Advances in Geotechnical Earthquake Engineering - Soil Liquefaction and Seismic Safety of Dams and Monuments

ADVANCES IN
GEOTECHNICAL
EARTHQUAKE
ENGINEERING SOIL
LIQUEFACTION AND
SEISMIC SAFETY OF DAMS
AND MONUMENTS

Edited by Abbas Moustafa

Advances in Geotechnical Earthquake Engineering Soil Liquefaction and Seismic
Safety of Dams and Monuments
Edited by Abbas Moustafa

Published by InTech
Janeza Trdine 9, 51000 Rijeka, Croatia

Copyright 2012 InTech
All chapters are Open Access distributed under the Creative Commons Attribution 3.0
license, which allows users to download, copy and build upon published articles even for
commercial purposes, as long as the author and publisher are properly credited, which
ensures maximum dissemination and a wider impact of our publications. After this work
has been published by InTech, authors have the right to republish it, in whole or part, in
any publication of which they are the author, and to make other personal use of the
work. Any republication, referencing or personal use of the work must explicitly identify
the original source.

As for readers, this license allows users to download, copy and build upon published
chapters even for commercial purposes, as long as the author and publisher are properly
credited, which ensures maximum dissemination and a wider impact of our publications.

Notice
Statements and opinions expressed in the chapters are these of the individual contributors
and not necessarily those of the editors or publisher. No responsibility is accepted for the
accuracy of information contained in the published chapters. The publisher assumes no
responsibility for any damage or injury to persons or property arising out of the use of any
materials, instructions, methods or ideas contained in the book.

Publishing Process Manager Igor Babic
Technical Editor Teodora Smiljanic
Cover Designer InTech Design Team

First published February, 2012
Printed in Croatia

A free online edition of this book is available at www.intechopen.com
Additional hard copies can be obtained from [email protected]

Advances in Geotechnical Earthquake Engineering Soil Liquefaction and Seismic Safety
of Dams and Monuments, Edited by Abbas Moustafa
p. cm.
ISBN 978-953-51-0025-6

Contents

Preface IX
Chapter 1 Lessons Learned from Recent Earthquakes
Geoscience and Geotechnical Perspectives 1
Robert C. Lo and Yumei Wang
Chapter 2 Lateral In-Situ Stress
Measurements to Diagnose Liquefaction 43
Richard L. Handy
Chapter 3 Review on Liquefaction Hazard Assessment 63
Neelima Satyam
Chapter 4 Liquefaction Remediation 83
Sarfraz Ali
Chapter 5 Simplified Analyses of Dynamic Pile Response
Subjected to Soil Liquefaction and
Lateral Spread Effects 113
Lin Bor-Shiun
Chapter 6 Non-Linear Numerical Analysis of
Earthquake-Induced Deformation of Earth-Fill Dams 139
Babak Ebrahimian
Chapter 7 Selection of the Appropriate Methodology for
Earthquake Safety Assessment of Dam Structures 167
Hasan Tosun and Evren Seyrek
Chapter 8 Earthquake Response Analysis and
Evaluation for Earth-Rock Dams 189
Zhenzhong Shen, Lei Gan, Juan Cui and Liqun Xu
Chapter 9 Recent Landslide Damming Events and
Their Hazard Mitigation Strategies 219
Ahsan Sattar and Kazuo Konagai
VI Contents

Chapter 10 Rate-Dependent Nonlinear Seismic
Response Analysis of Concrete Arch Dam 233
Xiao Shiyun
Chapter 11 Seismic Potential Improvement of Road Embankment 269
Ken-ichi Tokida
Chapter 12 Seismic Response Analysis and Protection of
Underground Monumental Structures
The Catacombs of Kom EL-Shoqafa,
Alexandria, Egypt 297
Sayed Hemeda
Chapter 13 Seismic Protection of Monolithic Objects of
Art Using a Constrained Oscillating Base 333
Alessandro Contento and Angelo Di Egidio
Chapter 14 Application of a Highly Reduced One-Dimensional
Spring-Dashpot System to Inelastic SSI Systems
Subjected to Earthquake Ground Motions 359
Masato Saitoh
Chapter 15 Numerical Prediction of Fire Whirlwind
Outbreak and Scale Effect of Whirlwind Behavior 383
Seigo Sakai
Chapter 16 The Vibration of a Layered
Rotating Planet and Bryans Effect 405
Michael Y. Shatalov, Stephan V. Joubert and Charlotta E. Coetzee

Preface

Despite the recent progress in seismic-resistance design of structures, earthquakes
remain the first natural hazard causing large life loss and massive property destruction
worldwide. The recent 2010 Haiti earthquake and the 2011 Japan earthquake are notable
examples on life and economic losses in developing and developed countries. The 2010
Haiti earthquake killed more than 250,000 persons and left a long-term suffers for the
people of that country. The 2011 Tohoku earthquake and the associated tsunami caused
enormous economy loss and massive destructions to engineering structures off the
Pacific coast of Tohoku in Japan. In fact, each new earthquake brings surprises with it
that teach earthquake and structural engineers new lessons. The field of earthquake
engineering has gained crucial advances during the last six decades or so starting from
the use of analog seismographs, digital seismographs to the use of modern technologies
and design methods such as sensors, structural control, health assessment and optimum
design of structures under dynamic loads.
This book sheds lights on recent advances in earthquake engineering with special
emphasis on soil liquefaction, soil-structure interaction, seismic safety of dams and
underground monuments, mitigation strategies against landslide and fire whirlwind
resulting from earthquakes.
The book contains sixteen chapters covering several interesting topics in earthquake
engineering written by researchers from several countries. Chapter 1 provides a
comprehensive review on lessons learned from earthquakes with special emphasis on
geoscience and geotechnical aspects. Chapters 2-6 are devoted to soil liquefaction during
earthquakes and its effect on engineering structures. Chapter 2 focuses on lateral in-situ
stress measurements to diagnose soil liquefaction. Chapter 3 deals with hazard
assessment due to soil liquefaction. The evaluation and remediation of soil liquefaction is
addressed in chapter 4. Chapter 5 tackles the problem of seismic response of piles with
soil liquefaction and lateral spread effects. Chapter 6 investigates the non-linear analysis
of induced deformations and liquefaction of earth dams.
Chapters 7-11 are related to seismic response analysis and safety assessment of dam
structures. Chapter 7 deals with the selection of appropriate technique for safety
assessment of dams. The seismic response and safety of earth-rock dams is studied in
chapter 8. Chapter 9 explores the recent landslide of damming events and their hazard
X Preface

mitigation strategies. In chapter 10, the rate independent non-linear seismic response
of arch dams is presented. Chapter 11 focuses on the seismic potential improvement of
road embankments. The response analysis of underground monuments under
earthquake ground motions is studied in chapter 12 with focus on the Catacombs of
Kom El-Shoqafa in Egypt. Chapter 13 studies the seismic protection of monolithic
objects of art using a constrained oscillating base. Chapter 14 examines the application
of a highly reduced one-dimensional spring-dashpot system to inelastic soil-structure
interaction systems under strong ground motions. Chapter 15 study the numerical
prediction of fire whirlwind out break due to earthquakes with emphasis on the recent
2011 Tohoku Japan earthquake. The last chapter of the book handles the vibration of a
layered rotating plant and Bryan's effect.
I hope this little effort benefits graduate students, researchers and engineers working
in the filed of structural/earthquake engineering. I'd like to thank authors of the
chapters of this book for their cooperation and effort during the review of the book.
Thanks are also to my teachers, C S Manohar, Indian Institute of Science, Sankaran
Mahadevan, Vanderbilt University and Izuru Takewaki, Kyoto University who put
my feet in the field of earthquake engineering and structural reliability.

Prof. Abbas Moustafa
Department of Civil Engineering,
Faculty of Engineering,
Minia University,
Egypt

1
Lessons Learned from Recent Earthquakes
Geoscience and Geotechnical Perspectives
Robert C. Lo
1
and Yumei Wang
2

1
Klohn Crippen Berger Ltd., Vancouver, B.C.
2
Sustainable Living Solutions LLC, Portland, Oregon
1
Canada
2
U.S.A.
1. Introduction
Earthquakes have been occurring long before human development, and will continue to
occur with or without human civilization. Natures forces behind earthquakes are powerful,
unstoppable and can be deadly. Recent earthquakes illustrate these destructive forces across
the globe. In Japans March 2011 disaster, nearly 24,000 persons perished or missing in the
worlds most seismically prepared country with advanced early warning systems for
tsunami and earthquake. In January 2010 at Haiti, a developing nation, even worse
devastation occurred with about a quarter million fatalities.
Earthquake disasters are often covered in the news media for a short time period. However,
after the media blitz fizzles out, the recovery period ensues. Recovery can involve extreme
socio-economic hardship - painful emotional losses, physical injuries, public health crisis,
widespread environmental contamination and loss of homes and businesses. This
readjustment could last for many years. With todays increasing population and economical
development in seismic hazard zones (in both developing and developed nations), the
global seismic risk is also going up. The field of earthquake science has seen many recent
advances, some involving geoscience and geotechnical issues. Synthesized in this chapter
are key advances gleaned from literature that can be applied towards risk management
decisions to reduce future loss of lives and socio-economic disruptions.
As members of the Earthquake Investigation Committee (EIC) of ASCE Technical Council
on Lifelines Earthquake Engineering (TCLEE), the authors have been involved in the
investigation for four (Sumatra, Wenchuan, Maule and Tohoku-Oki) of the six recent
earthquakes covered in this chapter, focusing on the geoscience and geotechnical aspects.
This chapter first highlights the characteristics and damages of these earthquakes: the
2004/2005 Sumatra, Indonesia, 2008 Wenchuan, China, 2010 Haiti, 2010 Maule, Chile,
2010/2011 Christchurch, New Zealand, and 2011 Tohoku-Oki (East Japan) earthquakes (see
Table 1). It then discusses some of the geoscience and geotechnical aspects of these
earthquakes with references to other relevant seismic events. Finally, it outlines the lessons
learned from these events in general as well as with respect to lifelines facilities, and draws
some conclusions.
Advances in Geotechnical Earthquake Engineering
Soil Liquefaction and Seismic Safety of Dams and Monuments

2
2. Recent earthquakes
2.1 General
Table 1 summarizes the characteristics and damages of the six recent events. It provides a
thumb-nail sketch of these events including: date and location, earthquake type and focal
mechanism, peak ground acceleration and Modified Mercalli Intensity, special features,
casualties, damages and general references. Three of these are tsunami-generating
subduction events of magnitude, Mw 8.8 to 9.1-9.3 (see Fig. 1), while the other three are
crustal events of magnitude, Mw 6.0 to 7.9, involving blind thrust, strike-slip/thrust or
reverse faults. Prominent features of these events are briefly outlined below.
2.2 Prominent features
2.2.1 2004 (Mw 9.1-9.3)/2005 (Mw 8.6) Sumatra, Indonesia earthquakes/tsunamis
The December 26, 2004 Sumatra earthquake was triggered by the rupture of a locked
segment of the fault plane at least 500 km long by 150 km wide between the subducting
Indo-Australian Plate and the upper Eurasian (Burma) Plate (see Fig. 2). Figure 3 shows the
computed vertical and horizontal components of surface displacements of the upper plate
based on a finite-fault model (EERI 2005, 2006, ASCE 2007, BSSA 2007).

Fig. 1. Major Circum-Pacific Subduction Earthquakes Since 1957. http://outreach.eri.u-
tokyo.ac.jp/ eqvolc/201103_tohoku/

Fig. 2. Three-Dimensional Sonar Imagery of Seabed off the coast of Sumatra Island. (BBC 2005)

Lessons Learned from Recent Earthquakes Geoscience and Geotechnical Perspectives

3

Table 1. Summary of Relevant Earthquake and Damage Data for Six Recent Earthquakes In
2004 to 2011

4

Table 1. (continued). Summary of Relevant Earthquake and Damage Data for Six Recent
Earthquakes In 2004 to 2011


5
The stealthy nature of tsunami onslaught of coastal inhabitants and international tourists
around the Bay of Bengal and Indian Ocean (see Fig. 4) in the morning after Christmas of
2004 and ensuing heavy casualty focused the worlds attention at the time. The tsunami run-
up height had considerable variation around the Indian Ocean, but ranged in general from 2
to 5 m, and reaching a maximum of 31 m in Sumatra (see Fig. 5). The event served as an
impetus to improve the tsunami-warning system for the countries in the region. Although
the subsequent smaller event on March 28, 2005 further south involved nominal tsunami
waves reaching 1 to 3 m height locally and did far less damage, it stirred up considerable
local fear due to the dreadful earlier event.
No strong-motion acceleration time histories were recorded in the epicentral region. In the
near-field northwest and north Sumatra, tsunami compounded earthquake-shaking
damage. In the far-field tsunami was the predominant cause of destruction. The severity of
tsunami damage was affected by many factors such as bathymetry, shoreline configuration
and topography, etc. which influenced the wave focusing, reflection and refraction; tsunami
run-up height; extent of inland inundation; flow velocity and scour, wave pressure, uplift
and debris impact force. EERI (2006) noted the following tsunami-related phenomena:
Maldives suffered moderate damage, although the coral-atolls archipelago rises only
about 2 m above the mean sea level. Since the islands rise from the seafloor steeply,
wave amplification was nominal.
The Indian mid-ocean ridges served as wave guides, and funnelled the tsunami away
from the tip of Africa.
The tsunami generating capacity of an earthquake is governed by the mass of the water
body suddenly displaced by the seafloor movement. The presence of the Nias and
Simeulue Islands reduced the affected water body during the 2005 earthquake, thus
induced relatively low tsunami.
Unlike tidal gauges that could be affected by harbour resonance, tsunameters can
indicate free-field tsunami height.

Fig. 3. Modelled Surface Displacements of the Upper Plate in Metres, for Vertical (Left) and
Horizontal (Right) Components, 2004. http://neic.usgs.gov/neis/eq_depot/2004/
eq_041226/neic_slav_ff.html

6

Fig. 4. Tsunami-Damaged Countries Around Indian Ocean, 2004.
(UN OCHA 2005)
Two types of leading waves of tsunami were modelled back in 1994: a leading
depression N-wave (LDN) and a leading elevation N-wave (LEN). Tide-gauge records
on Fig. 6 confirmed the validity of the earlier hydrodynamic modelling: LDN wave was
shown on Phuket, Thailand record, while LEN wave on Male, Maldives record.
Historically, the leading depression N-wave, i.e., that causing the initial receding of the
shoreline as the tsunami approaching, has been a death trap for many unwary
fishermen and beachcombers.

Fig. 5. Representative Tsunami Runup Heights along Shores of Indian Ocean, 2004.
(EERI 2006)


7

Fig. 6. Tide Gauge Records from Male, Maldives and Phuket, Thailand, 2004. (EERI 2006)
2.2.2 2008 (Mw 7.9) Wenchuan, China earthquake
The latent threat of the causative Longmenshan fault system was formally recognized by the
geoscience research community about a year prior to the 2008 event, but this finding did not
influence the seismic code at the time. The major seismic event of Mw 7.9 impacted a large
region in the southwest China, involving several provinces that were significantly under-
designed for the event. Figures 7 and 8 show the conditions of the old Beichuan town before
and after the earthquake.

Fig. 7. Old Beichuan Town Before Earthquake. http://www.eeri.org/site/meetings/us-
china-us-china-symposium

Fig. 8. Old Beichuan Town After Earthquake. http://www.eeri.org/site/meetings/us-
china-us-china-symposium

8
Due to the steep and rugged topography in the affected mountainous region, wide spread
landslides (see Figs. 9 and 10) have been major destructing factors, besides strong
earthquake shaking (EEEV 2008, EERI 2008). About 20,000 fatalities, near one-fourth of the
total, were caused by 15,000 geohazards in the form of landslides, debris flows and rockfalls,
with the largest landslide involving a volume of 1.1 billion m
3
. In the high, steep slopes (see
Fig. 10) along the 270 km long Longmenshan tectonic belt, the large vertical acceleration and
topographic amplification of ground motion have resulted in more than 10,000 potential
geohazard sites after the event (Yin et al. 2011).

Fig. 9. Landslides in Proximity of Old Beichuan Town. http://www.agu.org/news

Fig. 10. Landslides along Minjiang River near Wenchuan County. (EEEV 2008)


9
Up to 34 landslide lakes were formed in Sichuan and one in Gansu provinces, threatening
about 700,000 people living downstream. The largest one was located in Tangjiashan,
Beichuan County (see Fig. 11), with a 71-m high debris dam blocking the Shitingjian river
forming a lake about 800 m long and 600 m wide. The downstream flood-threatened area
had to be evacuated, and the debris dam breached by excavation and blasting to remove the
secondary flood hazard.
Figure 12 shows the number and severity of landslides per km of National Highway
Route #213 over the hanging wall versus foot wall. As expected in a thrust-fault earthquake
(Sommerville 2000), there is substantially more damage over the hanging wall as compared
to that over the foot wall. Similarly, there is more damage in the area along the earthquake
propagation direction than in the opposite direction.
The strong shaking with peak ground acceleration up to 0.98 g, ground failures and fault
displacements up to 2 to 4 m caused wide spread destruction of communities and
infrastructures. The long duration of strong shaking, over 100 seconds in general, was
detrimental to unreinforced masonry buildings, and non-ductile reinforced concrete
buildings that form the bulk of the building stock in the affected area. Figure 13 shows the
acceleration response spectra at Qingping Station in the epicentral region (Ventura et al.
2008). Superimposed on the figure is the design acceleration spectra for Vancouver, British
Columbia with Site Class C local soil condition, according to NBCC (2005) code for
comparison purposes. As typical in most earthquakes, vertical peak ground acceleration is
of similar value as its horizontal counterparts in the near field.

Fig. 11. Landslide Lake at Tangjiashan Landslide. Engineering Resilient Cities - Mahin - Oct
2008
Initially, accesses to remote areas were handicapped by the disruptions of highways and
railways. The prompt and orderly nation-wide rescue and restoration programs were
responsible for mitigating the suffering of affected population and the recovery of the region
to normalcy. The unique Chinese mechanism for emergency response, recovery and

10

Fig. 12. Landslide Activities above Hanging Wall versus Foot Wall. (EEEV 2008).
reconstruction involved communities and jurisdictions located far away from the
damaged areas. This twinning of communities in need and those to help accomplished
dual goals: sharing of enormous financial hardship; and cultivating camaraderie among
population. Figure 14 shows some of the officials who had spent several months in a
donated school to assist relocated residents from outlying communities including the
neighbouring province. Temporary dwelling units were set up across the earthquake
damaged region. Figure 15 shows that the communities thus set up have become new
settlements with all amenities to conduct normal life. Residents were finding work both
within the settlement and outside. Grain drying activity was seen in the foreground of the
figure (Lo 2009).

Fig. 13. Acceleration Response Spectra for Qingping Station in Epicentral Region. (Ventura
et al. 2008)


11

Fig. 14. School Donated for Temporary Accommodation of Survivors in Chongzhou, 2008.
A comprehensive three-year reconstruction program covered the management organization
and socio-economical structure for regional revitalization, and was carried out by the
twinned communities. A new Beichuan town was constructed in Yongchang City about
25 km downstream of the destructed town with many traditional architectural elements of
the local Qiang minority (see Fig. 16).

Fig. 15. Relocated Community in Temporary Accommodation of Chongzhou, 2008.

12

Fig. 16. City Scene Along Yongchang Boulevard, Yongchang, Sichuan New Beichuan
Town. http://www.skyscrapercity.com/ showthread.php?t=1222823
Figure 17 shows the Yongchang River Bank, and Figure 18, an apartment building
constructed in the new town with the support of a twinned city located near the northeast
coast of China, about 1,400 km away.

Fig. 17. Yongchang River Bank, Yongchang http://www.skyscrapercity.com/
showthread.php?t=1222823


13

Fig. 18. New Apartment Building Constructed with Support from Linyi City, Shandong
Province, 2010. http://www.eeri.org/site/meetings/ us-china-symposium
2.2.3 2010 (Mw 7) Haiti earthquake
Haiti has suffered devastating earthquakes similar to the 2010 event in the past, despite
recent seismic quiescence. The 2010 earthquake is caused by a combination of reverse and
left-lateral strike-slip faulting related to the Enriquillo-Plantain Garden fault system. The
threat of this specific fault to the population was not recognized prior to the event (EERI
2010, USGS/EERI 2010). There is no strong motion record for the main shock in Haiti. Peak
ground acceleration was estimated in the range of 0.3 to 0.45 g in the affected area. The
building stocks, consisting mainly of unreinforced masonry and non-ductile reinforced
concrete structures (see Fig. 19) including government buildings (see Fig. 20) and buildings

Fig. 19. Damaged Buildings Located on Hill Slope. http://www.worldcatastrophe.com

14
used by the United Nation Stabilization Mission Headquarters, are generally inadequate to
withstand this level of shaking (Fierro and Perry 2010). The earthquake fatality was
estimated at about 233,000 to 250,000, with 300,000 injured and 1 million homeless.
Subsequent to the earthquake outbreaks of cholera caused about 1,000 deaths as of October
2010 due to the poor hygiene condition.

Fig. 20. Damaged Presidential Palace in Port-au-Prince. http://alanrdennis.wordpress.com
After the event, USGS (2010) developed initial seismic hazard maps for Haiti to assist the
management of earthquake response and reconstruction effort. The increased seismic
hazard along the adjacent portion of the Enriquillo fault directly to the south of Port-au-
Prince must be considered in the post-earthquake reconstruction.
2.2.4 2010 (Mw 8.8) Maule, Chile earthquake/tsunami
In the 2010 Maule, Chile earthquake, the tsunami contributed close to half of the total
casualty of about 521 due to the timing of the event occurring in the early morning. Tsunami
damage to the over 500 km long coastline varied, as it came in during low tide and involved
3 to 4 major surges. Tsunami run-up height generally ranged from 3 to 9 m in the epicentral
region with the maximum reaching about 20 m along cliff faces.
The relative low casualty speaks for the seismic-resistant capacity of the general building
stocks. The fact that high-occupancy buildings were vacant at the time of earthquake also
helped to reduce casualty. Damaged or collapsed buildings suffer design flaws such as lack
of vertical continuity of load-bearing structural elements, relatively thin shear walls and lack
of detailing requirements for reinforcing steel in special boundary elements, etc.
(GEER 2010). After the earthquake, looting has been reported in the affected region until the
army was deployed to maintain law and order.
Figure 21 shows the modelled slip distribution over the fault plane in the land area. The
maximum fault slip along the coast line is about 300 cm. The earthquake occurred in the


15
summer with prevailing low water table. It reduced the occurrences of landslides and
liquefaction. The relatively long duration of earthquake shaking did trigger slope slumps in
marginally stable or wave-undermined natural slopes, and liquefaction and lateral
spreading in low-lying areas with high water table. Figure 22 shows a concrete pier used for
unloading fish catches from small vessels at the Coronel fishing port. Its failure appears to
be caused by the combination of earthquake shaking, foundation liquefaction and possibly
tsunami wave force.

Fig. 21. Fault Plan Slip Distribution over land area Maule, Chile.
http://tectonics.caltech.edu/sliphistory/2010_chile/index.html
The flow failure of a tailings dam triggered by the earthquake is shown in Figs. 23 and 24
(GEER 2010). Compared with a large number of tailings dam failures in the 1965 La Ligua
earthquake and two failures in the 1985 Santiago earthquake, tailings dam performance in
this event seems to show some improvement. Compacted engineering fills and structures
supported by improved ground using stone columns and micro-piles were reported to
perform well.

16

Fig. 22. Distorted Fish-Unloading Concrete Pier at Coronel Port, Chile. (TCLEE 2010)

Fig. 23. Birds-eye View of Las Palmas Tailings Impoundment Prior to Failure. (GEER 2010)

Fig. 24. Upper Scarp of Failed Tailings Impoundment. (GEER 2010)


17
2.2.5 2010/2011 (Mw 7.0) Darfield (Canterbury) / (Mw 6.1) Christchurch New Zealand
earthquakes
Both the September 2010 Mw 7.0 Darfield and February 2011 Mw 6.1 events were shallow
crustal earthquakes, which triggered wide-spread liquefaction in the fluvial and diluvial
deposits in the epicentral region (see Figs. 25 and 26). Peak ground accelerations spiking
above 1 g, often with higher vertical than horizontal acceleration, were reported in the
epicentral areas of both events. In the central business district of Christchurch, the largest
city on the South Island, peak acceleration reached about 0.7 g on February 22, 2011 as
compared to about 0.3 g on September 4, 2010. Figure 27 shows the horizontal acceleration
response spectra for Christchurch Hospital with 5% damping. It illustrates that the
aftershock is reflected by the current New Zealand 2,500-year design spectrum, while the
main shock is corresponding to the 500-year design spectrum. Proposal to increase the
design spectrum for periods less than 1.5 sec is being considered (EERI 2010/2011).

Fig. 25. Liquefaction Maps: Sept. 2010 (Red) and Feb. 2011 (Yellow). (GEER 2011)

Fig. 26. Liquefied Sand Deposited on Street, 2011. http://otilya.com/view/ News/Asia-
Pacific/New-Zealand

18
The smaller aftershock caused significant casualties and heavier structure damages in
Christchurch (see Figs. 28, 29 and 30) than the main event, because it occurred much closer
to the central business district, where many buildings had already suffered varying degrees
of damage due to the prior event.

Fig. 27. Horizontal Acceleration Response Spectra for Christchurch Hospital. (EERI 2011)

Fig. 28. Scene of Downtown Christchurch Captured by a Tourist During Earthquake, 2011.
http://otilya.com/view/christchurch_ as_the_quake_hits.html


19
Unreinforced masonry and non-ductile reinforced concrete buildings suffered most
structural damages, while liquefaction and lateral spreading contributed to foundation
failures or differential settlement of buildings and pipeline breaks. Non-structural damages
to building components and content caused food-supply shortage and slowed down
business recovery. Other damages were caused by rock falls in the Port Hills area.

Fig. 29. Two Survivors Comforting Each Other During Rescue Operation at Pyne Gould
Building, 2011. http://www.kiwiblog.co.nz/2011/03/News/Asia-Pacific/New-Zealand

Fig. 30. Topographic Amplification May Play a Role in Damaging an Unreinforced Masonry
Building, 2011. http://otilya.com/view/NewsAsia-Pacific/New-Zealand
2.2.6 2011 (Mw 9.0) Tohoku-Oki (East Japan) earthquake/tsunami
Figure 31 shows the modelled slip distribution over the fault plane which triggered the
Tohoku-Oki earthquake. The slip on the offshore rupture plane was about 30 m, but the slip

20
along the east coastline reduced to below 5 m. Figure 32 shows acceleration time histories
recorded in the earthquake region (NIED 2011). These records have been used to analyze
structures such as the study of the response of high-rise buildings to long-period ground
motions (Takewaki et al. 2011). It is worth to note that some of the records in Fig. 32 show
multi-sequences of earthquake excitations. Moustafa and Takewaki (2010) indicated that
such earthquake sequences would result in more severe structural damages due to the
accumulation of inelastic deformations. Substantial damages were suffered by the buildings
and infrastructures in such a major subduction event of long duration due to the combined
effect of earthquake shaking and tsunami onslaught. Earthquake subsidence caused some
coastal areas submerged in tidal water (see Fig. 33). JSCE (2011) provided an overview of the
characteristics of the earthquake and tsunami; and damage reconnaissance including:
geotechnical and structural damage, liquefaction, damage to dykes and levees, and bridges.

Fig. 31. Fault Plane Slip Distribution Tohoku-Oki, Japan. http://tectonics.caltech.edu/slip_
history/2011_taiheiyo-oki/index.html
Tsunami run-up height was in a range of 4 to 8 m, with a maximum of up to 38 m. The
existing tsunami-protection seawalls and gates in the coastal region were overrun in places
causing significant destruction. As a result, the numbers of casualty (15,840) and missing
people (3,546) tend to be high for Japan, a seasoned country having invested heavily in the
preparation for and defence against the frequent seismic and tsunami events. Emergency
response efforts were hampered by fuel shortage, telecommunication and transportation
disruption, damage to fire and police stations and hospitals and nuclear radiation.
The destructive power of earthquake and tsunami has long been experienced and
recognized by human race since its existence. However, 2011 Tohoku-Oki earthquake has


21
revealed another relatively modern threat, i.e., the trigger of a long-lasting and potentially
deadly nuclear incident. Salt water intrusion into pumping systems resulting in equipment
failures has been reported earlier such as in the 1993 Hokkaido and 2004 Sumatra events.
This relatively minor problem usually only led to delay in the restoration of the water and
sewer systems. Nevertheless, at the Fukushima Dai-Ichi nuclear generating station (see
Fig. 34), the disruption of the cooling systems of nuclear reactors and power failure resulted
in the meltdown of reactor fuel rods.

Fig. 32. Ground Acceleration Time histories Recorded in Earthquake Region. (NIED 2011)
It is reported that during risk assessment of the nuclear power station prior to the
earthquake, a tsunami expert was overruled by an executive, when the survivability of the
reactors in a potential tsunami event was raised. Thus, the design flaw at this plant was not
corrected, while another plant with a newer design survived the current event without
problem. Construction records also revealed that in 1967, the owner, Tokyo Electric Power,
excavated 25 metres off the 35-metre high natural ground where the reactors were to be
located. The lowering of the site grade appeared to facilitate equipment transportation as
well as pumping of seawater for cooling the reactors.
The current solution is to abandon the power station, and to eventually entomb the
damaged reactors in concrete after their cool-down, even though ongoing radiation leak
continues to contaminate the plant site, its surrounding land and sea five months after the
earthquake (August 2011).
A mandatory 15 per cent cut of peak-time power consumption is in force this summer (2011)
for large users in and around Tokyo. Only nineteen of the 54 Japanese reactors that were in
service prior to the March 11 earthquake are working in July 2011, because of local
opposition. It is uncertain how long will it take to bring the radiation problem at the
Fukushima Dai-Ichi plant site under control. Germany and Switzerland have indicated their

22
plan to phase out nuclear power plants, while other countries have also starting the
reassessment of their future nuclear-power development plans. Thus, the radiation concern
for all nuclear power plants has to be carefully evaluated, and robust defence measures
against natural and man-made hazards have to be implemented around the world.

Fig. 33. Waterfront Buildings in Onagawa, Japan Inundated by Tidal Water Due to
Earthquake Subsidence. http://tectonics.caltech.edu/slip_history/2011_taiheiyo-
oki/index.html

Fig. 34. Blast-Damaged Fukushima Dai-Ichi Nuclear Power Plant Unit 3 (left) and Unit 4
(right).http://pinktentacle.com/


23
3. Geoscience and geotechnical aspects
Geoscience and geotechnical aspects of those earthquakes covered in Section 2 and Table 1
are discussed here together with other relevant events (Lo and Wang 2008).
3.1 Geoscience aspect
Earthquake events inflicted heavy loss of life and property as well as disrupting
businesses. They illustrate the high risk of structure inadequacy in populated and
developed areas vulnerable to seismic hazard.
Tsunami hazard to coastal population can be mitigated by advanced warning and
physical barriers such as sea walls and gates. However, when warning system is either
not established (Sumatra 2004) or insufficient evacuation time is available due to
proximity to the fault rupture, and the physical tsunami barriers are overrun (Tohoku-
Oki 2011), the tragic loss of lives and properties is severe.
Earthquakes usually occur according to a typical pattern, starting with a single main
event, maybe preceded by foreshocks, followed by a series of smaller aftershocks that
would diminish in magnitude and number with time. Exceptions to this pattern are not
uncommon. In fact, the concept of earthquake conversations (Stein, 2003 and 2005)
suggests that the hypothesis of stress-triggering of earthquake may be the mechanism
controlling earthquake occurrence. As population centres around the world grow from
isolated metropolitans to more or less continuous industrialized belts, the following
earthquake patterns observed in recent years deserve more attention in order to
mitigate damages.
Numerous large aftershocks of the Sumatra 2004, Maule 2010 and Tohoku-Oki 2011
subduction events, and the rupture of an adjacent segment of the locked plate
boundary offshore such as the Nias 2005 event or the increased threat to the Tokyo
region after the Tohoku-Oki 2011 event, and a Mw 6.9 crustal (or intraslab) normal-
fault event on March 11, 2010 in the region of Libertado O Higgins in Chile
triggered by the February 27, 2010 subduction event;
Several large regional seismic events occurred within several months after the
Wenchuan 2008 event due to stress readjustment in the region;
The Christchurch 2011 aftershock of the Darfield 2010 event occurred much closer
to the population centre;
Turkey Kocaeli (August) and Dzce (November) earthquakes in 1999 - triggered by
the rupture of two adjacent segments of the North Anatolian transform fault; and
El Salvador January and February earthquakes in 2001 - an offshore intra-slab
event followed by an inland crustal event.
Thus, major seismic events of various source mechanisms could occur in adjacent areas
within a short time interval. This phenomenon has significant implications in
earthquake design and restoration. First, there is a potential threat to the safety of
personnel engaging in the emergency response. Secondly, any retrofitted structure may
have to undergo severe seismic test during or shortly after its repair. It may no longer
be acceptable to lower the design criteria for retrofitted structures, even though it has
been currently practiced in many countries as the experience of Christchurch in 2010
and 2011 shows.

24
Earthquake investigation tools and methods have continually been evolved with
significant recent advances, such as:
Global Positioning System (GPS) GPS has been increasingly deployed along
major faults to provide continuous deformation data over the tracked GPS stations;
Interferometric Synthetic Aperture Radar (InSar) InSar is a satellite remote-
sensing technique used to track areal ground deformation adjacent to faults. Both
GPS and InSar are used to monitor the ground deformation along major faults
and/or in highly seismic areas before and after earthquake occurrence; and
Satellite imagery, such as Google Earth and high-resolution imagery, and Light
Detection and Ranging (LiDAR) have become standard tools used for the study of
post-earthquake damages. A workshop was conducted by the Global Earth
Observation Catastrophe Assessment Network (GEO-GAN 2010) to review the use
of high-resolution satellite imagery with the coordinated efforts by volunteering
professionals to conduct rapid structure damage assessments, geotechnical studies
and landslide surveys. The development started in 2008 for studying the
Wenchuan earthquake, and expanded in 2010 for the Haiti earthquake, involving
verification of the assessment results by field reconnaissance on the ground (see
www.virtualdisasterviewer.com).
Digital seismographs Newly installed digital seismographs capture a wide spectrum
of seismic ground motions accurately and efficiently and can be easily transmitted to
regional centers for further data processing than their earlier analog counterparts. The
number of strong-motion accelerograms available for seismic design analyses have
increased drastically in recent years including from four of the six events discussed here
with the exception of the 2004/2005 Sumatra and 2010 Haiti events.
Coral as natural record of relative sea-land movement (see Fig. 35) Coral has been
successfully used by researchers to study co-seismic and inter-seismic ground
movements caused by inter-plate subduction earthquakes (Caltech 2011).
Internet The internet has been used by the United States Geological Survey for
conducting earthquake intensity survey, and by the earthquake investigators to
coordinate international investigation efforts through dedicated websites, links,
individual blogs and social media. Such widespread use of electronic communication
could potentially be harnessed as an effective, informal means to spread tsunami
warning, especially to distant coastal areas to reduce the likelihood of repeating the
2004 Sumatra tsunami tragedy.
Information technology - Modern equipment such as digital camera, GPS, laptop and
satellite communication, etc. all contributes to the efficiency of an individual investigator.
3.2 Tsunami and geotechnical aspects
Earthquake secondary effects refer to non-tectonic surface processes that are related to
earthquake shaking. They are often spectacular in expression and are main causes for loss of
life and property. These effects include: tsunami, landslide, rockfall, turbidite (dense,
sediment-laden flow offshore), liquefaction and lateral spreading, and site amplification, etc.
(Lo et al. 1996). Their preserved geologic signatures sometimes serve as paleoseismic
evidence for strong ground motion of prehistoric earthquakes (Yeats et al. 1997).


25
3.2.1 Tsunami effects
Forces associated with tsunami In addition to the inundation effect of rapidly moving
waves, tsunami also exerts additional physical forces to structures such as
hydrodynamic pressure including uplift (see Figs. 36, 37 and 38), debris impact and
wave scour (see Figs. 39 and 40). The overturning of seawalls could be caused by the
combined effect of tsunami hydrodynamic pressure, uplift and scour. Both passive and
active measures are used to mitigate tsunami damage. Passive measures are use of
buffer zone and/or natural barriers such as mangroves and sand dunes to avoid
building structures in near shore areas. Where structures are required in tsunami-
vulnerable zone, both inundation effect and tsunami-related forces have to be
considered in the overall design and operation of these structures. Well designed
concrete buildings, which have crawl space on the first floor to allow tsunami flowing
through, performed well (see Figs. 41 and 42). However, the soft-story effect under
seismic loading condition should be carefully evaluated.
The important impact of tsunami hydrodynamic pressure on the performance of port
facilities and bridges have been recognized for some time, vertical restrainers to
counteract the tsunami uplift force against bridge decks such as experienced in the
Maule and Tohoku earthquakes are proposed by Kawashima (2011). Policy to require
vessels leaving the ports and sailing to the sea upon receiving tsunami warning
would prevent damage to the vessels as well as port infrastructures from vessel
impact.

Fig. 35. Sea Level Changes Inferred from Truncated Coral Growth Rings. (Caltech 2011)
http://www.tectonics.caltech.edu/images/sumatra/coral_record_web.jpg

26

Fig. 36. Wharf Damaged by Wave Impact and Uplift, Bam Nam Kem, Phang Nga Province,
Thailand, 2004.

Fig. 37. Uplift of Bridge Deck In Spite of Longitudinal Stoppers Tohoku Earthquake.
(Kawashima 2011)


27

Fig. 38. Overturned Police Station, Onagawa, Japan, 2011.

Fig. 39. Pump Station Damaged by Wave Scour, Patong Beach, Phuket Province, Thailand,
2004.

28

Fig. 40. Scouring of Foundation Soils and Severing of Buildings Utility Lines, Phi Phi Island
Hospital, Thailand, 2004.

Fig. 41. Building with Crawl Space on the First Floor, Kamala Beach, Phuket Province,
Thailand, 2004.
3.2.2 Ground failure effects
Ground failure includes landslide, slope failure, liquefaction, lateral spreading, bearing
capacity failure, underground failure and ground displacement:
Landslide Correlations between the number of landslides and earthquake magnitude
have been established around the world. In regions with pyroclastic deposits, the
number of landslides for a given magnitude far exceeds those in other regions due to
the relative large void ratio and weak strength of these deposits. The areas threatened
by a landslide include the sliding land mass as well as the area traversed by the fast-


29
moving sliding debris (see Fig. 43). The presence of water in the soil mass, as reflected
in its degree of saturation, increases the probability of triggering a slide during an
earthquake. Thus, whether the occurrence of earthquake is in dry or wet season could
have significant difference in its impact. Heavy precipitation in earthquake stricken area
could trigger post-earthquake landslides, debris flows and/or lahars (mudflows of
volcanic debris). Moreover, the amount of water within the sliding mass and/or being
entrained along its moving path increases its run-out distance and, hence, downstream
impact. The local topographical detail along the debris travel path dictates the area
being buried by the slide debris. The formation of a landslide lake, caused by temporary
blockage of river by unstable slide debris, poses a special flood hazard. Timely
construction of a spillway to drain the lake is required to ameliorate the downstream
flood hazard such as the case with the Tangjiashan Landslide-Lake after Wenchuan
2008 earthquake (see Fig. 11).

Fig. 42. Close-up of Fig. 41, Showing Inter-connected Grade Beams and Columns.
Slope failure Failure in natural or man-made slope occurs where the available effective
shear strength is exceeded by prevalent shear stress due to the earthquake-induced pore
pressure increase and/or presence of inertia force during earthquake shaking. Slope
failure of highway and/or railway embankment could disrupt vital transportation link
and wrought havoc in the community life and regional economy as evidenced in all major
earthquakes. Cracks and upstream slope slump of the 26-m high Fena earth and rockfill
dam after Guam earthquake in 1993 required emergency repair by U.S. Navy Seebees.
Liquefaction and lateral spreading Liquefaction and lateral spreading could occur in
much gentle slopes or level ground, where the soil deposits involve relatively loose
cohesionless materials such as sand and silt and occasionally gravelly soils. Loss of
shear strength and associated ground movements would affect structures located in

30
areas with high groundwater table, such as low-lying areas, river banks and ports, and
lead to slope failure, lateral spreading, foundation horizontal displacement and
settlement or uplift, bearing-capacity failure, and rupture of lifelines. Emergency repair
of flood-control dykes or levees damaged by liquefaction of subsoils was often needed
such as after Tohoku-Oki earthquake (see Fig. 44). The Tohoku-Oki earthquake caused
levee damage at over 2,000 locations.

Fig. 43. Debris Flow at Huosigou, Chongzhou, China, 2008.
Bearing-capacity failure Foundation bearing-capacity failure also occurs in cohesive
soil due to the softening effect of earthquake shaking. The punch shear failure of
footings is associated with soil heave in the adjacent area. For tall and narrow buildings
with an aspect ratio (height/width) greater than 2, it could lead to severe tilting or


31
overturning of the building (see Fig. 45). This type of foundation behavior was observed
in Mexico City due to the presence of sensitive Mexico City clay during the 1985
Michoacan earthquake, and again in Adapazari City due to the presence of Holocene
alluvium during the 1999 Kocaeli earthquake.

Fig. 44. Levee Slope Repair and Sand Boils at Toe, Japan. (JSCE 2011)

Fig. 45. Overturned Building of High Aspect Ratio in Adapazari City, Turkey, 1999.
Surface expression of underground failure Sinkhole may occur during an earthquake.
The phenomenon appears to be puzzling until the cause is identified. Usually, sinkhole

32
occurs where pre-existing underground voids already formed prior to the earthquake.
These voids may be formed by solution of limestone, abandonment of mine workings
or piping erosion of fine soils by groundwater flow. In other situation, earthquake-
triggered failure of underground structure could also result in the formation of
sinkholes. The former examples are sinkholes occurred along highway culverts in
El Salvador and Nisqually earthquakes where piping erosion of backfill around the
culverts was present (see Fig. 46), and sinkholes occurred at the abutment of Matahina
Dam in New Zealand (Gillon and Newton 1991). The latter example is the sinkholes
formed due to the collapse of Bolu Tunnels after the 1999 Dzce earthquake. At
Christchurch, concern of voids formed underground due to massive loss of sand to the
surface because of liquefaction was raised (see Fig. 26).

Fig. 46. Sinkhole Formed by Collapse of Roadway During Earthquake, Washington, 2001.


33
Ground displacement Ground displacement includes tectonic movement or non-
tectonic surface processes. When it occurs within the footprint of structures, it affects
the structural stability and integrity. Where lifelines intersect fault traces, the impact of
potential ground displacement related to fault rupture is a major design challenge as
the experiences in Wenchuan, Kocaeli and Dcze earthquakes show. On the other hand,
differential settlement of approach fill and pile supported bridge deck at bridge
abutments is a frequently recurring scene, e.g., at Christchurch (see Fig. 47). Subsoil
volume decrease would lead to foundation settlement with no adjacent soil heave.
3.2.3 Ground shaking effect
Site amplification Site amplification of ground motion through soft soil deposits has
been well recognized and incorporated in the modern seismic design codes. However,
severe building damages suffered at soft sites at Adaparazi and Avcilar during 1999
Kocaeli earthquake reiterate the importance of this phenomenon. Avcilar is near
Istanbul and about 80 km west of the epicenter. The distance attenuation of ground
motion was more than compensated by local site amplification. Similarly, topographic
amplification of ground motion was also evident during Christchurch (see Fig. 30), and
all other recent earthquakes.
4. Lessons learned from earthquakes
4.1 General
Earthquake damage in a given area tends to reflect its social-economic condition. Thus, wide
spread damage to sub-standard structures and heavy loss of life seem to recur unfortunately
too frequently. Comartin et al. (2004) considered that reducing high loss of life from
earthquakes is the most important challenge facing the global earthquake engineering
community. Efforts underway to meet this challenge included: improving construction
practices in individual countries such as India and Columbia and promoting international
collaboration and information sharing. The EERI/IAEE World Housing Encyclopedia is a
valuable platform to spearhead this ongoing effort.
Conversely, in areas where earthquake-resistant design codes and engineering practices are
kept up-to-date, damage tends to be limited to older and inadequately designed structures,
or areas with special site conditions. However, this general observation may have
exceptions such as in Christchurch, New Zealand and Tohoku, Japan, when potential
seismic hazards are under-estimated.
Outlined in the following are some of the lessons learned from the recent earthquakes:
Tsunami warning Timely warning of tsunami and evacuation drill are prerequisites
for saving lives in such an event. Sumatra earthquake-tsunami is an unprecedented
tragedy to illustrate this point, and provided an impetus for the set up of a tsunami-
warning system for countries around the Indian Ocean. Adequate warning requires the
implementation of a system involving instrumentation on land and on ocean floor, and
international organization and communication protocol. The system has to be
maintained vigilantly to be effective. While warning against tsunamis triggered by
distant earthquakes is adequate in North America, further work is required for warning

34
and protection against tsunamis caused by nearby earthquakes as demonstrated in the
experience of 2010 Maule, Chile and 2011 Tohoku-Oki, Japan events.
For inhabitants living along vulnerable shorelines, running to high grounds or elevated
structures, when feeling earthquake shaking or seeing unusual signs of waves including
receding sea water or advancing wave, remains to be practical means to escape
drowning by tsunami. Tragedy in the recent Chilean and Japanese events, however,
remind us that the success of escape is not guaranteed, when the time window for
survival is measured in minutes rather than hours, especially for elder or handicapped
members of the population. The need for reliable near field (local) tsunami warning was
further demonstrated in the case of 2005 Nias earthquake. About three months after the
Sumatra earthquake, the tsunami-scared populace was all too eager to run for their life
after given the warning. Nevertheless, no significant tsunami was realized this time.
The ensuing evacuation commotion, and other false tsunami alerts in the wake of the
2004 Sumatra earthquake served to emphasize that the conduct of timely and reliable
tsunami warning and orderly evacuation remains both a technical and societal
challenge. Similar confusions to a lesser extent were also reported in some of the
tsunami alerts in the North America.
Tsunami vertical evacuation refuges - are an engineering option for enhancing public
safety against near-field earthquakes (FEMA 2008/2009). These structures should be
clearly identified as tsunami-proof refuges. They need to be designed for both
earthquake and tsunami loads, as well as to a safe height above the tsunami run-up,
allowing for rapid public ingress and upward movement (see Fig. 48). As a cost-
effective measure, any elevated structure in a tsunami-vulnerable zone such as water
tower, lighthouse or fire-lookout tower could be designed with additional capacity and
features to accommodate tsunami-escapees looking for an escape path.

Fig. 47. Settlement of Approach Fill at Bridge Abutment Formed a Bike-Jump for a
Youngster. http://otilya.com/view/News /Asia-Pacific/New-Zealand


35

Fig. 48. Vertical Evacuation Structure at Shirahama Beach Resort in Japan. (FEMA 2009)
Instrumentation for strong ground motion Installation and maintenance of modern
instrumentations for recording strong ground motion are important to characterize
earthquake motion. Further instrumentation to measure the response of critical
structures would improve the understanding and analysis of seismic behavior of the
instrumented structures. However, lack of strong motion data in epicentral areas at
Port-au-Prince, Haiti; Sumatra and Nias, Indonesia serve to indicate that it is not always
feasible to achieve this goal due to economic, political and/or technical constraints.
Moreover, potential tsunami inundation should also be considered when selecting sites
for strong motion instrumentation.
ASCE (2011) published a special issue on the selection and modification of
earthquake ground-motion for nonlinear dynamic analysis of structures. The suite of
17 papers provides useful information about the ground-motion input in the
nonlinear response-history analysis of structures. In particular, Moustafa (2011)
outlined a new framework for modelling design earthquake loads for inelastic
structures, using the method of critical excitations and adopting damage indexes to
reflect the structure damage level.
Land use Microzonation is to be prepared for identifying areas susceptible to tsunami
run-up, strong earthquake shaking, ground failures such as landslides, fault
displacements, slope failures and liquefaction, etc. Determination of appropriate land
use should then be guided by the identified hazards to mitigate potential loss.
Seismic design Update of seismicity hazard and upgrade of seismic design are to be
carried out on a regular basis. Although the replacement of vulnerable structures before
they are decimated by earthquake could prevent avoidable loss of life, funding required
for such an undertaking seems always difficult to come by, especially during the
current weak financial condition around the world. Up-to-date seismic design measures
should be implemented in new structures. Whenever possible, retrofit of existing
structures should be facilitated by legal requirements and/or financial incentives.

