground motions.

290





Figure 3. Geology map of Wu Feng.

2 SITE INFORMATION

2.1 Geography features

Figure 3 shows the geology map of Wu-Feng County. As shown in the figure, geography of Wu-Fengcan be divided into two parts: the east hillside and the west alluvium fans. Hillside area of Wu-Fengis mainly thick laterite (gravel with silty infill) formation lying above the sandstone, shale, and mudstone interlayer. The near surface shale and mudstone are generally highly weathered, and have

been eroded by creeks and gullies in east-west direction. Elevation of the hillside ranges from about

160 m above sea level at hilltop dropping to about 70 m above sea level within 1000 m distance. Thewest alluvium fans are fairly flat within less than 10 m difference in elevation (60∼50 m above sealevel). The alluvium plains of Wu-Feng are generally formed by two major river systems in centralTaiwan, Da-Gia River and Wu-Si River. Instead of thick sand deposits that most west coast plainsof Taiwan have, formation of plain area of Wu-Feng is laterite deposit interlayered with high finecontent sandy silt (ML) or silty sand (SM) deposit. The town center and most residential housingare located in the plain area. Agricultural area of Wu-Feng is also located in the alluvium plainwhere the old river channel and flooding plains used to be. During the 1999 Chi-Chi earthquake,the hillside area mainly suffered from destructive ground deformation caused by the fault rupture(Figure 4). However, in plain area of Wu-Feng, despite the laterite layers and high fine content of the

silty sand deposits, extensive liquefaction induced ground failures including subsidence and lateralspreading occurred during the Chi-Chi earthquake. In this paper, investigation and discussion arefocused on the plain area for major liquefaction induced ground failures.

2.2 Soil condition

In this study, total of 40 boreholes were collected to identify the soil conditions of Wu-Feng area.Table 1 summarizes 25 boreholes with depth of 20 m or more that were used to study liquefaction potential (NCREE, 2000; Chang, 2001, Lin, 2006). Figure 5 shows locations of most boreholes

291





Figure 4. Fault rupture at Wu-Feng junior high school.

Table 1. Basic information of collected boreholes.

Coordinates

No. N E Elevation m Depth M GWT depth m

BH-3 2662550 217761 46.48 30 −1.3

BH-6 2662767 218229 49.39 30 −3.2

BH-7 2661861 218608 56.6 30 −3.2

BH-8 2661560 218502 58.7 50 −0.5

BH-10 2661221 218349 56.88 30 −1.4

BH-11 2662203 218331 53.16 30 −0.9

BH-12 2660838 218153 58.97 30 −4.2BH-13 2662392 218519 55.23 30 −0.9

BH-14 2664115 215570 35.16 23.6 −3.5

C-8 2661574 218500 58.64 28.35 −0.5

C-15 2661485 218296 56.37 19.4 −2.5

B-1 2661188 218369 57.00 25 −2.7

B-2 2661150 218412 57.00 25 −2.3

B-3 2661147 218325 57.00 25 −2.75

B-4 2661106 218395 57.00 25 −2.55

B-5 2661109 218326 57.00 25 −2.7

B-6 2661091 218374 57.00 25 −1.9

B-7 2662169 218226 51.41 30.45 −1.9B-8 2661491 218261 55.62 20.45 −3.2

B-9 2662089 217875 50.07 20.45 −2.28

B-10 2661283 217784 53.81 33.5 −0.54

D-1 2662572 218272 51.00 30.45 −2.7

D-2 2662631 218284 52.00 30.45 −3.1

D-3 2662545 218343 52.00 30.45 −2.8

D-4 2662597 218365 52.00 30.45 −2.8

292





Figure 5. Studied area and locations of boreholes.

within the studied area. Figures 6 to 8 show the various cross section soil profiles in east-west and north-south directions respectively. As shown in the figures, thickness of the silty sand layers isgradually increasing towards to the west of the plain area of Wu-Feng. Ground water table is fairlyclose to surface and generally less than 2 m below surface in the plain area. Average SPT-N valueof the silty sand layers within 10 m depth is smaller than 10, with maximum values no more than

293





Figure 6. East-west soil profile I of Wu-Feng.

Figure 7. East-west soil profile II of Wu-Feng.

20 to a depth of 20 m. Moreover, this silty sand deposit of Wu-Feng was found to have high finecontent, yet less plasticity. For silty sand material, the fine content ranges from 20% to 40%, and the plastic index is almost zero. For sandy silt material, the fine content ranges from 40% to 70%,and the plastic index still has a value smaller than 7. The most serious liquefaction induced ground settlement was observed with the triangular zone among BH7, BH11, BH3, and BH8 boreholeswhere thickness of the silty sand deposits is over 15 m (Figure 9). Whereas, the lateral spreadingfailures were found scattered along banks of the Ger-Niou-Gen creek where C10 borehole locates.

294






Figure 8. North-south soil profile of Wu-Feng.

Figure 9. Soil profile of triangular zone among BH7, BH11, BH3, and WBH8.

For area where BH5 and BH11 locate, distinguished ground settlement was still observed despitethat the liquefiable soil layers were confined by clay layers or gravel layers. Mechanism of suchground subsidence is of great research interests.

At the time when the Chi Chi earthquake occurred, Taiwan had installed total 389 strong ground motion (SGM) monitoring stations over the island. Borehole BH8 is at the location of SGM stationTCU065, the Wu-Feng elementary school SGM station. Figure 10 shows the shear wave velocity

295






Figure 10. Shear wave velocity profiles of soils at SGM station TCU065 (BH8) (NCREE, 2008).

profile of borehole BH8 obtained by various geophysics methods including microtremer array,SASW, suspension P-S wave logging device, downhole, and refraction methods (NCREE, 2008).As shown in the figure, shear wave velocities of soil within 15 m below ground surface, where soilis mainly silty sand, range from 150 m/s to less than 200 m/s. It reaches 500 m/s at a depth of 50 mwhere the deep seated gravel layer locates.

3 CHARACTERIZATION OF STRONG GROUND MOTION

3.1 Strong ground motion recorded during Chi-Chi earthquake

During the Chi-Chi earthquake, surface fault rupture of Che-Lung-Pu fault has just stroke along thegeological boundary which separates the alluvium plain and hills of Wu-FengTownship. Because of theexposed surface fault rupture andits close distance from the epicenter, peak ground accelerationsrecorded at SGM station TCU065 were as high as 774 gal in east-west direction, 563 gal in north-south direction and 258 gal in vertical direction. Earthquake duration at Wu-Feng was as long as44 sec. Such a strong ground motion could be ranked as level II earthquake accordingly.

Figure 11 summarizes peak ground accelerations of Chi Chi earthquake recorded at four SGMstations near Wu-Feng. In Figure 11, there are three SGM stations (Dali, Wu-Feng, and Nantou)

296





Figure 11. Peak ground accelerations of Chi Chi earthquake recorded at four SGM stations near Wu-Feng.

along the Che-Lung-Pu fault rupture. Geological condition of Dali is mainly laterite gravel depositand that of Nantou is 10 m to 15 m silty sand layer over laterite gravel deposit. Figures 12 to 14show the acceleration histories of these three SGM stations near the Che-Lung-Pu fault. As shownin the figures, Wu-Feng has the highest PGA among three SGM stations along the Che-Lung-PuFault. Amplifying effects of ground motions, including larger acceleration amplitudes, and longer

bracket duration are clearly found in acceleration histories of Wu-Feng in all three directions. Morecomplicated frequency characteristics and multi-peak accelerations were also observed at Wu-Fengdespite that peak accelerations of Dali and Nantou were found at about 10 sec after the earthquake

started. Large ground accelerations which exceeded 300 gal were widely spread from 7 sec toalmost 30 sec after the earthquake started at Wu-Feng. These unique ground motion characteristicswere concluded as one of the major factors for so many liquefaction induced damages occurred atWu-Feng during the Chi-Chi earthquake.

For other two SGM stations shown in Figure 11, Puli is the closest station near the epicenter and soil conditionthere is mainly gravel laterite formation.Yuanlinis the site with thick sand deposit over 60 m in depth andfarther from the epicenter. Both Puli andYuanlinare away from Che-Lung-Pu faultrupture. During the Chi-Chi earthquake, Puli was suffered intense strong ground motion and had large numbers of buildings collapsed. A lot of soil liquefaction phenomena were found in Yuanlin,yet the building or facility damage level of Yuanlin was not as serious as those in Puli and Wu-

Feng. Figures 15 to 17 show the acceleration histories of Puli, Wu-Feng, and Yuanlin in east-west,north-south, and vertical direction accordingly. As shown in the figures, Puli suffered intense strongground shaking in 15 sec to 25 sec after the Chi-Chi earthquake started; yet much smaller ground accelerations were found in Yuanlin. Despite the low amplitude ground motions, soil condition and long duration might be the major factors to many soil liquefactions observed in Yuanlin.

The strong ground motion of Wu-Feng recorded during Chi-Chi earthquake has characteris-tics of higher peak ground acceleration, longer duration, and wider distribution of large ground accelerations than those of other stations nearby. These unique features had provided a “perfect”seismic environment for soil liquefaction to happen.

297





Figure 12. Acceleration histories of Dali,(TCU067) Wu-Feng(TCU065), and Nantou(TCU076) in East-West

direction.

3.2 Predicted ground accelerations

In order to access the ground motion distribution over the studied area, ground motion recorded at TCU065 were also used to back-calculate the bedrock motion and to predict ground motions atselected borehole locations. This analysis was done by using ProShake program. In this preliminaryanalysis, soil layer within analyzed boreholes are assumed to be horizontal. Moreover, the deepseated gravel layer with shear wave velocity over 500 m/s and depth over 50 m was assumed as the

298





Figure 13. Acceleration histories of Dali(TCU067), Wu-Feng(TCU065), and Nantou(TCU076) in

North-South direction.

bedrock in ProShake program (Lin, 2006). Figure 18 shows the result of interpretation in a form of peak ground acceleration contour. As depicted in the figure, locations with thick silty sand depositall present very high peak ground acceleration values. In summary, the unique soil condition and strong ground motion characteristics together created perfect panorama for soil liquefactions and thus caused serious damages in Wu-Feng during the Chi-Chi earthquake.

299





Figure 14. Acceleration histories of Dali(TCU067), Wu-Feng(TCU065), and Nantou(TCU076) in vertical

direction.

4 LIQUEFACTION INDUCED GROUND FAILURES

4.1 Reconnaissance results

Because of the unique strong ground motion characteristics and soil condition described in previous paragraphs, alluvium plain area of Wu-Feng has sufferedextensive ground failures including ground

300





Figure 15. Acceleration histories of Puli(TCU074), Wu-Feng(TCU065), and Yuanlin(TCU110) in East-West

direction.

subsidence, lateral spreading, and sand boils. Structures damages including building collapses,foundation settlement, and ground surface cracks that caused numbers of casualties also occurred during the Chi-Chi earthquake (Chang, 2001).

The soil liquefaction areas of Wu-Feng could be divided into four zones according to geologicalconditions and types of ground or structure failures. Zone 1 is from the area of BH13 down south toBH7 as shown in Figure 5. Soil condition of Zone 1 is largely thick silty sand deposit with scattered

301





Figure 16. Acceleration histories of Puli(TCU074), Wu-Feng(TCU065), and Yuanlin(TCU110) in North-South direction.

clay layers. This area was the commercial district of Wu-Feng. Extended ground subsidence caused by soil liquefaction was found in this Zone, and has caused serious building damages (Figure 19).South from Zone 1, Zone 2 is within the locations of C10, B1, B2, and B3 where Ger-Niou-GenCreek meets the Gan River. Soil condition of Zone 2 is the typical alluvium delta soil deposit whichcontains thick silty sand deposit with clay lenses. Serious lateral spreadings and building foundation

302





Figure 17. Acceleration histories of Puli(TCU074), Wu-Feng(TCU065), and Yuanlin(TCU110) in verticaldirection.

failures were found during the Chi-Chi earthquake in Zone2 (Figure 20). Zone 3 is parallel to Zone1 along the west edge of residential area of Wu-Feng from BH11 down to WBH8 and BH12. Sand

boils, ground subsidence, and ground surface cracks were widely over this area (Figure 21). Thiszone has similar soil condition as Zone 1. Luckily, minor structure damages occurred in this area

because of less housing and population densities. Zone 4 is the alluvium plain area where most

303





Figure 18. Calculated peak ground acceleration distribution during the Chi-Chi earthquake of the studied

area.

agricultural fields located. During the Chi-Chi earthquake, sand boils spread over the rice fieldsand fruit farms in this area. Soil condition of Zone 4 is mainly thick silty sand deposit with gravelinterlayers. There was only few building damage and ground subsidence found near the area of BH3 (Figure 22).

304





Figure 19. Ground failures and building damages at Zone 1.

Figure 20. Lateral spreading and foundation failures at Zone 2.

4.2 Survey results

Soil liquefaction induced ground failures at Wu-Feng was not well studied due to serious fatalitiesand limited research resources when the first phase reconnaissance work was conducted soon

305






Figure 21. Sand boils and ground surface cracking at Zone 3.

Figure 22. Sand boils and building foundation failures at Zone 4.

after the Chi-Chi earthquake. In an effort to further characterize damages of this unique Level IIearthquake site, the authors conducted a Geological Information System (GIS) based survey via

photos and available documentations in 2005 (Lin, 2006). Soil liquefaction induced ground failuressuch as sand boils, ground subsidence, and lateral spreading were documented from photos and

306






Figure 23. Mapping result of ground failures (sand boils, subsidences, and lateral spreadings).

existing reconnaissance reports. Such information was mapped onto the GIS maps through GlobalPositioning System (GPS) survey. Figure 23 shows the result of ground failure mapping. Similar GPS mapping was also applied to the structural damages. As shown in Figure 24, structural damagesinformation including building collapses, foundation settlement, and ground surface cracks wasalso collected and presented in the GIS map.

307





Figure 24. Mapping of Structural damages.

In addition to document the locations of ground failures and structural damages, the authorsalso conducted a settlement survey via available photos and GPS survey. Standard GPS reference

points within the Wu-Feng district were analyzed to gather calibrated settlement after the Chi-Chiearthquake. Local settlement was then extrapolated via sequence photos from the reference pointsto locations of interests including inspected boreholes and ground failure sites. Measurable objectsin photos such vehicles, manholes, and existing buildings were used as intermediate references

308





Figure 25. Predicted settlement contour.

or scales for such a post-event survey. This process was first conducted to locations where actualmeasurements were taken after the Chi-Chi earthquake for verification. Total 30 survey points wereanalyzed in order to cover the studied area in detail. Settlement contour was then generated usingGIS software as shown in Figure 25. The predicted settlement contour provides an overall pictureof the ground subsidence caused by liquefaction of Wu-Feng during the Chi-Chi earthquake. Asshown in the figure, delta areas where rivers intersect in alluvium plain had the most serious ground

309





subsidence ranged from 30 cm to 50 cm. Reasons for this particular occurrence are probably thethick silty sand deposit and high ground water table controlled by adjacent rivers. Large surfacedeformation were also found in locations along banks of Li-Yuan Creek and Ger-Niou-Gen Creek where lateral spreading were most likely to happen.

4.3 Factors of observationExcellent consistency of peak ground acceleration (PGA) distribution and settlement contour wasfound by comparing Figures 18 and 25. Survey results of ground failures and structural damagesshown in Figures 23, and 24 also have shown reasonable agreement to the calculated PGA and

predicted settlement contours. Factors observed from these f igures validate that chain reaction of strong ground motions and soil liquefactions is probably the major cause for extensive ground failures and serious building damages occurred in Wu-Feng during Chi-Chi earthquake.

5 CONCLUSIONS AND FUTURE RESEARCHES

In this paper, the authors present the case study result of soil liquefaction induced damages of Wu-Feng during Chi-Chi earthquake in detail. The serious damages caused by soil liquefactionin Wu-Feng are concluded as a consequence of near fault strong ground motion and the uniquegeological condition. Feature characteristics of strong ground motion of Wu-Feng during the Chi-Chi earthquake include long bracket duration, wide distribution of high ground accelerations, and the extreme peak ground acceleration. Soil condition at alluvium area of Wu-Feng where most soilliquefaction damages occurred is mainly thick silty sand deposit over deep seated gravel deposit.The silty sand was found to contain high fine content, yet with low to zero plasticity. For somelocations, the silty sand deposit is interlayered with clay layers or gravel layers. During the Chi-Chiearthquake, this unique soil deposit liquefied despite its high fine content and confinement of

clay or gravel layers. Moreover, the authors summarized results of post-event surveys includingmapping of ground failures, mapping of structural damages, and settlement prediction via photosin this paper. Relationship between liquefaction factors (strong ground motion and soil condition)and types of ground failures was preliminarily accessed through these survey results. Importantinformation obtained from this study including strong ground motion characteristics and types of ground failures are invaluable for improving our understanding in design and performance analysisof geotechnical structures under strong earthquakes.

Moreover, liquefaction induced ground failures at Wu-Feng also raised various research interestsin related engineering problems. Dynamic properties of the unique silty sand is of great researchinterests to gain further understanding in influence of non-plastic fine contents on dynamic behav-iors of granular material. Volume change behavior of the silty sand under seismic condition is also of interests to improve seismic performance analysis and failure prediction of geotechnical structures.Mechanism and behavior of soil liquefaction occurred with confinement of clay or gravel layersare also of great research interests. Finally, investigation and documentation procedures developed in this study provide valuable references for future reconnaissance works of similar events.

REFERENCES

Chang, Y. M., “Characterization of Soil Liquefaction at Wu-Feng Area,” Master Thesis, National Chung Hsing

University, Taichung, Taiwan. (2001)

Lin, C. C., “A Study of Soil Liquefaction-Induced Ground Settlement in Wu-Feng,” Master Thesis, National

Chung Hsing University, Taichung, Taiwan. (2006)

NCREE, “Geotechnical Reconnaissance Report of the Chi-Chi Earthquake.” (2001) NCREE, data support.

(2008)

310





Failures of soil structures during the 1999 Taiwan

Chi-Chi earthquake

C.C. Huang Department of Civil Engineering, National Cheng Kung University, Taiwan

ABSTRACT: Post-earthquake investigations were performed for three types of geotechnicalstructures, including geosynthetic-reinforced soil block walls, concrete-covered earth-core lev-ees, and soil retaining walls. Comparative studies were also performed based on the result of soil

borings, soil testing, and pseudo-static analyses to highlight the mechanism controlling differentseismic-resisting behavior. It was found that seriously damaged soil structures in the 1999 Chi-Chiearthquake were always intrinsic with deficiencies resulted from bad design and/or construction

practices. Two geosynthetic-reinforced modular block walls with excessive displacements or col-lapsed facing columns were found to have a vertical spacing of reinforcement equal to 0.8 m and a relatively loose backfill with an internal friction angle of about 30◦. A synergistic effect of largevertical spacing of reinforcement and low backfill strength accounted for the excessive lateralthrust and deformation of the wall. A seriously damaged earth levee with a plain concrete cover exemplifies another bad practice of placing a brittle structure on a deformable (or liquefiable)foundation. Some leaning-type soil retaining walls built on hillsides suffered great settlement and lateral movement during the earthquake. It was found that these damaged leaning-type retaining

walls had small base width to wall height ratios, resulting in low safety factors against failure evenunder static conditions.

1 INTRODUCTION

Figure 1 shows the location of the Che-Lung-Pu fault triggering the 1999 Chi-Chi earthquake(ML = 7.3, M L: magnitude on the Richter scale.) The collapsed fault had a total length of about80 km in approximately North-Sourth direction. The major earthquake occurred at 1:47 AM of September 21, 1999. The epicenter located at 120.82◦ E and 23.85◦ N, with a focus depth of 8 km.This earthquake claimed 2415 lives, 11305 wounded, and destroyed more than 100,000 buildings.

Ground accelerations records obtained near the sites investigated here, namely, TCU 052, TCU 078,TCU 076, and CHY 080, are shown in Figures 2, 3, 5, and 6, respectively. Overviews regardinggeological, geotechnical and structural engineering aspects of the Chi-Chi earthquake can be found in the Earthquake Disaster Mitigation (EDM) Technical Report No. 7 (2000).

2 DAMAGE OF GEOSYNTHETIC-REINFORCED MODULAR BLOCK WALLS

Figures 6(a)–6(d) show overviews of four geosynthetic-reinforced modular block walls (RMBWs)in the post-earthquake investigation of the 1999 Chi-Chi earthquake for Sites 1-4 shown in Figure

1. The modular block walls at Sites 1 and 2 suffered excessive bulging and/or collapsing of facing blocks. Huang (2000) and Huang et al. (2003) pointed out that the excessive bulging is associated with large settlement (or downward drag force) of the backfill adjacent to the facing and thesubsequent pull-out of the geogrid from the facing blocks. Figures 7(a) –7(d) show typical crosssections of the walls at Sites 1-4, respectively. Among them, Sites 1 and 2 were subjected to facingcollapse; Site 3 was subjected to a light damage of facing block displacements; Sites 4 showed nosign of damage.

Figures 7(a)–7(d) Show that vertical spacings of reinforcement, Sv = 0.8 m were used in the wallsat Sites 1 and 2; Sv = 0.6 m were for Sites 3 and 4. In addition, a high value of φ= 48◦ was obtained

311




Figure 1. Locations of Che-lung-pu fault, investigated Sites and seismographs.

Figure 2. Recorded horizontal ground acceleration

at station TCU 052 (near to Sites 1, 2, and 3).

Figure 3. Recorded horizontal ground acceleration

station TCU 078 (near to Site 4).

based on the result of direct shear tests on undisturbed samples from Site 4. Pseudo-static analyses

on these modular block walls based on a multi-wedge failure mechanism as shown in Figure 8suggested that the contribution of facing to the seismic stability of the wall is significant, in termsof the “critical horizontal seismic coefficient, k hc

” (see Figure 9) which represents the critical stateof full mobilization of soil-soil and block-block frictional strength along the failure line.

Figure 9 shows the Fs vs. k h relationships for the RMBW at Site 2 using an internal friction angleof the backfill, φs = 30.4◦, a cohesion of the backfill, cs = 0, a pull-out strength of geogrid from the

block-block interface (φ b−r = 21◦, c b−r = 7.8 kN/m), a ratio between vertical and horizontal seis-mic coefficients, λ= k v/k h = 0.2 and an inter-wedge friction angle of φBF = 0◦. Type-1 simulatesa conventional two-wedge analysis without considering the contribution of facing. Types 2 and 3

312







Figure 4. Recorded ground acceleration at stationTCU 076 (near to Site 5).

Figure 5. Recorded ground acceleration at stationCHY080 (near to Site 6).

Figure 6(a). A view of the collapsed GRS-MB wall

at Site 1. (after Huang et al., 2003).

Figure 6(b). A view of the collapsed GRS-MB wall


Figure 6(c). A view of the lightly damaged GRS-MB wall at Site 3. (after Huang et al., 2003).

Figure 6(d). A view of the undamaged GRS-MBwall at Site 4. (after Huang et al., 2003).

analyses take into account the stability function of facing. In Type-2 analysis, the shear strength of fiber reinforced plastic (FRP) rods used to align modular blocks was not considered. While it wastaken into account in Type-3 analysis. Based on the result of site investigation, the bulging mode of facing may inhibit the development of shear stress in these rods. Therefore, it’s more reasonable not

313










Figure 7(a). Cross section of the failed RMBW at Site 1. (after Huang et al., 2003).

Figure 7(b). A severely deformed section of the RMBW adjacent to the failed one at Site 1. (after Huang

et al., 2003).

to take into account the contribution of shear strength of these FRP rods in the stability analysis.Type-2 analysis is considered representative of the facing condition in the field. A Newmark-typeseismic displacement analysis as schematically shown in Figure 10(a) was performed and the resultshowed a good agreement between the observed and the calculated horizontal displacements of

314






Figure 7(c). A cross section of lightly damaged (or intact) RMBW at Site 3. (after Huang et al., 2003).

Figure 7(d). A cross section of intact RMBW at Site 4. (after Huang et al., 2003).

315






Figure 8. Force equilibrium in the modified two-wedge method.

Figure 9. Fs vs. k h relationships for the RMBW


Figure 10(a). Schematic diagram for calculating

seismic displacement of wall based on Newmark’s

theory.

the wall (δ3h=

390 mm for transitional mode, and δh=

440 mm for transitional + buckling modes,respectively, see Huang et al. 2003). Translational displacements (δ3h) shown in Figure 10(b) arehorizontal seismic displacements along the base of facing calculated using Newmark’s sliding block theory; horizontal displacements due to the buckling mode are the horizontal displacement at thecentral height of facing induced by the settlement of the crest of the facing column. In deriving therelationship between the settlement of the crest of the backfill (δ2v) and the horizontal displacementat the central height of the facing due to the buckling mode, the deformed facing is assumed to be acircular arc. Details of the displacement calculation for the buckling mode was reported by Huanget al. (2003).

316







Figure 10(b). Displacement vs. time relationships of the GRS-MB wall at Site 2. (after Huang et al., 2003).

Figure 11. The damaged levee at Site 5

(0K + 73).

Figure 12(a). Plane view of the collapsed levee at

Site 5.

3 DAMAGE OF LEVEES

Figure 11 shows an overview of a damaged levee at Site 5 near the Che-Lung-Pu fault. A plane viewof the damaged levee is shown in Figure 12(a). Boring data and a cross section is shown in Figure12(b). Boring log profile in Figure 12(b) indicates that the heavily damaged site is associated withrelatively weak soil strata with Standard Penetration Test (SPT) N-values ranged between 8 and 15in the backfill and foundation soils. This contrasts the N-values (N≥ 12) obtained in the the lightlydamaged levee at opposite side of the river as shown in Figure 12 (c). For the levees investigated here, plain concrete of about 200 mm thick is used to cover the earth core of the levees. It appears

317







Figure 12(b). A typical cross section for the collapsed levee (0K+048) at Site 5.

Figure 12(c). A typical cross section for a lightly damaged levee at the opposite side of Site 5.

that the deformability (or ductility) of the foundation soil, the earth core, and the plain concretewere greatly varied, resulting in bad seismic performance of the levee system.

4 DAMAGE OF SOIL RETAINING WALLS ON HILLSLOPES

Figure 13 shows a collapsed soil retaining wall supporting a highway embankment on a hillslopenear the south end of the Che-Lun-Pu fault (Site 6 in Figure 1). The plane view of the failure Site isshown in Figure 14(a). It can be seen in Figure 14(a) that the failed embankment consists a convexcurve in the planar view. From the standpoint of three-dimensional (3-D) stability of the slope,the convex portion of the slope is with a lower stability comparing with the straight portion of the slope. Figure 14(b) shows the boring log profile adjacent to the failed soil retaining wall. Theembankment was backfilled with colluvium with rock fragments. Because of the rock fragmentscontained in the colluvium, N-values are influenced to some extent. It can be seen, however, a weak layer with a relatively low N-value exists, which may dominate the seismic stability of the wall.

318






Figure 13. A view of the collapsed soil retaining wall at Site 6. (after Huang and Chen, 2004).

Figure 14(a). Top view of the severely damaged highway embankment at site 6. (after Huang and Chen,

2004).

