Takaji Kokusho-Earthquake Geotechnical Case Histories For Performance-Based Design-CRC (2009)

EARTHQUAKE GEOTECHNICAL CASE HISTORIES
FOR PERFORMANCE-BASED DESIGN
2009 Taylor & Francis Group, London, UK
Earthquake Geotechnical Case

Histories 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 = 7.2
Courtesy 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, India
Printed and bound in Great Britain by TJ International Ltd, Padstow, Cornwall
All rights reserved. No part of this publication or the information contained herein
may be reproduced, stored in a retrieval system, or transmitted in any form or by any
means, electronic, mechanical, by photocopying, recording or otherwise, without
written 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/Balkema
P.O. Box 447, 2300 AK Leiden, The Netherlands
e-mail: Pub.NL@taylorandfrancis.com
www.crcpress.com 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
About the Editor
Introduction
The 26 Case Histories
VII
IX
XI
XIII
Geotechnical performance of the Kashiwazaki-Kariwa Nuclear Power Station

caused by the 2007 Niigataken Chuetsuoki earthquake
T. Sakai, T. Suehiro, T. Tani & H. Sato
2006 large-scale rockslide-debris avalanche in Leyte Island, Philippines

R.P. Orense & M.S. Gutierrez
31
Slope failures during the 2004 Niigataken Chuetsu earthquake in Japan

T. Kokusho, T. Ishizawa & T. Hara
47
Slump failure of highway embankments during the 2004 Niigataken Chuetsu earthquake
K. Ohkubo, K. Fujioka & S. Yasuda
71
Fill slope failure of the Takamachi housing complex in the 2004

Niigataken Chuetsu earthquake
S. Ohtsuka, K. Isobe & T. Takahara
83
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
95
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
109
Quay wall displacements and deformation of reclaimed land during

recent large earthquakes in Japan
T. Sugano
119
River dike failures during the 1993 Kushiro-oki earthquake

and the 2003 Tokachi-oki earthquake
Y. Sasaki
131
Ground failures and their effects on structures in Midorigaoka

district, Japan during recent successive earthquakes
K. Wakamatsu & N. Yoshida
159
Behaviour of SCP-improved levee during 2003 Miyagiken-Hokubu Earthquake

A. Takahashi & H. Sugita
V
177

Preface
Performance-Based Design (PBD) is increasingly employed recently in structural design of buildings and infrastructural facilities in many countries. However, PBD has not yet been established
sufficiently in geotechnical engineering practice. Seismically induced ground deformation essential 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-strain
relationship 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 earthquakes
are 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 engineering structures. The next major challenge for the geotechnical engineering community is to shift
from the limit state design to the strain/deformation evaluation based on time/frequency-domain
calculations 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 the
conventional methods, uncertainties involved in the PBD become considerable in terms of seismic
input, large-strain soil properties, variability of soil properties, optional parameters in numerical
analyses, 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 analytical results is a sort of benchmark case histories with well-documented geotechnical and seismic
conditions. As one of the activities of TC4, ISSMGE in the 20052009 term, it has been planned
to publish a case history volume accommodating well-instrumented geotechnical and earthquake
data 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 Earthquake
Geotechnical 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, liquefaction, 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, raft
foundations, improved soils, etc.
The case histories included here are limited only for prototypes, excluding model tests, accompanied by earthquake records obtained nearby, geotechnical surveying data, structural data and other
pertinent parameters. Records, measurements and associated information relevant to case history
studies are accommodated as much as possible. Not only case histories of drastic failures but also
comparable 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 case
history 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. The
volume, 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. Tomohiro
Ishizawa, Chuo University, who contributed in formatting data files in CD-ROM is also gratefully
acknowledged. The great cooperation from a number of reviewers who generously served their
precious time in reviewing and editing the manuscripts both technically and grammatically is
gratefully 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 reliability
for 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 Electric
Power Industry (CRIEPI) in Japan in 1969.
Director of Siting Technology Division in CRIEPI in 1989
Professor 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 (20022003).
Chairman of Asian Technical Committee No.3 (Geotechnology for Natural Hazards)
of ISSMGE (19982005).
Chairman of Technical Committee No.4 (Earthquake Geotechnical Engineering &
Associated Problems), ISSMGE (since 2006)
IX

Introduction
In this case history volume, 26 case histories are accommodated in a sequence from newer to
older seismic events (CH01CH26) as listed in The 26 Case Histories. As shown in the table, the
subjects of the papers include slope failures, liquefaction, lateral spreading, soil subsidence, failure
of 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 motions
in liquefied soil.
In most of the papers, photographs and diagrams are originally prepared in full color and they
are sometime not clear enough to understand well in the black & white printed version. The readers
are therefore recommended to refer to the full color copies of the attached CD-ROM along with
the printed version.
Individual authors tried their best to make their case histories as quantitative as possible in terms
of earthquake ground motion records, geotechnical investigation data, etc. Some of the papers are
accompanied by digital data stored in the CD-ROM as indicated in the table. They are formatted in
the same order as in the table of contents and the associated data files can easily be accessible by
clicking 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, appropriate
literatures, 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 clicking
the 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 the
acknowledgments of their documents.
Takaji Kokusho
Chairman, TC4 (Earthquake Geotechnical Engineering and Associated Problems), ISSMGE
Professor of Civil Engineering Department, Chuo University, Tokyo, Japan.
XI
Figure 5.
Damage of equipment.
5
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
(2) In the Arahama side area, sand boils are observed on the TP 3 m ground level along the Unit 1
emergency intake channel, on the north side of the Unit 4 intake, and on the inland side of the
Unit 3 and 4 reactor buildings. In the Ominato side area, sand boils are observed on the south
side of the outlet and around the feed-water tank and parking lot. Representative sand boils are
shown 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.
channel.
Sand boil near Unit 1 emergency intake
Figure 12.
intake.
Figure 13.
Distribution of sand boils and cracks (Units 1-4, Arahama side area).
9
Sand boil on the north side of Unit 4
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 14, Arahama side area).
Figure 25.
Subsidence of ground near Unit 1 seawater heat exchanger building.
13
Figure 31. Thickness and extent of backfill soil.
16
Figure 32.
Groundwater level distribution (October/November 2007).
18
19
Figure 33.
Cross section of backfill and groundwater level observed in October/November 2007 (Arahama side area).
20
Figure 34.
Cross section of backfill and groundwater level observed in October/November 2007 (Ominato side area).
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. Doublecell 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
Figure 41.
Backfill soil grain size distribution curves.
Double amplitude of strain DA=1%

DA=2%
DA=5%
Excess water pressure ratio =95%
Cyclic shear stress ratio d / 2 0
0. 50
0. 45
0. 40
0. 35
0. 30
RRL20=0. 26
0. 25
0. 20
0. 15
0. 10
0. 05
0. 00
10
100
1000
Number
Number
of cycles
of cycles
of loading
of loading
Nc Nc
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 the
Kashiwazaki-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 1050 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 boils
and 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 deformation was observed. Laboratory tests were conducted using disturbed samples. Basic
characteristics 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 backfill
soil. Efforts are being made toward quantitative evaluation of subsidence near the building due to
the interaction between the building and soil, subsidence in unsaturated soil around the building
and 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 to
have 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 with
estimated volume of about 2025 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 slide
which is equivalent to 2.5 times the mean rainfall amount for the whole month of February, must
have played a major role on the instability of the slope. Minor earthquakes, the strongest of which
had a surface-wave magnitude of Ms = 2.6 and whose epicenter was determined to be 23 km west
of 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 landslide
generated 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 eastern
side 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. This
makes 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 is
worthy 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 mechanisms

have 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 one
of the most active geologic settings in the world. The locations of major fault lines in the Philippines
are shown in Fig. 2a. The Philippine Fault Zone (PFZ), a 1,200 km-long active fault system which
transects 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
(Ms = 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
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).
1500
250
200
1200
150
900
Cumulative
Rainfall
100
600
Daily
Rainfall
300
50
0
1/1
Cumulative Rainfall (mm)
Daily Rainfall (mm)
Landslide
0
1/11
1/21
1/31
2/10
2/20
Figure 4. Daily rainfall from 01 January28 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 February 2006, the area was drenched by 687.8 mm of rainfall, with the peak daily rainfall of 171.0 mm
occurring on 12 February (see Fig. 4). Because of concerns regarding possible flooding and landsliding, some residents of Guinsaugon left their homes and evacuated to safer places. After this, the
rainfall intensity somewhat subsided, with an intensity of 32.4 mm recorded the day prior to the
landslide. As a result, the people who sought refuge came back to their homes to resume normal
daily 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
Maasin, Southern Leyte Station (MSLP)

Digital Counts
10000
E-W Component
5000
0
Digital Counts
10000
N-S Component
5000
0
Digital Counts
10000
U-D Component
5000
0
103600H
103700H
103800H
103900H
104000H
Local Time
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 was
made with Mr. Virgilio Monghit who saw the slide as it occurred from a location only about 20 m
from the left side of the scarp (viewed from below the slide). Mr. Monghit owns the land close to
the scarp, which he planted with banana and coconut, and he was working on his land on the day
of the slide. The witness said that a few minutes before the slide, he felt ground shaking from an
earthquake. 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 the
mountain, 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 the
mountain created three clusters of rock mass which moved downhill in a wave-like fashion before
disintegrating and spreading further. Apparently, some large pieces of rocks and boulders were
observed flying and hopping above the ground.
Another survivor of the slide confirmed this observation. Once the rock mass started to disintegrate 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, the
ground 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 witness
Tony 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 explosion
and saw the top of the mountain came sliding down. There was a great cloud of dust and large wall
of earth moving towards him, so he turned and ran away as fast as he could, looking back only once
to check if the wall was near.
3.4 Post-failure behavior
Distinct element modeling of the slide performed by Gutierrez (2006) indicated that the rockslide
initially 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 face
of the mountain. The falling rock then slid along the bedding plane at the base of the overhanging
rock, 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 large
separations along existing fractures appear to subdivide the falling block into several clusters before
sliding and disintegrating. Initially, the blocks at the top of the scarp moved mainly downwards
along 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 wide
expanse covered by the moving debris can be attributed to the saturated rice field which served as
lubricating layer for the portion of the rock avalanche that extended to the valley floor. Based on
37
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 understood. 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 pressurization and connectivity of the faults close to Guinsaugon. In addition to causing potential fault
activation, water inflow from rainfall could have increased the unit weight making the overhanging
rock 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 residents
claimed 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
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
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 various
factors influencing failure modes, including geology and earthquake data. Then, six representative
slope 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. In
order to facilitate better understanding the background of the failures, pertinent data concerning
geology, 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 stones
and 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 parallel
in the north-south direction as indicated in Fig. 2, among which rivers are flowing almost in the
same direction from north to south. Mountains are about 500 m at the highest, and the slopes
are 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
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 MJ = 6.0 of 2004 Niigataken Chuetsu
earthquake.
Epicenter (degree)
Earthquake
MJ : JMA
magnitude
Occurrence
(y/m /d/time)
Latitude (N)
Longitude (E)
Depth (D) (km)
Main shock
After shock
After shock
After shock
After shock
6.8
6.3
6.0
6.5
6.1
2004/10/23/17:56
2004/10/23/18:03
2004/10/23/18:12
2004/10/23/18:34
2004/10/27/10:40
37.29
37.35
37.25
37.31
37.29
138.87
138.98
138.83
138.93
139.03
13
9
12
14
12
Table 2. JMA and K-net strong motion recording stations during 2004 Niigataken Chuetsu earthquake.
System
Site (Code)
Latitude (deg)
Longitude (deg)
MS Epicenter
distance (km)
Max. Acc
(EW) gal
JMA
JMA
JMA
K-net
Yamakoshi
Ojiya
Kawaguchi
Ojiya (NIG 019)
37.327
37.310
37.268
37.306
138.890
138.795
138.861
138.790
4.3
7.0
2.5
7.4
723
898
1676
1308
The digitized acceleration time histories of the main shock recorded in the damaged areas
including 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 the
P/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
05
510
1024
2433
3338
110
200
260
310
590
300
690
1700
1700
2100
1.8
1.8
1.7
1.9
1.9
Fine sand
Sandy gravel
Weathered mudstone
Slightly weathered mudstone
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 versus

slope 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 schematically
illustrated in Fig. 7;
Type-A: Deep slips parallel to sedimentation planes (parallel dip slip or daylight dip slip), in gentle
slopes 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 along
the slip plane. The displaced soil volumes were very large, translating ground surface with little
disturbance.
Type-B: Shallow slips not parallel to sedimentation planes (infacing dip slip) at steep slopes (>30
degrees). 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.
Slope
Higashi-Takezawa
Yokowatashi
Haguro-tunnel
Entrance
Naranoki
Kajigane
Musikame
Incident energy at baserock EIP (KJ/m2 )

Latitude Longitude
North
East
Main shock Aftershock Aftershock Aftershock Aftershock
MJ = 6.3 MJ = 6.0 MJ = 6.5 MJ = 6.1
Type (degree) (degree)
MJ = 6.8
A
A
B
37.304
37.330
37.329
138.906
138.827
138.895
428.6
382.4
398.3
91.7
50.8
95.9
21.0
22.4
18.9
143.0
103.1
134.2
27.4
14.8
23.9
B
C
C
37.322
37.309
37.349
138.912
138.899
138.886
393.8
430.8
356.1
107.8
89.9
91.3
18.3
21.2
16.7
139.7
141.0
121.1
26.6
25.2
20.7
in which M is the Richters Magnitude, although MJ = 6.8 (MJ is Japanese Earthquake Magnitude
almost equivalent to the Richters Magnitude) is used in the computation. The comparison of
the computed values with measured energies from vertical array records near epicentral areas
for the 1995 Kobe earthquake (Kokusho and Ishizawa 2007) and also for the 2004 NiigatakenChuetsu earthquake (Kokusho et al. 2008) has already confirmed that these simple equations can
approximate 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 is
about 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 MJ = 6.0 had already
occurred. Due to the absence of similar data just before the earthquake, air-photographs taken
in 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 two
time 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 2829 years before the
earthquake. 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 than
those of the 6 large slope failures.
The digitized data of pre/post-earthquake elevations for the 6 slopes are stored in the CD-ROM
of this volume. The horizontal and vertical coordinates are taken as the Global Coordinate VIII. The
coordinates (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 reference
nodes are given for each slope. Furthermore, in order to know the change of cross-sections of the
failed 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 observation
indicated 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 15. Idealization of sliding soil mass by a

rectangular block in Higashi-Takezawa.
Figure 14. Cross-sectional change of Higa

shi-Takezawa slope before and after the earthquake.
Figure 16. Cross-section of Higashi-Takezawa slope along (simplified after Yuzawa-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 before
the major restoration works are shown, in which soil types, water tables, estimated depths of the
slip plane are indicated. A part of the prefailure surface in the down-slope side is not available
in this chart, and the post-failure surface reflects some initial restoration works. Nevertheless, it
is readily understood that the 300-meters long soil block translated (along the slip plane of the
mudstone of 20 degrees on average) by 100 m horizontally and the front deformed and collided
56
Figure 30. Contour map (a) and air-photograph (b) of Haguro Tunnel entrance slope after the earthquake
(based on photographs taken in 28, Oct. 2004) (JSCE 2007).
before
400
after
slip plane
15-15
(r)av 111.0
300
200
400
60.2
13-13
Altitude (m)
200
36.1
4.52
11-11
300
27.6
170.0
200
400
h 50.7
8.18
300
400
24.55
94.5
232.0
09-09
300
200
Figure 32.
400
300
200
100
Horizontal distance (m)
Idealization of sliding soil mass by a
0 rectangular block in Haguro Tunnel entrance slope.
Figure 31. Cross-sectional change of Haguro