36
Emergency response and public education Drills of emergency response need to be
practiced regularly to enhance the capability of response agencies and personnel, and
public awareness. Preservation of certain damaged or failed structures in the
earthquake-stricken areas as memorial sites would go a long way to remind and
educate the public about the importance of adequate seismic design and earthquake
preparedness. Figure 49 shows the memorial park established at the preserved old
Beichuan townsite in honour of the victims, and Figure 50 shows the public taking a
sober earthquake lesson at the park.

Fig. 49. Memorial Park at Old Beichuan Town. http://www.eeri.org/site/meetings/ us-
china-symposium

Fig. 50. Educational Tour of Memorial Park. http://www.eeri.org/site/meetings/us-china-
symposium


37
Potential impacts on secondary damages by earthquakes such as tsunamis, fires, landslides,
debris flows, dam breaches, disruptions of lifelines and release of hazardous material
including radiation are to be considered in order to devise realistic response strategy, and to
plan for orderly implementation of long-term upgrade of existing facilities.
4.2 Lifelines facilities
Lifelines are those systems required for the survival and function of societies, and include:
water and wastewater, electric power, communications, transportations, natural gas and
liquid fuels, etc. Human societies often evolved on the nodal points of the rivers and ports
transportation networks located on flood plains. Thus, various geotechnical issues reviewed
in Section 3.2 affect the seismic performance of lifelines facilities, such as rupture of water
and waste water pipelines, disruption of power and communication lines, and damage to
bridges, power transmission and communication towers, etc. due to foundation failures,
liquefaction and lateral spreading.
Lessons applicable to lifelines facilities include the following:
Fault Displacement Fault displacement caused significant structure damages in the
Wenchuan earthquake. Earlier examples are the collapse of Arifiye overpass across the
Trans-European Motorway (TEM) in the Kocaeli earthquake (with about 3 to 3.5 m fault
displacement) and significant displacement and damage to the Bolu Viaducts in the
Dzce earthquake (with 1.5 m displacement of adjacent piers and 13
o
rotation of pile
caps across the fault). The impact of potential fault displacement on the performance of
the sub- and super-structures crossing active faults remains to be a serious challenge to
the designers.
Review and Upgrade The reliable service of a lifeline depends on proper functioning
of all its important components at all times. It requires regular review and upgrade of
these facilities in normal time as well as prompt repair of failed components after an
earthquake.
Redundancy Robustness of the lifelines relies on the built-in redundancy. As a certain
line or node of a given lifeline is down, the availability of alternate lines or nodes is
critical for maintaining the lifeline function during the critical period of emergency
response and restoration.
Two 380-kV transmission lines bypassed the Adapazari substation, which suffered
circuit-breaker damage during the Kocaeli earthquake. These two lines were critical to
power restoration after the earthquake. This type of redundancy requires careful
planning and consideration of a variety of potential damage scenarios based on
knowledge of potential earthquakes and physical conditions of the lifelines facilities.
Lack of redundancy in the cooling system of nuclear reactors at the Fukushima Dai-Ichi
nuclear power plants is the fatal flaw. It eventually led to the meltdown of the fuel rods.
It not only stopped power generation at the plants, but also disrupted and/or curtailed
power generation at other nuclear plants due to the shadow that its incidents casted on
the safety of nuclear power in general.
Tsunami Inundation One side effect of tsunami inundation during Sumatra
earthquake was to cause pump failures due to salt water intrusion. This delayed the

38
restoration of those water and sewer lines whose function relies on the pumps. Thus,
sea water intrusion needs to be considered and guarded against or avoided in the
design of water and wastewater systems. Moreover, contamination of agricultural
irrigation and groundwater supply systems by sea water intrusion deserves careful
attention.
5. Conclusions
Earthquake, with its potent destructive power associated with the sudden release of
enormous energy stored over an indefinite period of seismic acquiescence, strikes terror in
peoples mind around the world. This chapter reviews six recent earthquakes, which
occurred in 2004 to 2011, including three tsunami-generating subduction earthquakes
(along the coastlines of the Indian Ocean, South and North Pacific Ocean) and three
crustal events (in the interior of the Asia Continent and two islands in the Caribbean Sea
and South Pacific Ocean). These earthquakes affected both developed countries (New
Zealand, Chile and Japan) and developing countries (Indonesia, China and Haiti). Due to
the economic and resources constraint, the ability of a country to direct its resources to
cope with natural hazards inherent to its location and setting varies. Thus, both self
development and international cooperation and assistance are important to enhance the
overall capacity of each country to protect its population and economy against the ravage
of earthquake and its associated secondary hazards. Summarized in the following are
some of the observations gleaned and lessons learned and/or re-learned from recent and
other relevant earthquakes:
5.1 General observations
Post-earthquake investigation has evolved with the technology advancement over the
years. The institutional support of this activity has been augmented many times by the
individual voluntary efforts of professionals of various disciplines. The collaboration of
research and practicing experts in these critical occasions promotes the development of
appropriate technology that improves the survival chance of people and sustaining
performance of structures for different regions of the world.
The occurrence of a series of earthquakes in a given region, be it a main shock followed
by aftershocks, or a foreshock followed by the main shock, or a sequential rupture of
adjacent faults, deserves special attention in seismic design and post-earthquake
response and reconstruction.
The concurrent and/or secondary effect of earthquake, such as tsunami, ground failure,
breach of man-made or landslide dam, release of hazardous material and fire hazard
should always be considered. An even more serious and long-lasting hazard, i.e., the
meltdown of fuel rods and radiation leak at nuclear power plant is now added to this
list.
The level of casualty and structure destruction, in general, reflect the engineering
practice such as seismic building code and its enforcement, design and construction
practice and workmanship in an area. Numerous and large-scale landslides further
contributed to the destruction of the 2008 Wenchuan earthquake. Unreinforced


39
masonry and non-ductile concrete buildings were the main structure type that suffered
heavy damage in the 2008 Wenchuan, 2010 Haiti, 2010 Chile and 2011 Christchurch
earthquakes.
The heavy tsunami casualty in the 2004 Sumatra event was due to the absence of a
warning system, while the higher than normal casualty in the 2011 Tohoku-Oki event
was due to the unusually high tsunami run-up in places and relatively short time
window available for evacuation of people from the low-lying areas.
5.2 Lessons learned
The building code is a living document, which requires periodical review and update to
reflect current understanding of the regional and local site seismicity, structural
earthquake performance and latest design and technology. Any important new
findings, such as the recognition of new or higher seismic hazard than that envisaged
by the current code, should be carefully reviewed and implemented in the revised code
as soon as practical.
The high cost of re-building collapsed, demolishing unsafe, and retrofitting
damaged buildings has caused some re-examination of the main objective of
building code, i.e., to protect life safety of occupants and minimizing their injury.
Beyond the life-safety objective, each jurisdiction may find its own optimal
economic balance between the cost of initial design and construction, versus the
cost and delay of re-construction after an earthquake. The current trend of
structural design based on performance requirement will provide some flexibility
to allow the adoption of a design higher than the minimum code standard based on
life safety.
Schools and emergency-response facilities, including fire and police stations, and
hospitals, especially those located on weak subsoils prone to liquefaction, tsunami-
vulnerable zone, or site subject to geologic hazard such as landslides, rock falls, etc.,
should be relocated or retrofitted to meet the requirement of maintaining their function
after an earthquake.
Critical facilities, with high occupancies, or containing hazardous materials or
susceptible to radiation leak, or serving important lifeline functions to society (e.g.,
power generation and distribution and telecommunication facilities), or have other
sensitive parameters, should be designed and maintained at higher seismic
standard.
Major lifelines that are either co-located and/or are interdependent with respect to
maintaining their proper function should have special performance requirements in
order to avoid multiple and/or cascading lifelines failures.
Emergency response and resilience plans for the region, and local jurisdictions should
be developed and exercised. These plans should focus on robustness, redundancy,
resourcefulness, and rapidity.
Mutual-aid plans among adjacent jurisdictions, both locally and internationally, should
be pre-arranged prior to the occurrence of earthquake so that the precious time required
for effective rescue operations in the initial golden period of 72-hours will not be
wasted.

40
6. Acknowledgements
The authors wish to acknowledge the assistance they received from the librarians and
administration staff of Klohn Crippen Berger for their effort in literature search and
manuscript preparation, and from colleagues of ASCE TCLEE. The critical review and
helpful suggestions by Professor A. Moustafa is greatly appreciated.
7. References
ASCE (2007). Sumatra-Andaman Islands Earthquake Tsunami of December 26, 2004,
Technical Council on Lifeline Earthquake Engineering, Monograph No. 30.
ASCE (2011). Special Issue on Earthquake Ground-Motion Selection and Modification for
Nonlinear Dynamic Analysis of Structures, J. Struct. Eng. 137 (3).
BBC (2005). http://news.bbc.co.uk/2/hi/science/nature/4247409.stm
BSSA (2007). The 2004 Sumatra-Andaman Earthquake and the Indian Ocean Tsunami, Bull.
Seism. Soc. Am. 97, No. 1A, Jan.
BSSA (2010). The 2008 Wenchuan, China Earthquake, Bull. Seism. Soc. Am. 100, No. 5B,
Nov.
Caltech (2011). http://www.tectonics.caltech.edu/
Comartin, C. et al. (2004). A Challenge to Earthquake Engineering Professionals, Earthquake
Spectra, Vol. 20, No. 4, pp. 1049-1056.
EEEV (2008). General Introduction of Engineering Damage of Wenchuan Ms 8.0 Earthquake,
Jour. of Earthq. Eng. and Eng. Vib., Vol. 28. Supplement, October.
EERI Special Earthquake Reports. (2005 Mar. to Jul. Sumatra 2004/2005 Eqs.; 2008 Oct.
Wenchuan 2008 Eq.; 2010 April and May Haiti 2010 Eq.; 2010 Jun. and Oct. Maule
2010 Eq.; 2010 Nov. Darfield 2010 Eq.; and 2011 May Christchurch 2011 Eq.). EERI
Newsletters.
EERI (2006). Summary Report on the Great Sumatra Earthquakes and Indian Ocean
Tsunamis of 26 December 2004 and 28 March 2005, W.D. Iwan Tech. Ed., Aug. and
Earthquake Spectra, Vol. 22, Special Issue 3, June.
EERI (2011). The March 11, 2011 Great East Japan (Tohoku) Earthquake and Tsunami:
Societal Dimensions, August.
FEMA (2008/2009). Guidelines for Design of Structures for Vertical Evacuation from
Tsunamis, P646, May 2008, and Vertical Evacuation from Tsunamis: A Guide for
Community Officials, P646A, June 2009.
Fierro, E. and Perry, C. (2010). Preliminary Reconnaissance Report 12 January 2010 Haiti
Earthquake, PEER.
GEER (2010/2011). Geo-engineering Reconnaissance of the 2010 Maule, Chile
Earthquake, Version 2, May 25, 2010. and Geotechnical Reconnaissance of the
2011 Christchurch, New Zealand Earthquake,
http://www.geerassociation.org/, 2011.
GEO-CAN (2010). Remote Sensing and the GEO-CAN Community: Lessons from Haiti and
Recommendations for the Future, Global Earth Observation Catastrophe
Assessment, Network Workshop Report, June.


41
Gillon, M. and Newton, C. (1991). Abutment Repairs at Matahina Dam, Proc. of the 17th
Congress on Large Dams, ICOLD, Vienna.
JSCE (2011). Preliminary East-Japan Earthquake Damage Investigation Report, Japanese
Society of Civil Engineers, April 11. http://committees.jsce.or.jp/report/node/40
(in Japanese).
Kawashima, K. (2011). Damage of Bridges Resulted from East-Japan Earthquake of March
11, 2011, http://peer.berkeley.edu/news/wp-content/uploads/2011/04/East-
Japan-Earthquake-21.pdf (in English).
Lo, R.C., Henderson, P.W. and Morrison, K.I. and Finn, W.D. Liam (1996). Seismic
Geotechnical Considerations An Overview, 11th WCEE Paper 1853.
Lo, R.C. and Wang, Y. (2008). Geoscience and Geotechnical Aspects of Recent Earthquakes,
4
th
Conf. on Geotechnical Earthquake Engineering and Soil Dynamics, Sacramento,
Calif. May.
Lo, R.C. (2009). May 12, 2008 Wenchuan Earthquake - Lessons Learned from Its Impact,
Emergency Response and Recovery, ASCE TCLEE Conf. on Lifeline Earthquake
Engineering in a Multihazard Environment, Jun 28-Jul 1, Oakland, Calif.
Moustafa, A. and Takewaki, I. (2010). Modeling Critical Ground-Motion Sequences for
Inelastic Structures, Advances in Structural Engineering, Vol. 13, No. 4, pp. 665-
679.
Moustafa, A. (2011). Damage-Based Design Earthquake Loads for Single-Degree-of-Freedom
Inelastic Structures, ASCE J. Struct. Eng. 137 (3) pp. 456-466.
NBCC (2005). National Building Code of Canada, National Research Council.
NIED (2011). National Research Institute for Earth Science and Disaster Prevention,
http://www.bosai.go.jp/e/index.html.
PEER (2011). Preliminary Briefing on Japans Tohoku Earthquake, Aftershocks, and
Tsunami of March 11, 2011.
Somerville, P. (2000). Seismic Hazard Evaluation, 12
th
WCEE, Paper 2833.
Stein, R.S. (2003, 2005). Earthquake Conversations, Scientific American, pp. 72-79, January,
2003 and pp. 82-89, Sp Vol. 15, 2005.
Takewaki, I. Murakami, S., Fujita, K., Yoshitomi, S., and Tsuji, M. (2011). The 2011 off the
Pacific coast of Tohoku earthquake and response of high-rise buildings under
long-period ground motions, Journal of Soil Dynamics and Earthquake
Engineering.
UN OCHA (2005). Relief Web Map Center, United Nation Office for the Coordination of
Humanitarian Affairs, Jan. 3.
USGS (2004-2011). Significant Earthquake and News
http://earthquake.usgs.gov/earthquakes/eqinthenews/
USGS/EERI (2010). The Mw 7.0 Haiti Earthquake of January 12, 2010, USGS/EERI Advance
Reconnaissance Team Report, USGS Open-File Report 2010-1048.
USGS (2010). Documentation for Initial Seismic Hazard Maps for Haiti, Open File Report
1067.
Ventura, C., Elwood, K., and Huffman, S. (2008). Preliminary Report Wenchuan
Earthquake of May 12, 2008 - Reconnaissance Trips.

42
Yeats, R.S., Sieh, K. and Allen, C. R. (1997). The Geology of Earthquakes, Oxford Univ.
Press.
Yin, Y., Zheng, W., Li, X., Sun, P. and Li, B. (2011). Catastrophic Landslides Associated with
the M8.0 Wenchuan Earthquake, Bull. Eng. Geol. Environ. 70:15-32.
2
Lateral In-Situ Stress Measurements to
Diagnose Liquefaction
Richard L. Handy
Iowa State University
U.S.A.
1. Introduction
Death by earthquake usually involves toppling and crumbling of masonry walls, falling
roofs, and pancaking floors in multistory buildings. Residents of underdeveloped countries
are particularly vulnerable because structures only rarely are designed to resist earthquakes.
The safety of masonry construction could be improved with buttressing or cellular design,
but those procedures are little known and seldom used, suggesting that effective charity
should include education and innovation.
Even earthquake-resistant structures remain vulnerable to a hidden danger, that posed by
liquefaction. Most susceptible are saturated loose sands and silts where shaking movements
can cause individual soil grains to become dislodged and temporarily suspended in water,
turning the soil into a dense liquid analogous to quicksand. The process occurs at depth due
to increased horizontal shearing stresses, so gentle slopes can turn into landslides, buried
tanks can float up out of the ground, and buildings can sink or tip over. Fortunately the
destruction is relatively slow compared with walls crumbling and roofs falling, and can
allow enough time for people to escape with their lives. A famous example in Niigata, Japan,
where as shown in Fig. 1 multistory apartment buildings remained intact and in one

Fig. 1. Apartment buildings at Kawagishi-Cho, Niigata, Japan, tipped after the foundation
soil liquefied during the 1964 earthquake.
2011 Richard L. Handy

44
instance tipped far enough that residents could crawl out of windows and walk to safety
down the exterior wall.
Perhaps the most serious situation is where a potentially liquefiable soil supports a dam,
where a failure would be sudden and could take many human lives.
Liquefaction is soil specific, so building sites in earthquake areas can be tested to determine
their susceptibility. However, the tests are largely empirical and based on case histories. This
chapter examines the role of a more fundamental soil property, lateral in-situ stress, on soil
collapse and liquefaction, and emphasizes the use of a relatively new tool to measure that stress.
2. The nature of liquefaction
Since both lateral and vertical stresses in soils tend to increase with depth, they are
conveniently expressed in terms of a ratio of effective stresses, K
o
. During liquefaction, soil
grains are temporarily suspended in water: Effective or intergranular stresses are zero, and
there is no K
o
. Then the grains settle and displace water so it seeps upward, often taking
sand particles with it. The reconstituted soil density therefore is higher than it was prior to
liquefaction, so liquefaction can be a one-time event, mainly limited to geologically youthful
sandy deposits such as deltas, alluvial fans and floodplains, and to loose, man-made sandy
fill soil that is dumped without being compacted.
Liquefaction probably occurs with all major earthquakes, but in severely damaged areas the
evidence lays buried underneath the devastation. Prehistoric earthquakes can be dated from
sedimentary sequences that contain sand carried upward by the displaced water and
deposited as shallow cones at an earlier ground surface.
Stress ratios existing in soils prior to and after liquefaction have not yet been fully explored,
the major restraint being the difficulty of obtaining reliable in-situ stress data.
3. Current methods for evaluating the liquefaction potential of a soil
The pioneering work of Seed and Idriss (1971) developed criteria for identifying potentially
liquefiable soils from Standard Penetration Test (SPT) data. The method has since been
improved through the use of empirical corrections to the data, and is supplemented by cone
penetration tests (CPT) and measurements of shear wave velocity (Youd et al. 2001).
Schmertmann (1971) suggested that the SPT is essentially a side friction test, implying a
dependence of blow counts on horizontal stress existing and developed in the soil. Kelly and
Lutenegger (2004) show close relationships between blow counts and side friction measured
by torquing the SPT sampler, with the exact relationship depending on the type of soil. The
lateral stress ratio K
o
appears in several formulas for cone resistance aced on cavity expansion
theory (Yu and Mitchell 1998). Thus even though the lateral in-situ stress changes in the
neighborhood of a test instrument, it still may play an important role in a liquefaction process.
4. Measuring K
o

Vertical effective stress in soil can be obtained from depth, soil density, and elevation of the
groundwater table, but lateral stress must be measured. Empirical estimates of the lateral
stress ratio K
o
based on depth, density, friction angle, and overconsolidation ratio, do not
address effects or likelihood of liquefaction.

Lateral In-Situ Stress Measurements to Diagnose Liquefaction

45
One approach for measuring lateral in-situ stress in soil is to bore a hole and insert a
measuring device without disturbing the soil. This is the concept of the self-boring
pressuremeters. However, tests are time-consuming, and some plucking or shoving aside of
individual soil particles during drilling is unavoidable.
Other devices such as the Cone Penetration Test (CPT) and Dilatometer laterally displace
the soil to the extent of creating a lateral bearing capacity failure. Empirical relationships can
be developed based on pressure chamber testing, but the changes in lateral stress caused by
disturbance involve an unknown combination of soil responses including elastic
compression, consolidation and shearing. These occur simultaneously and create changes in
lateral stress that are larger and more variable than the stresses that are intended to be
measured. A third approach is to leave a flat pressure cell in place long enough for stresses
to equilibrate, but that ignores the permanence of soil arching action.
A different approach is used with the K
o
Stepped Blade.
4.1 The K
o
stepped blade
The methodology used with the K
o
Stepped Blade was adapted from analytical chemistry, to
intentionally introduce different levels of disturbance and extrapolate the results to a
condition of zero disturbance. This is accomplished with four blade steps ranging in
thickness from 3.0 to 7.5 mm, Fig. 2.

Fig. 2. Schematic showing the method used to interpret Stepped Blade data. A linear e-log p
relationship indicates consolidating behavior. Reprinted with permission from Handy and
Spangler (2007). The McGraw-Hill Companies, New York.

46
The blade is advanced incrementally into the bottom of a boring so each measurement is
made on a progressively increasing step thickness at the same depth. While the test is rapid,
enabling as many as 10 determinations in an hour, the blade cannot be pushed continuously,
but must be withdrawn and the boring extended deeper for each new test series. A blank
space at the top of the Blade allows four multi-stress measurements to be made at 127 mm (5
in.) depth increments in each series. Measurements are directional owing to the flat shape of
the Blade.
4.2 Interpretation of K
o
stepped blade data
Extensive measurements in the field and in laboratory simulations show a consistent
relationship between step thickness and the logarithm of pressure, illustrated by the graph
in Fig. 2 (Handy et al. 1982; White and Lutenegger 2008). Analogy to the familiar e-log p
linear relationship suggests that soil penetrated by the Blade must be laterally consolidating.
Data that do not meet this consolidation criterion are not included in the interpretations.
Stepped Blade steps are thin in order to minimize the tendency to create a lateral bearing
capacity failure. The two most common causes for data rejection are (1) elastic response that
increases the pressure on the thinnest step prior to a breakdown of the soil structure, and (2) a
lateral bearing capacity failure that relieves stress on the thickest step(s). The three behaviors,
elastic response, consolidation, and lateral bearing capacity failure, are illustrated in Fig. 3.
Elastic and bearing capacity responses must not be used in the extrapolations, and are readily
recognized from pressure decreases instead of increases on successive blade steps.

Fig. 3. Stress diagrams inferred from pressure measurements on different thickness steps,
showing the critical role of consolidation and reasons for rejecting specific data points.
An elastic concentration of stress most commonly occurs with the thin step in stiff soils, and
a relief of pressure from a bearing capacity failure most commonly occurs with the thick
step in soft soils. Usable data sequences therefore may have two, three or four data points. A
discriminating evaluation of each data point is required before using a regression analysis to


47
extrapolate the lateral in-situ stress, or the analysis will yield incorrect and erratic interpretations
including a low value for R
2
.
A linear regression analysis of the logarithm of measured pressures versus blade step
thickness yields a slope and intercept. The intercept is the lateral stress that existed prior to
insertion of the blade, and the intercept can be examined for consistency. A test performed
close to a hole or hard inclusion in the soil will change the slope and can be another reason
to question some data. However, since such a condition cannot be confirmed, questionable
data points are shown on graphs but are labeled with question marks.
The distribution of soil pressure across the face of the Blade was investigated by pulling
narrow friction strips (Handy et al. 1982) and emphasized the importance of focusing
measurements in the central area of each step.
4.3 The K
o
stepped blade in relation to other in-situ tests
In Fig. 4, soil response mechanisms interpreted from Stepped Blade data are applied to other
in-situ displacement soil tests. For example, it appears likely that the width/thickness ratio
of the Dilatometer causes a lateral bearing capacity failure, so empirical correlations are
used to estimate both stress and modulus from expansion of a metal diaphragm. A more
direct measure of modulus might be obtained from elastic response of soil to penetration by
the thinnest step of the Blade, a possibility that has been examined but not yet fully
investigated.

Fig. 4. Width: thickness ratios for various in-situ displacement test devices and (at top)
relationships to soil behavior determined from Stepped Blade tests. Consolidating behavior
is required to interprete lateral in-situ stress from Blade tests. The w/t ratio for the cone is
1.0. Modified from Handy et al. (1990)

48
4.4 Pneumatic pressure cells
The several requirements for pressure cells used in the Blade are (1) that they be as thin as
the thinnest Blade step, (2) that they measure pressure on a discrete centralized area, (3) that
they be durable and reliable, and (4) that when damaged they can be easily replaced or
repaired in the field. A back-pressured pneumatic cell was developed to meet these criteria
and is shown schematically in Fig. 5. An important advantage of the pneumatic cells is that
measurements involve gas pressures and do not require calibrations.
The flat surface of each cell is covered with Teflon over an elastomeric membrane to
minimize friction with the soil, and to create an internal seal between the two cell chambers.
The chambers are slowly pressurized with gas pressure while maintaining a small
differential between the two chambers of the cell. At the instant when the gas pressure
equals the soil pressure anywhere around the threshold, the membrane is lifted sufficiently
to cause an internal leak so pressures in the two chambers equalize. At that time pressure is
read and recorded, and pressures to both chambers are simultaneously vented to prevent
bulging of the membranes and ready the cell for the next pressure reading. Selector valves
are turned to direct the gas pressures to the next Blade step, and the test procedure is
repeated by simply closing the two relief valves. If desired, a reading can be repeated for
additional confirmation.

Fig. 5. Part of the secret of the K
o
Stepped Blade is the back-pressured pneumatic cells. At
the instant when gas pressure equals soil pressure, the differential pressure goes to zero so
the total pressure is read from the appropriate pressure gauge. After Handy and Eichner,
U.S. Pat. No. 4,662,213, 1987.
Experiments have shown that the Teflon cover is much more durable than thin sheet metal
but nevertheless can be scraped or punctured. If a leak occurs, positive gas pressure
prevents water and soil from entering the cell, and since cells plug in they can be quickly
replaced in the field. Experiments with electrical strain-gauged metal membranes showed
that they were not reliable under the severe test conditions imposed in the field.


49
A hollow rib attached to the back of the Blade carries the multiple gas lines and reinforces
against bending.
5. Test sites
Several test sites were selected in potentially liquefiable soils:
1. The Jackson Lake Dam was built to increase the water level in a natural glacial lake
located on the Snake River in northwestern Wyoming, U.S.A. The log-crib dam was
constructed in 1907 and failed three years later. It was replaced between 1911 and 1916
by a concrete and earth dam.
In 1964 the destructive nature of liquefaction was emphasized both by the Niigata
Earthquake in Japan and the Alaska Earthquake, and in 1976 the Jackson Lake Dam
foundation soil was determined to be a potentially liquefiable on the basis of Standard
Because a failure could be catastrophic, in 1986-89 the dam was removed and the gravelly
sand foundation soil was densified by Deep Dynamic Compaction, shown in Fig. 6. Stepped
Blade Tests were conducted to measure lateral pressures in the sand after the dam was
removed, both before and after the site was compacted.

Fig. 6. Deep Dynamic Compaction to prepare a gravelly sand foundation soil for the
reconstructed Jackson Lake Dam, Wyoming. Reprinted with permission from Handy and
Spangler (2007), The McGraw-Hill Companies, New York.
2. Considerable areas of poorly compacted sand fill exist in many bay areas including San
Francisco, California, where there is a history of liquefaction dating from the 1906
earthquake. A potentially liquefiable sand was identified at the San Francisco Naval
Shipyard, so experiments were conducted by the U.S. DOT-FHWA to determine the
effectiveness of driving displacement piles to densify the sand. K
o
Stepped Blade tests

50
were conducted to measure lateral pressures in the sand both before and after pile
driving.
3. A relatively recent development called Rammed Aggregate Piers, or the Geopier
method for soil reinforcement, is becoming increasingly popular for improving
foundation soils. As shown in Fig. 7, the rammer head is shaped to push aggregate
outward into the soil and thereby increase lateral stress. K
o
Stepped Blade tests were
conducted to measure those stresses at various depths and distances from the piers.
4. K
o
Stepped Blade tests conducted in collapsible loess soil to determine relationships
between collapsibility and lateral in-situ stress, and suggested a history of liquefaction
in a basal, saturated zone. This should affect slope stability, and massive landslides in
the thick loess in China have taken many human lives. The trigger usually is excessive
rains, but also can involve earthquakes.

Fig. 7. The Geopier hydraulically actuated rammer is shaped to direct part of the ramming
energy outward and increase lateral stress in the surrounding soil.
5.1 Test results from the Jackson Lake Dam site, Wyoming
The Deep Dynamic Compaction shown in Fig. 6 involved dropping an approximate 100
tonne weight about eight times at each location on a grid pattern. This left a pattern of
surface depressions that were 2 to 3 m deep. The results from K
o
Stepped Blade tests
conducted before dynamic compaction are shown in the left diagram in Fig. 8, and
afterwards in the right diagram in the same figure. The site was difficult because of angular
gravel-size rock fragments in the soil.
At the left in Fig. 8, below about 5.5 m (18 ft) depth Stepped Blade tests gave K
o
in the
neighborhood of 0.5 without giving consideration to the weight imposed by the dam. The
Jky relationship indicates that this corresponds to an angle of internal friction of 30
o
, which
is reasonable for normally consolidated sand.
In the same figure, the weight of the dam compressed the soil and increased lateral stress
down to a depth of about 4m (13 ft), indicating that normal consolidation penetrated to that
depth. Below that depth there is a transition to a K
o
of about 0.3, indicating


51
underconsolidation and a potential for collapse that can be attributed to the weight of the
dam.
At the right in Fig. 8, Deep Dynamic Compaction is shown to cause a substantial increase in
lateral stress such that the average K
o
is 1.0, perhaps from seepage forces created when the
weight of the overlying soil was supported by pore water pressure (Handy 2012). The
uneven distribution of lateral stress after compaction may be symptomatic of localized
shearing that concentrated stress in one area while reducing it in another.

Fig. 8. Lateral stresses measured in foundation sand at the Jackson Lake dam site, (left)
before and (right) after compaction. At the left, the soil was normally consolidated until after
the embankment was built, when the weight of the dam compacted the upper part but left
the lower soil underconsolidated and potentially liquefiable. At the right, compaction
increased the average K
o
to 1.0, indicating a history of induced liquefaction, and K
o
= 0.5
after the dam was replaced. From Handy (2012), modified from Handy et al. (1990).
In this connection the narrow range of lateral stresses measured before compaction, in a
range as low as 8 kPa or 10%, compared with 50% after compaction, indicates that the
variability is real and should not be assigned to random experimental error. Large variations
in lateral stress also have been measured within a compacted earth embankment (C.
Anderson, personal communication).
After compaction and restoration of the weight of the dam, K
o
is about 0.5, indicating that
compaction was successful in creating a quasi-normally consolidated condition, and the
dam should be safe from future collapse and liquefaction of the foundation soil.
5.2 Test results from the San Francisco Naval Shipyard, California
The San Francisco Naval Shipyard was first established in 1870 at Hunters Point in the
southeast part of the city. Starting in the early 1900s the area was expanded by dumping fill

52
sand, making this an appropriate site to measure the effectiveness of driving piles to
compact the sand. The discovery of contamination with nuclear and other wastes led to the
shipyard being closed in 1994.
Lateral stresses measured prior to pile driving were quite consistent, shown by data points
and the shaded area to the left in Fig. 9. A trend line drawn through the data and extended
upward to the y axis indicates a prior surcharge equal to the weight of about 5 m (16 ft) of
the same soil.

Fig. 9. Tests in loose sand at the San Francisco Naval Shipyard the slope of the lateral stress
line gives K
o
= 0.29 prior to pile driving, indicating a potential for liquefaction. As in the case
of DDC, pile driving increased the average K
o
to 1.0, indicating temporary liquefaction
down to the bottoms of the piles. Data are from Handy et al. (1990)
The K
o
value obtained from the slope of the trend line is 0.29, compared with a value of 0.5
that can be expected if the soil is normally consolidated. (A K
o
calculated without taking
into account the surcharge is meaningless as it becomes infinite at the ground surface.) The
measurements of lateral stress confirm that the soil is underconsolidated and potentially
liquefiable, and a suitable candidate for remediation.
Lateral stresses measured after pile driving were more erratic, but nevertheless cluster
around K
o
= 1.0 down to the bottom level of the piles. Below that depth, lateral stress
remained unaffected.
The K
o
value of 1.0 may be from seepage forces during temporary liquefaction that occurred
during pile driving. Liquefaction would be beneficial for enabling a wider distribution of
lateral stress, but also would tend to relieve passive pressures that otherwise would act to
increase skin friction. A liquid condition cannot sustain a passive ratio any more than it can
sustain an active ratio. The lack of passive pressure is consistent with a group effect of 1.0.


53
5.3 Test results after Geopier soil reinforcement
The Geopier method for soil reinforcement involves boring holes that typically are 760 mm
(30 in.) in diameter, and incrementally filling the holes with layers of rammed coarse
aggregate. As previously noted, the hydraulic rammer head, Fig. 7, is shaped to direct a
substantial portion of the ramming effort laterally outward into the soil.
5.3.1 K
o
no longer constant
Fig. 10 shows lateral stress measurements at different depths and distances from rammed piers
at the site of a stadium in Memphis, Tennessee. The soil was saturated loessial silt. Because of
the variability of the pier diameter caused by incremental ramming, data are plotted versus
distance from pier centers. K
o
is not constant with depth, which is quite a different distribution
of lateral stress from those in the preceding examples. Instead of K
o
it is the lateral stress at
particular radial distances that is constant with depth. This probably relates to the uniform
distribution of ramming energy throughout the length of a pier, and may be influenced by
compaction that proceeds from the bottom up instead of from the top down.
5.3.2 Another evidence for temporary liquefaction
In Fig. 10 the pier contact stress is obtained by extrapolation of data using plastic theory. The
open arrow pointing to the right in the figure indicates an undiminished transfer of radial
compaction stress outward to a radial distance of almost 2 m ((6 ft). This radial distribution
of stress is possible only if the soil temporarily behaved as a liquid.

Fig. 10. Radial stresses measured in saturated loess soil near Rammed Aggregate Piers in
Memphis, Tennessee. The data suggest a succession of behavioral modes that includes
temporary liquefaction. Modified from Handy and White (2006).

54
5.3.3 Where does the water go during Geopier ramming?
Rammed Aggregate Piers are constructed incrementally by compacting successive layers of
aggregate such that each layer has a final compacted thickness of about 0.3 m (1 ft). During
ramming, the data in Fig. 10 indicates that within the liquefied zone the pore pressure
exceeds 100 kPa (14.5 psi, 1 Tsf). Beyond a radial distance of 1.9 m, which is outside of the
liquefied zone, the decrease in radial stress is linear with distance. Elastic theory indicates
that this can occur only if there a lack of restraint from circumferential tension, which can be
explained by radial tension cracking. Radial cracks have been observed emanating outward
from piers, and can extend down to a depth where the fluid pressure exceeds the existing
lateral stress. In a normally consolidated soil below the groundwater table that can be as
deep as 20 m (60 ft).
Open cracks are an open invitation for injection by a soil-water mixture, and soil particles
carried in with the water can prop the cracks open and create a drainage gallery. That is the
concept used in the Hydrofrak process to enhance oil and gas production from wells.
After the fluid pressure dissipates, the injected cracks should create a ring of compression
that preserves stress in the elastic zone.
Cavity expansion theory allows an estimate of the post-drainage friction angle from the
slope of the line in the drained zone (Baguelin et al. (1978). This gives a friction angle of
about 40
o
, which represents a substantial increase over friction angles commonly measured
in loess (Handy and White 2006).
It will be noted in Fig. 10 that identical behaviors were observed at all three test depths.
5.3.4 How lateral stress affects preconsolidation pressure
Fig. 11 (a) represents a normally consolidated where intergranular friction acts to support
the grains and reduce the lateral pressure. If a high lateral stress is applied, it can reverse the
directions of the friction arrows, as shown in Fig. 11(b). Then when a vertical load is applied,
before the grains can slip and the soil can consolidate, the load must be sufficient to again
reverse the friction arrows, Fig. 11(c). This concept is further illustrated by the succession of
Mohr circles in the lower part of the figure. It means that imposing a high lateral stress can
cause a substantial increase in the preconsolidation pressure so within specific limits the soil
will tend to compress elastically instead of by consolidating. This can reduce total settlement
of a structure by an order of magnitude. This has been confirmed by load tests and is an
important feature in Geopier design.
It should be noted that in a conventional laboratory consolidation test the lateral stress is not
controlled, and in the conventional triaxial it is difficult to apply a lateral stress that exceeds
the vertical stress, so in both cases the stress ratio remains in a normally consolidating mode.
What if lateral stress later is released, for example by trenching? During compaction the
soil cohesion is greatly reduced by remolding, and then is recovered, a source of setup
factor for driven piles. Schmertmann (1991) suggests that even the strength of sand can
be doubled by aging, and a time-related recovery of strength is characteristic of
liquefaction (Youd et al. 1996).


55

Fig. 11. Model illustrating how a high imposed lateral stress can reverse friction directions at
grain contacts and increase the preconsolidation pressure without actually consolidating the
soil. Modified from Handy (2001).
5.4 Test results in collapsible loess
Loess is wind-deposited silt, and comes from lss, which is German for loose. Thick loess
deposits that are close to a river floodplain source are so porous that it does not retain a
groundwater table. Only lower parts of deposits are saturated, where the groundwater table
is perched on an underlying less permeable stratum.
The strength and ability to stand in high steep banks such as shown in Fig. 12 is attributed to
negative pore water pressure that is enhanced by a small content of active clay minerals, The
strength therefore is sensitive to the moisture content, and sudden collapse can occur upon
wetting. A concentration of surface water therefore can contribute to rapid and intense
gullying and can crate vertical chimneys. Severe conditions can result in an almost karst-
like terrain (Derbyshire et al. 1994).
5.4.1 Loess K
o
before collapse
Some of the early tests of the Blade were in collapsible loess, and gave K
o
values that were
so low, 0.1 to 0.2, that they were met with considerable skepticism. However, the data were
repeated and found to be consistent.

56
Low lateral stress probably results from dust settling into a nest of grasses and then being
washed down by rains until grains come into contact, where they are held in place by
negative pore water pressure or suction. As the loess accumulated, vegetation migrated
upward and left root channels that although often mineralized still may enhance drainage.
Densities as low as 100 kg/m
3
(64 lb/ft
3
) have been measured in loess actively being
deposited in Alaska.
The loess deposit investigated in this study is thinner and farther from the source than that
depicted in Fig. 12, in a relatively flat area in the western part of Omaha, Nebraska. It has a
weathered soil profile and probably has been subjected to modification of the lateral stress
through compaction. K
o
values in Fig. 13 indicate that except for near-surface modifications
the loess is collapsible down to a depth of about 5 m (16 ft).

Fig. 12. Western Iowa loess deposits are thick close to the Missouri River floodplain, where
glacial outwash was the major source of wind-blown silt. The author in his younger days is
sampling about half-way down the cut slope.


57
5.4.2 Loess K
o
after collapse
In Fig. 13 in the lowest 2 m (6 ft) of the deposit, four independent data points give K
o
= 1.0.
The reproducibility confirms the precision of the measurements, and the value may be an
indication of accuracy since it characterizes a liquid condition.
The highest K
o
in Fig. 13, at a depth of 1 m (3 ft) translates into a vertical pressure of about
100 kPa (14 psi), which can be explained by ordinary tire pressures.

Fig. 13. K
o
Stepped Blade data from Nebraska loess about 17 km (11 miles) from the
Missouri River floodplain source. From Handy (1994).
5.4.3 K
o
and partial collapse
Between depths of about 4 and 8 m (13 and 26 ft) in Fig. 13 there is a transition zone from
the lowest value of K
o
to K
o
= 1.0. Ordinarily values that are higher than those resulting from
normal consolidation are attributed to overconsolidation, but that appears to be unlikely at
this site since it would require that loess deposition would be interrupted by a major erosion
cycle.

58
The explanation may relate to a capillary fringe that includes saturation in the lower part,
where pore water pressures are negative and diminish to zero at the groundwater table.
The implication is that the soil may be only partly collapsed, increasing lateral stress to a
value that is higher than normally would occur but not high enough to create a K
o
= 1.0
liquid condition. To the authors knowledge this may be the first indication of a partial
collapse.
5.4.4 Geography of collapsible loess
Loess deposits become thinner and finer grained with increasing distance from a source.
Larger near-source thicknesses such as shown in Fig. 12 can be attributed to randomly
variable wind directions that pile up more soil when winds are blowing nearly parallel to a
source (Handy 1976).
A commonly used predictor for loess collapsibility is the saturation moisture content, which
can be calculated from the soil unit weight and mineral specific gravity, in relation to the
liquid limit. Loess containing less than 16 percent 2 m clay is very likely to be collapsible,
and one containing more than about 32 percent clay is not (Handy 1973).
5.4.5 Loess collapse and slope stability
A relationship between liquefaction of loess and slope stability is instantly appreciated if
one walks along the toe of an active landslide involving saturated basal loess. The soil
squeezing out of the basal zone only appears to be firm enough to walk on. The most likely
condition for sliding is after extensive rains that raise the groundwater elevation to a new
level in underconsolidated loess.
6. Conflicting observations
A K
o
that is less than about 0.5 for sand and 0.55 for silt indicates an underconsolidated
condition that is susceptible to collapse and liquefaction. This can be triggered by an
earthquake or other ground vibrations such as those induced by pile driving or compaction,
or in the case of loess soils, by saturation.
However, a low K
o
is not required for lateral compaction used in the Geopier process.
Whereas compaction from the top down can result in a constant K
o
with depth, Geopier
compaction results in a variable K
o
and constant lateral stress with depth, with the
developed lateral stress depending on radial distance from the piers.
What may explain this difference? One possibility is that the pier ramming process involves
compaction energy that is uniformly distributed with depth, resulting in a different stress
distribution than when compaction is from the ground surface down.
A Rammed Aggregate Pier is surrounded by a zone of highly compacted soil that should
help to contain the ramming energy for the next lift. That may not be the case for pile
driving, although this possibility is yet to be investigated. Pile driving from the top down
may generates a peripheral zone of sheared and dilated soil that is less able to confine the
pressure created by driving.


59
With normal consolidation, K
o
normally is in a range 0.5-0.6, whereas the K
o
acquired after
drainage of a liquefied soil is consistently 1.0. Why the difference? In both cases soil grains
are settling out of water that is under pressure.
One difference is that normal consolidation that occurs as vertical stress gradually is
increased activated intergranular friction as shown in Fig. 1(a), where it acts to partly
support the individual soil rains and reduce lateral stress. During liquefaction that friction
does not exist, and during drainage there are large seepage forces and a transfer from fluid
pressure to effective stress. The process would be difficult to simulate but might be
examined with laboratory tests.
7. Conclusions with regard to K
o
, collapse and liquefaction
A low K
o
is an indicator of a potential for collapse and liquefaction, provided that there is a
sudden change in conditions that is sufficient to break down the soil structure.
The defining value for collapse and liquefaction is a K
o
less than about 0.5-0.55. The angle of
sliding friction of quartz is about 25
o
, which according to the Jky relationship should result
in K
o
= 0.58 without a contribution from positive or negative dilatancy.
During liquefaction as water is ejected from soil its density increases, tending to make
liquefaction a one-time event that is mainly limited to geologically young or poorly
compacted man-made sand or silt deposits.
Lateral in-situ stress is a manuscript for stress history, as certain values are indicative of
normal consolidation, overconsolidation, and underconsolidation. For example, the
amount of removal of overburden often can be determined from a linear relationship
between lateral stress and depth. In a preciously collapsed soil K
o
is consistently found to
equal 1.0.
Artificial compaction results in a much higher variability of lateral stress than does
consolidation involving natural processes.
8. Conclusions with regard to the instrumentation
Lateral stress has been called the Holy Grail of soil mechanics (M. Jamiolkowski, personal
communication). It plays a major role in shallow and deep foundation bearing capacity, pile
skin friction, slope stability, and pressures on retaining walls.
The stated goal of the research sponsor, the U.S. DOT-FHWA, was to support
development of an instrument that would rapidly determine lateral in situ soil stress
within an arbitrary margin of 1 psi (7 kPa). In Fig. 11 the several data points for K
o
= 1.0
occupy a much narrower range, within about 0.4%. The narrow range of data in Fig. 8
fall in a range 6 kPa, which still is within the target range even though it includes a soil
variable.
The reproducibility and speed of the test make it useful not only to determine average
values, but also to define standard deviations that may be equally meaningful, for example
by diagnosing man-made compaction as opposed to natural consolidation.

60
The possible use of lateral in-situ stress to indicate a potential for collapse or liquefaction
was not anticipated during development of the test. However, it is not unusual for
a new approach to open new vistas and paths that can unexpectedly lead in diverse
directions.
The easy path to lateral stress is through computer modeling, and before that there was
mathematical modeling. However, every model should be tested against the real world
because otherwise it may be accepted but nevertheless flawed. Measurements inspired
Terzaghi to develop consolidation theory; it was not the other way around. Measurements
made in the 20
th
Century of pressures on retaining walls revealed inadequacies of 19
th
and
18
th
Century Rankine and Coulomb theories (Terzaghi 1943; Handy 1985). Measurements
of soil loads on buried pipes made over 100 years ago led to a theoretical model that
requires empirical corrections and only recently was shown to incorporate a simple and
obvious error (Handy 2004).
Lateral stress existing in soil at the toe of an embankment or earth dams should be an index
to their stability against sliding, by allowing a determination of an existing factor of safety.
Lateral stress existing in soil next to a shallow foundation can test and evaluate the safety
against a foundation failure, and can be helpful to determine if an additional load may be
allowable. Measurements of lateral pressures in soils behind a retaining wall target the
existing factor of safety and how it may change under varying conditions. With so many
potential uses the K
o
Stepped Blade may be underutilized.
9. Acknowledgements
Administrators who see a need and are guided more by their dreams than by their
worries tend to remain anonymous and seldom get their due. The following
organizations are listed chronologically according to their participation in the research
and development of the K
o
Stepped Blade: The U.S. DOT-FHWA; Soil Systems, Inc.,
Marietta, GA; Spangler Geotechnical Laboratory of Iowa State University, Ames, IA;
Eichner Engineering, Ames, IA; Handy Geotechnical Instruments, Inc., Madrid, IA;
GeoSystems, Inc., Lenexa, KS., Geopier Foundation Co., Inc., Moorsville, NC, and the
SBIR program of the U.S. Army Corps of Engineers WES, Vicksburg, MS. Many graduate
students worked with, tested and modified the Blade and made it work. It was a group
effort.
10. References
Baguelin, F., Jzquel, J. F., and Shields, D. H. (1978). The Pressuremeter and Foundation
Engineering. Trans Tech Publications, Clasuthal, Germany.
Derbyshire, E., Meng, X., Wang, J., Zhou, Z., and Li, B. Collapsible Loess on the Loess
Plateau of China. In Proc. Of the NATO Advanced Research Workshop on Genesis
and Properties of Collapsible Soils. Kluwer Academic Publishers, Netherlands, pp.
267-293.
Handy, R. L. (1973). Collapsible Loess in Iowa. Proceedings of the Soil Science Society of America
Vol. 37, No. 2, pp. 281-284.