Pseudo-static analysis for the failure section shown in Figure 14(c) was performed using a multi-wedge mechanism of the backfill (Figure 15a) and a bearing capacity failure mechanism as shownin Figure 15(b). The multi-wedge failure mechanism shown in Figure 15(a) consists of active

wedges ‘B’ and ‘F’, a wall and a passive wedge ‘P’. The safety factor against lateral sliding, Fsa, isdefined as the ratio between the ultimate shear resistance and the driving force along the base of the wall. The critical seismic coefficient k ha

is defined as the value of k h that fulfills the conditionof Fsa = 1.0. A critical value of k ha

= 0.069 is obtained when taking into account the stabilizingeffect of wedge ‘P’; k ha

= 0.007 is obtained when ignoring the stabilizing effect of wedge ‘P’, assummarized in Figure 16.

A safety factor against bearing capacity failure, Fsb, is defined based on the failure mechanismshown in Figure 15(b) which consists of an active, an transitional and a passive failure wedges(wedges AZ, RA, and PZ, respectively). Critical values of k h (namely, k h b

) is defined as the value

319






Figure 14(b). Soil log profile of the failed highway embankment at Site 6 (after Huang and Chen, 2004).

Figure 14(c). A typical section of the failed highway embankment at Site 6 (after Huang and Chen, 2004).

of k h that fulfills the condition of Fsb = 1.0. Fsb is defined as the ratio between the ultimate bearingcapacity induced by the failure mechanism consiting of failure wedges AZ, RA and PZ and theexternal load applied at the base of the wall. Figure 16 shows that a critical value of k h b

= 0.053 isobtained when the effect of inertial force is taken into account in the limit equilibrium calculation.A value of k h b

= 0.053 thus calculated is greatly different from that without considering the inertiaof soil mass, as indicated in Figure 16. Results of pseudo-static analyses for Site 6 (see Figure 16)also showed that stability of the wall is low when the passive resistance in front of the wall is nottaken into account (as shown by the curve of k ha

= 0.007). It is also shown that bearing capacity

320






Figure 15(a). Failure mechanism of horizontal sliding along the base of the wall. (after Huang and Chen,

2004).

Figure 15(b). Mechanism of bearing capacity failure of the wall. (after Huang and Chen, 2004).

failure can be a dominant failure mode (as shown by the curve of k h b= 0.053) when compared with

the lateral sliding failure mode with the aid of passive resistance.Figure 17 shows the results of Newmark-type seismic displacement analyses for the wall at

Site 6 using critical values of k ha and k h b

obtained in Figure 16. In Figure 17, δ2v representsthe vertical displacement of wedge ‘F’ behind the wall, and δ3v and δ3h represent vertical and horizontal displacements for the wall, respectively. To calculate vertical settlements δ2v and δ3v,

321






Figure 16. Fs vs. k h relationships for the wall at Site 6 (after Huang and Chen, 2004).

a compatibility diagram of the displacements for all failure wedges (B, F, and the wall) is used.Details of the displacement compatibility for the failure wedges was reported by Huang (2005). Inthis analysis, the N-S ground acceleration record obtained at station CHY080 during the Chi-Chiearthquake is used. This result indicates that the observed great displacements of the wall can beexplained by using k ha

= 0.007, suggesting that the passive resistance in front of the wall may be

absent during the earthquake. Figure 18 summarizes the result of a parametric study for Site 6 onthe effects of the internal friction angle of the active zone, φa, the internal friction angle of the

passive zone, φ p, and the internal friction angle of the foundation soil, φf . In this parametric study,φa = 40◦, 42◦, and 45◦ are used to simulate internal friction angles of the failure wedges ‘B’, ‘F’,and the wall base; φ p = 0◦, 30◦ and 36◦ are used to simulate internal friction angles of wedge ‘P’;φf = 29◦, 33◦, and 36◦ are used to simulate internal friction angles for the foundation soil. Thisfigure shows that for a probable internal friction angle of the backfill, φa = 42◦, a state of zero

passive resistance can be used to simulate the great lateral displacement observed (δ3h ≥ 3.5 m).The zero passive resistance state may be a result of a bad practice which ignores the compactionof the passive zone. This figure also shows that the influence of φf to the vertical displacement of the wall is significant. Therefore, the evaluation of the strength of foundation soils is important inassessing the seismic performance of soil retaining walls placed on the hillside.

5 CONCLUSIONS

Three types of geotechnical structures, namely, geosynthetic-reinforced modular block walls, soillevees with plain concrete covers, and soil retaining walls situated on the hillslope, were investigated after a major earthquake occurred in September 21, 1999 in central Taiwan. Comparative studies

322





Figure 17. Results of displacement calculations for Site 6. (after Huang, 2005).

on the seismic behavior of these structures were performed by conducting soil boring, testing and pseudo-static analyses. Results of the comparative study show that:

(1) Bad seismic performance of the geosynthetic-reinforced modular block walls was associated with improper design and construction practices. Seriously damaged reinforced modular block

walls were all reinforced with geogrids placed with a large vertical spacing (Sv) of 0.8m,contradicting to the design guidelines which suggests Sv ≤ 0.6 m. Insufficient compaction for the backfill soils is also accounted for the serious damage during the earthquake. It is considered that relatively loose backfill generates great lateral earth pressure and downward drag force

behind the modular block wall. As a result, the modular block facing exihibited excessive buldging associated with pull-out of the geogrids from the facing blocks.

(2) Earth-filled levees covered with plain concrete shells also exhibit anthor bad example in thedesign of hydraulic (or geotechnical, in a broader sense) structures. This practice is subjected toa major modification in the future, in terms of the overall flexibility (or ductility) of the levee,especially when the levees are placed on soft foundations which are prone to liquification,

excessive settlement or lateral spreading during earthquakes.(3) A great number of leaning-type soil retaining walls were used to stabilize highway embank-ments in hillslope areas of Taiwan. These structures have small base width-to-height ratiosand low safety factors against sliding, overturning and bearing capacity failures even under static conditions (except those with matrix suction or negative pore pressure). It was found thatthe safety factor against bearing capacity failure and the vertical displacement of the wall aresuceptible to the change of internal friction angle of the foundation soil. Therefore, a smalldecrease of the strength of foundation soils may cause excessive settlement of the wall. The

passive resistance in front of the wall plays a significant role in preventing lateral sliding of

323






Performance of buildings in Adapazari during the 1999 Kocaeli,

Turkey earthquake

J.D. BrayUniversity of California, Berkeley, CA, USA

R.B. SancioGolder Associates, Inc., Houston, TX, USA

ABSTRACT: A large number of structures in Adapazari, Turkey collapsed or were heavily dam-aged due to strong ground shaking or ground softening resulting from the August 17, 1999 Kocaeliearthquake. The softening of shallow silt deposits led to relative vertical displacement of buildingsinto the ground, building tilt, lateral translation of buildings, and broken underground utilities.Sediment ejecta were observed at some locations but not all. Following the earthquake, system-atic surveys of building damage and ground failure were completed, which were followed bycomprehensive subsurface investigations. These investigations are summarized, and the differ-ent types of foundation failures observed in Adapazari are described. The building survey datashow the interdependence of structural damage with ground failure. Areas within the city withsignificant ground failure commonly had relatively greater amounts of buildings with significantstructural damage. The mechanisms that might have led to the observed building performance are

described.

1 INTRODUCTION

Adapazari, the capital of the Sakarya Province, is home to approximately 180,000 people. The citysuffered the most building damage and life loss during the August 17, 1999 Kocaeli earthquake(Bray and Stewart 2000). According to Turkish Federal Government data, 5,078 buildings (27% of

the total stock) were either severely damaged or destroyed. The official loss of life in Adapazariwas 2,627, although the actual number was probably much higher.

Ground failure under and adjacent to buildings in Adapazari was pervasive. Hundreds of buildings settled and tilted into the ground; others also translated horizontally over the ground.Additionally, many of these buildings had structural damage. Rapid building damage and ground failure surveys were performed along four lines across the city (Bray and Stewart 2000). Moredetailed surveys of representative building sites were also performed throughout the city (Brayet al. 2004). With this wealth of post-earthquake reconnaissance data available, comprehensivesubsurface investigations along the survey lines and at the selected building sites were completed (Bray et al. 2004). “Undisturbed” sampling of the fine-grained soils largely believed to be respon-

sible for the observed ground failure in Adapazari was conducted to allow over a hundred cycliclaboratory tests to be performed (Bray and Sancio 2006).The site characterization data developed through these studies is summarized and made avail-

able in the accompanying electronic files (see the appendix). Earthquake ground motions in thevicinity of Adapazari are discussed in Bray et al. (2004) and representative records are pro-vided. Building performance data, including relative movements of their foundations, are also

provided. Representative cases of building performance at sites involving ground failure are pre-sented, and the prevalent mechanisms of ground movements that affected building performance aredescribed.

325




2 BUILDING TYPES IN ADAPAZARI

Adapazari is densely developed in most areas, primarily with 3 to 6 story reinforced concrete frame buildings and with some older 1 to 2 story timber/brick buildings. Reinforced concrete constructionis primarily non-ductile, with shallow, reinforced concrete stiff mat foundations located at depthsof typically 1.5 m due to shallow groundwater. The more common 3 to 6 story reinforced concrete

buildings were constructed with a beam-column structural frame. Shear walls are uncommon.Interior and exterior walls are built with cinder blocks and covered with stucco. The roof of theconcrete buildings are inclined and covered with clay tiles. Older buildings of 1 to 2 stories thatwere built with timber and solid clay bricks are also found, but are less prevalent.

Due to the poor ground conditions in Adapazari, the foundations of the reinforced concrete buildings are very robust compared to foundation systems commonly employed for buildings of these heights. These foundations generally consist of a 30 to 40 cm thick reinforced concrete matthat is stiffened with 30 cm wide and 100 cm to 120 cm deep reinforced concrete grade beams thatare typically spaced between 4 m and 6 m in both directions. The open cells between adjacent grade

beams are filled with soil and then covered with a thin concrete floor slab. The resulting foundation

is effectively a very stiff mat foundation that is about 1.5 m thick. Tilting of structures after the earth-quake without significant structural damage is generally attributed to the exceptional robustnessof these foundations, which allows the building to respond as nearly a rigid body (if the overlyingstructural system does not fail) while it undergoes significant differential downward movement,tilt, or lateral translation. The initial fundamental periods of these buildings are estimated to bewithin the range of 0.1 to 0.4 s.

3 EARTHQUAKE SHAKING IN ADAPAZARI

The nearby Sakarya accelerograph recorded a peak horizontal (east-west) ground acceleration,

velocity, and displacement of 0.41 g, 81 cm/s, and 220 cm, respectively. The north-south (fault-normal) component failed to record the main event, but it likely contained a pulse-like motiondue to forward-directivity. The Sakarya station is located in southwestern Adapazari at a distanceof 3.3 km from the fault rupture. It is situated on the floor of a small 1-story building (with no

basement) and is underlain by a shallow deposit of stiff soil overlying bedrock (averageVs ∼ 470 m/sover top 30 m, Rathje et al. 2003). Downtown Adapazari is located at a distance of about 7 km fromthe rupture, and motions there would differ from those at the accelerograph due to different site-source distances and ground response effects associated with the relatively soft and deep alluviumin the downtown area. Ground motions recorded at similar site-source distances on deep alluviumsuggest that the PGA in Adapazari was on the order of 0.35–0.45 g.

Figure 1 shows a representative soil column of downtown Adapazari based on a deep boring

performed by the General Directorate of Waterworks (DSI). Shear wave velocity (Vs) measurementsto a depth of 10 m were performed at this site using Spectral Analysis of Surface Waves (SASW)

by Cox (2001) and to 30 m using Rayleigh wave inversion by Anatolian Geophysical (O. Yilmaz2003, personal communication). Additionally, SASW data to 50 m depth is available at a nearbysite that has similar subsurface conditions. These data form the basis for the Vs profile shown inFigure 1 (average Vs ∼ 150 m/s over the top 30 m).

The level of shaking in Adapazari was sufficiently intense that the CSR induced in the criticalsoil layers within the upper 6 m by the earthquake were largely between 0.3 and 0.5 (Bray et al.2004). Over this range, SPT-based and CPT-based cyclic resistance ratio (CRR) curves are fairlyvertical. Hence, uncertainty in estimating the demand from this event for the free-field case does not

significantly affect the liquefaction triggering assessment. However, the liquefaction assessmentis highly sensitive to penetration resistance.

4 SUBSURFACE CHARACTERIZATION OF ADAPAZARI SOILS

The heart of the City ofAdapazari, which inTurkish means “island market,” lies in a plain formed byrecent fluvial activity of the meandering Sakarya and Çark rivers. As evidence of the active fluvial

processes in the Adapazari basin, a masonry bridge built in 559 AD across the basin’s primary

326




Figure 1. Generalized subsurface conditions and shear wave velocity profile in downtown Adapazari (from

Bray et al. 2004).

river is now 4 km west of its current alignment (Ambraseys and Zatopek 1969). Most of the cityis located over deep sediments (e.g., Endes et al. 1998, Rathje et al. 2000, Kudo et al. 2002, and Komazawa et al. 2002). The Adapazari basin is as deep as 200 to 400 m near the center of the city,

but the alluvium thins toward the hills in the southwest part of the city.The Adapazari basin is a former Pliocene-Pleistocene lake. The lake sediments are overlain

by late Pleistocene and early-Holocene alluvium transported from the mountains north and southof the basin. The shallow soils (depth < 10 m) are recent Holocene deposits laid down by theSakarya and Çark rivers, which frequently flooded the area until flood control dams were built.An organic sample retrieved from a depth of 4 m at a site within the city was dated to be only1,000 years old, and floods as recent as 50 years ago continued to deposit sediment throughoutthe city (Sancio 2003). Sands accumulated along bends of the meandering rivers, and the rivers

flooded periodically depositing nonplastic silts, silty sands, and clays throughout the city. Clay-richsediments were deposited in lowland areas where floodwaters created ponds (Onalp et al. 2001).

An extensive field investigation program was carried out at selected building sites and alongstreets surveyed previously in Adapazari to document the subsurface conditions and to identifythe soil deposits that might have had a detrimental effect on building performance during theearthquake. The site investigation program included 135 Cone Penetration Test (CPT) profilesand 46 exploratory borings with closely spaced Standard Penetration Tests (SPT) with energymeasurements. The details of this investigation have been described by Bray et al. (2003).

Many soil profiles are characterized as loose silts and silty sands in the upper 4 to 5 m whichoverlie clay deposits with some silty sand layers, although at several locations a 4 to 5 m-thick layer

of dense sand lies between the surficial silt/silty sand layer and the deeper clay layers and at other locations clayey soils replace the shallow deposits of silts (Bray et al. 2001, Onalp et al. 2001,Sancio et al. 2002, and Sancio 2003).

Sancio et al. (2002) developed four general subsurface profiles, Type 1 through Type 4, for four central districts of downtown Adapazari (see Figure 2), and subsequently classified the soilconditions along the lines into one of these four generalized soil profiles. Soil profiles Type 1through Type 3 contain soil deposits that are susceptible to significant strength loss as a result of cyclic loading. Soil profileType 4 does not contain soils susceptible to severe strength loss as a resultof earthquake shaking. The relative amount and degree of ground failure that occurred in the central

327





Figure 2. Generalized subsurface soil profiles in downtown Adapazari (from Sancio et al. 2002).

districts of Adapazari appears to be principally controlled by shallow soil deposits. Ground failure principally occurred in zones that contained shallow, saturated, loose silts that were susceptibleto cyclic mobility based on newly proposed criteria of Bray and Sancio (2006) that supersedesthe Chinese Criteria by Seed and Idriss (1982), which has been shown to be unconservative due

primarily to the clay-size criterion.

5 BUILDING DAMAGE AND GROUND FAILURE SURVEY DATA

5.1 General assessment

The widespread occurrence of ground failure in Adapazari made it conducive for the investigationof the response of buildings on shallow foundations. Excessive settlement and tilt of buildings onshallow foundations overlying sandy deposits have been previously documented in other earth-quakes, for example in Niigata after the 1964 Niigata earthquake by Seed and Idriss (1967) and others and in Dagupan City after the 1990 Luzon, Philippines earthquake byTokimatsu et al. (1992)and others. However, the phenomena observed in Adapazari are of particular interest because of

the f ine-grained nature of the soil deposits that underwent ground failure and because of the pre-dominance of moderate deformations that sometimes allowed fo the building to be inhabitable after limited repairs, if the corresponding structural damage was light.

Different forms of ground failure-related building damage were identified during the reconnais-sance missions that followed the earthquake, some of which are shown in Figures 3 and 4. Incidentsof ground failure can be categorized as follows:

Uniform vertical displacement : Many buildings in Adapazari sank into the ground, many timeswithout noticeable tilt, as is shown for the case of Figure 3a. At times, heave of the surroundingground was observed, as in the case of Figure 3b.

Vertical displacement with tilt : Some buildings experienced non-uniform vertical deformation,

causing the building to be condemned albeit devoid of structural damage, as shown in Figure 3c.Toppling of buildings, as depicted in Figure 3d, was typically observed in laterally unconstrained slender buildings, i.e., large ratio of building height (H) to its width (B).

Lateral translation: A previously undocumented failure mode was also observed in Adapazari.Some buildings translated laterally over the soil directly beneath their foundation. Two suchcases are depicted in Figure 4. In the f irst case, the structure displaced 31 cm away from the

previously adjacent sidewalk. In the second case, the structure displaced approximately 110 cmin the direction of an open alley mobilizing a passive wedge of soil. This building also translated 50 cm to 55 cm in the orthogonal direction, and displaced vertically a significant amount.

328





Figure 3. Examples of ground failure-induced building damage in Adapazari after the August 17, 1999

Kocaeli earthquake: a) vertical displacement, b) vertical displacement with ground heave, c) vertical

displacement with significant tilt, and d) bearing capacity failure (from Sancio et al. 2004).

5.2 Detailed survey data

A total of 719 structures were surveyed in Adapazari, which is about 4% of the building stock. Thedensity and height of construction was fairly consistent along the survey lines, so that variations indamage intensity are statistically meaningful. The degree of structural damage to a building wasdescribed using a modification of a system first proposed by Coburn and Spence (1992), whereeach building is assigned a Structural Damage Index ranging from D0 (no observed damage) to D5(complete collapse of the building or a story within the building). Information on observed vertical

building displacement or penetration relative to the adjacent ground, tilt, lateral movement, and eruption of sand boils was compiled by post-earthquake investigators, using the Ground FailureIndex described by Bray and Stewart (2000). GF0 corresponds to no observable ground failure and

GF3 to significant building penetration of more than 25 cm, building translation of more than 10 cm,or 3 degrees tilt. Detailed description of the structural damage index (D) and ground failure index(GF) is provided in Table 1 and Table 2, respectively. The building survey database is summarized in Table 3.

5.3 Findings

Some localities with severe ground failure also had significant structural damage, whereas othershad only moderate structural damage. Broad areas with ground failure and only light structural

329





Figure 4. Examples of lateral displacement of structures on mat foundations in Adapazari after the 1999

Kocaeli earthquake. Note the development of a passive resistance wedge in the photograph on the right (from

Sancio et al. 2004).

Table 1. Structural Damage Index (Bray & Stewart 2000, modified from Coburn & Spence 1992).

Index Description Interpretation

D0 No observable damage No cracking, broken glass, etc.

D1 Light damage Cosmetic cracking, no observable distress to load bearing structural

elements

D2 Moderate damage Cracking in load-bearing elements, but no significant displacements

across these cracks

D3 Heavy damage Cracking in load-bearing elements with significant deformations across

the cracks

D4 Partial collapse Collapse of a portion of the building in plan view (i.e., a corner or a wing

of the building)

D5 Collapse Collapse of the complete structure or loss of a floor

Table 2. Ground Failure Index (Bray and Stewart 2000).

Index Description Interpretation

GF0 No observable ground failure No settlement, tilt, lateral movement, or boils

GF1 Minor ground failure Settlement, < 10 cm; tilt of >3-story buildings< 1 deg;

no lateral movements

GF2 Moderate Ground Failure 10<< 25 cm; tilts of 1–3 deg; lateral movements< 10cm

GF3 Significant Ground Failure > 25 cm; tilts of >3 degrees; lateral movements> 25 cm.

damage were not prevalent, but they did exist. However, overall, the compiled data indicate that theseverity of structural damage generally increases with increasing levels of ground failure (Sancioet al. 2002). Sediment ejecta were observed within some of the ground failure zones, but ejectawere absent from many areas.

330





Table 3. Summary of buildings per line for which data

are provided in the paper.

Line # No. of Buildings

provided in Database

1 143

2 34

3 70

11∗ 84

∗the original survey included line 11 which was later

grouped with other data and re-named Line 4. The original

name is maintained in this paper.

Figure 5. a) 5-story high building A1 seen in the background. Building A2 is seen in the foreground (Photo

courtesy of D. Frost); b) Building A2 as seen 8 months after the earthquake. Note than building A1 has been

demolished.

6 REPRESENTATIVE OBSERVATIONS OF BUILDING PERFORMANCE

6.1 Site A – Two 5-story apartment buildings in Tul street

The two reinforced concrete 5-story apartment buildings shown in Figure 5 are located alongTül Street, between Yakin and Telli Streets, in the Cumhuriyet District of Adapazari. The GPScoordinates of the site are N 40.77922◦ and E 30.39487◦.

The building seen in the background of Figure 5a (designated Building A1 in Figure 6) experi-enced non-uniform relative downward movement of 10 to 15 cm at the SE corner and 150 cm at the

NW corner. According to Bray and Stewart (2000), the displacement at the NW corner is relatively precisely known because it was measured across the gate shown in Figure 7a. Associated with thisnon-uniform settlement, the building tilted 3◦ to the west and 5◦ to the north.

331





Figure 6. Plan view of Site A and location of subsurface exploration points.

Figure 7. a) Entrance gate to building A1, where 150 cm of settlement were measured at this location 8 days

after the earthquake; b) Entrance gate to building A2, where 30 cm to 35 cm of settlement were measured at

this location 8 months after the earthquake.

No significant foundation distress was observed in the garage at the base of Building A1, indi-cating that the foundation underwent essentially rigid-body settlement and rotation. The building’sstructural frame was essentially undamaged by the earthquake (Bray and Stewart, 2000). How-ever, Building A1 was demolished due to the excessive tilt of the building as a result of ground failure. Photographs that were taken 8 days after the Kocaeli earthquake occurred are shown in

332






Figure 8. a) East to west view of Tül street 8 days after the Kocaeli event. Note the cracking of the pavement

with apparent bulging. (Photo courtesy of J. P. Bardet); b) View of the northeast corner of building A1 in which

pavement cracking is also evident. However, bulging is not apparent on the east side of the building (Photo

courtesy of J. P. Bardet).

Figure 9. North-south cross section across the foundation and ground floor of building A1.

Figure 8. Bulging and cracking of the pavement is evident as a consequence of downward move-ment of Building A1. Measurements taken 8 months after the earthquake across the gate adjacentto Building A1, which is shown in Figure 7b, indicate that the settlement of this building was inthe range of 30 cm to 35 cm.

The buildings were irregularly shaped in their plan view (Figure 6). Building A1’s largest plandimension was 12 m in the E-W direction, and its minimum plan dimension was 8.4 m in the N-Sdirection alongYakin Street. The height of this building above ground was 13.6 m. Hence, its heightto width ratio was 1.6. The largest and smallest plan dimensions of Building A2 are 16.8 m in the

N-S direction and 12.9 m in the E-W direction, respectively. Its height is 13.6 m, so its height towidth ratio is 1.1.The structural drawings for both buildings indicate that they were built following the same basic

design. The thick mat foundation is located at a depth of 1.5 m below the sidewalk elevation. As isshown in Figure 9, the mat consists of 1.2 m-deep reinforced concrete tie-beams over a solid, 30 cm-thick slab. The space between the tie-beams was filled with soil, and the first floor nonstructuralconcrete slab placed over this fill. The building drawings also specify that the surface over whichthe foundation will be placed should be prepared with a layer of coarse aggregate over which a thinlayer of fine aggregate concrete is placed.

333






Figure 10. a) View from Meydan street of the north and east sides of the building, and b) view of the north

and west sides of the structure.

As shown in Figure 6, six Cone Penetration Tests (CPT), two of them with downhole seismicwave profiling (SCPTU), and four exploratory borings with the Standard Penetration Tests (SPT)at closely spaced intervals were performed to characterize the subsurface conditions at the site.Additionally, the field vane test was used at shallow depths in the boring identified as SPT-A4.

6.2 Site D – 4 story reinforced concrete building in Meydan street

The 4-story reinforced concrete building shown in Figure 10a and 10b is located along MeydanStreet, Çukurahmediye District, Adapazari as indicated in Figure 11.

According to Bray and Stewart (2000), this building experienced approximately 44 cm of verticalsettlement, 50 cm to 55 cm of lateral displacement to the west and 100 cm of lateral displacementto the south after the August 17, (1999), Kocaeli earthquake.

As depicted in Figure 12, this building has lateral dimensions of 11 m in the E-W directionand 9.8 m in the N-S direction. The height of the building, as indicated by the building drawingsis 12.75 m. However on June 16, 2000 the height of the building was measured to be 13.5 m.Considering the latter, the height to width ratio is 1.4.

Tilt measurements performed on the roof 10 months after the Kocaeli earthquake showed thatthe building is inclined towards the west and the south by approximately 1◦ and 2◦, respectively.Inspection of the structural components such as beams and columns showed no evidence of cracking

or distress. Non-structural components, such as infill walls and hardwood floors, also appeared undamaged.According to the building drawings, the building was built over a mat foundation consisting of

1.2 m-deep reinforced concrete tie beams over a solid, 30 cm-thick slab. It can be noticed in theE-W cross section of the structure and foundation of the building shown in Figure 13 that the firststory of the building has a greater height than the other stories as it is used for commercial purposes.

As shown in Figure 12, 3 CPTs, one of them with downhole seismic wave profiling (SCPTU)and 4 exploratory borings with the implementation of the SPT with energy measurements were

performed to identify and characterize the shallow subsurface soils at the site. Shelby tube samples

334





Figure 11. Map of Çukurahmediye district showing the location of the 4-story building studied.

Figure 12. Plan view of Site D and location of subsurface exploration points.

were retrieved at the boring named SPT-D2 and the samples were tested to obtain the stress historyand triaxial undrained shear strength of the shallow deposits.

7 MECHANISMS OF FOUNDATION FAILURE IN ADAPAZARI

Figure 14 depicts two common modes of building performance in Adapazari after the Kocaeliearthquake. The drawing to the left shows a stout building with a large mat foundation where its

width is much greater than the thickness of the underlying liquefiable silt deposit. The drawing tothe right depicts a slender building with a narrow foundation width.Based on the interpretation of the results of the in situ tests by Bray et al. (2004), the shallow

silt deposit was identified as the critical layer under most of the buildings studied in Adapazari.In general, deeper deposits (5 m< depth< 10 m) of silt and sand were often too dense to havecontributed significantly to the observed building performance. Moreover, soil specimens obtained from the deeper silt strata that were potentially liquefiable, which were sampled and then tested ina cyclic triaxial system, exhibited significantly greater cyclic strength than the shallow silt (Sancio2003). Although at some sites the deeper layers might have developed significant excess pore water

335






Figure 13. East-west cross section across the foundation, ground floor, and first floor of building D1.