Tunnel entrance slope before and after the earthquake.
and y = 147799 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 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 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 from the outside
nodes. Thus, detailed topographical changes due to the slope failure can be drawn from the digital
data in the CD-ROM.
62
Figure 48. Contour map (a) and air-photograph (b) of Mushikame slide after the earthquake (based on
photographs taken in 28, Oct. 2004) (JSCE 2007).
before
300
after
slip plane
1414
250
200
300
(r)av = 113.4
21.2
h = 44.0
82.0
1212
Altitude (m)
250
13.6
120.0
200
7.98
29.9
300
1010
145.0
250
15.1
200
300
190.0
0808
Figure 50. Idealization of sliding soil mass by a

rectangular block in Mushikame slope.
250
200
350
300
250 200 150 100

Horizontal distance (m)
50
Figure 49. Cross-sectional change of Mushikame

slope before and after the earthquake.
and y = 149999 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 every 0.5 m
apart are also given. Because of the errors involved and some other reasons, even the nodes outside
the failed slope show some non-zero elevation changes. In order to differentiate the outside nodes,
69

Slump failure of highway embankments during the 2004

K. Ohkubo & K. Fujioka
Nippon Expressway Research Institute Co. Ltd
S. Yasuda
Tokyo Denki University, Saitama, Japan
ABSTRACT: The 2004 Niigataken-chuetsu earthquake caused serious damage to highway

embankments of Kan-etsu Expressway in Japan. Severe slides of embankments occurred in a
hill zone. Some excess pore water pressure had to be generated in the saturated part of the fills
and 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 embankments and culverts in gravelly grounds. In addition, large settlements of culverts and lateral
spread of embankments occurred in clayey and sandy grounds. The settlement of the embankment
seemed to be induced due to the reduction of shear modulus of the filled materials and foundation
grounds.
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. Several
earthquakes, such as the 1978 Miyagiken-oki, the 1987 Chibaken-toho-oki and the 1995 Kobe
earthquakes 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 embankments 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-ETSU

EXPRESSWAY 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 the
earthquake. Total length of the closed expressways was 580 km. Emergency treatments were applied
to the damaged expressway embankments by filling, placing and spreading. Then all expressways
were able to open for emergency vehicles about 19 hours after the earthquake because no serious damage induced for expressway bridges and tunnels. About 13 days after the earthquake all
expressways 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
Collapse of embankment
Flow of
collapsed soil
Ground water table
(a) Type 1: Serious slide of the embankment on the sloping ground
Settlement of road surface

C-Box
Dense ground
(b) Type 2: Settlement of the embankment on the level ground without the deformation of
the ground
Settlement of road surface
C-Box
Separate at joint C-Box
Lateral flow
Lateral flow
Settlement ofground
Soft ground
(c) Type3: Settlement of the embankment and theculvert on the level ground with
the deformation of the ground
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 fill. 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 demonstrate
the 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 11.
Soil cross section between Yamamotoyama Tunnel and Yamaya PA.
Figure 12.
C-Box Kawaguchi 11.
Figure 14.
Soil cross section adjacent to the culvert box at C-Box Kawaguchi 11
Figure 13. Differential

embankment and culvert at.
77
settlement
between
100
Percent finer by weight (%)
90
226-2
226-2-S 1
226-2-S 3
226-2-S 4
80
70
60
50
40
30
20
10
0
0.001
Figure 15.
0.01
0.1
1
Grain size (mm)
10
100
Grain-size distribution curves of fill soils near C-Box Kawaguchi 11.

100
Differential
settlement between
embankment and
culvert (cm)
80
60
40
20
0
Ojiya IC
Settlement of
culvert (cm)
226.00
226.50
227.00
227.50
228.00
228.50
229.00
229.50
230.00
230.50 231.00
KP
226.50
227.00
227.50
228.00
228.50
229.00
229.50
230.00
230.50
100
80
60
Separation of joints
of culvert boxes
40
20
0
226.00
231.00
Ojiya IC
KP
C-Box
C-Box
Kawaguchi 22 Ojiya 2
S1
S: Settlement of embankment
S1: Differential settlement
between embankment
and culvert
S2: Settlement of culvert
J: Separation of joint of
culvert boxes
S2
C-Box
Kawaguchi 11
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

Fill slope failure of the Takamachi housing complex in the 2004

S. Ohtsuka & K. Isobe
Department of Civil and Environment Engineering, Nagaoka University of Technology, Japan
T. Takahara
Graduate 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 shown
in Fig. 1), the ground deformed significantly in a fill area, and artificial fill slopes around the
complex suffered considerable collapse. Damage to houses caused by ground deformation had been
repeatedly observed in past earthquakes such as the 1978 Miyagiken-oki and the 1995 Hyogokennanbu earthquakes. Since flat land in Japan is limited, the cut and fill technique is widelly applied to
the 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 in
valleys 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 the
complex.
This paper reports on the ground disaster involving damage to houses, ground deformation
and slope failure at the Takamachi housing complex. The correlation between damage to houses
and fill 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 surface
wave 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 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 gravitytype 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
Damaged houses ratio[%]
100.0
Figure 9(c).
80.0
13 %
11 %
60.0
40.0
18 %
67 %
69 %
58 %
12 %
10 %
22 %
21 %
Crack
Fill & Crack
Danger
Warning
No damage
76 %
20.0
0.0
25 %
All
Fill
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 retaining
walls 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 concrete
retaining wall supporting the fill was translated downward. The slope failure was 40 m wide at
the top, 53 m long and 19 m wide at the bottom. Waterholes were observed in the center of the
collapsed 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 as
at 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.

ri
Bo
ng
I
Unit: m
8000
0 20 40
8300
3000
300
Surface before earthquake
Bank
5
ri
Bo
Surface after earthquake
II
ng
Collapsed sediment
10
Oyama layer
Slope deposit
0 20 40
15
20
Figure 13.
Uonuma layer
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 view
and 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 was
obtained as 1.30 for the collapsed soil and 0.90 for the intact soil. The increase in void ratio of
90
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): 233243 (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 PerformanceBased 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
Figure 2. A uplifted manhole in Kushiro Town.
Katsuragi
43 11 8 22 2
Kiba
N
100
Water channel
Nichii side
Sattsuru Beniya side
No.65
16 55 131 28 2
12
3
0
18
103
36
No.9
200 (m) Kushiro timberyard
: Manhole
Number shows uplift of manhole (unit:cm)
Figure 3.
Town.
Height of uplifted manholes in Kushiro
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 the
uplift of sewage manholes in Nakashibetsu Town and Shibetsu Town. Damage occurred in peaty
ground also.
97
100 km
Maximum acceleration (gal)

N-S : North-South direction
E-W : East-West direction
U-D : Up-Down direction
Sapporo[HKD180]
N-S : 67.7gal
E-W : 65.9gal
U-D : 23.7gal
Kitami[HKD054]
N-S : 53.5gal
E-W : 53.9gal
U-D : 32.7gal
Obihiro[HKD095]
N-S : 151.4gal
E-W : 192.6gal
U-D : 97.4gal
Ikeda[HKD092]
N-S : 434.7gal
E-W : 609.5gal
U-D : 215.3gal
Hiroshima[HKD182]
N-S : 66.2gal
E-W : 56.2gal
U-D : 23.6gal
Kushiro[HKD077]
N-S : 311.2gal
E-W : 407.1gal
U-D : 163.6gal
Tomakomai[HKD129]
N-S : 86.6gal
Mukawa[HKD126]
E-W : 72.9gal
N-S : 189.4gal
U-D : 32.9gal
E-W : 189.8gal
U-D : 54.1gal
Shizunai[HKD106]
N-S : 140.9gal
E-W : 186.9gal
U-D : 66.0gal
Bihoro[HKD055]
N-S : 51.3gal
E-W : 68.2gal
U-D : 26.2gal
Hiroo[HKD100]
N-S : 810.0gal
E-W : 972.5gal
U-D : 461.1gal
Urakawa[HKD109]
N-S : 186.3gal
E-W : 240.7gal
U-D : 84.4gal
Urahoro[HKD091]
N-S : 390.4gal
E-W : 375.5gal
U-D : 325.2gal
Shiranuka[HKD085]
N-S : 258.8gal
E-W : 276.6gal
U-D : 151.0gal
Chokubetsu[HKD086]
N-S : 738.6gal
E-W : 784.9gal
U-D : 271.6gal
Figure 7. Maximum ground surface acceleration recorded by K-net during the 2003 Tokachi-oki
earthquake.
Figure 8. Locations of damaged and intact manholes in Onbetsu Town following the 2003 Tokachi-oki
earthquake.
99
Percent finer by weight(%)
100
D60 =0.346mm
D50 =0.242mm
D30 =0.149mm
D10 =0.0136mm
Uc =25
rs =2.589g/cm3
80
60
40
20
0
0.001
0.01
0.1
10
100
Grain size(mm)
Figure 9. A manhole in Onbetsu Town uplifted

during the 2003 Tokachi-oki earthquake.
Figure 10. Grain-size distribution curve of boiled

soil for sewage pipes in Onbetsu Town.
A'
Ground surface
No.8 Pump No.7
station Z=1.81m
Level
10m
No.10
Z=4.97m
No.9
Z=5.17m
No.5
Z=2.56m
No.6
Z=2.20m
No.3
Z=3.24m
No.4
Z=2.90m
No.1
Z=3.45m
No.2
Z=3.53m
EX-No.2
B-No.6
H-6 B-1
SPT-N value
0
50
SPT-N valu
0
50
SPT-N value
0
50
5m
0m
Sewage pipe
D=200mm
Sewage pipe
D=350mm
-5m
0m50m25m
Z: depth of pipe
: Sand
: Gravel
: Fill
Figure 11.
: Silt
: Clay
Soil cross section and depth of sewage pipes along
: Peat
line A-A
v
v
B'
Ground surface
No.118
D-2
Z=4.57m
Z=4.99m
D-1
Z=5.35m
D-5
No.123
B-13
Z=5.10m
No.122
No.121
No.120
No.119
D-3
Z=5.06m
Z=5.21m
D-4
H-7 B-1
SPT-N value
SPT-N value
Level
SPT-N value
10m
: Volcanic soil
in Onbetsu Town.
H-8 B-1
050
50
50
vv
5m
vvvv
v v v v
vvvvv
Sewage pipe
D=200mm
0m
-5m
0m
50m
Z: depth of pipe
: Fill
Figure 12.
: Gravel
: Sand
: Silt
: Clay
: Peat
Soil cross section and depth of sewage pipes along line B-B in Onbetsu Town.
100
v
v
v
v
v
v
: Volcanic soil
Figure 13. Locations of damaged and intact manholes in Toyokoro district of Toyokoro Town following
the 2003 Tokachi-oki earthquake.
Figure 14. Locations of damaged and intact manholes in Moiwa district of Toyokoro Town following
the 2003 Tokachi-oki earthquake.
101
Depth of water table (m)
Rate of damage to sewage pipes due to

the Niigataken-chuetsu earthquake (%)
50
100
0
0
Figure 26. Uplifted pipe found during restoration

work in Nagaoka City.
10
Clayey ground
Figure 27. Relationship between depth of water

table and amount of damage to sewage pipes (Technical Committee on the Sewer Earthquake Countermeasures, 2005).
Table 2. Detailed soil investigation conducted at sites of damage and undamaged sites in Nagaoka City, Ojiya
City and Kawaguchi Town (Partially quoted from Technical Committee on Sewer Earthquake Countermeasures,
2005).
Surface
natural
deposits
up to the
depth of
pipes
Replaced
soil
Damage to
manholes
and road
Nakazawa in Nagaoka City
Sakuramchi in Ojiya City
Kawaguchi Town
Damaged
site
Not damaged
site
Damaged
site
Not damaged
site
Damaged
site
Not damaged
site
Soil
type
Silt, Silty
sand
Silt
Sandy silt
Sand with silt,

Sandy silt
Clayey
soil
Sandy soil,
Clayey soil
SPT-N
value
Soil
type
2 to 4
0 to 5
Sand
with
gravel
Sandy
gravel
Gravelly
sand with
fines
Gravelly
sand with
fines
Gravelly
sand with
fines
Sand with
fines
Water
level
GL-0.65m
Deeper
than pipe
GL-1.1m
GL-1.38m
GL-0.2m
GL-0.9m
SPT
N -value
11 14
Degree of
compaction
74%
78 to 82%
81%
Relative
density
38 to 41%
Uplift of
manhole
40 cm
No
8 to 20 cm
No
24 cm
No
Road
cave-in
30 cm
No
20 cm
No
23 cm
No
107

Fluidisation and subsidence of gently sloped farming fields

reclaimed with volcanic soils during 2003 Tokachi-oki
earthquake in Japan
Y. Tsukamoto & K. Ishihara
Department of Civil Engineering, Tokyo University of Science, Japan
T. Kokusho
Department of Civil Engineering, Chuo University, Japan
T. Hara
Department of Civil Engineering, Wakayama National College of Technology, Japan
Y. Tsutsumi
Institute of Industrial Science, University of Tokyo, Japan
ABSTRACT: The fluidisation and subsidence of gently sloped farming fields took place in Tanno
area of Kitami city in Hokkaido, Japan, during 2003 Tokachi-oki Earthquake, which occurred at
4:50 am on September 26, 2003. The area of 35 metres wide and 150 metres long subsided and the
water channel located at the downstream area was filled with erupted subsurface soils for a length
of about 1 km. 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. The subsurface layer
consisting of loosely dumped deposits of volcanic soil was found to have been liquefied and erupted
onto the surface, which caused subsidence during the earthquake. Multiple series of Swedish weight
sounding tests were conducted and the soil profiles at several cross sections were estimated. Based
on the soil profiles estimated from the field sounding tests, the subsurface liquefiable soil layer
was detected. The progressive sequence of fluidisation and flow of subsurface soil and associated
ground subsidence is discussed.
1 INTRODUCTION
The intense shaking struck the regions in Hokkaido during 2003 Tokachi-oki Earthquake, which
occurred at 4:50 am on September 26, 2003. The epicentre was located off the south-east coast of
Hokkaido and the exact location was given as 41.78N, 144,97E. The focal depth was 42 km, and
the earthquake magnitude was M = 8.0, as reported by Japan Meteorological Agency. A number of
river dikes in Tokachi region suffered from cracks, slips and subsidence, whose river mouths were
faced with the south-east coast of Hokkaido. The cracks and distortions of road pavements, uplift
of sewage manholes, settlements of buildings due to liquefaction, and minor lateral spreading at
several ports were also found.
Far from the south-east coast of Hokkaido, the fluidisation and subsidence of gently sloped
farming field occurred in Tanno area of Kitami city during the earthquake. The ground surface of
35 metres wide and 150 metres long subsided, and the water channel located at the downstream
area was filled with erupted fluidised debris for a length of about 1 km. It was apparent that
the subsurface layer of volcanic soil beneath the subsided farming field experienced intensive
liquefaction and erupted onto the surface, and all the subsurface deposits flowed downstream.
The reconnaissance field survey was carried out and the outcome of the survey is described in
detail below.
109
Figure 2. Topographical location of site.
Figure 3.
Strong motion data records observed at K-NET station in Kitami.
111
Figure 6.
Location 1.
Figure 7.
Location 2.
Figure 8.
Location 3.
Figure 9.
Location 4.
Figure 10.
Location 5.
Figure 12.
Location 7.
Figure 11.
113
Location 6.
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 equal
to 980 N (100 kg). At each stage of this load increment, the depth of penetration is measured. The
weight 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 fixed 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 empirical
correlations 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 Nsw -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 Nsw can also be directly converted to the relative density Dr of sandy soils,
(Tsukamoto et al. 2004), as follows,
where 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 are
identified in Fig. 19. The data of the sounding tests are listed in Tables 1 to 13 of APPENDIX. The
soil profiles at five cross sections are estimated as shown in Figs. 20 to 24. The weak subsurface
soil layer can still be detected. It would be reasonable to assume that the overlying unsaturated top
stratum at the upstream portion prohibited the underlying liquefied layer from blowing directly
116