61
Handy, R. L. (1976). Loess Distribution by Variable Winds. Geological Society of America
Bulletin. Vol. 87, pp. 915-927.
Handy, R. L., Remmes, B., Moldt, S, and Lutenegger, A. J., and Trott, G. (1982). In-situ
Stress Determination with the K
o
Stepped Blade. Journal of the Geotechnical
Engineering Division of the American Society of Civil Engineers. Vol. 108, No. GT11,
pp. 1405-1422.
Handy, R. L. (1985). The Arch in Soil Arching. Journal of the Geotechnical Engineering Division
of the American Society of Civil Engineers. Vol. 111, GT 3, pp. 302-318.
Handy, R. L., Mings, C., Retz, D., and Eichner, D. (1990). Field Experience with the
K
o
Stepped Blade. Transportation Research Record. Issue 1278, pp. 125-
134.
Handy, R. L. (1994). A Stress Path Model for Collapsible Loess. In Proc. of the NATO Advanced
Research Workshop on Genesis and Properties of Collapsible Soils. Kluwer Academic
Publishers, Netherlands, pp. 33-49.
Handy, R. L. (2001). Does Lateral Stress really Control Settlement? American Society of Civil
Engineers Jour. of Geotechnical and Geoenvironmental Engineering. Vol. 1127, No.GT7,
pp. 623-626.32,
Handy, R. L. (2004). Anatomy of an Error. American Society of Civil Engineers Jour. of
Geotechnical and Geoenvironmental Engineering. Vol. 130, No. 7, pp. 768-
771.
Handy, R. L., and White, D. J. (2006). Stress Zones near Displacement Piers: I. Plastic and
Liquefied Behavior. American Society of Civil Engineers Jour. of Geotechnical and
Geoenvironmental Engineering. Vol. 132, No. GT1, pp. 63-71.
Handy, R. L., and Spangler, M. G. (2007). Geotechnical Engineering. The McGraw-Hill
Companies. ISBN-13:978-0-07-48120-5 and 10-0-07-148120-6. New York.
Handy, R. L. (2012, in review). Ko and Liquefaction at Jackson Lake Dam, Wyoming.
American Society of Civil Engineers Jour. of Geotechnical and Geonenvironmental
Engineering.
Kelley, S. P., and Lutenegger, A. J. (2004). Unit Skin Friction from the Standard Penetration
Test Supplemented with the Measurement of Torque. American Society of Civil
Engineers Jour. of Geotechnical and Geoenvironmental Engineering. Vol. 130, No. 5, pp.
540-543.
Schmertmann, J. (1979). Statics of the SPT. American Society of Civil Engineers Jour. of the
Geotechnical Engineering Divisioin. Vol. 105, No. GT5, pp. 655-670.
Schmertmann, J. (1991). The Mechanical Aging of Soils. American Society of Civil Engineers
Jour. of Geotechnical and Geoenvironmental Engineering. Vol. 117, No. 9, pp. 1288-1330.
Terzaghi, K. (1943). Theoretical Soil Mechanics. John Wiley & Sons, New York.
White, D. J., and Lutenegger, A. J., eds. (2008). The Papers of R. L. Handy. CTRE, Iowa State
University. ISBN 978-0-9820144-0-0, Ames, Iowa, U.S.A.
Youd, T. L., Idriss, I. M., Andrus, R. D., Arango, I., Castro, G., Christian, J. T., Dobry, R.,
Finn, W. D. L., Harder, L. F., Jr., Hynes, M. E., Ishihara, K., Koester, J. P., Liao, S. S.
C., Marcuson, W. F., III, Martin, G. R., Mithcell, J. K., Moriwaki, Y., Power, M. S.,
Robertson, P. K., Seed, R. B., and Stokoe, K. H., II (2001). Liquefaction Resistance
of Soils: Summary report from the 1996 NCEER and 1998 NCEER/NSF Workshops

62
on Evaluation of Liquefaction Resistance of Soils. Journal of the Geotechnical
Engineering Division of the American Society of Civil Engineers. Vol. 127, No. GT 10,
pp. 817-833.
Yu, H. S., and Mitchell, J. K. (1998). Analysis of Cone Resistance: Review of Methods.
American Society of Civil Engineers Jour. of Geotechnical and Geoenvironmental
Engineering. Vol. 124, No. 2, pp. 140-149.
3
Review on Liquefaction Hazard Assessment
Neelima Satyam
Earthquake Engineering Research Centre
International Institute of Information Technology Hyderabad
India
1. Introduction
Experience from past earthquakes has demonstrated the vulnerability of structures to
seismically induced ground deformation. During earthquake, soil can fail due to
liquefaction with devastating effect such as land sliding, lateral spreading, or large ground
settlement. The phenomenon of liquefaction of soil had been observed for many years, but
was brought to the attention of engineers after Niigata (1964) Alaska earthquakes (1964).
Liquefaction is a phenomenon in which the strength and stiffness of a soil is reduced by
earthquake shaking or other rapid loading. Liquefaction and related phenomena have been
responsible for tremendous amounts of damage in historical earthquakes around the world
(Borchardt, 1991). During the Bhuj earthquake on 26
th
January 2001 (M=7.7) lot of damages
had been occurred due to liquefaction and other ground failures (Rao and Mohanty, 2001a).
From these investigations it was observed that a vast majority of liquefaction occurrences
were associated with sandy soils and silty sands of low plasticity.
For the last four decades, investigations on understanding the liquefaction phenomena were
carried out and have resulted in several different perspectives in describing various
liquefaction-related phenomena. Liquefaction is looked upon as the condition at which the
effective stress reaches (temporarily) a value of zero by few, while others consider
liquefaction to have occurred when the soil deforms to large strains under constant shearing
resistance. The first phenomenon is referred to as cyclic mobility and the second as flow
liquefaction which may result in significant lateral deformations by either of them. To date,
most research into liquefaction hazards has concentrated on the question of liquefaction
potential, i.e., whether or not liquefaction will occur. The influence of liquefaction on the
performance of structures, however, depends on the effects of liquefaction. While estimation
of liquefaction effects has been improved by development of empirical procedures, the
uncertainty involved in predicting these effects is still extremely high. More reliable
prediction of structural performance requires more accurate prediction of liquefaction
effects (Steven L Kramer, et.al; 2001).
2. Mechanism of soil liquefaction
It is necessary to understand the mechanism of soil liquefaction, where it occurs and why it
occurs so often during earthquakes. Figure 1 clearly depicts the mechanism of soil
liquefaction. Liquefaction of soil is a process by which sediments below water table

64
temporarily lose shear strength and behave more as a viscous liquid than as a solid. The
water in the soil voids exerts pressure upon the soil particles. If the pressure is low enough,
the soil stays stable. However, once the water pressure exceeds a certain level, it forces the
soil particles to move relative to each other, thus causing the strength of the soil to decrease
and failure of the soil follows. During earthquake when the shear wave passes through
saturated soil layers, it causes the granular soil structure to deform and the weak part of the
soil begins to collapse.
The collapsed soil fills the lower layer and forces the pore water pressure in this layer to
increase. If increased water pressure cannot be released, it will continue to build up and
after a certain limit effective stress of the soil becomes zero. If this situation occurs then the
soil layer losses its shear strength and it can not certain the total weight of the soil layer
above, thus the upper layer soils are ready to move down and behave as a viscous liquid. It
then is said that soil liquefaction has occurred.

Fig. 1. Mechanism of Soil Liquefaction
2.1 Stress condition at liquefaction
The basic difference between solid state and liquid state of a substance is that the substance
in its solid state shows resistance to deformation when subjected to external forces, whereas


65
the substance in its liquid state does not have this property. Therefore, the process of
transformation of any substance from solid state into a liquid state is substantially a process
of diminishing in shear resistance of the material. The shear resistance of cohesion less soil is
mainly proportional to the intergranular pressure and the co-efficient of friction between
solid particles which is usually given by the following relationship
t
s
= o tan | (1)
Since o = (o - u )
t
s
= (o-u) tan| (2)
where, t
s
is shear resistance,o is the effective normal stress, o is the total normal stress, u is
the pore water pressure and | is the angle of internal friction in terms of effective stress.
The condition for liquefaction is o= 0 then u will be equal to o. This can be defined in terms
of Lambe parameters as
q = (o
1
- o
3
) = (o
1
-o
3
) (3)
p = (o
1
+ o
3
) (4)
p =1/2 (o
1
+ o
3
) (5)
where, o
1
and

o
3
are maximum and minimum total principal stress, o
1
and o
3
maximum
and minimum effective principal stress respectively . Then the condition for liquefied soil
will be q = 0, p = 0 and p = u.
Although the condition of soil liquefaction obeys the same condition as explained above, the
mechanism of liquefaction process is different. Three typical mechanisms of soil liquefaction
are identified as explained below:
2.2 Liquefaction caused by seepage pressure only: Sand boils
If the pore water pressure in a saturated sand deposit reaches and excesses the overburden
pressure, the sand deposits will float or boil and lose entirely its bearing capacity. This
process is nothing to do with the density and volumetric contraction of sand. Therefore, it
has been usually considered as a phenomenon of seepage instability. However, according to
the mechanism behavior of the material, it also belongs to the category of soil liquefaction.
2.3 Liquefaction caused by monotonous loading or shearing: Flow slide
The concept of critical void ratio has been suggested by Casagrande. The skeleton of loose
saturated sand exhibits irreversible contraction in bulk volume under the action of
monotonous loading or shearing, which cause increase of pure water pressure and decrease
of effective stress and finally brings about an unlimited flow deformation.
2.4 Liquefaction caused by cyclic loading or shearing: Cyclic mobility
With various experimental techniques and testing apparatus it has been revealed that
cohesion less soil always show volumetric contraction at low shear strain level, but may

66
dilate at higher shear strain level depending upon the relative density of soil. Therefore,
under the action of cyclic shearing a saturated cohesion less soil could show liquefaction at
time intervals when shear strain is low, but may regain shear resistance in time intervals
when the shear strain level is higher. A sequence of such sort of intermittent liquefaction
would bring about the phenomenon of cyclic mobility with limited flow deformation. If the
saturated cohesion less soil was loose enough to keep contraction at high shear strain level,
then it also could come out to be an unlimited flow deformation.
3. Evaluation of liquefaction potential
The liquefaction potential of any given soil deposit is determined by a combination of the
soil properties, environmental factors and characteristics of the earthquake to which it may
be subjected. Specific factors that any liquefaction evaluation desirably takes into account
include the following:
SOIL PROPERTIES
- Dynamic shear modulus
- Damping characteristics
- Unit weight
- Grain size characteristics
- Relative density
- Soil structure
ENVIRONMENTAL FACTORS
- Method of soil formation
- Seismic history
- Geologic history (aging, cementation)
- Lateral earth pressure coefficient
- Depth of water table
- Effective confining pressure
EARTHQUAKE CHARACTERISTICS
- Intensity of ground shaking
- Duration of ground shaking
Some of these factors cannot be determined directly, but their effects can be included in the
evaluation procedure by performing cyclic loading tests on undisturbed samples or by
measuring the liquefaction characteristics of the soil by means of some in-situ tests. The
evaluation of liquefaction potential is based on two approaches (i) macroscopic evaluation
and (ii) microscopic evaluation. Evaluation of liquefaction potential can be done using field
tests or laboratory tests. The preliminary assessment of the liquefaction potential of a soil
deposit over a large area in a seismically active region can be done using the following
indices, which are characteristics of liquefiable soils:
- Mean grain size,D
50
= 0.02- 1.0 mm
- Fines content < 10%
- Uniformity coefficient < 10


67
- Relative density < 75%
- Plasticity index < 10
- Intensity of an earthquake > VI
- Depth < 15m
There are three different ways to predict liquefaction susceptibility of a soil deposit in a
particular region. They are (a) Historical criteria, (b) Geological and Geomorphological
criteria and (c) Compositional criteria (Kramer, 2000). According to historical criteria soils
that have liquefied in past can liquefy in future also. With the help of past earthquake
records one can predict the liquefaction in future.
The type of geological processes that created a soil deposit has strong influence on its
liquefaction susceptibility. Deposits formed by rivers, lakes, and wind and by man made
deposits particularly those created by the process of hydraulic filling are highly susceptible
to liquefaction. It also depends on soil type. Uniform graded soils are highly susceptible
than well-graded soil deposits also, soils with angular particles are less susceptible than soils
with rounded particles.
Cohesive soils with the following properties are vulnerable to significant strength loss under
relatively minor strains (Seed et al.1983) i.e. if percent finer than (0.002 mm) is less than 30
percent, liquid limit less than 35 percent and if the moisture content of the insitu soil is
greater than 0.9 times the liquid limit (i.e., sensitive clays).
In addition to sandy and silty soils, some gravely soils are potentially vulnerable to
liquefaction. Most gravely soils drain relatively well, however gravelly soils are also
liquefiable when the voids are filled with fine particles and if it is surrounded by less
pervious soils, drainage can be impended and may be vulnerable to cyclic pore pressure
generation and liquefaction.
Gravels tend to be deposited in a more turbulent depositional environment than sands or
silts, tend to be fairly dense, and so generally resist liquefaction. Accordingly, conservative
preliminary methods may often suffice for evaluation of their liquefaction potential. For
example, gravely deposits that can be shown to be pre-Holocene in age (older than about
11,000 years) are generally not considered susceptible to liquefaction. Andrus and Stokoe
(2000) compiled 225 liquefaction case histories from the United States, Taiwan, Japan and
China. Among these case history sites 90% of the liquefied soils had a critical layer thickness
of less than 7 m, an average depth below land surface of less than 8 m, and water table
depth is at less than 4 m below ground surface.
3.1 Field methods
The use of insitu testing is the dominant approach in common engineering practice for
quantitative assessment of liquefaction potential. Calculation of two variables is required for
evaluation of liquefaction resistance of soils. They are as follows:
1. The seismic demand on a soil layer, expressed in terms of CSR and
2. The capacity of the soil to resist liquefaction, expressed in terms of CRR.
The models proposed by Seed and Idriss (1971), Seed and Peacock (1971), Iwasaki (1978)
and Robertson and Wride (1998) methods are extensively used for predicting liquefaction

68
potential using field data. Youd et al. (2001) reviewed in detail the available field methods
available for the evaluation of liquefaction potential of soils.
3.1.1 SPT based methods
Standard penetration test is widely used as an economical, quick and convenient method for
investigating the penetration resistance of non-cohesive soils. This test is an indirect means
to obtain important design parameters for non-cohesive soils. The use of SPT as a tool for
evaluation of liquefaction potential began to evolve in the wake of a pair of devastating
earthquakes that occurred in 1964; the 1964 Great Alaskan Earthquake (M=9.2) and 1964
Niigata Earthquake (M=7.5), both of which produced significant liquefaction related
damage.
It should be ensured that the energy of the falling weight is not reduced by friction between
the drive weight and the guides or between rope and winch drum. The rods to which the
sampler is attached for driving should be straight, tightly coupled and straight in alignment.
For driving the casing, a hammer heavier than 63.5 kg may be used. Standard Penetration
Test set up and accessories are Standard split spoon sampler, 65 kg hammer, guide pipe
assembly, anvil and drill rod.
In the standard penetration test, a standard split spoon sampler is driven into the soil at the
bottom of a borehole by giving repeated blows (30-40 blows per minute), using a 65 kg
hammer released from a height of 75 cm. The blow count is found for every 150 mm
penetration. If full penetration is obtained, the blows for the first 150 mm are ignored as
those required for the seating drive. The number of blows for the next 300 mm of
penetration is recorded as the Standard Penetration Resistance, called the N value.
- If number of blows to drive 15 centimeters exceeds 50, the test has to be repeated.
- If the stratification is homogeneous and denseness is not very erratic, the spacing for the
test depth can be increased suitably beyond a depth of 6 meters or so.
- Wide variations in N-value at given depth along the section would show heterogeneity
of the subsoil and denseness.
3.1.1.1 Factors affecting test results
i. Effective over burden pressure: Effective over burden pressure affects results
considerably.
- Desai (1968) reported that N value at shallow depths underestimates relative density.
N-value corrected for the surcharge effects represent normal pressure is changed.
- It was proved that positive or negative pore pressure developed in fine sands and silty
sands depend upon the denseness of the sub-soil and thus the effective normal
pressure against split spoon sampler is altered.
- It was observed that the removal of 5 meters of soil affected N value considerably. In
cohesive soils N-value do not reflect the precompression load and shear strengths if
soils are partly saturated.
ii. Grain size and shape
- Gravels have reduced friction and penetration resistance, which will block the SPT
sampler and gives erratic results.


69
- The particle size effect on N value is prominent if 30 % soil particles are less than 0.1
mm and soils is saturated or dry.
- In case of silty fine sands and very fine sand, positive or negative pore pressure can
be generated depending on the state of compactness and N values will change
according
iii. Degree of saturation
- N values will be reduced by 15 % due to saturation and this is more pronounced in case
of loose soils.
- Penetration resistance increased due to saturation in case of denser soils while in case
of loose, fine and silty sand N value is considerably reduced.
3.1.1.2 Corrections applied in standard penetration test
i. Corrections due to Overburden
It is an established fact that SPT blows are greatly affected by the overburden pressure at
the test point. The effect of length or weight of the driving rods is not so pronounced and
may be neglected. According to their investigations, Terzaghi-Peck correlation between
SPT blows and density index is valid under an overburden pressure of approximately 280
kPa.
The curves are based on results for air dry and partially wetted, cohesion less sands and are
considered conservatively reliable in all sands, saturated or unsaturated. But it is generally
felt that the corrections provide over-estimate of density index.
For interpretation and correlations of SPT results the current thinking is to adopt 100kPa
(1kg/cm
2
)

as the reference overburden pressure and the N blows corrected for this pressure
are called the normalized or corrected values, Nc.
Nc=Cn*Nr (6)
Cn = 0.77log
10
(
20
' o
), where >0.25 kg/cm
2
(7)
where, Nr= Observed N value in the field and Cn = Correction factor
The another simple relation for the correction factor Cn that greatly cover more research
work on the correction factor carried in the USA was given in Eqn 8 as below.
Cn=
100
' o
(8)
where, in kPa
ii. Corrections due to Dilatancy
In submerged very fine or silty sands below the water table, the observed value of N may be
too great (compared to the penetration resistance of permeable submerged soils of equal
density index) if the void ratio is below the critical voids ratio which corresponds
approximately to N=15. Submerged fine sands and silty sands offer increased resistance due

70
to excess pore water set up during driving and unable to dissipate immediately (dilatancy
effect). The corrected value of N is defined in IS: 2131(1981) is as follows
N=15+
1
2
(Nc-15) (9)
where, Nc is corrected value after over burden correction
Wherever both the overburden and submerged corrections are necessary, the overburden
correction is applied first.
3.1.1.3 Seed and Idriss (1971) method
The initial approach for evaluating behavior of soils in the field during dynamic loading
was developed by Seed and Idriss (1971). The procedure is referred to as the simplified
procedure, and involves the comparison of the seismic stresses imparted onto a soil mass
during an earthquake (Cyclic Stress Ratio, CSR) to the resistance of the soil to large
magnitude strain and strength loss (Cyclic Resistance Ratio, CRR). The CSR estimation is
based on the estimated ground accelerations generated by an earthquake, the stress
conditions present in the soil, and correction factors accounting for the flexibility of the soil
mass (Youd and Idriss 1997). Seed and Idriss developed this empirical method by
combining the data on earthquake characteristics and in-situ properties of soil deposits,
which is widely used all over the world for the assessment of liquefaction hazard. For
earthquakes of other magnitudes, the appropriate cyclic strength is obtained by multiplying
with a factor called magnitude scaling factor MSF. The factor of safety against liquefaction,
F
L
can then be estimated as the ratio of CSR and CRR.
3.1.1.4 Seed and Peacock (1971) method
In the Seed and Peacock (1971) method, the average shear stress
av
will be computed same
as in Seed and Idriss method. Using corrected SPT N value and the proposed chart by Seed
and Peacock,
Z
can be calculated at the desired depth of the soil strata. If
av
>
Z
then soil
will liquefy at that zone.
3.1.1.5 Iwasaki et al. (1982) method
Iwasaki et al. (1982) proposed a simple geotechnical method as outlined in the Japanese
Bridge Code (1991). In this method, soil liquefaction capacity factor R, is calculated along
with a dynamic load L, induced in a soil element by the seismic motion. The ratio of both is
defined as liquefaction resistance. The soil liquefaction capacity is calculated by the three
factors, which take into account the overburden pressure, grain size and fine content. In this
method it is assumed that the severity of liquefaction should be proportional to the
thickness of the liquefied layer, proximity of the liquefied layer to the surface, and the factor
of safety of the liquefied layer.
The prediction by the liquefaction potential index is different than that made by the
simplified procedure of Seed and Idriss (1971). According to Toprak and Holzer (2003), the
simplified procedure predicts what will happen to a soil element whereas the index predicts
the performance of the whole soil column and the consequences of liquefaction at the


71
ground surface. Sonmez (2003) modified this method by accepting the threshold value of 1.2
of factor of safety as the limiting value between the categories of marginally liquefiable to
non-liquefiable soil.
The NCEER workshops in 1996 and 1998 resulted in a number of suggested revisions to
the SPT based procedure. Cetin et al. (2000) reexamined and expanded the SPT case
history database. The data set by Seed et al. (1984) had 125 cases of liquefaction/ no
liquefaction in 19 earthquakes, of which 65 cases pertain to sands with fines content s
5%, 46 cases had fines content between 6 and 34% and 14 cases had > 35%. Cetin et al.
(2000) used their expanded data set and site response calculations for estimating CSR to
develop revised relationships. Idriss and Boulanger (2004) presented a revised curve
between CSR and modified SPT value based on the reexamination of the available field
data.
3.1.2 CPT based method
The CPT test has become one of the most common and economical methods of subsurface
exploration. The cone penetrometer is pushed into the ground at a standard velocity of 2
cm/sec and data is recorded at regular intervals (typically 2 or 5 cm) during penetration.
The results provide excellent stratigraphic detail and repeatability provided proper care has
been taken in calibration of the equipment (transducers and electronics). The cone
penetrometer is instrumented to record a number of different parameters, with the most
common being the force of the tip, the force of the sleeve, and the pore pressure behind the
tip. Cone penetrometers have also been used to provide or measure electrical properties,
shear wave velocities, visual images of the soil, acoustic emissions, temperature and water
samples.
The CPT is a versatile sounding method that can be used to determine the materials in a soil
profile and their engineering properties. The equipment consists of a 60
0
cone, with 10 cm
2

base area and a 150 cm
2
friction sleeves located above the cone. A sensor is attached for
measuring tip resistance, pore pressure and sleeve resistance. To evaluate the potential for
soil liquefaction it is important to determine soil stratification and in-situ soil state. The CPT
is an ideal in-situ test to evaluate the potential for soil liquefaction because of its
repeatability, reliability, continuous data and cost effectiveness.
3.1.2.1 Robertson and Wride (1998) method
A simplified method to estimate cyclic shear resistance (CSR) was developed by Seed and
Idriss (1971) based on maximum ground acceleration at the site as under:
CSR = t
av
/o
0
= 0.65 (MWF) (o
0
/ o
0
) (a
max
/g) r
d
(10)
MWF = (M)
2.56
/173

(11)
where, MWF is the magnitude weighting factor and M is the earthquake magnitude,
commonly M = 7.5
Seed et al. (1985) also developed a method to estimate the cyclic resistance ration (CRR) for
clean sands and silty sands based on the CPT using normalized penetration resistance.

72
The cone penetration resistance q
c
can be normalized as
q
c1N
= C
Q
(q
c
/p
a
)

(12)
C
Q
= ( P
a
/ o
0
)
n
(13)
where, C
Q
is normalized factor for cone penetration resistance, P
a
is the atmosphere of
pressure in the same units as o
0
and n ia an exponent that varies with soil type (= 0.5 for
sands and 1 for clays) and q
c
is the field cone penetration resistance at tip. The normalized
penetration resistance (q
c1N
) for silty sands is corrected to an equivalent clean sand value
(q
c1N
)
CS
as
(q
c1N
)
CS
= K
C
q
c1N
(14)
where, K
C
is the correction factor for grain characteristics and is defined as below by
Robertson and Wride (1998).
K
C
= 1.0 for I
C
s 1.64

(15)
K
C
= -0.403 I
C
4
+ 5.581 I
C
3
21.63 I
C
2
+33.75 I
C

17.88 for I
C

>1.64 (16)
If I
C

> 2.6 , the soil in this range are likely to clay rich or plastic to liquefy. I
C

is the soil
behavior type index and is calculated as
I
C

= [(3.47 log Q)
2
+(1.22 +logF)
2
]
0.5

(17)
where Q is normalized penetration resistance
= [(q
c
-o
0
)/P
a
][ P
a
/o
0
]
n
(18)
F = [f
s
/ (q
c
-o
0
)]* 100% (19)
where f
s
being the sleeve friction stress
CRR
7.5
= 0.833[
1
( )
1000
c N cs
q
] +0.05 if
1
( )
c N cs
q <50 (20)
CRR
7.5
= 93[
1
( )
1000
c N cs
q
]
3
+0.08 if 50 s
1
( )
c N cs
q <160

(21)
where,
1
( )
c N cs
q is clean sand cone penetration resistance normalized to approximately 100
kPa (1atm). Then, using the equivalent clean sand normalized penetration
resistance
1
( )
c N cs
q , CRR can be estimated from the Fig. 2.
The CPT based liquefcation correlation was reevaluated by Idriss and Boulanger (2004)
using case history data compiled by Shibata and Teparaksa (1988), Kayen et al. (1992),
Boulanger (2003) and Moss (2003).
Moss (2003) has provided a most comprehensive compilation of field data and associated
interpretations. He used friction ratio R
f
instead of the parameter I I
C,
soil behavior type index
and examined for the cohesion less soils with fines content greater than or equal to 35%.


73

Fig. 2. Calculation of CRR from CPT q
c1N
(Youd et al., 2001)
3.1.3 Shear wave velocity (V
S
) based methods
The shear wave velocity based procedure has advanced significantly in recent years, with
improved correlations and more databases as summarized by Andrus and Stokoe (2000).
Shear wave velocity can be determined either by subsurface geophysical method or by
surface geophysical method as explained earlier Liquefaction potential can be evaluated
from the shear wave velocity (V
S
) using the following three methods. These procedures can
be useful particularly for sites where it is difficult to penetrate or sample soils.
3.1.3.1 Andrus and Stokoe (2000) method
Andrus and Stokoe have carried extensive research into the use of shear wave velocity as an
index of liquefaction resistance. Various investigators have developed relationships between
shear wave velocity and liquefaction resistance. This section presents detailed guidelines for
applying the procedure described in Andrus and Stokoe that was developed using
suggestions from two workshops and following the general format of the Seed and Idriss
(1971) simplified procedure. Andrus and Stokoe (2000) developed liquefaction resistance
criteria from field measurements of shear wave velocity Vs. This method form the basis for
the currently accepted shear wave velocity criteria for liquefaction potential assessment.
Liquefaction hazard assessment based on this method is done and a liquefaction hazard
map of Delhi region hazard map is prepared.

74
Shear wave velocity Vs is corrected similar to SPT N value using the atmospheric pressure
Pa and initial effective vertical stress,
0
. The cyclic resistance ratio (CRR) is determined
empirically at various depths using the correlation developed between CRR and the shear
wave velocity for the liquefaction assessment. The detailed analysis procedure is outlined in
the later chapter.
3.1.3.2 Hatanaka, Uchida and Ohara (1997) method
Hatanaka et al. (1997) performed a systematic research relating the undrained cyclic shear
strength of high quality undisturbed gravel samples to the shear wave velocity measured
insitu V
S1
is used for correcting the effect of effective confining stress on V
S
by using the
following equation.
V
S1
=
3/8
'
( )
98
o
s
V
v
(22)
The value of 3/8 in the above equation is the average value of 0.5 and 0.25 which covered
the test result. It is also important to know the K
O
value for converting the liquefaction
strength obtained in laboratory R
LAB
to that in the field R
INSITU
based on the following
equation
R
INSITU
= 0.9 {
(1 2 )
3
+
o
K
} R
LAB
(23)
The variation of shear wave velocity with confining pressure for various soil conditions was
given by Hardin (1963).
3.1.3.3 Tokimatsu, Yamazaki, and Yoshimi (1986) method
This method is proposed by Tokimatsu et al. (1986). The working principle in this procedure
is that the liquefaction strength has a good correlation with elastic shear modulus for a given
soil under a given confining pressure. The procedure is outlined in Fig. 3.The procedure
shown on the left corresponds to the shear wave velocity measurement in-situ. Based on the
measured shear wave velocity, Vs the elastic shear modulus in the field G
OF
can readily be
determined. The procedure on the right involves laboratory tests on a specimen
reconstituted from the sample obtained at the site. Both liquefaction test, the shear modulus
at small shear strain G
OL,
of the specimen is measured and compared with G
OF.
If they are
equal, the liquefaction test has to be done on the same specimen. If they are not equal that
usually means that G
OF
is larger than G
OL,
cyclic shear stresses are applied to the specimen
until the shear modulus reaches the value; then the liquefaction test is conducted. In order to
take the effects of void ratio (e) and confining stress (o
c
) on the G
O
is normalized with
respect to e and o
c
as described as
G
O
=
{ } min
( )( ') o
o
n
G
F e
(24)

min
( ) F e = (2.17 -
min
e )
2
(1+
min
e ) (25)
where, n is a factor whose value equals to 2/3.


75

Fig. 3. Outline of Method (Tokimatsu et al., 1986)
In this method, it is important to determine G
OL
from laboratory test. G
OL
value highly
depends on the confining stress at laboratory. As a result it is an important work to
reasonably estimate the k
o
value of insitu soils.
Tsurumaki et al. (2003) verified the simplified procedure using new data for different kinds
of gravelly soils by undrained cyclic triaxial tests on undisturbed sample of gravelly soils
obtained by the in-situ freezing sampling method. The undrained cyclic strength increases
with increasing shear wave velocity.
3.2 Laboratory methods
3.2.1 Cyclic triaxial test
The cyclic triaxial test is the most commonly used test for the measurement of dynamic soil
properties at high strain levels. This test simulates the liquefaction phenomenon during
earthquakes by appling cyclic shear to the saturated sandy soil under undrained condition.
Axial loading is applied in steps to the specimen and the shear strain and shear stress at
various levels of loading. Five to ten levels of shear strain amplitudes are chosen from the
range of 10
-5
to 10
-2
for testing. Dynamic deformation characteristics are influenced by the
Evaluation of Insitu
Liquefaction
Liquefaction
Test
G
OF
= G
OL

Preshear of
Prestress
Obtain Undisturbed Sample Measure Vs
G
OF
= *Vs
2

Measure G
OL


76
effective confining pressure during the test. When an undisturbed sample of normally
consolidated soil is obtained, the effective vertical pressure at the depth of sampling is
isotropically applied by cell pressure to avoid the influence of over consolidation. In order
to obtain in-situ shear modulus of the soil from the laboratory test results, correction of these
results is necessary so that a shear modulus corresponding to the average effective principal
stress at the sample depth is obtained.
Seed and Lee (1966) were the first to reproduce liquefaction in a cyclic load triaxial test on
loose and dense sands and concluded that liquefaction occurs more easily in sandy soils
having higher void ratios and void ratio remaining constant, lower the effective confining
pressure higher the liquefaction susceptibility.
3.2.2 Cyclic direct simple shear test
The cyclic direct simple shear test is capable of reproducing earthquake stress conditions
much more accurately than the cyclic triaxial test. It is most commonly used for liquefaction
testing. In this test, a short cylindrical specimen is restrained against lateral expansion by
rigid boundary platens, a wire reinforced membrane or with a series of stacked rings. By
applying cyclic horizontal shear stresses to the top or bottom of the specimen, the test
sample is deformed in the same way as an element of soil subjected to vertically
propagating S waves. In recent years, simple shear devices that allow independent control
of vertical and horizontal stresses have been developed. To better simulate actual
earthquake conditions, Pyke (1979) used a large-scale simple shear apparatus. It is
concluded that cyclic strength is related to the relative density of the soil and cyclic stresses
that cause liquefaction in simple shear were less than those causing liquefaction in triaxial
shear.
3.2.3 Cyclic torsional shear test
Many of the difficulties with cyclic triaxial and cyclic shear test can be overcomed with
cyclic torsional shear test. This is mostly used to determine stiffness and damping
characteristics over a wide range of strain levels. It allow isotropic or an isotropic initial
stress conditions and can impose cyclic shear stresses on horizontal planes with continuous
rotation of principal axes. Dobry et al.(1995) used strain controlled cyclic torsional loading
along with stress controlled axial loading of solid specimens and has proven effective for
measurement of liquefaction behavior. Torsional testing of solid specimens, however
produces shear stresses that range from zero along the axis of the specimen to a maximum
value at the outer edge. To increase the radial uniformity of shear strains, a hollow
cylindrical cyclic torsional shear apparatus were also developed. While hollow cylinder tests
offer perhaps the best uniformity and control over stresses and drainage. Ishihara and Li
(1972) developed a torsinal triaxial shear test and conducted strain controlled tests on solid
cylinders of saturated sands. These tests helped in establishing relationship between cyclic
triaxial tests, cyclic simple shear tests and the torsional triaxial test.
3.2.4 Shake table test
Shake table tests of many sizes are being used for liquefaction studies on saturated soil
samples prepared in a container, fixed to a shaking platform and vibrated at the desired


77
frequency for a prescribed time. A surcharge is placed on the sample to provide the
confining pressure. The measurements of acceleration, pore water pressure and settlements
are made during the test. Shaking tables utilize a single horizontal translation degree of
freedom, but shake table with multiple degrees of freedom have also been developed.
Kokusho (1987) developed a numerical model based on shake table test.
4. Magnitude scaling factor
Seed and Idriss (1982) introduced a correction factor termed as magnitude scaling factor
(MSF). This factor can be used to scale up or scale down the CRR based curves upward or
downward, depending upon the earthquake magnitude. Figure 4 gives the curves proposed
by various authors for different earthquake magnitudes.
4.1 Seed and Idriss (1982) scaling factor
Seed and Idriss (1982) developed a set of MSF from average number of loading cycles for
various earthquake magnitude and laboratory test results. These MSF have been routinely
applied in engineering practice.

Fig. 4. Magnitude Scaling Factor Derived by Various Investigations (Youd et al., 2001)
Idriss (1995) developed a revised set of magnitude scaling factor and was defined as
MSF = (
2.24
2.56
10
w
M
) (26)
Earthquake Magnitude, M
W
M
a
g
n
i
t
u
d
e
S
c
a
l
i
n
g
F
a
c
t
o
r
,
M
S
F

78
The revised scaling factors were higher than the original scaling factors for magnitude < 7.5
and somewhat lower than the original factors for magnitude > 7.5 relative to the original
scaling factors the revised factors lead to a reduced calculated liquefaction hazard for
magnitude < 7.5 but increase calculated hazard for magnitude > 7.5. National Center for
Earthquake Engineering Research (NCEER, 2001) workshop participants suggested this
revised scaling factors as lower bound for MSF values.
4.2 Ambraseys (1988) scaling factor
Ambraseys (1988) analyzed the liquefaction data and calculated cyclic stress foe sites that
did or did not liquefy versus N
60
. Based on this he developed an empirical equation that
define CRR as a function of N
60
and moment magnitude. For magnitude less than 7.5, MSF s
suggested by Ambraseys are significantly larger than both the scaling factors developed by
Seed and Idriss. For magnitudes > 7.5, Ambraseys factors are significantly lower and much
more conservative. It is not recommended for hazard evaluation because of few data
constrain.
4.3 Andrus and Stokoe (1997) scaling factor
Andrus and Stokoe developed scaling factors by drawing bounding curves for sites where
surface effects of liquefaction were or were not observed for earthquake magnitudes of 6, 6.5
and 7.0.MSF for magnitude < 6 and > 7.5 were extrapolated from the following equation
MSF = (
7.5
w
M
)
2.56
(27)
For magnitude < 7.5 the MSF s proposed by Andrus and Stokoe are rather close in value to
the MSFs proposed by Ambraseys. For magnitude > 7.5 Andrus and Stokoe MSFs are
slightly smaller than the revised MSFs proposed by Idriss.
4.4 Youd and Noble (1997) scaling factor
Youd and Noble (1997) used a probabilistic and logistic analysis to analyze case history data
from sites was or were not reported following past earthquakes. They defined three sets of
MSFs for different magnitude ranges and with different probabilities of liquefaction (P
L
)
occurrence.
P
L
< 20 % MSF = [
3.81
4.53
10
M
] for M
W
< 7 (28)
P
L
< 32% MSF = [
3.74
4.33
10
M
] for M
W
< 7 (29)
P
L
< 50% MSF = [
4.21
4.81
10
M
] for M
W
< 7.75 (30)
The NCEER (2001) workshop report provides a useful insight in to the choice of MSF. For
magnitude < 7.5, the lower bound for the recommended range is the new MSF proposed by


79
Idriss. The suggested upper bound is the MSF proposed by Andrus and Stokoe (1997). The
upper bound values are consistent with MSFs suggested by Ambraseys (1988) and Youd
and Noble(1997a) for P
L
< 20%. For magnitude > 7.5, the new factors recomnded by Idriss
should be used. These new factors are smaller than the original Seed and Idriss (1982)
factors. The relations by Ambraseys (1998) and Arango (1996) give significantly larger MSF
values for earthquake magnitudes less than 7. Table 1 gives the MSF values given by various
researchers.

Magnitude
M
Seed
and
Idriss
(1982)
Ambra-
seys
(1988)
Arango (1966) Andrus
and
Stokoe
(1997)
Youd and Noble (1997)
Distance
Based
Energy
Based
P
L
< 20% P
L
< 32% P
L
< 50%
5.5 1.43 2.86 3.00 2.20 2.80 2.86 3.42 4.44
6.0 1.32 2.20 2.00 1.65 2.10 1.93 2.35 2.92
6.5 1.19 1.69 1.60 1.40 1.60 1.34 1.66 1.99
7.0 1.08 1.30 1.25 1.10 1.25 1.00 1.20 1.39
7.5 1.00 1.00 1.00 1.00 1.00 --- --- 1.00
8.0 0.94 0.67 0.75 0.85 0.80 --- --- 0.73
8.5 0.89 0.44 --- --- 0.65 --- --- 0.56
Table 1. Magnitude Scaling Factors Suggested by Various Researchers (Youd and
Noble,1997)
During the Niigata earthquake (M=7.5) on June 16, 1964, widespread liquefaction was
observed in the low-lying areas. The liquefaction was accompanied by foundation failure
and failure of retaining structures. It is important to note, however, that the structures
themselves suffered very little damage; they essentially settled and rotated as rigid bodies
under the loss of bearing capacity. Several of the apartment buildings were later jacked back
to a vertical position and underpinned with new foundations. Nearly 20 years after the
earthquake, excavation beneath the building, as part of the construction of upgraded
foundations for an increase in the height of the building, revealed that the piles had been
extensively damaged during the Niigata earthquake. Large displacements of the buildings
were driven by gravity and resulted in performance that would be considered as failure
by almost any definition, even though the structural elements of the buildings were
virtually undamaged. In the 1964 Niigata earthquake, a number of RC buildings settled and
tilted due to a loss of bearing capacity.
During the 1990 Luson earthquake, substantial building damage occurred in 2 to 4 story
buildings resting on liquefied ground in Daguapan City. During the 1999 Kocaeli
earthquake in Turkey, settlement or tilting of buildings due to liquefaction was widespread
in Adapazari. Although liquefaction in this case was less severe than in Niigata probably
because the soil contained a lot of fines, many 5 to 6 story buildings suffered heavy
structural damage due to uneven settlement of at most 2 m .Riverbanks, levees, and dikes
suffered liquefaction induced settlement and sliding failures during the 1993 Kushiro-Oki
earthquake, the 2003 Tokachi-Oki earthquake and many other earthquakes. During the 2003

80
Tokachi-Oki earthquake, the right banks of the Tokachi River suffered severe damage due to
liquefaction-induced deep sliding and crest settlement . Another failure mode is the uplift of
subsurface structures such as pipelines, buried tanks, pools, etc. Concrete manholes and
pipes of sewage systems are raised because of liquefaction in surrounding soils during the
1993 Kushiro-Oki earthquake, the 2003 Tokachi-Oki earthquake, and the 2004 Niigata ken-
Chuetsu earthquake. Although surrounding ground appeared stable during earthquakes,
sands backfilling buried structures liquefied and buoyant pressure in the liquefied fill raised
them over 1.5 m. The detailed liquefaction assessment requires several inputs regarding the
site specific geological, geophysical, geotechnical, seismotectonic, ground motion
parameters and their effects on the structures considering their design aspects, local soil
conditions which would enhance the earthquake effects like soil amplification, liquefaction
of soils etc.
5. References
Ambraseys N. (1988) Engineering Seismology, J. Earthq. Eng. and Struct.Dyn.,17, p.66
Andrus R.D. and Stokoe K.H. (1997) Liquefaction Resistance of Soils from Shear Wave
Velocity, Proc. NCEER Workshop on Evaluation of Liquefaction Resistance of Soils, 89-
128.
Andrus R.D. and Stokoe K.H. (2000) Liquefaction Resistance Based on Shear-Wave Velocity,
J .Geotech. Engg. Div., ASCE., (126)11:1015-1025.
Arango I. (1996) Magnitude Scaling Factors for Soil Liquefaction Evaluations, J.Geoecht.
Engg. Div., ASCE.,.122(11), 929-936.
Borcherdt R.D., Wentworth C.M., Glassmoyer G., Fumal T.,Mork, P. and Gibbs, J (1991) On
the Observation, Characterization, and Predictive GIS Mapping of Ground
Response in the San Francisco Bay Region,California, Proc of 4th International
Conference on Seismic Zonation, 3, 545-552.
Boulanger, R.W. 2003. High overburden stress effects in liquefaction analysis. Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, 129(12): 10711082.
Cetin K.O., Seed R.B., Moss R.E.S., Der Kiureghian A., Tokimatsu K. H., L.F. Jr., and Kayen,
R.E. (2000). Field Case Histories for SPT-Based In Situ Liquefaction Potential
Evaluation, Geotechnical Engg. Research Report UCB/GT-2000/09.
Dobry, R., Tabaoda, V. and Liu.L. (1995) Centrifuge Modeling of Liquefaction Effects During
Earthquakes, Proc. of 1st International Conference on Earthquake Geotechnical Engg..3,
1291-1324.
Desai M.D (1968) Subsurface, Exploration by dynamic penetrometers. Nasal Pub. & printers,
Surat, 17 pp.
Hatanaka M., Suzuki Y., Kawasaki T. and Masaaki E. (1988) Cyclic Undrained Shear
Properties of High Quality Undisturbed Tokyo Gravel, J. Soils and Foundations, (28)
4, 57-68.
Hatanaka M., Uchida A. and Ohara J. (1997) Liquefaction characteristics of a gravelly fill
liquefied during the 1995 Hyogo-ken Nanbu Earthquake, Soils and Foundations, 1,
3,107-115.
IDRISS I.M. and BOULANGER R.W. (2004) Semi-Empirical Procedures for Evaluating
Liquefaction Potential During Earthquakes. Proc. of 11th SDEE and 3
rd
Conference.
University of California, Berkeley IS: 2131 (1981) Method for Standard Penetration
Test for Soils


81
Iwasaki T., Tokida K., Taksuoko F., Watanabe S., Yasuda S. and Sato H.(1982) Microzonation
for Soil Liquefaction Potential using Simplified Methods, Proc. of 3rd International
Conf. on Microzonation, 3, 1319-1330.
Kayen R. E., Mitchell J. K., Seed R. B., Lodge A., Nishio S., and Coutinho R. (1992)
Evaluation of SPT, CPT, and Shear Wave Based Methods for Liquefaction Potential
Assessment using Loma Prieta Data, Proc. of 4
th
Japan-U.S. Workshop on Earthquake
Resistant Design of Lifeline Facilities and Countermeasures for Soil Liquefaction,
Hamada, 177-204.
Kramer S,L. (2000) Geotechnical Earthquake Engineering, Prentice Hall, NewJersey, 653 P.
Kokusho T. (1987) In-situ Dynamic Soil Properties and their Evaluation, Proc.8th Asian
Regional Conf. of SMFE, Kyoto, II: 215-240.
Moss R.E.S., Seed R.B., Kayen R.E., Stewart J.P., Youd T.L., and Tokimatsu, K. (2003) Field
Case Histories for CPT-Based Insitu Liquefaction Potential Evaluation. UC Berkeley
Geoengineering Research Report No.UCB/GE-2003/04
Pyke R. M. (1979) Nonlinear Soil Models for Irregular Cyclic Loadings, J.Geotech. Engg. Div.,
ASCE, 105(6), 715-726
Rao K.S. and Mohanty W.K. (2001a) The Bhuj Earthquake and Lessons for the Damages, IGS
News, 33, 3-10.
Robertson P.K. and Wride C.E. (1998) Evaluating Cyclic Liquefaction Potential Using the
Cone Penetration Test, J. Canadian Geotechnical Engg.,(35)3, 442-459.
Seed H. B. Tokimatsu K. Harder L. F. and Chung, R. M. (1985) Influence of SPT Procedures
in Soil Liquefaction Resistance Evaluations, J.Geotechnical Engg. Division, ASCE,
(111) 12, 1425-1445.
Seed H.B and Idriss I.M. (1971) Simplified Procedure for Evaluating Soil Liquefaction
Potential, J. Soil Mechanics and Foundations, ASCE, (97) SM9,1249-1273.
Seed H.B. and Lee K.L. (1966) Liquefaction of Saturated Sands During Cyclic Loading,
Journal of the Soil Mechanics and Foundation Division, ASCE, 92, 105-134
Seed H.B., and Idriss I.M. (1982) Ground Motions and Soil Liquefaction During
Earthquakes: Earthquake Engineering Institute Monograph, p134.
Seed H.B., Idriss I.M., and Arango I. (1983) Evaluation of Liquefaction Potential Using Field
Performance Data, J. Geotechnical Engg., ASCE, 109 (3)458-482.
Seed H.B. and Peacock W.H. (1971) Test Procedures for Measuring Soil Liquefaction
Characteristics, J. Soil Mechanics and Foundations Div., ASCE, 152-167.
Shibata T. and Teparaksa W. (1988) Evaluation of Liquefaction Potential of Soils Using Cone
Penetration Tests, Soils and Foundations, (28) 2, 49-60.
Steven L. Kramer.,Ahmed.W. Elgamal (2001) Modelling Soil Liquefaction Hazards for
Performance-Based Earthquake Engineering: Pacific Earthquake Engineering
Research Centre,p 5-8.
Tokimatsu K., Yamazaki T. and Yoshimi Y. (1986) Soil Liquefaction Evaluations by Elastic
Shear Moduli, Soils and Foundations, 26, 25-35
Tsurumaki.S K.N., Yuichi K., Uchida A. and Babasaki R. (2003) Study on the Simplified
Procedure for Evaluating the Undrained Cyclic Strength for Gravelly Soils, Proc. of
17th International Conference on Structural Mechanics in Reactor Technology (SMiRT
17)Prague, Czech Republic, August 17 22.
Youd T.L., Idriss I.M., Andrus R.D., Arango I., Castro G., Christian J.T., Dobry R., Finn
W.D.L., Harder L.F.Jr., Hynes M.E.,Ishihara K., Koester J.P., Liao S.S.C., Marcuson

82
W.F.III., Martin G.R., Mitchell J.K., Moriwak, Y., Power M.S., Robertson P.K., Seed
R.B. and Stokoe K.H.II. (2001) Liquefaction Resistance of Soils: Summary Report
from the 1996 NCEER and 1998 NCEER/NSF Workshops on Evaluation of
Liquefaction Resistance of Soils, J. Geotechnical Engg., ASCE, (127)10,817-833.
Youd, T. L., And Noble, S. K. (1997). Magnitude scaling factors. Proc., NCEER Workshop on
Evaluation of Liquefaction Resistance of Soils, Nat. Ctr. for Earthquake Engrg. Res.,
State Univ. of New York at Buffalo, 149165.
4
Liquefaction Remediation
Sarfraz Ali
NUST College of Civil Engineering
Pakistan
1. Introduction
Naturally occurring soils are heterogeneous materials due to their varying composition
broadly termed as granular and cohesive in view of their grain size distribution, and or
contractive and dilative according to physical behaviour. Their engineering behaviour is
affected by composition, moisture conditions, stress history, boundary conditions, future
loading conditions, seismic conditions, etc. The intensity and magnitude of such variable
factors make the engineering behaviour of soils under static and dynamic conditions
variable from one point to another and is thus regarded as nonlinear and anisotropic.
Behaviour under low and high moisture conditions is significantly different under dynamic
loading such as earthquakes, machine vibrations, traffic vibrations, etc.
Engineering behaviour of soils under loads is evaluated in terms of shear strength which
comprises of parameters known as cohesion, angle of internal friction, effective stress, and
pore pressure. Under static conditions (monotonic loading) when loads reach to limit where
pore pressure equals effective stress, saturated loose granular soils loose their strength and
behave like a liquid. This phenomenon is called liquefaction. While behaviour of medium to
dense saturated sands under static loading is dilative in nature, under cyclic loading of
certain magnitude and shaking, undergoes failure similar to liquefaction called cyclic
mobility. Civil engineering structures built on sites of loose saturated and dense
cohesionless soils are therefore at risk of liquefaction and cyclic mobility with potential of
damage to the structures and need improvement to safeguard against damages by
liquefaction or cyclic mobility.
In the wake of failures due to flow slides of loose saturated sand slopes and hydraulic
shells of dams, the concept of liquefaction was introduced by Arthur Casagrange in
between 1935 and 1938. Large scale damages due to liquefaction came in the lime light
during 1964 earthquakes in Anchorage, Alaska, and Nigata in USA and Japan and in
Loma Prieta earthquake of 1989 in USA. Geotechnical engineering concepts of shear
strength, effective stress, and pore water pressures, etc. were introduced by that time and
the geotechnical engineers were able to explain the reasons of such damages; the loss of
strength of loose saturated sand soils due to cyclic stresses caused by the earthquake
waves called, liquefaction. Since then, exhaustive research efforts have been made to
understand phenomenon of liquefaction due to monotonic and dynamic loadings,
procedures to evaluate liquefaction, and remediation potential hazards due to
liquefaction.
Soil Liquefaction and Seismic Safety of Dams and Monuments 84
2. Background
On October, 8, 2005, Kashmir area of Pakistan was struck by a 7.8 magnitude of earthquake.
Keeping in view, damages of public life and infrastructure (roads, life lines, etc.), this was
one of deadliest earthquakes in the history of Pakistan and the world. The destruction zone
remained in the mountainous regions where soil sediments are mostly devoid of loose
saturated sands and therefore damages due to liquefaction were not noticed. However, in
the backdrop of damages caused by this earthquake, building codes were revised with
stringent measures on evaluation of all earthquake related hazards and construction of
earthquake resistant structures.
Soon after this earthquake, a firm had planned to construct a heavy workshops complex
in an industrial area located some 65 km west of Islamabad. The planned RCC framed
workshops had covered area of 25000 m and individual column loads varied from 1200
kN to 2300 kN. Geotechnical site investigation revealed that the soil at construction site is
composed of loose pockets of alluvial deposits of Indus River comprising silty and sandy
strata with varying degree of fines in different layers. Besides potential hazard of
liquefaction, the site had low bearing capacity. Allowable bearing capacity (BC) of 100 kPa
was available against 150 kPa required to support the foundation. Deep Dynamic
Compaction DDC technique being the cheapest as compared to other techniques was
selected to remediate liquefaction hazard and to improve the BC to 160 kPa. Uncertainty
both in defining subsurface conditions and soil response to a specific soil improvement
technique necessitated detailed testing of the site to evaluate effectiveness of the designed
program.
3. Purpose and scope
The DDC project at Attock industrial area being first soil improvement project in the
country was viewed as a good opportunity to enhance practical aspects of this technique.
While the project was going on, the opportunity was used to carry out research with a view
to formulate liquefaction evaluation methodology, devise methods to monitor degree of
improvement and effectively measure and manage ground vibrations. This chapter while
highlighting literary aspects of this technique provides detailed account of planned
objectives and outcomes. The author has proposed a practical procedure for undertaking
soil improvement project based on his personal experience of this real time project. The
scope of the chapter is limited to practical aspects of case history project and covers
following:
Approach adopted for site characterization.
Methodology for liquefaction evaluation.
Liquefaction Remediation.
Ground Vibrations.
Quality Controls.
Conclusions and Recommendations.
4. Site characterization
Evaluation for liquefaction involves identifying nature of soils and loading conditions that
will cause the liquefaction to occur. While determining liquefaction susceptibility, there

Liquefaction Remediation 85
could be two possibilities; firstly the soil deposits found in natural state of stresses and
subjected to earthquake shaking and secondly the natural soil deposit subjected to
construction of structures (static loading) and expected dynamic loading from machine or
traffic vibrations. Evaluation for liquefaction must therefore encompass site characterization
and well considered loading conditions (past stress history, stresses due to structures,
earthquake, traffic/machine vibration, etc.).
Site characterization comprising geotechnical and geophysical techniques of the potential
project site and up to a depth of influence of the planned structures should be performed to
evaluate all aspects affecting engineering behaviour of the site during useful life of the
planned structures. Various aspects of site characterization are described separately and
shown in Figure 1.
4.1 Geology
Geological structures such as faults folds, etc., type of sediment (marine, plains,
mountainous, etc.), age of sediments, degree of consolidation, past stress history,
stratigraphy, etc.
4.2 Topography
Topographic details such as location of the site, general relief of the site and adjacent areas,
flow of storm water towards the site, elevation, etc.
4.3 Hydrology and boundary conditions
Hydrological and boundary conditions such as depth of ground water table and its
fluctuation with seasonal variation (if any) due to rains, drainage pattern of the site and
adjacent areas, proximity to water bodies affecting moisture conditions of the site such as
floods, ground water flow regimes, etc.
4.4 Regional tectonic setting and seismicity
Aspects related to regional tectonic setting and seismicity such as proximity to fault lines,
type of fault lines, past earthquake tsunami history, local seismic codes on horizontal and
vertical peak ground accelerations, duration of shaking, past history of earthquake damages
such as ground ruptures, liquefaction, collapse of building, destruction of life lines, etc. A
site specific seismic hazard analysis will be very useful tool in evaluation of more realistic
dynamic parameters and is recommended to be performed.
4.5 Field and laboratory investigation
Field and laboratory investigation should be conducted to develop SPT/CPT/DCPT
profiles (these would later be used for comparing pre to post compaction degree and
depth of improvement), unit weight, moisture content, Atterberg Limits, shear strength
parameters, shear wave velocity (this will be used to assessing ground vibrations), etc.
Depth is an important aspect for evaluation of liquefaction hazard. Generally it should be
equivalent to the depth of influence of the planned structures or 20 m depth whichever is
larger. Depth of influence of structures can be calculated using stress distribution theories
available in the literature.
4.6 Geotechnical profile
Finally a pre improvement geotechnical profile of the soil deposit at site should be
developed consisting of type of layering, percentage fines, moisture content, and
SPT/CPT/DCPT variation, shear wave velocity in tabular or graphical pattern as shown in
Table 1 and Figure 2 (more columns can be added for pre to post compaction comparison).
This format will be a useful ready reckoner during the execution phase of the project and for
pre to post compaction comparison.