Heave

0 5 10 15 20 m

HORIZONTAL SCALE

0

5

10

0

5

10

ML

CH

CL w/ Silt

Clayey Silty Sandy

Heave

Figure 14. Modes of failure of stout and slender buildings in Adapazari.

pressures and softened considerably, their contribution to the observed building performance is judged to be less significant than that of the looser shallow soil layers. In many cases, the response of the upper loose, low plasticity silt clearly dominated the building response (e.g., as in the case of lateral translation of buildings). Moreover, ground failure and building performance at soil profile

336





Type 3 sites (Figure 2), which had nonliquefiable, dense sand underlying the shallow liquefiablesoils, were fairly consistent with those observed at soil profiles Type 1 and Type 2, which had marginally liquefiable silt/sand deposits below the highly liquefiable, shallow soil. Hence, thesoils below a depth of 5 m appeared to have relatively little impact on observed performance of

buildings at these sites.Without considerable error, it can be assumed that only the saturated, loose, nonplastic-to-

low plasticity silt layer (ML) shown in Figure 14 lost significant strength during the earthquake.Additionally, based on eye witness accounts of people who ran out of buildings and encountered building punching into the ground, it can be reasonably assumed that most of the deformationsoccurred during the period of strong earthquake shaking.

The near-fault earthquake-induced shear stresses in the free-field with estimated peak ground accelerations of 0.35 g to 0.45 g at the ground surface (Bray et al. 2004) were sufficient to generate

positive excess pore water pressures and subsequent loss of strength and stiffness of the saturated shallow, loose soil in the free-field. Dynamic analyses by Travasarou et al. (2006) indicate that thecyclic stress ratios induced near and under the edge of a building’s mat foundation were even higher than those induced in the free-field. Hence, the soils in these areas generated even larger positive

excess pore water pressures. Due to the weight of the overlying structure, high excess pore water pressures were also generated directly underneath the building.The generation of high positive excess water pressures under and near the stout building would

cause the soils susceptible to cyclic strength loss to soften and displace laterally under the weightof the building (the failure mode shown at the left of Figure 14). Thus, the soil likely squeezed laterally due to earthquake-induced deviatoric shear under the extra load imposed by the heavy

buildings, which induced some ground heave (e.g., Figure 3b). Additionally, due to the hydraulicgradients developed from the contrast of relatively high pore water pressures under the structuresto the relatively low pore water pressures in the free field, partial drainage of these soils likelyoccurred, which allowed some volumetric strain to accumulate during earthquake strong shaking.This mechanism helps to explain the numerous observations of buildings punching into the ground

with no discernable ground heave (e.g., Figure 3a). The earthquake-induced soil deformations aresufficiently large for the soil to dilate eventually and recover its shear strength and once againwithstand the static weight of the structure.

Subsequent cycles of earthquake shaking generated or sustained positive excess pore water pressures, which in turn with ground and building shaking produce additional vertical deformationof the building foundation. These stress cycles soften the soil such that it undergoes cycles of strength loss with limited strain potential, i.e. cyclic mobility, and this “ratcheting” effect alsocontributes to building settlement.

Conventional methods used to calculate deformations along potential sliding surfaces (i.e., Newmark-type analysis) are not directly applicable, because the soil in this case is not slidingalong a surface but deforming under the dynamic and static load imposed by the building. The

implementation of the finite element method to model this problem is perhaps a better way to ana-lyze the deformations to which soil elements are subjected. These analyses should be the work of continuing research into the performance of buildings over liquefiable ground to develop method-ologies that allow design engineers to estimate seismically induced settlement of foundations over

potentially liquef iable ground (e.g., Dashti et al. 2008).The other failure mechanism, which is shown at the right of Figure 14, is more representative

of a conventional bearing capacity-type failure. In this case, as in the one previously described,the generation of positive excess pore water pressures causes the soil to temporarily lose strength.Additionally, horizontal ground shaking causes the building to apply an overturning moment atthe foundation level, or equivalently, an eccentricity of the vertical load. The magnitude of the

overturning moment and thus the eccentricity is a function of the seismic response of the buildingand the height of the building.If the mat foundation is narrow, the effect of the eccentric load is greater, because it causes stress

concentrations over a smaller area of the mat foundation. When this stress approaches or exceedsthe seismic bearing capacity of the soil (i.e., considering the reduction of strength due to excess porewater pressure generation), the building begins to tilt. As tilting is initiated, the area over which thestresses are applied is reduced, thus the magnitude of the stress increases. Under these conditions, a

progressive failure is possible. Continuing tilt will cause toppling unless the bearing capacity of thesoil increases sufficiently due to dilation or due to an increase of effective stress due to dissipation

337




of excess pore water pressure, or cessation of shaking which causes the overturning moment toreduce significantly.

8 CONCLUSIONS

A large number of buildings in Adapazari experienced earthquake-induced ground failure as wellas significant structural damage. Although structural damage in Adapazari was primarily due tostrong ground shaking and poor structural design and construction, building damage also appearsto be related to ground failure, with significant building damage being more likely to occur in areasof significant ground failure.

The observations of ground failure and building damage prompted several surveys and in-depth studies of the ground conditions in Adapazari. The results of these surveys and subsurfaceinvestigations are documented in this paper and the accompanying electronic files.

In Adapazari, vertical displacement of buildings into the ground during the 1999 Kocaeli earth-quake appears to be related to the applied static and dynamic contact pressure at the foundation of these buildings. Building settlement is also affected by a large number of other variables that cannot

be independently assessed through these case histories, although useful insights can be gleaned from the available data. The development of engineering tools for evaluating the consequences of seismically induced excess pore water pressure and the effects of soil liquefaction/softening on

building performance warrants more attention.

ACKNOWLEDGEMENTS

Financial support was provided by the National Science Foundation (NSF) under Grant CMS-

0116006. Initial reconnaissance efforts were also supported by the NSF through the GeoengineeringExtreme Events Reconnaissance (GEER) Association. Dr. Turan Durgunoglu and his ZETASCorporation provided the laboratory equipment in Istanbul, as well as the drilling equipmentin Adapazari. Professor Akin Onalp and his laboratory at Sakarya University provided labora-tory support for the initial subsurface characterization testing. Other support was provided byZETAS Corporation and our Turkish colleagues, and this and all support is greatly appreciated.The efforts of those who participated in the GEER post-earthquake reconnaissance effort and theinitial subsurface characterization program, i.e., T.L. Youd, J.P. Stewart, E.M. Rathje, D. Frost,J-P Bardet, A. Onalp, T. Durgunoglu, R.B. Seed, O. K. Cetin, E. Bol, M.B. Baturay, C. Christensen,T. Karadayilar, A. Ansal, A. Barka, R. Boulanger, D. Erten, I.M. Idriss, A. Kaya, T. O’Rourke,

and D. Ural, are acknowledged, especially the collection of building data by Jonathan Stewart, LesYoud, and Curt Christensen, which was so useful in this study.

REFERENCES

Bird, J.F., Sancio, R.B., Bray, J.D., and Bommer, J.J. 2004. The ground failure component of earthquake

loss estimations: A case study for Adapazari Turkey. Proceedings of the 13th Conference on Earhquake

Engineering , Vancouver, B.C., Canada. Paper No.803.

Bray, J.D. and Stewart, J.P. 2000. Damage patterns and foundation performance in Adapazari. Kocaeli,

Turkey Earthquake of August 17, 1999 Reconnaissance Report, Earthquake Spectra; 16 (Suppl. A):

163–189.Bray, J.D., Sancio, R.B., Youd, T.L., Durgunoglu, H.T., Onalp, A., Cetin, K.O., Seed, R.B., Stewart, J.P.,

Christensen, C., Baturay, M.B., Karadayilar, T., and Emrem, C. 2003. Documenting Incidents of Ground

Failure Resulting from the August 17, 1999 Kocaeli, Turkey Earthquake. Data Report Characterizing Sub-

surface Conditions. Geoengineering Research Report No. UCB/GE-03/02, Univ. of California, Berkeley,

May 15 (provided as an appendix to this paper).

Coburn A, Spence R. 1992. Earthquake Protection. Wiley, Chichester.

Cox, B.R. 2001. Shear Wave Velocity Profiles at Sites Liquefied by the 1999 Kocaeli, Turkey Earthquake.

M.S. Thesis, Utah State University, Logan, Utah.

338




Dashti, S., J.D. Bray, M.R. Riemer, D. Wilson. 2008. Centrifuge Experimentation of Building Performance on

Liquefied Ground. Proc. Geotechnical Earthquake Engineering and Soil Dynamics IV, ASCE Geotechnical

Special Publication No. 181, Zeng, D. et al., Ed., May 18–22, Sacramento, CA.

Endes, H., Kurtulus, C., and Asilhan, M. 1998. Adapazari bedrock depth investigation study. Subsurface

investigation group. Kocaeli University, February, Vol. 1, 39 pages (in Turkish).

Komazawa, M., Morikawa, H., Nakamura, K., Akamatsu, J., Nishimura, K., Sawada, S., Erken, A., and

Onalp, A. 2002. Bedrock structure in Adapazari, Turkey – a possible cause of severe damage by the 1999Kocaeli earthquake. Soil Dynamics and Earthquake Engineering. 22: 829–836.

Kudo, K., Kanno, T., Okada, H., Ozel, O., Erdik, M., Sasatani, T., Higashi, S., Takahashi, M., and Yoshida, K.

2002. Site-specific issues for strong ground motions during the Kocaeli, Turkey, earthquake of 17 August

1999, as inferred from array observations of microtremors and aftershocks. Bulletin of the Seismological

Society of America. 92(1): 448–465.

Liu, H. 1995. An empirical Formula for Evaluation of Buildings Settlements due to Earthquake Liquefaction.

Proc. of the 3rd International Conference on Recent Advances in Geotechnical Earthquake Engineering

and Soil Dynamics. 1: 289–293.

Liu, H., and Dobry, R. 1997. Seismic response of Shallow Foundation on Liquefied Sand. Journal of

Geotechnical and Geoenvironmental Engineering. 123(6): 557–567.

Meyerhof, G.G. 1964. Shallow Foundations. Proc. of the ASCE Conference on Design of Foundations for Control of Settlements. Northwestern University, Evanston, Ill, June.

Onalp, A., Arel, E., and Bol, E. 2001. A general Assessment of the Effects of 1999 Earthquake on the Soil-

Structure Interaction in Adapazari. In Jubilee Papers in Honor of Prof. ErgunTogrul , 10, ICSMFE, Istanbul,

Turkey.

Rathje, E., Idriss, I. M., and Somerville, P. 2000. Strong ground motions and site effects. Earthquake Spectra,

Supplement A, V. 16. Chap. 4: 65–96.

Rathje, E.M., Stokoe, K.H., and Rosenblad, B. 2003. Strong Motion Station Characterization and Site Effects

During the 1999 Earthquakes in Turkey. Earthquake Spectra. 19(3): 653–675.

Sancio, R.B. 2003. Ground Failure and Building Performance in Adapazari, Turkey. Dissertation in Partial

Satisfaction of the Requirements for the Degree of Doctor of Philosophy, University of California, Berkeley,

Fall 2003.

Sancio, RB, Bray JD, Stewart JP Youd TL, Durgunoglu HT, Onalp A, Seed RB, Christensen C, Baturay MB,

Karadayilar T. 2002. Correlation between ground failure and soil conditions in Adapazari, Turkey. Soil

Dynamics and Earthquake Engineering . 22: 1093–1102

Seed, H.B. and Idriss, I.M. 1967. Analysis of soil liquefaction: Niigata earthquake. Journal of the Soil

Mechanics and Foundation Division, ASCE . 93(3): 83–108.

Seed, H. B. and Idriss, I. M. 1982. Ground Motions and Soil Liquefaction During Earthquakes. EERI

Monograph, Berkeley, California, 134 pages.

Shahien, M.M. 1998. Settlement of Structures on Granular Soils Subjected to Static and Earthquake Loads.

Ph.D. Thesis. University of Illinois at Urbana-Champaign, Urbana, Illinois.

Tokimatsu, K., Kojima, H., Kuwayama, S., Abe, A., and Midorikawa, S. 1992. Liquefaction-induced damage

to buildings in 1990 Luzon earthquake. Journal of Geotechnical Engineering, ASCE . 120(2): 290–307.

Travasarou, T., Bray, J.D., and Sancio, R.B. 2006. Soil-Structure Interaction Analyses of Building Responses

During the 1999 Kocaeli Earthquake. Proc. 8th US Nat. Conf. EQ Engrg., 100th Anniversary Earthquake

Conference Commemorating the 1906 San Francisco Earthquake, EERI, Paper 1877.

Yoshimi, Y. and Tokimatsu, K. 1977. Settlement of Buildings on Saturated Sand During Earthquakes. Soils

and Foundations. 17(1): 23–38.

Youd, T.L., and Perkins, J.B. 1987. Mapping of Liquefaction Severity Index. Journal of Geotechnical

Engineering, ASCE . 113(11): 1374–1392.

DATA APPENDICES

This paper is provided together with the following data contained in a companion CD:

1) UC Berkeley Geotechnical Research Report No. UCB/GE-03/02, “Documenting Incidents of Ground Failure Resulting from the August 17, 1999 Kocaeli, Turkey Earthquake: Data ReportCharacterizing Subsurface Conditions, May 15, 2003.” This report is provided as a file named “Bray et al 2003Adapazari_data_report.pdf”. In addition, the companion interactive html-based

pages are provided in a folder named “Adapazari Subsurface Data.” Click on the file labeled “INDEX-Start file.html” to access this information. The files in the html-based web pages

339




contain descriptions of the sites investigated and all of the soil boring and CPT data collected during the site investigation carried out in the year 2000 to characterize the subsurface conditionsin Adapazari. Data are provided as pdf reports and in electronic format when available.

2) Microsoft Excel™ spreadsheet, which is named “Adapazari+Building Performance”, con-taining the structural damage index (D) and ground failure index (GF) collected during the

post-earthquake reconnaissance survey as described in Table 3. When available, building plan

dimensions and reliable estimates of building settlement are provided.3) Two drawings of four central Adapazari districts showing the approximate location of CPTs as

well as the block numbers used as nomenclature in the survey lines. The drawings, which arenamed “Adapazari+CPT” and “Adapazari+Blocks”, respectively, are provided in .pdf format.

4) Ground motion recording of earthquake ground shaking during the August 17, 1999 Kocaeliearthquake in the vicinity ofAdapazari, Turkey. This is the Sakarya 090 acceleration-time history,which is named “Sakarya 090 ERD Recording-PEER, that was recorded by ERD and processed

by the Pacific Engineering for the PEER Strong Motion Database.

340




Figure 1. Map showing distribution of structural damage and liquefied area.

Figure 2. Particle orbits of ground displacement during main shock.

Figure 3. Response spectra of recorded strong ground motions.

developed to a lesser extent in central Port Island and southern Rokko Island where the fills had been treated or consist of soils containing significant amounts of clay.

Figure 2 shows the particle orbits of ground displacements that have been computed by double-integration of the recorded strong motion in the areas during the main shock (CEORKA, 1995;Sekisui House, 1996). The peak cyclic ground displacements were 35 cm at a non-liquefied site

on Rokko Island, and 46 cm and 55 cm near liquefied areas on Port Island and in Fukaehama. Itis estimated that about two thirds of these displacements resulted from shear strains induced inthe reclaimed fills. The acceleration response spectra for damping ratios of 10% range around 0.3–0.5 G for periods less than 0.5 s, as shown in Figure 3.

2.2 Characteristics of piles for building foundations

The piles used for building foundations in the areas include precast concrete piles, steel pipe (S) pilesand cast-in-place concrete (CC) piles. Most precast piles are hollow with outer diameters typically

342







Table 1. Case histories of pile foundations in the Kobe earthquake (after Tokimatsu and Asaka, 1998).

Building Building No. Dimension Depth of Pile Pile Pile Axial Ground

or site type of (m) pile head type diameter length force water

story (m) (mm) (m) (MN) table (m)

A RC 4 10.6×16.5 1.2 PHC-A 350 22 0.3 3

B RC 5 11.7×25.2 1.7 SC+PHC-A 600 31 0.9 2.2

C RC 4 24.8×75.6 1.73–2.13 SC+PHC-B 500 35–42 0.63 3

D RC 6 12× 106.4 1.78 S 500 30 0.66 3

E NA NA 42× 50 1 SC+PHC-A 500 33 0 2

F RC 3 23.3×7.5 1.65 PC-A 400 20 0.39 2

G RC+S 2 187× 18 2.5 S 406.4 27.55 0.63 3.5

H SRC NA 41.8×16.1 2.10–2.30 CC 1500 47–48 0.14 1.65

Figure 4. Relation of bending moment with curvature of piles.

ranged from 35 to 60 cm, which contrasts with solid cast-in-place concrete piles with diameterstypically over 100 cm. The precast concrete piles encountered in the Kobe area include prestressed concrete (PC) piles made before 1980s and prestressed high strength concrete (PHC) piles madeafter 1980s. Both PC and PHC piles have three different capacities for a given diameter, i.e., TypesA, B, and C in ascending order of capacity. To strengthen the capacity and ductility, steel pipereinforced concrete (SC) piles and reinforced prestressed concrete (PRC) piles have been also used.

Table 1 summarizes the characteristics of pile foundations of buildings/sites described in this paper, herein called Buildings/Sites A-H, and Figure 4 shows the estimated relations of bendingmoment with curvature for the piles. In the following, Mu is the bending moment at concretecrashing at extreme compression fiber of PC, PHC, and SC piles or at which compression fiber strain of S pile reaches a limiting value; My is the bending moment at yielding of tension bars of PC and PHC piles or at yielding of steel at extreme tension fiber of SC and S piles; and M c is the

bending moment at concrete cracking at extreme tension f iber of PC and PHC piles.

3 OUTLINE OF PILE FERFORMANCE FROM FIELD INVESTIGATION

A large number of performance of pile foundations during the Kobe earthquake have been presented, based on field investigation including excavation of pile heads (e. g., Kansai Branch of ArchitecturalInstitute Japan (AIJ), 1996; AIJ et al., 1998). In addition to integrity tests for piles (Figure 5), severalmethods were used for detectingpile damage below the ground surface. Borehole cameras (Figure 6,Oh-oka et al., 1996) have identified damage portions and severity (Figure 7), and inclinometers(Figure 8, Shamoto et al., 1996) have provided data to estimate deformed shapes with depthof precast hollow piles. The main findings from the field investigation are summarized below(Tokimatsu et al, 1996; Tokimatsu and Asaka, 1998; Tokimatsu et al, 1998; and Tokimatsu, 2003).

343





Figure 5. Integrity test for detecting pile damage. Figure 6. Borehole camera survey for detecting pile

damage.

Figure 7. Crack detected by borehole camera.

Figure 8. Inclinometer for detecting

deformation of pile.

3.1 Damage to piles in non-liquefied area

In the nonliquefied area where damage to superstructures concentrated, the damage features of piles are summarized as follows:

1) Quite a few buildings tilted with their superstructures remaining intact.2) Most of these buildings suffered extensive shear failure of their pile heads, as illustrated in

Figure 9, leading to complete demolition or significant restoration including underpinning of the building, as shown in Figure 10.

3) Such damage was particularly significant in buildings with a high aspect ratio that could piles.

3.2 Damage to piles in liquefied area

In the liquefied level ground, the damage features of piles are summarized in Figure 11 and asfollows:

1) Most non-ductile precast concrete (PC) piles and many pre-stressed high-strength concrete(PHC) piles bearing on firm strata below liquefied layers suffered severe damage oftenaccompanied by settlement and/or tilting of their superstructures (Figs. 11(a)(b) and 12);

344








Figure 9. Shear failure of pile.

Figure 10. Extensive restoration of pile foundation

(Courtesy of Dr. Y. Suzuki).

Figure 11. Performance of pile foundations in liquefied level ground.

2) Failures of those piles concentrated at the top and the bottom of the liquefied layer rather thanat the pile head (Figs. 11(a)(b)));

3) Follower piles consisting of upper ductile steel reinforced concrete SC piles and lower PHC piles (Type-A) with a low moment capacity also damaged only near the bottom of the liquefied layer (Fig. 11(c)(d), Nagai, 1997; Horikoshi & Ohtsu, 1996);

4) Follower piles consisting of upper ductile SC and lower PHC piles (Type B), having a momentcapacity greater than that of PHC (Type A) piles, however, did not show any visible damage(Fig. 11(e), Fujii et al., 1996);

5) Ductile steel pipe (S) piles and SC piles appeared not to have experienced any significant damage(Fig. 11(f), e.g. Japanese Association of Steel Pipe Piles (JASPP), 1996);

6) Low-rise buildings supported on friction piles embedded within fills often performed well,while they settled with the surrounding ground (Fig. 11(g), Fig. 13); and

7) Buildings, if supported on undamaged piles bearing on firm soils, often suffered large verticalgaps induced by liquefaction-induced ground settlement (Fig. 11(h), Fig. 13).

345







Figure 12. Tilted building in liquefied ground.Figure 13. Three-story building supported on fric-

tion piles with foundation supported on point bearing

piles on the left.

Figure 14. Tilted building in laterally spreading

ground (Building F in Fig. 24).

Figure 15. Performance of pile foundations in

Laterally spreading ground (Tokimatsu, 1999).

3.3 Damage to piles in laterally spreading area

In the area where liquefaction-induced lateral spreading occurred, particularly near the waterfront

of artificial islands, the damage features of piles were extensive, often accompanied by a tilt of a building (Fig. 14), and are summarized as follows (Fig. 15):

1) Damage was not limited to nonductile PC and PHC piles but extended to some ductile S piles (Satake et al., 1997) and large cast-in-place concrete piles of diameters greater than 1 m(Tokimatsu et al, 1996; Kuwabara & Yoneda, 1998);

2) Failures of those piles concentrated near the pile head (Fig. 16) or at the top and the bottom of the liquefied layer (Figs. 17 and 18);

3) Damage to pile caps and/or foundation beams often preceded or accompanied damage to ductileand/or high strength S and cast-in-place concrete (CC) piles themselves;

4) Piles within a building near the waterfront showed different failure and deformation modes inthe direction perpendicular to the shoreline as shown in Figure 15 (Tokimatsu et al., 1997), whilethose away from the waterfront showed similar deformation modes (Satake et al., 1997); but

5) When facing the span side of the building with the sea on the left, the sea side pile cap rotated clockwise, whereas the land side pile cap rotated counterclockwise (Fig. 15, Oh-oka et al.,1997b), indicating that the stresses acting on pile head vary within a building; and

6) CC piles surrounded by cement column walls or continuous diaphragm walls did not suffer any serious damage (BTL Committee, 1998), but their superstructure often suffered extensivedamage (BTL Committee, 1998).

346








Figure 16. Damage to pile head due to lateral ground

spreading.

Figure 17. Damage to pile due to lateral ground

spreading detected by borehole camera.

Figure 18. Damage to pile in lateral spreading area.

3.4 Lessons learnt from field investigation

The significant contrast in damage pattern of building between non-liquefied and liquefied area is probably due to the following:

1) Soil liquefaction deamplified the ground motions particularly in the period range less than 1 s,reducing the damage to superstructure of buildings in the liquefied area.

2) Soil liquefaction increased the ground displacement and thus kinematic effects on piles, leadingto the damage to pile foundations.

The above trends indicate that, in addition to horizontal forces and overturning moments imposed on pile heads from superstructures, kinematic forces induced by dynamic and permanent ground displacements had a significant impact on pile damage particularly in liquefied and laterally spread-ing areas. The damage to piles without vertical loads confirms the significant effects of ground movements. The difference in failure and deformation modes of piles within a building near thewaterfront as shown in Figures 15(a) and (d) probably reflects rapid changes in horizontal ground

347







Figure 19. Boring log and pile characteristics

and damage at Building A.

Figure 20. Boring log and pile characteristics and

damage at Building B.

displacement. In addition, Figure 15 suggests that the effect of lateral ground movement on foun-dation damage is more serious in the span direction than in the longitudinal direction. Described in the following chapter are several case histories of pile foundations for which extensive field investigation was made.

4 CASE HISTORIES OF PILE FOUNDATIONS IN LIQUEFIED GROUND

Building A, which is a four-story building supported on 35 cm diam PHC piles in Mikagehama(Shamoto et al., 1997), tilted largely. Figure 19 (a) shows a boring log at the site. The field survey

by Shamoto et al (1977) showed that the piles cracked near the pile head as well as near the bottom

of the fill as shown in Figure 19 (b).Building B, which is a five-story building supported on the SC+PHC (Type A) piles in the center

of Fukaehama, tilted by 1/29 to the northeast after the quake without any damage to superstructure(Nagai, 1997). 50–60 cm diam SC piles 6 m long, and PHC (Type A) piles 13 and 12 m long wereused as the upper, and middle and lower piles. The piles penetrated to a depth of about 33 m througha reclaimed f ill with a thickness of about 10 m (Fig. 20 (a)). An excavation survey suggested thatthe pile heads inclined to the southwest, the opposite direction of the tilting of the building, by1/30-1/23. In addition, a boring into a hollow pile through the pile cap suggested that the pile wasdamaged and bent largely at a depth between 9.6 and 11.3 m below the top of the pile cap (Fig. 20(b)), which appears to be near the bottom of the liquefied layer.

Building C, which is a four-story building supported on the SC+PHC (Type B) piles in an

untreated area on Port Island, suffered neither differential settlement nor structural damage (Fujiiet al., 1996). 50 cm diam SC piles 8 m long, and PHC (Type B) piles 13 and 14 m long were used as the upper, and middle and lower piles. The piles penetrated to a depth of about 37 m through athick reclaimed fill (Fig. 21 (a)). An excavation survey with pile integrity tests from exposed pileheads suggested that the piles had no damage.

Building D, which is one of five-story school buildings supported on the 50 cm diam S piles 30 mlong in Fukaehama, experienced insignificant damage and settlement/tilting, though the ground surface around the building settled by 10 to 50 cm (JASPP, 1996). Figure 22 (a) shows the boringlog at the site. An excavation survey, however, showed that a large residual deformation occurred at the horizontal plate connecting the pile head and the pile cap (Fig. 22 (b)).

A case history of a group of piles at Site E (Horikoshi and Ohtsu, 1996) also provides a good basis to evaluate the effects of ground displacement, since neither inertia force nor vertical load from the superstructure was imposed on the piles during the earthquake. The piles 33 m long (SC

pile 5 m long+PHC piles 13 m and 15 m long) with diameters of either 40 or 50 cm had beendriven through a reclaimed f ill in the center of Fukaehama about 350 m away from the shoreline.The pile heads were located at a depth in between 0.5 and 1.5 m with a ground water table of 2 m.The piles experienced the earthquake before placing any pile caps or foundation beams and failed at a depth of about 8 m, i.e., the interface between gravelly medium sand fill with SPT N-values of 5 and gravelly coarse sand fill with SPT N-values of 12 (Fig. 23 (a),(b)).

348






Figure 24. Map showing location of Building F and vectors of permanent ground displacement (after

Tokimatsu et al, 1998).

Figure 25. Section and foundation plan of Building F (after Tokimatsu et al, 1998).