Quay wall displacements and deformation of reclaimed land during

recent large earthquakes in Japan
T. Sugano
Port and Airport Research Institute, Yokosuka, Japan
ABSTRACT: To investigate, understand and prevent earthquake damage to port facilities, damage investigation, numerical simulations and model tests are powerful tools for improving initial
and restoration design procedures. However, it is difficult to account for global interpretation and
understanding of the failure mechanism of port facilities considering water-soil-structure interaction without actual cases of failure. In this paper, typical case histories of earthquake-induced
failures at two ports during three different earthquakes are introduced in detail.
1 SEISMIC EFFECT ON PORT STRUCTURES

In a modern society, it is very important to maintain the logistic system functioning immediately even after tremendous earthquakes. In Japan, an earthquake prone country, more than
1,000 commercial ports have experienced severe earthquakes. The experience in case of the 1995
Hyogoken-Nambu Earthquake (Kobe earthquake) in particular gave a lot of lessons to all concerned with port facilities and logistic industries (management, administration, operation, shipping,
shipper, contractor, etc.).
In port facilities, a quay wall is considered as main structures. As shown in Figure 1, seismic
waves are generated along a fault and they propagate through rock, traveling to the surface of the
bed rock at a site of interest. The ground motions then propagate through the local soil deposits,
reaching the ground surface and impacting port structures. In case of a quay wall, the earthquake
motion acts format the base and also behind the structure. In many cases, liquefaction occurs in
loose sand layers and reclaimed lands, and brings serious damage during earthquakes.
Figure 1.
Propagation process of seismic waves to a quay wall (International Navigation Association, 2000).
119
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 vertical
directions, 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, especially
Kobe City economy, for several years.
The Kobe Port covers an area about 6 km long by 12 km wide including two man-maid islands
as 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 15. Comparison of the acceleration time history of 1993 Kushiro-Oki and 2003 Tokachi-Oki
Earthquakes (at the Kushiro Port, ground surface).
5 CONCLUDING REMARKS
The aforementioned quay wall damage suggests that:
1) Damage to port structures is often associated with significant deformation of a soft or liquefiable soil deposit. Therefore, if any potential for liquefaction exits, implementing appropriate
remediation 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 collapses. Hence design method based on displacements and ultimate stress states are desirable over
conventional force based design methods for defining the comprehensive seismic performance
of port structures.
3) Most damage to port structures is the results of soil-structure interaction during the earthquake
shaking. Therefore, seismic analysis and design should also take into the account, both the
geotechnical and structural aspects of the port structures.
Based on the lessons from the damage during the 1995 Hyogoken-Nambu earthquake, a seismic
Performance-Based-Design methodology (Japan Port and Harbour Association, 1999) was introduced. In order to assess/evaluate the seismic performance, new simulation techniques need to be
introduced in the technical standards for port facilities in Japan. However, in practice, it is not
easy 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)
Figure 17.
EW
NS
UD
2003 Tokachi-Oki Eq.
Surface
Downhole
576
202
347
154
149
66
1993 Kushiro-Oki Eq.
Surface
Downhole
343
268
450
203
362
121
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 as
model test data and numerical simulation data, and feed them back to practice with appropriate
interpretation.
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: 135152.
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: 181207.
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 Geotechnical Aspects of the January 17 1995 HyogokenNambu Earth-quake: 119136.
International Navigation Association, 2000. Seismic Design Guidelines for Port Structures:.8. Rotterdam:
Balkema
Japan 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 Earthquake and the 2003 Tokachi-oki Earthquake. During the Kushiro-oki Earthquake in 1993, it was
found 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 case
showed the importance of drainage to prevent the infiltrated rainwater from accumulating within
a 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 was
visible at the surface. Evidence suggests that the cracks were caused by expansion of the bottom
width of the dike.
1 INTRODUCTION
Japan has suffered from recurrence of earthquake disasters since ancient days, and damages
have 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-oki
Earthquakes in Hokkaido.
Hokkaido in Japan has suffered from repetitive seismic damages to flood protection dikes especially from large scale earthquakes off its southeast coast which faces the subduction zone between
the 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 the
data obtained at locations along the Shiribeshi-Toshibetsu River after the 1993 Hokkaido-nanseioki Earthquake (Sasaki et al. 1997) which revealed the failure caused by stretching of the bottom
of the dike. Lowering the water level within the dike body was effective in mitigating earthquake
damage during the 1994 Hokkaido-toho-oki Earthquake. Valuable lessons have been gained also
from 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 Earthquakes
till 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 the
2003 earthquake, a very interesting type of failure was discovered. The bottom section of the dike
was stretched and cracks opened in the body of the dike at a section of the dike with no visible
abnormality on its surface.
Stretch type deformation is considered to adversely affect the water-cutoff performance of a
dike especially in raining condition, but its engineering features are not well studied. It is one
of the phenomena that should be elucidated in order to appropriately assess the flood protection
performances of a river dike in near future.
131
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
mode
Length
(m)
Ratio
(%)
Length
(m)
Ratio
(%)
1
2
3
4
5
6
375
8,085
700
4.1
88.2
30
1,122
7,371
1,601
0.3
11.1
Total
9,168
10,124
100
0.1
7.6
100
72.8
15.8
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 conditions
of 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 Kushiro
Retarding 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 from
its damage, and the sections damaged during the 1994 Hokkaido-toho-oki Earthquake. The sections shown by solid line are the damaged area during the 1993 Kushiro-oki Earthquake, and those
shown 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 history
The 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 is
the first class river that has a stream length of 154 km and a basin area of 2,510 km2 . The Kushiro
River 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 control
works 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 Kushiro
River 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 the
Retarding Basin and their construction history since 1952 are shown in Figure 7 (KDCO 1994). The
dike had a crest elevation of 9.3 m (dike height was about 7 m), a crest width of 8 m, a berm width
of 3 m, slope inclination of 1:2 and a base width of about 4550 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 Marsh
Kushiro 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 facilities
so that nature is preserved.
The Marsh is covered by peat layer at its surface, and the thickness of the peat deposit varies
from place to place. The thickness is over 4 m in the middle of the Retarding Basin and is about 3 m
in 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 Figure
8 (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
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 Figure
13. It was also recognized that traces of liquefied sand intruded into some of these cracks (See
Figure 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. The
sketches of other types result from a synthesis of the observation records at excavating pits, where
investigations were conducted at some selected points to confirm the cracks. Each type showed the
following 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 top
of the slopes. More cracks were found on the top of the riverside slope than that of the landside
slope 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 the
dike.
Compared to the crack depths estimated from the appearance of the ground surface immediately
after 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 cracks
that 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 the
dike 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 dike
A cross section of the dike including the foundation at 9 km 850 (the cross section of the largest
deformation) of the left side bank is shown in Figure 16. It should be noted that the bottom surface
of the dike subsided for about 2 m at most from the surrounding ground due to consolidation
settlement, 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 20.
Suspected failure process.
Figure 21.
Crest settlement vs. bottom bending.
145
Figure 24.
Restoration of left side bank.
Figure 25.
Soil improvement.
Table 3. PGA at Hirosato during the 1994 Hokkaido
Toho-oki Earthquake.
Direction
Dike
(gal)
Ground
(gal)
LG
TR
UD
157
189
43
303
406
187
A cross section of the improved ground is shown in Figure 25. A longitudinal transition area
was 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 to
quickly discharge rain water after the water seeps into the dike.
2.3.6 Hokkaido-toho-oki Earthquake
Six months after the completion of the restoration works for the sections damaged by the
1993 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 vertical
were observed at the Kushiro Meteorological Station, which corresponded to about 4560 % of the
PGA at the same station during the 1993 Kushiro-oki Earthquake. At Hirosato Observatory, a PGA
of 300400 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 the
ground 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
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 from
1987 to 2001.
The dike at this section was now about 7 m high above the surrounding ground, with a crest
width of 9 m, a bottom width of about 75 m and side slopes of 1:5.
150
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 the
foundation. 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 in
the excellent report by Kohno & Sasaki (1969) involving a Tokachi River dike failure during the
1968 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 although
the consolidation settlement was symmetric about the dike center. This implies the need for more
detailed study of the water retention characteristics of fill material.
From the damage morphology in the Kushiro Retarding Basin dike, it was found that actual
depths of cracks within dike were sometimes deeper than that detected from dike surface, and it
was found at some sites, that cracks were not limited to the ones that could be identified on the
surface 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 dike
surface. At this site, it was deduced from the pull out of sluice segments at the bottom of the dike
was 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 is
concluded 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 long
spread 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 materials
at an occasion of their failure are also hardly known exactly. Further, as damaging earthquakes
do 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 case
histories of past performance as described in this paper.
ACKNOWLEDGEMENT
The author is grateful to the Hokkaido Development Bureau for permission to use all the date
concerning 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 destructiveness during Kushiro-oki, January 15th of 1993, Earthquake, Report of the Building Research Institute.
(134): 1127.
Finn, L.W.D., SasakiY. and Wu G. 1997. Simulation of response of the Kushiro River dike to the 1993 KushiroOki and 1994 Hokkaido Toho-Oki earthquakes. Proc., 14th International Conference on Soil Mechanics
and Foundation Engineering. (1): 99102.
Hokkaido Development Bureau (HDB) 1997. Report of research committee on damage caused by the Kushirooki 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): 10031015.
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): 924.
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.: 235240.
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.: 6168.
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-Nagoya97: 553558.
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 Earthquake Geotechnical Engineering during 14th International Conference on Soil Mechanics and Foundation
Engineering. Hamburg/Germany: A. A. Balkema,: 307316.
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
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 10. Rupture and separation of connection between

manhole and a sewage pipe during the 1993 Kushiro-oki
earthquake (Courtesy of Kushiro City Government).
Figure 9. Deformation of a sewage
pipeline 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 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 liquefaction 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 liquefaction 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 separated 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 from
the figure that the areas of the heaviest concentration of the damage during the four earthquakes
172
The National Research Institute for Earth Science Disaster Prevention Science and Technology Agency
(NIEDSTA). 1993. Prompt Report on Strong-Motion Accelerograms 41: National Research Institute for
Earth Science and Disaster Prevention.
(NIEDSTA. 1994. Prompt Report on Strong-Motion Accelerograms 44: National Research Institute for
Earth Science and Disaster Prevention.
(NIEDSTA). 2003. Prompt Report on Strong-Motion Accelerograms 112: National Research Institute
for Earth Science and Disaster Prevention.
Wakamatsu, K. & Yoshida, N. 1995. Ground deformations and their effects on structures in Midorigaoka
Districts, Kushiro City, during the Kushiro-oki earthquake of January 15, 1993. Proceedings of the 5th
U.S.-Japan Workshop on Earthquake resistant Design of Lifeline Facilities and Counter measures for Soil
Liquefaction. Snowbird, U.S.A.: NCEER.
176
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 (with
SCPs) is smaller, and (2) SPT-N values in the sandy soil layers at the berm (Borehole B-5) are
slightly 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. As
mentioned in the above section, the levees distant from the river mouth (about 8 to 17 km from the
182
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, 15471548 (in Japanese).
National Institute for Land and Infrastructure Management and Public Works Research Institute. 2003. Preliminary Report of the 2003 Miyagiken-Hokubu Earthquake, Civil Engineering J., Vol.45, No.9, 1015 (in
Japanese).
National Institute for Land and Infrastructure Management, Public Works Research Institute and Building
Research Institute. 2005. Report of the 2003 Miyagiken-Hokubu Earthquake, 77pp (in Japanese).
Sekizawa, S., Sato, S., Okada, S. & Hiratsuka, T. 2005. Damege simulation of Narusegawa river dike during the
2003 Miyagiken-Hokubu Earthquake, Proc. 40th Japan National Conference on Goetechnical Engineering,
20072008 (in Japanese).
184
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