Fig. 1. Parameters fir site characterization

Depth
(m)
Type of
Soil
Fine
Contents
(%)
Moisture
Content
(%)
SPT N-Values
Shear Wave Velocity
(m/sec)
Pre
compaction
Post
compaction
Pre
compaction
Post
compaction
Table 1. Geotechnical Profile

Fig. 2. Geotechnical Soil Profile
5. Methodology adopted for evaluation of liquefaction
Susceptibility of soil deposit at site to liquefaction should be determined through multiple
approaches since remediation of liquefaction hazard is a cost intensive event that would
raise the overall project cost. Various criterions evolved by researchers. Historical,
geological, and compositional criterion suggested by Kramer, (1996) were used to evaluate
liquefaction susceptibility.
5.1 Evaluation for liquefaction susceptibility
Evaluation for liquefaction susceptibility can be made according To Historical
Geologic Compositional criterion proposed by Kramer (1996) and empirical correlations
based on SPT N-values. Historically, the construction site of research project is located in
Himalayan seismic zone having a long history of earthquakes in the past and prediction of
earthquakes in the future therefore the site demanded an evaluation for seismic hazards.
Geologically the construction site is composed of poorly sorted an alluvial stratum of recent
age with un-corrected SPT N-value of less than 12. Compositionally, the soil layers present
at depths from 3.5 to 5.5 m, containing fines as low as 6%, low SPT N-values, and rise of
Depth
(m)
Soil Type
SPT N-Value
5 10 20
1
2
3
4
5
6
7
8
Compacted Fill
(Silty Sandy Clay)
Poorly Graded
Sand with Silt
Silty Clayey Sand
Well Graded Sand
with Clay
Silty Clayey Sand
Fines
75%
6%
16%
10%
36%
Depth
(m)
Soil Type
SPT N-Value
5 10 20
1
2
3
4
5
6
7
8
Compacted Fill
(Silty Sandy Clay)
Poorly Graded
Sand with Silt
Silty Clayey Sand
Well Graded Sand
with Clay
Silty Clayey Sand
Fines
75%
6%
16%
10%
36%
Fines
75%
6%
16%
10%
36%
ground water table to a 5 m (from existing 12 m) in monsoon season renders the project site
susceptible to liquefaction. Since al sites susceptible may not actually liquefy, therefore
evaluation for initiation of liquefaction must be performed to actually establish or rule out
the potential hazard of liquefaction.
5.2 Evaluation for initiation of liquefaction
The next step after evaluation of liquefaction susceptibility is the evaluation for initiation of
liquefaction because it might happen that a soil at a particular site is susceptible to
liquefaction but devoid of monotonic and dynamic loading conditions necessary to initiate
it. Its occurrence requires a loading that is strong enough to initiate or trigger liquefaction.
Most popular methods for evaluation of initiation of liquefaction are "cyclic stress approach"
and "cyclic strain approach" (Kramer, 1996). In this paper, initiation of liquefaction is
evaluated using cyclic stress approach. An earthquake of magnitude 7, peak ground
acceleration of 0.16g, and ground motion of 0.4 seconds were used for evaluation for
initiation of liquefaction for the case history project.
5.3 Characterization of site specific dynamic loading
Site specific analysis should be carried out for the expected monotonic loading and the
existing soil conditions. Site specific dynamic analysis for a particular site requires
characterization of expected dynamic loading (earthquake, machine, traffic, etc.) in terms of
a level of uniform cyclic shear stress that is applied for an equivalent number of cycles
(Kramer, 1996). Equation (1), proposed by Seed et al. (1983) can be used for calculation of
equivalent cyclic shear stresses. In equation (1),
cyc
= cyclic shear stress at the time of
earthquake,
max
a

= peak ground surface acceleration,
v

= total vertical stress, g =
acceleration due to gravity,
d
r = value of stress reduction factor for a given depth (0.98 to
0.96 for depth of 3.5 to 5.5 m depth, Seed and Idriss, 1971).

max
0.65
cyc
v d
a
g r

(1)
5.4 Characterization of liquefaction resistance
Cyclic shear resistance should be calculated at various depths using equation (2), (Kramer,
1996).

,
'
o
cyc L L
v
CSR
(2)
Where '
o
v
= initial vertical effective stress and
L
CSR = cyclic stress ratio. SPT N
Field
values
were corrected to (N
1
)
60
by applying appropriate corrections which include corrections for
overburden, energy, borehole diameter, rod length, and sampling.
5.5 Factor of safety
Factor of safety against liquefaction is expressed as FS
L
=
, cyc L cyc
. Liquefaction occurs
when FS
L
is less than 1 (Kramer, 1996). Factor of safety should be checked at various depths

to ascertain initiation of liquefaction or otherwise. Factor of safety of the research project
was less than 1 prior to DDC, Table 2.

Depth (m) (N
1
)
60

L
CSR
'
o
v
(kN/m
3
)
, cyc L

cyc

FS
L

3.5 6 0.06 64.01 3.84 6.52 0.59
4.5 7 0.07 71.43 5.00 8.19 0.61
5.5 8 0.08 80.12 6.41 9.96 0.64
Table 2. Cyclic Shear Resistance of Soil and Factor of Safety (FS
L
)
6. Liquefaction remediation
Various techniques are available to improve a potential project site to a desired level of site
performance vis--vis potential hazards. Selection of site will depend on factors such as
geotechnical site conditions, required degree of improvement, availability of technology,
cost, time, etc.
6.1 Soil improvement techniques to remediate liquefaction
Some of the techniques suitable for remediation of liquefaction hazards are; soil replacement
(if desired depth of improvement is shallow), deep dynamic compaction, vibrocompaction,
grouting, deep soil mixing, etc. In order to select a most efficient and economical
remediation technique requires a comprehensive review of existing geotechnical site
investigation, susceptibility to liquefaction hazard, compactibility of soil at site, cost-benefit
analysis of techniques vis--vis desired degree and depth of improvement. Since this
chapter deals with remediation of liquefaction hazard using DDC, only details pertaining to
this technique will be discussed at length hereafter.
6.2 Deep dynamic compaction technique
Deep dynamic compaction technique has effectively been used for remediation of
liquefaction hazard in Pakistan and other countries. The technique has also been used
successfully for improvement of Municipal Solid Waste sites for construction of buildings,
parks, recreational facilities, etc. The technique though simple, requires careful design,
construction, and quality controls as described to ensure desired results. It involves
repeatedly dropping of heavy weights from a crane on each impact point. The impact
energy of the falling weight densifies the soil at depth. DDC program generally consist of
two or more passes followed by ironing pass to compact the top ground surface, Figures 3
and 4. The weight of tamper, height of fall, number of drops, and grid spacing is selected to
achieve required depth and degree of improvement. The mechanism of densification for
unsaturated soil is analogous to large scale proctor compaction. For loose saturated granular
soils, the impact from heavy weights liquefies the soils and particles are rearranged to a
denser configuration. An impact of heavy falling weights generates ground vibrations
which limits its use in built up areas although there are techniques to minimize ground
vibrations such as trenches, air bags, etc.
6.2.1 Objective of improvement
The author is of the opinion that a site that requires remediation for liquefaction hazard
would also need improvement for soil bearing capacity. Considerations for a soil
improvement project should therefore include type of structure, type of foundation, bearing
capacity vis--vis depth and degree of envisaged improvement. Existing site conditions
must be carefully reviewed while selecting and designing a site improvement project. It
might happen that the planned site is undulating and requiring fill, or water logged, or may
contain vegetation and waste material of old disused structures. While preparing a site for
DDC, following must be ensured:
All fills must be of a material suitable to act as a granular blanket.
Old pavement and floors of the demolished buildings must be ripped and removed and
back filled with granular material to facilitate transfer of impact energy to deeper
layers.
Underlying services and trenches must be removed and suitably back filled.
Preferably, the site should be fairly levelled.
Multiple options must be considered to improve the site and finalized only after the detailed
investigation has been completed for example, for lighter structures or area requiring
shallower depth and degree of improvement, different option can be considered than the
heavier and sensitive structures. The layout of the planned structures may even be re-
adjusted to place critical structures on one segment and remaining ones on the other side.
The type of foundation must be considered in post improvement scenario, it might be
possible to place the foundation for lighter structures even on the surface thus reducing
overall cost of the project.

Fig. 3. Dynamic Compaction Equipment (heavy crane, tamper, release mechanism) and process


Fig. 4. Diagrammatic layout of primary and secondary pass during DDC
6.2.2 Project performance needs
Candidate civil engineering project requiring evaluation for liquefaction hazard may
include low and high rise buildings, life lines (water supply lines, gas pipe lines), bridges,
etc. Usually a large project comprises of number of structures with varying specifications,
depth of influence and performance needs. It is important to note that all structures within a
project may not need soil improvement therefore it would be important to identify
structures demanding improvement vis--vis their criticality and user needs.
6.2.3 Availability of technology and experience
DDC technique requires heavy duty cranes with lifting and drop arrangements for a tamper
from different heights. The tamper can be made of steel, concrete or concrete with steel
casing. The tamper should have air vents through its body to enhance impact energy. There
could be many unforeseen in such projects therefore an adequate reserve of essential crane
spares should be available at site.
6.2.4 Time and cost
Since this technique requires mobilization of very equipment to the site, like all other
projects a cost benefit analysis of DDC compared with other options such as avoidance, soil
replacement, deep soil mixing, grouting etc. is a pre-requisite for adoption of this technique.
LEGEND
Primary Pass
Secondary Pass
6 m
6 m
3 m
3 m
3 m
6 m
6 m
LEGEND
Primary Pass
Secondary Pass
Apart from the cost, DDC is a simple technique but is a time consuming process.
Availability of time must therefore be considered right at the outset of the project. Generally
this technique is suitable for large projects that do not have very tight timelines.
6.2.5 Preliminary DDC design to evaluate improvement capacity of soil
Determination of compactibility or improvement potential of soil deposit at a site needs no
emphasis and depends on factors such as particle size, liquid limit, relative density, type of
layers, and depth of ground water table. Although DDC is suitable for most of the soils
including municipal solid waste site, its best effectiveness has been reported in granular
soils. For cohesionless soils, depth of improvement is proportional to square root of the
energy per blow.
A preliminary DDC program can be designed from correlations proposed by Menard and
Broise (1975), Lukas (1986) given below as Equation 3 and 4 respectively. Attention must be
paid to the applied energy range over the project area for Equation 2 and n values for
different soils when using Equation 3.
D
max
= WH (3)
D
max
= n WH (4)
As a thumb rule, grid spacing of primary pass of DDC should not be less than the desired
depth of improvement with secondary (and tertiary if spacing of secondary pass is more
than 3 m) in between the primary pass. An ironing pass with an overlap of 1/3
rd
of the
tamper diameter is good enough to compact the upper soil layer. To finalize the number of
passes, number of drops at each impact point, a comprehensive experimental design is
required to record both degree and depth of improvement in horizontal and lateral extents.
Improvement and vibrations should be verified after pre-selected number of drops, author
suggests; firstly after half the number of drops, secondly after next 2/3
rd
number of drops
and finally after all the drops are completed. Keeping in view large number of variables
such as site and soil conditions, moisture conditions, depth of ground water table, etc., the
author is of the view that best mean to achieve desired improvement is through well
planned, executed, and monitored preliminary DDC program. Minimum area for of 24m Xx
24 m is recommended for preliminary DDC.
6.3 Experimental design for evaluation of improvement for field trial of preliminary
DDC program
Pre to post compaction improvement can be evaluated by comparing pre to post compaction
electrical resistivity profiles, SPT N-values, crater depth, quantum of backfill material, shear
wave velocity, etc. As far SPT is concerned, a value should be selected as the criteria to
terminate further drops. For alluvial soil deposits with 5 m desired depth of improvement,
an SPT value of 15 was selected as limit for discontinuing drops for the case history project.
There could be different ways to evaluate improvement during preliminary design and field
trial stage of the project. To observe improvement vis--vis number of drops or impact
energy and intensity of ground vibrations, author used a layout shown in Figure 5 which
proved very effective and can be used for future projects.

Evaluation of improvement under impact points, between the craters, and laterally away
from impact points is very important, as shown in Figures 6, 7, 8, respectively. With a
thoughtful evaluation methodology, actual improvement is possible to assess. For the
research project, the layout of test craters and location of boreholes was designed with a
view to keep the distance between pre and post compaction boreholes as minimum as
practically feasible. In this research, pre to post compactions SPT were performed within a

Fig. 5. Layout of Compaction Site, Test Craters, Vibration Measurement Points and Location
of ERT Survey
distance of 2 m. Total of 9 test craters; crater no. 1 through crater no. 9, were used in this
research. Total of 48 boreholes were drilled and 384 SPT performed to evaluate
improvement at different points. Detail of experimental design is given as, refer to Figure 5:
improvement after 5 blows was evaluated at crater no 3, 4, and 7
improvement after 10 blows was evaluated at crater no 2, 6, and 8
lateral improvement after 5 blows was evaluated around crater no. 1, 5 and 7
lateral improvement after 10 blows was evaluated around crater no 2, 6, and 8
Improvement at middle of adjacent craters was evaluated between crater no 1 & 2, crater
no 5 & 6 and crater no 8 & 9 after full scale compaction i.e. after primary, secondary and
ironing pass no compaction was carried out within 9 m of the test craters.
Locations of all boreholes has been referenced to the center of impact point and are
given in terms of tamper diameter D (D = 2.4 m)

Fig. 6. Location of boreholes for evaluation of improvement under impact point
6.4 Field trial and finalization of dynamic compaction program
Once the field trial is complete, improvement in depth, between impact points and in lateral
direction must be evaluated. To assess improvement in realistic manner, details of the field
trial of research project is explained in the ensuing paragraphs. SPT were performed two
0 D*
0.25 D*
0.5 D*
Side View of a C rater
Plan View of a C rater
P
r
e

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

P
o
s
t

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

P
o
s
t

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

* D = Tamper Diameter, 2.4 m
0 D*
0.25 D*
0.5 D*
Side View of a C rater
Plan View of a C rater
P
r
e

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

P
o
s
t

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

P
o
s
t

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e


weeks after full scale compaction. Improvement under impact point was evaluated at crater
# 2, crater # 6 and crater # 8 (upper halves). Improvement was also evaluated at the middle
of crater # 1 and crater # 2, crater # 5 and crater # 6, crater # 8 and crater # 9. Improvement
in lateral direction was measured only at two places, i.e., towards lower side of crater # 7
and upper side of crater # 8, refer to Figure 5.

Fig. 7. Location of boreholes for evaluation of improvement in lateral direction

Fig. 8. Location of boreholes for evaluation of improvement at the middle of adjacent craters
6.4.1 Improvement at crater # 1 (upper half) and middle of crater # 1 and crater # 2
Layout of BH for evaluation of improvement under impact point and at the middle of two
adjacent craters (crater # 1 and crater # 2) is shown in Figure 9. Pre-to-post compaction SPT
profile after full scale compaction for improvement under impact point at crater # 2 and at
the middle point of crater # 1 and crater # 2 is shown in Figure 10.
0 D*
1.75 D*
Plan View
of a C rater
P
r
e

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

P
o
s
t

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

P
o
s
t

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

Side View
of a C rater
2.4 D*
3 D*
0 D*
1.75 D*
Plan View
of a C rater
P
r
e

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

P
o
s
t

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

P
o
s
t

C
o
m
p
a
c
t
i
o
n

B
o
r
e
h
o
l
e

Side View
of a C rater
2.4 D*
3 D*
3 m
Post Compaction Borehole
3 m
Post Compaction Borehole
It is evident from Figure 10, that directly under impact point on upper side of crater # 2;
significant improvement was noted in top 3.5 m strata. Moderate improvement was
observed in strata between depths of 3.5 m to 5 m. Improvement was marginal in strata
depth of 5 m to 7 m. Improvement below 7 m depth remained insignificant. At the middle of
crater # 1 and crater # 2, improvement was slightly less than the improvement under impact
point however still the improvement was significant in upper 3.5 m strata. Improvement
below 3 m depth at the middle of these craters was almost same as that of improvement
under impact point of crater # 2.

Fig. 9. Layout of BH for Evaluation of Improvement after Full scale Compaction at Crater #
1 (Upper Half) and Middle of Crater # 1 and Crater # 2

Fig. 10. Pre-to-post Compaction N (Field) Comparison of Improvement at Crater # 1(Upper
Half) and Middle of Crater # 1 and Crater # 2
Crater # 1 Crater # 2
1.5 m
(0.625 D)
Post-compaction
Test Points
1.5 m
(0.625 D)
1.5 m
(0.625 D)
Post-compaction
Test Points
1.5 m
(0.625 D)
0
1
2
3
4
5
6
7
8
9
5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
N (Field)
D
e
p
t
h
,

m
Pre-comp, at 0.25 D
Post-comp, at 0.25 D
Post-comp, at 1.5 m (0.625 D)

Layout of BH for evaluation of improvement under impact point and at the middle of crater
# 5 and crater # 6 is shown in Figure 11. Pre-to-post compaction SPT profile after full scale
compaction for improvement under impact point at crater # 6 and at the middle point of
crater # 5 and crater # 6 is shown in figure 12.

Fig. 11. Layout of BH for Evaluation of Improvement after Full scale Compaction at Crater # 5
(Upper Half) and Middle of Crater # 5 and Crater # 6

Fig. 12. Pre-to-post Compaction N (Field) Comparison of Improvement at Crater # 5 (Upper
Half) and Middle of Crater # 5 and Crater # 6
Under impact point, improvement was significant in upper 4 m strata, moderate in strata
depth of 4 m to 5 m and marginal from 5 m to 6 m. At the middle of crater # 5 and crater # 6,
improvement was slightly less than the improvement under impact point however still the
improvement was significant in upper 3 m strata. Improvement below 3 m depth at the
Post-compaction
Test Points
1.5 m
(0.625 D)
1.5 m
(0.625 D)
Post-compaction
Test Points
1.5 m
(0.625 D)
1.5 m
(0.625 D)
0
1
2
3
4
5
6
7
8
9
5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
N (Field)
D
e
p
t
h
,

m
Pre-comp, at 0.25 D
middle of these craters was same as that of improvement under impact point of crater # 6
with slight variation at depths of 4 m and 8 m.
Layout of BH for evaluation of improvement under impact point and at the middle of crater
# 8 and crater # 9 is shown in Figure 13. Pre-to-post compaction SPT profile after full scale
compaction for improvement under impact point at crater # 8 and at the middle point of
crater # 8 and crater # 9 is shown in Figure 14.

Fig. 13. Layout of BH for Evaluation of Improvement after Full scale Compaction at Crater # 8
(Upper Half) and Middle of Crater # 8 and Crater # 9

Fig. 14. Pre-to-post Compaction N (Field) Comparison of Improvement at Crater # 8 and
Middle of Crater # 8 and Crater # 9 (Figure 5.17)
Post-compaction
Test Points
1.5 m
(0.625 D)
1.5 m
(0.625 D)
Post-compaction
Test Points
1.5 m
(0.625 D)
1.5 m
(0.625 D)
0
1
2
3
4
5
6
7
8
9
5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
N (Field)
D
e
p
t
h
,

m
Pre-comp, at 0.25 D

Directly under impact point on upper side of crater # 8; significant improvement was noted
in top 4 m strata. Moderate improvement was observed in strata between depths of 4 m to
5.5 m. Improvement was marginal in strata depth of 5.5 m to 7.5 m. Improvement below 7.5
m depth was insignificant. At the middle of crater # 8 and crater # 9, improvement was
slightly less than the improvement under impact point in the upper 2 m strata, still the
improvement was significant. Improvement below 2 m depth at the middle of these craters
was same as that of improvement under impact point of crater # 8.
6.4.4 Lateral improvement , lower side of crater # 7
Pre-to-post compaction N (Field) comparison after full scale compaction for improvement in
lateral direction on lower side of crater # 7 is shown in Figure 15. In lateral direction,
improvement was moderate in upper 2.5 m strata and marginal from 2.5 m to 3.5 m strata
from centre of impact point to distance of 1.75 D. Improvement was marginal in the upper
1.5 m strata from 1.75 D to 3 D from centre of impact point. Improvement below this depth
was insignificant.

Fig. 15. Pre-to-post Compaction N (Field) Comparison of Lateral Improvement at Lower
Side of Crater # 7
6.4.5 Lateral improvement, upper side of crater # 8
Pre-to-post compaction N (Field) comparison after full scale compaction for improvement in
lateral direction on upper side of crater # 8 is shown in Figure 16. Improvement was
moderate in upper 3 m strata and marginal from 3 m to 5 m strata from centre of impact
point to distance of 1.75 D. Improvement was marginal in the upper 1.5 m strata from 1.75 D
to 3 D from centre of impact point. Improvement below this depth was insignificant.
0
1
2
3
4
5
6
7
8
9
5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
N (Field)
D
e
p
t
h
,

m
Pre-comp, at 2.4 D of Crater # 7
Post-comp, at 3 D

Fig. 16. Pre-to-post Compaction N (Field) Comparison of Lateral Improvement at Upper
Side of Crater # 8
7. Quality controls
At the end of field trial, DDC program can be finalized with respect to number of passes,
number of drops, grid spacing of all the passes (primary, secondary, tertiary, ironing etc.),
and height of fall. During field trial, improvement monitoring mechanism is finalized in
terms of crater depth vis--vis number of drops, quantum of backfill material, and increase
in penetration resistance (SPT, CPT, or DCPT). Since it may not be possible to perform so
many penetration tests, more convenient means would still remain crater depth, and back
fill material. Author also used pre to post compaction electrical resistivity profiles as a tool
to monitor depth and degree of improvement. It is however emphasized that electrical
resistivity profile can be tricky in case of rains or increase in moisture conditions at the site.
In such circumstances, only an experienced geophysical or geotechnical engineer will be
able to interpret profiles. While pre to post compaction resistivity profiles can be compared
easily in 2-dimensional perspective, monitoring of improvement through crater depth and
back fill will require tabulated format which must be kept at site and filled in carefully.
Consequent to results of field trial of the designed compaction program, 5 cm settlement
between any two successive blows was selected as acceptable compaction criteria for
achievement of desired improvement. Final crater depths after 5 blows and 10 blows are
shown in Figure 17. Observations on crater depths are:
Almost same trend in crater depths has been observed at all craters
Crater depths increased with increase in no. of blows
Crater depths are more for initial 5 blows (around 96 cm) than for next 5 blows (around
37 cm)
At crater # 7 to crater # 9, crater depths are more for initial 5 blows and less for next 5
blows.
0
1
2
3
4
5
6
7
8
9
5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
N (Field)
D
e
p
t
h
,

m
Pre-comp, at 2.4 D of Crater # 8
Post-comp, at 3 D

Slightly more crater depths for 5 blows at crater # 7 to crater # 9 are because strata at
this location were relatively loose
Crater depths of research project have been plotted and compared with normalized crater
depths of DDC case histories of non-collapsible soils proposed by Mayne et al., 1984 as
shown in Figure 18. Comparison of crater depths indicates that crater depths of research
project are towards lower limit of DDC case histories. Since relationship of normalized
crater depths and no. of blows is nonlinear therefore best fit curve has been obtained using a
second order polynomial equation, given in Figure 18.
Relationship of crater depths and depth of improvement of research project was nearly
linear as shown in Figure 19; however the depth of improvement was more for larger crater
depths. A best fit line was obtained with linear equation (equation 5) where depth of
improvement was dependant variable and crater depth was independent variable. The
equation of the best fit line gives somewhat higher values of depth of improvement for
crater depths of 0.5 m to 1 m and lower depth of improvement at crater depths of more than
1.5m. In view of the lower depth of improvement of project after full scale compaction, the

Fig. 17. Crater Depth Measurements after 5 and 10 Blows
results of equation 5 would be towards lower bound of the expected range of improvement
of a similar compaction program under similar soil conditions. Since R
2
value of the
equation is 0.98, equation 51 shows a good correlation.
D = 3.7 d 0.11 (5)
Where,
D = Depth of improvement, m
d = depth of crater, m
25
35
45
55
65
75
85
95
105
115
125
135
0 1 2 3 4 5 6 7 8 9 10 11
Number of Drops
C
r
a
t
e
r

D
e
p
t
h
,

c
m
Crater # 2 (10 Blows)

Fig. 18. Comparison of Normalized Crater Depths Vs No. of Drops with the Range proposed
by Mayne et al., (1984)

Fig. 19. Plot of Crater Depths vs. Depth of Improvement
y = -0.0007x
2
+ 0.0132x + 0.005
R
2
= 0.9944
0.00
0.05
0.10
0.15
0.20
0 1 2 3 4 5 6 7 8 9 10
Number of Blows
N
o
r
m
a
l
i
z
e
d

C
r
a
t
e
r

D
e
p
t
h
Reaearch Case Study
Case Histories - Lower Limit
Case Histories, Upper limit
Poly. (Reaearch Case Study)
y = 3.65x - 0.11
R
2
= 0.98
0
1
2
3
4
5
6
0.00 0.50 1.00 1.50
Crater Depth, m
D
e
p
t
h

o
f

I
m
p
r
o
v
e
m
e
n
t
,

m


Fig. 20. ERT Profiles, Crater # 1 to Crater # 6
7.1 Evaluation of improvement from ERT profiles
Pre-to-post compaction ERT profiles of strata at crater # 1 to crater # 6 are shown in Figure
20. The profiles are based on relative densities of various layers in the strata therefore the
colour legend of pre and post-compaction profiles indicate different resistivity ranges. At
crater # 1 to crater # 6, over all pre-compaction resistivity range of 1.60 -m to 112 -m
increased to post-compaction range of 3.82 -m to 350 -m.
Pre-to-post compaction ERT profiles of strata at crater # 7 to crater # 9 are shown in Figure
21; the overall increase in resistivity was more in this case. Pre-compaction resistivity range
of 0.370 -m to 509 -m increased to post-compaction resistivity range of 6.34 -m to 1054
-m. Salient of pre-to-post compaction ERT profiles are:
Post-compaction ERT profiles confirm in general, the 5 m depth of improvement
evaluated from SPT
Relatively loose strata are also indicated at crater # 1, crater # 2 and crater # 8
Keeping in view the same nature of strata and SPT results at all other craters, presence
of loose strata could be attributed to some error in recording of resistivity data
a. Pre-compaction
b. Post-compaction
Crater # 6
Crater # 5 Crater # 4 Crater # 3 Crater # 2
Crater # 1
a. Pre-compaction
b. Post-compaction
Crater # 6
Crater # 1
a. Pre-compaction
b. Post-compaction
Crater # 6
Crater # 1
More no. of BH could have better explained the variation in resistivity in between test
craters

Fig. 21. ERT Profiles, Crater # 7 to Crater # 8
8. Ground vibrations
Ground vibrations from dynamic compaction can be dangerous to the adjacent structures
and therefore needs to be determined prior to compaction operation. Effect of vibrations on
various structures and human beings is shown in Table 3 and 4. Vibrations were predicted
using Scaled Distance Approach proposed by Mayne (1985). To assess attenuation effect of
this trench, vibrations were measured across the trench using Linear Variable
Displacement Transducer (LVDT)-an approach analogous to Impulse Response Function
approach proposed by Svinkin (1996a, 1997), refer to Figure 5. In a dynamic compaction
project in Pakistan, a 2 meter wide and 4 meter deep trench was excavated at the edge of the
compaction site to reduce vibrations to avoid any damage to a building some 50 meters
away from the site. Vibrations were measured across the trench at a distance of 25 m to
assess the attenuation effect of trench and compare with the value obtained from empirical
correlations. Vibrations were also measured on compaction site after the primary pass to
assess the effect of primary impacts on the magnitude of vibrations.
Linear Variable Displacement Transducers (LVDT) and accelerometer was used for
measurement of vibrations. These were installed on top of 1 inch dia, 1.25 m long -
Pre-compaction
Post-compaction
Pre-compaction
Post-compaction

deformed steel bar. The steel bar was drilled into ground with only the top surface visible at
the ground surface 2 inch thick foam was placed under the base of the LVDT assembly to
minimize inertia effect of the LVDT assembly. Layout of the test is shown in Figure 22 and
23. Data was recorded for 10 consecutive drops at a distance of 25 m from point of impact
across the trench and at distance of 25 m at place where primary pass was completed. The
data recorded was numerically differentiated to obtain peak particle velocity (PPV) and was
further differentiated to obtain acceleration.

Type of Structure and their Natural Frequencies
Safe level of Vibrations
(mm/s)
Reinforced or framed structures, industrial and heavy
commercial buildings at 4 Hz and above
50
Un-reinforced or light framed structures, residential or light
commercial type buildings at 4 Hz 15 Hz
15-20
Un-reinforced or light framed structures, residential or light
commercial type buildings at 15 Hz 40 Hz and above
20-50
Table 3. Effect of vibrations on structures (British Standard 7385: Part 2-1993).

Level of Vibrations
(mm / sec)
Effect of Vibration
0.1 not noticeable
0.15 nearly not noticeable
0.35 seldom noticeable
1.00 always noticeable
2.00 clearly noticeable
6.00 strongly noticeable
14.00 very strongly noticeable
17.8 severe noticeable
Table 4. Effect of vibrations on human beings (British Standard 7385).

Fig. 22. Layout of Vibration Measurement Points at 25 m Distance
25 m
17 m 2 m
6 m
Primary Pass Completed
T
r
e
n
c
h
Un-compacted Area
Sensors
25 m
17 m 2 m
6 m
T
r
e
n
c
h
Un-compacted Area
Sensors
25 m
17 m 2 m
6 m
T
r
e
n
c
h
Un-compacted Area
Sensors
25 m
17 m 2 m
6 m
T
r
e
n
c
h
Un-compacted Area
Sensors

Fig. 23. Installation arrangement of LVDT and accelerometer
The vibrations are estimated through empirical correlations or measured with the help of
instruments such as portable seismograph, accelerometers, velocity transducers, linear
variable displacement transducers (LVDT), etc. Mayne et al. (1984) proposed an empirical
relationship between PPV and inverse scaled distance to estimate minimum and maximum
range of PPV against various scaled distances and is shown in Figure 3. Inverse scaled
distance is square root of the compaction energy, WH divided by the distance, d from the
impact point. From Figure 24, a minimum PPV of 7 mm/sec and maximum PPV of 70
mm/sec were estimated at a distance of 25 m from point of impact.
Empirical Equation (6), proposed by Rollins and Kim 1992 [6] for estimation of the PPV, is
based on field monitoring data of several dynamic compaction projects. PPV in these
projects was measured using portable seismograph. The frequency of these vibrations
ranged from 5 - 40 Hz. At a distance of 25 m away from impact point, a PPV of 14.16
mm/sec is estimated using Eq. (6).

20
WH
PPV
d
1.03
(6)
Where,
PPV = Peak Particle Velocity, mm/sec
W = Weight, tons
H = Height of fall, m
d = Distance from point of impact, m

Accelerometer
LVDT
1.5 m Long Deformed Steel Bar
Flushed to Ground
Data Cables Connected to Laptop
Through Signal Amplifiers
Accelerometer
Accelerometer
LVDT
1.5 m Long Deformed Steel Bar
Flushed to Ground
Data Cables Connected to Laptop
Through Signal Amplifiers
Accelerometer


Fig. 24. Peak Particle Velocity (PPV) Vs Scaled Distance, Estimated PPV of Research Project
shown in Yellow Lines
8.1 Across isolation trench
Across isolation trench, maximum PPV of 22.5 mm / sec was measured at 25 m distance
from the point of impact as shown in Figure 25. This maximum velocity was obtained at 9
th

consecutive drop. PPV could not be measured at distance of 50 m point either due to very
weak signals received by sensors or some malfunctioning of sensors.
8.2 Over partially compacted site
Over partially compacted site (a place where primary pass of compaction was completed), a
maximum PPV of 14.1 mm / sec on 10
th
consecutive drop was calculated at a distance of 25
m away from point of impact as shown in Figure 25 and Table 5 and accelerogram is shown
in Figure 26 and 27. Vibrations at a distance of 50 m away from impact point were very low
and are not mentioned. The best fit curve is given by exponential equation. Increase in PPV
at 25 m point across trench with each successive drop indicates that the PPV increased with
increase in soil density due to DDC. An overall decrease in PPV after completion of first
pass indicates that the initial pass created several denser / loose soil mediums. The primary
impact points at grid spacing of 6 m centre to centre with un-compacted soil in between
caused almost 50 % attenuation of PPV.


Fig. 25. Measured Ground Vibrations

No of
Blow
PPV Across Trench
(at 25 m Distance)
mm/sec

PPV Over Partially
Compacted Site
(at 25 m Distance)
mm/sec
1 13.3 10.3
2 14.9 11.3
3 17.1 12.9
4 19.1 13.2
5 20.3 13.1
6 20.6 14
7 21.3 13.8
8 21.4 13.6
9 22.5 13.3
10 22.4 14.1
Table 5. Range of PPV for Various Numbers of Blows

Fig. 26. Accelerograms for successive blows
y = 0.0559e
0.2278x
R
2
= 0.9786
0
2
4
6
8
10
12
0 2.5 5 7.5 10 12.5 15 17.5 20 22.5 25
PPV, mm / sec
N
o
.

o
f

B
l
o
w
s
Vibrations at 25 m distance
Across Trench
Vibrations at 25 m Distance
over Partially Compacted
Site
Expon. (Vibrations at 25 m
distance Across Trench)


Fig. 27. Accelrogram of PPV at 10th blow
DDC technique can remediate liquefaction potential and enhance bearing capacity up to
desired limits at desired depths. Factor of safety evaluated after DDC is given in Table 6
which shows an increased value of factor of safety at three different depths.

Depth
(m)
(N
1
)
60

L
CSR
'
o
v

(kN/m
3
)
, cyc L
.
cyc

FS
L

3.5 23 0.24 71.43 17.14 7.46 2.3
5.5 14 0.17 82.12 13.96 9.46 1.5
7.5 12 0.15 92.81 13.92 11.41 1.2
Table 6. Cyclic Shear Resistance of Soil after Dynamic Compaction and Factor of Safety (FS
L
)
9. Conclusion and recommendations
In the light of research conducted for remediation of liquefaction hazard, improvement
under impact points, lateral direction after 5 blows, 10 blows and full scale compaction,
following conclusions are presented:
directly under impact points, the bearing capacity of soil improved to 160 kPa upto a
depth of 5 m.
maximum improvement under impact points occurred at depths from 2 m to 4 m.
sharp decrease in improvement is observed below 4 m depth.
improvement was negligible below 6.5 m depth.
with the increase in no. of drops from 5 blows to 10 blows, the degree of improvement
also increased from maximum 35 blows to 45 blows.

Fig. 28. Suggested Procedures for DDC Project

increase in no. of blows from 5 to 10, had negligible effect on degree of improvement
below 6.5 m depth.
in upper 2 m of strata, improvement at the middle of any two adjacent impact points
was comparatively less than improvement under the impact point.
From results of soil improvement project of alluvial soils, it is concluded that deep dynamic
compaction technique can remediate liquefaction potential and enhance bearing capacity
upto desired limits at desired depths. The compacted layers at top of strata and reduced
grid spacing can significantly reduce the overall depth of improvement. In the light of this
research project, a flow chart in Figure 28 is suggested for more efficient and effective DDC
program.
10. References
Bowles, E. Joseph [1997]. Foundation Analysis and Design. McGraw-Hill Book Company,
Inc., New York.
Lee, F.H., and Gu, Q., [2004]. Method for Estimating Dynamic Compaction Effects on
Sand. J. of Geotch. and Envmt. Eng., ASCE. No. 130(2), pp. 139-152.
Lukas, R.G. [1986]. Dynamic Compaction for Highway Construction. FHWA, US DOT, Report
No. FHWA/RD-86/133.
Mayne, P.W., Jones J.W., and Dumas, J.C., [1986]. Ground Response to Dynamic
Compaction.J. of Geotech. Eng., ASCE, No. 110(6), pp. 757-774.
Menard, L., and Broise, Y. [1976]. Theoretical and Practical Aspects of Dynamic Consolidation.
Ground Treatment by Deep Compaction, The Institute of Engineers, London. pp. 3-
18.
Meyerhof, G. G., [1976]. Bearing Capacity and Settlement of Pile Foundation. J. of
Geotech. Eng. Div. Am. Soc. of Civ. Eng., No. 102(3), pp. 195-228.
Oshima, A., and Takada, N. [1998]. Evaluation of Compacted Area of Heavy Tamping by
Cone Point Resistance. Proc. Int. Conf. Centrifuge 98., Kimura, Kusakabe, and
Takemurs, eds., Balkema, Rotterdam, pp. 813-818.
Rollins Kyle, M., and Kim, Ji Hyoung [1994]. US Experience with Dynamic Compaction of
Collapsible Soils, in In-Situ Deep Soil Improvement, Geotechnical Special Publication,
Am. Soc. of Civ. Eng., New York, pp. 27-43.
Khalid, Muhammad Mumtaz (1995). An investigation into high water table problem at
Attock District, an MSc Thesis, National University of Science and Technology
(NUST), Pakistan.
Kramer, L., Steven (1996). "Geotechnical earthquake engineering." Prentice Hall Series. 351-
354, 369-376.
Kyle M. Rollins (1994). In-Situ Deep Soil Improvement, Geotechnical Special Publication No.
45, 37-38.
M/S High Tech, (2004). Report on geotechnical engineering investigation for Attock District,
Pakistan.
Seed, H.B., Idriss, I.M. (1971). Simplified procedures for evaluating soil liquefaction
potentials, J. of Soil Mechanics and Foundation, ASCE, Vol. 107, 12491274.
Seed, H.B., Idriss, I.M., and Arango (1983). "Evaluation of liquefaction potential using field
performance data." Journal of Geotechnical Engineering, ASCE. 109, 458-482.
Ali, Sarfraz. Evaluation of Deep Dynamic Compaction Technique. Thesis in Masters Degree
in Geotechnical Engineering, National University of Science and Technology
(NUST), Islamabad, Pakistan, 2007.
Wood R.D. Vibration Screening with Wave Barriers. A look Back for Future Geotechnics,
1999, 325-348.
Svinkin, R. Mark. Prediction and Calculation of Ground Vibrations. Proceedings of the 24
th

Annual Members Conference of the Deep Foundation Institute, Dearborn,
Michigan, 1999.
Menard, L. and Broise, Y. Theoretical and Practical Aspects of Dynamic Consolidation.
Geotechnique, 25, No. 1, 3-18, 1975.
Mayne et al. Ground Response to Dynamic Compaction. Journal of Geotechnical Engineering,
1984, 110. No. 6, 345-349.
Rollins, Kyle M. and Kim Ji Hyoung. US Experience with Dynamic Compaction of Collapsible
Soils. Geotechnical Special Publication No. 1994, 26-43.
5
Simplified Analyses of Dynamic Pile
Response Subjected to Soil Liquefaction
and Lateral Spread Effects
Lin Bor-Shiun
Geotechnical Enigineering Research Center, Sinotech Engineering Consultants,
Taiwan
1. Introduction
Pore water pressures greater than effective soil stress and subsequent liquefaction are
known to occur in saturated sand deposits subjected to earthquake excitations. Liquefaction
of soils can result in a reduction of soil strength and yields large settlement via lateral
spreading. For superstructures supported on pile foundations embedded in such soils,
these effects can be devastating. For example, the 1964 Niigata earthquake in Japan
damaged the foundation piles under one of the piers of the 12 spans, 207 meter long
Showa Bridge. After the earthquake, an excavation survey of damaged piles indicated that
bending failure occurred due to the lateral spreading of river bed soils (Hamada, 1992).
Similarly, in the 1994, when Northridge earthquake occurs, river bank areas between
Santa Clarita and Fillmore, Highway 23 crosses over the Santa Clara River, where sand
boils were observed near a bridge pier. Cracks induced by lateral spreading were found
approximately 4.5 m away from the pier (Stephen et al., 2002). Afterward, in the 1995
Kobe Earthquake, quay walls along the coastline of Kobe moved up to several meters
toward the sea as a result of lateral spreading (Tokimatsu and Asaka, 1998). Some papers
also discuss liquefaction-induced lateral spread under the foundations of long-span
bridges subjected to spatially-varying ground motions. (Abbas and Manohar 2002; Zerva
and Zervas, 2002; Wang et al., 2004; Zerva 2009). More recent devastating earthquakes
such as the March 2011 Tohoku earthquake in Japan and the January 2010 Haiti
earthquake can be found and reported by EERI (Earthquake Engineering Research
Institute) and USGS (United States Geological Survey).
Recent research has focused on understanding the transfer of forces between a pile and the
surrounding layered soil during liquefaction (Hamada, 1992; Meyersohn, 1994; Tokimatsu,
2003, Bhatachaya et al., 20022004; Jefferies and Been, 2006). Excavation surveys by
Hamada (1992) clearly showed that foundation piles are especially susceptible to damage at
the interface between liquefied and non-liquefied layers. This observation was also verified
by Meyersohn (1994) and Lin et al. (2005) with static numerical techniques.
In terms of static design for pile foundations, the current mechanism of failure assumes that
the soil pushes the pile. The Japanese Road Association Code (JRA, 1996) has incorporated
this concept. The code advises civil engineers that the non-liquefied layer acts passive

114
pressure on the piles and liquefied layer offers thirty percents of overburden pressure when
designing piles against bending failure due to lateral spread. Other codes such as USA code
(NEHRP, 2000) and Eurocode 8, part 5 (1998) also have specifications about the problem
(Bhattacharya et al., 2005). In the other hand, Tokimatsu (2003) investigated that the
equivalent earth pressure acting on the pile during liquefaction in shaking table tests can be
defined as the seismic passive pressures subtracting the seismic active pressures. This
concept was also verified with the centrifuge tests by Haigh and Madabhushi (2005) and
Madabhushi et al (2010). In a design process, engineers need the limit states to define the
serviceability of members according to the safety of performances to structures (Priestley et
al., 1996; Kramer and Algamal, 2001).
When piles are subjected to lateral spreading, lateral forces are exerted directly on the
embedded depth of piles within liquefied layer. There are generally two methods to analyze
this phenomenon. The first method is called the Force-based method. Using an explicit
numerical procedure, earth pressure is applied onto the piles based on a viscous flow model
(Chaudhuri et al. 1995; Hamada and Wakamatsu, 1998; Lin et al., 2010). In order to
effectively use the force-based method, several soil parameters must be known. Also, the
force-based method can account for the effect of soil topography. In the second method,
known as the Displacement-based method, observed or computed lateral ground
displacements are transmitted by theoretical soil springs on the whole pile system
(Tokimastu and Asaka, 1998; Ishihara, 2003; Chang and Lin, 2003; Cubrinovski and Ishihara,
2004; Preitely et al., 2006). The second method has several advantages such as being able to
choose a soil spring model that matches the complexity of the soil stratum. Also, nonlinear
material effects can be considered.
This chapter investigates pile response to loading caused by liquefaction using the
EQWEAP (Earthquake Wave Equation Analysis for Pile) numerical analysis procedure
(Chang and Lin, 2003; Chang and Lin, 2006; Lin et al., 2010). Both a displacement and forced
based form of EQWEAP are used. Methodology and case study comparisons with results of
these two procedures are presented separately. The chapter ends with a final synthesis of
observations and conclusions drawn from the two methods.
2. Methodology
2.1 Brief overview
The Winkler foundation model is often used in analyzing the deformation behaviors of the
pile foundations. For solutions of the dynamic Winkler foundation model, or the so-called
beam on dynamic Winkler foundation (BDWF) model, the wave equation analysis, initially
proposed by Smith (1960), has been suggested for the driven piles. To make the wave
equation analysis more accessible at the time-domain, the author (Chang and Yeh, 1999;
Chang et al., 2000; Chang and Lin, 2003) has suggested a finite difference solution for the
deformations of single piles under superficial loads. Such formulations can be extended for
the case where the piles are subjected to seismic ground shaking. Prior to analysis of the pile
system shown in Figure 1, the seismic induced free-field excitation behavior of the soil
stratum needs to be obtained. A description of the soil stratum behavior during excitation
provides a one-dimensional soil amplification solution for the site. For the site of interest,
time dependent earthquake records are used with the modified M-O method to calculate the
Simplified Analyses of Dynamic Pile Response
Subjected to Soil Liquefaction and Lateral Spread Effects

115
dynamic earth pressure coefficients (Zhang et al., 1998). Liquefaction potential at various
depths of the site is evaluated for the limited pore water pressure ratio (Tokimatsu and
Yoshimi, 1983). and numerical methods such as the finite element method or the mechanical
model which models the discrete model of the pile system (Bathe, 1982) .