Two piles labelled S-7 and N-7 in Figure 25, one on the seaside and the other on the landside,were examined using a television camera and a slope-indicator that was invented for hollow piles(e.g., Shamoto et al., 1996). These instruments were in turn inserted into the piles after either coring or removing their pile caps. The slope-indicator can provide separately the slope angles of two orthogonal components of a pile. Either integrating or differentiating of the measured slopeangle with depth can yield the horizontal displacement and curvature of the pile. The details of

350






Figure 26. Boring log and variation with depth of displacement, curvature and damage to piles of Building

F (after Tokimatsu et al, 1998).

Figure 27. Map showing location of Building G and vectors of permanent ground displacement (after

Tokimatsu et al, 1998).

the test equipment and procedure are described elsewhere (Shamoto et al., 1996; Oh-oka et al.,

1996, 1997).Figures 26(b),(c) shows the deformation patterns estimated by the slope-indicator in the spandirection where the displacements of the piles are largest. The horizontal displacements at both

pile heads are almost the same, being equal to 60–80 cm, and consistent with those obtained by theaerial photographic survey. In spite of the same displacements at both pile heads, the deformation

patterns with depth are extremely different. Namely, Pile N-7 inclines simply towards the sea,whereas Pile S-7 bent largely with a failure at a depth of 5 m.

Figures 26 (d),(e) shows the variation of curvature of the piles with depth. Also shown inFigure 26 (f) is the distribution of cracks from television observation. The variations of curvatureare consistent with the failure patterns in such a way that the failure and cracks concentrate wherelarge curvatures occur. This indicates that the slope-indicator used herein can provide deformation

and curvature patterns of piles with a reasonable degree of accuracy. Based on Figure 26 (b)–(f),failures occur at three parts in Pile S-7, i.e., a depth of about 5 m as well as near the pile head and the interface between fill and natural deposit; however, they occur at only two parts in Pile N-7,i.e., near the pile head and the interface between fill and natural deposit.

5.2 Two-story building in Fukae-hama

Building G was located in the northwest of a reclaimed island called Fukae-hama (Figure 27) onthe south of Fukae. The distances from the northwest corner of the building to the northern and

351






Figure 28. Foundation plan of Building G (after Tokimatsu et al, 1998).

Figure 29. Boring log and variation with depth of displacement, curvature and damage to piles of Building

G (after Tokimatsu et al, 1998).

western waterfronts were about 100 m. Figure 28 shows the foundation plan of the building. Thetwo-story building constructed in late 1960s was supported on steel pipe piles 406 mm in diameter with a thickness of 9.5 mm and about 27 meters long. Each pile cap, either 3.4 m by 2.2 m and 1.05 m high or 2.2 m by 2.2 m by 0.75 m high, had 6 or 4 piles. Figure 29 (a) shows a boring logobtained after the quake. A fill about 15-m thick comprising gravelly sands overlies a layer of densesand and gravelly sand that occurs at depths of about 17 m. Underlying this layer are those of densesands and sandy gravels. The groundwater table is located at a depth of about 3.5 meters.

The damage to the superstructure of the building in the 1995 event was minor except for the buckling of longitudinal reinforcements and the spalling of concrete at three columns on the westside. However, most anchor bolts that connected the column with the girder at the second floor

level were broken. Many sand volcanoes observed after the quake indicated the occurrence of soilliquefaction in and around the building. In addition, the asphalted floor in the building as wellas the ground surface nearby settled by as much as 50–60 cm, creating large vertical gaps wherefoundation beams existed. The vectors shown in Figure 27 indicate the magnitude and directionof horizontal ground displacements determined from the aerial photograph survey. The ground displacements are found to have occurred mostly towards the sea, amounting to 3-4 meters near the waterfront, and decreasing to less than 50 cm near the building. The ground displacementcomponent in the longitudinal direction of the building appears to be larger than that in the spandirection.

A television camera was inserted into the hollow space of 17 piles labelled Nos. 1–18 except

for No. 9 in Figure 28. The presence of reinforced plates inside some piles prevented observations below that depth. No local bucking was detected over the length investigated. An inclinometer survey was also made for the same piles except for Nos. 5 and 9 (Satake et al., 1997).

Figure 29 (b)–(e) shows the distribution of the two orthogonal horizontal displacements withdepth for the piles for which the investigation was possible to a depth below the reclaimed fill.The X and Y axes in the figure correspond to the longitudinal and span directions of the building,respectively. Also shown in Figure 29 (f) are the vector displacements relative to the pile tip,

projected on the horizontal plane. Figures 29(b)–(e) indicates that all piles deform above a depth of 12–15 m, resulting in large curvatures at that depth. The depth where large curvatures occur is close

352






Figure 30. Map showing investigated building and ground displacement (Tokimatsu, 2003).

Figure 31. Boring log with distribution of

cracks detected by borehole camera survey(Kuwabara and Yoneda, 1998).

to the bottom of the reclaimed fill. The displacements of pile heads range from 40–100 cm, with their longitudinal components slightly larger than the span components. In addition, the displacementsnear the waterfront are larger than those away from the waterfront. The average displacements inthe longitudinal and span directions are about 70 and 50 cm for Piles Nos. 1 to 4 and 45 and 35 cmfor Piles Nos. 6 to 18. These vector displacements are consistent with those of the ground surfacearound the building estimated from the aerial photographic survey.

Katoh et al. (1997) also showed that, in return for no damage to steel pipe piles, cracks and

spalling of concrete occurred near the pile heads. When facing the southern longitudinal side of the building, all pile caps rotated counterclockwise about their span axes. In contrast, when facingwestern span side of the building, the pile caps on the seaside rotated clockwise while those on thelandside rotated counterclockwise. The difference in pile cap rotation in the span direction would

be consistent with that observed at Building F, although such a difference cannot be identified clearly from Figure 29.

5.3 Under-construction building

Figure 30 shows a map showing the location of Building H (Kuwabara and Yoneda, 1998; and

Tokimatsu et al, 2000). The building, hatched in the figure, was situated on a reclaimed land inOsaka Bay, about 25 m and 90 m inland from the eastern and southern coastlines, respectively.Figure 31 (a) shows a boring log of the site. The surface layer to a depth of 15–16 m is a reclaimed

fill consisting of weathered granite soils, which is underlain by layers of Holocene clay and sandygravel to a depth of about 27 m. Underlying below is a thick Pleistocene layer of sandy gravel,which is considered to be the bearing stratum for the pile foundation of the building. The ground water table was located at a depth of about 2 m.

The building was a 7 by 2 spanned steel frame structure, 41.8 m long in the east-west direc-tion and 16.1 m wide in the north-south direction (Figures 32 and 33), which was supported on

353






Figure 32. Investigated building after earthquake (Tokimatsu, 2003).

Figure 33. Vector slope angles of pile heads observed after earthquake (Kuwabara and Yoneda, 1998).

cast-in-place concrete piles 47–49 m long, having diameters of 1.2–1.7 m. The building, six storiesat its completion, experienced the 1995 Hyogoken-Nambu earthquake when the steel frames of thethird floor were constructed. Thus, the piles carried only 2% of the long-term design load duringthe earthquake. The concrete of the cast-in place piles and the foundation beams were placed about

85 and 24 days prior to the earthquake, respectively.It is conceivable that, during the earthquake, the reclaimed fill down to a depth of 15 m liquefied.

Figure 30 also shows vector displacements of the ground surface and buildings in and around thesite, which have been detected by an aerial photographic survey. The quay walls on the southernand eastern coastlines had moved seawards by 2.5 and 3 m, respectively, and the ground surfacearound the building was displaced southeast by about 0.5–1.0 m.

To characterize the damage features of the foundations, the ground was excavated up to a depthof 2 m to expose some pile heads, and the their slope angles and damage features were examined (Kuwabara and Yoneda, 1998). The vector slope angles thus determined for the exposed pile headsare indicated in Figure 33. In the E-W (longitudinal) direction, the piles on the east (sea) side of

the building inclined westwards (i.e. landwards), while the piles on the west (land) side inclined eastwards (i.e. seawards). In the N-S (span) direction, in contrast, all the piles inclined southwards(i.e. seawards), with slope angles slightly larger on the south piles than on the north piles. Besides,the slope angles in the N-S direction tend to be larger than in the E-W direction.

Vertical coring was made from the centers of Piles Y2 and Z1 to examine their failure and deformation modes, which pierced the piles at depths of 12.0 and 15.2 m, respectively. Additionalvertical coring made at a point 37.6 cm apart from the center of Pile Y2 also pierced the pile at adepth of 15.0 m. Considering that the diameters of the piles are 1.4 and 1.5 m respectively, the pilesabove 15 m depth had been displaced south-southeastwards by at least 90 cm, i.e., about 22 cm

354






eastwards and 85 cm southwards. A television camera into the bored holes of the piles as well asdirect observation of the pile heads showed that concrete cracking concentrated at depths smaller than 6 m and at depths below 12 m, as shown in Figure 31(b).

6 CONCLUSIONS

Field performance of various pile foundations that experienced soil liquefaction and lateral spread-ing in the 1995 Hyogoken-Nambu earthquake have been compiled and summarized, together withtheir failure and deformation modes identified by surveys using borehole cameras and/or slope-indicators. The field performance of pile foundations identified by the surveys has shown thefollowing:

1) In the nonliquefied area where damage to superstructures concentrated, quite a few buildingstilted with their superstructures remaining intact, due to shear failure of their pile heads.

2) In the liquefied area, many buildings supported on non-ductile piles suffered foundation distresswith their superstructures remaining intact, due to failures of the piles at the interface betweenliquefied and non-liquefied layers.

3) In the laterally spreading area, similar but more extensive foundation distress occurred in whichthe piles of a building near the waterfront often showed different failure modes in the direc-tion perpendicular to the waterfront, while those away from the waterfront showed similar deformation patterns.

Thesignificant difference in damagepatternof piles among non-liquefied, liquefied, andlaterallyspreading areas is mainly due to the following:

1) The nonliquefied surface soil amplified the ground motions significantly, leading to theextensive damage to buildings or otherwise the shear failure at the pile heads.

2) Soil liquefaction deamplified the ground motions particularly in the period range less than 1 s,reducing the damage to superstructure in the liquefied and laterally spreading areas.

3) Soil liquefaction and lateral spreading increased the ground displacement and thus kinematiceffects, leading to the damage to pile foundations.

4) Spatial variation of ground displacement in the laterally spreading area had a significant effecton the failure modes of piles within a building.

ACKNOWLEDGMENTS

The field case histories described herein was made possible through the post-earthquake field investigation and their compilation conducted by various organizations including but not limited to the Committee on Building Foundation Technology against Liquefaction and Lateral Spreading,Japan Association for Building Research Promotion; the Committees on Damage to BuildingFoundations in Hyogoken-Nambu Earthquake, both Architectural Institute of Japan and KansaiBranch of AIJ; and Japanese Association for Steel Pile Piles. Professor Fumio Kuwabara, NipponInstitute of Technology, kindly provided the information concerning the case history of damaged cast-in-place concrete piles used in this paper. The strong motionaccelerograms at the Higashi-KobeBridge station were provided by the Public Works Research Institute, Ministry of Construction.The authors express their sincere thanks to the above organizations and person.

REFERENCES

Architectural Institute of Japan 1988. Recommendations for design of building foundations, 430pp. (in

Japanese).

Architectural Institute of Japan 1990. Ultimate strength and deformation capacity of buildings in seismic

design, 715pp. (in Japanese).

Architectural Institute of Japan et al. 1998. Report on the Hanshin-Awaji Earthquake Disaster, Building Series

Volume 4, Wooden Structure and Building Foundations (in Japanese).

355





Damage investigation on the foundations of the Hanshin Expressway

Route 5 caused by the 1995 Hyogoken-Nambu earthquake

N. Hamada & F. Yasuda Hanshin Expressway Co. Ltd., Osaka, Japan

A. NakahiraCTI Engineering Co., Ltd., Osaka, Japan

T. TazohShimizu Corporation, Tokyo, Japan

ABSTRACT: The 1995 Hyogoken-Nambu earthquake caused damage to the foundations of the bridges on the Hanshin Expressway Route 5, which connects seven reclaimed islands in OsakaBay. Soil liquefaction and the lateral spreading it induced occurred widely in reclaimed landsduring the earthquake. In-ground horizontal displacements from the surface to a depth of −51.5 mdetailing both effects of soil liquefaction and the lateral spreading were successfully observed

by inclinometers which had been installed prior to the earthquake. The damage investigation onthe pile cracks was carried out on 119 pile foundations of Route 5 in total by using a bore-holecamera and nondestructive integrity sonic tests. The damage to the pile foundation of a bridge on

Route 5 caused by the 1995 earthquake is also presented in detail. The pier of the bridge moved approximately one meter laterally toward the river; the quay wall moved approximately three meterslaterally in the same direction and subsided by more than one meter.

1 DAMAGE OVERVIEW IN HANSHIN EXPRESSWAY ROUTE 5

1.1 Damage investigation

The Hanshin Expressway Route 5 (Bayshore line) connects the reclaimed lands between the citiesof Osaka and Kobe. It is an elevated highway structure from Tempozan in Osaka city to RokkoIsland in Kobe city. Route 5 has been relatively newly constructed and newer design specif icationshave been applied. This is why the damage was relatively slight in spite of the short distance fromthe epicenter of the 1995 Hyogoken-Nambu Earthquake. That said, the eastern approach span of the Nishinomiyako Ohashi bridge collapsed (Figure 1). Foundation structures of long-span bridgesclose to quay walls moved toward the waterway, which caused damage not only to the foundations,

but also to the bearings and girder end members of superstructures (Figure 2).

Figure 1. Collapse of the approach span of

Nishinomiyako Ohashi bridge.

Figure 2. Collapse of quay walls near Higashi Kobe

Ohashi bridge.

357






Table 1. Investigation methods used in the survey area.

Methods Purpose Number Unit Survey points

Nondestructive Test Damage to Pile 167 foundation

(Sonic Integrated Test)

Bore Hole Camera Crack of Pile 240 Foundation

Survey

Observation of Pile Top Crack of Pile Top 6 Foundation P134,P161,P164,P208,

P211,P216

Horizontal Loading Test Damage to Pile 1 Foundation P208(Uozakihama)

Boring Survey (SPT) Variation of 42 Hole Side and center of each

Underground reclaimed land

Excavation Survey Observation of 2 Site Minami-Ashiyahama

(Test Pit) Underground Koshienhama

Inclinometers Displacement of 2 Hole Minami-Ashiyahama

Underground

Aerial Survey Displacement of 366 Foundation All foundations of main

Foundation and access bridges

*) P134, P161, etc; pier number of the bridge

Figure 3. Location of investigation.

The relative distance between piers and the tilt angle of piers were measured prior to restorationwork. It was found that the maximum displacement of the pier toward the waterway was approxi-mately one meter. Some investigations (listed in Table 1) were carried out in order to ascertain the

damage condition of foundations, and determine the damage level by taking into consideration thecrack conditions and the residual displacement of foundations.A location surveywas conductedafter the earthquake regarding horizontal displacements of piers,

abutments and the surrounding grounds. Horizontal displacements and angles of foundations werecalculated using the results of a GPS survey and the original coordinates from the completion of construction. Ground displacements were calculated by comparing the targets identified by aerial

photos before and after the earthquake. The location of the investigation is shown in Figure 3.Figure 4 shows the soil property of the investigation sites. The thickness of each layer, the N-value,and the length of pile in Figure 4 are shown by averaged value of all the data obtained at 7 reclaimed

358





Figure 5. Residual displacement of ground and footing.

Table 2. Foundation types of Hanshin Expressway Route 5.

Direct Caisson RC piles Consecutive underground

Classification basics basics basics wall basics Total

Abutment 0 0 21 0 21

(0.0%) (0.0%) (100.0%) (0.0%) (100.0%)

Pier 0 52 280 13 345

(0.0%) (15.1 %) (81.2%) (3.8%) (100.0%)

borehole camera. The precision level of the nondestructive integrity sonic test was proved by using

test piles and comparing the results attained by using a borehole camera.In comparison with the above three methods, direct and indirect inspections are effective methods

in terms of precision and reliability, but nondestructive inspection is effective in terms of economyand concision. As the first step, the nondestructive test was carried out from several times to fortytimes at most for each foundation to select possible damaged piles. The second step, indirect test

by using a borehole camera, was carried out for the piles that previously showed relatively highturbulence of reflection waves among selected piles.

1.2 Variation of soil characteristics before and after the earthquake

Reclamation material at seven reclaimed lands from Rokko Island to Naruohama Island consistsmainly of Masa-soil or sand including gravels. Ground damage occurred in the above mentioned reclaimed lands during the 1995 earthquake. Standard penetration tests were carried out in thesereclaimed lands after the earthquake to compare with the N-value obtained before the earthquake.The results are shown in Figure 6. It is much more likely that N-values increase on the western side,and decrease on the eastern side, as the boundary between Nishinomiyahama and Koshienhama.Liquefaction was considered to be one of the reasons for the difference in the N-values before and after the earthquake. To clarify this point, the relative settlement of the ground close to piers wasmeasured after the earthquake. Figure 7 shows the settlement at the respective reclaimed lands.

360





Figure 6. Average SPT N-value of every reclaimed land before/after the earthquake.

Figure 7. Average settlement of reclaimed lands. Figure 8. Relationship between increase/decrease

of SPT N-value and settlement of reclaimed lands.

Reclaimed land situated further to the west has a greater amount of soil settlement. It is estimated

that the volume change of the fill layer occurred due to liquefaction. The relationships between theincrease and decrease of the N-values and the settlement are shown in Figure 8. A correlation isrecognized in that the larger the settlement, the greater the increase in N-value.

2 IN-GROUND DISPLACEMENTS RESULTING FROM SOIL LIQUEFACTIONAND THE LATERAL SPREADING IN MINAMI-ASHIYAHAMA ISLAND

Minami-Ashiyahama is a reclaimed island with an area of 125.6 hectares in Ashiya Port in OsakaBay, which was constructed over a period of 25 years from 1971 to 1996 (completed one year

after the 1995 earthquake). In Minami-Ashiyahama, in-ground horizontal displacements from thesurfacetoadepthof −51.5 m detailingboth effects of soil liquefactionandthe lateral spreadingweresuccessfully observed by inclinometers which had been installed before the earthquake (HanshinExpressway Public Corporation, 1996b, Nanjo et al. 1997, Tazoh et al. 2000). The observed in-ground displacement data is extremely valuable for earthquake engineering. There had been norecorded data of the actual behavior of soil liquefaction and the horizontal spreading available inthe past.

Figure 9 shows the map of the reclaimed land of Minami-Ashiyahama, where in-ground hori-zontal displacements were measured at points No. 1 and 2. The measurement points are located

361







Figure 9. The locations of point No. 1 where in-ground displacements were measured in the reclaimed island

of Minami-Ashiyahama.

near the Hanshin Expressway Route 5 which crosses Minami-Ashiyahama. A channel exists at adistance of 70 m from the points in the north direction, and a quay wall is constructed along thechannel. The in-ground displacements obtained at point No. 2 are unreliable because of corrosionand the cut of the inclinometer, and accordingly this data was not examined in this paper.

Figure 10 shows the measured in-ground horizontal displacements to a depth of −51.5 m at point No. 1. The horizontal displacements in the north-south direction (N-S) are larger than those in theeast-west direction (E-W). The surface displacement (N-S) reached 13 cm (relative displacementobtained with the displacement at the depth of −51.5 m regarded as zero). The ground around

point No. 1 moved in the opposite direction away from the channel. It should be noted that this phenomenon differs from the well-known occurrence in which the ground resulting from soilliquefaction moves seawards in waterfront areas.

The following conclusions can be drawn through comparisons between measured in-ground horizontal displacements (N-S) and the soil profile obtained from a borehole test near point No. 1.

1) Layers of Pleistocene gravel deposits (D g 1−1, D g 1−2, and D g 1−3) appear to a considerable extentin the south direction. The amplification of the displacement at the boundary between D g 1−3

and its lower layer of Pleistocene sand (D s1) becomes greater.2) The displacements in the Holocene clay (Ac) and Holocene sand (A s) layers are small compared

with those of the Pleistocene gravel layers (D g 1−

1 –D g 1−

3).3) The amplification between the f ill layers (B2−2 –B3) is extremely large. The possibility of soilliquefaction in these layers is enormous.

4) The amplification in the fill layer of B2−2 decreases, while the fill layer of B2−1 increases. Thereason for this occurrence can be considered to be a result of the incidence of soil liquefaction.

5) The amplification in the surface layer of the fill (B1) is very small.

Figure 11, which was taken three months after the earthquake, shows the damage to the quay walland to the embankment. The point and direction in which the photograph was taken are illustrated

362





Figure 10. In-ground horizontal displacements of N-S and E-W components at point No. 1 from the surface

to a depth of −51.5 m.

Figure 11. The damage to the revetment and embankment during the 1995 earthquake. (The photo was taken

three months after the earthquake.)

in Figure 9. Several ten centimeter-wide cracks can be identified at the foot of the embankmentand on the quay wall. The stability of the embankment is not threatened by the damage.

3 DAMAGE INVESTIGATION ON PILE FOUNDATIONS

3.1 Outline of the damage investigation

The damage investigation on the cracks of piles was carried out to 119 pile foundations of theHanshin Expressway Route 5 in total, which are located in the seven reclaimed islands in HyogoPrefecture from Naruohama Island inAmagasaki city to Rokko Island in Kobe city. These reclaimed

363






islands are located in a region 20 to 30 km from the epicenter of the 1995 earthquake. The crack investigation on piles was carried out by the borehole camera test and the nondestructive integritysonic test using impulse elastic waves. The measurement of the deformation of foundations and grounds was performed by a GPS survey and aerial photogrammetry (Hanshin Expressway PublicCorporation 1996a, Matsui et al. 1998, Matsui et al. 1999).

The following results are presented.

1) The displacement of the foundations and the grounds, and the cracking of the piles.2) The correlation between the pile cracks and the soil profiles at both the vicinity of the quay

walls and the inlands, which is based on the analysis on the relationship between the distancefrom the quay walls to the pile foundations and the pile cracks.

3) The differences of the effect of displacement of the grounds on the pile cracks at the vicinityof the quay walls and the inlands.

We defined the vicinity of a quay wall and an inland as a site where the distance from a quay wallto a pile foundation was less than a hundred meters and more than a hundred meters, respectively,

based on the conclusion demonstrated in paragraph 1.1.

3.2 Investigation results

The investigation results on the pile cracks are shown in Figure 12. The contour lines in the figureare drawn based on the data of the number of pile cracks for each incremental depth of 2 m. The soillayers are classified into the four layers of B layer (fill layer), A c layer (Holocene clay layer), Dsg

layer (soil deposit consisting of a Pleistocene sandy soil and clay layer), and D g layer (Pleistocenesandy gravel layer) based on the boring explorations carried out in the design. All piles are supported in the D g layer of the Pleistocene sandy gravel layer for which the N-values are more than 40–50.

The horizontal displacement of the foundation and ground is defined as the component in thedirection of the bridge axis. The positive and negative values of the displacement represent the

displacements toward to quay wall and to the inland, respectively. The distance between the quaywall and the center of the foundation is defined by the minimum distance in the direction of the

bridge axis.It is clear from Figure 12 that the cracks are concentrated at the pile-heads, and also appear at

some depths of the piles, and that the horizontal displacements of the foundations and groundsare large at the sites in the vicinity of quay walls and become small at the inland sites. From theinvestigation which used a borehole camera, it is found that most of the cracks are transverse acrossthe pile sections and that diagonal cracks are not generated. Therefore it can be inferred that bendingmoments are dominant as the critical force applied to the piles during the 1995 earthquake.

3.3 Damage to pilesThe correlation between the density of the number of the pile cracks and the distance from the quaywalls to the pile foundations for each soil layer is shown in Figure 13(a) –(c). Figure 13(d) showsthe average crack density for all soil layers in each pile. It can be seen from this figure that thecrack density is large in the vicinity of the quay walls and becomes smaller as the distance fromthe quay walls increases. The crack density at the inland sites is relatively small, but these cannot

be ignored. Regarding the crack distribution at each soil layer, the cracks are concentrated in theB layer. Then many cracks appeared in the Ac layer and few in the Dsg and Dg layers. The cracksin each layer are relatively larger in the vicinity of the quay walls than those at the inland sites. Itseems that the crack density in each soil layer has a relatively constant value with some variation

regardless of the distances from the quay walls.The distributions of the number of cracks by depth are shown in Figure 14. Figure 14(a) and (b)represent the average of the number of 14 pile foundations in the vicinity of the quay walls and thatof 105 pile foundations at the inland sites, respectively. Curves showing the total number of cracksand the number of cracks of more than 0.5 mm in width are shown in these f igures.

It can be seen from Figure 14(a) that the cracks in the vicinity of a quay wall are concentrated ata depth of 1 m from the pile-heads and also concentrated in the upper half section of the fill layer (B layer). Under the lower half section of the fill layer, the cracks are distributed at the upper and lower ranges of 4–5 meters from the soil layer boundaries between layers B and Ac and between Ac

364




Figure 12. Damage investigation of ground and foundations in seven reclaimed islands.

3 6

5





Figure 14. Average crack numbers in piles by depth.

Figure 15. Relationship between residual dis-

placement of ground and crack density of pile.

Figure 16. Relationship between PL-value and

residual displacement of ground.

4 DAMAGE TO THE PILE FOUNDATION OF A HIGHWAY BRIDGE CAUSEDBY SOIL LIQUEFACTION AND ITS LATERAL SPREADING

4.1 Outline of the bridge and the damage

The damaged bridge on the Hanshin Expressway Route 5 shown in Figure 17 has three spans of Gerber-steel-box girders 310 meters in length which cross a river running between the artificialislands of Nishinomiyahama and Minami-Ashiyahama. The bridge was completed in 1993, two

367







Figure 17. The damage to a bridge on the Hanshin Expressway Route 5 caused by the 1995 earthquake.

Figure 18. Plan view of the pier and piles of the damaged bridge.

years before the 1995 earthquake. The rigid-framed steel pier (pier No. 134 shown in Figure 17(a))on Minami-Ashiyahama Island is supported by 56 piles (4 rows by 14 columns) as shown inFigure 18, which are cast-in–place concrete piles, 34 meters in length and 1.5 meters in diameter.

Reclamation of the Minami-Ashiyahama was started in 1971 and reclamation in the proximity of the expressway was completed in around 1990. The soil used for the reclamation was decomposed granite soil excavated in Awaji Island. The soil deposit consists of a fill layer (depth of 0–14 m)

368






Figure 19. The residual deformations of the pier and the revetment and the damage to the piles.

mainly composed of gravel, a Holocene clay soil layer (14–20 m), a Holocene sandy soil layer (20–28 m), and a Pleistocene sandy soil layer (under 28 m). The soil profile of the ground can beseen in Figure 19(b). The ends of the piles are embedded in the Pleistocene sandy soil layer.

The samples of sandy soil were gathered by block soil sampling and freezing soil sampling.Block soil sampling and freezing soil sampling are the best methods for sampling sandy soil. Theresults of laboratory tests on the samples gathered by these methods were used for analyses withoutany correction.