mainshock
aftershock
2003/7/26
0:13
7:13
16:56
5.5
6.1
5.3
12.8
12.9
13.3
5 (Kanan)
6 (Kanan)
6 (Kanan)
National Inst. (2003)
The 726 Eq., on the other hand, caused intensive damage in a relatively small area of Miyagi
Prefecture. More than 1,000 houses had totally collapsed; more than 10,000 houses had partially
collapsed. 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 failures
were 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 the
same 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 away
from the epicenter, as shown in Figure 1.
More than 180 landslides, slope failures and rock failures occurred around the Asahiyama Hill
during the 726 Eq. (Irasawa, 2003). A landslide with similar magnitude and configuration to the
Dateshita landslide occurred after the main shock at Nishisaruta, Kanan-cho. The Nishisaruta
landslide, on the other hand, was very close to the source area of the earthquake as shown in
Figure 1. Although the angle of the original slope was steeper than that of Dateshita landslide, the
average 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 the
earthquake. Therefore, the collapsed soil possibly contained much water originally.
Traces of mud splashes were observed at some places. Figure 4b) shows traces of mud splashes
remaining 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 a
rice field as shown in Figure 4d). The traces evidenced that the flow speed of the slide was not
slow. 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 soil
snaked along the slope down to my house. My wife, who left the house immediately after the
shaking, was buried to the waist. Another witness who lived in the house that was broken by the
slide also said: After the shaking intensified, a propane tank that was set outside the house wall
shot into the house, breaking the wall with mud; a supporting column of the house sounded as
though it had broken. Then, I got out of the house after shutting off the electricity. It took about
6090 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 distance
of about 120180 m over about 6090 seconds. Therefore, the averaged velocity is estimated to
be about 1.33.0 m/s. Konagai (2003) also suggested that the velocity at the center of the stream
line was about 67 m/s based on kinematic analyses of traces of mud splashes. Moreover, these
evidences 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 the
observation of traces of mud splashes on the trees along both sides of the slide.
188
189
Figure 3.
Soil profiles and the results of sounding tests (courtesy of Miyagi Prefectural Government).
Table 4. Summary of physical and mechanical properties for undisturbed samples at Nishisaruta landslide.
d (g/cm3 )
sat (g/cm3 )
w(%)
Gs
e
k (g/cm3 )
G0 (kPa )
f (degree)
p (degree)
CSR
Dry density
Wet density
Water content
Specific gravity
Void ratio
Coefficient of permeability
Compression index
Swelling index
Initial shear modulus
Internal friction angle
Phase transformation angle
Cyclic shear stress ratio (N = 20, DA = 5%)
Figure 15.
1.35
1.78
32.1
2.71
1.01
9.2 102
1.42 102
6.27 103
7.65 104
43.8
26.0
0.34
Cyclic behavior of collapsed soil at Nishisaruta landslide.
isotropic consolidation tests, CD tests, stress-controlled undrained cyclic shear tests with constant
stress amplitudes, and strain-controlled undrained cyclic shear tests with gradually increasing strain
amplitude. Shear tests were performed with a conventional triaxial test apparatus.
Next we address cyclic undrained deformation behavior. Figure 15 shows the cyclic stress-strain
relation 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 was
50 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 same
as that with the B value of 0.95 (Tsukamoto et al., 2002). The mean effective stress gradually
decreased 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 behaviors
reflected normal cyclic behavior of fine-graded sand.
Kokusho et al. (2004) conducted undrained triaxial tests with undisturbed samples under saturated 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: the
confining 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 and numerical simulations, features of the Nishisaruta
landslide are summarized as follows:
(1) The subsurface soil of the slope was a fill with fine-graded sand, which originated from
sandstone on the hill.
198
(2) The main failure of the slide occurred at the upper part of the slope. The slope angle at the
collapsed 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 the
collapsed 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 the
lower rice field. The residual strength, after undrained cyclic shear loading, of the collapsed soil
was much larger than that of the collapsed soil at Dateshita (Kokusho et al. 2004). Therefore, the
steeper slope angle here possibly allowed the detached soil to spread widely.
The numerical simulation conducted by the present authors is available for understanding the
failure time sequence (Uzuoka et al, 2005). The simulations suggested that the saturated fill liquefied 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 mechanical
soil tests and preliminary numerical simulations were introduced. We summarized the results as
follows.
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 landslide 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. Saturated fill liquefy during the earthquake because the maximum horizontal acceleration at the surface
was estimated to about 300 gal. Moreover, even unsaturated fill was presumably fluidized during
shaking, losing the initial shear strength and spread on the rice field.
During the 726 Eq., a landslide with similar magnitude and configuration to the Dateshita
landslide occurred after the main shock at Nishisaruta, Kanan-cho. The Nishisaruta landslide was
also one of the largest landslides induced by the 726 Eq. Some features were different from the
Dateshita 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 angle
of 20 30 ) was steeper than that of the Dateshita landslide. Rainfall was an important feature
of 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 principal
motion.
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 on
the 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 the
reconnaissance team on the earthquakes in the Japanese Geotechnical Society provided valuable
information 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 strong
motion records in the July 26 earthquake. Graduate and undergraduate students in the Geotechnical
Laboratory, Tohoku University, were very helpful for the site investigation and laboratory tests. The
authors 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): 4454 (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 Ground Motion waveforms 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): 160179.
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): 2532 (in Japanese).
Kawakami, H. Konishi, J. and Saitoh, Y. 1985. Mechanism of slope failures by the Naganoken-seibu earthquake
1984 and the characteristics of pumice, Tuchi-to-Kiso, JGS, 33(11): 5358 (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): 546561 (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.): 437444.
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): 115.
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: 910 (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, 20852086 (in Japanese).
Mishima, S. & Kimura, H. 1970. Characteristics of landslides and embankments failures during the Tokachioki
earthquake, Soils and Foundations, 10(2): 3951.
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): 4758.
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: 115132.
Sassa, K. Fukuoka, H., Scarascia-Mugnozza, G. and Evans, S. 1996. Earthquake-induced-landslids: distribution, motion and mechanics, Special Issue on Geotechnical Aspects of the January 17, 1995
Hyogoken-Nambu Earthquake. Soils and Foundations, 5364.
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):
93104.
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): 519528.
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): 8799.
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): 149163.
200
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
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)

Length of berth panel (m)
Live loading (kN/m2 )
Total live load, LL (MN)
Dead loading (kN/m2 )
Total dead load, DL (MN)
Earthquake coefficient,
Earthquake force, FE1 and FE2 (MN)
23.2
22.9
32.3
17.1
23.5
12.5
0.1
2.1
24.7
65.7
49
79.5
8.1
13.2
0.12
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 37. Typical soil borehole profiles in the dry cargo berthing area.
20 percent. The SPT blow counts (N-values) of these layers indicate that the soils are potentially
liquefiable under strong and sustained shaking. However, their depth and high fines content may
have inhibited liquefaction during the earthquake. Even without a detailed site response analysis,
it may be inferred that the upper layers of the soft natural soils were prone to amplify the transient
ground motions during earthquakes. This has also been proved through ground response analyses
by Sitharam and Govindaraju (2004) and Kumar (2006). It may also be said that the transient ground
221
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 41.
2008).
Soil profiles adjacent to the building and its relation to the liquefaction potential (After Dash et al,
obtained ground motion during the earthquake. Key information, particularly the ground motion
data, which helps in understanding the behavior of man-made constructions such as buildings,
bridges, dams and ports in the epicentral region, unfortunately were of poor quality and in some
cases were missing. The gap could somehow be filled by subsequent studies of available near field
velocity data and SRR data at a few other places.
The earthquake apparently did not produce primary surface rupture but induced liquefaction in
large areas. However, only a relatively small amount of data pertaining to liquefaction resulting
from the earthquake has been collected till date. Nevertheless, initial observation regarding the
size, orientation and spatial distribution of liquefaction related features suggest certain trends and
raise several important issues which will be certainly of interests to designers and engineers. Bhuj
earthquake has revealed that there is a need for identifying sites that are prone to liquefaction by
means such as microzonation. Microzonation mapping depending on the liquefaction resistance of
soils should be carried out on the basis of soil types and the types of structures that are vulnerable to
strong shaking. With good liquefaction potential and susceptibility maps as a starting point, public
officials and private property owners can make informed decisions about how to concentrate limited
224
resources to manage and reduce the risk in a developing country like India. These will also serve
as reliable tools for the performance-based design of geotechnical structures.
The important questions here are: how the new structures be designed and how safe are the
existing structures? Damage to various structures also brought into the light the shortcomings
of the existing codal provisions. Therefore, there is a need to improve the codal provision with
emphasis on extensive geotechnical investigations and understanding the seismic behavior of subsoils. This seems to be the first step that has to be taken to mitigate similar destruction during future
earthquakes in that area. Enforcement of applicable design codes for engineered constructions and
improvement of seismic resistance of non-engineered construction would now face unparallel
challenge, if such catastrophe are to be avoided in the future.
ACKNOWLEDGMENTS
The authors would like to acknowledge the following institutions and individuals whose data were
instrumental in shaping this manuscript: (1) Earthquake Engineering Research Institute (EERI),
California, USA for supplying the numerous data and figures included in this paper, (2) Kandla
Port Trust, Gujarat, India for delivering various soil data and original figures, (3) Prof. R.N.
Iyengar and Prof. T. G. Sitharam of Indian Institute of Science, Bangalore, India for supplying the
data on ground motion, (4) Dr. L. Govindaraju of Bangalore University, India for supplying the
digital strong motion record and various other soil data, and (5) Dr. N. Itagaki of Akita Prefectural
University, Akita, Japan for providing several field reconnaissance data of the earthquake.
REFERENCES
Bardet, J.P., Rathje, E.M., and Stewart, J. 2002. Ground failures and geotechnical effects (Ports), Bhuj, India
Earthquake of January 26, 2001reconnaissance report, EERI, California, USA: 101130.
Boulanger, R.W., and Idriss, I.M. 2005. New criteria for distinguishing between silts and clays that are
susceptible to liquefaction versus cyclic failure, 25th Annual Conference of the United States Society on
Dams: 357366.
Dash, S.R., Govindaraju, L., and Bhattacharya, S. 2008. On the probable cause of the failure of Kandla port and
customs office tower during the 2001 Bhuj earthquake, Proc. of the 14th World Conference on Earthquake
Engineering, Beijing, China, CD-ROM.
Dash, S.R., Govindaraju, L., and Bhattacharya, S. 2009. A case study of damages of the Kandla Port and
Customs Office tower supported on a matpile foundation in liquefied soils under the 2001 Bhuj earthquake,
Journal of Soil Dynamics & Earthquake Engineering, 29(2): 333346.
Earthquake Engineering Research Institute (EERI). 2001. Preliminary observations on the origin and effects
of the January 26, 2001 Bhuj (Gujarat, India) Earthquake, EERI Special Earthquake Report, California,
USA.
Hengesh, J.V., and Lettis, W.R. 2002. Geology and tectonic setting, Bhuj, India Earthquake of January 26,
2001reconnaissance report, EERI, California, USA: 722.
Idriss, I.M., and Boulanger, R.W. 2004. Semiempirical procedures for evaluating liquefaction potential during
earthquakes, 11th International Conference on Soil Dynamics and Earthquake Engineering, 1: 3256.
Indian Meteorological Department (IMD). 2002. Bhuj Earthquake of January 26, 2001, A consolidated
document, Government of India.
Indian Port Association. 2007. Coordination of business plans for major ports in India, Consolidated Port
Development Plan, Vol. 1, Main Report.
IS 1893, 2002. Criteria for earthquake resistant design of structures, Bureau of Indian Standard, New Delhi,
India.
Iyengar, R.N., and Raghu Kanth S.T.G. 2006. Strong ground motion estimation during the Kutch, India
Earthquake, Pure and Applied Geophysics, 163: 153173.
Kandla Port Trust. 2000. Advantage Kandla, Port of the Millennium, Ganghidham, Kutch, India.
Kumar, J. 2006. Ground response of Ahmedabad city during Bhuj Earthquake: a case study, Electronic Journal
of Geotechnical Engineering, Paper 2006-0657.
Mahindra Acres. 2000. Ninth cargo berth at Kandla, Preliminary Report on Analysis and Design to Kandla
Port Trust, Mahindra Acres Consulting Engineers Limited, Ganghidham, Kutch, India.
Malik, J.N., Merh, S.S., and Sridhar, B. 1999. Paleo-delta complex of Vedic Saraswati and other ancient rivers
of northwestern India, Indian Geological Survey Memoir, 42.
225

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 Mw = 7.7 occurred in the southeast coast of El Salvador on 13 January 2001, causing widespread damage to buildings and several kinds of civil
engineering structures due to ground shaking and earthquake-induced ground failures, including
several large-scale landslides. The most tragic among these landslides occurred on the steep northern flank of the Blsamo Ridge, where an estimated 200,000 m3 of soil slid. Once mobilized, the
landslide material behaved as semi-fluid mass and traveled northward an abnormally long distance
of about 700 m into the Las Colinas neighborhood of Santa Tecla and covered many houses, burying
more than 500 people. The slide materials involved pyroclastic deposits, i.e., silty sands and sandy
silts. 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 Mw = 7.7 occurred in the Republic of El Salvador in Central America. This was the first major earthquake of the third millennium and the
fifth destructive earthquake to affect the small Central American republic in 50 years. This was
followed 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 Salvadors total population, was
affected by the earthquake. Economic losses were set at US$1.6 billion, which is equivalent to 12%
of the countrys 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 landslides
occurred off the steep northern flank of Blsamo Ridge, about 2 km west of San Salvador (the
countrys 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 vertical
drop from the ridge to the terminus was about 160 m. The slide materials, estimated at about
200,000 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 the
volume of slide materials involved is rather unusual.
This paper focuses on the characteristics of the earthquake and strong motion records, as well
as the features of the Las Colinas slope, the characteristics of landsliding event, and its impact
on the environment. The type of soil involved and the conditions that led to in the long run-out
distance 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 possible
use 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 subducting
at 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. Due
to their shallow foci and coincidence with large population centers, these earthquakes have been
responsible 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 Blsamo Ridge, the northern
boundary of the Cordillera de Blsamo. The ridge lies west-southwest of San Salvador, with elevations up to 1100 m above sea level. The landscape is rugged, with the ridges rising by 200300 m
from 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 Blsamo is underlain by the Blsamo Formation, which consists of mafic
volcanic breccias, lavas, welded ignimbrites, and other well-indurated volcanic rocks. In the area,
the top of the Blsamo Formation is marked by weathered soil layer that contains enough clay
and fine-grained material that it poses a barrier to downward-draining surface layer that percolates
through the porous overlying, young volcanic deposits (Jibson and Crone, 2001).
The surface soil in the Blsamo Ridge consists mainly of volcanic tuff sediments, commonly
poorly consolidated young pyroclastic and epiclastic deposits (Schmidt-Thom, 1975). The brown
and yellowish medium to fine grained layers are referred to locally as Tobas color caf, while the
white deposit, the Tierra blanca, consists of light-gray to white, fine-grained pumice ash. Within
this ash are numerous large blocks of pumice which are commonly concentrated in layers, although
they 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 intense
rainfall and under earthquake shaking. Schmidt-Thom (1975) pointed out that these thick, poorly
consolidated volcanic deposits are especially sensitive to erosion by surface water, and the erosion
is particularly intense if the protecting overgrowth and, along with it, the soil cover were originally
lacking 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 Blsamo Ridge is high. Earthquake-induced landslides are
common, 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 distances
along 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 Meteorology
Department 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 (see
Table 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 the
dry 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
1999
2000
1958 mm
1504 mm
1454 mm
2037 mm
1303 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 southsoutheast of the capital, San Salvador (see Fig. 1). The source parameters for this earthquake, as
obtained by different agencies, are given in Table 2. It can be seen that they are remarkably consistent
in terms of size and depth. The epicenter of this earthquake, considered as among the strongest to
hit El Salvador in recent years, was located within the Caribbean plate above the subducting Cocos
Plate.
Figure 1. USGS focal mechanism solutions and CIG aftershocks distribution of the 13 January 2001 earthquake (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)
Depth (km)
Magnitudes
Agency
17:33:32
17:33:46
17:33:30
17:33:31
13.049
12.97
12.868
12.8
88.660
89.13
88.767
88.8
60
56
60
60
Mw = 7.7, Ms = 7.8, mb = 6.4

Mw = 7.7, Ms = 7.8, mb = 6.4
Mw = 7.7
Mw = 7.6
NEIC
HRV
CASC
ERI
NEIC National Earthquake Information Center.