Fig. 1. Discrete system of the single pile
The earthquake motions can be decomposed into vertical and horizontal components. Pore
water pressure effects are accounted for using an excess pore water pressure model. Soil
deformation, seismic loading, resistance, damping and the inertia forces of the soil relative
to time are applied to the pile segments and used to solve for the corresponding pile
displacements. Figure 2 shows the layout of the described superposition procedure.

WEA for Seismic
Motions
Free Field Response

Fig. 2. Superposition of the free-field analysis and WEA

116
Formulations can be derived from the wave equations of the piles. Analysis of the
foundations can be performed assuming unloaded or time dependant sustained loading
conditions. With proper boundary conditions at the pile head, interactions of the structural
system can be modeled. The above procedure is known as EQWEAP, which mainly
concerns the nonlinear behaviour of liquefied soil induced permanent ground displacement
rather than piles. Figure 3 illustrates the flow chart for the EQWEAP

Fig. 3. Flowchart summarizing the numerical procedures of the analysis
2.2 EQWEAP: Displacement-based method
2.2.1 Wave equation of pile foundations concerning soil liquefaction
To make the wave equation analysis of the deformations of single piles under superficial
loads more accessible in the time-domain, several authors (Chang and Yeh, 1999; Chang et
al., 2000; Chang and Lin, 2003) have suggested a finite difference solution. Such
formulations can be extended to the case where the piles are subjected to seismic ground
shaking. Assuming force equilibrium, the governing differential equations of the pile
segment exciting laterally can be written as:

4 2 2
4 2 2
( , ) ( , ) ( , )
( , )
( , ) 0
p p p
p p p p x s s
u x t u x t u x t
u x t
E I A P C K u x t
t x t x
c c c
c
+ + + + =
c c c c
(1)
where ( )
p s
u u u = = relative pile displacements,
p
u = absolute pile displacements,
s
u =the
absolute soil displacements,
p
E =Young's modulus of the pile;
p
I = moment of inertia of the
pile,
p
= uniform density of the pile,
p
A = cross-section area of the pile;
x
P = superstructure
loads,
s
C and
s
K = damping coefficient and stiffness of the soils along the pile, and x is
ordinate variable, and t represents for time.. For earthquake loading transmitting from the
soils, Eq. (1) can be expanded using the central difference formula as shown below:

117
( )
( 2, ) (4 ) ( 1, )
(6 2 2 ) ( , )
1
, 1
(4 ) ( 1, ) ( 2, )
( ) ( , 1) [ ( , 1) ( , 1)] ( , )
p p
p
p
p p
p s s s
u i j B u i j
A B D u i j
u i j
B u i j u i j A C
A C u i j C u i j u i j Du i j
+ + + (
(
+
(
+ =
(
+ +
(
(
+ + +

(2)
where
4
2
p p
p p
A x
A
E I t
A
=
A

2
x
p p
P x
B
E I
A
=
4
2
s
p p
C x
C
tE I
A
=
A

4
s
p p
K x
D
E I
A
=
Eq. (2) indicates that the absolute pile displacements under the earthquake excitations can
be solved directly from the absolute displacements of the adjacent soil. A major advantage
of this method is that the matrix analysis is not required in solving for the pile
deformations. One can simply use a free-field analysis to obtain the liquefied soil
displacements using the excess pore water pressure model(as described in Section 2.22)
and then substitute the displacements into Eq. (2) to obtain the desired solutions. This is
similar to those suggested in the multiple-step analysis of the soil-structure interaction
problems. In addition, equations describing the lateral excitations of the highest and
lowest elements of the pile should be modified using proper boundary conditions listed
as follows.
Top of the pile:
a. Free head:

( )
3
3
,
p
t
p p
y x t
P
E I x
c
=
c
;
( )
2
2
,
p
t
p p
y x t
M
E I x
c
=
c
(3)
b. Fixed head:

( )
3
3
,
p
t
p p
y x t
P
E I x
c
=
c
;
( ) ,
0
p
y x t
x
c
=
c
(4)
At the tip of the pile:

( )
2
2
,
0
p
y x t
x
c
=
c
;
( )
3
3
, ( , )
0
p p
p p X
y x t y x t
E I P
x x
c c
+ =
c c
(5)
where
t
M and
t
P are the external moment and load applied at the pile head. The discrete
forms of these equations can be derived with the central difference schemes. Detailed
derivations can be found in Lin (2006).
2.2.2 Soil stiffness and damping
For discrete models of the various soil types (sand, clay, etc.), both of the stress-
displacement curves (t-z, q-z and p-y equations) and the Novaks dynamic impedance
functions are used popularly in practice. The former, which is established empirically from
the in situ pile load tests, can be used for substantial load applied slowly. The later, initially

118
suggested by Novak (1972, 1974 and 1977) for the soils around the piles subjected to small
steady-state vibrations, is able to capture the dynamic characteristics of the soil resistances
and energy dissipations. Soil displacements close to a pile subjected to dynamic loading are
nonlinear (Prakash and Puri, 1988; Nogami et al, 1992; Boulanger et al., 1999; El Naggar and
Bently, 2000). El Naggar and Bently (2000) used a nonlinear soil model that incorporated a
p-y curve approach to predict dynamic lateral response of piles to soil movement. The
computed responses were found compatible with the results of the statnamic pile test. The
nonlinear stiffness of the p-y equations is adapted in this investigation. The corresponding
soil stiffness is described as below:

2
0
1
2 2
2 0 0 1
1 1 0
8 (1 )(3 4 ) 1
(3 4 ) 1 ln 1
m
NL
r
G v v
r
k
r r r
v
r r r
t
(
| |
(
+
|
(
\ .

=
(
| | | | | |
(
+ +
| | |
(
\ . \ . \ .

(6)
where
0
r is the pile radius,
1
r is the outer radius of the inner zone, v is Poissons ratio of
the soil stratum, and
m
G is the modified shear modulus of the soils. A parametric study
shows that a ratio
0 1
/ r r of 1.1-2.0 yields the best agreement.
For the damper, a transformed damping model is used. Equivalent damping ratios, D, of the
soils at steady-state excitations are first computed from the Novaks dynamic impedance
functions, K* where
K* = K
real
+ i K
imag
= K(e) + i eC(e) ~ K(e) (1+ 2iD) (7)

( )
( )
2 ( )
C
D
K
e e
e
~ (8)
In the above equations, K(e) and C(e)are the frequency-dependent stiffness and damping
coefficient of the impedance. For simplicity, the computed damping ratios are incorporated
with the static stiffness Kst to model the kinematics of the soil. The revised damping
coefficient c(e)can be written as:

( ) 2 / st c DK e e ~
(9)
Decomposing the actual load-time history into a series of small impulses, the damping
coefficient c(t) can be obtained by integrating a damping function c(t) to a set of unit
impulses of the actual load-time history. Knowing that D=C(e)e/2K(e), the associated
geometric damping ratios can be computed. Modeling the values of D(e)/e and assuming
that they are symmetric with respect to the ordinate, a mathematical expression of the
damping can be written as:

( )
B
s c t AK t
=
(10)
where A and B are the model parameters (Chang and Yeh, 1999; Chang and Lin, 2003).

119
2.2.3 Modeling soil liquefaction
Soils affected by induced pore-water pressure reduce the lateral resistance of the piles. This
study utilized an excess pore water empirical model to complete effective stress analysis
(Martin et al., 1975; Finn et al, 1977; Finn and Thavaraj, 2001), and obtain free-field motions
under liquefaction. Kim (2003) successfully predicted the excess pore-water pressure
resulting in soils subjected to earthquake shaking by verifying results with laboratory tests.
This model can be divided into undrained conditions and drained conditions as follows:
a. Undrained condition:

1
vd
w
p
r w
u
n
E K
c A
A =
| |
+
|
\ .
(11)
where
w
u A = an increase in pore water pressure;
vd
c A = an increment in volumetric strain;
r
E = one dimensional rebound modulus at an effective stress ( '
v
o );
p
n = porosity, and
w
K =
bulk modulus of water.
For saturated sand
w r
K E >> and therefore

w r vd
u E c A = A (12)
According to simple shear test, the volumetric strain increment (
vd
c A ) is a function of the total
accumulated volumetric strain (
vd
c ) and the shear strain ( ). The relationship is given by

2
3 [ 1]
[ ] 1 2 [ 1]
4 [ 1]
( )
vd i
vd i vd i
vd i
C
C C
C
c
c c
c

A = +
+
(13)

[ ] [ ]
1
n
vd n vd i
i
c c
=
= A
(14)
where [i] = ith time step or cycle; and
1
C ,
2
C ,
3
C , and
4
C are constants depending on the
soil type and relative density. An analytical expression for rebound modulus (
r
E ) at any
effective stress level ( '
v
o ) is given by

' 1
'
0
2
( )
( )
m
m n v
r v
E
mk
o
o
=
(15)
where
'
0 v
o is initial value of the effective stress; and
2
k , m and n are experimental
constants for the given sand.
b. Drained condition:
If the saturated sand layer can drain during liquefaction, there will be simultaneous
generation and dissipation of pore water pressure (Sneddon, 1957; Finn et al. 1977). Thus,
the distribution of pore-water pressure at time (t) is given by

120
( )
w vd
r r
w
u k u
E E
t z r z t
c c c c c
= +
c c c c
(16)
where u = the pore-water pressure; z = the corresponding depth; and k = the permeability ;
and
w
r is the unit weight of water. Before conducting the free-field analysis, the adequate
shear modulus (Seed and Idriss, 1970) may be determined from the following equation

0.5
2
1000 ( ' )
m
G K o = (17)
where
2
K is a parameter that varies with shear strain and '
m
o is the mean effective stress.
Pore water pressure will increase during shaking and leads to a decrease of effective stress.
In some situations, pore-water pressure equals overburden stress in sand deposits and may
liquefy. The initial shear modulus can be calculated from the initial effective stress. Then, G
is modified due to the shear strain and pore water pressure under liquefaction. The
modified value is substituted in place of the former one and convergence of solutions is
obtained using an iterative manner.
In addition, to avoid over-predicting the excess pore water pressure and ensure
compatibility with practical observations, it is suggested to use the pore water pressure ratio
(
u
r ) to accurately control soil liquefaction levels (Lee and Albaisa, 1974; DeAlba et al., 1976;
Tokimatsu and Yoshimi, 1983). The equation is given by

1
1
1 1
sin (2 1)
2
u L
r F
o|
t

= + (18)
where , are the experienced constants, and
L
F is the safety factor of liquefaction. In
order to use the above formulas, the liquefaction potential analysis of the site needs to be
conducted prior to the analysis.
2.2.4 Free field analysis
The one-dimensional seismic excitations of soils onto the piles are computed from a free-
field response analysis for the site of interest. Such an analysis can be conducted using the
finite element technique, or be simply solved for using the 1-D wave propagation model and
the lumped mass analysis. For simplicity, the lumped mass model is selected. To analyze the
equations of motion of the soil layer under the earthquake excitations, the relative
deformations of the structural system are obtained with the base accelerations induced by
the earthquake. Base motions of the site are obtained by modifying the seismic accelerogram
recorded at the ground surface of that site. This is done simply by obtaining the frequency-
spectrum of the accelerogram, and then multiplying it with the analytical transfer function
represented for the ratios of the accelerations occurring at the base (bedrock) and those at
the ground surface of that site (Roesset, 1977). This computation would complete a
frequency-domain convolution and prepare a base-acceleration spectrum to solve for the
corresponding accelerogram. To have consistent results for a specific site, one must be very
cautionous about the wave velocities and the thickness of the soil layers used in the
analyses. Crosschecks are required for vertical and horizontal excitations to ensure that the

121
analytic parameters are rational. Notice that the discrete solutions of the wave equations are
in terms of the displacements only. To obtain the timedisplacement history of the soils, a
baseline correction procedure (Kramer, 1996) is suggested to eliminate the integral offsets of
the velocities and displacements appearing after the quake excitations. The responses of the
free-field using the above procedure have been checked with the solutions of FEM as shown
in Figure 4. Using this simplified model just be only computed one-way ground response
depending on the inputted seismic motions. And, despite the simplicity of the geometry, an
exact solution of the full model, and a detailed analysis of the phenomenon, have not
perfectly been achieve (Schanz and Cheng, 2000).

-16
-8
0
8
16
24
D
i
s
p
l
a
c
e
m
e
n
t

(

c
m

)
0 10 20 30 40
Time ( sec )
-0.5
0.0
0.5
R
e
l
a
t
i
v
e

D
i
s
p
l
a
c
e
m
e
n
t

(

c
m

)
Kobe (NS)
Kobe (NS)
21.339
18.888
0.159
0.366
2x2 Pile Groups
ABAQUS
WAVE EQUATION

Fig. 4. Comparison of numerical results from WEA and FEM
2.3 EQWEAP: Force-based method
2.3.1 Wave equation of pile foundations concerning lateral spread
The wave equation describing a single pile under lateral loads can be derived based on a
force equilibrium of the pile segments shown in Figure 1 as follows,

( ) ( ) ( )
( )
4 2 2
4 2 2
, , ,
,
x
u x t u x t u x t
EI A P P x t
x t x
c c c
+ + =
c c c
(19)
where u is the lateral pile displacement relative to the soil, E is the Young's modulus of the
pile, I is the moment inertia of the pile, is mass density of the pile, A is the cross-sectional
area of the pile,
x
P are the superstructure loads, ( ) , P x t is the time-dependent loading due to
laterally spreading at various depths, x is ordinate variable, and t represents for time. Using
explicit finite difference schemes, the discrete form of Eq. (19) can be written as

122

( )
( ) ( ) ( )
( ) ( )
( ) ( ) ( )
( )
1
1 1
1 1
1 1
2, 4 1,
2 2 6 ,
1
, 1
4 1, 2,
, 1
u i j B u i j
A B u i j
u i j
A B u i j u i j
A u i j C
( + + +
(
+ +
(
+ =
(
+
(
(
+

(20)
where
4
1
2
A x
A
EI t
A
=
A
;
2
1
x
P x
B
EI
A
=
;
4
( , ) P x t x
EI
A
.
For the initial condition, ( ) , u i j and ( ) , 1 u i j are set to zero. Equation 20 can only calculate
the responses of piles under lateral loads along the lenfth of the pile. The head and tip can not
be solved for. With proper boundary conditions (see Eq. 3~4), the other equations can then be
derived. While the liquefaction-induced dynamic earthquake pressures are computed, the pile
responses at various depths can be solved through the above formulations.
2.3.2 Dynamic earth pressure
Since Okabe (1926) and Mononobe and Matsuo (1929) introduced the concept of dynamic
lateral pressure, many reports and practical works have been conducted in this manner
(Ishibashi and Fang, 1987; Richard et al., 1990; Ishibahi et al., 1994; Budhu and Al-karni,
1993; Richard et al., 1993; Soubra and Regenass, 2000). Tokimatsu (1999, 2003) and Uchida
and Tokimatsu (2005) determened several factors that affect the response of a pile in
saturated sand by using a shaking table tests. They suggested that the total earth pressure
acting on the foundation, when neglecting the friction between foundation and soil (see Fig.
5), is define as:

E EP EA
P P P Q F = =
(21)
where
E
P is total earth pressure,
EP
P and
EA
P are earth pressures on the active and passive
sides, Q is shear force at the pile head, and F is total inertial force from the superstructure
EA
P
EP
P
1
F
2
F
Q
F: Inertial Force(=F
1
+ F
2
)
P
EA
: Active Earth Pressure
P
EP
: Passive Earth Pressure
Q: Shear Force

Fig. 5. Schematic layout of forces acting on Foundation (from Uchida and Tokimatsu, 2005)

123
and foundation. In addition, Haigh and Madabhushi (2005) have verified that the adjacent
stresses of single piles subjected to lateral spreading forces would range between the active
state and the passive state through centrifuge modeling.
Based on the Mononobe-Okabe method, Zhang et al. (1998) successfully derivates the time-
dependent coefficients of earth pressure under active and passive states that involves the
motions of soils and foundations. One can also modify the plane strain model of soil wedge
to extend it to be three dimensional analysis. The descriptions and formulations of the
coefficients of active and passive earth pressures are referred to Zhang et al. (1998).
2.3.3 Modeling lateral spread
For lateral spread induced by liquefaction, the soil properties such as the unit weights and
the friction angles of the soils could be corrected based on the calculated pore water
pressure ratios. There are two ways depicting the weakness of soils during liquefaction
(Matsuzawa et al. 1985; Ebeling and Morrison, 1993). Those equations are given by
' (1 )
s s u
r = (22)

1
' tan [(1 )tan ']
eff u
r | |
= (23)
where '
s
is the unit weight of the soil,
s
is the effective unit weight of the soil, ' | is the
friction angle of the soil, and '
eff
| is the effective friction angle of the soil.

0
Z
1
Z
2
Z
i
Z

Fig. 6. Distribution of earth pressure along a pile

124
Figure 6 illustrates the distributions of earth pressures along the pile with the discrete blocks
and nodes. According to the geometry of pile (see figure 7) and Eq. (24), the lateral forces at
various depths are determined by
( ' ) ( )
E s E
i
P ZK B = (24)
where
i
Z is the corresponding depth of node,
E
K is the equivalent dynamic coefficients of
earth pressure (i.e.
E EP EA
K K K = ), and B is the loaded width of the pile body ( /2 d t = ,
where d is the pile diameter).

EP
P
EA
P
2 / d B t =

Fig. 7. The loaded width of the pile body due to lateral spreading
3. Practical simulation
In the following section, two case studies are presented, one of which focuses on pile
foundation damagess caused by the Niigata earthquake in Japan (Hamada, 1992) and the
other which focuses on foundation pile cases damaged during the 1995 Kobe earthquake.
The Niigata earthquake case study utilize the displacement-based EQWEAP method, in
which the free-field and the wave equation analysis are both performed to calculate the
dynamic responses of piles under liquefaction. In The Kobe earthquake case studies, the
force-based EQWEAP method is utilized to assume lateral flow induced forces on the piles.
Dynamic earth pressures caused by lateral spreading of the liquefied layers are first
generated and used to model forces exerted on the piles where the deformations of piles
occur. These results show the pile failure pattern validate the applied methodology.
3.1 Case study: Pile damages due to soil liquefaction
The Niigata Family Court House was a four-story building located on the left bank of the
Shinano River. The building was supported on a concrete pile foundation (Figure 8) each
pile of the foundation having a a diameter of 35 cm and length of 6 to 9m. During the
earthquake, the pile foundations were damaged by liquefaction-induced ground
displacement. Excavation surveys showed that two piles (No.1 pile and No.2 pile) had
severe cracks (Figure 9). They were conjecturally crushed by excessive bending moments at

125
the interface between liquefied and non-liquefied layers as shown in Figure 9. According to
aerial photographs of the area, the permanent ground displacement in the vicinity of
building moved approximately 1.1m and the maximum displacement of No.1 pile and No.2
pile were respectively 50 cm and 70cm. For simplification, the entire soil system could be
assumed as an upper layer and a lower layer. The upper layer from the ground surface to
the depth of 8m is classified as medium-dense sand. The lower layer from the depth of 8 to
11m is classified as dense sand. The time history of earthquake record adopted the NS-
component of the 1964 Niigata Earthquake as illustrated in Figure 10.

Fig. 8. Footing and foundation beams of Niigata Family Court House (from Hamada, 1992)

Fig. 9. Damage to piles and SPT-N values in sit u (from Hamada, 1992)

126

0 20 40 60 80 100
Time (sec)
-0.20
0.00
0.20
A
c
c
e
l
e
r
a
t
i
o
n

(
g
)
0.134
Niigata - Ns Component

Fig. 10. Time history of Niigata Earthquake (NS Component)
The initial shear modulus of the soils at the any depth can be calculated by Eq. (17). The
distribution of shear modulus is similar to the hyperbolic form observed in gibson soils and
increases with the depth. The determination of pore water ratio pressure (
u
r ) and reduction
factors (
E
D ) versus depth can be estimated by the liquefaction potential method suggested
by Tokimatsu and Yoshimi (1983) with Eq. (18) for various levels of liquefaction. Moreover,
one can conduct EQWEAP analysis to obtain the liquefied free-field response considering
the effect of pore water. The excess pore pressure ratios at different depths are shown in
Figure 11. It was found that the soil layer reached a liquefied state gradually after 2.8
seconds.
Figure 12 shows the time histories of ground motions. The maximum displacement of the
ground which takes place at the surface is 47.3 cm at about 10 seconds. The liquefied layer
( 100%
u
r = ) ranging between the depths of 2 m to 8m displaces by 30 cm to 45 cm (see
figure 12). The displacements reduce to about 3 cm below the liquefied layer for
14 ~ 45%
u
r = as shown in Figure 12. Figure 13 indicate the maximum displacements of
piles at various depths from wave equation analysis. Based on the results form Figure 13,
the peak value occured at the pile head and the relative displacements between the pile
head and pile tip are 50 cm and 69 cm. The maximum bending moments of piles are
shown in Figure 14 and those peak values would also occur approximately at the interface
between liquefied and non-liquefied layers. Comparing the numerical results by
Meryersohn (1994), the computed values are nearly consistent with the ones reported. In
the meantime, the peak shear forces of piles also occur at this zone. Therefore, the
excessive bending moment and shear zone of the pile is again revealed in this study using
the suggested procedures.

127
0.0 1.0 2.0 3.0 4.0 5.0
0.0
0.5
1.0
0.0 1.0 2.0 3.0 4.0 5.0
0.0
0.5
1.0
E
x
c
e
s
s

P
o
r
e

P
r
e
s
s
u
r
e

R
a
t
i
o

(
R
u
)
0.0 1.0 2.0 3.0 4.0 5.0
Time (sec)
0.0
0.5
1.0
At depth of 4m
At depth of 5m
At depth of 6m

Fig. 11. The time history of excess pore pressure ratios at different depths

128

-50
0
50
-50
0
50
D
i
s
p
l
a
c
e
m
e
n
t

(
c
m
)
0 20 40 60 80 100
Time (sec)
-50
0
50
34.9
At Pile Head
At Depth of 1m
At Depth of 2m
44.7
47.3
-40
0
40
-40
0
40
D
i
s
p
l
a
c
e
m
e
n
t

(
c
m
)
0 20 40 60 80 100
Time (sec)
-40
0
40
30.7
38.1
At Depth of 3m
At Depth of 4m
At Depth of 5m
33.7
-50
0
50
-50
0
50
D
i
s
p
l
a
c
e
m
e
n
t

(
c
m
)
0 20 40 60 80 100
Time (sec)
-50
0
50
42.4
3.26
At Depth of 6m
At Depth of 7m
At Depth of 8m
35.69
-4
0
4
-4
0
4
D
i
s
p
l
a
c
e
m
e
n
t

(
c
m
)
0 20 40 60 80 100
Time (sec)
-4
0
4
3.19
3.22
At Depth of 9m
At Depth of 10m
At Depth of 11m
3.2

Fig. 12. Time histories of ground motions at different depths

129
0 50 100 150
0
1
2
3
4
5
6
7
8
9
0 50 100

Fig. 13. Maximum pile displacements for No.1 Pile and No.2 Pile
0 40 80
0
1
2
3
4
5
6
7
8
9
Mcr
My Mu
0 50 100
Mcr
My Mu

Fig. 14. Maximum pile bending moments for No.1 Pile and No.2 Pile

130
3.2 Case study: Pile damages due to lateral spread
Mikagehama is a man-made island in the port area of Kobe home to a number of liquefied
propane gas (LPG) and oil tanks. During the 1995 Kobe Earthquake, the soils underlying the
foundatinos of tanks liquefied. A quay wall moved seaward and lateral spreading of the
backfill soils damage the piles supporting the tanks. Oil-storage tank TA72 is chosen to be a
target, which is located in the west part of the island about 20m from the waterfront. Figure
15 illustrates the cross sectional view of tank and underlying pile foundations. The tank has
a diameter of 14.95 m and its storage capacity is about 2450 kl. It is supported on 69 precast
concrete piles each with the length of 23 to 24 m and diameter of 45 cm. The water table is
estimated at the depth of 2 to 3 m. Sand compaction piles were conducted to increase the
SPT-N values of the Masado layer around the outside of Tank TA72.
According the relation between the bending moment ( M) and curvature () where
0
D is
the diameter of pile and N is axial load on pile head, one can know that the cracking
bending moment (
cr
M ), the yield bending moment (
y
M ) and the ultimate bending moment
(
u
M ) are 105 kN-m, 200 kN-m, and 234 kN-m respectively. The ultimate shear strength is
232 kN with regards to ACI (1998). Ishihara and Cubrinovski (2004) have utilized bore-hole
cameras and inclinometers to inspect the damages of the piles. Their results for pile No. 2
are shown in Figure 15. The main failure field was located at a depth of 8 to 14 meters where
the piles were found to have developed many cracks. Moreover, pile No. 2 had wounds due
to large deformations where lateral spreading of liquefied soils developed along the weak
interface. Quantifying damage of structures caused by earthquakes in terms of Park and
Ang damage indices, an index that provides a measure of structure damage level, gave a
value of 0.8, signifying the piles were in a near state of collapse. (Park and Ang, 1985;
Moustafa, 2011).

Fig. 15. Cross sectional view of Tank TA 72 and its foundation (from Ishihara and
Cubrinovski, 2004)

131
In this study, the length of pile is assumed to be 24 m with a diameter of 45 cm. Seismic
record of the NS-component of 1995 Kobe Earthquake is adopted. According to the field
data, distributions of pore water ratio pressure (
u
r ) versus the depths can be estimated by
evaluating the liquefaction potential of that site. With all the required data and
incorporating with the modified M-O model (Zhang et al, 1998), the dynamic coefficients of
earth pressure are computed as shown in Figure 16. Also, the unit weight of the soil is
reduced by
u
r (refer to Eq. 22). When obtaining those dynamic earth forces to insert and
execute the wave equation analysis, the time histories of displacements along the pile can be
illustrated as shown in Figure 17. The displacement of the pile head oscillates significantly
with time, but the peak value is smallest. As the depth increases, the peak displacement of
pile becomes larger. Those peak displacements along the pile are shown in Figure 18(a). The
maximum value among them occurs at the pile tip about 52.7 cm and the maximum relative
displacement between the pile top and the pile bottom is estimated about 44.7 cm. The
deformed shape of the pile is similar to pile No. 2. It can be found that the maximum
bending moments which exceed the ultimate bending moment at depths of 2 to 23 m and
that this zone is the mose dangerous zone.. With regards to the shear failure, the weak
interface exists at a depth of 11 m, in which the maximum shear force is close to the ultimate
(Figure 18b~18c). The above observations are agreeable to field investigations reported by
Ishihara and Cubrinovski (2004).
0
5
10
K
E
P
0 10 20 30 40 50
Time (sec)
0
1
2
K
E
A
8.1
1.5

Fig. 16. Dynamic coefficients of earth pressure

132

-60
0
60
8.02
-60
0
60
14.19
-60
0
60
27.19
-60
0
60
D
i
s
p
l
a
c
e
m
e
n
t

(
c
m
)
30.4
-60
0
60
37.8
0 10 20 30 40 50
Time (sec)
-60
0
60
At Ground Surface
At Depth of 4m
At Depth of 8m
At Depth of 12m
At Depth of 16m
At Depth of 20m
47.13

Fig. 17. Time histories of lateral displacement along the Tank TA72 No.2 pile

133
0 50
0
2
4
6
8
10
12
14
16
18
20
22
24
0 200 400 600
Mcr My Mu
0 50 100 150 200 250
Vu =232 kN
(a) Maximum displacement (cm) (b) Maximum bending moment (kN-m) (c) Maximum shear force (kN)
44.7cm

Fig. 18. Maximum structural response along the Tank TA72 No.2 pile
(a) Maximum displacement; (b) Maximum bending moment; (c) Maximum shear force
4. Conclusions
EQWEAP is a simplified but effective procedur to analyze the dynamic pile-soil interaction
under the earthquake. In the analysis, the pile deformations are obtained solving the
discrete wave equations of the pile, where the seismic ground motions are pre-calculated
from one-dimensional lumped mass model assuming a free-field condition or dynamic earth
pressure are directly exerted onto the pile. This chapter presented both displacement- and
forced-based form of the EQWEAP analysis method along with two comparative case
studies: Using wave equation analysis and the EQWEAP method, pile response to
liquefaction has been computed and compared to the case histories of the Niigata
earthquake records. Case histories of the Kobe earthquake show that the lateral spreading
can be a major cause to damage the piles. Specifically conclusions for the displacement and
forced based EQWEAP methods can be summarized as follows:
1. Based on the suggested numerical procedure using EQWEAP (Chang and Lin, 2003; Lin
et al., 2011), one can evaluate the motions of the soil stratum and the pile foundations at
various depths to estimate the occurrence of pile damages and patterns of failure. This
procedure provides a simplified but rational dynamic analysis to the pile foundation
design work.
2. The use of the empirical excess pore pressure model for liquefaction can be applicable
to soils underneath the liquefiable layers using a minimum pore pressure ratio. The
pore pressure ratio should be calculated using the empirical formula suggested by
Tokimatsu and Yoshimi (1983) providing that the factors of safety against liquefaction
are known.

134
3. Not only the interfaces between the liquefied and non-liquefied layers can exert
excessive bending moments and shear stress, but also the layer contrast of the soils can
yield similar effects. Engineers need to be more careful in designing pile shafts that are
susceptible to fail due the liquefaction resulting from earthquakes and the layer
contrast.
4. The wave equation analysis can be used to model the pile responses under lateral
spread due to earthquake. The modified M-O model (Zhang et al., 1998) incorporating
reduction methods for soil parameters were successfully used to represent the dynamic
earth pressures of the lateral spread. The numerically determined pile deformations
were similar to deformations discovered at piles actually affected by lateral spread. In
advance, if nonlinear behaviour of pile such as the moment-curvature relationship and
complexity of pile geometries can also be considered simultaneously in this method, the
results would be enhanced to capture detailed mechanism and definite performance of
piles foundations.
5. Acknowledgement
The special thank goes to my colleague, Mr. Jeff Keck. The typeset and revision that he gave
truly help to complete this task. The assistance is much indeed appreciated.
6. References
Abbas, A.M. and Manohar, C.S. (2002). Critical spatially varying earthquake load models for
extended structures. Journal of Structural Engineering, Vol. 9, No.1, pp. 39-52.
American Concrete Institute (1995). Building Code Requirements for Structural Concrete, ASCE,
USA
Bathe, K.J. (1982). Finite Element Procedure in enigeering analysis, Prentice Hall, Englewood
Cliffs, New Jerwey, 735 p.
Bhattacharya, S., Bolton, M.D. and Madabhushi, S.P.G. (2005). A Reconsideration of the Safety
of Piled Bridge Foundation in Liquefiable Soils, Soils and Foundations, JGS, Vol. 45, No.
4, pp. 13-25.
Bhattacharya, S., Madabhshi, S.P.G. and Bolton, M.D. (2002). Pile Instability during
Earthquake Liquefaction, Engineering Mechanics Conference, ASCE , Seattle, July.
Bhattacharya, S., Madabhushi, S.P.G. and Bolton, M.D. (2004). An alternative Mechanism of
Pile Failure in Liquefiable Deposits during Earthquakes, Geotechnique, Vol.3, No.54,
, pp. 203-213.
Boulanger, R.W., Curras, C.J., Kutter, B.L., Wilson, D.W. and Abghari, A., (1999). Seismic
Soil-Pile-Structure Interaction Experiments and Analyses, Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, Vol.125, No.9, pp. 750-759.
Budhu, M. and Ai-karni, A. (1993). Seismic Bearing Capacity of Soils, Geotechnique, Vol.43,
No.1, pp.187-187.
Chang, D.W. and Lin, B.S. (2003). Wave Equation Analyses on Seismic Responses of
Grouped Piles, Procds., 12th Asian Regional Conf. on Soil Mechanics and Geotechnical
Engineering, Singapore, August, pp. 581-586.
Chang, D.W. and Yeh, S.H., (1999). Time-domain Wave Equation Analyses of Single Piles
Utilizing Transformed Radiation Damping, Soils and Foundations, JGS, Vol. 39, No.
2, pp. 31-44.

135
Chang, D.W., Roesset, J.M. and Wen, C.H., (2000). A Time-domain Viscous Damping Model
Based on Frequency-dependent Damping Ratios, Soil Dynamics and Earthquake
Engineering, Vol. 19, pp. 551-558.
Chaudhuri, D., Toprak, S., and ORourke, T.D. (1995). Pile Response to Lateral Spread: a
Benchmark Case, Lifeline Earthquake Engineering, Procds., 4th U.S. Conference, San
Francisco, California, pp. 1-8.
Curbrinovski, M and Ishihara, K. (2004). Simplified Method for Analysis of Pile Undergoing
Lateral Spreading in Liquefied Soils, Soils and Foundations, JGS, Vol. 44, No. 5, pp.
119-133.
DeAlba, P., Seed, H.B. and Chan C.K. (1976). Sand Liquefaction in Large-Scale Simple Shear
Tests, Journal of Soil Mechanics and Foundations Division, ASCE, Vo1.102, GT9, pp.
909-927.
Ebeling, R.M., and Morrison, E.E. (1993). The Seismic Design of Waterfront Retaining
Structures, Office of Navy Technology and Department of the Army, NCEL
Technical Report R-939, Naval Civil Engineering Laboratory, Port Hueneme, CA, 256
p.
El Naggar, M.H. and Bentley, K.J. (2000). Dynamic Analysis for Laterally Loaded Piles and
Dynamic p-y Curves, Cananian Geotechnical Journal, Vol.37, pp. 1166-1183.
Eruocode 8 (1998). Design Provisions for Earthquake Resistance of Structure-Foundations,
Retaining Structures and Geotechnical Aspects, European Committee for
Standardization, Brussels.
Finn, W.D.L. and Thavaraj, T. (2001). Practical Problems in the Seismic Analysis of Bridge
Pile Foundations, Procds., 10th Int. Conf. Computer Methods and Advances in
Geomechanics, Tucson, Arizona, USA, pp. 1011-1018.
Finn, W.D.L., Lee, K.W., and Martin, G.R. (1977). An Effective Stress Model for Liquefaction,
Journal of the Geotechnical Engineering Division, ASCE, Vol. 103, No. SM7, pp. 657-
692.
Haigh, S.K. and Madabhushi S.P.G. (2005). The Effects of Pile Flexibility on Pile-Loading in
Laterally Spreading Slops, Proc. Int. Workshop Simulation and Seismic Performance of
Pile Foundations in Liquefied and Laterally Spreading Ground, ASCE, 14 p.
Hamada, M.1992. Large Ground Deformations and Their Effects on Lifelines1964
Niigata Earthquake, in Case Studies of Liquefaction and Lifeline Performance
During Past Earthquakes, Vol. 1, Japanese Case Studies, Technical Report NCEER-92-
0001, NCEER, Buffalo, NY, USA.
Hamada, M. and Wakamatsu, K. (1998). A Study on Ground Displacement Caused by Soil
Liquefaction, JSCE, Journal of Geotechnical Engineering, No. 596/-43, pp. 189-208.
Ishibashi, I. and Fang, Y.S., (1987). Dynamic Earth Pressures with Different Wall Movement
Modes, Soils and Foundations, JGS, Vol. 27, No. 4, pp. 11-22.
Ishibashi, I., Osada, M. and Uwabe, T. (1994). Dynamic Lateral Pressures due to Saturated
Backfills on Rigid Walls, Journal of Geotehnical Engineering, Vol.120, No.10, pp. 1747-
1767.
Ishihara, K. (2003). Liquefaction-Induced Lateral Flow and Its effects on Foundation Piles,
5th National Conference on Earthquake Engineering, Istanbul, Turkey, May, 28 p.
Ishihara, K. and Cubrinovski, M. (2004). Case Studies on Pile Foundations undergoing
Lateral Spreading in Liquefied Deposits, Proceeding of 5th International Conference on
Case Histories in Geotechnical Engineering, New York, Paper SOAP 5.

136
Japanese Road Association (JRA) (1996). Specification for Highway Bridges, Part V, Seismic
Design.
Jefferies M. and Been K. (2006). Soil Liquefaction, Taylor and Francis, New York.
Kim, S.I. (2003). Liquefaction Potential in Moderate Earthquake Regions, Procds., 12th Asian
Regional Conf. on Soil Mechanics and Geotechnical Engineering, Singapore, August,
pp.1109-1138.
Kramer, S.L. (1996). Geotechnical Earthquake Engineering, McGraw-Hill Book Co., Inc., New
York.
Kramer S. and Algamal A. (2001). Modeling Soil Liquefaction Hazards for Performance-Based
Earthquake Engineering, PEER Report 2001/13.
Lee, K. L. and Albeisa, A. (1974). Earthquake Induced Settlements in Saturated Sands,
Journal of Soil Mechanics and Foundations Division, ASCE, Vo1.100, GT4, pp. 387-406.
Lin, B. S., (2006). Structural Analyses for Pile Foundations Subjected to Soil Liquefaction and
Lateral Spreading, Doctors Dissertation, Tamkang University, Taiwan, 487 p.
Lin, B. S., Chang, D.W, and Ho, H. C. (2010). Pile Responses under seismic earth pressures of
Lateral Spread Using Wave Equation Analysis, ASCE, Geotechnical Special
Publication, N0. 201, Dynamic Soil-Structure Interactions: Part I - Proceedings of Sessions
of GeoShanghai 2010, pp. 109-117.
Lin, S.S., Tseng Y. J., Chiang C.C. and Hung C.L. (2005). Damage of Piles Caused by
Laterally Spreading Back Study of Three Cases, Workshop on Simulation and
Seismic Performance of Pile Foundation in Liquefied and Laterally Spreading Ground,
University of California, Davis, March.
Madabhushi, G., Knappett, J., Haigh, S. (2010). Design of Pile Foundation in Liquefiable Soils,
World Scientific, Inc.
Martin, G.R., Finn, W.D.L., and Seed, H.B. (1975). Fundamentals of Liquefaction under
Cyclic Loading, Journal of the Geotechnical Engineering Division, ASCE, Vol. 101, No.
GT5, pp. 423-438.
Matsuzawa, H., Ishibashi, I, and Kawamura, M. (1985). Dynamic Soil and Water Pressure of
Submerged Soils, Journal of the Geotechnical Engineering, ASCE, Vol. 111, No. 10, pp.
1161-1176.
Meyersohn, W.D. (1994). Pile Response to Liquefaction-induced Lateral Spread, Doctors
Dissertation, University of Cornell, USA.
Mononobe, N. and Mastuso, H. (1929) On the Deternation of Earth Pressure During
Earthquake, Proceedings, World Engineering Congrss, 9 p.
Moustafa, A. (2011). Damage-Based Design Earthquake Loads for Single-Degree-Of-
Freedom Inelastic Structures, Journal of Structural Engineering, ASCE, Vol. 137, No.
3, pp. 456-467.
National Earthquake Hazard Reduction Program (NEHRP) (2000). Commentary (Federal
Emergency Management Agency, USA, 369) for Seismic Regulations of New Buildings
and Other Structures.
Nogami, T., Otani, J., Konagai, K. and Chen, H.L. (1992). Nonlinear Soil-pile Interaction
Model for Dynamic Laterally Motion, Journal of Geotechnical Engineering, ASCE,
Vol.118, No.1, pp. 89-106.
Novak, M. (1974). Dynamic Stiffness and Damping of Piles, Canadian Geotechnical Journal,
Vol.11, pp. 574-598.

137
Novak, M. (1977). Vertical Vibration of Floating Piles, Journal of the Engineering Mechanics
Division, ASCE, Vol. 03, No. M1, pp. 153-168.
Novak, M. and Beredugo, Y.O. (1972). Vertical Vibration of Embedded Footings, Journal of
Soil Mechanics and Foundations Division, ASCE, Vol. 98, No. SM12, pp. 1291-1310.
Okabe, S. (1926). General Theroy of Earth Pressures, Journal of the Japan Society of Civil
Engineering, Vol.12, No.1.
Park Y.J., and Ang, A.H.S. (1985). Mechanistic Seismic Damage Model for Reinforced
Concrete, Journal of Structural Engineering, ASCE, Vol.111, No.ST4, pp. 722-739.
Prakash, S. and Puri, V.K. (1988), Foundations for Machines: Analysis and Design, John Wiley
and Sons, Canada, pp. 438-458.
Priestley, M.J.N., Seible F. and Calvi, G.M. (1996). Seismic Design and Retrofit of Bridges, John
Wiley & Sons, Inc.
Richard, R., Elms, D.G. and Budhu, M. (1990). Soil Fluidization and Foundation behavior,
Proc. 2nd International Conference Recent Advance in Geotechnical Engineering and Soil
Dynamics, pp. 719-223.
Richard, R., Elms, D.G. and Budhu, M. (1993). Seismic Bearing Capacity and Settlement of
Foundations, Journal of Geotechnical Engineering Division, ASCE, Vol. 119, No. 4, pp.
662-674.
Roesset, J.M. (1977), Soil Amplification of Earthquake, Numerical Methods in Geotechnical
Engineering, McGraw-Hill Book Co., Inc., New York, pp. 639-682.
Schanz, M. and Cheng, A.H.-D (2000). Transient Wave Propagation in a One-Dimensional
Poroelastic Column, Acta Mechanica, Vol.145, No.1-4, pp. 1-18.
Seed, H.B., and Idriss, I.M. (1970). Soil Moduli and Damping Factors for Dynamic Response
Analysis, Report NO.EERC 75-29, Earthquake Engineering Research Center,
University of California, Berkeley, California.
Smith, E.A.L. (1960). Pile Driving Analysis by the Wave Equation, Journal of Soil Mechanics
and Foundation Divisions, ASCE, Vol.86, No.SM4, pp. 35-61.
Sneddon, I.N. (1957), Elements of Partial Differential Equations, McGraw-Hill Book Co., Inc.,
New York, pp. 274-275.
Soubra, A. H. and Regenass, P. (2000). Three-dimensional Passive Earth Pressure by
Kinematical Approach, Journal of Geotechnical and Geoenvironmental Engineering, Vol.
126, No. 11, pp. 969-978.
Stephen, E. D., Nason, J. M., Mark, G. B., and Bryan, J. W. (2002). Assessment and Mitigation of
Liquefaction Hazards to Bridge Approach Embankments in Oregon, Oregon State
University, FHWA-OR-RD-03-04, Nov. 2002.
Tokimatsu K. and Yoshimi Y. (1983). Empirical Correlation of Soil Liquefaction Based on
SPT N-Value and Fines Content, Soil and Foundations, JSSMFE, Vol. 23, No.4, pp. 56-
74.
Tokimatsu, K. (1999). Performance of Pile Foundations in Laterally Spreading Soils, Procds,
2nd Int. Conf. Earthquake Geotechnical Engineering, Vol.3, pp. 957-964.
Tokimatsu, K. (2003). Behavior and Design of Pile Foundations Subjected to Earthquakes,
Procds., 12th Asian Regional Conf. on Soil Mechanics and Geotechnical Engineering,
Singapore, August.
Tokimatsu, K. and Asaka, Y. (1998). Effects of Liquefaction-induced ground displacement on
pile performance in the 1995 Hyogoken-Nambu earthquake, Soils and Foundations,
Special Issue, No. 2, pp. 163-178.