369





Figure 19(a) illustrates the residual deformations of the pier and the quay wall caused by the1995earthquake. The top of the pier moved approximately one meter horizontally toward the river. Thequay wall moved about three meters horizontally in the same direction and subsided by more thanone meter. Figure 17(b), taken two weeks after the earthquake, shows the ground failure around the

pier. The occurrence of soil liquefaction around the bridge was identified by evidence of large sand boil flow across the entire surface of the surrounding ground. Severe damage to the girder-shoes

on the pier on Minami-Ashiyahama Island were found; however, there was no structural damage tothe pier itself. The pier located on Nishinomiyahama Island moved 0.6 meters horizontally toward the river (Hanshin Expressway Public Corporation et al. 1996, Nanjo et al. 2000).

4.2 Damage to the piles

Three investigation methods were adopted to survey the damage to the piles of the pier, namelydirect inspection by means of excavation around the pile-heads, nondestructive integrity sonic testutilizing impulse elastic waves, and indirect inspection utilizing a borehole camera. Figure 19(b)shows the result of the investigation using the borehole camera.

Inspection of the first borehole (borehole A) was halted at the depth of 21 m due to contact with areinforcing bar. The second borehole (borehole B) was located at a distance of 40 cm from boreholeA. The condition of the cracks in boreholes A and B was different. Many cracks appeared betweenthe pile-head and the depth of 12 m in borehole A. On the other hand, cracks at the depths of 10 m,13–16 m, 20–23 m, and 35 m were observed in borehole B. The difference in the occurrence of thecracks between boreholes A and B represents the fact that the pile was remarkably deformed inone direction. The deformation of the pile shown in Figure 19(b) was obtained by evaluating theconditions of the opening or closing of the cracks. The field investigation identified one meter of horizontal displacement toward the river at the pile-head.

5 SUMMARY

The main conclusions of our study are:

1) Measured displacements showed that the quay walls have almost no effect on the foundation at thesites longer than 100 meters away from the quay walls on all of the reclaimed lands. In comparisonwith the three methods used to investigate pile damage, direct and indirect inspections areeffective methods in terms of precision and reliability, but nondestructive inspection is effectivein terms of economy and concision. It is found that the fill layer was more compacted at themore settled land together with an increase in the N-value.

2) The amplification between the fill layers (B2−2 –B3) is extremely large. The possibility of soil

liquefaction in these layers is enormous. The amplification in the fill layer of B 2−2 decreases,while the fill layer of B2−1 increases. The reason for this occurrence can be considered to be aresult of the incidence of soil liquefaction. The amplification in the surface layer of the fill (B1)is very small.

3) The pile damage is relatively severe in the vicinity of the quay walls, as is the ground displace-ment. The crack density of the piles at the inland sites is concentrated at the pile-heads and cracks in the vicinity of the quay walls appear not only at the pile-heads but also at the boundaryof the soil layers between the liquefied fill layer and the Holocene clay layer.

4) The pier of the bridge supported by 56 piles on the Hanshin Expressway Route 5 moved approx-imately one meter laterally toward the river; the quay wall moved approximately three meterslaterally in the same direction and subsided by more than one meter.

REFERENCES

Hanshin Expressway Public Corporation, Hanshin Expressway Engineering Control Center, 1996a. Damage

Investigation on Foundations of Bridges in Reclaimed Lands (in Japanese): Hanshin Expressway Public

Corporation.

Hanshin Expressway Public Corporation, 1996b. Investigation on the Seismic Damage to Bridge Foundations

in Reclaimed Land (in Japanese): Hanshin Expressway Public Corporation.

370




Iwasaki, T., Tatsuoka, F., Tokita, K., Yasuda, S., 1980. Estimation of Degree of Soil Liquefaction During

Earthquakes, Tsuchi-to-Kiso, Vol. 28, No. 4, 23–29 (in Japanese): The Japanese Geotechnical Society.

Japan Road Association, 1996. Specifications for Highway Bridges, Part V Earthquake Resistant Design (in

Japanese): Japan Road Association.

Matsui, T., Nanjo, A., Yasuda, F., Nakada, Y., Imada, Y., 1998. Analytical Method and the Applicability of

Damage Investigation on Piles by Nondestructive Integrity Sonic Test, Proceedings of Japan Society of

Civil Engineers, No. 596, 261–270 (in Japanese).Matsui, T., Nanjo, A., Yasuda, F., Nakahira, A., Kuroda, C., 1999. Cause of Seismic Damage to Piles of Road

Bridges in Seashore Reclaimed Land, Proceedings of Japan Society of Civil Engineers, No. 638, 259–271,

(in Japanese).

Nanjo, A., Yasuda, F., Kubota, K., Hisaki, E., 1997. Investigation on In-ground displacements before and

after the Hyogo-ken Nambu Earthquake, Proceedings of Japan Society of Civil Engineers, 380–381 (in

Japanese).

Nanjo, A., Yasuda, F., Fujii, F., Tazoh, T., Ohtsuki, A., Fuchimoto, M., Nakahira, A., Kuroda, C., 2000.

Analysis of the Damage to the Pile Foundation of Highway Bridge Caused by Soil Liquefaction and Its

Lateral Spread due to the 1995 Great Hanshin Earthquake, Proceedings of Japan Society of Civil Engineers,

No. 661/I-53, 195–210 (in Japanese).

Tazoh, T., Ohtsuki, A., Fuchimoto, M., Kishida, S., Nanjo, A., Yasuda, F., Kosa, K., Fujii, F., Tamba, Y., Nakahira, A., Kuroda, C., 1998. Residual Displacements in Liquefied Soil Deposits, Proceedings of the

Eleven European Conference on Earthquake Engineering, 721–730.

371





Damage to subway station during the 1995 Hyogoken-Nambu

(Kobe) earthquake

N. YoshidaTohoku Gakuin University, Tagajo, Japan

ABSTRACT: A detailed reconnaissance survey was made at the Daikai subway station, whichwas the first subway structure that had completely collapsed due to earthquakes. The station isa box frame structure with a column at the center, measuring 17 m wide and 7.17 m high in the

exterior dimensions and 120 m long. A complete collapse occurred at more than half of the center columns, which resulted in the failure and collapse of the ceiling slab and subsidence of subsoilover the station by more than 2.5 m at maximum. Many diagonal cracks were also observed on thewalls in the transverse direction. Judging from the damage patterns, a strong horizontal force wasimposed on the structure from the surrounding subsoil. Evidence of damage due to vertical forcewas not clearly observed. Finally, characteristics to be used in the future analysis in order to makethe mechanism of collapse clear are shown through the numerical analyses.

1 INTRODUCTION

The Hyogoken-nambu (Kobe) earthquake of January 17, 1995 caused severe damage to variousstructures. Among these, the damage to the subway was one of the amazing events, becauseunderground structures had been considered to be relatively safe from earthquake effects compared to the structures above the ground. During the Kobe earthquake, however, damage to subwayoccurred at many locations as shown in Figure 1. Damage to the Daikai station was the most severe;more than 30 columns completely collapsed, and ceiling slabs deformed extensively, resulting inabout 2.5 m subsidence of the national road running above the subway.

Damage to underground line-shaped structures with small dimensions have been reported during past earthquakes (e.g., 1964 Niigata earthquake (Hamada, 1992), 1978 Miyagiken-oki earthquake(Tohoku Branch of JSCE investigation team, 1980), 1983 Nihonkai-chubu earthquake (JSCE

Figure 1. Damaged subways and damage patterns.

373





investigation team, 1986), and 1993 Kushiro-oki earthquake (JSSMFE investigation team, 1994)).Damage to these structures was caused mainly by the large ground deformation due to liquefactionor differential ground movement near the interface between hard and soft grounds. However, reporton damage to large-scaled underground structure is very few. Separation of construction joints onside walls at the intersection between the underground and ground surface sections in soft ground was reported at a subway during the 1985 Mexico earthquake (Kawashima, 1994). Differential

movement at a flexible joint between a ventilation tower and tunnel, and water leak near the ven-tilation tower were reported during the 1989 Loma Prieta earthquake, USA (ADEP (Associationfor Development of Earthquake Prediction) investigation team, 1990). This damage was, however,very light.

The Daikai station was the first subway structure that had been completely collapsed during anearthquake. As shown in Figure 1, damage to the subway system was reported not only at thisstation but also at many other locations although the degree of damage was less than that in theDaikai station. In this paper, we intend to describe the damage in detail as it was, and to make thecause of the failure clear through in-situ investigation as well as numerical analyses.

2 DESIGN AND CONSTRUCTION OF THE DAIKAI STATION

The Daikai station is located on the Tohzai line, a subway system operating to the west fromdowntown Kobe city. Near this station, the subway runs right under the national road No. 28. Asshown in Figure 1, the Japan Meteorological Agency seismic intensity in this area was evaluated to be 7, which is nearly equivalent to MM seismic intensity scale of 10.

Since the longitudinal direction of the subway is at an angle of about 40 degrees from the east-west direction, it is not convenient to use bearing to point out the direction. Instead, as shown inFigure 2, we will use “mountain side” and “sea side” to point out the transverse direction, and “Shinkaichi side” and “Nagata side” for the longitudinal direction, where Shinkaichi and Nagata

are names of subway stations next to the Daikai station.Construction of the Daikai station was begun in August 1962 using the cut-and-cover method,and it was completed on January 31, 1964. Soldier beam and sheathing boards were employed toexcavate below the ground surface to a depth of about 12 m.

Figures 2 and 3 show plan and cross-section of the station. The station is a two-story reinforced concrete underground structure; B2 floor consists of platforms and rail lines and the B1 floor is aconcourse with a ticket barrier. The thickness of the overburden soil is about 4.8 m at section 1-1and 1.9 m at section 2-2.

The main part of the B2 floor is a box type frame structure with columns at the center, measuring17 m wide and 7.17 m high in the outside dimension, and it is 120 m long in the longitudinal

Figure 2. Plan of Daikai station.

374





Figure 3. Typical section of Daikai Station.

Figure 4. Load distribution considered during the

design.

Figure 5. Reinforcing steel arrangement of the

center column.

direction. There are 35 center columns in total along the longitudinal direction. The center columnis 3.82 m high and has a cross-section of 0.4 m×1.0 m, and the distance between columns is 3.5 m.These columns are supported by the upper beam with 1.6 m deep and the lower beams with 1.75 mdeep. Thicknesses of the ceiling and base slabs are 0.80 and 0.85 m, respectively, and the thicknessof the side walls is 0.7 m above the platform and 0.85 m below the platform. There are utility rooms(electric facility room and switching station room) under the concourse, therefore the walls areheavily loaded in this region.

The frame was designed based on a consideration of the weight of the overburden soil, lateralearth pressure, and weight of the frame under ordinary loading conditions as shown in Figure 4,

but the earthquake load was not taken into account, with the normal method being used at that

time of the design. Under these loads, a center column is subjected to N = 4410 kN (no shear force nor bending moment works at the center column because the structure and external loads aresymmetric); top of the lateral wall is subjected to M =191 kNm, N = 444.3 kN, and Q=97.1 kN,and the bottom of the lateral wall is subjected to M = 463.5 kN, N =517,4 kN, and Q=211.5 kN

per unit length in the longitudinal direction, where M , N and Q denotes bending moment, axialforce and shear force, respectively, and the so called lateral wall is the outermost wall of theunderground station. The design strength of the concrete was 23520 kN/m2 for the center columnand 20580 kN/m2 for the other structural components, and design the yield stress of the reinforcingsteel bar was 23.52 kN/cm2. Round steel bars with diameters from 16 to 25 mm were used to thewalls and slabs, and 32 mm diameter bars were used in the center column. A transverse hoop with

9mm diameter was placed at every 350 mm in the center column. Figure 5 shows the reinforcingsteel arrangement of the center column. Since the cross-sectional area of the center column was predetermined and was small, high design strength concrete was used and many reinforcing steel bars were placed in order to sustain design axial load. The allowable axial force (one-third of designstrength) is 4439 kN, which is slightly larger than the design axial force of 4410 kN.

Two tests were made after the earthquake in order to evaluate the strength of the concrete. Strengthof 37240 kN/m2 with standard deviation of 2646 kN/m2 was obtained by Schmidt hammer tests.Average compression strength of the 8 cylindrical specimens taken from the center column was39690 kN/m2.

375







Figure 6. Borehole investigation before the earthquake.

3 GEOMORPHOLOGICAL AND GEOLOGICAL SETTING AT DAIKAI STATION

The Daikai station is located west of Kobe port, and is about 15 km northeast of the epicenter of the Kobe earthquake. It is geomorphologically located in a low-lying area with a ground surfaceelevation of about 5 m, which extends along the southern toe of Rokko Mountain.

Rokko Mountain, with an elevation of about 400 m in the western part near the Daikai station,is composed of granite formed during the Cretaceous period. Many active faults run along thesouthern piedmont line of the mountains, one of which runs immediately north of the Daikai

station.The geology of the site is essentially soft sandy silt of back marsh origin overlying medium to

coarse granitic sand containing gravel, which, in turn, overlies soft Holocene marine clay and densePleistocene gravel.

A soil investigation was made in 1959 prior to the construction of the subway. Figure 6 shows boring logs obtained from tests on the west and east sides of the station. The soil investigationwas also made after the earthquake in February, 1995, which included PS logging. The result of this investigation is shown in Figure 7 along with soil profiles. The depth of the water table was

between 6 and 8 meters after the earthquake, which is lower than that before the earthquake (about3 m; see Figure 6). Referring to another source (Kobe city, 1980), generally the depth of base (SPT

N -value> 50) was deep on the west side and consists of silty or clayey surface soil. The depthof base becomes shallow toward east; that at the Daikai station site is about 15 m and that at theShinkaichi station is less than 10 m. In addition, sand becomes more predominant toward the east.

In addition, standard penetration and cone penetration tests were made near the structure inorder to determine the properties of the fill material. Decomposed granite soil was used as the fillmaterial. Figure 8 compares the SPT N -values with N -value computed from the cone penetrationtests. The N -value of the fill is about 10 at all the depths except near the bottom. Figure 9 showsthe grain size distribution and Figure 10 shows strain dependent characteristics of undisturbed samples of the fill material obtained from a dynamic deformation characteristics test. The test was

376





Figure 7. Settlement of ground surface and soil profile. Boring No. 4 is fill whose SPT- N value is shown in

Figure 8.

Figure 8. SPT N -value of fill materials.

made under isotropic initial stress states (IC) with confining pressures of 49 and 98 kN/m2, and an

anisotropic initial stress state (AC) with principal stresses of 49 and 29.4 kN/m2.

4 DAMAGE DUE TO EARTHQUAKE

4.1 General features of the damage

Figure 11 shows a schematic diagram of the damage in the longitudinal direction. Since the finishwith concrete block and tile were made on the interior surface of the wall, they had to be removed to

377






Figure 12. Schematic figure showing the damage patterns in the transverse direction. Numbers in (a) to (f)

are clear height in m measured after the damage, and numbers in (g) to (f) denote crack width in mm.

Figure 12 shows a schematic diagram of the damage in the transverse direction at severalcross-sections.

The general features of the damage were that the center column completely collapsed on the

Nagata side of the station resulting in the settlement of the ceiling slab as well as the overburdensoils above the station. Many cracks were observed in the longitudinal direction on the walls aswell as in the transverse direction.

A train just passed the Daikai station at the time of the earthquake, but was not scheduled tostop at the station. The motor man stopped the train after feeling the earthquake and escaped to theground surface with passengers through the Daikai station. Judging from the interview with thisman, it seems that the center column had already collapsed when they passed the station. Vehicletransportation on the national road No. 28, which runs over the station, however, was possible duringthe day the earthquake occurred. The ground with 90 m long and 23 m wide gradually settled downto more than 2.5 m at maximum. The contour lines of the settlement of the road surface are shownin Figure 7, which was measured on January 28. Ground settlement probably occurred beneath the

pavement slab at the time of the earthquake, but the pavement slab sustained its own weight dueto its rigidity. Figure 13 shows settlement of the ground surface. Evidence of liquefaction such assand boil was not observed near the station.

4.2 Center column at B2 floor

Referring to Figure 2, the station can be divided into 3 zones in the longitudinal direction alongthe station depending on the structural system: zones composed of column 1-23, column 24-29,

379





Figure 13. National road No. 28 running above the Daikai station. The 5 story building on the left is the

municipal Daikai apartment house.

Figure 14. View of the Daikai station from the

Nagata side. Center column collapsed completely and the ceiling slab subsided.

Figure 15. View of the Daikai station from the

Shinkaichi side. Damage was relatively less compared with the Nagata side.

Figure 16. Damage to center column No. 10 where

damage was the severest.

Figure 17. Damage to center column No. 20.

and column 30-35, which are designated as zones A, B and C, respectively, hereafter (see Figure11). Zones A and C are a one story box frame structure whereas zone B has utility rooms adjacentto the platform as well as the B2 floor (concourse).

Damage was the most severe at zone A, in the Nagata side. Almost all of the center columnscompletely collapsed and the ceiling slab fell down. As a result, the original box frame structuredistorted to an M-shaped section as shown in Figure 12. Typical damage to the center column isshown in Figure 14 to Figure 18.

380









Figure 18. Damage to center column No. 23. Figure 19. Damage to center column No. 34. Onlyevery other reinforcing bar buckled indicating that

the zigzag arrangement of the reinforcing steel (See

Figure 5) worked well to retain the core concrete.

There were two types of collapse pattern in the center column in zone A; collapsed portion bendsinto one direction as shown in Figure 16 and Figure 17, and Figure 12(b)(c), and in two directionsin a symmetrical manner as shown in Figure 18 and Figure 12(a)(d). The collapsed portion was

pushed toward the mountain side in columns 7, 11, 12, 16 and 19, and toward the sea in columns

2-6, 9, 10, 20, and 21, whereas the collapsed portion was pushed out into both directions in columns8, 13, 14, 17, 18, and 22-24.In zone B, as shown in Figures 11(d) and 12(d), the collapse of the column occurred in the upper

portion and reinforcing steel buckled into a symmetrical shape in columns 24 and 25. The upper longitudinal beam connecting the center column was bent at a point between columns 25 and 26.The small separation of the corner concrete of the center column was observed at the mountainside of upper portion and at the sea side of lower portion, in columns 26, 27 and 28.

Although the structural system in zone C was the same as that for zone A, damage was less inzone C compared with that in zoneA. Figure 12(f) shows the damage to column 31 where the lower

part of the center column is damaged and ceiling slab settled about 5 cm.Figure 19 shows a close-up of the damaged portion of column 34. It was noted that every other

bar buckled. As shown in Figure 5(a), buckling of the reinforcing steel was constrained by thehoop steel as well as by the bars placed in a zigzag manner connecting the reinforcing steel onthe opposite sides of the column. Damage shown in Figure 19 indicates that this additional zigzagdistribution of reinforcing steel works well in confining the core concrete.

The clear height between the upper and lower beams was measured at a point between the center columns, which is shown as h in Figure 11(d). The minimum clear height is 1.15 m. Compared withthe design clear height of 4.02 m, settlement of the ceiling slab is about 2.9 m maximum, which isabout the same as the settlement of the ground surface shown in Figure 6.

Cracks were observed in the upper beam between columns 32 and 33 supposedly caused by punching shear on the column.

4.3 Railroad level and base slab

The level of the railroad is measured every 5 meters. The differences of levels between the adjacent points are from 8 to 28 mm, therefore the gradient at each interval is between 1.6 and 5.6%. Themeasured gradient value is scattered around the design value of 3%. The scatter was larger near the

boundary of the station. It is not known whether this measured scatter existed before the earthquakeor was caused by the earthquake.

381






Cracks were not observed on the base slab, and water leakage was also not observed throughthe base slab. Water leakage was, however, found at some of the construction joints.

4.4 Ceiling slab

In zoneA, the ceiling slab kinks andcracks150–250 mm wide, appeared in thelongitudinaldirectionabout 2.15 to 2.40 m from the center line of the column as shown in Figure 11(a), Figure 14 and Figure 20. This portion was found to coincide with the portion where sagging reinforcing steel barsin the ceiling slab were bent upward so as to use them efficiently under the action of load shown inFigure 4. In addition, cracks in the transverse direction were noted at nearly equal distances on theceiling slab, the widths of which extended up to 70 mm. Almost all of them were located along theedge of the center column. Cracks were located at about 45 degrees in the longitudinal directionnear the end on the Nagata side. In addition, the separation of cover concrete was observed over almost the entire area near the haunch and the intersection between the lateral wall and ceiling slab.

Many small cracks in the diagonal direction were observed on the ceiling slab near the boundary between zones A and B. Cracks in the diagonal direction were also observed at the Shinkaichi side

of zone B.Only a few cracks were observed in the longitudinal direction in zone C.

4.5 Lateral wall

The damage pattern in the lateral walls is shown in Figure 11(b)(c)(e)(f). Since a concrete block masonry wall finished with tile was constructed at the inside of the wall, the damage to the lateralwall cannot be seen from inside. These finisheswere removed except in the portionwhere designated as “region not investigated” in the figure. Although the damage patterns are similar in both themountain and sea side walls, it was a little more severe in the sea side wall. In this wall, separation of

cover concrete was observed both near the top and bottom haunches, and vertical cracks 0.1–1 mmwide, run from bottom to top. In addition, diagonal cracks probably caused by large shear stresswere observed at the Nagata side end of zone A.

According to the investigation of the exterior surface as shown in Figure 11(b)(f), wide cracksin the longitudinal direction were observed along with the intersection with the haunch. Therefore,as emphasized in Figure 12(a), as a result of the collapse of the center column in zone A, a kink occurred at the upper part of the lateral wall and the intermediate part of the ceiling slab, but notat the end of the ceiling slab.

Under the platform, a significant separation of cover concrete was observed on the mountainside lateral wall for a length of about 40 m; 30 m to 72.5 m from the end on the Nagata side, and this separation of cover concrete was seen on the sea side lateral wall for a length of about 30 m;

35 m to 65 m from the end on the Nagata side. Water leakage was also observed at these separated locations sufficient enough to dampen the area and sometimes occasionally actual flow of water could be seen. Cracks were also observed on the columns that support the platform. In general, thecrack width was wider on the railroad side than in the wall side, which indicates that the platformwas pushed toward inside.

Crack patterns were found to be similar in the lateral wall under the platform. Looking from the Nagata side, three vertical cracks were observed within the first 10 m. From that, cracks were notobserved for about 60 m long. Vertical cracks at equal distances were observed between 60 m and 90 m. It was concluded that these cracks were originally caused by thermal effects during concretesolidification process and the crack width was enlarged due to the earthquake.

After the collapse of the center column and ceiling slab, the lateral wall tilted toward inside.The relative movement of the bottom of the upper haunch from the platform was 2.7 cm on the seaside wall and 1.5 cm on the mountain side wall at about 10 m from the end of the Nagata side. Atabout 30 m distance from the end at the Nagata side where the settlement of the ceiling slab wasthe most severe, both the seaside and mountain side walls tilted by about 9.0 cm and about 5.9 cm,respectively. In addition, at a location 50 m from the Nagata side, the sea side and mountain sidewalls tilted bout 1.6 cm and 4.5 cm, respectively. The tilt of the lateral wall was hardly observed inzone C.

Deformation of the lateral wall and the slab were hardly observed in zone B.

382




Figure 20. Damage to ceiling slab; view toward Nagata side from column No. 18.

Figure 21. Damage to transverse wall at the electricfacility room.

Figure 22. Damage to transverse wall in the sea side

at the end of Nagata side.Figure 23. Diagonal cracks appeared on the lat-

eral wall on the mountain side at the end of the

Nagata side.

In zone C, diagonal cracks were observed on the lateral wall near the end of the Shinkaichi side.Apart from the end, there were a few vertical cracks. They would have appeared soon after the

construction due to thermal effects, and the width was enlarged due to the earthquake.

4.6 Transverse wall

There are several walls in the transverse direction: both ends, utility rooms, etc. The damage patterns of these walls are shown in Figure 12(g) –(j) and Figure 23 to Figure 21. Diagonal crackswere observed on all the walls.

4.7 Electric facility room and switching station room

These rooms were constructed on B2 floor adjacent to the platform and under the concourse. Theyare surrounded by concrete walls. Many diagonal cracks were observed in the walls in the transversedirection as shown in Figure 12(i). Very little damage to the lateral wall was observed.

4.8 Ventilation tower

There was a ventilation tower at about 70 m south-east from the Daikai station. Figure 24(a) showsthe location and details of the tower, and Figure 24(b) shows detailed reinforcement. As shown inFigure 24(c), there found distortion of about 4 cm at about 4 m from the ground surface.

383








Figure 24. Ventilation tower.

This indicates that the relative displacement of the ground surface from GL-4m was larger than4 cm. This displacement is very important because the analysis clarifying the mechanism of thedamage must explain it at the same time. It indicates that this distortion can be used to evaluate theaccuracy of the analysis.

5 MECHANISM OF COLLAPSE BASED ON DAMAGE PATTERN

Damage of the center column occurred to various degrees from complete collapse to slight damagein the Daikai station. It is very difficult to clearly define the mechanism of the damage to the Daikaistation by only observing the completely collapsed columns, because the effect of gravity (axialload) becomes predominant in the residual deformation pattern. Therefore, it is better to observecolumns that were only slightly damaged.

For column 28, which is one of the slightly damaged columns, a small separation of the columncover concrete was observed at two locations: sea side (bottom) and mountain side (top) of the col-umn. This indicates that the horizontal force acting in the transverse direction toward the mountainside caused the damage to the column.

The mechanism becomes clearer by focusing the investigation on the slightly damaged column

such as columns 22–26 and 29–34. The predominant damage occurred either at the bottom or atthe top of the column. This damage seems to have been caused by the bending moment; bendingmoment around the longitudinal axis seems to be predominant whereas that in the transversedirection is much less effective. This also indicates that the horizontal force acting in the transversedirection caused damage to the column.

Based on the observation of these columns, the mechanism of the damage of the collapsed columns in zone A is explained as follows: 1) Due to the strong horizontal force, the member reaches its strength under the combination of bending moment and shear force acting near theend of the column, which resulted in the collapse of the end of the column. 2) The load carrying

384





capacity of the box frame was reduced, and therefore the excess relative horizontal displacementoccurred. Once the horizontal displacement becomes large, additional moment due to axial load (so called P - effect) increases greatly resulting in the complete collapse of the center column.

This mechanism can explain most of the damage which occurred at the Daikai station. As seenin Figure 11, damage was the most severe in zone A. This can be explained by the fact that therewere more walls in the transverse direction in zones B and C than in zone A. Actually, as shown in

Figure 12(h)(i), many diagonal cracks were observed on these walls, which indicates that a largehorizontal force also acted in zones B and C. Since the transverse wall carried much of the load,damage to the center column was small in this zone. The damage which occurred in the municipalDaikai apartment house and ventilation tower also suggests the existence of a strong horizontalforce.

Horizontal force must act not only in the transverse direction but also in the longitudinal direction.Damage caused by the horizontal load in the longitudinal direction, however, is rarely observed.This can be accepted by considering that the total wall dimension is larger in the longitudinal thanin the transverse direction. Evidence of the horizontal force may be seen in the diagonal cracks onthe lateral wall at both ends of the station.