HRV Harvard Geophysical Laboratory.
CASC Central America Seismological Consortium.
ERI Earthquake Research Institute, University of Tokyo.
229
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: 1g = 980 gal.
The numerous slope failures that have been observed along the slopes of the Blsamo Ridge can
be classified into three categories: (1) shallow slope failures in the hillsides, typically involving
movement 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 surface
soil or the weathered zone of the rock formation and movement is generally translational. It is likely
that the surface soil or the thickly weathered portion at the hilltop was stripped off by tensile forces
generated 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 slopes
of road cuts.
In addition, it was observed that the slope failures, at least those that occurred in Blsamo 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 distinctly
observed; 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 sites
roughly 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 Blsamo Ridge, caused the
greatest loss of life in a single location from the 13 January earthquake. In this section, the features
of the landslide are discussed in detail.
4.1 Features of the Las Colinas landslide
4.1.1 General damage
As mentioned earlier, the most disastrous landslide associated with this earthquake occurred in Las
Colinas, located in Nueva San Salvador (Santa Tecla), southwest of San Salvador, where the sliding
232
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 dwellings
standing close together.
236
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.
2001).
Cross-section of the Las Colinas slope showing the geological stratum (after Lotti & Associati,
4.4 Possible causes of the landslide

Based on the field survey, three factors can be speculated as possible triggering mechanism for the
Las 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 amplification
There is a strong possibility that the magnitude of shaking during the earthquake may have been
magnified at the crest because of the topography of the site. Based on the microtremor measurements, the lateral components of the ground motions are amplified at the top, and a 1 Hz
240
as the accompanying deforestation, may have contributed, to some extent, to the vulnerability of
the site to landslide.
The Las Colinas landslide caused the greatest loss of life in a single location during the January
13, 2001 earthquake. A large volume of soil mass, estimated at 200,000 m3 , was thrown off the
rim of the Blsamo Ridge and covered hundreds of residential houses, resulting in 500 deaths.
Microtremor measurements and ground response analyses suggest that earthquake shaking was
amplified by the rock mass and the steep ridge topography. Indications of potential liquefaction of
the pumiceous sandy layer were also observed. The landslide transformed into a flowslide despite
the low water content and the relatively low potential energy. Based on laboratory tests and field
observation, the long runout distance can be attributed to the presence of weakly bonded materials
which are easily crushed during fast shearing, allowing rapid and progressive decrease in the
available strength. The dipping direction of the weak soils towards the slope and the flat zone at
the bottom of the ridge with buildings and roads aligned in the flow direction also contributed to
the long travel distance.
ACKNOWLEDGMENTS
The first and third authors would like to acknowledge the assistance of the members of the
JSCE Reconnaissance Team and the El Salvadorian counterparts during the visit to the site in
2001. The microtremor measurements presented in the paper were performed by Dr. P. Mayorca,
while the soil tests were conducted by Prof. T. Yamamoto. The second author would like to thank
Dr. W. Vargas-Monge and Engr. J. Cepeda.
REFERENCES
Baxter, S. 2001. Personal Communication.
Bommer J.J., Benito, M.B., Ciudad-Real, M., Lemoine, A., Lopez-Menjivar, M.A., Madariaga, R.,
Mankelow J.M., Mendez de Hasbun, P., Murphy, W., Nieto-Lovo, M., Rodriguez-Pineda, C.E., Rosa, H.
2002. The El Salvador earthquakes of January and February 2001: context, characteristics and implications
for seismic risk, Soil Dynamics and Earthquake Engineering, 22, 389418.
Crosta, G.B., Imposimato, S., Roddeman, D., Chiesa, S. and Moia, F. 2005. Small fast-moving flow-like
landslides in volcanic deposits: the 2001 Las Colinas landslide (El Salvador), Engineering Geology, 79,
185214.
Devoli, G., Egger, C., Ferres, D. and Rubio J. 2001. Estudio geolgico preliminary de la ladera norte de la
Sierra del Blsamo, Santa Tecla, Departamento de La Libertad, El Salvador. Gelogos del Mundo.
Faccioli, E., Battistella, C., Alemani, P. and Tibaldi, A. 1988. Seismic microzoning investigations in the
metropolitan area of San Salvador, El Salvador following the destructive earthquake of October 10, 1986,
Proc., International Seminar on Earthquake Engineering, Innsbruck, 2865.
Ishihara, K. 1984. Post-earthquake failure of a tailings dam due to liquefaction of pond deposit, Proc.,
International Conference on Case Histories in Geotechnical Engineering, St. Louis, Vol. 3, 11291143.
Japan Society of Civil Engineers, JSCE 2001. The January 13, 2001 Off the Coast of El Salvador Earthquake,
Tokyo, Japan, 112pp.
Jibson, R.W. and Crone, A.J. 2002. Observations and recommendations regarding landslide hazards related to
the January 13, 2001 M-7.6 El Salvador Earthquake, USGS Open File Report 01141, 21pp.
Konagai, K., Johansson, J., Mayorca, P., Yamamoto, T., Miyajima, M., Uzuoka, R., Pulido, N.E., 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), 115.
Lotti, C. & Associati-Enel. Hydro 2001. Investigacion Geotecnical Integral, Informe Final, Sintesis Geologica
y Modelos Numericos, Anexo: 2.2.1.1 El Salvador, Technical Report, financed by BID, property of
Ministerio de Medio Ambiente y Recursos Naturales, 37pp.
Mendoza, M.J., Dominguez, L., Melara, E.E. 2001. Deslizamientos y flujo de tierras en la ladera Las Colinas,
Nueva San Salvador, El Salvador, C.A., disparado por el sismos del 13 de Enero de 2001. Proceedings of
the Second Iberoamerican Conference on Earthquake Engineering, Madrid; 77180.
242

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 cooperative
research 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 successfully
recorded in the deployed LSST arrays. Acceleration time-histories and associated response spectra
from surface and subsurface arrays and from containment responses are reported herein to provide
a field data base for further applications.
1 INTRODUCTION
The effect of soil-structure interaction (SSI) is important for the seismic design of massive structures. Although a variety of SSI analysis techniques and associated computer codes have been
developed, when used for prediction analysis, their results may vary significantly. In order to
quantify 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 1980s at Lotung,
Taiwan, under the cooperation of Electric Power Research Institute (EPRI) and Nuclear Regulatory
Commission (NRC) from USA, and the Taipower Company (TPC) of Taiwan (Tang, 1987). Two
scaled (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 surrounding 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 magnitudes 4.57.0 have been recorded. These data have been applied to evaluate some current available
SSI methods (EPRI, 1991a, b; Chen et al., 1990). The research results were evaluated to develop
guidelines for soil-structure interaction analysis (Hadjian et al., 1991).
An international consortium was organized in 1990 to perform a second phase seismic test at
Hualien, named the Hualien LSST project (Tang et al., 1991). Since the Lotung test was installed in
a soft site, a stiff soil site at Hualien was selected this time to construct a quarter scale containment
model and a tank model for field tests. The containment model and surrounding areas are densely
instrumented to monitor earthquake responses. This paper aims at reporting the Hualien project
and 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 the
cooperation 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 LEnergie Atominque (CEA), Electricite de France (EdF) and FRAMATOME
(France), Korea Institute of Nuclear Safety (KINS), Korea Electric Power Corporation (KEPCO)
245
Figure 1.
Geological map and stations of SMART-2 at Hualien LSST site (Honsho, 1990).
and Korea Power Engineering Company (KOPEC) (Korea) and Taipower Company (Taiwan). The
primary objective of this program was to study the effects of soil-structure interaction on the
response of a nuclear power plant containment structure and a storage tank structure. A quarter-scale
containment model structure was built on sand-gravel deposits, and a dense monitoring network
was installed for measuring the seismic responses of the model structure and surrounding soil.
Hualien, on the east coast of Taiwan, is a seismically active area located on the collision zone of
the Eurasian Plate and the Philippine Sea Plate. Strong earthquakes have occurred very frequently
in this area in the past . For this reason, the 2nd Strong Motion Array (SMART-2) was installed along
the east coast of Taiwan for seismic studies (Chiu et al., 1994). Taking the advantage of the existing
SMART-2 array, the site for the second phase LSST was selected at Meilun, Hualien, at the north
tip of the SMART-2 array as shown in Fig. 1. By using the dense array of seismographs in LSST as
well as in the SMART-2, it is anticipated that strong earthquake data can be well recorded at this
area, thus providing a valuable database for seismic ground motion and soil-structure interaction
analyses.
2.1 Instrumentation
2.1.1 Seismic arrays (surface array and vertical array)
At the Hualien test site, a dense strong motion array, including three surface linear arrays and
three downhole arrays, was installed to monitor earthquake ground motions around the model. The
layout of the Hualien LSST arrays is shown in Fig. 2.
Three surface arrays are deployed along three radial lines away from the containment model,
with 120 apart. These radial lines are designated ARM1, ARM2 and ARM3 as shown in Fig. 2. On
each surface array, five seismograph stations are installed at a distance of 6.49 m, 11.57 m, 21.46 m,
36.41 m and 56.61 m, respectively, from the center of the containment model. The three vertical
arrays are located under the stations A15, A25 and A21 as shown in Fig. 2, and therefore they are
designated A15, A25 and A21 vertical arrays, respectively. In each vertical array, four downhole
accelerometers are installed at the depths of 5.3 m, 15.8 m, 26.3 m and 52.6 m, respectively. In
246
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 the
SMART-2 array was installed at a depth of 100 m below the ground surface. The purpose of this
station was to provide earthquake information at deeper elevations. Since A15 and A25 vertical
arrays 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 model
and will be regarded as the near-field array. For each station in the surface and vertical arrays, the
accelerometers 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 thickness
of 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 surface
sandy soil layer. Then the excavated pit was backfilled with compacted crushed stone around the
containment model. Inside the containment model, a total of 15 accelerometers were installed at
four 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 geologic
map 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 coastal
area 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
Table 2. Peak accelerations recorded at each station during the Chi-Chi earthquake.
Vertical array station
Surface array station
Peak Acc. Value (gal)
Containment station
Name
EW
NS
Name
EW
NS
Name
EW
NS
D11
D12
D13
D14
D21
D22
D23
D24
D25
D26
D28
30.77
30.02
27.95
21.92
30.56
28.79
27
23.65
27.01
24.57
21.03
97.09
75.87
59.9
40.55
88.07
72.32
52.94
46.07
114.86
81.81
52.26
67.81
60.73
47.18
46.07
71.17
55.75
45.45
44.85
66.24
57.43
46.82
A11
A12
A13
A14
A15
A21
A22
A23
A24
A25
A31
A32
A33
A34
A35
29.99
32.83
33.02
32.12
31.89
31.93
29.75
33.22
32.66
28.18
31.41
25.82
27.49
26.85
27.66
109.44
104.35
107.64
98.1
86.03
96.88
97.93
108.76
117.54
118.93
89.26
97.8
99.6
105.76
105.19
73.59
69.09
65.66
65.77
73.92
76.13
72.76
77.67
87.08
84.48
80.66
69.62
62.44
63.96
62.97
BAE
BAS
BAW
BAN
WLE
WLN
WHE
WHS
WHW
WHN
RFE
RFS
RFW
RFN
RFC
32.04
32.44
30.51
34.09
34.07
35.21
74.28
35.62
34.05
39.21
35.29
38.88
36.01
34.94
31.58
100.46
97.63
98.3
102.84
97.32
107.43
145.9
105.81
106.65
108.96
119.18
118.42
111.34
112.94
112.5
75.01
78.74
78.33
78.1
79.4
80.78
119.79
101.74
101.96
101.59
126.42
129.35
126.3
126.54
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 frequency

of 2 Hz, but in the NS direction at a frequency of 3 Hz. Besides, it can be observed that the
maximum responses in all three ARMs increased a little with respect to the distance from the
containment.
251
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 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 Hualien
LSST 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, 9931024
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, 371395.
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, 463475.
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 the
Lotung 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 1115, 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.151152.
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, 11571167.
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 Backfill), Report, Tokyo Electric Power Company, Tokyo, Japan.
257
Figure 2. The Chiufenerhshan dip slope landslide looking from the south to the north (by M.L. Lin).
Figure 3. The blow out point near the crest of the Chiufenerhshan dip slope landslide (by M.L. Lin).
seven of them located on the hanging wall side, and the other five stations located on the foot
wall side of the fault. A comparison was made of the characteristics of ground motions recorded
at the hanging wall side stations and those at the foot wall side. It was found that the strong
motions recorded on either side appeared to have different characteristics (Lin et al. 2000a). Since
the Chiufenerhshan landslide was located on the hanging wall side, records of the strong motion
stations on the same side should be used for determination of the ground motion at the landslide. The
highest peak ground acceleration in the horizontal direction is 989 gal E-W recorded by TCU084
station, and the highest vertical peak ground acceleration is 416gal recorded by station TCU071
out of the seven strong motion stations on the hanging wall side. The records in the horizontal
261
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
Figure 6. The elevation differences from 1980 to 1999 in Chiufenerhshan landslide area (after Lin et al.,
2000).
1200
Before earthquake
After earthquake
1100
Elevation (m)
1000
900
800
700
A'
600
500
400
0
500
1000
1500
2000
2500
3000
Distance (m)
Figure 7. The A-A cross-section slope profile of the Chiufenerhshan landslide before and after sliding.
263
Ac (gal)
Ac (gal)
900
450
0
-450
-900
TCU071
900
450
0
-450
-900
TCU074
900
450
0
-450
-900
TCU076
900
450
0
-450
-900
TCU089
900
450
0
-450
-900
TCU084
900
450
0
-450
-900
TCU079
900
450
0
-450
-900
TCU078
20
25
30
35
40
45
50
55
Time Elapse (Sec.)
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 2896 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 cohesion
rather 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, uniaxial
compression 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 uniaxial
compression test result ranges from 1339 to 3734 kPa with the average of 2911 kPa. The result
of 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 to
32 degree with the average of 28 degree. The peak friction angle ranges from 28 to 50 degree with
the average of 39 degree (Chen, 2001).
5 CAUSES OF THE LANDSLIDE
The weather condition before the Chi-Chi earthquake was fairly dry without much precipitation
in the nearby area of Chiufenerhshan landslide. Among the rain fall gauge stations installed by
265
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 field 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 field 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 in
horizontal, 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 most
important factor for causing the Chiufenerhshan landslide is the strong ground motion induced by
the Chi-Chi earthquake, which causes the sliding to occur along the bedding of weaken formation.
269
Figure 15. The bore hole logs of bore hole number 5 to 10 (modified from Tseng, 2004).
Table 3. Results of direct shear test of bore hole samples (p denotes peak strength and r denotes residual
strength; Tseng, 2004).
Bore Hole
Depth (m)
Peak
cohesion,
Cp (kPa)
Peak friction
angle, p
(deg)
Residual
cohesion,
Cr (kPa)
Residual
friction angle,
r (deg)
BH-1
BH-2
BH-3
BH-4
BH-5
10.0510.45
11.5011.95
12.4012.65
15.3015.70
34.1034.40
96
28
66
71
0
36
35
35
36
23
4
3
3
3
0
33
31
33
33
20
BH-6
28.5029.00
25
23
BH-6
39.4039.60
24
23
BH-7
21.1021.30
24
23
BH-7
36.7037.00
25
23
BH-8
BH-9
BH-10
38.7338.95
14.8014.95
48.8249.00
78
91
87
31
34
32
3
3
6
27
29
27
Rock type
Sandstone
Sandstone
Sandstone
Sandstone
Fractured sandstone
with shale
Fractured sandstone
with shale
Fractured sandstone
with shale
Fractured sandstone
with shale
Fractured sandstone
with shale
Sandstone with shale
Sandstone
Sandstone with shale
6 CONCLUSIONS
On September 21, 1999, the Chi-Chi earthquake with the local magnitude of 7.3 struck central
Taiwan and caused extensive slope failures in central Taiwan. A large scale dip-slope landslide
known as the Chiufenerhshan landslide near the epicenter occurred, which caused severe casualties
and losses of properties. In this paper, the case history and geological, geomorpgollogical properties, and causes of the Chiufenerhshan landslide are presented. The possible ground motions caused
270
20
18
16
Rainfall(mm)
14
12
EQ
10
8
6
4
2
0
1
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Day
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 was
also 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 acceleration
thus 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 strong
ground motion induced by the earthquake.
REFERENCES
Central Geological Survey, 2000, The geological map of Taiwan 1:50,000, Puli
Chang, 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: 293309.
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, 199209.
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
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 17. The isoseismal map and the acceleration records of main shock and the first triggered event
(redrawn from Shin, 1999).
Elevation (m)
1300
CHY080
1100
Cho-Lan formation
slip surface of 1999
900
Chin-Shui shale formation
700
Ta-Wuo sandstone member
500
300
500
1000
1500
2000
2500
3000
3500
4000
Distance (m)
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 the
first 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 first
triggered event nearby and the vertical ground acceleration played an important role in triggering
the landslide.
7 CONCLUSIONS
On September 21, 1999, the Chi-Chi earthquake with the local magnitude of 7.3 struck central
Taiwan and caused extensive slope failures. A large scale dip-slope landslide known as the Tsaoling
285
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. 3147.
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. 34.
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. 114118.
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. 95112.
Hung, J.J., 1980. A study on Tsao-Ling rockslides, Taiwan (in Chinese). Journal of Engineering Environment,
vol. 1, pp. 2939.
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, 15th16th, pp. 459467.
Hung, J.J., 1999. Historical photographs of Tsao-Ling rockslides (in Chinese). Sino-Geotechnics, vol. 76,
pp. 113124.
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. 518.
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 the Tsao-Ling landslide area under earthquake. Masters 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, 15th16th, pp. 459467.
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. 113123.
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, 199209.
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. 114.
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 1823. (in Chinese)
Yeng, K.T., 2000. The residual strength of Chin-Shui shale in relation to the slope stability of Tsao-Ling.
Masters 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 the
fault 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 reconnaissance
and surveyed results of liquefaction induced ground failures are presented. Lastly, conclusions on
factors that possibly contribute to such extensive ground failures and future research suggestions
are 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 Chi
earthquake possessed very unique strong ground motion characteristics. Structure performance
and geotechnical hazards such as soil liquefaction under such strong ground motion are of great
interests 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 within
30 km range from the epienter of the Chi-Chi earthquake. Surface fault rupture of Che-LungPu fault has just stroke along the foothill line which is the geological boundary separating the
alluvium 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 at
Wu-Feng was as high as 774 gal in east-west direction and duration was as long as 44 sec. Such
strong 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 is
also 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) formation
lying above the sandstone and shale interlayers. This hillside area mainly suffered from destructive
ground deformation caused by the fault rupture. Instead of thick sand deposit like adjacent plain
sites in central Taiwan, formation of plain area of Wu-Feng is laterite deposit interlayered with
high fine content silty sand deposit. Surprisingly liquefaction induced ground failures including
subsidence and lateral spreading have occurred extensively over most plain area of Wu-Feng during
Chi-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 that
needed special attentions as well as limited research resources when the first phase reconnaissance
work was conducted. Detailed survey and analysis was not done until 2005 via photos and available
documentations by the authors (Lin, 2006). In this paper, the authors will first present soil condition
and characterization of the near fault strong ground motion of Wu-Feng during the Chi-Chi earthquake. Secondly, both reconnaissance and surveyed results of liquefaction induced ground failures
are 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 improve
our understanding on liquefaction induced ground failures caused by strong ground motions.
290
Figure 4.
Fault rupture at Wu-Feng junior high school.
Table 1. Basic information of collected boreholes.