138
Uchida, A. and Tokimatsu, K. (2005). Comparison of Current Japanese Design for Pile
Foundations in Liquefiable and Laterally Spreading ground, Proc. Int. Workshop
Simulation and Seismic Performance of Pile Foundations in Liquefied and Laterally
Spreading Ground, ASCE, 10p.
Wang, J.-N., Mesa , L., and Sizemore J. (2004). Seismic Design of the Cooper River Bridge,
Proceedings of Geo-Trans 2004, ASCE, California, July.
Zerva, A. (2009). Spatial Variation of Seismic Ground Motions, CRC Press.
Zerva, A., and Zervas V. (2002). Spatial Variation of seismic ground motions: an overview,
Applied Mechanics Reviews, Vol.55, No.3, pp. 271297.
Zhang, J. M., Shamoto, Y., and Tokimatsu, K. (1998). Seismic Earth Pressure for Retaining
Walls under Any Lateral Displacement, Soils and Foundations, Vol. 38, No. 2, pp.
143-163.
6
Non-Linear Numerical Analysis of Earthquake-
Induced Deformation of Earth-Fill Dams
Babak Ebrahimian
University of Tehran
Iran
1. Introduction
From January 2010 to March 2011, eight major earthquakes of magnitude 6.1 or greater, one
of which was 9.0, have shaken the globe. These destructive events, such as January 2010
Haiti earthquake and March 2011 Tohoku Japan earthquake, have resulted in numerous
impacts on different lifeline facilities such as dams (Takewaki, 2010; Eberhard et al., 2010;
Rathje, 2010; Ashford et al., 2011; Harder et al., 2011; Matsumoto et al., 2011). During these
recent earthquakes as well as the past strong ones, numerous earth-fill dams have failed and
enormous loss of life and significant property damage have been caused (Sherard et al.,
1963; Pinto, 1993; Ozkan, 1998; Krinitzsky & Hynes, 2002; Basudhar et al., 2010; Matsumoto
et al., 2011). Therefore, evaluation of stability conditions of earth-fill dams is a major issue of
concern in seismic areas. Earthquake-induced deformations in an earth-fill dam may lead to
overtopping and consequently to severe losses in terms of property and human lives.
Hence, an accurate evaluation of seismic stability of earth-fill dam should be employed to
guarantee safer conditions for dams during earthquakes.
The performance of earth-fill dams, subjected to seismic action, can be evaluated through
different approaches including force-based pseudo-static methods, displacement-based
sliding block methods and dynamic analysis (Sherard, 1967; Seed et al., 1975). Since the 1971
San Fernando earthquake in California, understanding the effects of earthquake actions on
dams has been significantly progressed (USCOLD, 1992). Gazetas (1987) discussed the
historical developments of theoretical methods for estimating the seismic response of earth-
fill dams to earthquake. He outlined important features, advantages and limitations of the
methods. Pseudo-static approach, the most common method, is widely used in engineering
practice to assess the seismic stability of earth-fill dams. This approach is quite simplistic by
which the complex aspects of seismic behaviour are represented in terms of static forces and
the dam stability is expressed in terms of an overall factor of safety. The response of dam to
earthquake may be related to several factors such as dam geometry, mechanical properties
of construction soil materials, distributions of pre-seismic stresses and pore water pressures
inside the dam body, and input motion characteristics. Most of these factors are partially or
totally neglected by the approaches traditionally adopted for assessing the seismic safety of
earth-fill dams. For instance, some earthquake parameters such as frequency content and
duration which significantly affect the soil response are neglected in the pseudo-static
approach (Ambraseys, 1960). On the other hand, studying the seismic response of earth-fill
dam is complex and, in general, requires dynamic analysis methods with different levels of

140
sophistication in terms of proper problem formulation, characterization of material
properties and modelling of soil stressstrain behaviour. In dynamic analysis methods using
numerical simulation techniques, comprehensive analysis of earth-fill dam responses to
dynamic loading is allowed. The development of geotechnical computation and numerical
modelling offers interesting facilities for dam response analysis, considering complex issues
such as soil non-linearity, evolution of pore water pressures and real earthquake records. In
this regard, Prevost et al. (1985) presented 2D and 3D non-linear dynamic finite element (FE)
analyses of an earth-fill dam, based on non-linear hysteretic analysis using multi-surface
plasticity theory. Lacy & Prevost (1987) proposed a general and efficient numerical
procedure for analyzing the seismic response of earth-fill dams. In their procedure, the
dams were considered as non-linear two-phase systems. They outlined appropriate coupled
dynamic field equations for the response of two-phase soil system. Abouseeda & Dakoulas
(1998) studied the non-linear seismic behaviour in earth-fill dam-foundation interaction
using boundary element (BE) and finite element (FE) methods. Chen & Harichandran (2001)
studied the stochastic response of Santa Felicia earth-fill dam, in southern California, to
spatially varying earthquake ground motion (SVEGM). They used SVEGM model in which
the propagation of seismic waves is taken into account. Cascone & Rampello (2003)
investigated the seismic stability of an earth-fill dam using decoupled displacement
analysis. Ming & Li (2003) conducted a full coupled analysis of failure and considered the
remediation of Lower San Fernando Dam. They used a critical state model, incorporating
the concept of state dependent dilatancy for describing the soil behaviour over full loading
ranges. Adalier & Sharp (2004) studied the seismic behaviour and remediation of an
embankment on a liquefiable foundation. Papalou & Bielak (2004) studied the non-linear
seismic response of earth-fill dams with canyon interaction. In their developed FE-based
method, the dam was idealized as a shear beam and the surrounding medium as a half-
space. The dams non-linearity was considered using multi-yield surface plasticity theory.
Ebrahimian & Vafaeian (2005) considered the seismic response of earth-fill dams during
earthquake using 2D full non-linear dynamic analysis. They used finite difference (FD)
method and adopted the Mohr-Coulomb elastic-perfectly plastic constitutive model to
describe the stress-strain relation of the soil. They focused on the seismic behaviour of very
high earth-fill dams. Wang et al. (2006) presented the dynamic analyses in which a non-
linear, effective-stress-based soil model is employed. They used bounding surface
hypoplasticity model for sand and implemented the model into a 2D finite difference (FD)
program. The advantages of the proposed non-linear approach, conducted for a rock-fill
dam, were illustrated by comparing the obtained results with those of equivalent linear
approach. The models capability was demonstrated by evaluating the seismic performance
of an earth-fill dam. Siyahi & Arslan (2008) carried out the transient dynamic time history FE
simulations to investigate the performance of earth-fill dams under seismic excitation. Then,
they studied different failure modes of earth-fill dams as the earthquake aftermath. Sica et
al. (2008) studied the effect of loading history on the seismic response of earth-fill dams.
They considered the static and seismic behaviours of a real case-history using coupled
dynamic approach. The approach was solved numerically by FE method. Rampello et al.
(2009) studied the response of an earth-fill dam to seismic loading using the displacement-
based analysis and the FE effective-stress dynamic analysis. The FE analysis was carried out
using a constitutive model which was capable to reproduce the soil non-linearity and
calibrated against laboratory measurements. They also investigated the effects of assumed
input motion and bedrock depth on the seismic response of earth-fill dam. Ebrahimian
(2009) presented a numerical modelling of seismic behaviour of an earth-fill dam rested on

Non-Linear Numerical Analysis of Earthquake-Induced Deformation of Earth-Fill Dams

141
liquefiable foundation. The numerical simulation was carried out using effective-stress-
based, full coupled non-linear dynamic analysis. In this regard, Finn-Byrne model with
extended Masing rules was used to model the pore water pressure generation in the
liquefied soils. Ebrahimian (2011) investigated the dynamic behavior of earth dams by using
a full non-linear dynamic finite difference analysis. The effects of input motion
characteristics and dam reservoir condition on the dynamic response of earth dams were
identified in this study. For this purpose, three real earthquake records with different levels
and PGAs were used as the input motions.
In many parts of the world, the repetition of mediumstrong intensity earthquake ground
motions at brief intervals of time has been observed. The design philosophies for dams in
seismic regions are based on multi-level design approaches, which take into account more
than a single damageability limit state. According to these approaches, a sequence of seismic
actions may produce important consequences on the dam safety. In fact, dams have been
among the first structures that have been designed systematically against different
earthquake levels. Since 1989, the ICOLD guidelines have introduced several levels of
seismic loading, namely the Operating Basis Earthquake (OBE), Maximum Credible
Earthquake (MCE), Maximum Design Earthquake (MDE) and Safety Evaluation Earthquake
(SEE). However, the terms MDE or SEE are used as substitutes for the MCE. In order to
analyze the behaviour of dams for specified levels of seismic hazard, several requirements
should be considered. The seismic input and performance levels associated with
serviceability, damage control, and collapse prevention are also defined. A thorough review
about the different earthquake levels for dam design has been given in Wieland (2008).
Amadio et al. (2003) analyzed the effects of repeated earthquake ground motions on the
response of single-degree-of-freedom systems (SDOF) with non-linear behaviour.
Accordingly, a comparison study was performed to investigate the effect of a single seismic
event on the originally non-damaged system for different hysteretic models in terms of
pseudo-acceleration response spectra and damage parameters. They showed that the
elasticperfect plastic system is the most vulnerable one under repeated earthquake ground
motions. Moustafa & Takewaki (2010) modelled ground motions of multiple sequences that
produce the maximum damage in the structure. It was shown that critical repeated
acceleration sequences produce larger structural damage compared to single critical
earthquakes. Afterwards, Moustafa (2011) developed a new framework to model the design
earthquake loads for inelastic structures. New measures of the structure performance that
were based on energy concepts and damage indices were introduced in his paper.
Concerning the seismic-resistant design of dams, several international standards have been
developed by scientific communities in the past 25 years. However, only few countries have
their own guidelines and regulations for seismic design of dams. Therefore, the ICOLD
Bulletins and the local seismic building codes (e.g., Eurocode 8) are used as references. In
brief, other famous international codes are USCOLD (United States Committee on Large
Dams), US Army Corps of Engineers (USACE), ANCOLD, (Australian National Committee
on Large Dams), IITK-GSDMA Guidelines for Seismic Design of Earth Dams and
Embankments and Canadian Dam Association Guidelines for Dam Safety.
In this chapter, a numerical study of seismic behaviour of earth-fill dams overlaying bedrock
subjected to real earthquake records is presented. For this purpose, full non-linear dynamic
finite difference (FD) analysis is employed incorporating a simple elastic perfectly plastic
constitutive model and Rayleigh damping. The former is used to describe the stress-strain
response of the soil and the latter to increase the hysteretic damping level. The effect of non-

142
linear soil behaviour is then considered in the analysis from the very beginning of
earthquake loading. In order to precisely explain the soil response under general cyclic
loading, Masing rules (Masing, 1926) are implemented into the constitutive model. Soil
stiffness and hysteretic damping change with loading history. Firstly, the procedures of
calibrating the constructed numerical models with centrifuge test data as well as real case
history are presented and explained. Moreover, some important aspects of model calibration
are discussed. Long Valley earth-fill dam, subjected to the 1980 Mammoth Lake earthquake,
is analyzed for the real case history and the obtained numerical results are compared with
the real ones, measured at the site in both time and frequency domains. The computed
values show relative good agreements with the measured ones. It is shown that the Masing
rules, combined with the simple elastic-plastic model, offer reasonable numerical
predictions. A comprehensive parametric study is also carried out to identify the effects of
dam height, input motion characteristics, soil behaviour and strength of shell materials on
the seismic response of earth-fill dams. It is demonstrated that the fundamental aspects of
seismic behaviour of earth-fill dams can accurately be captured by the current numerical
procedure. It should be mentioned that this study does not consider the fluid-skeleton
interaction, which may have significant effects on the seismic response of earth-fill dams.
2. Methodology
2.1 Numerical modeling procedure
Here, numerical analysis is conducted using FLAC program, based on a continuum finite
difference discretization applying Lagrangian approach (Itasca, 2004). Every derivative in
the set of governing equations is directly replaced by algebraic expression written in terms
of field variables (e.g., stress or displacement) at discrete point in space. Regarding dynamic
analysis, explicit finite difference scheme is applied to solve the full equation of motion
using the lumped grid point masses derived from the real density of surrounding zone. The
calculation sequence first invokes the equations of motion for deriving new velocities and
displacements from stresses and forces; then, strain rates are derived from velocities, and
new stresses from strain rates. Every cycle around the loop corresponds to one time step.
Each box updates all grid variables from known values which are fixed over the time step
being executed (Fig. 1).

Fig. 1. Basic explicit calculation cycle (Itasca, 2004)


143
The equation of motion, in the simplest form, relates the acceleration ( du dt ) of a mass (m)
to the applied force (F) which may vary with time. Newtons law of motion for the mass-
spring system is:

du
m F
dt
=
(1)
In a continuous solid body, Eq. (1) is generalized as follows:

ij
i
i
j
u
g
t x
o

c
c
= +
c c
(2)
where, = mass density; t = time; x
j
= components of coordinate vector; g
i
= components of
gravitational acceleration (body forces);
ij
= components of stress tensor; i = components in
a Cartesian coordinate frame.
For problem analysis, the strain rate tensor and rotation rate tensor, having the velocity
gradients, are calculated by the following equations:

1
2
j
i
ij
j i
u
u
e
x x
( c
c
= + (
c c
(

(3)

1
2
j
i
ij
j i
u
u
x x
e
( c
c
= (
c c
(

(4)
where,
ij
e = components of strain rate;
ij
e = components of rotation rate;
i
u = components
of velocity.
The specific mechanical relationship is used in order to obtain the stress tensor as below:

( )
o o k = , ,
ij ij ij
M e (5)
where, M = specific rule of behaviour; k = history parameters (based on the specific rules
which may or may not exist).
The problem selected here is the simplified representation of typical earth-fill dam geometry.
The dam section is a symmetric zone section with central clay core rested on bedrock,
( )
0.5
5 H
3

Fig. 2. Geometry of dam

144
as shown in Fig. 2. Five earth-fill dam cross sections of different heights (H = 40, 80, 120, 200
and 280 m) are analyzed.
2.1.1 Constitutive model
Mohr-Coulomb constitutive relation is used to model the behaviour of soil. The failure
envelope for this model corresponds to a Mohr-Coulomb criterion (shear yield function)
with tension cutoff (tensile yield function). Stress-strain relationship is linear elastic-
perfectly plastic. Linear behaviour is defined by elastic shear and bulk modulus. While,
plastic behaviour is determined by the angle of internal friction and cohesion of the soil.
Shear modulus of sandy shell materials is calculated from the following formula (Kokusho
& Esashi, 1986):

( )
2
0.6
max
(2.17 )
8400
1
m
e
G
e
o
' =
+
(6)
where, G
max
= maximum (small strain) shear modulus in kPa; e = void ratio;
m
o' = mean
effective confining stress in kPa; Poissons ratio is considered as 0.3 for the shell materials.
Shear modulus of clayey core materials is calculated by the below formula (Hardin & Black,
1968):

( )
( )
2
0.5
max
2.973
3270
1
m
e
G
e
o
' =
+
(7)
Poissons ratio for the core materials is taken as 0.45.
Here, the basic elastic-perfectly plastic model is modified in order to better fit with the
curves of shear modulus and damping ratio derived from the experimental data. This
modified model can predict the seismic behaviour and the associated deformations of earth-
fill dams. Masing behaviour is implemented into FLAC via FISH subroutine (Itasca, 2004) in
order to represent more accurately the non-linear stress-strain behaviour of soil that follows
the actual stress-strain path during cyclic loading. Masing model consists of a backbone
curve and several rules that describe the unload-reload behaviour of soil and the cyclic
modulus degradation. Backbone curve can be constructed by the modulus reduction curves
coupled with the small strain modulus (G
max
). Unload-reload rules can similarly be
formulated to reproduce the hysteretic damping values expected from the standard curves
of damping ratio versus shear strain (e.g., Seed et al., 1986; Vucetic & Dobry, 1991). These
formulations are described later.
In this study, shear modulus and damping ratio curves, proposed by Seed et al. (1986) for sandy
soils and Vucetic & Dobry (1991) for clayey soils, are adopted as the references. The geotechnical
properties of earth-fill dam materials, used in the analyses, are presented in Table 1.

Region
wet

sat

Porosity
(n)
C K
(kN/m
3
) (kN/m
3
) (kPa) degree (cm/s)
Core 20 20.5 0.45 0.41 80 8 10
-7
Shell 22 23.0 0.30 0.33 - 40 10
-2
Table 1. Soil properties


145
2.1.2 Boundary conditions
The geotechnical problems can be idealized by assuming that the regions far from the area
of interest extend to infinity. The unbounded theoretical models should be truncated to the
manageable size by using the artificial boundaries for minimizing the computation time as
well as avoiding the outwards propagating waves form reflecting to the model. The viscous
boundary, developed by Lysmer & Kuhlemeyer (1969), is used in the current calculations. In
this case, independent dashpots are used in the normal and shear directions at the model
boundaries, as shown in Fig. 3. During the static analysis, the bottom boundary is fixed in
both the horizontal and the vertical directions and the lateral boundaries only in the
horizontal direction. In dynamic analysis, when the dam is laid on the foundation (and not
on the bedrock), lateral boundaries are changed into free-field boundaries, available in the
FLAC library, in order to eliminate the wave reflections from the truncated boundaries.
2.1.3 Element size
Numerical distortion of propagating wave can occur in dynamic analysis as a function of
modelling condition. The numerical accuracy of wave transmission is affected by both the
frequency content of input wave and the wave speed characteristics of system. Kuhlemeyer
& Lysmer (1973) showed that for an accurate representation of wave transmission through
the soil model, the spatial element size should be smaller than 1/10 to 1/8 of the wavelength
associated with the highest frequency component of input wave i.e.,

9
L

A = (8)
where, = wave length associated with the highest frequency component that contains
significant energy. Considering the above mentioned criteria, the element size is defined
small enough to allow the seismic wave propagation throughout the analysis.

Fig. 3. Free field boundaries in dynamic analysis (Itasca, 2004)

146
2.1.4 Damping
Material damping in soil is generally because of its viscosity, friction and plasticity
development. Indeed, the role of damping in the numerical models is the reproduction of
energy losses in the natural systems subjected to dynamic loads. The dynamic damping is
provided in the model by Rayleigh damping option available in FLAC. Rayleigh damping
was primarily used in analyzing the structures and elastic continua in order to damp the
natural oscillation modes of system. Rayleigh damping R
d
=5% is used to compensate the
energy dissipation in the media (Itasca, 2004). In the dynamic analysis incorporated
plasticity constitutive models, considerable amount of energy dissipation can occur during
the plastic flow. In the calculations of such cases, minimal percentage of hysteretic damping
(e.g. 2%) is considered as well. The dams natural frequency is determined as the Rayleigh
damping parameter by Fourier analysis of its free response, as shown in Fig. 4. The
fundamental frequency (f
1
= 1.71 Hz) of the dam with 40 m height is shown in this figure
and those of dams with different heights are tabulated in Table 2.
2.1.5 Time step
The governing equations in time should be integrated incrementally for completing the
numerical solution. The solution time step should be small enough in order to accurately
define the applied dynamic loads and guarantee the stability and convergence of solution.
In this regard, time step is about 10
-6
second in the current FLAC model.

Fig. 4. Fourier amplitude spectrum of free horizontal motion at the dam crest

First mode of vibration
Dam Height (m)
40 80 120 200 280
f
1
(Hz.) 1.7 0.88 0.60 0.36 0.26
Table 2. Fundamental frequency


147
2.2 Input excitations
Selecting dynamic input motion is an important task in the seismic evaluation processes. In
non-linear dynamic analysis, the expected earthquakes should be expressed as a set of
ground motion time histories. For more correct evaluation, the input motions which offer an
appropriate range of dam responses should be selected in the adaptable time history
realizations. This procedure may be intractable due to the number of time-history
realizations. However, in reality, the level of earthquake responses, probably achieved by
physical system, is limited. Quantifying such responses demands good understanding of the
seismic response of the system as well as the ground motion parameters that characterize
the damage potential of seismic input (USCOLD, 1999). Different parameters can be
employed to identify the severity and damage potential of a certain acceleration time
history, assumed as the representative of expected earthquake ground motion; peak ground
acceleration value (PGA) is of such kind. The use of this descriptor is intuitively natural
since accelerations and resulting inertial forces are directly related by Newtons second law.
However, there is no direct relation between PGA and structural response at the dominant
natural frequencies of most typical dams. Moreover, large PGA values are not sufficient for
generating response conditions which lead to significant damage. Despite these limitations,
PGA is still the fundamental parameter used to judge the damage potential of certain
acceleration histories. On the other hand, the seismic response of system is strongly affected
by the frequency content of earthquake. Therefore, the better characterization of a given
input motion can be achieved by using some forms of spectral representations. In particular,
using Fourier amplitude spectrum is at the core of earthquake engineering practice.
However, such characterizations do not provide direct description of the duration or time
variation features of a given input motion.
Based on above, in this chapter, three different real acceleration time histories are selected
from a database of earthquake records: Tabas, PGA=0.93g in MCE level; Naghan,
PGA=0.72g in MDE level; San Fernando, PGA=0.21g in DBE level. In the dynamic analysis
of dams, the scaled records are filtered to the maximum frequency of 10 Hz, transferred to
the inside bedrock formation by standard de-convolution analysis and applied at the base
of numerical model. The information of earthquake records are summarized in Table 3, and
their corresponding acceleration time histories and Fourier amplitude spectra are shown in
Figs. 5 & 6, respectively.

Earthquake Station Date

M
Closest
Distance
PGA (g)
PGV PGD
(km) (cm/s) (cm)
Tabas Tabas 1978 7.4 94 0.93 121.4 94.58
Naghan Naghan 1977 5.4 75 0.72 46.20 61.00
San Fernando Pasadena 1971 6.6 19 0.21 10.90 2.320
Table 3. Earthquake records data

148

-10
-5
0
5
10
0 5 10 15 20 25
Time (Sec.)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
2
)

(a) Tabas

-10
-5
0
5
10
0 5 10 15 20 25
Time (Sec.)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
2
)

(b) Naghan

-2.5
0
2.5
0 5 10 15 20
Time (sec.)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
2
)

(c) San Fernando

Fig. 5. Input acceleration time histories


149
0
2
4
6
0 5 10 15 20 25
Frequency (Hz)
F
A
S

(
m
/
s
)

(a) Tabas

0
0.5
1
1.5
2
0 5 10 15 20 25
Frequency (Hz)
F
A
S

(
m
/
s
)

(b) Naghan

0
0.2
0.4
0.6
0.8
0 5 10 15 20 25
Frequency (Hz)
F
A
S

(
m
/
s
)

(c) San Fernando
Fig. 6. Fourier amplitude spectra
2.3 Full non-linear dynamic analysis
Equivalent linear analysis is the common method used for evaluating the seismic behaviour
of earth structures. In this approach, first, the responses are linearly analyzed using the
initial values of damping ratio and shear modulus. Then, the new values of damping ratio
and shear modulus are estimated, using maximum value of shear strain and laboratory
curves. These values are used for redoing the analysis. This procedure is repeated several
times until the material properties show no variation. Therefore, no non-linear effect is
directly captured by this method as it assumes linearity during the solution process. Strain-
dependent modulus and damping functions are considered roughly in order to approximate
some effects of non-linearity (damping and material softening).

150
In the non-linear analysis, employed in this study, the non-linear stress-strain relationship is
directly followed by each zone. Damping ratio and shear modulus of the materials are
calculated automatically at different strain levels. The real behaviour of soil, under cyclic
loading, is non-linear and hysteretic. Such behaviour can be simulated by Masing model
(Masing, 1926), which can model the dynamic behaviour of soil. In this model, the shear
behaviour of soil may be explained by a backbone curve as:
( )
( )
max
max max
1 /
bb
G
F
G

t
=
+
(9)
where, F
bb()
= backbone or skeleton function; = shear strain amplitude; G
max
= initial shear
modulus;
max
= maximum shear stress amplitude.
Stress-strain curve follows the backbone curve in the first loading, as shown in Fig. 7(a);
however, for explaining the unload-reload process, the above equation should be modified.
If load reversal occurs at the point (
r
,
r
), stress-strain curve follows the path given by the
below formula:

2 2
r r
bb
F
t t (
=
(

(10)
In other words, the shapes of unload-reload curves are similar to that of backbone curve
(with the origin shifted to the loading reversal point) except they are enlarged by a factor
of 2, as shown in Fig. 7(b). The Eqs. (9) & (10) describe the Masing behaviour (Masing,
1926).
Masing rules seem not to be enough for precise explanation of soil response under general
cyclic loading. Finn et al. (1977) developed modified rules to describe the irregular loading.
They suggested that unloading and reloading curves follow the concerning two rules. If the
new unloading or reloading curve exceeds the last maximum strain and cut the backbone
curve, it will follow the backbone curve up to meeting the next returning point, as shown in
Fig. 7(c). If a new unloading or reloading curve passes through the previous one, it will
follow the former stress-strain curve, as shown in Fig. 7(d). According to this model, the
tangent shear modulus can be defined at the points on the backbone and new reloading-
unloading curves by the Formulas (11) & (12), respectively, as:

2
max
max
max
1
t
G
G G

t
(
= +
(

(11)

2
max
max
max
1
2
t r
G
G G
t
(
= +
(

(12)
Based on the results, obtained in this research, the shear stress decreases as the number of
load cycles increases; it means that shear stress-strain curves are more inclined. In this
study, Masing rules are implemented into FLAC via a series of FISH functions in order to
simulate the non-linear stress-strain relationships.


151
(
r r
)
G
max
G
max

(a) Rule 1 (b) Rule 2

(c) Rule 3 (d) Rule 4

Fig. 7. General patterns of loading, unloading and reloading paths in Masing model
3. Numerical results and discussions
3.1 Validation analysis
In this research, one-zone sample is simulated using the unit cell as shown in Fig. 8, in order
to validate the implementation of Masing rules in FLAC program. The one-zone sample
consists of sandy soil and a periodic motion is exerted at its base. Vertical loading is
established only by gravity and then, the Equilibrium stresses are installed in the soil. The
stress/strain loops of the sample are shown in Fig. 9(a) for several cycles. According to the
figure, shear modulus decreases as shear strain increases. It seems that the hysteretic model
can reasonably handle the multiple nested loops. Energy dissipation and shear stiffness
degradation are clearly observed during seismic loading, as shown in Fig. 9(b). Shear
modulus reduction curve, obtained in this study, follows well the empirical relation
proposed by Seed et al. (1986) and the test data.

152

Fig. 8. One-zone model in FLAC for simulating cyclic simple shear test
-300
-250
-200
-150
-100
-50
0
50
100
150
200
250
-0.0015 -0.0010 -0.0005 0.0000 0.0005 0.0010
Shear strain
S
h
e
a
r

s
t
r
e
s
s

(
k
P
a
)

Numerical Prediction
Seed et al. (1986)
Test data

(a) (b)
Fig. 9. Simulation results of one-zone sample: (a) hysteresis loop, and (b) comparing the
shear modulus reduction curves
The results obtained from the numerical analyses are compared with those of experimental
ones in order to evaluate the capability of proposed model. One of the centrifuge tests
related to the embankment which was performed in VELACS project (VErification of
Liquefaction Analysis using Centrifuge Studies (Arulanandan and Scott, 1993, 1994)) is
chosen as a reference. It is attempted to create almost similar conditions for both laboratory
model test and numerical model, as shown in Figs. 10(a) & (b), respectively. The numerical
results are presented and compared with those obtained from the corresponding centrifuge
test data. The computed (numerical simulations) and measured (centrifuge tests) results are
shown in Figs. 11(a) & (b). According to these comparisons, the reference numerical model
can rationally predict the seismic behaviour of earth-fill dam.
Finally, the models ability in simulating the seismic behaviour of earth-fill dam during a
real earthquake is verified by a real well-documented case history. In this regard, the results
of non-linear dynamic analysis of Long Valley (LV) earth-fill dam in California subjected to
the 1980 Mammoth Lake earthquake (Griffiths and Prevost, 1988) are presented. Then, the
results are compared with the real measurements recorded at the site and also with the
results presented by previous authors. LV dam is located in Mammoth Lake area


153
(California) in close proximity of active faults. The dam is a rolled earth-fill one formed
mainly with the impervious zone. The dam has maximum height of 55 m, 182 m length at
the crest, and upstream and downstream slopes of 3h/1v. The LV dam was instrumented in
the 1970s with a multiple-input-output array; it has 3 accelerometer stations to monitor the
boundary conditions, and 5 stations to record the dam response (Fig. 12(a)). Thus, the array
comprised a total of 22 accelerometers linked to a common triggering mechanism.

(a) (b)
Fig. 10. Model configuration: (a) schematic sketch of dam in centrifuge container box, and
(b) numerical grid constructed in FLAC

(a) (b)
Fig. 11. Measured time histories versus calculated ones at the middle of dam height: (a)
acceleration, and (b) vertical displacement
In May 1980, a series of 6 earthquakes occurred in the Mammoth Lakes area. The
magnitudes of these earthquakes were M
L
= 4.9 - 6.7, and the induced peak accelerations at
the crest centre was 0.5g in the upstream-downstream direction (x direction, as shown in
Fig. 13(a)) during the strongest event. Extensive arrays of 22 input-output (excitation-
response) accelerations were recorded, providing a valuable information source of the dam
seismic responses over a wide range of deformation levels. In this study, the dam is
subjected to the input motion, recorded downstream at the outlet during Mammoth Lake
earthquakes. The first 12 seconds of the recorded acceleration is used with data point at 0.02
second intervals and the peak acceleration is 0.135g in the upstream-downstream (x)
direction and 0.084g in the vertical direction (y).
The cross section of LV dam is shown in Fig. 12(b) and its detailed information is found in
(Griffith and Prevost, 1988). The numerical grid constructed in FLAC is presented in Fig.
12(c). The input accelerations are applied in the horizontal and vertical directions of the
model base. Free Field boundary conditions are exerted to the lateral boundaries of
numerical model.

154
This research focuses on the computed acceleration at the crest which can be compared
directly with the measured values. Previously, LV dam has been analyzed by: Lai & Seed,
1985; Elgamal et al., 1987; Griffiths & Prevost, 1988; Yiagos & Prevost, 1991; Zeghal & Abdel-
Ghaffar, 1992; Woodward & Griffiths, 1996. The first natural frequency, obtained in this

(a)

(b)

(c)
Fig. 12. Long Valley earth-fill dam: (a) schematic view of Long Valley canyon, earth-fill dam
and installed instrumentation array acceleration, (b) cross section, and (c) numerical grid for
dynamic analysis


155
study, is presented in Table 4 and compared with the other solutions available in the
literature. The results of present study are reasonably in close agreement with those of other
relevant numerical investigations. Here, the crest acceleration responses of LV dam are
computed and compared with those recorded at the site in both time and frequency
domains.

Frequency
Spectral
Analysis
2D FE
Analysis
3D FE
Analysis
Elasto-
plastic FE
Analysis
Elasto-
plastic FE
Analysis
Present
Study
Griffith &
Prevost
(1988)
Griffith &
Prevost
(1988)
Griffith &
Prevost
(1988)
Yiagos &
Prevost
(1991)
Woodward
& Griffiths
(1996)
f (Hz.) 1.85 1.76 1.95 1.987 1.79 1.71
Table 4. First natural frequency of Long Valley dam
Fig. 13(a) shows the computed horizontal acceleration of the crest; it indicates that the
amplification occurs between the base and the crest. The magnification factor of peak
amplitude at the crest is about 5.47 over the peak base amplitude. The crest response,
computed in the horizontal direction, is compared with the measured values, as shown in
Fig. 13(b); the dashed line corresponds to the computed response there. Excellent overall
agreement is achieved between the computed and measured values; however, the computed
values show higher amplitudes. The frequency contents of two time records are compared
in the form of Fourier amplitude spectra (FAS), as shown in Fig. 13(c). Their peaks are in
close agreement although the computed values show rather more energy associated with
the fundamental frequency around 1.8 Hz. The frequency content of the up/down stream
motion, presented in Fig. 13(c), shows that the energy is concentrated just at the frequencies
below 2 Hz.
In the vertical direction, the calculated acceleration shows low agreement with the measured
values, as shown in Fig. 14(a). According to this figure, the plots of vertical acceleration are
superimposed at the base and crest. This excitation is considerably noisier and less intensive
in the vertical direction in compared with that of horizontal one. The maximum
accelerations at the crest, recorded in the vertical and horizontal directions are 0.172g and
0.64g, respectively. The computed accelerations in the vertical direction are compared with
the measured values in the crest of LV dam, as shown in Fig. 14(b). According to this figure,
the computed values have generally lower amplitudes in compared with those of measured
values. The Fourier amplitude spectra of these time histories are given in Fig. 14(c). The
measured values show a broad band of frequencies, none of which is dominant. The
computed values also contain a broad band of frequencies, but with clear peaks in the
ranges of 2-3 Hz and 5-6 Hz. It should be mentioned that the time and frequency domain

156

(a) (b)

(c)
Fig. 13. Comparing computed and measured time histories in horizontal direction: (a) crest
and base acceleration, (b) crest acceleration, and (c) Fourier amplitude spectrum

(a) (b)

(c)
Fig. 14. Comparing computed and measured time histories in vertical direction: (a) crest and
base acceleration, (b) crest acceleration, and (c) Fourier amplitude spectrum


157
results show more appropriate accordance in the horizontal direction in compared with
those of the vertical direction.
The results, obtained in the validation analysis of LV dam, in term of crest acceleration are
given in Table 5 and compared with the other numerical results presented by previous
authors. According to the comparisons, the numerical procedure, presented in this study,
can properly capture the fundamental aspects of seismic behaviours of earth-fill dams. As
mentioned earlier, the numerical model is then used for parametric studying of hypothetical
earth-fill dams due to the satisfactory modelling of validation cases.

Yiagos &
Prevost
(1991)
Woodward &
Griffiths (1996)
Present
Study
Measured
values
Maximum horizontal
Acceleration (g)
0.53 0.80 0.61 0.40
Minimum horizontal
Acceleration (g)
-0.65 -0.68 -0.50 -0.50
Table 5. Comparing the Numerical results obtained for Long Valley dam
3.2 Parametric study
Here, the analyses are carried out to investigate the effects of dam height, input motion
characteristics, soil behaviour and strength of shell materials on the seismic behaviour of
earth-fill dams. The effects of different earthquakes are studied on the horizontal permanent
deformations, permanent shear strains and maximum accelerations, as shown in Fig. 15. The
values have been induced at the crests of dams with different heights. The displacements are
shown in Fig. 15(a) and the relevant shear strains in Fig. 15(b). It is clear that the shear strain
variation is similar to displacement. The horizontal displacements and shear strains in the
dam body increase with dam height increasing. The calculated values are much higher in
Tabas earthquake and the failure occurs in the dam body. According to Fig. 15(a), the
maximum horizontal displacement computed at the crest of dam is about 94 cm at the end
of Tabas earthquake. It can be observed in Figs. 15 (a) & (b), that increasing in the input
motion energy leads to significant increase of displacements and shear strains. Fig. 15(c)
illustrates the coupled effects of dam height and earthquake type on the maximum
acceleration induced at the dam crest. According to the figure, the crest acceleration
decreases as the dam height increases and no amplification is seen maybe due to more
flexible behaviour, larger damping and larger developed plastic zones, observed in higher
dams. Therefore, because of these factors, more energy is absorbed in higher dams in
compared with that in the shorter ones. It can be seen in Fig. 15 (c) that the accelerations in
the dam crest are more reduced in higher dams comparing with the smaller ones. It should
be mentioned that PGA of Naghan earthquake (0.72g) is much higher than that of San
Fernando earthquake (0.21g). However, the created displacements and shear strains in the

158
dam crest caused by Naghan earthquake are close to those of San Fernando input motion. It
can be concluded that using just PGA parameter is not sufficient for evaluating the effect of
a certain earthquake time history on the dam response. Therefore, other earthquake
parameters such as effective duration, magnitude and frequency content should be
considered in the analysis.

(a) (b)

(c)
Fig. 15. Computed values of: (a) permanent horizontal displacements, (b) permanent shear
strains, and (c) induced maximum accelerations at dam crest versus dam height for different
earthquakes
Failure mechanism with permanent shear strain contour in the dam body is shown in Fig.
16, regarding two different heights at the end of Naghan earthquake. The slip surface is
much deeper and more obvious in the dam with 280 m height (Fig. 16(b)) in compared with
that of 120 m height (Fig. 16(a)).

(a) (b)
Fig. 16. Failure surface with shear strain contour in dam body at the end of Naghan
earthquake: (a) dam height=120 m, and (b) dam height=280 m


159
A dam with 40 m height is subjected to the mentioned earthquakes and chosen as a
reference with two different behaviours, elastic and elastic-perfectly plastic, in order to
investigate the effect of soil behaviour on the seismic response of dam body. As it is
expected, regarding linear elastic behaviour, smaller displacements and shear strains are
observed along the dam height, Figs. 17(a) & (b). However, large amplification occurs
especially for the strongest earthquake, Fig. 17(c). It means that plasticity reproduce more
energy dissipation during dynamic loading. In such cases, the accelerations are reduced
across the dam height and therefore become lower than the base acceleration. According to
Fig. 17(a), maximum displacement occurs at about Z/H = 0.88 in linear elastic behaviour,
while it happens at the crest of dam in linear elastic-perfectly plastic behaviour.
Furthermore, in elastic-perfectly plastic behaviour, the dynamic induced residual
(permanent) displacement increases largely in the upper part of dam, especially for Tabas
and Naghan earthquakes, confirmed in the previous research works (Ohmachi and
Kuwano, 1994; Ozkan et al., 2006). That is why the crest should especially be considered in
designing the embankment dams, due to the stronger shaking at the upper parts, for
avoiding undesirable deformations. The distribution of shear strain is extremely non-linear
along the dam height in the stronger earthquakes, as shown in Fig. 17(b). In the elastic dams,
maximum acceleration occurs in the dam crest, as shown in Fig. 17(c). In the acceleration
profile of Tabas earthquake, a special increase is seen along the dam centerline at Z/H = 0.38.

(a) (b)

(c)
strains, and (c) induced maximum accelerations along the dam height for elastic and elastic-
plastic behaviours subjected to different earthquakes

160
Strength of dam materials is an important parameter which can significantly affect the
seismic response of dam. In this regard, different friction angles are assumed accompanying
with different dam heights, subjected to Naghan earthquake, in order to clarify the above
mentioned effect. Fig. 18(a) shows horizontal displacement values versus dam height for
different friction angles of shell materials. The variations of shear strains in the crests of
dams, with different heights, are shown in Fig. 18(b) for different friction angles of shell
materials. As it is expected when the friction angle increases, the horizontal displacement
and shear strain in the dam crest decrease. It can be seen in the above figure that the
variation of friction angle causes no significant change in the displacement and shear strain
regarding 40. However, the variation is more significant in = 45, compared with the
lower friction angles; the highest displacement values correspond to = 30. The horizontal
displacements computed at the crests of dams with 40 and 120 m heights are about 16 and
13 cm, respectively, and their shear strains are about 3.5 and 2.5, respectively. The
maximum acceleration induced at the top of dam decreases as the friction angle decreases or
the dam height increases, as shown in Fig. 18(c). Considering larger friction angles for the
shell materials (e.g., = 45) leads to about 70% increase in the dynamic amplification. The
computed maximum crest acceleration of dam with 40 m height is about 0.89g for = 45,
while that of 120 m height is 0.52g. When the dam height deceases, the horizontal
displacement and shear strain increase but the acceleration decreases at the crest of dam. All
variations are linear for = 45, but for the other friction angles are slightly non-linear, as
shown in Fig. 18.
0.00
0.05
0.10
0.15
0.20
0.25
30 50 70 90 110 130
Height (m)
H
o
r
i
z
o
n
t
a
l

D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
=30
=35
=40
=45

0.0000
0.0005
0.0010
0.0015
0.0020
0.0025
0.0030
0.0035
0.0040
0.0045
0.0050
30 50 70 90 110 130
Height (m)
S
h
e
a
r

S
t
r
a
i
n
=30
=35
=40
=45

(a) (b)
3
4
5
6
7
8
9
10
30 50 70 90 110 130
Height (m)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
2
)
=30
=35
=40
=45

(c)
strains, and (c) induced maximum accelerations at dam crest versus dam height for different
friction angles of shell materials


161
3.3 Lessons learned
The author experienced several interesting points and noteworthy items during numerical
model calibration and numerous dynamic analyses which are listed below:
- In full non-linear dynamic analysis, soil stiffness degradation is automatically taken
into account upon constitutive model of soil and just the initial shear modulus is
needed as an input parameter. Therefore, it is important to be sure that in the numerical
model, the trend of shear modulus decrease and damping ratio increase are in
agreement with those of laboratory test results during dynamic loading.
- The poissons ratio of about 0.5 should not be used in the calculating of bulk modulus
of undrained soil layers (such as clay) in the analysis. Otherwise, bulk modulus
increases irrationally and the time step of analysis decreases rigorously and
consequently the calculation time increases excessively. Therefore, the poissons ratio
should not be more than 0.45 in such cases.
- If a raw acceleration record from a site is used as a time history, then FLAC model
may exhibit residual displacements once the motion is finished. This arises from the fact
that the integral of complete time history may not be zero. Therefore, the process of
baseline drift correction should be performed in such cases.
- The input motion should be filtered before being applied to the FLAC grid in order to
eliminate all high frequency components form it.
- The stages of construction should be considered in the numerical simulation of earth-fill
dams. In the present study, the stages are: initial state of foundation (if any); layer by
layer dam replacement; applying the hydrostatic water pressure due to the replacement
of dam reservoir; seepage analysis in the dam body; mechanical adjustment to new flow
field; and finally dynamic analysis. Regarding the mentioned stages, one is run to
equilibrium and then the next stage is started. However, construction sequences have
much greater effects on static results than dynamic results.
4. Conclusions
This chapter presents the non-linear seismic behaviour of earth-fill dams using explicit finite
difference method. In this regard, a simple elastic-perfectly plastic constitutive model with
Mohr-Coulomb failure criterion is used to describe the stress-strain response of the soil.
Here, Rayleigh damping is used to promote the level of hysteretic damping during dynamic
analysis. Masing rules are implemented into the constitutive model to precisely explain the
non-linear soil response under general cyclic loading. The numerical model is then
calibrated using centrifuge test data as well as field data. The field data are obtained in real
measuring of Long Valley earth-fill dam subjected to the 1980 Mammoth Lake earthquake.
The results of dynamic analysis, obtained in this study, are compared with the real
measurements of Long Valley dam in terms of accelerations computed at the crest of dam in
both time and frequency domains. The proposed numerical model can properly reproduce
the overall seismic behaviours of earth-fill dams, their qualities and quantities, under
earthquake loading conditions, confirmed by validation analyses. After validation, the
effects of dam height, real earthquake loading, soil behaviour and strength of shell materials
on the seismic response of earth-fill dams are evaluated through a comprehensive

162
parametric study. The effect of dam height on the non-linear seismic behaviour is
particularly focused in this research. The following conclusions are obtained based on the
performed parametric study:
- If the dam materials keep their elastic behaviours during dynamic loading, then the
horizontal acceleration increases along the dam height (from the base to the top). In this
case, the higher dams show larger amplifications, especially if the natural periods of
their bodies coincide with the periodical nature of earthquake waves.
- When the dam body shows non-linearity or the materials go towards plastic behaviour
during a strong shaking, the attenuation of acceleration waves in the dam body
becomes more effective. Consequently, the amplitudes of earthquake accelerations
decrease when moving from the base towards the top.
- According to the non-linear elastic-plastic analyses, when the height of the dam
increases, then the strongest dynamic loading (Tabas earthquake) induces plasticity in
large parts of the dam body. In fact, strong earthquakes are more effective in changing
the material behaviour from elastic to plastic condition in comparison with weak
earthquakes.
- The higher dams are more flexible than the smaller ones. Consequently, the flexibility
affects the shear strains which influence the shear modulus degradation and
attenuating coefficient. All these effects are on the trend of weakening the accelerations
along the height.
- Soils with less strength (suppose low friction angle) go towards yielding by small
amount of dynamic force which cause the attenuation of acceleration along the dam
height in the weak materials compared with the strong ones.
- Regarding a dam subjected to the earthquake with lower energy, the dam body
behaves as an elastic material. Therefore, the induced seismic accelerations inside the
dam body become larger from the base of dam to its top. In this case, small plasticity
zones are developed in the dam body and the dam remains safe during dynamic
loading.
- Finally, non-linear dynamic analysis shows that plasticity should be considered in the
investigation of seismic response of earth-fill dams, because of which the acceleration of
the dam crest decreases and the displacements and shear strains of dam body as well as
the energy dissipation increase. All these can significantly affect the seismic response of
earth-fill dams.
5. References
Abouseeda, H. & Dakoulas, P. (1998). Non-linear Dynamic Earth Damfoundation
Interaction using a BEFE Method. Journal of Earthquake Engineering and Structural
Dynamics, Vol.27, pp. 917936
Adalier, K. & Sharp, M.K. (2004). Embankment Dam on Liquefiable Foundation-Dynamic
Behavior and Densification Remediation. Journal of Geotechnical Geoenvironmental
Engineering, Vol.130, No.11, pp. 12141224


163
Amadio, C.; Fragiacomo, M. & Rajgelj, S. (2003). The Effects of Repeated Earthquake Ground
Motions on the Non-linear Response of SDOF Systems. Earthquake Engineering &
Structural Dynamics, Vol.32, No.2, pp. 291308
Ambraseys, N.N. (1960). The Seismic Stability of Earth Dams, Proceedings of the Second World
Conference on Earthquake Engineering, pp. 13451363, Tokyo, Japan
Arulanandan, K. & Scott, R.F. (1993). Verification of Numerical Procedures for the Analysis
of Soil Liquefaction Problems, Conference Proceedings of VELACS, Vol. 1, Balkema,
Rotterdam
Arulanandan, K. & Scott, R.F. (1994). Verification of Numerical Procedures for the Analysis
of Soil Liquefaction Problems, Conference Proceedings of VELACS, Vol. 2, Balkema,
Rotterdam
Ashford, S.A.; Boulanger, R.W.; Donahue, J.L. & Stewart J.P. (2011). Geotechnical Quick
Report on the Kanto Plain Region during the March 11, 2011, Off Pacific Coast of
Tohoku Earthquake, Japan, Geotechnical Extreme Events Reconnaissance (GEER)
Basudhar, P.K.; Kameswara Rao, N.S.V.; Bhookya, M. & Dey, A. (2010). 2D FEM Analysis of
Earth and Rockfill Dams under Seismic Condition, Fifth International Conference on
Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics and
Symposium in Honor of Professor I.M. Idriss, San Diego, California
Cascone, E. & Rampello, S. (2003). Decoupled Seismic Analysis of an Earth Dam. Soil
Dynamics and Earthquake Engineering, Vol.23, pp. 349365
Chen, M. & Harichandran, R.S. (2001). Response of an Earth Dam to Spatially Varying
Earthquake Ground Motion. Journal of Engineering Mechanics, Vol.127, pp. 932939
Eberhard, M.; Baldridge, S.; Marshall, J.; Mooney, W. & Rix, G. (2010). The M
w
7.0 Haiti
Earthquake of January 12, 2010: USGS/EERI Advance Reconnaissance Team, Team
Report, Vol.1
Ebrahimian, B. & Vafaeian, M. (2005). Effects of Dam Height on the Seismic Response of
Earth Dam, Proceedings of 7th International Conference on Civil Engineering, Tarbiat
Modarres University, Tehran, Iran
Ebrahimian, B. (2009). Numerical Modeling of the Seismic Response of an Earth Dam
Founded on Liquefiable Soils, Proceedings of 2nd International Conference on Long
Term Behavior of Dams, E. Bauer, S. Semprich, G. Zenz, (Eds.), pp 610-615, Graz
University of Technology, Graz, Austria
Ebrahimian, B. (2011). Numerical Analysis of Nonlinear Dynamic Behavior of Earth Dams.
Frontiers of Architecture and Civil Engineering in China, Vol.5, No.1, pp. 2440
Elgamal, A.M.; Abdel-Ghaffar, A.M. & Prevost, J.H. (1987). 2-D Elastoplastic Seismic Shear
Response of Earth Dams: Application. Journal of the Engineering Mechanics ASCE,
Vol.113, No.5, pp. 702-719
Finn, W.D.L.; Lee, K.W. & Martin, G.R. (1977). An Effective Stress Model for Liquefaction.
Journal of Geotechnical Engineering Division ASCE, Vol.103, No.6, pp. 517-553
Gazetas, G. (1987). Seismic Response of Earth Dams: Some Recent Developments. Soil
Dynamics and Earthquake Engineering, Vol.6, No.1, pp. 148
Griffith, D.V. & Prevost, J.H. (1988). Two- and Three-Dimensional Dynamic Finite Element
Analyses of the Long Valley Dam. Geotechnique, Vol.38, pp. 367-388

164
Harder, L.F.; Kelson, K.I.; Kishida, T. & Kayen, R. (2011). Preliminary Observations of the
Fujinuma Dam Failure Following the March 11, 2011 Tohoku Offshore Earthquake,
Japan, Geotechnical Extreme Events Reconnaissance (GEER)
Hardin, B.O. & Black, W.L. (1968). Vibration Modulus of Normally Consolidated Clay.
Journal of Soil Mechanics & Foundation Engineering ASCE, Vol.84, No.2, pp. 1531-1537
Itasca Consulting Group, Inc. FLAC (Fast Lagrangian Analysis of Continua), Version 4. 2004,
Minneapolis, MN
Kokusho, T. & Esashi, Y. (1981). Cyclic Tri-Axial Test on Sands and Coarse Material,
Proceedings of 10th International Conference on Soil Mechanics and Foundation
Engineering, (Quoted by Ishihara 1986)
Krinitzsky, E.L. & Hynes, M.E. (2002). The Bhuj, India, Earthquake: Lessons Learned for
Earthquake Safety of Dams on Alluvium. Engineering Geology, Vol.66, pp. 163-196
Kuhlmeyer, R.L. & Lysmer, J. (1973). Finite Element Method Accuracy for Wave
Propagation Problems. Journal of the Soil mechanics and Foundation Division ASCE,
Vol.99, No. SM5, pp. 421-427
Lacy, S.J. & Prevost, J.H. (1987). Nonlinear Seismic Response Analysis of Earth Dams. Soil
Dynamics and Earthquake Engineering, Vol.6, No.1, pp. 4863
Lai, S.S. & Seed, H.B. (1985). Dynamic Response of Long Valley Dam in the Mammoth Lake
Earthquake Series of May 25-27 1980, Report No. UCB/EERC-85/12, Earthquake
Engineering Research Center
Lysmer, J. & Kuhlmeyer, R.L. (1969). Finite Element Method for Infinite Media. Journal of
Engineering Mechanics ASCE, Vol.95, No. EM4, pp. 859-877
Masing, G. (1926). Eigenspannungen und Verfestigung Beim Messing, Proceedings of 2nd
International Congress on Applied Mechanics, Zurich.
Matsumoto, N.; Sasaki, T. & Ohmachi, T. (2011). The 2011 Tohoku Earthquake and Dams,
ICOLD 89
th
Annual Meeting, Lucerne, Switzerland
Ming, H.Y. & Li, X.S. (2003). Fully Coupled Analysis of Failure and Remediation of Lower
San Fernando Dam. Journal of Geotechnical and Geoenvironmental Engineering,
Vol.129, No.4, pp. 336349
Moustafa, A. & Takewaki, I. (2010). Modeling Critical Ground-Motion Sequences for
Inelastic Structures. Advances in Structural Engineering, Vol. 3, No.4, pp. 665-679
Moustafa, A. (2011). Damage-based Design Earthquake Loads for Single-degree-of-freedom
Inelastic Structures. Journal of Structural Engineering, Vol.137, No.3, pp. 456-467
Ohmachi, T. & Kuwano, J. (1994). Dynamic Safety of Earth and Rock Fill Dams, Proceedings
of a course, T. Ohmachi, J. Kuwano, (Eds.), Balkema, Rotterdam
zkan, M.; Yzyazicioglu, M. & Aksar, U.D. (2006). An Evaluation of Gldrcek Dam
Response during 6 June 2000 Orta Earthquake. Soil Dynamics and Earthquake
Engineering, Vol.26, No.5, pp. 405419
zkan, M. (1998). A Review of Considerations on Seismic Safety of Embankments and Earth
and Rock-Fill Dams. Soil Dynamics and Earthquake Engineering, Vol.17, pp. 439458
Papalou, A. & Bielak, J. (2004). Nonlinear Seismic Response of Earth Dams with Canyon
Interaction. Journal of Geotechnical and Geoenvironmental Engineering, Vol.130, No.1,
pp. 103110
Pinto, S.P. (1993). Soil Dynamics and Geotechnical Earthquake Engineering. Balkema, Rotterdm


165
Prevost, J.; Abdel-Ghaffar, A.M. & Lacy, S. (1985). Nonlinear Dynamic Analyses of an Earth
Dam. Journal of Geotechnical Engineering, Vol.111, No.7, pp. 882897
Rampello, S.; Cascone, E. & Grosso, N. (2009). Evaluation of the Seismic Response of a
Homogeneous Earth Dam. Soil Dynamics and Earthquake Engineering, Vol.29, pp.
782798
Rathje, E.M. (2010). Case History: The Geotechnical Aspects of the Haiti Earthquake,
ISSMGE Bulletin, Vol.4, Issue 3
Seed, H.B.; Lee, K.L.; Idriss, I.M. & Makdisi, F.I. (1975). The Slides in the San Fernando Dams
during the Earthquake of February 9, 1971. Journal of the Soil mechanics and
Foundation Division ASCE, Vol.101, No. GT7, pp. 651688
Seed, H.B.; Wong, R.T.; Idriss, I.M. & Tokimatsu, K. (1986). Moduli and Damping Factors for
Dynamic Analyses of Cohesionless Soils. Journal of Geotechnical Engineering ASCE,
Vol.112, No.11, pp. 1016-1032
Sherard, J.I.; Woodward, R.J.; Gizienski, S.F. & Clevencer, W.A. (1963). Earth and Earth-Rock
Dams. John Wiely and Sons, Inc
Sherard, J.L. (1967). Earthquake Considerations in Earth Dam Design. Journal of the Soil
mechanics and Foundation Division ASCE, Vol.93, pp. 377401
Sica, S.; Pagano, L. & Modaressi, A. (2008). Influence of Past Loading History on the Seismic
Response of Earth Dams. Computers and Geotechnics, Vol.35, pp. 6185
Siyahi, B. & Arslan, H. (2008). Nonlinear Dynamic Finite Element Simulation of Alibey Earth
Dam. Environmental Geology, Vol.54, pp. 7785
Siyahi, B. & Arslan, H. (2008). Earthquake Induced Deformation of Earth Dams. Bulletin of
Engineering Geology and the Environment, Vol.67, pp. 397403
Takewaki, I. (2011). Preliminary Report of the 2011 off the Pacific Coast of Tohoku
Earthquake. Journal of Zhejiang University-SCIENCE A (Applied Physics &
Engineering), Vol.12, No.5, pp. 327-334
USCOLD (US Committee on Large Dams). Observed performance of dams during earthquakes.
Committee on Earthquakes, 1992, July, Denver, CO
USCOLD (US Committee on Large Dams). Updated guidelines for selecting seismic parameters
for dam projects. Committee on Large Dams, 1999, Denver, CO
Vucetic, M. & Dobry, R. (1991). Effects of the Soil Plasticity on Cyclic Response. Journal of
Geotechnical Engineering ASCE, Vol.117, pp. 89107
Wang, Z.L.; Makdisi, F.I. & Egan, J. (2006). Practical Applications of a Nonlinear Approach
to Analysis of Earthquake-Induced Liquefaction and Deformation of Earth
Structures. Soil Dynamics and Earthquake Engineering, Vol.26, pp. 231252
Wieland, M. (2008). Large Dams the First Structures Designed Systematically against
Earthquakes, The 14
th
World Conference on Earthquake Engineering, October 12-17,
2008, Beijing, China
Woodward, P.K. & Griffiths, D.V. (1996). Non-linear Dynamic Analysis of the Long Valley
Dam. Computer Methods and Advances in Geomechanics, Vol.11, No.6, pp. 635-644
Yiagos, A.N. & Prevost, J.H. (1991). Tow-phase Elasto-plastic Seismic Response of Earth
Dams: Applications. Soil Dynamics and Earthquake Engineering, Vol.10, No.7, pp. 371-
381

166
Zeghal, M. & Abdel-Ghaffar, A.M. (1992). Analysis of Behavior of Earth Dam Using Stron-
Motion Earthquake Records. Journal of Geotechnical Engineering ASCE, Vol.118,
No.2, pp. 266-277
7
Selection of the Appropriate Methodology
for Earthquake Safety Assessment
of Dam Structures
Hasan Tosun
1
and Evren Seyrek
2
1
Eskisehir Osmangazi University, Civil Engineering Department, Eskisehir
2
Dumlupinar University, Civil Engineering Department, Kutahya
Turkey
1. Introduction
Important lessons which reflect the seismic performance of dams under large earthquakes
are available in the literature, (1971, San Fernando earthquake; 1985, Mexico earthquake;
1999, Kocaeli earthquake; 2001, Bhuj earthquake; 2008, Wenchuan earthquake). Seismic
behavior of dams subjected to these severe earthquakes shows that earthquake safety of
dams is an important phenomenon in dam engineering and requires a more comprehensive
seismic studies.
Dams built on the site with high seismicity have a high-risk potential for downstream life
and property. Active faults near dam sites can cause to damaging deformation of the
embankment. In general, strong ground shaking can result instability of the dam and
strength loss of foundations. (Seed et al., 1969; Seed et al., 1975; Jansen, 1988; Castro et
al.,1985). In the last decade, large earthquakes have killed many thousands of people and
caused economic devastation, commonly as a result of building failures in seismic events.
Therefore, meaningful seismic parameters are needed to perform a satisfactory evaluation of
dam structure (Tosun, 2002).
ICOLD (1989) stated that safety concerns for embankment dams subjected to earthquakes
involve either the loss of stability due to a loss of strength of the embankment of foundation
materials or excessive deformations such as slumping, settlement, cracking and planer or
rotational slope failures. To obtain preliminary information about seismic parameters, the
simplified procedures can be used. If the materials used in embankment are not susceptible
to loss of strength and the hazard and risk ratings are low, the simplified analyses are
entirely sufficient to define the seismic evaluation parameters. The safety concerns for
concrete dams subjected to earthquakes involve evaluation of the overall stability of the
structure, such as verifying its ability to resist induced lateral forces and moments and
preventing excessive cracking of the concrete. For analyzing the loads, different procedures
are performed. In the simplified analyses, peak ground motion parameters and response
spectra are sufficient to define the seismic evaluation parameters. It is suggested the finite
element method to be used for analyzing of most dams in high risk or hazard class.