The source of the horizontal force is the next question. Earth pressure may be one of the factors,however does not seem to be the predominant factor, because it works symmetrically. Responseof the frame may not be a specific cause of damage because all the structural elements except thecenter column come into contact with soil, and therefore do not behave independent of the resultingmovement of the subsoil. The most probable is a relative displacement between the base and ceilinglevels caused by the subsoil movement. As a result of the response of the subsoil profile due to theearthquake, the relative displacement between the ceiling and base levels appears and results in ahorizontal force acing on the subway structure. This type of displacement may be negligible in thecase of a small structure, but it can be very effective in a large structure such as the Daikai stationand/or the nonlinear behavior of the subsoil profile becomes predominant.

In addition, the difference in the thickness of the overburden soil may affect the extent of damage

between zones A and B. Since the thickness of the overburden soil is larger in zone A (4.8 m) thanin zone B (1.9 m), the inertia force acting on the ceiling slab in zone A may be larger than that inzone B.

At present, the evaluation of the load transferred from the overburden soil to the ceiling slab isdifficult. It is clear that not all the inertia force of the overburden soil was transferred to the ceilingslab, however, the research on the evaluation of the soil block which provides the inertia force tothe subway has not yet been undertaken.

Evidence that vertical force affects the collapse of a station during an earthquake has not beendefinitely confirmed.

6 SUMMARY OF NUMERICAL ANALYSES

In general there may be several explanations on the mechanism by which the structure is collapsed.In such a case, it is important to examine whether the same mechanism or approach can explainsurvived or slightly damaged structures nearby, or to predict observed behavior related to thedamage of the structure. In the case of the Daikai subway station, there are at least two good examples to ensure that the assumed mechanism is relevant. The one is the damage to the Nagatastation and the other is distortion of the ventilation tower.

The Nagata station is located one station west to the Daikai station. This station was also damaged during the Kobe earthquake, but it was not so severe compared with the Daikai station; about 1/3 of

the center column were damaged. The structural design was almost the same with each other and theground condition was also similar. Therefore, analysis on this station may be a good milestone toexamine the validity of the analysis. In other words, the analysis that explains the cause of collapseof the Daikai station must also obtain successful result on an analysis on the Nagata station.

As described in the preceding, the permanent displacement of about 4 cm was observed close tothe Daikai station. This is also a good milestone to show the validity of the analysis.

The input earthquake is also an important factor to explain the mechanism of damage. It is a rarecase that the earthquake motion is recorded at the site under investigation; the earthquake recordsobtained apart from the investigated site, therefore, must be used.

385




Figure 25. FEM mesh and material property of soil for the Daikai station.

Figure 26. Finite element mesh and material property used in the analysis at Nagata Station.

In the dynamic response analysis made by the authors (Nakamura et al, 1996; Yoshida and

Nakamura, 1996; Yoshida et al., 1996.), we made effort to respond to these questions. In thischapter, material properties used in these analyses are introduced to assist in the future analysestogether with the summary of numerical analysis.

The finite element mesh used in the nonlinear earthquake response analyses for the Daikai and Nagata stations are shown in Figures 25 and 26, whereγ denotes unit weight, V s denotes shear wavevelocity and ν denotes Poisson’s ratio. Material properties used in the analyses are written in thefigures. The ground at the Daikai station is modeled into level ground considering that the boreholedata at D-1 and B-3 (mountain and sea sides of the station) are nearly identical (see Figure 7). Thefill surrounding the station is also considered. The station is modeled into a box-type frame with a

386






Figure 27. Strain dependent characteristics of soil.

Figure 28. Waves applied at the base of the model.

column at the center and a rigid region at the beam-to column intersection. Shear modulus of theframe is set 12.7 MPa/m2, Poisson’s ratio is 0.2, and unit weight is 24 kN/m3.

Strain dependent shear modulus and damping characteristics used in the analysis is shown inFigure 27, where result of the dynamic deformation characteristics test is used for the fill materialand the empirical equation proposed by Yasuda and Yamaguchi (1985) is employed to evaluate thenonlinear characteristics of the soil other than the fill. Layer numbers are counted from the upper layer in Figure 25.

Two earthquake motions, recorded during the earthquake, are employed in the analysis. Theone is the record at the Kobe University, which is treated as outcrop wave at the base becausethe site is located on the rock. The other is the record at Port Island. The multiple reflection,nonlinear, one-dimensional earthquake response analysis is employed to separate incident and reflected waves at GL-83 m, and the incident wave is multiplied by two in order to obtain outcropmotion. Both the horizontal component in the transverse direction and the up-down component areapplied simultaneously. It is noted that the Kobe University station is located a little farther fromthe epicenter than the Daikai station, whereas Port Island is closer than the station. However, theground shaking at the depth of GL-83 m is obviously smaller than that at the ground surface.

The nonlinear moment-curvature relationship is modeled into a tri-linear model by considering

elastic-perfectly plastic behavior for the reinforcing bar and e-function for the concrete. The yield stress of the reinforcing bar is 31.2 kN/m2, which is larger than the nominal value because both theeffect of strain hardening and scattering of the yield stress are considered. Compressive strengthof the concrete is set 38 kN/m2, which is an average value obtained by the Schmidt hammer testof the remaining structure. The nonlinear analysis indicates that the center column is collapsed byshear at first. Then, moment at the exterior wall and slab reaches yielding of the reinforcing bar.However, they did not reach the collapse or ultimate bending moment.

Relative displacement of the ground between the depths of both slabs is 2.7 cm under the Kobeuniversity wave. This displacement is sufficient to cause damage to the Daikai station, but it seems

387






Figure 29. Horizontal load versus relative displacement relationship.

to be a little smaller than the displacement occurred at the ventilation tower. It may be reasonable, because Kobe University is located a little farther from the epicenter than the Daikai station. Under the ground shaking of Port Island wave, the relative displacement increases to 3.64 cm. Therefore,the Port Island wave is more preferable in the analysis of the Daikai station.

Horizontal load versus relative displacement relationship is shown in Figure 29. By comparingwith the Daikai station, the load carrying capacity is larger at the Nagata station and the relativedisplacement at the failure is smaller compared with those at the Daikai station. This seems toindicate that Nagata station is easier to collapse than the Daikai station. However, the actual

phenomena have inverse tendency. This can be explained by comparing the ground displacementnear the station. At the Daikai station, the relative displacement at failure by the static analysis is2.85 cm, whereas that by the dynamic analysis is 3.64 cm, which is much larger than the failure

displacement. On the other hand, the relative displacement at failure obtained by the static analysisis 2.55 cm at the Nagata station and that by the dynamic analysis is 2.7 cm. They are nearly thesame order although the displacement by the dynamic analysis is a little larger. This error may becaused since contribution of transverse wall is not taken into account, and because of the error inevaluating the material property of soil.


The Daikai station was the first subway structure that had completely collapsed due to the earth-

quake. A detailed reconnaissance survey of the damage was made in order to determine the behavior of the station during the earthquake. Based on the study of the damage pattern, the mechanism of the collapse of the station is considered to be as follows:

The B2 floor of the station was subjected to a strong horizontal load, which caused deformationof the box frame structure. In zone A where the amount of wall in the transverse direction is small,the center column initially collapsed due to a combination of bending and shear resulting fromthe deformation of the box frame. Then, as a result of the relative displacement between the topand bottom of the column, additional moment induced by gravity of the overburden soil became

predominant, resulting in the failure of the column. Since the walls in the transverse direction carrymost of the horizontal force in zones B and C, damage to the column was much smaller compared

with that in zone A. Instead, many diagonal cracks appeared on the walls in the transverse direction,such as walls at both ends of the station and walls in the utility rooms. In addition, less overburdensoil is present in zone B because of the existence of the B1 floor (concourse), which reduced theinertia force transferred from the soil to the ceiling slab.

Mechanism of the failure of the Daikai subway station is explained by the nonlinear analysis.It is confirmed that the center column collapsed by the combination of bending and shear, but thelateral wall and slabs did not completely fail although the tensile reinforcing bar yielded. Accuracyor effectiveness of the method is examined by two approaches, i.e., to explain the distortion of the ventilation tower and to compare damage at the Nagata station, in addition to the damage

388





investigation. Judging from the permanent displacement at the ventilation tower near the Daikaistation, the earthquake input to the station should be much larger than the record at the Kobeuniversity and close or a little larger than the record at Port Island. The Port Island site is closer to the epicenter than the Daikai station, but the depth of GL-83m is deeper than the base of theanalysis; therefore, this can be a reasonable result.

In the future, if one wants to analyze the Daikai station in order to show another mechanism

responsible for the damage, he must explain two other phenomena, i.e., distortion of the ventilationtower, and slight damage to the Nagata station.

ACKNOWLEDGEMENT

This paper is summary of several papers compiled by the author and his colleagues, such as Iidaet al., 1996, Nakamura et al, 1996, Yamato et al., 1995, Yoshida and Nakamura, 1996, and Yoshidaet al., 1996. The author thanks their contributions.

REFERENCES

ADEP investigation team. 1990. Reconnaissance report of the 1989 Loma Prieta earthquake, Association of

Development of Earthquake Prediction (in Japanese)

AEDP (Association for Earthquake Disaster Prevention). 1998. Strong motion array record database, No. 3

CEORKA (The Committee of Earthquake Observation and Research in the Kansan Area,). 1995. Earthquake

record during the 1995 Hyogoken-nambu earthquake, http://www.ceorka.org/

Hamada, M. 1992. Large Ground deformations and their Effects on lifelines: 1964 Niigata earthquake, case

studies of liquefaction and lifeline performance during past earthquakes, Technical Report NCEER-92-0001,

National Center for Earthquake Engineering Research, Buffalo

Iida, H. et al. 1996. Damage to Daikai Subway station, Special Issue of Soils and Foundation, JGS, pp. 283–300

JSCE investigation team. 1986. Reconnaissance report of the 1983 Nihonkai-chubu earthquake, Japan Society

of Civil Engineering (in Japanese)

JSSMFE investigation team. 1994. Reconnaissance report of the damage during the 1993 Kushiro-oki and

Noto-hanto earthquakes, Japan Society of Soil Mechanics and Foundation Engineering (in Japanese)

Kawashima, K. 1994. Earthquake resistant design of underground structure, Kajima Shuppan (in Japanese)

(in Japanese)

Kobe city. 1980. Geology of Kobe, Kobe city (in Japanese)

Nakamura, S., Suetomi, I. and Yoshida, N. 1996. Estimation of aseismic ground displacement around Daikai

Subway station based on earthquake damage, Proc., 31st Symposium of Geotechnical Engineering , pp.

1275–1276 (in Japanese)

Tohoku Branch of JSCE investigation team. 1980. Reconnaissance report of the 1978 Miyagiken-oki

earthquake, Tohoku Branch of Japan Society of Civil Engineering (in Japanese)Yamato, T. et al. 1995. Damage to Daikai Subway station, Kobe Rapid Transit System and estimation of its

reason during Hyogoken-nambu Earthquake, Jour. of Geotechnical Engineering, Proc. of Japan Society of

Civil Engineer , No. 537/I-35, pp. 303–320 (in Japanese)

Yasuda, S. and Yamaguchi, I. 1985. Dynamic shear modulus obtained in the laboratory and in-situ, Proc.,

Symposium on Evaluation of Deformation and Strength of Sandy Gravels , JSSMFE, pp. 115–118 (in

Japanese)

Yoshida, N. and Nakamura, S. 1996. Damage to Daikai Subway station during the 1995 Hyogoken-nambu

Earthquake and its investigation, Proc., Eleventh World Conference on Earthquake Engineering , Acapulco,

Mexico, Paper No. 2151

Yoshida, N. et al. 1996. Dynamic analysis of Daikai Subway station, Kobe Rapid Transit System, Proc. Seminar

on Dynamic Analysis on the Great Hanshin-Awaji Earthquake Disaster , JGS, pp. 38–53 (in Japanese)

389


http://www.ceorka.org/






Vertical array records during 1995 Hyogoken-Nambu earthquake

by Kobe city, KEPCO and other organizations

Y. IwasakiGeo-Research Institute, Osaka, Japan

ABSTRACT: This paper describes vertical array strong ground motion data obtained during theHyogoken Nambu Earthquake (Kobe Earthquake) of 1995. The wave forms and the geotechnicalconditions are provided as well as other related information for a number of sites. Based upon a

series of vertical array data for aftershocks at Port Island, changes of geotechnical characteristicsafter the liquefaction were analyzed. The results shows sudden decrease of P-S wave velocitiesduring the liquefaction and then gradually recovered to the original states. Some of these data areavailable for readers who want to work on the vertical array records.

1 INTRODUCTION

During the Hyogoken-Nambu Earthquake of 1995, strong ground motions have been recorded at various sites. In geotechnical earthquake engineering, it is important to study such dynamic

behavior of soils as amplification of surface soils as well as liquefaction phenomena through

in-situ records of strong ground motion. Most of their records are at the ground surface and or ground floor of a structure.

If the strong ground motions in underground are obtained by a vertical array of seismographs, theresearch in this field is expected to advance much in depth.After the Hyogoken-Nambu Earthquake,KiK-net data, obtained by nearly 1000 sets of vertical array recording system deployed by NIED(National Research Institute of Earthquake and National Disaster, Japan) are available throughInternet service.

This paper introduces records of the vertical arrays that have been obtained during the KobeEarthquake of 1995.

2 VERTICAL ARRAY RECORDS

The promotion committee to establish database of strong ground motion records in Japan worked tocompile array data and published the results in 1998. The data by the committee contains most of theground strong motion data of the vertical and horizontal array systems in Japan including Hyogoken

Nambu Earthquake as well as such other array system being operated by Tokyo University at ChibaExperimental site. There were 18 organizations in Japan that have been operating horizontal and vertical array system that had provided data to the committee. Among them, five organizationswere operating the vertical array system during the earthquake at 10sites in total shown in Fig.1and listed in Table-1.

Disaster Control Research Center, Tohoku University, operates monitoring system of ground motion in Sendai city since 1983 including vertical array system. The data are planned to be openin near future.

3 VERTICAL ARRAY RECORDS DURING HYOGOKEN-NAMBU EQ

The recorded strong ground motions for each site are plotted from Figures.15 to 22 together withgeotechnical conditions as well as dynamic characteristics at the site, where the information is

391




Figure 1. Location map of vertical array sites during Hyogoken Nambu Earthquake.

Table 1. Vertical array data during Hyogoken Nambu Earthquake.

Altitude Latitude No of

Site Name (degree) (degree) depths installed depth Organization

1 Port Island 34.670 135.208 4 GL, 16 m, −32m, −83m Kobe-city

2 YAMAZAKI 35.060 134.603 2 GL-0.7 m, −30 m KEPCO

3 TAKASAGO 34.753 134.783 3 GL, −25 m, −100 m4 TECHNICAL 34.743 135.442 3 GL, −24.9 m, −97 m

RESEARCH

5 KAINAN PORT 34.150 135.192 3 GL, −25 m, −100 m

6 Higashi-Kobe Bridge 34.707 135.296 2 GL-2, −30m Hanshin

7 INAGAWA 34.836 135.427 2 GL-2, −30 m Exprees Way

8 MKO(Miyakojima) 34.704 135.523 2 GL, −70.39 m Ohbayashi Corp.

9 TIS(TAISYO) 34.650 135.478 2 GL, −57.49 m

10 TKMF(TAKAMI) 34.690 135.462 2 GL-1.5 m, −30 m Konoike Corp.

available. Since the surface layers show smaller S-wave velocities than deeper ones in these arraysites, ground motions at the surface are generally amplified compared to those at deeper depths.However, some of the records of the acceleration on the surface are found significantly decreased compared to those in the underground. These are probably caused by liquefaction.

There are several papers on the array data itself and research based upon array data. Generalcharacteristics of Port Island records are discussed by Iwasaki et al. (1996). Sugito et al.(1996)discussed some errors of angles of installation of the accelerometers for Kobe Port Island and four sites of KEPCO. He indicated that the directional error might become significant effects on

392





Figure 2. Comparison of ground condition before and after the Earthquake of 1995.

the analysis of liquefaction. Sato et al. (1996) also discuss the directional errors in addition tonon-linear and liquefaction effects on records at KEPCO sites.

Wakamatsu, et al. (1995) discussed characteristics of ground motion in Osaka based uponobservation of microtremors and compared a ratio of H/V characteristics with those of strongmotions. Fujii et al. (1996) reports that change of natural period of the structure depends upon rather

ground motion than the structural displacement. Yamazaki et al. (1997) simulated 3D behavior of bridge structure of Higashi Kobe based upon input motions obtained during the earth-quake.Ground motion records obtained at depth were used as the input motion for preliminary liquefactionanalysis. The simulated results correspond well to the observed surface record that shows longer

period of pulse after liquefaction. Kokusho et al. (1998) studied vertical array records at severalsites and focused rotation errors and the strain dependent characteristics of soils.

4 PORT ISLAND RECORDS AND SITE CONDITION

Port Island site was a manmade land in front of Kobe-city where the coastal sea was reclaimed

by excavated weathered granite transported from mountain areas behind the city. Fig.2 shows acomparison of site investigation results before and after the earthquake for the bore hole where thevertical array was installed. It is noted that the SPT N-values in the filled layer at the top surface wasabout 5 before the earthquake and increased to about N=10 due to post liquefaction densificationof the filled layer. In contrast, no significant increase of S-velocity occurred.

Figs.3 and 4 show the NS and UD components of strong ground motion records of the mainshock. In Fig.3, the liquefaction is recognized to start about 14seconds because high frequencycomponent are lost in the record at the ground surface compared to the record at G.L.-83m.

During the time window from 14 to 20seconds, the NS component contains only the long period of one to two seconds. Horizontal motions that are mainly caused by vertically traveling shear waves are strongly affected by liquefaction because the rigidity of the liquefied layer becomes very

small. In contrast, much shorter period waves are dominant in the UD component in the sametime window as shown in Fig.4. The vertical motions are less affected by liquefaction because theP-wave velocity depends upon dynamic compressibility of soil, which is insensitive to liquefaction.

5 AFTERSHOCKS IN THE PORT ISLAND RECORDS

Figure. 5 shows a full length of the recorded ground motion at the Port Island site for UD and NScomponents at two depths of G.L.0 m and G.L.-16 m after the main shock with a time length of

393





Figure 3. NS component of the main shock at Port Island site.

360seconds. There are several aftershocks following the mainshock. Here, these aftershocks will beinvestigated to time dependent changes of amplification, P and S-wave velocities, and predominantfrequency of the ground after the liquefaction.

5.1 Change of ratio of maximum amplitude at surface to a depth of G.L.-16 m

The maximum amplitudes of the main shock and aftershocks for horizontal and UD componentsat G.L.00 m and G.L.-16 m are listed in Table 2. Ratio of the maximum amplitudes at G.L.0 mto G.L.-16 m are listed in Table-3, and plotted in Figure.6. In the record of UD-component of G.L.-16 m, there are spike type pulses with very large amplitudes. These spikes are neglected here,

because these pulses are not recorded in the deeper depths and unreliable.

The ratio of the maximum amplitude of the main shock was less than one (0.67) for the horizontalcomponent and larger than one (2.23) for the vertical component. The horizontal motion was de-amplified, while the vertical motion was amplified as a result of liquefaction. The amplituderatios for the horizontal components were decreased from 0.67 of the mainshock to about 0.3 for three aftershocks that occurred between 34 and 208seconds from the start of the mainshock. Theaftershocks that occurred after around 300seconds show some increase of the ratio.

On the other hand, the ratios of the UD components show the similar tendency of decreasingafter the mainshock, keeping rather constant for a while, and increasing at about 300seconds.Based upon the fact that the ratio of the horizontal component begins to increase around 250 to300seconds, it is considered that the excess pore water pressure built up during the main shock has

maintained by that time.

5.2 Estimation of velocity under liquefaction based upon aftershocks

The aftershock records was divided into two sections of P-wave and S-wave. In Figure.10, onlythe shear wave part is shown after the S-wave arrival. Figure.7 shows a cross correlation function

between two records of NS component at G.L.00 m and G.L.-16 m.The cross correlation function shows its maximum value at t= 0.23seconds, which is considered

to be the travel time of S-wave from the depth of G.L.-16m to G.L.00. The estimated value of the

394





Figure 4. UD component of the main shock at Port Island site.

Figure 5. Strong ground motion record of duration 360sec from triggered time (UD and NS-component).

395






Table 2. Change of maximum amplitude of main shock and aftershocks.

UD(gal) EW(gal) NS(gal) Horizontal Com.(gal)

time/elapsed

time(sec) GL00m GL-16m GL00m GL-16m GL00m GL-16m GL00m GL-16m

10/0 555.0 *248.7 284.0 543.0 314.0 565.0 426.0 636.0

44/34 24.0 18.6 7.3 30.0 9.3 20.0 9.6 30.0

184/174 33.8 59.3 12.4 60.0 15.9 81.2 16.6 60.0

218/208 65.6 59.3 19.6 70.7 22.7 94.3 26.1 100.3

303/293 23.9 14.3 8.1 18.8 9.4 18.2 10.0 21.0

317/307 10.0 11.7 7.2 10.7 8.9 16.5 9.6 17.3

time in record/elapsed time from the beginning of the main shock: Horizontal

Horizontal Component is root sum squired of EW and NS in time domain.

*248.7 is the maximum value although some larger motions of spike tyoee was recorded in later section.

Table 3. Change of Amplitude Ratio.

Elapsed time (sec) UD EW NS Hor.Com.

0 2.23 0.52 0.56 0.67

34 1.29 0.24 0.47 0.32

174 0.57 0.21 0.20 0.28

208 1.1 1 0.28 0.24 0.26

293 1.68 0.43 0.52 0.48

307 0.85 0.67 0.54 0.55

Figure 6. Change of maximum Amplitude ratio of main and aftershocks with elapsed time.

shear wave is 76 m/sec. The same procedure was applied to obtain other aftershocks for P-waveand S-wave and the estimated velocity values are listed in Tab 4 and plotted in Figure.8.

It was difficult to obtain time duration of S-wave for aftershocks with elapsed time shorter than200seconds, hence the delay time was evaluated for only three aftershocks. The velocities measured

by PS-logging before and after the earthquake are also shown in Table 4 and Figure 8.

396





Figure 7. Estimation of elapsed time for shear wave travel from −16 m to 0 m by cross correlation.

Table 4. Change of velocities after liquefaction.

P-wave S-wave

Elapsed time delay time Vp delay time Vs

(sec) Dt(sec) (m/sec) Dt(sec) (m/sec)

PS-log.(Aug, 1991) 780 – 210

34 0.10 160

174 0.08 200

208 0.08 200 0.30 53

293 0.07 229 0.21 76307 0.07 229 0.12 133

PS-log.(June, 1995) 1,410 210

Shear wave velocity once decreased to less than 40m/s corresponding to liquefied conditionrecovers after 200 seconds. It should be noted that P-wave velocity decreased also when theliquefaction took place.

The P-wave, originally Vp=780 m/s, decreased to 160 m/s and gradually recovered to about200 m/c. The original P-wave velocity that was less than that in water of about 1.5 km/s indicating

the layer was not fully saturated condition. If it was fully saturated condition, the decrease in Vpmight have been much less than the observation.

5.3 Change of transfer function between G.L.-16 m to G.L.0 m

Figure 9 shows spectral characteristics of NS components at G.L.-16 m and G.L.0 m of the after-shock shown in Figure 7. The lowest frequency among three peaks is f = 1.1Hz that correspondsto the fundamental frequency of the top surface layer of 16 m in thickness with S-wave velocity of 70 m/s.

397





Figure 8. Change of velocities after the mainshock.

Figure 9. Spectral ratio of NS-component of acceleration at G.L.00 to G.L.16(308–315 sec).

Figs.10 and 11 show time-dependent change of the spectral ratio of the NS and UD components,respectively, with different elapsed time. There is a significant difference between them after theliquefaction. In the horizontal motion, the spectral ratios are lower than one within 200seconds of the elapsed time.

On the other hand, the spectra ratios of UD components shown in Fig.11 are mostly greater thanone. The predominant frequency tends to increase with the elapsed time, presumably correspond-ing to the increase of Vp in the liquefied layer. Fig.12 shows a linear relationship between theP-wave velocities listed in Table-4 and the predominant frequencies shown in Fig.11 for the sameaftershocks.

398






Figure 10. Change of spectral ratio of horizontal NS component at G.L.00 to G.L.16.

Figure 11. Change of spectral ratio vertical UD component of G.L.00 to G.L.16.

The predominant frequency fp, the resonant frequency based upon multiple reflection of P-wave,is formulated for a layer of 16 in thickness as

399






Figure 12. Relationship between P-wave velocity and Predominant Frequency.

fp: predominant period (Hz)This equation is superposed in Fig.12, which fits well with the data points.

6 ROTATION ERRORS IN THE VERTICAL ARRAY RECORDS

Rotation error is the difference of angle of orientation of the installed seismometer and the intended direction. It has been reported that the directions of the axis of seismometers at different depths ina vertical array were not the same and need to be corrected.

Sato et al. (1995), Sugito et al. (1996), and Kokusho et al.(1998) discussed these rotation errorsof the vertical arrays for several sites based upon ground motion records during the Hyogoken

Nambu Earthquake of 1995. Sato et al. (1995) and Sugito et al. (1996) had estimated the rotationerror by comparison of orbits of strong ground motions at underground with those at the surface.Kokusho et al. (1998) applied the maximum coherence method to estimate the offset angle fromthe orientation of the installed seismometers at the ground surface.

Table-5 shows the estimated offset angles of the installed directions of seismometers for four sites reported by three different papers. The four sites are Port Island and three vertical arrays thatwere operated by KEPCO.

The correction of the original record is obtained by rotation by the offset angle anti-clockwisefor the plus and clockwise for the minus numbers.

Figure 16 shows distribution of these offset angles against depth for three papers. It may be seenthat the offset angle tends to become larger with the installed depth. It is natural that it becomesdifficult to set the axis of the seismometer with correct direction for the deeper position duringinstallation.

However, it should be noticed that the direction of the seismometer at the ground surface may

sometimes installed with opposite direction of offset angle of 180 degrees.Sugito et al. (1996) indicated that there are two sites where an installed horizontal direction wasoffset angle of 180degrees that is opposite polarity of the sensor direction. Under this situation,simple procedures of either orbit comparison or the coherent method do not give correct offsetangle.

Figure 14 shows relationship of the off set angle estimated by Sato et al. and other two estimates.If these offset agrees each other, the plotted points should be on a straight line with the inclinationof 45 degrees. The large magnitude of dispersed plots in Figure 14 indicate the need of further research to establish proper method of obtaining offset angles in a vertical array.

400





Table 5. Rotation error of installed seismometers in vertical arrays.

Directional offset (degrees)

installed Sato et al. Sugito et al. Kokusho et al.

Site Name No. of depths depth (1996) (1996) (1998)

1 Port Island 4 GL 0 0 0

16m 0 0 11

32m 0 0 8

83m 15 22 27

3 Takasago 3 GL 0 0 0

25m −30 −20 6

100m 0 16 −37

4 Technical Research 3 GL 0 180(NS) 0

24.9m 0 0 6

97m −34 46 −37

5 Kainan port 3 GL −60 180(EW) 0

25m 0 67 70100m 0 46 46

Site location number corresponds to Figure 1

Figure 13. Distribution of rotational errors estimated by Sato, Sugito, and Kokusho.

7 OTHER ARRAYS DURING THE HYOGOKEN NAMBU EARTHQUAKE

Other array data listed by Table-1 are shortly described here. Geotechnical conditions are provided together with ground motions in the most sites.