Coordinates
No.
Elevation m
Depth M
GWT depth m
BH-3
BH-6
BH-7
BH-8
BH-10
BH-11
BH-12
BH-13
BH-14
C-8
C-15
B-1
B-2
B-3
B-4
B-5
B-6
B-7
B-8
B-9
B-10
D-1
D-2
D-3
D-4
2662550
2662767
2661861
2661560
2661221
2662203
2660838
2662392
2664115
2661574
2661485
2661188
2661150
2661147
2661106
2661109
2661091
2662169
2661491
2662089
2661283
2662572
2662631
2662545
2662597
217761
218229
218608
218502
218349
218331
218153
218519
215570
218500
218296
218369
218412
218325
218395
218326
218374
218226
218261
217875
217784
218272
218284
218343
218365
46.48
49.39
56.6
58.7
56.88
53.16
58.97
55.23
35.16
58.64
56.37
57.00
57.00
57.00
57.00
57.00
57.00
51.41
55.62
50.07
53.81
51.00
52.00
52.00
52.00
30
30
30
50
30
30
30
30
23.6
28.35
19.4
25
25
25
25
25
25
30.45
20.45
20.45
33.5
30.45
30.45
30.45
30.45
1.3
3.2
3.2
0.5
1.4
0.9
4.2
0.9
3.5
0.5
2.5
2.7
2.3
2.75
2.55
2.7
1.9
1.9
3.2
2.28
0.54
2.7
3.1
2.8
2.8
292
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 at
selected borehole locations. This analysis was done by using ProShake program. In this preliminary
analysis, soil layer within analyzed boreholes are assumed to be horizontal. Moreover, the deep
seated gravel layer with shear wave velocity over 500 m/s and depth over 50 m was assumed as the
298
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 suffered extensive ground failures including ground
300
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 fatalities
and 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 II
earthquake site, the authors conducted a Geological Information System (GIS) based survey via
photos and available documentations in 2005 (Lin, 2006). Soil liquefaction induced ground failures
such as sand boils, ground subsidence, and lateral spreading were documented from photos and
306
Figure 24.
Mapping of Structural damages.
In addition to document the locations of ground failures and structural damages, the authors
also 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-Chi
earthquake. Local settlement was then extrapolated via sequence photos from the reference points
to locations of interests including inspected boreholes and ground failure sites. Measurable objects
in 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 actual
measurements were taken after the Chi-Chi earthquake for verification. Total 30 survey points were
analyzed in order to cover the studied area in detail. Settlement contour was then generated using
GIS software as shown in Figure 25. The predicted settlement contour provides an overall picture
of the ground subsidence caused by liquefaction of Wu-Feng during the Chi-Chi earthquake. As
shown in the figure, delta areas where rivers intersect in alluvium plain had the most serious ground
309
Figure 4. Recorded ground acceleration at station

TCU 076 (near to Site 5).
Figure 5. Recorded ground acceleration at station

CHY080 (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

at Site 2. (after Huang et al., 2003).
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-MB

wall 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 was
taken 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, its more reasonable not
313
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 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).
Site 5.
Plane view of the collapsed levee at
3 DAMAGE OF LEVEES
Figure 11 shows an overview of a damaged levee at Site 5 near the Che-Lung-Pu fault. A plane view
of the damaged levee is shown in Figure 12(a). Boring data and a cross section is shown in Figure
12(b). Boring log profile in Figure 12(b) indicates that the heavily damaged site is associated with
relatively weak soil strata with Standard Penetration Test (SPT) N-values ranged between 8 and 15
in the backfill and foundation soils. This contrasts the N-values (N 12) obtained in the the lightly
damaged 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 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 0.8 m,
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 the
design of hydraulic (or geotechnical, in a broader sense) structures. This practice is subjected to
a 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 embankments in hillslope areas of Taiwan. These structures have small base width-to-height ratios
and 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 that
the safety factor against bearing capacity failure and the vertical displacement of the wall are
suceptible to the change of internal friction angle of the foundation soil. Therefore, a small
decrease 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
Figure 18.
Results of parametric study on seismic displacements for the wall at Site 6. (after Huang, 2005).
the wall and also providing an extra safety factor against lateral sliding during earthquakes.
However, the role of passive zone to the stability of the wall tend to be ignored in current design
and construction practices.
REFERENCES
EDM Technical Report No. 7 (2000) Report on the Chi-Chi, Taiwan earthquake of September 21, 1999,
Earthquake Disaster Mitigation Research Center (EDM), The Institute of Physical and Chemical Research
(RIKEN), Miki, Hyogo Prefecture, Japan.
Huang, C.C. (2000) Investigations of soil retaining structures damaged during the Chi-Chi earthquake,
Journal of the Chinese Institute of Engineers, Vol. 23, No.4, pp.417428.
Huang, C.C., Chou, L. H. and Tatsuoka, F. (2003) Seismic displacements of geosynthetic-reinforced soil
modular block walls, Geosynthetics International, Vol. 10, No. 1., pp. 223.
Huang, C.C. and Chen, Y.H. (2004) Seismic stability of soil retaining walls situated on slope, J. Geotech.
and Geoenviro. Engrg., ASCE, Vol. 130, No. 1, pp. 4557.
Huang, C.C. (2005) Seismic displacements of soil retaining walls situated on slope, J. Geotech. and
Geoenviro. Engrg., ASCE, Vol. 131, No. 9., pp. 11081117.
324

Performance of buildings in Adapazari during the 1999 Kocaeli,

Turkey earthquake
J.D. Bray
University of California, Berkeley, CA, USA
R.B. Sancio
Golder Associates, Inc., Houston, TX, USA
ABSTRACT: A large number of structures in Adapazari, Turkey collapsed or were heavily damaged due to strong ground shaking or ground softening resulting from the August 17, 1999 Kocaeli
earthquake. The softening of shallow silt deposits led to relative vertical displacement of buildings
into 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, systematic surveys of building damage and ground failure were completed, which were followed by
comprehensive subsurface investigations. These investigations are summarized, and the different types of foundation failures observed in Adapazari are described. The building survey data
show the interdependence of structural damage with ground failure. Areas within the city with
significant ground failure commonly had relatively greater amounts of buildings with significant
structural 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 city
suffered 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 Adapazari
was 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). More
detailed surveys of representative building sites were also performed throughout the city (Bray
et al. 2004). With this wealth of post-earthquake reconnaissance data available, comprehensive
subsurface 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 responsible for the observed ground failure in Adapazari was conducted to allow over a hundred cyclic
laboratory tests to be performed (Bray and Sancio 2006).
The site characterization data developed through these studies is summarized and made available in the accompanying electronic files (see the appendix). Earthquake ground motions in the
vicinity of Adapazari are discussed in Bray et al. (2004) and representative records are provided. Building performance data, including relative movements of their foundations, are also
provided. Representative cases of building performance at sites involving ground failure are presented, and the prevalent mechanisms of ground movements that affected building performance are
described.
325
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 13 deg; lateral movements < 10 cm
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 the
severity of structural damage generally increases with increasing levels of ground failure (Sancio
et al. 2002). Sediment ejecta were observed within some of the ground failure zones, but ejecta
were absent from many areas.
330
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, indicating that the foundation underwent essentially rigid-body settlement and rotation. The buildings
structural frame was essentially undamaged by the earthquake (Bray and Stewart, 2000). However, 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 13.
East-west cross section across the foundation, ground floor, and first floor of building D1.
Heave
Heave
0
ML
CH
CL w/ Silt
Clayey Silty Sandy
10
10
HORIZONTAL SCALE
Figure 14.
10
15
20 m
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
of excess pore water pressure, or cessation of shaking which causes the overturning moment to
reduce significantly.
8 CONCLUSIONS
A large number of buildings in Adapazari experienced earthquake-induced ground failure as well
as significant structural damage. Although structural damage in Adapazari was primarily due to
strong ground shaking and poor structural design and construction, building damage also appears
to be related to ground failure, with significant building damage being more likely to occur in areas
of significant ground failure.
The observations of ground failure and building damage prompted several surveys and indepth studies of the ground conditions in Adapazari. The results of these surveys and subsurface
investigations are documented in this paper and the accompanying electronic files.
In Adapazari, vertical displacement of buildings into the ground during the 1999 Kocaeli earthquake 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 CMS0116006. Initial reconnaissance efforts were also supported by the NSF through the Geoengineering
Extreme Events Reconnaissance (GEER) Association. Dr. Turan Durgunoglu and his ZETAS
Corporation provided the laboratory equipment in Istanbul, as well as the drilling equipment
in Adapazari. Professor Akin Onalp and his laboratory at Sakarya University provided laboratory support for the initial subsurface characterization testing. Other support was provided by
ZETAS 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 the
initial 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. ORourke,
and D. Ural, are acknowledged, especially the collection of building data by Jonathan Stewart, Les
Youd, 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):
163189.
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 Subsurface 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 1822, 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 1999
Kocaeli earthquake. Soil Dynamics and Earthquake Engineering. 22: 829836.
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): 448465.
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: 289293.
Liu, H., and Dobry, R. 1997. Seismic response of Shallow Foundation on Liquefied Sand. Journal of
Geotechnical and Geoenvironmental Engineering. 123(6): 557567.
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 SoilStructure Interaction in Adapazari. In Jubilee Papers in Honor of Prof. Ergun Togrul, 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: 6596.
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): 653675.
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: 10931102
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): 83108.
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): 290307.
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): 2338.
Youd, T.L., and Perkins, J.B. 1987. Mapping of Liquefaction Severity Index. Journal of Geotechnical
Engineering, ASCE. 113(11): 13741392.
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 Report
Characterizing Subsurface Conditions, May 15, 2003. This report is provided as a file named
Bray et al 2003 Adapazari_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
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 doubleintegration 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. It
is estimated that about two thirds of these displacements resulted from shear strains induced in
the reclaimed fills. The acceleration response spectra for damping ratios of 10% range around
0.30.5 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) piles
and cast-in-place concrete (CC) piles. Most precast piles are hollow with outer diameters typically
342
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 slope
angle 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 span
direction where the displacements of the piles are largest. The horizontal displacements at both
pile heads are almost the same, being equal to 6080 cm, and consistent with those obtained by the
aerial 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 in
Figure 26 (f) is the distribution of cracks from television observation. The variations of curvature
are consistent with the failure patterns in such a way that the failure and cracks concentrate where
large 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) on
the south of Fukae. The distances from the northwest corner of the building to the northern and
351
BTL Committee 1998. Research Report on liquefaction and lateral spreading in the Hyogoken-Nambu
earthquake (in Japanese).
Chuo-Kaihatsu Corporation 1995. Reconnaissance report on the 1995 Hyogoken-Nanbu earthquake Great
Hanshin Disaster, (Hyogoken-Nanbu earthquake).
Committee of Earthq. Obs. and Res. in the Kansai Area (CEORKA) 1995. Digitized strong motion records in
the affected area during the 1995 Hyogoken-Nanbu earthquake.
Fujii, S., Cubrinvski, M., Hayashi, T, Shimazu, S., and Tokimatsu, K. 1996. Response analysis of buildings with
a pile foundation on a liquefied ground, Proc., 31nd Japan National Conf. on Geotechnical Engineering,
Vol. 1, pp. 11391140 (in Japanese).
Horikoshi, K. and Ohtsu, H. 1996. Investigation of PC piles damaged by the Hyogoken-nanbu earthquake,
Proc., 31st Japan National Conf. on Geotechnical Engineering, Vol. 1, pp. 12271228 (in Japanese).
Ishihara, K. and Yoshimine, M. 1992. Evaluation of settlements in sand deposits following liquefaction during
earthquakes, Soils and Foundations, Vol. 32, No. 1, pp. 173188.
Ishihara, K., Yoshida, K., and Kato, M. 1997. Lateral spreading of liquefied deposits during the 1995
Kobe earthquake, Proc., 3rd Kansai Int. Geotechnical Forum on Comparative Geotechnical Engineering,
pp. 3150.
Japanese Association for Steel Pipe Piles. 1996. Investigation report on steel pipe pile foundations in the
Hyogoken-Nambu earthquake- Part II, 156pp.
Japan Road Association 1980, 1997. Specifications for road bridges, Vol. IV (in Japanese).
Kansai Branch of Architectural Institute of Japan 1996. Report on case histories of damage to building foundations in Hyogoken-Nambu earthquake, Report presented by Committee on Damage to Building Foundations,
400pp. (in Japanese).
Kuwabara, F. and Yoneda, K. 1998. An investigation on the pile foundations damaged by liquefaction at the
Hyogoken Nanbu earthquake, Journal of Struct. Constr. Engrg., AIJ, No. 507 (in Japanese).
Nagai, K. 1997. Lessons concerning foundation design in Hyogoken-nambu earthquake, Kenchiku Gijyutsu,
No. 564, 8493 (in Japanese).
Oh-oka, H., Iiba, M., Abe, A., and Tokimatsu, K. 1996. Investigation of earthquake-induced damage to pile
foundation using televiewer observation and integrity sonic tests, Tsuchi-to-kiso, JSG, Vol. 44, No. 3,
pp. 2830 (in Japanese).
Oh-oka, H., Onishi, K., Nanba, S., Mori, T., Ishikawa, K., Koyama, S., and Shimazu, S. 1997a. Liquefactioninduced failure of piles in 1995 Kobe earthquake, Proc., 3rd Kansai Int. Geotechnical Forum on Comparative
Geotechnical Engineering, pp. 265274.
Oh-oka, H., Katoh, F., and Hirose, T. 1997b. An investigation about damage to steel pipe pile foundations due
to lateral spreading, Proc., 32nd Japan National Conf. on Geotechnical Engineering, Vol. 1, pp. 929930
(in Japanese).
Seed, H. B., Tokimatsu, K., Harder, L. F., and Chung, R. M. 1985. Influence of SPT procedures in soil
liquefaction resistance evaluations, Journal of Geotechnical Engineering, Vol. 111, No. 12, pp. 14251445.
Satake, K., Oh-oka, H., and Tokimatsu, K. 1997. Investigation of earthquake-induced damage to steel pipe
pile foundation, Proc., 32nd Japan National Conf. on Geotechnical Engineering, Vol. 1, pp. 927928 (in
Japanese)
Sekisui House 1996. Rokko Island city, Strong motion record during the 1995 Hyogo-ken Nanbu earthquake
and its analysis (in Japanese).
Shamoto, Y., Sato, M., Futaki, M., and Shimazu, S. 1996. A site investigation of post-liquefaction lateral
displacement of pile foundation in reclaimed land, Tsuchi-to-Kiso, JSG, Vol. 44, No. 3, pp. 2527 (in
Japanese).
Tokimatsu, K., Mizuno, H., and Kakurai, M. 1996. Building damage associated with geotechnical problems,
Soils and Foundations, Special Issue, pp. 219234.
Tokimatsu, K., Oh-oka, H., Shamoto, Y., Nakazawa, A., and Asaka, Y. 1997. Failure and deformation modes
of piles caused by liquefaction-induced lateral spreading in the 1995 Hyogoken-nambu earthquake, Proc.,
3rd Kansai Int. Geotechnical Forum on Comparative Geotechnical Engineering, pp. 239248.
Tokimatsu, K. and Asaka, Y. 1998. Effects of Liquefaction-induced ground displacements on Pile Performance
in the 1995 Hyogoken-Nambu Earthquake, Soils and Foundations, Special Issue, pp. 163177.
Tokimatsu, K., Oh-oka, H., Satake, K., Shamoto, Y. and Asaka Y. 1998. Effects of Lateral Ground Movements
on Failure Patterns of Piles in the 1995 Hyogoken-Nambu Earthquake, Proc., Geotechnical Earthquake
Engineering and Soil Dynamics 3rd Conference, GEO-INSTITUTE ASCE, 2, 11751186.
Tokimatsu, K. 1999. Performance of pile foundations in laterally spreading soils, Earthquake Geotechnical
Engineering, Vol. 3, pp. 957964.
Tokimatsu, K. 2003. Behavior and Design of pile foundations subjected to earthquakes, Proc., 12th ASIAN
Regional Conference on Soil Mechanics and Geotechnical Engineering, Vol. 2, 10651096.
356