168
The seismic hazard study at a dam site basically depends on the seismicity of a region, the
types of structures involved and the consequences of failure. FEMA (2005) states that the
design and evaluation of dams for earthquake loading should be based on a comparable
level of study and analyses for each phase of the study including seismo-tectonic, geological,
geotechnical and structural investigations. The level of study should reflect both the
criticality of the structure and the complexity of the analysis procedures. Basic seismic
studies for earthquake safety assessment of dams generally rely on existing seismological
studies, available site data and simplified methods of design or evaluation developed for
similar projects or structures. In other words, these studies use preliminary values of the
ground motions obtained from existing studies, a simplified structural analysis and a
general assessment of soil liquefaction and deformation. Detailed seismic studies involve the
use of site-specific studies in evaluating the earthquake hazard and dynamic analyses for
determining the response of project features to seismic loading. Detailed geological studies
should define the seismic tectonic province, characterize the site, and investigate all faults
that can be the source of ground shaking at the dam site.
Extensive field exploration and testing programs are necessary to perform earthquake safety
assessment. The earthquake history, earthquake recurrence relationship and strong motion
records should be defined by seismological investigation (Erdik et al, 1985). Structural
investigations should consider all relevant factors that affect the seismic hazard at the
specific site and the actual dynamic performance of the structure. Geotechnical studies
should relieve the types and spatial distribution of foundation and embankment materials
and the engineering properties of soil and rock, liquefaction potential of the foundation and
embankment soils, stability of natural and artificial slopes and estimation of deformations.
The final results of all studies should be used as a basis for making design or evaluation
decisions and for designing remedial measures.
2. Selection of design and safety evaluation earthquake
The selection of site-dependent seismic input is an important stage for determining the
safety evaluation of dam structures. The earthquakes have been specified by different
terms such as the Operating Basis Earthquake (OBE), the Maximum Credible Earthquake
(MDE) and the Safety Evaluation Earthquake (SEE). These earthquakes have been defined
by separate organization with different value. FEMA (2005) has meaningfully defined
those earthquakes in recent. Their short definitions are given below as based on this
guideline of FEMA.
2.1 Operating Basis Earthquake (OBE)
This earthquake is defined for the ground motions at the site to be expected to occur within
the service life of the project. FEMA (2005) states that the associated performance
requirement is that the project function with little or no damage and without interruption of
function. The OBE means to protect against economic losses from damage or loss of service.
Consequently, the return period of the OBE can be based on economic considerations.
2.2 Maximum Credible Earthquake (MCE)
According to FEMA (2005), this is the largest earthquake magnitude that could occur along
a recognized fault or within a particular seismo-tectonic province or source area under the

Selection of the Appropriate Methodology for Earthquake Safety Assessment of Dam Structures

169
current tectonic framework. The loading resulting from MCE can be exceeded for
probabilistic methods, which is discussed later on, for high return period faults close in,
such as North Anatolian Fault in Turkey and San Andreas Fault in USA.
2.3 Maximum Design Earthquake (MDE) or Safety Evaluation Earthquake (SEE)
The MDE is the earthquake that produces the maximum level of ground motion for which a
structure is to be designed, while SEE is defined just for safety evaluation. These
earthquakes may be considered as earthquakes which are equal to the MCE or to a design
earthquake less than the MCE. As based on the FEMA (2005), the associated performance
requirement for the MDE or SEE is that the project performs without catastrophic failure,
such as uncontrolled release of a reservoir, although significant damage or economic loss
may be tolerated.
Earthquake ground motions at a particular site are estimated through a seismic hazard
evaluation. The geologic and seismologic inputs needed for completing a seismic hazard
evaluation. Two different methods are widely used for describing earthquake ground
motions for seismic design. These are the deterministic seismic hazard analysis (DSHA) and
the probabilistic seismic hazard analysis (PSHA). According to Kramer (1996), DSHA
considers a seismic scenario that includes a four-step process. This procedure gives rational
solutions for large dams because it provides a straightforward framework for evaluation of
the worst situation. The DSHA procedure is outlined as below and introduced schematically
in Figure 1.

Fig. 1. Four steps of deterministic seismic hazard analysis (Kramer, 1996)
1. Identification and characterization of all earthquake sources capable of producing
significant ground motion at the site.

170
2. Selection of a source-to-site distance parameter for each source zone.
3. Selection of the controlling earthquake.
4. Definition of hazard at the site in terms of the ground motions produced at the site by
controlling earthquake.
PSHA has allowed the uncertainties in the size, location and rate of recurrence of
earthquakes, as well as in the variation of ground motion characteristics with earthquake
size and location, to be explicitly considered in the evaluation of seismic hazards (Figure 2).
This method is generally similar with DSHA and outlined as follows:
1. Identification and characterization of earthquake sources.
2. Characterization of seismic activity and temporal distribution of earthquake recurrence.
3. Determination of ground motion with use of predictive relationships.
4. Prediction of ground motion parameters

Fig. 2. Four steps of probabilistic seismic hazard analysis (Kramer, 1996)
DSHA and PSHA have been performed by different researchers in specific sites according to
their project goals and importance (Chandler et al. 2001; Fat-Helbary and Tealb 2002; Al-
Homoud, 2003; Ardeleanu et al. 2005; Simeonova et al. 2006; Orhan et al. 2007; Nakajima et
al. 2007; Tosun and Seyrek, 2006; Tosun et al. 2007; Seyrek and Tosun, 2011).
Both probabilistic and deterministic methods have a role in hazard and risk analyses
performed for decision-making purposes. One method may have priority over the other,
depending on the seismic environment and the scope of the project (McGuire 2001).
McGuire (2001) claims that the analysis of a specific site usually requires a probabilistic
approach, but a deterministic check on the resulting decision is appropriate. Generally
seismic sources contribute to the seismic hazard and risk at a site, and the integration of
these through a probabilistic analysis provides the most insight.


171
Orozova and Suhadolc (1999) express that all the earthquakes with different magnitudes
and distance influences the seismic hazard at a site and PSHA correctly reflects the actual
knowledge of seismicity. Another advantage of probabilistic approach is that offers a
rational framework for risk management by considering the frequency or probability of
exceedance of the ground motion against which a structure or facility is designed (Bommer
and Abrahamson 2006).
A severe criticism of PSHA came from Castanos and Lomnitz (2002), who consider that the
problem with PSHA is that its data are inadequate and its logic is defective. They suggest
that the deterministic procedures especially when coupled with engineering judgment to be
more reliable and more scientific.
3. Methods of analysis
In this section, seismic hazard analysis results for two different methodology presented in
previous section will be discussed by using acceleration values of dam site locations
subjected to different seismicity and geological setting.
For the seismic hazard analysis of the dams in Turkey, all possible seismic sources were
identified and their potential was evaluated in detail, as based on the guidelines given by
Fraser and Howard (2002). Various seismic source models and active fault maps have been
previously reported in Turkey (Ycemen, 1982; Erdik et al. 1985; Erdik et al. 1999; arolu et
al. 1992; Kayabal and Akn, 2003; Ulusay et al. 2004). These seismic-source models have
been modified taking advantage of recent neotectonic and seismic data for Turkey (Figure 3).
The data about 20th century instrumentally recorded earthquakes for Turkey and vicinity

Fig. 3. Seismo-tectonic map of Turkey

172
were collected by the Bogazici University Kandilli Observatory and Earthquake Research
Institute, National Earthquake Monitoring Center. It should be noted that moment
magnitude scale is used for hazard calculations.
To research the effect of seismic hazard models on the results, Altinkaya, Ayhanlar, Boztepe,
Hatap ve Kilickaya dam sites are considered (Figure 4). Three of them are located in
Yesilirmak basin and others are in Kizilirmak basin. The neotectonics of the region around
the dams is governed by four major elements: (1) the North Anatolian Fault (NAF), which is
the main structural feature in the basin, (2) the Ezinepazari Fault, which lies in south-west
direction as a secondary feature of NAF, (3) Ecemis Fault, which starts from Mediterranean
Sea at the south and has approximately same direction as Ezinepazari Fault. (4) Shear zone
including secondary faults at the central part of basin (Figure 4).
The North Anatolian Fault is one of the best-known strike-slip faults in the world, because
of its significant seismic activity and well developed surface features. It is approximately
1,500 km long and extends from eastern Turkey at the east to Greece at the west. Its width
ranges from a single zone of a few hundred meters to multiple shear zones of 40 km. This
fault produces very large earthquakes, which have resulted in the death of one thousand
people and severe structural damages. The Ezinepazari fault which is a secondary branch of
NAF has approximately 260 km length and extends to the central part of Anatolia in south-
west direction. Its width can be defined by a single zone of a few hundred meters. This
zone also comprises the Bala fault, which recently generated the earthquake near Ankara
city, and Merzifon fault, which is known as the source of large earthquakes occurred in
the past. The Ecemis fault, presently called as Central Anatolian Fault Zone, is also a
strike-slip fault, which is located at the southern part of basin. It is composed of several
Fault segments and its width ranges from 2 to 15 km. Kocyigit and Beyhan (1997) stated
that this fault connects to the North Anatolian Fault and extends to south as far as the
Hellenic trench, reaching up to several hundred kilometers in length. The shear zone,
which is located from the central part to the south-west of Kzlrmak basin, contains
structural features such as Krsehir, Gumuskent, Delice, Akpinar, and Bala faults. Seyrek
and Tosun (2011) remark that this zone has low seismic activity, although an earthquake
with moderate magnitude occurred in 1938.
During the analysis process, seismic zones and earthquakes within the area having a radius
of 100 km around the dam site were considered. Deterministic and probabilistic seismic
hazard analyses were performed by the computer program DAMHA, which was developed
at the Earthquake Research Center, Eskiehir Osmangazi University. DAMHA is capable of
performing the deterministic and probabilistic seismic hazard analyses on digitized tectonic
map of Turkey. This program includes databases for earthquakes, faults, area seismic
sources and attenuation equations (Seyrek, 2009). This program calculates the probabilistic
hazard for three hazard level as OBE, MDE and SEE. These levels correspond to the return
periods of 144, 475 and 2475 years respectively.
For horizontal peak ground acceleration calculation, eight separate predictive relationships
(Campbell, 1981; Boore et al. 1993; Campbell and Bozorgnia, 1994; Ambraseys et al. 1995;
Boore et al. 1997; Glkan and Kalkan, 2002; Kalkan and Glkan, 2004; Ambraseys et al. 2005)
were considered. It is clear that the use of several attenuation laws can result in more
reliable evaluation than a single relationship (ICOLD 1989). It was noted that Glkan and


173
Kalkan (2002) and Ambraseys et al. 2005 were derived from Turkey earthquakes. Other
attenuation relationships used for this study were selected owing to the similarities between
the mechanisms of North Anatolia Fault Zone and San Andreas Fault.

Fig. 4. Locations of dams on seismo-tectonic map
One of the most important parameters of source seismicity is the size (magnitude) of the
maximum earthquake. The general assumption is that one-third to half of the total length
of the fault would break when it generates the maximum earthquake (Mark, 1977). In
this study, for each source the maximum earthquake magnitude was determined using
the empirical relationships proposed by Wells and Coppersmith (1994) as given in Table
1. Where M
w
is the earthquake moment magnitude and L is the fault rupture length in
km.

Fault type Equation
Strike slip M
w
= 5.16+1.12logL
Reverse M
w
= 5.00+1.22logL
Normal M
w
= 4.86+1.32logL
All M
w
= 5.08+1.16logL
Table 1. Relationship between earthquake magnitude (M
w
) and rupture length (L)
Once the maximum earthquake magnitude is determined for each seismic source, a linear
regression is performed to estimate the coefficients of Gutenberg-Richter (1944) relationship
using the computer program DAMHA. Orhan et al. (2007) express that the records which
have a magnitude equal to or greater than 4.0 is more credible in Turkish earthquake
catalogue.

174
4. Analyses and results
For the seismic hazard analyses of the dam sites, a detailed study was performed. Local
geological features and seismic history referred in previous section were used to quantify
the rate of seismic activity in the basin. To present the effect of methodology on seismic
hazard results, five dam sites are chosen. Below, details of analysis results will be given for
each dam site.
4.1 Altinkaya dam
Altinkaya dam with a storage capacity of 5763 hm
3
is also located on a Kizilirmak river
(Figure 5). It was designed as rockfill dam and its construction was finished in 1988. Its
height from river bed is 195 meter. It produces electricity of 1632 GWh per year with an
installed capacity of 700 MW. Altinkaya dam is one the most important dam projects in
Kizilirmak basin.

Fig. 5. View of the Altinkaya dam
As a result of detailed evaluation, two seismic sources are considered for hazard
calculations. Seismic parameters used for hazard assessment are given in Table 2.
Results of deterministic and probabilistic analyses are given in Table 3. It should be noted
that each PGA value introduced for a dam site in these tables represent the average of those
obtained from eight different attenuation relationships discussed in previous section. Total
seismic hazard curve of Altinkaya dam site is presented in Figure 6. PGA value for SEE level
is 0.25 g.


175
Zone no Fault no Fault name
Fault
type*
M
max
a** b**
3
3-1 North Anatolian Fault Zone Segment SS* 7.7
5.172 0.684
3-2 North Anatolian Fault Zone Segment SS 7.4
18
18-1 Bala Fault N 6.7
5.457 0.878 18-2 Ezinepazari Fault SS 7.9
18-3 Merzifon Fault SS 6.9
* SS : Strike slip N: Normal
**from Gutenberg-Richter (1944) law
Table 2. Hazard parameters of seismic sources for Altinkaya dam site

Critical
Zone
Critical
Segment
Closest
distance (km)
PGA (g)
DSHA PSHA
50
th
percentile 84
th
percentile OBE MDE SEE
3 3-3 39.8 0.11 0.19 0.13 0.18 0.25
Table 3. DSHA and PSHA results for Altinkaya dam site

Fig. 6. Seismic hazard curve for Altinkaya dam site

176
4.2 Kilickaya dam
Kilickaya dam is one of the large dams located in Yesilirmak basin and was constructed on
Kelkit river for energy and flood control purposes. It was designed as rockfill dam and
finished in 1990. When the reservoir is at normal capacity, the facility impounds 1400 hm
3
of
water with a reservoir surface area of 64 km
2
. It has a height of 103 m from river bed. It
produces the electricity with an installed capacity of 124 MW (Figure 7).

Fig. 7. View of the Kilickaya dam
North Anatolian Fault Zone and Malatya Ovacik Fault Zone are taken into consideration for
hazard calculations. Seismic parameters used for hazard assessment are given in Table 4.

Zone no
Fault
no
Fault name Fault type* M
max
a** b**
3
3-1 North Anatolian Fault Zone Segment SS* 7.7
5.172 0.684
15
15-1 Malatya Ovacik Fault Zone SS 7.4
4.084 0.626 15-2 Malatya Ovacik Fault Zone SS 7.2
15-3 Malatya Ovacik Fault Zone SS 7.2
* SS : Strike slip
Table 4. Hazard parameters of seismic sources for Altinkaya dam site


177
The analyses show that the most critical zone is North Anatolian Fault Zone and closest
distance from this seismic source to dam site is 11.0 km. PGA value for 84
th
percentile is
greater than SEE level (Table 5). Total seismic hazard curve of Altinkaya dam site is
presented in Figure 8. PGA value for 84
th
percentile is greater than SEE level.

Critical
Zone
Critical
Segment
Closest
distance
(km)
PGA (g)
DSHA PSHA
50
th
percentile
84
th

percentile
OBE MDE SEE
3 3-4 11.0 0.37 0.60 0.17 0.26 0.40
Table 5. DSHA and PSHA results for Kilickaya dam site

Fig. 8. Seismic hazard curve for Kilickaya dam site
4.3 Hatap dam
Hatap dam with a body volume of 1.25 hm
3
was designed as rockfill dam with a clay core.
The construction of the dam was finished in 2009. It is located on the Hatap creek and has

178
42-m height from river bed (Figure 9). Main purposes are irrigation and water supply. When
the reservoir is at normal capacity, the facility impounds 11.6 hm
3
of water with a reservoir
surface area of 1.02 km
2
.
The project site are affected from source zone 3, 17, 18 and 32. Seismic parameters used for
hazard assessment are given in Table 6.

Zone
no
Fault
no
Fault name
Fault
type*
M
max
a** b**
3
3-1
North Anatolian Fault Zone
Segment
SS* 7.7
5.172 0.684
3-2
Segment
SS 7.4
3-3
Segment
SS 7.4
3-4
Segment
SS 7.9
3-5
Segment
SS 7.6
17
17-1 Sorgun Fault SS 7.2
2.655 0.413
17-2
Sarikaya Akdagmadeni
Fault
SS 7.3
18
18-1 Bala Fault N 6.7
5.457 0.878 18-2 Ezinepazari Fault SS 7.9
18-3 Merzifon Fault SS 6.9
32
32-1 Gumuskent Fault SS 7.4
3.054 0.476
32-2 Kirsehir Fault SS 6.6
32-3
Akpinar-Kirsehir Fault
Zone
SS 6.7
32-4
Akpinar-Kirsehir Fault
Zone
SS 6.9
32-5 Delice(Yerkoy) Fault SS 7.2
*SS: Strike slip N: Normal U: Unknown
Table 6. Hazard parameters of seismic sources for Hatap dam site
Seismic hazard analysis results are presented in Table 7. The critical segment is Ezinepazari
Fault with a closest distance of 17.7m from dam site. Figure 10 gives the probabilistic hazard
curve based on different attenuation relationships. Average PGA values are 0.16, 0.24 and
0.37 g respectively.


179

Fig. 9. View of the Hatap dam

Fig. 10. Seismic hazard curve for Hatap dam site

180

Critical
Zone
Critical
Segment
Closest
distance
(km)
PGA (g)
DSHA PSHA
50
th
percentile
84
th

percentile
OBE MDE SEE
18 18-2 17.7 0.27 0.45 0.16 0.24 0.37
Table 7. DSHA and PSHA results for Hatap dam site
4.4 Ayhanlar dam
Ayhanlar dam is located on Kiziloz creek with a storage capacity of 21.8 hm
3
(Figure 11). It
was designed as earthfill dam and its construction was finished in 2003. Its body volume is
1.2 hm
3
and height from river bed is 35 m. Main purpose is to perform irrigation.

Fig. 11. View of the Ayhanlar dam
From seismo-tectonic map of the investigation area, four zones with eleven segment are
included into the seismic hazard analyses. Details of the seismic hazard parameters of each
source are given in Table 8.
PGA values of dam site are given on the basis of deterministic and probabilistic approach
(Table 9). The critical segment is Gumuskent Fault with a magnitude of 7.4. Figure 12
presents the probabilistic hazard results by means of seismic hazard curve. PGA values for
return period of 475 and 2475 years are 0.16 and 0.30g respectively.


181
Zone
no
Fault
no
Fault name
Fault
type**
M
max
a*** b***
16
16-1 Ecemis Fault Zone Segment SS 7.4
4.805 0.814
16-3 Derinkuyu Fault N 6.5
16-4 Karsanti-Karaisali Fault Zone SS 7.2
16-5 Deliler Fault SS 7.3
17
17-1 Sorgun Fault SS 7.2
2.655 0.413
17-2 Sarikaya Akdagmadeni Fault SS 7.3
32
32-1 Gumuskent Fault SS 7.4
3.054 0.476
32-2 Kirsehir Fault SS 6.6
32-3 Akpinar-Kirsehir Fault Zone SS 6.7
32-4 Akpinar-Kirsehir Fault Zone SS 6.9
32-5 Delice(Yerkoy) Fault SS 7.2
33 33-1 NA* U 6.8 4.112 0.846
* NA: Non available
**SS: Strike slip N: Normal U: Unknown
***from Gutenberg-Richter (1944) law
Table 8. Hazard parameters of seismic sources for Hatap dam site

Critical
Zone
Critical
Segment
Closest
distance
(km)
PGA (g)
DSHA PSHA
50
th
percentile
84
th

percentile
OBE MDE SEE
32 32-1 6.9 0.38 0.64 0.09 0.16 0.30
Table 9. DSHA and PSHA results for Hatap dam site

182

Fig. 12. Seismic hazard curve for Ayhanlar dam site
4.5 Boztepe dam
Boztepe dam with a storage capacity of 14.2 hm
3
is also located on a Boztepe creek (Figure
13). It was designed as earthfill dam and its construction was finished in 1984. Its height
from river bed is 27 meter. Main aim of this dam is irrigation.

Fig. 13. View of the Boztepe dam


183
Five seismic source zone are included into the analyses. Seismic parameters used for hazard
assessment are given in Table 10.

Zone
no
Fault
no
Fault name
Fault
type*
M
max
a** b**
8
8-1 Elbistan Fault SS 6.9
5.050 0.831
8-2 Surgu Fault SS 6.6
10
10-1
East Anatolian Fault Zone
Segment
SS 7.0
5.585 0.925
10-2
Segment
SS 6.8
11
11-1
Segment
SS 7.0
5.882 0.948
11-2
Segment
SS 6.6
11-3
Segment
SS 7.0
11-4
Segment
SS 6.7
11-5 Tut Fault SS 6.3
14
14-1 Karatas-Osmaniye Fault Zone SS 6.8
5.280 0.818
14-2 Karatas-Osmaniye Fault Zone SS 6.5
16
4.805 0.814
16-3 Derinkuyu Fault N 6.5
16-4 Karsanti-Karaisali Fault Zone SS 7.2
16-5 Deliler Fault SS 7.3
*SS: Strike slip N: Normal U: Unknown
***from Gutenberg-Richter (1944) law
Table 10. Hazard parameters of seismic sources for Boztepe dam site

184
Results of deterministic and probabilistic analyses are given in Table 11. Average PGA
values from eight different attenuation relationships are seen in this table. Total seismic
hazard curve of Boztepe dam site is presented in Figure 14. PGA value for SEE level is
0.24 g.

Critical
Zone
Critical
Segment
Closest
distance
(km)
PGA (g)
DSHA PSHA
50
th
percentile
84
th

percentile
OBE MDE SEE
10 10-2 27.6 0.11 0.19 0.13 0.18 0.24
Table 11. DSHA and PSHA results for Boztepe dam site

Fig. 14. Seismic hazard curve for Boztepe dam site
5. Discussions
Turkey, which has at least 1200 large dams, is one of the most seismically active regions in
the world and major earthquakes with the potential of threading life and property occur
frequently here. It is obvious that many of the large dams planned in Turkey are located in


185
zones of moderate-high seismicity. Thus it is very important that dams to be resistant to the
strong earthquakes. Many of guidelines about earthquake safety of dams promise different
methodology for dam site locations with different seismicity.
The seismic hazard of a dam site is based on the peak ground acceleration. This value
derived from the defined design earthquake produces the main seismic loads. For
preliminary study, the existing map of seismic zones can be used to estimate the seismic
hazard of a dam site. However, authors believe that the detailed seismic hazard analyses
should be performed for safety evaluation of existing dams. Because, the main
requirement of an earthquakeresistant design of a dam is to protect public safety and
property. Therefore, seismic criteria and analysis parameters for dams should be selected
more conservatively than for conventional structures since the failures are more
disastrous.
To reveal the effects of methodologies on seismic hazard results, five dam sites are chosen
and repetitive analyses are performed. Deterministic and probabilistic methods give
different results for each dam sites. It is obvious that deterministic method (84
th
percentile)
gives the maximum value for dams within a near seismic source zone. For the dams which
are not close to the seismic sources, deterministic PGA value is lower than SEE level. But it is
necessary to perform a lot of examination for locations with low to high seismicity. Another
important effect on these differences is sensitiveness of attenuation equations to distance
and earthquake magnitude. For further study, analysis results should be compared for each
attenuation equation.
Authors state that seismic performance of dams within the near source zone must be
reevaluated in detail. Large dams which have high-risk class should be evaluated with
highest priority as a part of the National Dam Safety Program. Both DSHA and PSHA must
be performed for these dams.
6. References
Al-Homoud, A., 2003, Evaluation of strong motion acceleration for embankment dam
design considering local seismotectonics, Natural Hazards, 29, p: 37-56.
Ambraseys, N.N., 1995, The prediction of earthquake peak ground acceleration in
Europe,Earthquake Engineering and Structural Dynamics, Vol. 24, 467-490.
Ambraseys, N.N., Douglas, J., Karma, S.K. and Smit, P.M., 2005, Equations for the
estimation of strong ground motions from shallow crustral earthquakes using data
from Europe and the Middle East: horizontal peak ground acceleration and spectral
acceleration, Bulletin of Earthquake Engineering, 3, 1-53.
Ardeleanu, L., Leydecker, G., Bonje,r K., Busche, H., Kaise,r D., Schmitt, T., 2005,
Probabilistic seismic hazard map for Romania as a basis for a new building code.
Natural Hazards Earth Syst Sci 5:679684.
Bommer, JJ, Abrahamson, N., 2006, Why do modern probabilistic seismic hazard analyses
often lead to increased hazard estimates? Bull Seismol Soc Am 96(6):19671977.

186
Boore, D.M., Joyner, W.B. and Fumal, T. E., 1993, Estimation of response spectra and peak
accelerations from Western North American Earthquakes: An interim report, U.S.
Geol. Surv. Open-File Rept. 93-509, 72 p.
Boore, D.M., Joyner, W.B. and Fumal, T.E., 1997, Equations for estimating horizontal
response spectra and peak acceleration from Western North American earthquakes:
A summary of recent work, Seis. Res. Let., 68, 128-153.
Campbell, K.W., 1981, Near-source attenuation of peak horizontal acceleration. Bulletin
Seism. Soc. Am., vol. 71, N.6, 2039-2070.
Campbell, K.W. and Bozorgnia, Y., 1994, Near-source attenuation of peak horizontal
acceleration from worldwide accelerograms recorded from 1957 to 1993, Proc.
Fifth U.S. National Conference on Earthquake Engineering 3, Chicago, Illinois,
283-292.
Castanos, H. and Lomnitz, C., 2002, PSHA: is it science?, Engineering Geology, 66, 315-
317.
Castro, G., Poulos, S.J., Leathers, F. 1985, Re-examination of Slide of Lower San Fernando
Dam. ASCE, Journal of Geotechnical Engineering, V.111.
Chandler, A.M., Chan, L.S. and Lam, N.T.K., 2001, Deterministic seismic hazard parameters
and engineering risk implications fort he Hong Kong region, Journal of Asian Earth
Sciences, 20, 59-72.
Erdik, M., Doyuran, V., Glkan, P., Akka, N. 1985, Evaluation of Earthquake Hazard in
Turkey with Statistical Approach. Middle East Technical University Earthquake
Engineering Research Center, Ankara, 116 p, (in Turkish).
Erdik, M., Biro, Y.A., Onur, T., Sesetyan, K. and Birgoren, G., 1999, Assessment of
earthquake hazard in Turkey and neighboring rgions, Annali Di Geofisica, Vol. 42,
No. 6, 1125-1138.
Fat-Helbary, R., Tealb, A.A., 2002, A study of seismicity and earthquake hazard at the
proposed Kalabsha dam site, Aswan, Egypt, Natural Hazards, 25, p: 117-133.
FEMA 2005, Federal Guidelines for Dam Safety-Earthquake Analyses and Design of
Dams.
Fraser, W.A. and Howard, J.K., 2002, Guidelines for use of the consequence-hazard matrix
and selection of ground motion parameters, Technical Publication, Depatment of
Water Resources, Division of Safety of Dam, 7p.
Glkan, P. and Kalkan, E., 2002, Attenuation modeling of recent earthquakes in Turkey,
Journal of Seismology, 6, No. 3, 397-409.
ICOLD 1989, Selecting Parameters for Large Dams-Guidelines and Recommendations.
ICOLD Committee on Seismic Aspects of Large Dams, Bulletin 72.
Jansen, R.B. (ed) 1988, Advanced Dam Engineering for Design, Construction and
Rehabilitation. Van Noswtrand Reinhold, New York, 884p.
Kalkan, E., Glkan, P., 2004, Site-dependent spectra derived from ground motion records in
Turkey. Earthquake Spectra 20(4):11111138.
Kayabal, K. and Akn, M., 2003, Seismic hazard map of Turkey using the deterministic
approach, Engineering Geology, 69, 127-137.
Kocyigit, A., Beyhan, A., 1997, A new intracontinental transcurrent structure-the central
anatolian fault zone, Turkey. Tectonophysics 284:317336.


187
Kramer, S.L., 1996, Geotechnical earthquake engineering, Prentice Hall, New Jersey, 653
p.
Mark, R.K., 1977. Application of linear statistical models of earthquake magnitude
versus fault length in estimating maximum expectable earthquakes. Geology 5,
464 466.
McGuire, R.K., 2001, Deterministic vs. probabilistic earthquake hazards and risks, Soil
Dynamics and Earthquake Engineering, 21, 377-384.
Nakajima, M., Choi, I., Ohtori, Y. And Choun, Y., 2007, Evaluation of seismic hazard curves
and scenario earthquakes for Korean sites based on probabilistic seismic hazard
analysis, Nuclear Engineering and Design, 237, 277-288.
Orhan, A., Seyrek, E. and Tosun, H., 2007, A probabilistic approach for earthquake hazard
assessment of the Province of Eskisehir, Turkey, Nat. Hazards Earth Syst. Sci., 7,
607614.
Orozova, I.M. and Suhadolc, P., 1999, A deterministic-probabilistic approach for seismic
hazard assessment, Tectonophysics, 312, 191-202.
Seed, H.B., Lee, K.L., and Idriss, I.M. 1969, Analysis of Sheffield Dam failure. Journal of Soil
Mechanics and Foundations-ASCE, 95(SM6), 14531490.
Seed, H.B., Lee, K.L., Idriss, I.M., and Makdisi, F.I. 1975, The slides in the San Fernando
Dams during the Earthquake of February 9, 1971, J. Geotech. Eng.-ASCE, 101(GT7),
651688.
Simeonova, S.D., Solakov, D.E., Leydecker, G., Busche, H., Schmitt, T., Kaiser, D., 2006,
Probabilistic seismic hazard map for Bulgaria as a basis for a new building code.
Nat Hazard Earth Syst Sci 6:881887.
Seyrek, E., 2009, Numerical solution models for seismic hazard analysis of dam sites and a
case study. PhD thesis, Eskisehir Osmangazi University
Seyrek, E., Tosun, H., 2011, Deterministic approach to the seismic hazard of dam sites in
Kizilirmak Basin, Turkey, Natural Hazards, 14 p. (DOI 10.1007/s11069-011-9795-7,
Published online 01 April 2011)
Saroglu, F., Emre, O., Kuscu, I., 1992, Active Fault Map of Turkey. General Directorate of
Mineral Research and Exploration, Ankara, Turkey
Tosun, H. 2002, Earthquake-Resistant Design for Embankment Dams. Publication of
General Directorate of State Hydraulic Works, Ankara, 208 p.(in Turkish).
Tosun, H., Seyrek, E., 2006, Seismic studies on Turkish dams, International Water
Power&Dam Construction, February 2006, 20-23.
Tosun, H., Zorluer, ., Orhan, A., Seyrek, E., Trkz, M., Sava, H., 2007, Seismic hazard
and total risk analyses for large dams in Euphrates Basin in Turkey, Engineering
Geology, Volume 89, Issues 1-2 , 12 January 2007, 155-170.
Ulusay, R., Tuncay, E., Snmez, H. and Gkeolu, C., 2004, An attenuation relationship
based on Turkish syrong motion data and iso-acceleration map of Turkey,
Engineering Geology, 74, 265-291.
Wells, D.L. and Coppersmith, K.J., 1994, New empirical relationships among magnitude,
rupture length, rupture width, rupture area, and surface displacement, Bulletin of
the Seismological Society of America, Vol. 84, No. 4, 974-1002.

188
Ycemen, S., 1982, Seismic Risk Analysis, Publication of Middle East Technical University,
Ankara (in Turkish)
8
Earthquake Response Analysis and
Evaluation for Earth-Rock Dams
Zhenzhong Shen, Lei Gan, Juan Cui and Liqun Xu
State Key Laboratory of Hydrology-Water Resources
and Hydraulic Engineering, Hohai University
China
1. Introduction
An earthquake is any shaking of the ground, usually intense shaking of the ground - caused
by either natural sources or by humans. There are many different types of earthquakes. The
most common is tectonic earthquake, which occur when rocks in the earth's crust break due
to geological forces created by movement of tectonic plates. Another type of earthquake is
called volcanic earthquake, occurring in conjunction with volcanic activity. And also there
are various types of earthquakes caused by man-made activities, which caused directly by
human involvement, has been recorded as a result of water filling large dams, development
of mineral, geothermal and hydrocarbon resources, waste injection, underground nuclear
explosions and large-scale construction projects. This type of earthquake includes collapse
earthquake, explosion earthquake and induced earthquake due to water filling large dams
(Talwani, 1997; Chen & Talwani, 1998) and so on. Earthquake is the rapid vibration of the
earth surface whose acceleration motion can easily cause the destruction of hydraulic
structures. The relative deformation of the hydraulic structures increases when the inertial
force caused by the earthquake increases suddenly. For the hydrodynamic pressure acting
on the reservoir, dam foundation may appear relative deformation or different stage
deformation. For example, those parts of concrete dams, including dam crest, fold slope,
orifice inside the dam and corridor, are easily cracked, which may damage and then cause
the leakage of the dam body, even the collapse of the structures after the earthquake. As it to
the earth-rock dam, it is easy to appear cracks in dam in the process of the earthquake,
meanwhile, dam foundation or dam body may collapse caused by earthquake liquefaction if
the sands inside them are not fully roller compaction. Thus earthquake may cause the
unsafe of the hydraulic structures, even dam breach which may threatened the life and
property safety of the people downstream the river.
On May 12, 2008, a magnitude destructive earthquake (M
s
=8.0), with the epicenter in
Wenchuan County, struck Sichuan province, in southwestern China, killing more than
100,000 people. The earthquake was the largest and most destructive to strike China since
the 1976 Tangshan earthquake, which killed more than 240,000 people (Pei et al., 2010; Wang
et al., 2010; Cao et al., 2011). On January 12, 2010, a massive 7.3-magnitude earthquake has
struck the Caribbean nation of Haiti. The earthquake struck about 15km (10 miles) south-

190
west of Port-au-Prince which is the capital city of Haiti. Approximately 230,000 people died
immediately or during ensuing weeks, mostly due to acute trauma (Kenneth, 2011). The
most devastating earthquake (M
s
=9.0) in Japan after the 1923 Great Kanto earthquake hit the
eastern Japan in the afternoon of March 11, 2011. Nearly 30,000 people were killed or missed
in the Tohoku earthquake and the ensuing monster tsunami (Takewaki et al., 2011). In this
paper, the non-linear FEM is adopted to calculate the Wenchuan earthquake response. Then,
the earthquake-resistance safety of the Bikou dam is evaluated.
1.1 Bikou earth core dam
The Bikou hydropower station is located on the Bailong River in Gansu province of China,
where is about 240km away from the epicenter of Wenchuan earthquake, China, May 12,
2008(Chen et al., 2008; Yu et al., 2009; Lin et al., 2009). The Bikou earth core rockfill dam is
a 100m graduate earth-rock dam (Gu, 1989). According to the earthquake loss survey and
preliminary report of the Bikou hydropower project suffered 5.12 Wenchuan
earthquake, some damage appeared inside Bikou dam, but no dam slope slide and
instability damaged phenomenon happened after the earthquake (Zhou et al., 2009; Pan,
2009).The normal water level of the reservoir is 707.0m, and its relative reservoir capacity
is 5.2110
8
m
3
. The dam is earth core rockfill dam with height of 101.8m, where the typical
cross section of the dam is shown as in Fig.1. The ratio of the upstream slope of the dam
are 1:1.8 and 1:2.3 respectively, and the downstream slope are 1:1.7, 1:1.85, 1:2.2 and 1:1.5
respectively. Its width of dam crest is 7.6 m at elevation 710.00m, and the crest length of
dam is about 297.36 m. The dam is located in a roughly symmetric valley of V-type, the
slope of river bank is about 50-60 degree, and the one of right bank is about 35-45 degree.
The depth of overburden layer on the riverbed is about 25-34 m. The core wall in the
riverbed area is built on the natural sandy gravel layer, the width of core wall crest is
4.0m at elevation 709.00-710.00m, and the ratio of the upstream slope of the core wall are
1:0.2 and 1:0.25 respectively, the section of the dam along the dam axis is shown as in
Fig.2. Two concrete cut-off walls are adopted to connect the soil core wall and the
bedrock. The first one built along the dam axis inserts the bedrock with the thickness
1.30m and crest the elevation 630.00m. And the second concrete one is located at the dam
axis downstream 12m, inserting the bedrock with the thickness 0.80m and the crest
653.00m. The spillway buildings are made up of one spillway on the right bank, one
desilting tunnel and two spillway tunnels in the left bank.
Design W.L
703.00
Normal W.L
707.00
Check W.L
708.80
715.30
710.00(dam crest)
691.00
670.00
661.190
Design W.L
627.55
Check W.L
628.90
Normal W.L 617.02
Dead W.L
685.00
624.00
647.00
663.71
680.00
cut-off wall
thickness 130cm
overburden
cut-off wall
thickness 80cm
overburden
624.00
658.00
loam
core
wall
dam axis

Fig. 1. Typical cross section of the dam

Earthquake Response Analysis and Evaluation for Earth-Rock Dams

191
left bank
right bank
730
720
710
700
690
680
670
660
650
640
730
720
710
700
690
680
670
660
650
640
714.30
715.30
dam crest
710.00 normal water level 707.00
spillway
loam core wall
original ground
cut-off wall line
curtain grouting line
653.00 the first cut-off wall line
the second cut-off wall line
625.50

Fig. 2. Section of the dam along the dam axis
1.2 Geological conditions
In the dam site area, the fractures are developed, mainly are the bedding extrusion fracture
zones and the slow obliquity faults. Most of the fractures are the tension and twist fractures
of steep obliquity, and the slow obliquity fractures in the SN, NNE and EW directions. The
faulted structures are mainly as follow: the fault F9, F1, F6 and F14, broken bandwidth 0.3 ~
3.0m, mainly for fault mud and rock. The fracture zones are mainly made up of fault mud
and rock, with width of 0.3-3.0m.
2. Calculation model
2.1 Static analysis model
2.1.1 Control equation of FEM
According to the displacement solution, the basic balance equation of the nonlinear finite
element method is as follows
( ) { } { } K u u R ( =

(1)
Where ( ) (

K u is overall strength degree matrix, { } u node displacement array and{ } R node
load array.
This equation can be solved by the incremental early strain gauge method, where the basic
balance equation holds:
| |{ } { } { }
0
A = A + A K u R R (2)
Here, { } Au is nodal displacement incremental array, { } AR node load incremental array and
{ }
0
AR the equivalent node load array of initial strain.

192
In the static analysis, the soil materials and overburden layer materials are regarded as
elastic-plastic object, where the Duncan-Zhang's nonlinear-elastic (E-B) model (Duncan &
Chang, 1970) is carried out. The concrete and bedrock are regarded as linear elastic objects,
which obey the generalized Hook's law.
2.2 Dynamic analysis model
2.2.1 Control equation of FEM
The dynamic equilibrium equation can be described as follows after the calculation domain
is discretized by the finite element method

| |{ } | |{ } | |{ } ( ) { }
+ + =

o o o M C K F t
(3)
Here, , ,

o o o are nodal displacement, nodal velocity and nodal acceleration respectively,
( ) { }
F t nodal dynamic loading, | | M mass matrix, | | K strength degree matrix, | | C damping
matrix obtained by| | | | | | = + e e C M K , where e is the first mode for vibration frequency
and the damping ratio.
Eq.(3) can be gradually integral solved by the Wilson linear acceleration method
( Wilson u method), which can instead by the following equations.
{ } { }
( =

o
t
t
K F (4)
| |
| | | |
2
6 3
( = + +

M C
K K
dt dt
(5)

{ } { } | |{ } | |{ } = + +
t
t t
t
F F M A C B (6)

{ } { } { } { }
2
6 6
2

= + +

o o o
t t dt
t dt t dt
A
dt dt
(7)
{ } { } { } { }
3
2 +
2

= +

o o o
t t dt
t dt t dt
dt
B
dt
(8)

{ } { } { }
3
=
o o
t t
t
B
dt
(9)
{ } { } { }
2
6
=
o o
t t
t
A
dt
(10)
The iterative method is adopted with the consideration of the variations of dynamic shear
modulus and damping ratio in calculation process after the average dynamic shear strain
change. In the iteration process the convergence criterion is


193

1
0.1

<
i i
i
G G
G
(11)
Here,
i
G is the new shear modulus and
1 i
G the last shear modulus. The maximum
iteration times is taken for 6 to lest the iterative calculation into dead circulation.
In the dynamic analysis (Mejia et al., 1981a, 1982b; Shen et al., 2006a, 2010b), the equivalent
nonlinear viscoelastic model is applied in the dynamic calculation and analysis, the soil
materials and the overburden layers are assumed to be viscoelastic bodies. They reflect
nonlinearity and hysteretic nature of the dynamic stress-strain relation by using of the
equivalent dynamic shear modulus Gand equivalent damping ratio , which can be
expressed as the function between the equivalent shear modulus and damping ratio and the
dynamic shear strain amplitude. The key point of this model is to confirm the relationship
between the maximum dynamic shear modulus
max
G and the average effective stress
0
' o . In
this paper, the Hardin-Drnevich model (Hardin & Denevich, 1972a, 1972b) is used, where
the dynamic shear modulus and damping ratio can be calculated as follows

max
1
=
+

G
G (12)

max
1
=
+

(13)

0
max
| | '
' =
|
\ .
o
n
a
a
G K p
p
(14)
Where
0
' o is average effective stress,
a
P atmospheric pressure,
'
K modulus coefficient
and n is modulus exponent. Here,
max
G ,
0
' o and
a
p have the same dimension. The
relation curves of dynamic shear strain to dynamic shear modulus and damping ratio
can be obtained by dynamic tri-axial test (Xenaki & Athanasopoulos, 2008; Zegha &
Abdel-Ghaffar, 2009). When having dynamic calculation, the related relation curve can be
inputted directly, and then interpolate and extent the values according to the strain values
for the calculation.
The dynamic model of contact face elements referred to the test results of Hohai University,
China. The relationship between shear rigidity K and dynamic shear strain is described as
follow

max
max
1
=
+
t
f
K
K
MK
(15)
The shear rigidity K and damping ratio have the following relationship:

194

max
max
1
| |
= +
|
\ .

K
K
(16)

0.7
max
= o
n
K C
, tan = t o o
f n
(17)
Where o
n
is the normal stress on contact face, o the angle of internal friction on contact
face,
max
the maximum damping ratio and M, C are the test parameters.
2.3 Stability evaluation
The stress of element is used to calculate the factor of safety and evaluate its stability, thus
the position of slide surface can be obtained. According to the Mohr-Coulomb criterion,
the regions whose local safety factors are less than 1.0 are combined together to obtain the
most dangerous multi-slip surface. The factor of safety on the surface is defined as the
ratio of anti-sliding force to sliding force, and then the relationship between factors of
safety and time can be obtained during the earthquake period. In this way, the anti-
sliding stability is evaluated by stress when considering the unstable duration of
earthquake.
During an earthquake, the dynamic strength of rockfill material is not always less than its
static strength, at least we can assume it is equal to static strength. Taking compressive stress
as positive and tensile stress as negative, after obtaining the static stress and dynamic stress
under an earthquake by finite element method, the local factor of safety of an element can be
calculated by the following formula

( )
1 3
1 3
2 cos 2 sin +
=
o o
o o
d
s
c u
LF
(18)
Where c is cohesion, the angle of internal friction and
1
o ,
3
o are the maximum and
minimum principal stresses.
The
d
u can be calculated by the formula as follow

( )( )
1 3
1
3
+ +
=
o o
d d
d
u
(19)
Here, is Poissons ratio and
1
o
d
,
3
o
d
are dynamic stresses respectively.
3. Working behavior of Bikou earth-rock dam
According to the actual engineering conditions of the Bikou earth core rockfill dam, the
3-D non-linear FEM static and dynamic models of the dam are set up to calculate and
analyze the stress and deformation characteristics of dam, including the deformation and
stress of the dam shell, core wall, two cut-off walls, overburden layers, bedrock and so
on. In order to consider the construction process, the process of filling the dam is divided
into 15 levels to simulate, and the reservoir impoundment process is divided into 3
levels.