7.1 Vertical array by KEPCO

KEPCO (Kansai Electric Power Company) operates seismic monitoring system at several importantfacility sites. At Yamazaki site, the soil profile consists of top 3m of soft weathered soil, 6m of weathered rock, and andesite rock down to G.L.-37 m. No information of PS wave velocity isavailable. Takasago site consists of loose sand (N=10-20) and gravel formation (N>=50) down

401





Figure 14. Off set angle estimated by Sato et al. vs. by Sugito et al. and Kokusho et al.

to 13.5 m and dense gravel layers including some stiff clay. Technical Research Center site locatesat soft alluvial plain between Osaka and Kobe. A soft sandy silty layer, including some sand and gravel, of 11.5m in thickness (N=10) is followed by alternation of dense sand (N =50) and stiff

clay (N=10) each about 5 m thick on average. Kainan Port site has a medium dense sand (N= 10– 30) of 17 m in thickness on the top surface followed by alternation of clay and sand down to 83.5 mwhere base rock is reached.

7.2 Vertical array by Hanshin Express Way Co.

There are two sites of vertical array system at Higashi Kobe bridge and Inagawa operated byHanshin Express Way Co. Higashi Kobe bridge was constructed as a cable stay type of three spanswith center and side spans of 485 m and 200 m in 1993. Several accelerometers were installed atkey points of the bridge structure with two vertical array points in the ground near by the center

Tower. Yamazaki et al.(1997) reported the structural response analysis by the earthquake includingthe ground, suggesting the liquefaction during the earthquake based upon the special feature of therecorded waves in the vertical array. Inagawa site is medium dense sand and gravel(N = 20–40) atthe surface underlain by weathered rock ground. No information of P-S wave velocities is availablefor the two sites.

7.3 Vertical array by Ohbayashi and Konoike Corporations

Miyako-jima, Taisyo, and Takami sites are all related with high rise buildings where seismometersare installed beneath the foundation ground. Sawai et al.(1995) report some nonlinear effects of ground characteristics upon the ground motion records obtained.

8 AVAILABILITY OF THE VERTICAL ARRAYS DIGITAL RECORDS

The digital records of the strong motion array observation during the Hyogoken Nambu Earthquakemay be obtained from the Shinsai Yobo Chosakai at the following address.

The Shinsai Yobo Chosakai26-20, 5-chome, Shiba, Minato-ku, Tokyo, 108-8414

402





Figure 15. Takasago site, Kepco. Figure 16. Technical Research site, Kepco.

The strong ground motion records by KEPCO are obtained by request to Kansai Electric Power Co. Inc. as follows,

Department of Civil and Structural EngineeringKansai Electric Power Co.3-6-16, Nakanoshima, Kita-ku

403





Figure 17. Kainan Port site, Kepco. Figure 18. Yamazaki site, Kepco.

Osaka, 530-8270Port Island records are included in the CD-Rom attached in this volume with key to open. To

obtain the key, reader is requested to send his request to the following author’s address.E-mail address: [email protected]

404








Figure 19. Miyakojima site, Obayashi. Figure 20. Taisyo sites, Obayashi.

9 CONCLUSIONS

The vertical array records during the 1995 Hyogoken Nambu earthquake are introduced here and the typical waves of recorded earthquakes are shown. It is intended to provide basic characters of these records and information to those who want to obtain these data.

405





Figure 21. Kobe Higashi Bridge, HanshinExpressWay. Figure 22. Takami-site, Konoike.

Port Island ground motion had recorded a long duration of 360seconds, which contains severalaftershocks as well as the mainshock. Based upon a successive aftershocks of G.L.00 m and G.L.-16 m, the changes of the characteristics of the liquefied ground were obtained as follows,

1. Amplitude of horizontal and vertical motions at the surface are compared with those of G.L..-16 m. The ratio of the amplitudes shows clear de-amplification in horizontal motions during

406





the main shock and earlier aftershocks with gradual recovery from the de-amplification withelapsed time during later aftershocks.

2. Theratio of the vertical ground motions shows amplification duringthe mainshock, but decreased to show de-amplification during aftershocks with gradual recovery with time.

3. S-wave velocity decreased to the level at which S-wave arrival of aftershocks was not identifiable but gradually recovered about 200 s after the mainshock.

4. P-wave velocity also decreased significantly but also recovered with time.5. The time dependent change in predominant periods of horizontal and vertical motions

corresponds to the above mentioned changes of P and S wave velocities.6. It should be noted that SPT N-value has increased two times from N=5 t o N=10, no significant

increase of S-velocity was reported.7. Correction of rotation error of installed seismometer have been tried but still remains uncertainty

and needs to establish systematic correction procedure that can take into account polarity as wellas rotation.

The author expresses his sincere thanks to Kobe city and KEPCO for their generous permissionsto agree to provide their valuable array data if the reader requested. The digital data in this paper

are from No.3, Strong Motion Array Observation, March, 1998, Shinsai Yobo Kyokai.

REFERENCES

Kokusho, T. and Matsumoto Masaki. 1998. Nonlinearity in Site amplification and Soil Properties during the

1995 Hyogoken Nambu Earthquake, Special Issue on Geotechnical Aspect of the January 17, No. 2, Soils

and Foundations, 1–9, Tokyo: Japanese Geotechnical Society

Sato, K., Kokusho T., Matsumoto M., and Yamada E, 1996. Nonlinear seismic response and soil property

during strong motion, Special Issue of Soils and Foundations 41–52, Tokyo: Japanese Geotechnical Society

Iwasaki Y. and Tai M. 1996. Strong motion records at Kobe Port Island, Special Issue of Soils and Foundations

29–40, Tokyo: Japanese Geotechnical SocietySugito, M., Sekiguchi K., Yashima A., Oka F., Taguchi Y., and Kato Y. 1966. Correction of orientation error of

borehole strong motion array records obtained during the South Hyogo Earthquake of Jan. 17, 1995, J. of

Structural Eng./Earthquake Eng. Vol. 12, No. 3., No. 531.51–63, Tokyo: JSCE

Sawai N., Fujii. A.,Yokoyama H., Matsutani T., Ishida J., and Kobori J. 1996. Investigation on the High rise RC

Apartment Housing at Takami subjected to Hyogo-ken Nambu Earthquake, (6) On the record by Earthquake

Observation – Part 3, Proc. of An. Conv. of AIJ, paper number 21216, Tokyo: AIJ (in Japanese)

Yamazaki. F. et al., 1997. Response simulation of Higashi Kobe Bridge during Hyogo-ken Nambu Earthquake,

Proc. of 24th Symp. of Earthquake Engineering in Japan, pp. 613–616, (in Japanese)

Strong Motion Array Observation No. 3, 1998. Tokyo: Shinsai Yobo-Kyokai

407





Observed seismic behavior of three Chilean large dams

R. Verdugo Department of Civil Engineering, University of Chile, Chile

G. PetersCMG Ingenieros, Chile

ABSTRACT: Seismic shaking of large dams constructed with coarse material, such as rockfill,cobbles and gravels, is normally associated with settlements that, depending on both shaking

intensity and degree of compaction of the coarse material, can be significant. The observed seismicresponse of Cogoti and Santa Juana dams, with heights of 83 and 113 m, respectively, is presented.In the case of Cogoti Dam, an accumulated seismic settlement of 1.2 m has been measured, whichinduced damage in part of the concrete face, without involving any mechanical instability. On theother hand, Aromos Dam, a zoned dam with a height of 42 m, supposed to be founded on liquefiableground according to SPT data, underwent seismic settlements smaller than 10 cm during the 1985Chilean earthquake of Magnitude 7.8. The post seismic analysis suggested that liquefaction wasactually restricted because the liquefiable soils were confined by rather dense sandy materials.

1 INTRODUCTION

Chile is located on the southwestern portion of the American Continent, and a significant part of itsterritory is controlled by a subductive seismic environment associated with the collision between

Nazca and South America plates. The convergence rate among these plates is estimated to be in therange of 65 to 90 mm/year. As a result of this tectonic interaction, most of Chilean territory havea high rate of seismicity that includes the largest ever recorded ground motion; the 1960 Valdiviaearthquake with an estimated Magnitude 9.5 (Madariaga, 1998).

On the other hand, an important number of large dams are located along the Los Andes Rangetaking advantage of both available natural hydraulic resources and elevation. Moreover, there alsoexist some dams located in different valleys. Considering the high seismic activity that normally

takes place in Chile, all these earth structures are supposed to be designed to withstand strongearthquakes. In this context, the seismic response observed at three different Chilean high dams,Cogoti, Santa Juana and Aromos, is presented.

2 COGOTI DAM

2.1 Technical characteristics of the dam

Cogoti Dam is located in the confluence of Pama and Cogoti rivers, in Limari Province of Chile,about 65 km South of Ovalle city (see Fig. 1), at an elevation close to 575 m above sea level and.

It retains a reservoir capacity of 150 millions m

3

when the water is at the spillway crest elevation(654.8 m). In the left abutment, a lateral spillway excavated in rock and without gate allows a water flow of 5000 m3/s.

Cogoti Dam is a concrete face rock fill dam (CFRD) completed in 1940, although the main bodywas finished earlier in 1938. The 160 m long embankment, with a maximum height of 82.7 m and a crest width of 8 m, was constructed by placing blasted rock in the site without compaction. Theupstream slope has a 1.42–1.47: 1 (H:V) inclination and the downstream slope has 1.47–1.50:1(H:V). A cross-section through the embankment and the plan view are shown in Figs. 2 and 3,respectively.

409




Figure 1. General location of Cogoti Dam.

Figure 2. Cross-section of Cogoti Dam.

In the first 15 m of the dam height rock particles with a maximum size of 1.5 meters were used,which were just dumped by gravity at the dam site. Then, the same material limited to a maximumsize of 1.3 meters was placed by mechanical means, which induced a slight compaction generated

by the passage of trucks during the construction. The technical specifications required that eachlayer of rock-fill be washed thoroughly with water pressure (water head of 60 m) using a quantityequivalent to three times the volume of the sluiced layer. However, the available technical records of construction indicate that most of the time the required washing was not fully satisfied. Therefore,it is possible to conclude that the body of Cogoti dam is associated with a rock-fill material whichis under a poor state of densification.

410






Figure 3. Plan view of Cogotí Dam.

The cross-section of the gorge is shown in Fig.4, where it can be seen that the profile is a rather unsymmetric, with the right abutment less steep and with a change in its inclination. It will be shownlater that a concentration of the crest settlements occurred above this point of slope change. The faceslab is made of a series of square concrete plates of 10 m× 10 m, which have a thickness varyingfrom 80 cm at the base to 20 cm at the top, where it continues as a short vertical retaining wall of 1 m high. The concrete face is resting on a layer with an average thickness of 3 m and consistingof hand-placed small rock particles. Just beneath the vertical and horizontal joints a continuous

ditch was constructed to provide a better support to the joints. This ditch system is approximately0.7 m deep and 1.2 m wide. The joints between concrete plates were sealed by means of copper waterstops of 1.5 mm thickness and 60 cm wide. Additionally, the openings between plates werefilled with asphalt. The distribution of the individual concrete plates is shown in Fig. 3.

The base of foundation has a reinforced concrete cut-off wall that was initially designed with adepth of 3 m, but was finally built with a maximum depth of 20 m and an average depth of 7 m.On the other hand, a concrete wall with a variable depth in the range of 1.4 to 2 m is basically the

plinth along the abutments.Regarding the geology of the site, the embankment was placed in a deep and narrow gorge eroded

by Cogoti River, which consist mainly of andesitic rocks. Three important structural alignments

have been recognized at the dam site: N45

0−

75

◦

E, E-W and N-S. These systems have beenidentified as part of the historical leakages observed throughout the abutments and foundation of the dam.

A general view of this dam is shown in Fig. 5.

2.2 Seismic history of Cogoti Dam

Cogoti Dam has been shaken by several earthquakes, being three of them being serious in terms of causing damages to the embankment. These earthquakes are described below.

411





Figure 4. Gorge profile.

Figure 5. General view of Cogoti Dam.

The first shaking that stroke this dam occurred onApril 6, 1943, and known as Illapel earthquake,

with a 7.9 Magnitude and an epicentral distance from the dam site of about 90 km. The peak ground acceleration (PGA) estimated at the dam site is 0.19 g (Arrau et al., 1985).The second important earthquake that affected the dam seems to have happened in 1949.

However, no record of this seismic event is available.The third seismic event that hit the dam causing substantial settlements occurred on October 14,

1997 at 22:03, local time. This event has been reported with Magnitudes 7.1 and 6.8 by the USGSand Servicio Chileno de Sismología, respectively. The hypocenter has been located at Latitude−30.9, Longitude −71.2 and at 58 km depth. The resulting epicentral distance from the dam siteis 16 km. In the seismic station of Illapel, the peak ground acceleration reached 0.27 g.

412






Figure 6. Chronological history of settlements.

2.3 Seismic settlements

The chronological history of measured settlements since 1938 at different control points at the crestof the embankment is presented in Fig. 6. The location of the control points are shown in Figs. 3and 7. Additionally, the distribution of the settlements along the crest at different times is shownin Fig. 8.

These data indicate that the maximum vertical settlement does not take place in the sectionassociated with the maximum height of the rockfill. Unexpectedly, at least for the authors, themaximum vertical settlement occurs above the point where the bedrock undergoes a slope change.The maximum settlement corresponds to control point M7, whose chronological history is shownin Fig. 9.

The maximum seismic settlements, SS, are:

– 1943 Earthquake SS = 41.7 cm, (Control point M-10) – 1949 Earthquake SS = 12.3 cm, (Control point M-7) – 1997 Earthquake SS = 25.3 cm, (Control point M-8)

At present, the total maximum seismically-induced settlement reaches a value close to 79 cm,which represents almost 1% of the dam high.

413





Figure 7. Control points and settlement distribution along the crest.

Figure 8. Settlements along the crest at different times.

At the end of 2001, the total maximum settlement reached 138.7 cm, representing a settlementof 1.7% of the height of the dam after 63 years.

The analysis of the static settlements is presented elsewhere (Verdugo, 2001).

2.4 Observed damages

After the 1943 earthquake, several longitudinal cracks showed up along the crest, with lengths of 30 to 40 m. Also some transversal cracks were visible in different sections of the crest. Settlementsin the concrete face were reported. The most important damage was associated with a large dis-

placement of the rockf ill that involved the whole downstream slope. This situation was considered risky and therefore, the slope was immediately repaired.

414






Figure 9. Chronological history of maximum settlement.

The leakage of the dam was increased from 500 liters/sec in 1944 to 2600 liters/sec in 1988,when the concrete face was repaired.

The other seismic events that hit the dam have not caused any further significant damage ascompared to what happened during the 1943 earthquake.

3 SANTA JUANA DAM

3.1 Dam description

Santa Juana is a rockfill concrete face dam located in the bed of Huasco River and located approx-

imately 17 km east of Vallenar City, II Region of Chile. It was completed in 1995, and it has acapacity of 166 millions m3.Santa Juana Dam, with a height off 113.4 m and a crest length of 390 m, was constructed with

rock particles with a maximum size of 1 and 0.65 meters in the upstream and downstream supportingshoulders, respectively. All the materials were adequately compacted. A cross section showing thedistribution of these two materials is presented in Fig. 10. Inclination of 1.5:1 (H:V) and 1.6:1(H:V) were designed for the upstream and downstream slope, respectively.

The concrete face has a variable thickness from 45 cm at the base to 30 cm at the top, where itis transformed into a parapet wall.

415





Figure 10. Cross-section of Santa Juana Dam.

Figure 11. Profile of the valley and settlement distribution.

The shape of the throat of the rather narrow valley eroded by Huasco River is shown in Fig. 11.The profile is rather regular and it is possible to assume that the profile totally consist of bedrock.A cutoff wall made of plastic concrete was built in the fluvial material located at the river bed.

Control points along the crest of the dam were installed to monitor the settlements, which areindicated in Fig.12. Accelerometers were installed at different locations in order to study the seismicresponse of the dam, and the locations are indicated in Fig. 12. The accelerometer A3 is installed in rock.

3.2 Observed settlements

Fig. 13 presents the measured settlements at different control points located along the dam crest. Itis observed that the vertical deformation is in some way shifted to the right, confirming the resultsobserved in Cogoti Dam that suggest that the shape of the bedrock has an important effect on thedistribution of the vertical settlements, static and seismic.

The maximum settlement takes place at the control point MC-3-5, and the chronological historyis shown in Fig. 14. It can be observed that the settlements after 4 years of operation are quite small

416






Figure 12. Accelerometers and control points of settlement.

Figure 13. Vertical settlements along the crest.

(1 cm) and it seems that the earthquake that hit the dam did not generate any particular change inthe pattern of deformation.

3.3 Recorded earthquakes

Between 1997 and 2001, a total of 22 seismic events were recorded by the accelerometersinstalled in Santa Juana Dam. Table 1 lists the recorded earthquakes sorted according to their peak

417






Figure 14. Chronological history of observed settlements at MC 3-5 control point.

accelerations. Earthquake No. 1 has peak accelerationat the crest of 0.226g(E-W), while earthquake No. 22 has peak acceleration at the crest 0.0088 (E-W).

3.4 Observed seismic response

The largest earthquake that has hit the dam occurred in October 14, 1997 at 22:03, local time,which has been already described above. This earthquake induced peak accelerations recorded bythe instruments and these are presented in Table 2.

According to these data the component E-W is amplified at the top:

An attempt to evaluate the natural period of the dam has been carried out. Accordingly thetransfer functions between “rock – crest” and “toe – crest” were computed. This can be done bydividing the amplitudes of the Fourier transforms of the motions recorded at crest and rock and

418





Table 1. Recorded earthquakes.

Earthq. Lat. Long. Depth (km) Date Mag. Crest Spillw. Toe Rock

1 30.445 71.197 55 14/10/1997 6.8 E-W 0.226 0.078 0.104 0.048

Vertical 0.157 0.056 0.111 0.04

N-S 0.124 0.072 0.073 0.049

2 28.343 72.029 95 18/07/1998 5.1 E-W 0.132 0.132 0.051 0.132

Vertical 0.087 0.064 0.036 0.132

N-S 0.064 0.067 0.040 0.132

3 28.456 70.797 124.2 24/12/2001 4.9 E-W 0.113 0.114 0.078 0.054

Vertical 0.104 0.103 0.09 0.033

N-S 0.091 0.117 0.075 0.054

4 29.268 71.299 54.3 08/10/2000 4.9 E-W 0.104 0.051 0.046 0.026

Vertical 0.084 0.034 0.042 0.016

N-S 0.075 0.046 0.056 0.025

5 29.253 71.638 33.7 09/07/2000 4.8 E-W 0.067 0.039 0.018 0.03

Vertical 0.081 0.046 0.011 0.018

N-S 0.037 0.054 0.01 0.0326 30.442 71.315 50.2 06/11/2001 5.1 E-W 0.051 0.009 0.018 0.005

Vertical 0.026 0.004 0.013 0.005

N-S 0.022 0.01 0.011 0.005

7 29.399 72.128 0 14/09/1998 4.5 E-W 0.049 0.018 0.022 0.011

Vertical 0.039 0.017 0.029 0.011

N-S 0.047 0.024 0.034 0.018

8 28.078 71.257 228.6 16/06/1998 4.5 E-W 0.044 0.049 0.033 0.01

Vertical 0.038 0.026 0.019 0.007

N-S 0.026 0.038 0.021 0.014

9 29.477 71.821 43.2 04/09/2001 4.6 E-W 0.033 0.038 0.014 0.005

Vertical 0.013 0.005 0.009 0.003 N-S 0.017 0.009 0.013 0.006

10 27.879 70.421 94 09/08/2001 4.6 E-W 0.033 0.01 0.018 0.007

Vertical 0.017 0.005 0.012 0.006

N-S 0.015 0.01 0.015 0.008

11 27.041 70.741 222.8 22/11/1997 4.9 E-W 0.026 0.007 0.018 0.006

Vertical 0.025 0.008 0.015 0.013

N-S 0.018 0.009 0.015 0.01

12 29.567 71.9 0 01/09/2007 4.8 E-W 0.025 0.013 0.019 0.006

Vertical 0.02 0.009 0.013 0.005

N-S 0.013 0.012 0.014 0.008

13 28.502 71.39 23.8 17/08/2000 4.4 E-W 0.025 0.007 0.018 0.009Vertical 0.014 0.005 0.011 0.006

N-S 0.016 0.009 0.01 0.008

14 31.099 71.86 25.8 12/01/1998 6 E-W 0.023 0.004 0.011 0.003

Vertical 0.014 0.004 0.008 0.01

N-S 0.013 0.004 0.009 0.003

15 28.655 69.982 37.2 21/08/2001 3.9 E-W 0.023 0.011 0.014 0.005

Vertical 0.015 0.005 0.012 0.003

N-S 0.016 0.013 0.015 0.006

16 28.741 71.935 224.6 05/09/1997 4.8 E-W 0.021 0.006 0.011 0.005

Vertical 0.013 0.006 0.015 0.004

N-S 0.016 0.008 0.012 0.00617 29.513 72.26 34.2 03/09/1998 4.9 E-W 0.02 0.009 0.015 0.007

Vertical 0.016 0.007 0.009 0.006

N-S 0.014 0.009 0.01 0.008

18 28.365 68.951 131.6 16/10/2001 5.3 E-W 0.018 0.006 0.01 0.005

Vertical 0.02 0.004 0.009 0.004

N-S 0.016 0.009 0.011 0.006

(Continued )

419




Table 1. (Continued)

Earthq. Lat. Long. Depth (km) Date Mag. Crest Spillw. Toe Rock

19 28.95 71.181 163.1 05/10/1998 4.1 E-W 0.018 0.011 0.015 0.007

Vertical 0.017 0.007 0.014 0.005

N-S 0.014 0.014 0.014 0.01

20 29.162 70.802 218.4 28/09/1998 4.4 E-W 0.016 0.007 0.014 0.006

Vertical 0.015 0.005 0.011 0.003

N-S 0.015 0.012 0.012 0.006

21 28.152 71.148 87.7 12/08/2001 4.3 E-W 0.012 0.006 0.009 0.008

Vertical 0.013 0.005 0.007 0.005

N-S 0.01 0.01 0.008 0.011

22 28.049 68.12 394.2 29/01/1999 5 E-W 0.009 0.001 0.003 0.001

Vertical 0.007 0.001 0.003 0.003

N-S 0.005 0.002 0.004 0.001

Table 2. Peak recorded accelerations (g).

Accelerometer E-W N-S U-D

A1 0.078 0.072 0.056

A2 0.226 0.124 0.157

A3 0.048 0.049 0.040

A4 0.104 0.073 0.111

crest and toe. In the present analysis, the ratio of the undamped velocity response spectra of the tworecords were used. As shown by Hudson (1956; 1979), the undamped velocity response spectrum,given by:

is an upper bound of the Fourier amplitude spectrum:

The values of the peaks are almost identical in most cases and the numerical smoothing that isalways necessary to apply to the Fourier Transform is avoided. Additionally, the ratio of responsespectra are less sensitive to noise and the peaks are easier to identify. Typical results are shown inFig. 15.

The periods associated with the maximum value of the response spectral ratio were identified

and considered as predominant periods. The analysis of the available data is summarized in Fig. 16.It can be seen that a systematic predominant period for different earthquakes and different ratios(“crest – rock” and “crest – toe”) is obtained. According to these results, the predominant period of Santa Juana Dam in the E-W direction would be: T= 0.4 seconds.

If the classical expression of the predominant period is used:

420




Figure 15. Typical result of spectral ratio Crest – Toe.

Figure 16. Predominant period computed in each earthquake.

Where, H and Vs represent, respectively, the height of the dam (113.4 m) and the shear wavevelocity of the fill (≈700 m/sec). The theoretical predominant period would be: T= 0.42 seconds.

The match between empirical and theoretical values is quite good, although the geometry of SantaJuana Dam (crest length/high: 390/113.4= 3.4) does not satisfy the condition of plane deformation

(crest length/high> 4).

4 AROMOS DAM

4.1 Dam description

Completed in 1979, Aromos Dam is located in the Province of Quillota, approximately 90 km tothe north-west of Santiago (capital city of Chile). The reservoir has a capacity of 60.3 million m3.

421






Figure 17. Cross-section of Aromos Dam.

Figure 18. Plan view of dam-site.

The embankment is placed in a narrow zone of Limache Creek, about 5 km downstream of theconfluence with Aconcagua River. The cross section and plan view are presented in Fig. 17 and 18, respectively.

Aromos is a zoned dam 42 m high and 220 m long along the crest. The upstream and downstreamslopes are 3.75:1 (H:V) and 2.75:1 (H:V), respectively. The embankment consists of a core of f inesoils with supporting shoulders made of gravelly sand. It is resting on fluvial soil deposits consist

of sandy soil materials. Both abutments consist of weathered granite that improves with depth. Theshape of the throat is shown in Fig. 19.A plastic concrete wall of 80 cm in thickness and a maximum depth of 22.5 m is buried into

the foundation ground, with a part into the clay core, as impervious system to reduce infiltrationthroughout the fluvial deposit existing below the dam. It is important to mention that in severalsectors the plastic concrete wall did not reach the bedrock; however, the information indicates thatthese sectors were injected. Detail of this injection is not available.

The grain size distribution bands of the soil materials constituting zone 2, zone 3 and core are presented in Figs. 20. The Atterberg Limits indicate that the fines of the core material are classified

422






Figure 21. SPT N-values measured in the upstream zone.

with the efforts performed to pull out this equipment revealed that the ground was susceptible of undergoing liquefaction. Consequently, the engineers in charge of the dam construction worried about this issue and a great concern regarding the seismic stability of the dam arose.

As a result of this anomalous situation, geotechnical services of a well reputed internationalconsulting office were appointed, so an external evaluation about the actual risk of liquefactioncould be obtained. To assess the required seismic stability of Aromos dam, a complementarygeotechnical site investigation was performed. Boreholes with Standard Penetration Testing (SPT)were carried out. The available information associated with site investigation is presented below.

The Aromos dam was completed and its operation was delayed until the results of the requested seismic stability analysis became available.

4.3 Geotechnical site investigation

The main concern of the site investigation was to clarify the potential occurrence of liquefactionin the foundation that could trigger a major seismic failure. Two field exploration programs werecarried out; in 1979 and 1981, which basically consisted of standard penetration testing.

According to the area where the SPT were performed, the results were separated into threegroups:

– upstream zone – downstream zone – spillway zone

Additionally, the results for each sector were separated into two sets:

– Those where SPT N-values are predominantly greater than 15–20 blow/feet – Those where there are several SPT N-values smaller than 15–20 blow/feet

The results are shown in Figs. 21 to 23 for the upstream zone, downstream zone and spillwayzone, respectively. These results are not normalized to an overburden pressure of 1 kg/cm2.

These data show significant number of boreholes where the SPT N-values presents values below15–20 blow/30 cm, suggesting that liquefaction phenomenon could takes place under a strongmotion as the one expected for the area.

424





Figure 22. SPT N-values measured in the downstream zone.

Figure 23. SPT N-values measured in the spillway zone.