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. Nakahira
CTI Engineering Co., Ltd., Osaka, Japan
T. Tazoh
Shimizu 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 Osaka
Bay. Soil liquefaction and the lateral spreading it induced occurred widely in reclaimed lands
during the earthquake. In-ground horizontal displacements from the surface to a depth of 51.5 m
detailing 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 on
the pile cracks was carried out on 119 pile foundations of Route 5 in total by using a bore-hole
camera 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 meters
laterally 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 cities
of Osaka and Kobe. It is an elevated highway structure from Tempozan in Osaka city to Rokko
Island in Kobe city. Route 5 has been relatively newly constructed and newer design specifications
have been applied. This is why the damage was relatively slight in spite of the short distance from
the 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 bridges
close 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
365
Figure 12. Damage investigation of ground and foundations in seven reclaimed islands.
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 in
Figure 18, which are cast-inplace 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 014 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 (1420 m), a Holocene sandy soil layer
(2028 m), and a Pleistocene sandy soil layer (under 28 m). The soil profile of the ground can be
seen 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. The
results of laboratory tests on the samples gathered by these methods were used for analyses without
any correction.
369
Iwasaki, T., Tatsuoka, F., Tokita, K., Yasuda, S., 1980. Estimation of Degree of Soil Liquefaction During
Earthquakes, Tsuchi-to-Kiso, Vol. 28, No. 4, 2329 (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, 261270 (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, 259271,
(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, 380381 (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, 195210 (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, 721730.
371

Damage to subway station during the 1995 Hyogoken-Nambu

(Kobe) earthquake
N. Yoshida
Tohoku Gakuin University, Tagajo, Japan
ABSTRACT: A detailed reconnaissance survey was made at the Daikai subway station, which
was the first subway structure that had completely collapsed due to earthquakes. The station is
a 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 subsoil
over the station by more than 2.5 m at maximum. Many diagonal cracks were also observed on the
walls in the transverse direction. Judging from the damage patterns, a strong horizontal force was
imposed on the structure from the surrounding subsoil. Evidence of damage due to vertical force
was not clearly observed. Finally, characteristics to be used in the future analysis in order to make
the 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 various
structures. Among these, the damage to the subway was one of the amazing events, because
underground 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 subway
occurred 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 in
about 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
Figure 3. Typical section of Daikai Station.
Figure 4.
design.
Load distribution considered during the
Figure 5. Reinforcing steel arrangement of the

center column.
direction. There are 35 center columns in total along the longitudinal direction. The center column
is 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 m
deep. Thicknesses of the ceiling and base slabs are 0.80 and 0.85 m, respectively, and the thickness
of 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 are
heavily loaded in this region.
The frame was designed based on a consideration of the weight of the overburden soil, lateral
earth 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 are
symmetric); 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, axial
force and shear force, respectively, and the so called lateral wall is the outermost wall of the
underground station. The design strength of the concrete was 23520 kN/m2 for the center column
and 20580 kN/m2 for the other structural components, and design the yield stress of the reinforcing
steel bar was 23.52 kN/cm2 . Round steel bars with diameters from 16 to 25 mm were used to the
walls 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 reinforcing
steel 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 design
strength) 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. Strength
of 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 was
39690 kN/m2 .
375
80
60
40
20
0
0.01
Figure 9.
0.1
1
10
Grain size (mm)
Grain-size distribution of fill material.
1.0
25
0.8
20
0.6
0.4
15
IC (0.5)
IC (1.0)
AC (0.5,0.3)
10
5
0.2
0.0
0.0001 0.001
Figure 10.
0.01
0.1
Shear strain, (%)
Damping ratio,h(%)
Shear modulus ratio, G/Gmax
Percent finer by weight
100
0
10
Dynamic deformation characteristics.
Figure 11. Schematic figure indicating the damage patterns in the longitudinal direction. Here, h in (d)
denotes measured clear height, circled number denotes column number, and small size number in the figure
represents crack width in mm
check the cracks. The locations where finish was not removed are shown as region not investigated
in this diagram. The overburden soil had to be removed in order to see the cracks on the exterior
surface of the wall. Locations where this was done are shown as region investigated in the figure.
Checking of cracks on the exterior surface of the ceiling slab was not undertaken because of the
reconstruction work in progress.
378
Figure 24. Ventilation tower.
This indicates that the relative displacement of the ground surface from GL-4m was larger than
4 cm. This displacement is very important because the analysis clarifying the mechanism of the
damage must explain it at the same time. It indicates that this distortion can be used to evaluate the
accuracy 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 damage
in the Daikai station. It is very difficult to clearly define the mechanism of the damage to the Daikai
station by only observing the completely collapsed columns, because the effect of gravity (axial
load) becomes predominant in the residual deformation pattern. Therefore, it is better to observe
columns that were only slightly damaged.
For column 28, which is one of the slightly damaged columns, a small separation of the column
cover concrete was observed at two locations: sea side (bottom) and mountain side (top) of the column. This indicates that the horizontal force acting in the transverse direction toward the mountain
side caused the damage to the column.
The mechanism becomes clearer by focusing the investigation on the slightly damaged column
such as columns 2226 and 2934. The predominant damage occurred either at the bottom or at
the top of the column. This damage seems to have been caused by the bending moment; bending
moment around the longitudinal axis seems to be predominant whereas that in the transverse
direction is much less effective. This also indicates that the horizontal force acting in the transverse
direction 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 the
end of the column, which resulted in the collapse of the end of the column. 2) The load carrying
384
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 comparing
with the Daikai station, the load carrying capacity is larger at the Nagata station and the relative
displacement at the failure is smaller compared with those at the Daikai station. This seems to
indicate 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 displacement
near the station. At the Daikai station, the relative displacement at failure by the static analysis is
2.85 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 analysis
is 2.55 cm at the Nagata station and that by the dynamic analysis is 2.7 cm. They are nearly the
same order although the displacement by the dynamic analysis is a little larger. This error may be
caused since contribution of transverse wall is not taken into account, and because of the error in
evaluating the material property of soil.
The Daikai station was the first subway structure that had completely collapsed due to the earthquake. 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 deformation
of 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 from
the deformation of the box frame. Then, as a result of the relative displacement between the top
and 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 carry
most 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 overburden
soil is present in zone B because of the existence of the B1 floor (concourse), which reduced the
inertia 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 the
lateral wall and slabs did not completely fail although the tensile reinforcing bar yielded. Accuracy
or 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
Figure 1.
Location map of vertical array sites during Hyogoken Nambu Earthquake.
Table 1. Vertical array data during Hyogoken Nambu Earthquake.

Site Name
1
2
3
4
5
6
7
8
9
10
Port Island
YAMAZAKI
TAKASAGO
TECHNICAL
RESEARCH
KAINAN PORT
Higashi-Kobe Bridge
INAGAWA
MKO(Miyakojima)
TIS(TAISYO)
TKMF(TAKAMI)
Altitude
(degree)
Latitude
(degree)
No of
depths
installed depth
Organization
34.670
35.060
34.753
34.743
135.208
134.603
134.783
135.442
4
2
3
3
GL, 16 m, 32 m, 83 m
GL-0.7 m, 30 m
GL, 25 m, 100 m
GL, 24.9 m, 97 m
Kobe-city
KEPCO
34.150
34.707
34.836
34.704
34.650
34.690
135.192
135.296
135.427
135.523
135.478
135.462
3
2
2
2
2
2
GL, 25 m, 100 m
GL-2, 30m
GL-2, 30 m
GL, 70.39 m
GL, 57.49 m
GL-1.5 m, 30 m
Hanshin
Exprees Way
Ohbayashi Corp.
Konoike Corp.
available. Since the surface layers show smaller S-wave velocities than deeper ones in these array
sites, 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. General
characteristics 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 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
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 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 locates
at 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 m
where 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 by
Hanshin Express Way Co. Higashi Kobe bridge was constructed as a cable stay type of three spans
with center and side spans of 485 m and 200 m in 1993. Several accelerometers were installed at
key 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 including
the ground, suggesting the liquefaction during the earthquake based upon the special feature of the
recorded waves in the vertical array. Inagawa site is medium dense sand and gravel(N = 2040) at
the surface underlain by weathered rock ground. No information of P-S wave velocities is available
for 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 seismometers
are 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 Earthquake
may be obtained from the Shinsai Yobo Chosakai at the following address.
The Shinsai Yobo Chosakai
26-20, 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 Engineering
Kansai Electric Power Co.
3-6-16, Nakanoshima, Kita-ku
403
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, Hanshin Express Way.
Figure 22. Takami-site, Konoike.
Port Island ground motion had recorded a long duration of 360seconds, which contains several
aftershocks 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
2.
3.
4.
5.
6.
7.
the main shock and earlier aftershocks with gradual recovery from the de-amplification with
elapsed time during later aftershocks.
The ratio of the vertical ground motions shows amplification during the mainshock, but decreased
to show de-amplification during aftershocks with gradual recovery with time.
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.
P-wave velocity also decreased significantly but also recovered with time.
The time dependent change in predominant periods of horizontal and vertical motions
corresponds to the above mentioned changes of P and S wave velocities.
It should be noted that SPT N-value has increased two times from N = 5 to N = 10, no significant
increase of S-velocity was reported.
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 well
as rotation.
The author expresses his sincere thanks to Kobe city and KEPCO for their generous permissions
to 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, 19, 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 4152, Tokyo: Japanese Geotechnical Society
Iwasaki Y. and Tai M. 1996. Strong motion records at Kobe Port Island, Special Issue of Soils and Foundations
2940, Tokyo: Japanese Geotechnical Society
Sugito, 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. 531.5163, 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. 613616, (in Japanese)
Strong Motion Array Observation No. 3, 1998. Tokyo: Shinsai Yobo-Kyokai
407
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 maximum
size 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 each
layer of rock-fill be washed thoroughly with water pressure (water head of 60 m) using a quantity
equivalent 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 which
is under a poor state of densification.
410
Figure 4.
Gorge profile.
Figure 5.
General view of Cogoti Dam.
The first shaking that stroke this dam occurred on April 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 USGS
and Servicio Chileno de Sismologa, 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 site
is 16 km. In the seismic station of Illapel, the peak ground acceleration reached 0.27 g.
412
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 settlement
of 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. Settlements
in the concrete face were reported. The most important damage was associated with a large displacement of the rockfill that involved the whole downstream slope. This situation was considered
risky and therefore, the slope was immediately repaired.
414
Table 1. Recorded earthquakes.