4.4 Results and recommendations proposed in 1981

In August 1981, the international specialists concluded that the soil deposits where the AromosDam had been founded presented a risk of liquefaction, although they recognized that it should notcomprise the whole foundation. According to this result, the following two main countermeasureswere proposed:

– A berm of compacted material conf ining one third of the upstream shell, with a height of 13 m – Drill a battery of drainage columns in the area between the toe of the dam and the spillway.

The available information indicates that only some drainage columns were actually installed at the toe of the dam. Due to the economic situation existing in Chile at that time, nothing else

425






Figure 24. Settlements recorded in the crest of Aromos Dam.

Figure 25. Chronological history of settlements at the upstream side of the crest.

427






Figure 26. Chronological history of settlements at the downstream side of the crest.

Figure 27. Relative displacements in the spillway walls.

More studies being conducted in this area in order to have a better understanding of the absenceof liquefaction in sandy soils with very low SPT N-values.


The observed seismic behavior of three existing dams located in Chile has been presented.

428






REFERENCES

Arrau, L., Ibarra, I. and Noguera, G. 1985: “Performance of Cogoti Dam under seismic loading,” Proceedings

of Symposium on Concrete Face rockfill dams – design, construction and performance . Michigan, USA.

De Alva, P., Seed, H.B., Retamal, E. and Seed, R.B. 1987: “Residual strength of sand from dam failure in the

Chilean earthquake of March 3, 1985,” UCB/EERC-87/11.

Hudson, D. 1956: “Response Spectrum Technique in Engineering Seismology,” Proc. 2nd World Conferenceon Earthquake Engineering .

Hudson, D. 1979: “Reading and Interpreting Strong Motion Accelerograms,” Earthquake Engineering

Research Institute, California Institute of Technology.

Madariaga, R. 1998: “Seismicity of Chile,” (In Spanish) Física de la Tierra (Madrid), (10):221–258.

Saragoni, R., González, P. and Fresard, B. 1986: “Analysis of the recorded acceleration histories of March 3,

1985 earthquake,” The March 3, 1985 Chilean earthquake, CAP. (In Spanish).

Verdugo R. 2001: “Evaluation of deformation modulus of coarse materials from the analysis of dam behavior”,

Proc. of the XV International Conference on Soil Mechanics and Geotechnical Engineering . Turkey.

430





Liquefaction-induced flow slide in the collapsible loess deposit

in Tajik

K. IshiharaCivil Engineering Department, Chuo University, Tokyo, Japan

ABSTRACT: In the suburb of Dushanbe, Republic of Tajikistan of ex-USSR, an earthquake of magnitude 5.5 took place on January 23, 1989. In this event, extensive liquefaction developed inthe loess deposit of Aeolian origin in the gently sloping hilly terrain and lead to a series of catas-

trophic landslides accompanied by a large-scale mudflow. In contrast to the hitherto known casesof liquefaction which have usually occurred in water-sedimented sand deposits, the liquefaction inTajik was unique and novel in that it occurred unexpectedly in wind laid deposit of silt in a semi-arid region. The reasons for such liquefaction are thought to be the collapsible nature of highly porousloessal silt which had been wetted by irrigation water over the past year. The complete collapse of the loess structure due to the additional action of the seismic shaking appears to have lead to thecatastrophic landslide.

1 INTRODUCTION

Development of liquefaction and consequent occurrence of slumping or flow slides in sandydeposits have been recognized as the major phenomena leading to catastrophic damage to theground during earthquakes. Actual cases of such failure have been reported by many investigators(Dobry & Alvarezs, 1967, Seed, 1987, Ishihara, 1990). Most of the cases ever encountered how-ever, involved liquefaction-induced flow slides in the deposits sedimented under water, whether

placed artificially or naturally. Thus, the studies of in-situ liquefaction have been confined to thosecases which occurred in lowland areas near the waterfront or in deposits underlying water-retainingembankments or dams.

It was, therefore, a great surprise to observe a series of catastrophic flow slides during a recentearthquake which took place in nearly flat terrain in a semi-arid region where the ground is covered

by a thick mantle of silts of Aeolian origin. The earthquake occurred in the suburb of Dushanbe,

capital city of Republic of Tajikistan, and the liquefaction induced thereby triggered a chain of landslides leading to a catastrophic mudflow. This event seems to be unique and previously unknown

phenomenon. Thus, in the following pages, the new features of the earthquakes and the liquefaction-induced mud flow will be described, together with the mechanism of liquefaction occurring in thecollapsible loess deposit.

2 TAJIK EARTHQUAKE OF JANUARY 1989

At 5:02 a.m. on January 23, 1989, an earthquake with a magnitude 5.5 shook a village area called

Gissar about 30 km southwest of Dushanbe, the capital of the Republic of Tajikistan which bordersonAfghanistan. The location of the epicenter is shown in the map of Figure l. The focal depth of thisevent is reported to have been about 35 km. As shown in Figure 2, the affected area is located on arelatively flat basin-like plain which developed in front of the southern flank of the Pamir Mountainrange. The Dushanbe River flows southward on a fan deposit through the city of Dushanbe and merges into the Kahirnighan River as shown in Figure 3. The epicentral area is located at Gissar near the junction of the Kahirnighan River and the Halaka River. The tremor was felt over the epicentralarea and a network of strong motion seismographs registered the motions at several stations asindicated in Figure 4. In the city of Dushanbe, the peak horizontal ground accelerations during the

431




Figure 1. Location of the epicenter of 1989 Jan. 23 earthquake.

Figure 2. Map of Republic of Tajikistan.

432






Figure 3. Map of Dushanbe and its vicinity.

main shock were of the order of 50 to 90 gal. The time histories of acceleration recorded at Cymbulif closest to the epicenter are shown in Figure 5, where it can be seen that the main shaking lasted for about 4 seconds with a peak acceleration of 125 gal in the north-south direction. Extrapolatingfrom the magnitudes of recorded motions in its vicinity, the epicentral area is supposed to have¥undergone a shaking with a peak ground acceleration of the order of 150 gal. In view of thismagnitude of shaking and the moderate degree of observed structural damage, the intensity of shaking at the epicentral area is estimated to have been 7 in MSK (Soviet scale in 12 degree),as accordingly indicated in Figure 4. Class 7 in MSK is equivalent to the class 7 in Modified Mercalli scale in USA and to the class 4 in terms of the Japanese Meteorological Agency scale.The occurrence of the earthquake is purported to be associated with a fault movement of the order of 30 cm as indicated in Figure 3, but it was not possible for the authors to trace any clearly visibleevidence of the fault on the ground surface.

3 LANDSLIDE AND DEBRIS FLOW

In the village of Gissar, there are about 500 farmer’s houses and barns constructed of wood withadobe type walls. Even in the most severely affected area, the degree of structural damage wassuch that houses were partially destroyed and thus the destruction due to the earthquake shakingitself was moderate and limited to small local area.

The most striking feature of the damage was a series of landslides and debris flow which took place over the gently sloping hilly terrain consisting of windblown deposits “loess”. There werefour landslides in the affected area of Gissar as indicated in plan view of Figure 6. The landslidesturned into a mudflow of vast scale and buried more than 100 houses in 5 meters of mud. Anestimated 270 villagers died or are missing in the debris.

433





Figure 4. Intensity distribution of the 1989 Jan. 23 earthquake.

Figure 5. Horizontal accelerations during the main shock of the earthquake.

434






Figure 6. Plan view of the slide area in Gissar.

Figure 7. A plan view of Sharara slide area.

3.1 Sharara slide

The landslide at Sharara is about 800 m in frontage and the debris spread out as far downward as 300 m, from the original toe of the hill as shown in Figure 7. Many houses in the villageimmediately downhill were buried in mud resulting in the largest number of causalities. A viewlooking northward from the top of the bluff is shown in the photograph in Figure 8, where twowater storage tanks are seen perching on the debris. These tanks had been installed on the hilltopto supply water for domestic use and for agricultural irrigation. Just beyond the scarp of the slideon the east side, a pumping station remains intact as indicated in Figure 7. The pumping stationwas used to pump up water to the storage tanks or directly to the water channel for irrigation.Approximate cross sections of the slide are shown in Figure 9, where it may be seen that the broken

435






Figure 8. A view to the north from the hill top at Sharara.

Figure 9. Cross sections of the slide at Sharara.

blocks of loess soil moved out and turned into debris flow. The depth of sliding surface is estimated to be about 30 m at the bluff line. It is to be noticed that the bluff line produced by the Sharara slideis almost coincident with the line of the water channel installed on the shoulder of the hill to supplywater for the farmland over the hills. Thus, it appears likely that the water in the channel had beeninfiltrating into the loess deposit over the years and the slide was initiated from the failure of loesssoils near the channel weakened by water invasion. In fact, near the distal end of the debris flow,muddy water was seen spurting and oozing from amid broken pieces of the loessal soils.

3.2 Firma slide

The site of the Firma slide is located about 500 m west of the Sharara slide as shown in Figure 10.This slide, having developed along the same shoulder line, appears to be a continuation of the slideat Sharara. Thus, the features and conditions of the occurrence of the slide are almost identicalto those at Sharara explained above. In fact, a small ditch about l.0 m wide and 0.5 m deep had

been excavated along the scarp line to supply water for the farmland on the gently sloping hills back of the scarp line. Thus, the loessal soils wetted by water inf iltration appear to be responsiblefor causing the slide during the earthquake. As shown in a cross section in Figure 11, there wasa considerable spreading of the soil mass, extending outward about 100 m from the toe of theslide. This characteristic feature is indicative of the fact that a considerable amount of water was

436






Figure 10. A plan view of Firma slide area.

Figure 11. Cross section of the slide at Firma.

Figure 12. Scarps of Firma slide looking from the north.

involved in the sliding mass of soils. Considerable ground cracking was produced over the hillslopes extending about 50 m rearward from the slide scarp, indicating that overall movement of soils took place over a relatively wide area behind the slide. A photograph looking southward atthe scarp is shown in Figure 12. It can be seen that the scarp is nearly vertical indicative of the factthat the soil block fell off from vertical cleavage or fissures.

437







Figure 13. A plan view of May 1 slide area.

Figure 14. Cross sections of the slide at May 1.

3.3 May 1 slide

The location of a small slide in the hamlet of May 1 is indicated in Figure 13. This slide encompasses

an area approximately 100 m in width and 100 m in length. As shown in a cross section of Figure 14,the slide scarp is located at the shoulder of a hill where unlined water channel about 3 m wide and 2 m deep was excavated. As was the case with the other slides, the water invasion into the loessdeposit appears to have weakened the soil prior to the advent of the seismic shaking. Unlike theother slides, a pressure ridge about 7 m high was formed at the toe, because of the buttress actionof the firm ground in front of the slide. The pressure ridge is clearly seen in the photograph inFigure 15.

At the bottom of a chink between broken blocks of soil as indicated in Figure 13, a cone penetration test was conducted by means of a portable hand cone test device. The cone has anapex angle of 30◦ and a cross sectional area of 3.23 cm2. The result of the cone test is presented in

Figure 16 in terms of the penetration resistance, qc-value, plotted versus depth. It may be seen thatthe cone resistance is about 12 kg/cm2 at a depth of about 2 m from the bottom of the chink wherethe sliding zone appears to have developed.

3.4 Okuli slide

The largest and most cataclysmic phenomenon was the multiple slides at Okuli which developed ina slightly depressed area over hilly farmlands. The plan view of the slide is shown in Figure 6. Thegenetic portion of the slide extends westward over a length of about 1.5 km from the escarpment

438






Figure 15. May 1 slide looking eastward.

Figure 16. Cone penetration test result in May 1 slide.

at the east (see Figure 17). Taking into account the fact that the slide encompasses an area about850 m in frontage and 15 m in depth on the average, the total volume of the soil mass is estimated to be approximately 20 million cubic meters. At least two slides seem to have been triggered independently from the hillsides on the north which then merged into the main stream of the mud

flow. The shallow slide indicated by B in Figure 6 took place on the terrace of high relief withits scarp located near the hilltop and after the soil moved over the gentle slope the debris flow jumped into the main stream. There are many traces of violent mud flow remaining on the exposed hard soils, indicating evidence of liquefaction and consequent muddy flow of loess soil duringthe earthquake (see Figure 18). The slides indicated by A and C in Figure 6 appears to have beeninitiated along the line of the water channel. It appears likely that a small slide initially induced atthe toe might have retrogressed backwards over a distance of 1.5 km to the east and sidewards tothe north as well as to the south. Over the rugged sliding area, the original ground was broken intomany blocks producing an extremely irregular and hummocky surface.

439






Figure 17. A view from the north over the eastern escarpment of Okuli slide.

Figure 18. A view looking eastward at mud flow from the northern hill (Okuli slide).

Amid broken blocks, there were spreads of soil mud having presumably sprung out of the in-depthsliding zone, suggesting that liquefaction of saturated loess soil occurred during the earthquake.The sliding mass of the loess soil turned into a huge-scale debris flow and traveled through a

distance of about 2 km on nearly flat surface of the ground. The debris covered an area as large asabout 1.5 million square meters, as indicated in Figure 6. Near the distal tongue of the debris flowat Okuli-poen, tens of farmer’s houses and barns were buried in 2 to 5 meters of mud. A photographin Figure 19 shows buried houses near the end of the debris. Several cone penetration tests were

performed on the rugged surface, as indicated in Figure 6, by means of the same portable conedevice as used in the May i slide area. Typical results of the cone tests are shown in Figure 20,where it may be seen that in the sliding zone at depths of about 7 to 8 m, qc-value was found to beabout 1∼5 kgf/cm2.

4 SOIL CONDITIONS IN THE AFFECTED AREA

The area affected by the earthquake has topography of gently sloping hilly terrain. The terrainis covered by a mantle of loess which was deposited windblown from Karakumy desert west of Tajikistan and Kirgistan Republics during the Pleistocene era. The windlaid material is silt and tanto light-brown in color. The soil conditions in this area were investigated by the Geological Bureauof Tajikistan Academy of Science in Dushanbe. The soil profile in the north-south cross sectionin the middle of the slide area (B-B section in Figure 6) is shown in Figure 21. The cross sectionalong the longitudinal direction of the slide (A-A section in Figure 6) is presented in Figure 22.

440






Figure 19. A buried house in the debris at Okuli.

Figure 20. Cone penetration test results at Okuli slide area.

It may be seen that the loess deposit covers the ground surface to a depth of about 30 to 40 m,underlain by gravelly sand which varies in thickness to a maximum of 200 m.

The results of investigations by the Geological Bureau of Tajikistan Academy of Science show

that the ground water table is located about 5 m from the ground surface, but the water contentdiminishes around the depth of 20 m which is underlain by dry layers of the loess. It is also reported that, while the layer of loess about 3 m in thickness below the ground water table has a water contentof about 20%, the loess layer below this level does possess a high water content with a maximumof 40%.

The zone of such a highly saturated layer of loess is indicated in Figures 21 and 22. On the other hand, the laboratory tests have shown that the liquid limit and plastic limit of the loess is around 30 and 20, respectively, giving a value of about 10 for the plasticity index. Thus, the natural water content of the highly saturated loess is in excess of the liquid limit. Figure 23 shows the grain size

441






Figure 21. North-south cross section B-B in Figure 6.

Figure 22. Cross section thorough A-A

in Figure 6.

Figure 23. Gradation curve of the loess soil.

442







distribution curve of the loess. It may be seen that the loess is comprised of 80% of silt and 20%of clay fraction, and that there is no sand fraction.

5 CHARACTERISTICS OF LOESS DEPOSITS

It is well known that windblown deposits of soils characteristically possess vertical cleavage planesand, therefore, when a deposit is excavated, soil blocks tend to fall off leaving nearly verticalscarps. In steep-sided gullies in the windlaid deposit, splitting off of columnar sections of soil

blocks is commonly observed. The wind-deposited loess in Gissar is no exception and there arevertical scarps developing widely along the head line of the slides in Sharara, Firana, May 1 and Okuli. It is also known (Clevenger, 1958; Dudley, 1970; Chen, 1987) that the windlaid loess istan to light brown in color, light-weight and consists of matrix of cemented silt with a number of interspersed macro and micro pores which are purported to be remains of root holes. Generally,the loess deposit is devoid of clear stratification and is highly crumbly and brittle in a dry state,

being powdered easily between fingers. The loess is known to be composed of silt-sized particles

of quartz and feldspars bonded together by a small fraction of montmorillonite-type clay. As isgenerally the case, the loess in Gissar area contains about 20% clay as indicated by the gradationcurve in Figure 23. The loess is known as a material having a matrix structure which is collapsiblewhen wetted with water. When maintained dry, it is reasonably strong and incompressible and the

porous structure may persist even under a fairly large overburden pressure. However, once wetted,the loess may lose its stability. Because of its highly porous nature, water can easily infiltrate intothe pores whereby breaking down the matrix structure. As a result, a significant amount of decreasein bulk volume and loss of shear strength can take place under sustained loads leading to largesettlements or failure of structures founded on such loess deposits. This kind of characteristics isoften referred to as hydraulic collapsibility.

Another feature of engineering significance is the fact that disturbed loess is of low-plasticity

with its Atterberg limits plotting near the A-line in the plasticity chart. The plasticity index of the loess is generally around 10. Thus, if collapse is triggered with a sufficient amount of water invasion in excess of the liquid limit, the disturbed loess being devoid of cohesion tends to easilyslump and flow. It is known that any soil with a low plasticity index has a great potential to developliquefaction and flow-type failure (Ishihara & Koseki, 1989).

With the above-mentioned characteristics of the loess taken together, it is expected with good reasons that the loess deposits in Gissar area has been invaded with a large quantity of water and narrowly on the verge of hydraulic collapse, and when subjected to the seismic shock loading, itinstantaneously developed liquefaction and resulting mud flow.

6 MECHANISM OF LIQUEFACTION AND DEBRIS FLOW

The lands in the Gissar area had been left uncultivated through geological time. However, with therecent development of township and increasing population in the urban area of Dushanbe, the landshave been cultivated and used as agricultural farmlands to produce wheat and cotton. To match theneeds for water supply in the agricultural lands, a network of water channels was constructed inthe hilly area in Gissar. Some sections of the water canal were lined with concrete but most wereunlined. The water for irrigation was pumped up to the hilltop, and was distributed to each patchof farmlands from two large storage tanks or, directly through a network of open canals. The water in the canals had been leaking and permeating into the ground through the vertical fissures in the

loess deposit over the years and the large quantity of absorbed water had been filling up the poresin the loess. As indicated schematically in Figure 22, the ground water table was found to lie about5 m from the ground surface but the water content of the loess at a depth greater than about 20 mwas found to be small. It was also discovered that the water content In the layer between depths of about 7 m and 17 m is greater than the liquid limit. Thus, the water content might have probably

been distributed throughout the depth of the loess deposit as shown in Figure 24. At depths a fewmeters below the ground water table, the pores of the loess were probably only partly filled withwater, but as the water pressure increased with increasing depth, the pores probably became fullysaturated, producing a state of over saturation with a water content in excess of the liquid limit.

443




Figure 24. Hydraulically collapsible state of loessal deposit.

Figure 25. Conditions of earth development.

The exact depth of the vertical cleavage in the loess deposit is not known, but the maximum depthat which vertical cracks can remain open may be inferred by comparing the vertical overburden

pressure with the uniaxial compressional strength of the soil element, as illustrated in Figure 25.Suppose the vertical stress at depth Z given by γ t z exceeds the compressional strength, q u, and

then the soil element at this depth will fail whereby producing a large deformation in the lateraldirection. If there exist an open crack to this depth, it would be closed by this lateral bulge. Thereforethe depth at which a crack is closed may be given by

where γ t is unit weight of the soil. The exact value of the compressional strength of the loess atGissar is not known, but in view of the cementation developed in the matrix structure, the strengthmay be inferred roughly to be within the range of q u = 200∼ 400 kN/m2. Then, assuming theunit weight to be approximately γ t = 15 kN/m2, the depth of crack penetration is estimated to beabout 15 to 25 m. Considering a dominant role played by the openness or closure of cracks, thegross permeability of the loess deposit would probably have been distributed through the depth

444






as shown in Figure 24. It is to be noted that at the depth between 15 and 20 m, the permeabilitycoefficient decreases sharply, prohibiting the water from migrating farther into the deeper portionof the deposit. This would account for the fact that the measured water content in the deeper depositwas small of the order of 10%. By retaining a large quantity of water, the loess deposits betweenthe depths of about 7 and 17 m appear to have been in a state of impending hydraulic collapseeven prior to the advent of the earthquake. In fact, it is reported that a noticeable amount of ground

settlements had be observed over the surface of the firm land prior to the advent of the earthquake.This fact indicates the hydraulic collapse had partially occurred in the loess deposit in this area.However, since the topography in Gissar area is so gentle, that the down slope component of thegravity force was not large enough to cause landsliding and therefore the ground remained stableas a whole before the occurrence of the earthquake. When the earthquake occurred in this area,the saturated loess already in a precarious state underwent shaking and total collapse of the loessstructure was provoked, leading to liquefaction of the silt-sized material. Moreover, because of thelow-plasticity of the silt, the liquefied silt began to flow out even on a nearly flat ground surface,carrying large masses of soil. In the case of the slide in Okuli, the mudflow traveled through adistance of 2.0 km as indicated in Figure. 6. In the Okuli slide, the slide appears to have started

first in a small section adjacent to the toe, and a series of sequential slides must have retrogressed from one section to another until it reached the place of the final escarpment on the east. This typeof progressive failure is likely to develop in soils which require only small magnitude of drivingforce to cause failure. It is to be noted that no landslide and mudflow in such a gentle slope canindeed take place with such a large scale except by the shaking of an earthquake.

7 SIMPLE ANALYSIS FOR LIQUEFACTION

It is apparent that the liquefaction in the loess deposit was caused by the combined action of water infiltration and seismic shaking. The loess deposit had been in a metastable state on the verge

of complete failure due to the invasion of water and as such even a small magnitude of seismicshock would great enough to produce a state of liquefaction and a consequent catastrophic mud flow. Thus, when examining the triggering mechanism, any analysis of liquefaction will require aknowledge of cyclic resistance of the loess which has been soaked to a precarious state in whichthe hydraulic collapse is about to take place. The cyclic resistance of a loess in such a state ’has not

been investigated and, in addition, in the absence of any data -from in-situ penetration tests in theloess deposit in Gissar area, it is difficult to estimate the cyclic resistance of the intact loess except

by means of a back analysis based on a known intensity of shaking during the earthquake. The back analysis to evaluate the cyclic strength can be made by calculating the through the formula,maximum shear stress ratio, γ max/σ

v, through the formula,

where denotes a peak value in an irregular time history of acceleration on the ground surface,γ max/σ

v, is a corresponding peak value of dynamic shear stress acting on a horizontal plane atdepth z. g is the gravitation. σ v and σ v indicate the total and effective vertical stress at depth z,respectively.

In the present case, the maximum acceleration is inferred to have been of the order of 150 gal, based on the data recorded at Cymbulif about 5 km southwest of the epicenter as indicated in Figure4. The depth at which liquefaction was triggered may be taken typically as being about 10 m. Thus,

the total and effective vertical stress are computed to be σ v = 160 kN/m2

and σ v = 110 kN/m2

,respectively, assuming the unit weight is γ t = 15 kN/m3. With these values inserted in Eq. (2) themaximum stress ratio is obtained as

This value may be taken as being equal to the maximum stress ratio required to trigger theliquefaction in the loess deposit. The strength as above may be alternatively expressed in terms of

445




Table 1. Results of post-earthquake stability analysis.

Average depth Average slope Residual strength Cone resistance

Site H (m) α (degree) Su(kN/m2) qc(kgf/cm2)

Sharara (A-A section) 10 9 13.5

Firma 7 13 14.5

1-May 10 12 8.0 3∼6

Okuli 15 0.5 2.0 1∼4

the cyclic stress ratio causing liquefaction in 20 cycles of load application. For this transformation,the mean effective confining stress or the confining stress in the equivalent triaxial test conditionσ o,is calculated asσ o = (1− 2K o)σ v/3= 2/3 ·σ v where K o = 0.5 is the coefficient of earth pressure atrest. In view of the shock-like time history of the seismic shaking as shown in Figure 5, a correctionfactor of 0.55 may be used to obtain an equivalent amplitude of cyclic axial stress, σ d required tocause liquefaction in 20 cycles in the triaxial loading test (Ishihara, 1977). Thus, in terms of thecyclic stress ratio causing liquefaction in 20 cycles, the value in Eq. (3) is expressed as

This value is approximately equal to the cyclic strength of Toyoura sand compacted to a relativedensity of about 30%.

8 STABILITY ANALYSIS FOR FLOW SLIDE

In all the slides in the Gissar area as described above, it is apparent that the flow-type failure took place following the triggering of liquefaction. By considering the geometry of the mud flow it is possible to conduct post-earthquake stability analysis and to back-estimate the residual strengthor steady-state strength of the liquefied loess material (Ishihara 1990). This analysis consists of

back-calculating values of undrained shear strength of soils, along a known sliding plane, which isrequired to maintain the deformed configurations of the slopes in equilibrium. This can be achieved merely by putting the factor of safety equal to unity in the conventional formula of stability analysis.In view of the fact that the strength in the steady state is known not to be affected by the confining

pressure, the residual strength is assumed to take a constant value all the way along the sliding

surface. The values of the residual strength, Su, back-calculated for each of the typical crosssections in Figure 9, 11, 14 and 22 are presented in Table l. It may be seen that the residual strengthin the slides in Sharara and Firma is of the order of 15.0 kN/m2, whereas the strength in May 1 and 2. The range of appropriate Okuli is as small as 2.0 to 8.0 kN/m2 values in the cone penetrationresistance obtained at the sites of May 1 and Okuli is read off from the data in Figures 16 and 20 and listed in Table l. With an aim of establishing an empirical correlation between the cone resistanceand undrained residual strength of silty sands, the results of several case history studies wereshown by Ishihara et al. (1990) in a diagram in which back-calculated value of undrained strengthare plotted versus the cone tip resistance obtained at the corresponding sites of landsliding. Thisdiagram is shown in Figure 26, on which the values of Su and qc obtained in the present study aresuperimposed. It may be seen that the present data plot within the range of the previous field data.

9 CONCLUSIONS

At the time of the 1989 January 23 earthquake in the suburb of Dushanbe, Republic of Tajikistan,liquefaction developed in the Aeolian loess deposit and resulted in a series of landslides accompa-nied by a large-scale mudflow. The occurrence of liquefaction is attributed to the highly collapsiblenature of porous loessal deposits which had been wetted to a depth of about 15 m by water used for

446




Figure 26. Correlation between residual strength and cone resistance.

agricultural irrigation. The windlaid loess consisting mainly of silt-sized soil is thought to have beenin a barely stable state on the verge of hydraulic collapse before the earthquake, and this lead easily

to a complete collapse in a form of liquefaction upon being further subjected to seismic shaking.The complete slumping and long-distance flowage of the mud flow on the nearly flat ground wasexplained by the fact that the loess-forming silt is of low plasticity with a plasticity index of about10. The result of the post-stability analyses made for four sites of landslides in Gissar indicated themobilized undrained residual strength of the silt to have been probably in the range between 2.0and 15.0 kN/m2 which approximately coincides with similarly evaluated values of residual strengthin other case studies of flow failures.


takaji kokusho-earthquake geotechnical case histories for performance-based design-crc (2009)

Documents