Earthq.
Lat.
Long.
30.445
71.197
28.343
Depth (km)
Date
Mag.
55
14/10/1997
6.8
72.029
95
18/07/1998
5.1
28.456
70.797
124.2
24/12/2001
4.9
29.268
71.299
54.3
08/10/2000
4.9
29.253
71.638
33.7
09/07/2000
4.8
30.442
71.315
50.2
06/11/2001
5.1
29.399
72.128
14/09/1998
4.5
28.078
71.257
228.6
16/06/1998
4.5
29.477
71.821
43.2
04/09/2001
4.6
10
27.879
70.421
94
09/08/2001
4.6
11
27.041
70.741
222.8
22/11/1997
4.9
12
29.567
71.9
01/09/2007
4.8
13
28.502
71.39
23.8
17/08/2000
4.4
14
31.099
71.86
25.8
12/01/1998
15
28.655
69.982
37.2
21/08/2001
3.9
16
28.741
71.935
224.6
05/09/1997
4.8
17
29.513
72.26
34.2
03/09/1998
4.9
18
28.365
68.951
131.6
16/10/2001
5.3
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
E-W
Vertical
N-S
Crest
Spillw.
Toe
Rock
0.226
0.157
0.124
0.132
0.087
0.064
0.113
0.104
0.091
0.104
0.084
0.075
0.067
0.081
0.037
0.051
0.026
0.022
0.049
0.039
0.047
0.044
0.038
0.026
0.033
0.013
0.017
0.033
0.017
0.015
0.026
0.025
0.018
0.025
0.02
0.013
0.025
0.014
0.016
0.023
0.014
0.013
0.023
0.015
0.016
0.021
0.013
0.016
0.02
0.016
0.014
0.018
0.02
0.016
0.078
0.056
0.072
0.132
0.064
0.067
0.114
0.103
0.117
0.051
0.034
0.046
0.039
0.046
0.054
0.009
0.004
0.01
0.018
0.017
0.024
0.049
0.026
0.038
0.038
0.005
0.009
0.01
0.005
0.01
0.007
0.008
0.009
0.013
0.009
0.012
0.007
0.005
0.009
0.004
0.004
0.004
0.011
0.005
0.013
0.006
0.006
0.008
0.009
0.007
0.009
0.006
0.004
0.009
0.104
0.111
0.073
0.051
0.036
0.040
0.078
0.09
0.075
0.046
0.042
0.056
0.018
0.011
0.01
0.018
0.013
0.011
0.022
0.029
0.034
0.033
0.019
0.021
0.014
0.009
0.013
0.018
0.012
0.015
0.018
0.015
0.015
0.019
0.013
0.014
0.018
0.011
0.01
0.011
0.008
0.009
0.014
0.012
0.015
0.011
0.015
0.012
0.015
0.009
0.01
0.01
0.009
0.011
0.048
0.04
0.049
0.132
0.132
0.132
0.054
0.033
0.054
0.026
0.016
0.025
0.03
0.018
0.032
0.005
0.005
0.005
0.011
0.011
0.018
0.01
0.007
0.014
0.005
0.003
0.006
0.007
0.006
0.008
0.006
0.013
0.01
0.006
0.005
0.008
0.009
0.006
0.008
0.003
0.01
0.003
0.005
0.003
0.006
0.005
0.004
0.006
0.007
0.006
0.008
0.005
0.004
0.006
(Continued)
419
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 wave
velocity 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 Santa
Juana 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 to
the north-west of Santiago (capital city of Chile). The reservoir has a capacity of 60.3 million m3 .
421
Figure 19.
Profile of the gorge.
Figure 20.
Grain size distribution bands of materials used in the dam.
as CL. Therefore, the core material is actually clayey sand. It is interesting to mention that materials
2 and 3 are rather similar to each other.
4.2 Amazing finding at the end of construction
When the construction of the embankment was almost finishing, a bulldozer operating in the
river bed, near to the downstream toe, suddenly sank several centimeters. This evidence, together
423
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 Aromos
Dam had been founded presented a risk of liquefaction, although they recognized that it should not
comprise the whole foundation. According to this result, the following two main countermeasures
were proposed:
A berm of compacted material confining 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 absence
of 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
Figure 28.
Identification of boreholes with SPT N-values lower than 1520 blow/30 cm.
Cogoti Dam completed in 1938 is a good example of the magnitude of seismic settlements and
its relation with static settlements in a non compacted rockfill. Additionally, it shows the important
damage that may occur on the concrete face caused by seismic settlements. Moreover, the behavior
of Cogoti Dam shows how noble is a rock-fill material for being used in dam construction, since
even large seismic settlements did not cause any problem in the dam stability.
Santa Juana Dam completed in 1995 has been subjected to several earthquakes, which have been
recorded by four accelerometers distributed in the dam and in the site. The data clearly show the
phenomenon of amplification and the existence of a predominant period which is quite similar to
the theoretical value. These results are very useful for future earthquakes because any change in
this natural period can be quickly interpreted in terms of change in the mechanical behavior of the
body of the dam. Accordingly, it is strongly recommended the use of this type of instrumentation
in dams that permits to detect any potential local failure immediately after an earthquake.
Aromos Dam represents a remarkable case to learn that the actual occurrence of liquefaction in a
ground with heterogeneous conditions still requires a deeper analysis and a better understanding of
the interaction phenomena. It seems that whenever liquefiable materials are enclosed or surrounded
by dense soils which are not liquefiable, the whole behavior of the ground is controlled by the more
competent material and liquefaction can not take place.
ACKNOWLEDGEMENTS
The work reported here is a part of a research program on dam behavior carried out by the
Direccin de Obras Hidrulicas of the Chilean Ministry of Public Work and IDIEM (Instituto
de Investigaciones y Ensayes de Materiales) of University of Chile. Special thanks are given to
Carolina Vergara for all the work done during the analysis.
429
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 Conference
on 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) Fsica de la Tierra (Madrid), (10):221258.
Saragoni, R., Gonzlez, 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. Ishihara
Civil 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 in
the loess deposit of Aeolian origin in the gently sloping hilly terrain and lead to a series of catastrophic landslides accompanied by a large-scale mudflow. In contrast to the hitherto known cases
of liquefaction which have usually occurred in water-sedimented sand deposits, the liquefaction in
Tajik 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 porous
loessal 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 the
catastrophic landslide.
1 INTRODUCTION
Development of liquefaction and consequent occurrence of slumping or flow slides in sandy
deposits have been recognized as the major phenomena leading to catastrophic damage to the
ground 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 however, 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 those
cases which occurred in lowland areas near the waterfront or in deposits underlying water-retaining
embankments or dams.
It was, therefore, a great surprise to observe a series of catastrophic flow slides during a recent
earthquake 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 liquefactioninduced mud flow will be described, together with the mechanism of liquefaction occurring in the
collapsible 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 borders
on Afghanistan. The location of the epicenter is shown in the map of Figure l. The focal depth of this
event is reported to have been about 35 km. As shown in Figure 2, the affected area is located on a
relatively flat basin-like plain which developed in front of the southern flank of the Pamir Mountain
range. 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 epicentral
area and a network of strong motion seismographs registered the motions at several stations as
indicated 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. Extrapolating
from 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 this
magnitude 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 visible
evidence of the fault on the ground surface.
3 LANDSLIDE AND DEBRIS FLOW

In the village of Gissar, there are about 500 farmers houses and barns constructed of wood with
adobe type walls. Even in the most severely affected area, the degree of structural damage was
such that houses were partially destroyed and thus the destruction due to the earthquake shaking
itself 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 were
four landslides in the affected area of Gissar as indicated in plan view of Figure 6. The landslides
turned into a mudflow of vast scale and buried more than 100 houses in 5 meters of mud. An
estimated 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 village
immediately downhill were buried in mud resulting in the largest number of causalities. A view
looking northward from the top of the bluff is shown in the photograph in Figure 8, where two
water storage tanks are seen perching on the debris. These tanks had been installed on the hilltop
to supply water for domestic use and for agricultural irrigation. Just beyond the scarp of the slide
on the east side, a pumping station remains intact as indicated in Figure 7. The pumping station
was 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 slide
is almost coincident with the line of the water channel installed on the shoulder of the hill to supply
water for the farmland over the hills. Thus, it appears likely that the water in the channel had been
infiltrating into the loess deposit over the years and the slide was initiated from the failure of loess
soils 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 slide
at Sharara. Thus, the features and conditions of the occurrence of the slide are almost identical
to 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 infiltration appear to be responsible
for causing the slide during the earthquake. As shown in a cross section in Figure 11, there was
a considerable spreading of the soil mass, extending outward about 100 m from the toe of the
slide. 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 hill
slopes 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 at
the scarp is shown in Figure 12. It can be seen that the scarp is nearly vertical indicative of the fact
that the soil block fell off from vertical cleavage or fissures.
437
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 depth
at 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 t z exceeds the compressional strength, qu , and
then the soil element at this depth will fail whereby producing a large deformation in the lateral
direction. If there exist an open crack to this depth, it would be closed by this lateral bulge. Therefore
the depth at which a crack is closed may be given by
where t is unit weight of the soil. The exact value of the compressional strength of the loess at
Gissar is not known, but in view of the cementation developed in the matrix structure, the strength
may be inferred roughly to be within the range of qu = 200 400 kN/m2 . Then, assuming the
unit weight to be approximately t = 15 kN/m2 , the depth of crack penetration is estimated to be
about 15 to 25 m. Considering a dominant role played by the openness or closure of cracks, the
gross permeability of the loess deposit would probably have been distributed through the depth
444
Figure 26.
Correlation between residual strength and cone resistance.
agricultural irrigation. The windlaid loess consisting mainly of silt-sized soil is thought to have been
in 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 was
explained by the fact that the loess-forming silt is of low plasticity with a plasticity index of about
10. The result of the post-stability analyses made for four sites of landslides in Gissar indicated the
mobilized undrained residual strength of the silt to have been probably in the range between 2.0
and 15.0 kN/m2 which approximately coincides with similarly evaluated values of residual strength
in other case studies of flow failures.
ACKNOWLEDGEMENTS
This paper is the reproduction of a paper published in Soils and Foundations, Vol. 30, No. 4, pp.
7387.
In conducting the field investigations at the landslide sites at Gissar, Republic of Tajikistan the
overall cooperation offered by the Institute of Seismic Resistance Construction and Seismology,
and the Geological Bureau, Tajikistan Academy of Sciences, was essential and most helpful. The
authors wish to express their sincere gratitude to Drs. K. Morzoev, A. Ischuk and D.Y. Pabrobich,
for their kindness in arranging the field trip and in offering much important information on the
earthquake and soil conditions. The authors also wish to express their thanks to Drs. D. Semyonov
and I. Zasursky of the USSR Academy of Science at Moscow for their kind help for arranging a
trip to Tajik.
REFERENCES
Chen Z.Y. 1987. Problems of Regional Soils, Proc. 8th Asian Regional Conference Soil Mechanics and
Foundation Engineering 2: 167190. Kyoto: Japan.
Clevenger, W.A. 1958. Experiences with Loess as Foundation Material, Transaction of ASCE, 2916: 151169.
Dobry, R. & Alvarez, L. 1967. Seismic Failures of Chilian Tailings Dams, Proc ASCE 93(SM6): 237260.
Dudley, J.H. 1970. Review of Collapsing Soils, Journal of ASCE SM3: 925947.
Gibbs, H.J. & Bara5 J.P. 1967. Stability Problems of Collapsing Soil, Journal of ASCE, (SM4): 577594.
447
Ishihara, K. & Koseki, J. 1989. Cyclic Shear Strength of Fines Containing Sands, Proc. Earthquake Geotechnical Engineering, 12th International Conference on Soil Mechanics and Foundation Engineering: 101l06.
Rio de Janeiro
Ishihara, K., Yasuda, S. & Yoshida, Y. 1990. Liquefaction Induced Flow Failure of Embankments and Residual
Strength of Silty Sand, Submitted to Soils and Foundations.
Ishihara, K. 1977. Pore Water Pressure Response and Liquefaction of Sand Deposits during Earthquakes,
Proc. Dynamic Methods in Soil and Rock Mechanics2: 161193. Karlsruhe
Mitchell, J.K. 1976. Fundamentals of Soil Mechanics, John Wiley & Sons Inc:7679.
Seed, H.B. 1987. Design Problems in Soil Liquefaction, Journal ASCE, Vol. 113p GT8s: 827845.
448

Takaji Kokusho-Earthquake Geotechnical Case Histories For Performance-Based Design-CRC (2009)

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Takaji Kokusho-Earthquake Geotechnical Case Histories For Performance-Based Design-CRC (2009)

Transféré par

Droits d'auteur :

Formats disponibles

EARTHQUAKE GEOTECHNICAL CASE HISTORIES

FOR PERFORMANCE-BASED DESIGN

2009 Taylor & Francis Group, London, UK

Earthquake Geotechnical Case

2009 Taylor & Francis Group, London, UK

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

2009 Taylor & Francis Group, London, UK

Earthquake geotechnical case histories for performance-based design Kokusho (ed)

Geotechnical performance of the Kashiwazaki-Kariwa Nuclear Power Station

2006 large-scale rockslide-debris avalanche in Leyte Island, Philippines

Slope failures during the 2004 Niigataken Chuetsu earthquake in Japan

Fill slope failure of the Takamachi housing complex in the 2004

Fluidisation and subsidence of gently sloped farming fields reclaimed

Quay wall displacements and deformation of reclaimed land during

River dike failures during the 1993 Kushiro-oki earthquake

Ground failures and their effects on structures in Midorigaoka

Behaviour of SCP-improved levee during 2003 Miyagiken-Hokubu Earthquake

Earthquake geotechnical case histories for performance-based design Kokusho (ed)

Earthquake geotechnical case histories for performance-based design Kokusho (ed)

About the Editor

Earthquake geotechnical case histories for performance-based design Kokusho (ed)

Subsidence around Unit 1 light oil tank.

Rupture of outdoor fire protection piping.

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

Sand boil near Unit 1 emergency intake

Sand boil on the north side of Unit 4

Figure 22. Cracks in ground near Unit 5 seawater

Figure 23. Cracks along a buried duct of Unit 5

Subsidence contour map (Units 14, Arahama side area).

Subsidence of ground near Unit 1 seawater heat exchanger building.

Figure 31. Thickness and extent of backfill soil.

Groundwater level distribution (October/November 2007).

2009 Taylor & Francis Group, London, UK

2009 Taylor & Francis Group, London, UK

Locations of exploratory boreholes for measurement of backfill soil.

Physical properties of backfill soil near the building.

Backfill soil grain size distribution curves.

Double amplitude of strain DA=1%

Cyclic shear stress ratio d / 2 0

Earthquake geotechnical case histories for performance-based design Kokusho (ed)

2006 large-scale rockslide-debris avalanche in Leyte

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

geological, geomechanical and hydrological processes, and no conclusive triggering mechanisms

2 GEOLOGIC, TOPOGRAPHIC AND TECTONIC SETTING

Cumulative Rainfall (mm)

Daily Rainfall (mm)

3.2.1 Rainfall records

Maasin, Southern Leyte Station (MSLP)

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

2 GENERAL BACKGROUNDS OF SLOPE FAILURES

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

Depth (D) (km)

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

Figure 7. Schematic image of 3 types of slope

Figure 8. Horizontal slope displacements versus

2.3 Classifications of slope failures

Incident energy at baserock EIP (KJ/m2 )

Figure 15. Idealization of sliding soil mass by a

Figure 14. Cross-sectional change of Higa

Idealization of sliding soil mass by a

0 rectangular block in Haguro Tunnel entrance slope.

Figure 31. Cross-sectional change of Haguro

Figure 50. Idealization of sliding soil mass by a