195
3.1 Initial stress field of dam
In order to calculate the earthquake responses of dam, the initial stress field of dam must be
determined firstly. Thus, the 3-D finite element model is created to simulate the construction
process of dam and reservoir impounding, and the initial stress field of dam before
earthquake can be obtained.
3.1.1 Finite element model
According to the actual situation of the Bikou dam, the 3-D finite element model of the dam
and its foundation is set up, which has simulated the geometry and material partition of the
dam and its foundation. The control section super-element finite element mesh automatic
division technique is adopted to generate the information of finite element model, and
furthermore the super-element can be encrypted to form finite element. Based on
requirements of structural characteristics, stage loading and forming the super-element grid,
a set of control sections with 19 sections horizontal are selected. Then the dam and its
foundation are discretized, and the supper-element grid is built up. After the super-element
gird is discretized further more, the super-element gird is created whose total nodes are
31523 and total element numbers are 30087. The finite element mesh of the dam and its
foundation and its core wall are shown as in Fig.3.
Select domain of the calculation model. 1) The distance from the upstream boundary to the
dam axis of the river bed section is about 363.40m (approximately 1.0 times of dam height).
2) The distance from the downstream boundary to the dam toe of the river bed section is
around 401.00m. 3) The distance from the left boundary to the right boundary is about
698.18m (approximately 2.0 times of dam height). 4) The vertical distance from the bottom
boundary to the dam foundation surface is about 241.80 m.
The coordinate system is set up as following. The X-axis is along stream from up to down
with zero at dam axis, the Y-axis is along dam axis from right bank to left bank, and the Z-
axis is vertical corresponding to the elevation.

(a) dam body and foundation (b) core wall
Fig. 3. Finite element grid
Flow

196
3.1.2 Material parameters
The Duncan-Chang's nonlinear-elastic (E-B) model is carried out for the core wall and each
district materials of the dam shell. According to indoor tests and engineering experience, the
static material parameters of the dam are shown as in Table 1 and Table 2. And the
parameters of the contact surface are shown as in Table 3.

Material types
3
kN m
E kPa
v
Concrete 24.5 1.810
7
0.20
Bedrock 26.0 1.010
7
0.25
Core wall settlings 24.0 8.010
3
0.35
Table 1. Parameters of linear elastic model

Material types
/kNm
-3

c
/kPa
K
n
f
R
b
K
m A
/
ur
K
ur
n
core wall W 20.7 100 190 0.42 0.81 120 0.86 2.1 450 0.30
S 21.1 100 150 0.42 0.81 115 0.50 1.6 150 0.33
filter W 22.1 0 750 0.40 0.75 460 0.80 4.8 600 0.33
S 23.3 0 650 0.40 0.75 460 0.77 4.4 600 0.33
rockfill W 21.2 0 1000 0.45 0.80 400 0.75 6.0 1500 0.33
S 23.6 0 800 0.45 0.80 380 0.70 5.6 1500 0.35
ballast W 21.4 0 600 0.45 0.80 350 0.69 4.1 860 0.30
S 23.3 0 400 0.45 0.80 300 0.67 3.6 860 0.30
gravel
(dam)
W 22.1 0 900 0.50 0.75 460 0.80 7.5 1700 0.25
S 23.3 0 850 0.50 0.73 460 0.77 7.5 1700 0.23
gravel
(mud)
W 22.5 0 500 0.45 0.74 300 0.79 3.3 900 0.30
S 23.3 0 300 0.45 0.72 200 0.77 2.8 900 0.30
pebble W 17.9 0 850 0.40 0.81 400 0.70 5.2 1300 0.33
S 21.3 0 850 0.40 0.81 400 0.70 5.2 1300 0.33
gravel S 23.9 0 700 0.50 0.70 320 0.78 7.6 1400 0.45
Table 2. Parameters of Duncan-Chang's model


197
Material o

/
s
K
/10
4

s
n
fs
R
s
C
/kPam
-3

Loam /concrete 14.0 2.30 0.69 0.75 0
Sandy gravel / concrete 32.2 4.00 0.65 0.75 0
Loam /Loam 17.3 1.80 0.70 0.88 68
Bedrock /concrete 35.0 6.00 0.20 0.80 0
Bedrock /the settlings 30.0 3.00 0.70 0.80 0
Table 3. Parameters of contact surface
3.1.3 Displacement and stress field of dam
By use of the stage loading method to simulate the construction process of dam and
reservoir impounding, the displacement field and stress field of dam under normal water
level condition are obtained. Here, as examples, the displacement and stress distribution on
the maximum transverse cross section of dam (Y=210m) and dam axis maximum
longitudinal section of dam (X=-10m) are shown as in Fig.4 ~Fig.6.
Here, the displacement along the coordinate axis direction is positive, that is, the horizontal
displacement along the flow with the direction of upstream point to the downstream is
positive, the positive horizontal displacement along the dam axis direction is from the right
bank to the left, and the positive vertical displacement is from bottom to up. The unit of the
displacement is mm in some figures. The compressive stress is positive, and the tensile
stress is negative. The unit of the stress is kPa in some figures.
1. Dam shell
Under normal water level condition, the maximum horizontal displacement pointing to the
upstream and the downstream respectively is -289mm and 1132mm.The maximum value
appears in the lower crust, close to the slope. The dam horizontal displacement distribution
along the dam axis direction shows that the horizontal displacement value along dam axis
direction is small. And the maximum settlement occurs in the upstream dam shell which is
near the dam axis in the middle of the river. The maximum vertical displacement of the dam
body is 1273mm, accounts for about 1.2% of the maximum height.
The maximum first principal stress of the dam body is 2534kPa, the maximum second
principal stress is 2445kPa and the maximum third principal stress is 1842kPa. The
maximum principal stress of the upstream and downstream dam shell occurs at the
bottom of dam near the dam axis. The shear stress level of most rock-fill units are less
than 0.85, no the shear failure zone appears in the dam body. It indicates that the dam is
stable.

198
2. Core wall
The maximum horizontal displacement pointing to the downstream of the loam core wall is
812mm.The maximum horizontal displacements of core wall along dam axis direction is
181mm, pointing to the left bank. And the maximum vertical displacement of core wall is
1413mm, appearing in the upstream of the core wall near the dam axis where the elevation
is 657.00m at the deepest valley section. The maximum first principal stress of the core wall
is 1839kPa, the maximum second principal stress is 1373kPa, and the maximum third
principal stress is 1256kPa.

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(a) X-displacement
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(b) Y-displacement
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(c) Z-displacement

Fig. 4. Displacement distribution on section Y=210m (unit: mm)
Under normal water level condition, although core wall stress is smaller than depletion
layer stress, but no tensile stress appeared, so the loam core wall won't produce pull crack.
Meanwhile, for the strength and modulus of core wall is lower and the core wall is easy to
adapt to deformation, core wall stress levels are lower. It indicates that the core wall is
stable.


199
3. Cut-off wall
The maximum horizontal displacement of the first cut-off wall is 469mm, pointing to the
downstream, and the vertical displacement is -112mm, and that of the second cut-off wall
are 583mm and -129mm respectively. The maximum first principal stress of the first
concrete cut-off wall is 9806kPa, the maximum third principal stress is 714kPa, and that of
the second one are 16844kPa and 1094kPa respectively.

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(a) the first principal stress (unit: kPa)
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(b) the second principal stress (unit: kPa)
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(c) the third principal stress (unit: kPa)
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(d) stress level
Fig. 5. Stress distribution on section Y=210m

200

-20 -10 0 10 20
620
630
640
650
660
670
680
690
700
710
-20 -10 0 10 20
620
630
640
650
660
670
680
690
700
710

-20 -10 0 10 20
620
630
640
650
660
670
680
690
700
710
-20 -10 0 10 20
620
630
640
650
660
670
680
690
700
710

(a) X-displacement (b) Z-displacement

Fig. 6. Displacement distribution of core wall on section Y=210m (unit: mm)
3.2 Earthquake responses of dam
3.2.1 Finite element model
The finite element model what can be used to analyze earthquake responses of the dam is
build, which is the same as the model for the static analysis.
3.2.2 Dynamic parameters and calculation condition
The time history analysis method is used and the equivalent nonlinear viscoelastic model is
chosen as the constitutive model of soil, which assumes that the soil of dam body and
foundation overburden are considered as viscoelasticity, and equivalent shear modulus G
and equivalent damping ratio are applied to reflect the nonlinearity and hysteretic nature
this two properties of soil dynamic stress-strain relation. The dynamic parameters of dam
are shown as in Table 4 and Table 5.


201
Material
parameters
Soil Material
loam filter rockfill ballast Gravel
(dam)
Gravel
(mud)
pebble Gravel
(foundation)
K 375 1875 2282 1147 2052 696 1701 1603
n 0.63 0.61 0.55 0.58 0.54 0.57 0.55 0.54
Table 4. Maximum dynamic shear modulus parameters of dam materials

Material
parameters
Soil Material
loam filter rockfill ballast Gravel
(dam)
Gravel
(mud)
pebble Gravel
(foundation)
a
K 0.4653 1.5232 2.2814 1.6118 1.6653 1.4226 2.1450 1.6603
a
n 1.1883 1.2100 2.0871 1.5464 1.5553 1.2002 2.0322 1.4876
v
K 1.8252 2.0632 2.3221 1.7877 1.7623 1.5308 2.2895 1.7222
v
n 1.7119 1.9121 2.2007 1.6206 1.5989 1.3359 2.1659 1.5098
Table 5. Residual strain parameters of dam materials
-5.00
-4.00
-3.00
-2.00
-1.00
0.00
1.00
2.00
3.00
4.00
5.00
0 5 10 15 20 25 30
t/s
A
x
/

m
/
s
*
s

Fig. 7. The input curve of earthquake acceleration of bedrock in horizontal direction

202
Under the Wenchuan earthquake condition, in the dam site of Bikou hydropower station,
the seismic intensity is comprehensively evaluated as degree 9. Correspondingly, the
bedrock in the dam site has a peak level acceleration of 404cm/s
2
and its seismic acceleration
curve of bedrock in horizontal direction is shown as Fig.7, which is in up-down stream
direction and in dam axis direction. When calculating, the seismic acceleration value of
vertical direction is regarded as two thirds of the horizontal one. Meanwhile, the dynamic
water pressure is evaluated by the additional mass method called Westergaard method, that
using equivalent additional mass instead of the dynamic water pressure to stack with the
quality of the dam body during the earthquake.
3.2.3 Earthquake responses
Under the conditions of Wenchuan earthquake intensity, the earthquake response of the
dam are analyzed, including the maximum acceleration response, displacement response
and stress response of dam body, loam core and cut-off walls, and the earthquake induced
permanent deformation (Serff et al., 1976; Taniguchi et al., 1983; Kuwano & Ishihara, 1988;
Cascone & Rampello, 2003; Elia et al, 2011) of the dam is also obtained. The maximum
values of earthquake responses analysis results are shown as in Table 8.The distributions of
the dynamic analysis main features (Idriss et al., 19773; Ahmed-Waeil et al., 1990a, 1993b;
Zhu et al., 2003) are shown as in Fig.8~Fig.14.
1. Acceleration response
Calculation results show that the natural vibration period of Bikou dam is about 0.74
seconds. The maximum acceleration response of the rockfill respectively is 4.08m/s
2
along
the river direction, 4.43m/s
2
along the dam axis direction and 3.20m/s
2
along the elevation
direction. For the core, the maximum acceleration response along the above three directions
are 3.89m/s
2
,4.14m/s
2
and 3.14m/s
2
respectively. Along the dam axis direction, the
acceleration response of the dam is the greatest, followed by the acceleration response along
the river direction and the acceleration response along the elevation direction is the smallest.
When earthquake comes, the upstream and downstream dam shell materials, which have a
relatively long distance to the cut-off wall, are slightly affected. The corresponding
horizontal acceleration response is basically proportional to the dam height and the
maximum one appears near the upstream and the downstream slope surfaces which are
close to the dam crest. The seismic acceleration response of a point is related to the vertical
distance between the point and the bottom of the river valley; the height is greater, the
corresponding seismic response is greater. As the slide slope of the right bank is slower than
the left bank, the maximum horizontal acceleration response appears much closer to the
right bank. From the middle of river bed to the both banks, the horizontal acceleration
response decrease gradually which is related to the dam body type and boundary
conditions. From the distribution of acceleration response on cross section, it can be seen
that near the right bank, as the base of the downstream dam body is higher than the
upstream and overburden layer exist at the bottom of the upstream dam body, the
acceleration response of the upstream dam body is greater than the downstream dam body.
The situation is opposite near the left bank, as deep overburden layer exists at the bottom of
the downstream dam body and the downstream dam slope is much steeper than the
upstream, the acceleration response of the downstream dam body is greater than the
upstream dam body.


203
2. Displacement response
The maximum displacement response of the rock-fill is 25.92mm along the river direction,
16.99mm

along the dam axis direction and 12.39mm along the elevation direction. For the
loam core, the maximum acceleration response along the above three directions are
25.61mm, 16.77mm and 12.33mm respectively.
The displacement responses are not so great. Along the river direction, the displacement
response of the dam is the greatest, followed by the displacement response along the dam
axis direction. The smallest displacement response is along the elevation direction. The
distribution of displacement response is almost the same with acceleration response. Under
the seismic action, the displacement response is basically proportional to the dam height
and the maximum one appears near the upstream and the downstream slope surfaces which
are close to the dam crest. The seismic acceleration response of a point is related to the
vertical distance between the point and the bottom of the river valley, the height is greater,
corresponding seismic response is greater. The maximum horizontal displacement response
appears near the dam crest above the deepest site of river valley. As overburden exists at the
left bank, the maximum vertical displacement appears much closer to the left bank. From
the center of river bed to the both banks, the displacement response decreases gradually
which is related to the dam body type and boundary conditions. From the distribution of
displacement response on cross section, the displacement response of the dam is not only
related to the grade of upstream and downstream slope, but also the depth and distribution
of overburden. Close to the right bank, as the base of the downstream dam body is higher
than the upstream and overburden layer exists at the bottom of the upstream dam body, the
displacement response of the upstream dam body is greater than the downstream dam
body. The situation is opposite near the left bank. As deep overburden layer exists at the
bottom of the downstream dam body and the downstream dam slope is much steeper than
the upstream, the displacement response of the downstream dam body is greater than the
upstream dam body.
3. Stress response
For the rockfill of dam, the maximum first principal stress is 425kPa, the maximum second
principal stress is 352kPa and the maximum third principal stress is 203kPa. For the core, the
maximum first principal stress is 265kPa, the dynamic tensile stress is 221kPa; the maximum
second principal stress is 215kPa, the dynamic tensile stress is 178kPa; the maximum third
principal stress is 192kPa, the dynamic tensile stress is 157kPa. For the first concrete cut-off
wall, the maximum first principal stress is 3031kPa, the maximum second principal stress is
compressive stress with a value of 591kPa and the maximum third principal stress is
compressive stress with a value of 363kPa. For the second concrete cut-off wall, the
maximum first principal stress is 4482kPa, the maximum second principal stress is
compressive stress with a value of 683kPa and the maximum third principal stress is
compressive stress with a value of 422kPa.
The stress response of rockfill and core along the river direction is the greatest, followed by
the displacement response along the dam axis direction and the displacement response
along the elevation direction is the smallest. The strong stress response of the core happens
on the major river bed where the loam core connects to the both banks. The big and small
principal stress response of the rock-fill become stronger with the increase of cover depth,

204
thus the contour lines are parallel to the dam slope, and the maximum dynamic compressive
stress and maximum dynamic tensile stress appear at the bottom of the dam near the
deepest river valley. At the same elevation, the stress response of the dam body is much
stronger than the core. As the great difference of deformation modulus between concrete
and loam, stress concentration appears near the cut-off wall. Besides, at the downstream
side of cut-off wall, dynamic stress response is very strong. From the distribution of stress
response on longitudinal profile, it can be seen that the stress response at the bottom of the
dam on the right bank is a bit stronger than on the left bank, the dynamic stress response is
great at the connecting part between dam body and both bank slopes, especially near the
place where the section of bank slope varies, stress concentration of strong dynamic stress
response appears; and with the increase of peak acceleration, the stress response becomes
stronger. So, more attentions should be paid to these weak areas.
Great dynamic stress appears at the top of the concrete cut-off wall, so does in the
connection part between bedrock and the cut-off wall close to the bank slope. The stress
response of concrete is proportional to the peak seismic acceleration, thus with the increase
of peak acceleration, the dynamic tensile stress and compressive stress become greater. The
maximum dynamic tensile stress of the cut-off wall is shown as in Table 6. Overall, the
dynamic compressive stress of concrete is small and the dynamic tensile stress is relatively
much greater, so the monitoring to the strength of the cut-off wall should be strengthened to
ensure the safety operation of the dam.
During the earthquake, the maximum shear stresses of the dam are respectively 236kPa. On
the cross section of the dam, the maximum dynamic shear stress and the dynamic shear
stress become greater gradually from the upstream side and downstream side to the dam
axis; but in the middle of the core, the dynamic shear stress decreases greatly, and at the
bottom of dam, the dynamic shear stress response of rock-fill and transition material are
very great. So, as the significant difference of the filling materials, the maximum shear stress
responses are very significant in the connection parts of materials. The shear stress response
of rock-fill is very great on the bank slope where the sections vary, and shear stress
concentration appears in the local areas of where the sections vary at both banks; in the
main river bed where the core connects to both bank, the shear stress is great.
4. Earthquake induced permanent deformation
The maximum permanent horizontal displacement is 74mm along the river direction, 47mm

along the dam axis direction and the maximum permanent vertical displacement or
settlement is 239mm. Taking no account of the thickness of the overburden on the dam
foundation, the maximum dam height is 101.8m, then the permanent settlement induced by
earthquake is about 0.23 percents of the dam height.
Due to the upstream water pressure, the earthquake induced permanent deformation of the
core along the river orientation points to the downstream, so does the upstream dam shell
materials near the dam crest. Along the dam axis direction, the earthquake induced
permanent deformation is not so great and the maximum one appears near the upstream
dam slope. It can be seen that in the dam abutments of both banks, the displacements of the
dam body point to the center of the river valley; near the dam axis, the displacement and its
variation gradient at the right bank are greater than the left bank. Affected by the concrete


205
Item Value
maximum
absolute
acceleration
response
/ ms
-2

Rockfill up-down stream direction 4.08
dam axis direction 4.43
vertical direction 3.20
Core wall up-down stream direction 3.89
maximum
displacement
response
/mm
Rockfill up-down stream direction 25.92
Core wall up-down stream direction 25.61
maximum stress
response of
rockfill
/kPa
Rockfill 1
st
principal stress 425/-422
2
nd
3
rd
Core wall 1
st
2
nd
3
rd
1
st
cut-off wall 1
st
2
nd
3
rd
2
nd
cut-off wall 1
st
2
nd
3
rd
Earthquake induced permanent
deformation/mm
stream direction
(up/down stream)
74/-49
dam axis direction
(left /right bank)
47/-26
vertical direction(settlement) -239
maximum shear stress response / kPa 236
Table 6. Earthquake responses of Bikou dam by 3-D dynamic FEM

206
cut-off wall, the ultimate settlement of dam body below the elevation of 650m is small. And
the settlement of the dam body above the river valley is basically uniform from the right
side to the left, and the permanent deformation increases with the elevation increases. The
maximum settlement appears near the downstream dam crest which is close to the dam
axis. In general, the earthquake induced permanent deformations in both conditions are not
great.

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(a) X-direction (section Y=210m)

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(b) Y-direction (section Y=210m)

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(c) Z-direction (section Y=210m)

Fig. 8. Distribution of maximum absolute acceleration response (m/s
-2
)


207
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(a) X-displacement (section Y=210m)

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(b) Y-displacement (section Y=210m)

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(c) Z-displacement (section Y=210m)

100 150 200 250 300 350
600
650
700
100 150 200 250 300 350
600
650
700

(d) X-displacement (section X=-10m)
Fig. 9. Distribution of maximum displacement response (mm)

208

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(a) the first principal stress

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(b) the second principal stress

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(c) the third principal stress

Fig. 10. Distribution of maximum stress response on section (Y=210m) (unit: kPa)


209

140 160 180 200 220 240 260 280 300 320
600
620
640
140 160 180 200 220 240 260 280 300 320
600
620
640

(a) the first principal stress
140 160 180 200 220 240 260 280 300 320
600
620
640
140 160 180 200 220 240 260 280 300 320
600
620
640

(b) the second principal stress
140 160 180 200 220 240 260 280 300 320
600
620
640
140 160 180 200 220 240 260 280 300 320
600
620
640

(c) the third principal stress

Fig. 11. Distribution of maximum stress response of the second cut-off wall (unit: kPa)

210

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

Fig. 12. Distribution of maximum dynamic shear on section (Y=210m) (unit: kPa)

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(a) X-direction
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(b) Y-direction
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(c) Z-direction
Fig. 13. Distribution of earthquake induced permanent deformation on section(Y=210m)
(mm)


211
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(a) t=5s

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(b) t=15s

-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700
-200 -100 0 100 200 300 400
600
650
700

(c) t=25s

Fig. 14. Distribution of safety factors at different times during earthquake on section
(Y=210m)
4. Earthquake-resistance safety evaluation
4.1 Comparing the static calculation results and monitoring data
In order to monitor the consolidation settlement in construction and operation, 5
collimating lines are set for the Bikou Hydropower Station soil core dam(Fig.18), which
are respectively up dam 0-010.1m(708.00m elevation), under dam 0+007.8m(709m
elevation), under dam 0+041.8m(691m elevation), under dam 0+093.0(670m elevation) and
under dam 0+142.7(650m elevation). Only up dam section 0-010.0m got observed on
December 17, 1975, and other sections started observation two years later after storage for
the construction of crest parapet wall and downstream slope and drainage. Monitoring
date of dam surface horizontal displacement and settlement before earthquake are shown
as in Fig.16 and Fig.17.
Among all the monitoring sites, only section 0-010.0m started observation once storage,
while other sections started observation relatively late. Therefore, displacement of these 5
sections cannot directly be compared with that of finite element calculation. As section 0-

212
010.0m got monitored earlier, its monitoring date should be compared with modified results
of finite element calculation, which can be shown as in Fig.16 and Fig.17.
According to the settlement distribution (Fig.18) along dam axis direction of section 0-
010.0m, settlements of monitoring points on crest are approximately proportional to their
corresponding rock-fill thickness at the same place. Crest position where locates on the
deepest valley floor has the biggest settlement, and settlement on right bank is larger than
that on left bank. Settlement distribution discipline of finite element calculation is roughly
the same with that of monitoring results. Additionally, calculation values are larger than
monitoring data, since dam settlement started monitoring later than corresponding
calculation situation, then displacement before monitoring was included in the calculation
results.
X = 0+093.0m
D7-3
D15-1
D8-3 D9-3 D10-3 D11-3 D13-3 D15-3
X = 0+041.5m
X = 0+007.8m
X = 0-010.0m
D15-2 D13-2 D11-2 D10-2 D9-2 D8-2 D7-2
D9-1
D8-1
D10-1 D11-1
D13-1
D13-4 D11-4 D10-4 D9-4 D8-4
D8-5 D9-5 D10-5
D11-5
D13-5
F8-1 F7-1 F10-1 F11-1 F14-1 F15-1 dam crest
691.00m
670.00m

Fig. 15. Layout of monitoring point
Considering the starting observation time of section 0-010.0m, settlements of monitoring
points are values after reservoir started storage. Thus, calculation results of finite element
analysis should include the settlement values when dam body filling had been finished, that
is, settlements incremental in the operation and completion periods. The calculation values
are smaller than values of monitoring. Its because that calculation model only considers
principal deformation and ignores the rheological of dam body. From the comparison of
crest settlement above, calculation model and parameters are reasonable and fundamentally
match the reality.


213
-30
-20
-10
0
10
20
30
40
100 150 200 250 300 350
along dam axis direction/m
H
o
r
i
z
o
n
t
a
l

d
i
s
p
l
a
c
e
m
e
n
t

b
e
f
o
r
e
e
a
r
t
h
q
u
a
k
e
/
m
m
x=-10
x=7.8
x=41.8
x=93
x=142.7

Fig. 16. Distribution of horizontal displacement measured value before the earthquake
-600
-500
-400
-300
-200
-100
0
100 150 200 250 300 350
s
e
t
t
l
e
m
e
n
t

b
i
f
o
r
e

e
a
r
t
h
q
u
a
k
e
/
m
m
x=-10
x=7.8
x=41.8
x=93
x=142.7

Fig. 17. Distribution of settlement measured value before the earthquake
-1000
-800
-600
-400
-200
0
70 120 170 220 270 320 370
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

o
f

d
a
m

c
r
e
s
t

i
n

t
h
e
s
e
e
t
i
o
n

x
=
-
1
0
m
/
m
m
settlement in completion period settlement in operation period
settlement incremental monitoring settlement

Fig. 18. Comparison of the settlement measured value and calculated value before the
earthquake on the section (X=-10.0m)

214
4.2 Results of earthquake-resistance analysis
In the process of dynamic calculation, safety factor development of every element,
excluding bed rock, concrete cut-off walls, wave wall and joints, has been completely
recorded throughout earthquake. The safety factor is defined as the ratio of shear strength
to shear stress (including both static and dynamic shear stress) of element potential failure
surface. As it to analyze the seismic safety of upstream and downstream, the anti-slide
stability coefficient is calculated on pseudo-static method. Sweden Slice Method is
adopted for slope stability analysis, where both horizontal and vertical seismic actions are
considered.
Calculation condition for slope anti-slide stability can be seen in Table 7. The smallest
safety factors of upstream and downstream slopes are shown as in Table 8, and the
potential sliding surfaces are shown as in Fig.19. As the compacted earth dam design
specification (SL274-2001) says that the smallest anti-slide stability factor under seismic
action should be no less than 1.15. Under the condition BK-S1, slope stability of Bikou
dam cannot meet the present criterion. However, on the conclusion of Bikou station
injuring survey under 5.12 Wenchuan earthquake and its initial analysis report, only
some joints connecting crest and body to embankment were damaged and needed
amending, yet the whole dam was safe in general. Under the seismic action, safety factor
of Bikou dam does not satisfy demand, however, the dam did not suffer from sliding or
slope instability during the 5.12 Wenchuan earthquake. So, it is suggested that smaller
safety factors should be adopted. For example, a level 2 dam has a safety factor ranging
from 1.05 to 1.10. In a word, the safety factor distribution suggests that the Bikou dam has
some partial scopes where the factors are less than 1 near the crest and upstream and
downstream slopes. But the scales are sporadic and will not lead the failure of dam
slopes.

Condition Dam slope Peak
acceleration
/g
Upstream water
level/m
Downstream
water level/m
BK-S1 Upstream and
downstream
0.404 707.00 617.02
Table 7. Condition for dam stability analysis

Condition Peak acceleration/g Dam slope Fs
BK-S1 0.404 upstream 1.11
downstream 1.19
Table 8. Results for dam stability analysis


215

Circle coordinates(-57.60,752.02)
Circle coordinates(54.67,774.87)
Downstreamwater level 617.02m
Crest elevation 711.80
Normal water level 707.00m
Sliding surface
Loam
core
wall
Sliding surface

Fig. 19. Dangerous sliding surface position schemes of dam slopes (condition BK-S1)
5. Conclusions
In this paper, according to the actual engineering conditions of Bikou earth core rockfill
dam, the 3-D non-linear FEM model of the dam is set up for calculating earthquake
response by the dynamic time-history analysis method. By simulating the process of the
filling of dam body and reservoir impounding, firstly the original static stress field of dam
body is obtained under the normal water level. Then by the dynamic time-history analysis
method, the earthquake responses of the dam, including acceleration response,
displacement response, stress response and the earthquake induced permanent
deformation of the dam are obtained for inputting the earthquake with peak acceleration
of bedrock 404cm/s
2
. Under the action of the seismic with peak acceleration 404cm/s
2
,
the acceleration response along the up-downstream direction is maximal with the value of
4.08m/s
2
, the one along the dam axis is secondary with the value of 4.43m/s
2
, and the one
along vertical direction is the minimum with the value of 3.20m/s
2
. The maximum
displacement response of dam in the three directions are 25.92mm, 16.99mm, and
12.39mm respectively, and the maximum principle stress responses of rockfill body are
425kPa, 352kPa, and 203kPa respectively. The maximum earthquake induced permanent
settlement is 239mm, about 0.23 percents of the dam height. It is shown that the
earthquake responses of the dam are close to the recorded data. The theories and methods
for analyzing the earthquake responses of the earth dam here are feasible, and the results
are consistence with the real situation of post-earthquake.
6. Acknowledgement
This paper was supported by the Key Project of National Natural Science Foundation of
China (Grant NO. 40930635) and Non-profit Projects Research of Ministry of Water
Resources of China (Grant NO. 200901070). Grant NO 2009586012, entitled State Key

216
Laboratory of Hydrology-Water Resources and Hydraulic Engineering provided partial
support for this work.
7. References
Ahmed-Waeil, M. E., & Ramana, V. G. (1993). Dynamic behaviour and seismic response of
El Infiernillo dam. Earthquake Engineering & Structural Dynamics, 1993, Volume 22
Issue 8, pp. 665-684
Ahmed-Waeil, M. E., Ronald, F. S., Mohamed, F. S., et al. (1990). La Villita Dam
Response During Five Earthquakes Including Permanent Deformation. Journal
of Geotechnical Engineering (ASCE), Vol.116, No.10, October 1990, pp. 1443-
1462
Cao, Z. Z., Youd, T. L., and Yuan X. M. (2011). Gravelly soils that liquefied during 2008
Wenchuan, China earthquake, Ms=8.0, Soil Dynamics and Earthquake Engineering,
2011, Vol. 31, pp.1132-1143.
Cascone, E., & Rampello, S. (2003). Decoupled seismic analysis of an earth dam, SOIL
DYNAMICS AND EARTHQUAKE ENGINEERING, 2003, Vol. 23, pp.349-
365
Chen, H. Q., Xu, Z. P., and Lee, M. (2008). Wenchuan Earthquake and seismic safety of
large dams, Journal of Hydraulic Engineering, 2008, Vol.39, No.10, pp.1158-
1167
Chen, L. Y., & Talwani, P. (1998). Reservoir-induced seismicity in China, PURE AND
APPLIED GEOPHYSICS, 1998, Vol. 153, No. 1, pp.133-149
Duncan, J. M., and Chang, C. Y. (1970). Nonlinear analysis of stress and strain in soil,
Journal of the Soil Mechanics and Foundation Division, ASCE, 96(SM5), pp.l629-
l653
Elia, G., Amorosi, A., Chan A, H. C., et al. (2011). Fully coupled dynamic analysis of an
earth dam, GEOTECHNIQUE, 2011, Vol.61, pp.549-563
Gu, G. C. (1989). Earthquake Engineering for Earth-rock Dams, Hohai University Press, Nanjing
, 1989
Hardin, B. O., & Denevich, V. P. (1972). Shear moduls and damping in soil: design equations
and curves. ASCE, 98 (SM7), pp.667-92
Hardin, B. O., & Denevich, V. P. (1972). Shear modulus and damping in soil: measurements
and parameter effects. ASCE, 98 (SM6), pp.603-624
Idriss, I. M., Lysmer, J., Huang, R., et al. (1973), A computer program for evaluating the
seismic response of soil structures by variable damping finite element procedures,
Berkeley, College of Engineering University of California.
Kenneth, G. (2011). Dermatology Aboard the USNS COMFORT: Disaster Relief
Operations in Haiti After the 2010 Earthquake, Dermatologic Clinics, 2011, Vol. 29,
No.1, pp.15-19
Kuwano, J., & Ishihara, K. (1988). Analysis of permanent deformation of earth dams due to
earthquake, Soils and Foundations, Vol, 28(01), pp.41-55
Lin, P., Wang, R. K., and Li, Q. B. (2009). Effect Analysis of Structural Safety of Typical Large
Dams in Wenchuan 8.0 Earthquake, Chinese Journal of Rock Mechanics and
Engineering, 2009,Vol.28, No.6, pp. 1261~1268


217
Mejia, L. H. (1981). Three dimensional dynamic response analysis of earth dam, Report No.
UCB/EERC-81/15. Berkeley, University of California
Mejia, L. H., Seed, H. B., & Lysmer, J. (1982). Dynamic analysis of earth dams in three
dimensions, Journal Geotechnical Engineering Division (ASCE), 1982, 108(GT12), pp.
1354-1376
Pan, R. Y. (2009). Analysis of the recheck of the earthquake damage of Bikou Core Rockfill Dam,
Hohai University, Nanjing,2009
Pei, S. P., Su, J. R., Zhang, H. J., et al. (2010). Three-dimensional seismic velocity structure
across the 2008 Wenchuan M
s
8.0 earthquake, Sichuan, China, Tectonophysics, 2010,
Vol.491, pp.211-217
Serff, N., Seed, H. B., Makdisi, F. I., & Chang, C. K. (1976). Earthquake induced deformations
of earth dams, Report No. EERC/76-4, Earthquake Engineering Research Center,
University of California, Berkely, 1976
Shen, Z. Z., Wen, X.Y., and Lv, S.X. (2006). Analysis of earthquake responses for Jiudianxia
concrete face rockfill dam, L. Berga et al eds, Dams and Reservoirs, Societies and
Environment in the 21st Century, Published by Taylor & Francis/Balkema, London,
2006, Vol. 1, pp.925930
Shen, Z. Z., Xu, L.Q., and Wang, W. (2010). Earthquake response of Xika concrete face
rockfill dam by EFM, Engineering, Science, Construction, and Operations in Challenging
Environments, 2010 ASCE, pp.463-472
Takewaki, I., Murakami, S., Fujita, K., et al. (2011). The 2011 off the Pacific coast of Tohoku
earthquake and response of high-rise buildings under long-period ground motions,
Soil Dynamics and Earthquake Engineering, 2011, (In press)
Talwani, P. (1997). On the nature of reservoir-induced seismicity, PURE AND APPLIED
GEOPHYSICS, 1997, Vol. 150, pp.473-492
Taniguchi,E., Whiteman, R.V., & Warr, W. A. (1983). Prediction of earthquake induced
deformation of earth dams, Soils and Foundations, 1983, Vol. 23(4)
Wang, W. X., Sun, W. K., and Jiang, Z. S. (2010). Comparison of fault models of the 2008
Wenchuan earthquake (M
s
8.0) and spatial distributions of co-seismic deformations,
Tectonophysics, 2010, Vol.491, pp.85-95
Xenaki, V. C., & Athanasopoulos, G. A. (2008). Dynamic properties and liquefaction
resistance of two soil materials in an earthfill dam - Laboratory test results,
SOIL DYNAMICS AND EARTHQUAKE ENGINEERING, 2008, Vol. 28, pp. 605-
620
Yu, H. Y., Wang, D., Yang, Y.Q., et al. (2009). The preliminary analysis of strong ground
motion records from the M
s
8.0 Wenchuan Earthquake. Journal of Earthquake
Engineering and Engineering Vibration, 2009, Vol.29, No.1, pp.1~13
Zeghal, M., & Abdel-Ghaffar, A. M. (2009). Evaluation of the Nonlinear Seismic Response of
an Earth Dam: Nonparametric System Identification, JOURNAL OF EARTHQUAKE
ENGINEERING, 2009, Vol.2009. pp. 384-405
Zhou, J. P., Yang, Z. Y., Fan, J. X., et al. (2009). Seismic Damage Investigation on Large and
Medium Sized Hydropower Projects in Wenchuan Earthquake Area. Chinese
Journal of Water Power, 2009,Vol.35,No.5, pp.1-6

218
Zhu, S., Wen, S.Q., and Huang, Y. M. (2003). Deformation and Stress Calculation for A
200m High CFRD, Journal of Hohai University (Natural Sciences ) ,2003, 31(6):631-
634
9
Recent Landslide Damming Events and
Their Hazard Mitigation Strategies
Ahsan Sattar and Kazuo Konagai
Institute of Industrial Science, University of Tokyo
Japan
1. Introduction
Earthquakes often strike the vulnerable parts of our society. A 7.0-magnitude earthquake
shattered Port-au-Prince, the capital of Haiti. Based on historical records in the Port-au-
Prince area, the last large earthquake happened in 1770. Therefore, throughout its history,
the capital of the country had never experienced any deadly earthquake, but long-lasting
political violence. Being the poorest country in the Americas as per the Human
Development Index, due attention has never been paid to seismic preparedness, causing the
catastrophic devastation with 316,000 casualties to occur (Reuters, 12 January 2010). Even in
Japan with advanced technologies in disaster preparedness, the earthquake and tsunami of
March 11th 2011 revealed vulnerabilities of disaster prevention schemes/ systems given the
tsunami heights exceeding the determined design heights. These devastations often cause
long lasting problems. In Haiti 2 million people are managing to get by on food rations. In
Japan, huge amount of tsunami debris are just piled up in temporary damp yards with the
fear that they might have been radiation-contaminated, thus discouraging all attempts for
quick rehabilitations.
One of the most serious and costliest post earthquake problems is the stabilization of
earthquake-induced landslide and debris masses. Shaking from the January 12, 2010, Haiti
earthquake caused devastating structural damage and triggered many landslides that blocked
roads, dammed rivers and streams, and threatened infrastructure in many parts of Haiti.
About 4000 to 5000 landslides were estimated by the USGS (Randall at el., 2011). Many of the
landslides blocked stream drainages and impounded lakes. In Japan, it has never been given a
front-page coverage that total 4 billion JPY is spent annually even to this date to stabilize
unstable debris mass of total 200 million m
3
blocking the upper reaches of the 50km long
Joganji River in Tateyama Caldera; the debris mass of total 400 million m
3
at that time was
originated in the 1858 Hietsu Earthquake with the estimated magnitude of 7 to 7.1.
A landslide mass that blocks water and impounds a lake at its upstream is referred to as a
landslide dam. Landslide damming events are very common around the world. They
pose hazards to the population at both their upstream and downstream areas because of
inundations due to lake impoundment and slope stability problems associated with such
impoundment, and the risk of downstream flooding, respectively. Researches on landslide
dams have been focused mostly on their distribution and failure potential (e.g, Costa &
Schuster 1988, Ermini & Casagli 2003), and less attention has been given to their emergency
treatments. A retrogressive review of the emergency response and mitigation treatment for

220
recent landslide damming events around the world highlights the problems and uncertainty
involved in mitigating the risk from landslide dams.
The simplest and the most commonly used mitigation measure has been the excavation of a
spillway which is normally constructed at the lowest point at the crest of a landslide deposit.
The excavation of the spillway has been proven to be the best engineering technique for
mitigating risks posed by landslide dams, and its application can be traced about 500 years
back (Bonnard, 2006). Sometimes blasting is also done to help excavating the channel;
however blasting is seldom done for complete removal of large landslide masses. Drainage
of the dammed lakes by means of pumps and siphons is a relatively less common treatment
and is limited to small lakes. Tunnels or diversion channels are also rarely adopted as
mitigation measures owing to the cost and time required.

Fig. 1. True-color image of Tangjiashan landslide dam and dammed lake captured by
NASAs Earth Observing-1 (EO-1) (a) on 3
rd
June 2008 prior to overtopping and (b) on 8
th

June after overtopping but before the breaching of dam
A retrogressive review of the emergency response and mitigation strategies for recent
landslide damming events around the world is presented in this chapter. Four landslide
dams are presented for this purpose, consisting of the Tangjiashan landslide dam which was
triggered by the 2008 Wenchuan earthquake in China, the Higashi Takezawa landslide dam
that was triggered by the 2004 Mid-Nigata earthquake in Japan, the Attabad landslide that
was formed in 2010 in Hunza Pakistan, and the Hattian Bala landslide dam that was formed
by the 2005 Kashmir earthquake in Pakistan. A brief description of the mitigation measures
adopted for the dams is given and failures of Tangjiashan landslide dam and Hattian
landslide dam are also explained. A detailed description of the geomorphologic changes
resulting from the Hattian landslide dam is also provided to elucidate long lasting hazards
that landslide dams pose to their surrounding areas.
2. Tangjiashan landslide dam
The Ms 8.0 Wenchuan earthquake of China occurred on 12 May 2008 that have reportedly
created 256 landslide dams most of which were formed along the YingxiuBeichuan
(Lonmenshan) fault, which was dislocated over its 270km stretch in the earthquake (Qiang
at el., 2009). The Tangjiashan landslide dam (N31 50' 39.54", E104 25' 45.93") is located in
Dammed
Lake
Tangjiashan
Landslide Scar
Beichua
n city
Jiangjiang
River
1000 m
3
rd
June 2008
(a)
Dammed
Lake
Spillway
8
th
June 2008
200 m
Landslide
dam
(b)

Recent Landslide Damming Events and Their Hazard Mitigation Strategies

221
northern part of the fault zone, and impounded a lake on the Jiangjian River (Fig. 1). This
lake behind the 124 m high landslide dam had a maximum storage capacity of 320 million
m
3
(Liu at el., 2010). The dam posed a great threat of flooding to the 1.2 million people living
along the lower reach of the river, therefore required immediate mitigation measures.
Flow stablizes
Discharge 6420
cumecs
Bolder blockage
removed/Breaching
Overtopping
Spillway excavation
completed
Lake volume
200million m
Spillway excavation
Lake volume
110million m
2008 Wenchuan
Earthquake,
Landslide trigger
1
2
-
M
a
y
-
0
8
1
4
-
M
a
y
-
0
8
1
6
-
M
a
y
-
0
8
1
8
-
M
a
y
-
0
8
2
0
-
M
a
y
-
0
8
2
2
-
M
a
y
-
0
8
2
4
-
M
a
y
-
0
8
2
6
-
M
a
y
-
0
8
2
8
-
M
a
y
-
0
8
3
0
-
M
a
y
-
0
8
1
-
J
u
n
-
0
8
3
-
J
u
n
-
0
8
5
-
J
u
n
-
0
8
7
-
J
u
n
-
0
8
9
-
J
u
n
-
0
8
1
1
-
J
u
n
-
0
8
1
3
-
J
u
n
-
0
8
3
3

Fig. 2. Timeline of Tangjiashan landslide dam emergency response
2.1 Emergency response and mitigation measures
The water level of the dammed lake initially rose at a rate of 1.2 m/day under the average
flow rate of 80-90m
3
/s, and later on the rate decreased to 0.5m/day (Liu at el., 2010). Six
towns and two cities are located downstream of the dam that were threatened to be
inundated by the dam-break-caused flood. The dam break studies showed that a
catastrophic flood would occur if no mitigation measures were adopted. Till 23
rd
of May the

Fig. 3. Higashi Takezawa landslide dam and the dammed lake (Satellite imagery from
Google Map)
Landslide deposit
Impounded
lake
Imokawa
River
200 m
N37 18 7.4,E138 54 1.7
Source area
N

222
total volume of the lake had reached 110 million m
3
and then 200 million m
3
till the end of
May, 2008. The figure 2 presents the timeline of the Tangjiashan landslide dam emergency
response. The lake hypsometric curve showed that if the lake level was kept at 740.3 m, the
storage capacity of the lake would be reduced to 238 million m
3
instead of 320 million m
3

corresponding to the natural potential overtopping level of 755.0 m (Liu at el., 2010).
Accordingly the civil works were carried out to excavate a spillway to minimize the
potential dam breach damage. It took 7 days work from May 26 to 31, to excavate a 13 m
deep and 8 m wide channel covering 475 m in length. The primary intension for the
excavation of the spillway was to reduce the total volume of the lake and meanwhile to
create a controlled flood to drain the lake. About 275,000 people evacuated from the
downstream areas for the expected flood.
2.2 Retrospective review of breaching event
The lake level reached 740 m, and the water started overflowing through the channel on
June 7 (Qiang at el., 2009, Liu at el., 2010). The breaching process did not initiated
immediately; instead the water kept rising for two days. A huge block slowed the channel
erosion which was later removed by blasting on 8
th
June. However, no rapid flow occurred
till 10
th
June when the water level reached 742 m. Then the outflow increased rapidly, and
the channel walls and bed eroded rapidly. The outflow discharge peaked at 6420 m
3
/s on
June 10, and the erosion stopped on 12
th
June. Downstream areas were inundated where the
people were evacuated before the breaching. The debris mass from the breached dam were
transported and deposited along the downstream reaches. The emergency work for
Tangjiashan landslide dam was a successful case history of response where the early
identification of the hazard had made it possible to map out necessary measures for the
area.
3. Higashi Takezawa landslide dam
The M6.8 Mid-Niigata Prefecture earthquake of 23
rd
October 2004 triggered numerous
landslides and formed more than 50 landslide dams along the Imokawa River and its
tributaries. Among these landslide dams the most crucial dam was the Higashi Takezawa

Fig. 4. Emergency works at Higashi Takezawa landsldide dam. Photo taken on 20.12.2004 by
Takeji & Tadashi, 2005
(a)
(b)


223
landslide dam (N37 18' 16.41", E138 54' 16.13"), the largest landslide dam triggered by
the earthquake (Fig. 3). The landslide originated from the left bank of the Imokawa River
at Higashi Takezawa district. The landslide mass was about 350 m long and had volume
up to 1 million m
3
(Sassa, 2005). The blockage started to impound a lake and soon the
water started to inundate the Kogomo village. Due to the possible danger of overtopping
and successive breaching of the dam emergency measure were taken to prevent the
failure.
Due to the danger of possible failure of the dammed lake, it was necessary to reduce the
total capacity of the lake. Two possible mitigation measures were considered to cope with
the flood hazard that the landslide dam posed. One of the possible measures was to remove
the blockage to the original river level to drain the lake completely. This option would favor
people living both upstream and downstream reaches of Imogawa River. However, there
was a concern that removing only the toe part of the landslide mass would reactivate the
movement of the remaining landslide mass. Removing the entire landslide mass would not
be considered as a cost-effective solution. Therefore, the landslide mass was planned to be
consolidated with the combination of diversion pipes and spillway. This meant that in
exchange for abandonment of the upstream Kogomo village, the relevant authorities have
secured the lower reach of the dam. Inhabitants of Kogomo Village were offered alternate
sites for their new dwellings.
The following steps have been taken for stabilizing the landslide mass: twelve pumps
were installed first to lower and keep the lake water level 11.5 m below the overflow
point. In addition, the outflow level was raised by 3 m with sand bags at the outlet.
Keeping in view the pumping power for this temporary system, diversion pipes were
installed in case water overtopped the landslide mass (Fig. 4). Finally, a 280m long
concrete-faced spillway was constructed with its cross-section large enough for snow
melts.
Closed circuit televisions (CCTV) were installed to monitor the conditions in the catchment
in case of emergency. Other monitoring activities included observations of cracks,
installation of the water level gauges and monitoring the surface displacement of the
collapsed slope.
4. Attabad landslide dam
On 4
th
January 2010 a landslide (36 18 50N, 74 49' 13E) occurred at Attabad in Hunza,
Gilgit Baltistan. The debris mass dammed the Hunza river and formed a lake which was 120
m deep at the saddle point of the debris mass. At about 3 km downstream of the landslide
dam, there is a site of an older landslide that blocked a 30 km long lake that formed in 1958
and eventually breached and caused a catastrophic flood. Prior to the 2010 Attabad
landslide, cracks were being observed by the villagers at Attabad, which were being
progressively lengthened. The area was declared as high hazard area by the local
authorities, and evacuations of the affected slopes was being implemented by the end of
2009.

224

Fig. 5. The true-color image of the Attabad landslide dam (36 18 50N, 74 49' 13E) and
lake captured by Advanced Land Imager (ALI) on NASAs Earth Observing-1 (EO-1)
satellite, on (a) 16
th
March 2010 and (b) 23
rd
March 2010
A lake immediately started to develop at upstream of the debris mass. The impoundment
created huge problems for the 25,000 people living at upstream of the landslide dam.
Mitigation studies were conducted (Petley, 2010) and as an emergency repose a 30 m deep
spillway was planned to be excavated to reduce the maximum possible volume of the lake.
Seepage points emerged progressively at six different locations till April 2010. Meanwhile,
closed circuit television, (CCTV) system was installed to issue a warning in case of much
feared seepage failure. The excavation of the spillway started on 30
th
January 2010. The
surface silty-clay deposit created problems during the excavation works as the construction
machinery were bogged down with the muddy conditions at the site, and the work was
stopped when the excavation had reached 15 m deep across the crest till 16
th
May 2010. The
water started overflowing on 29
th
May 2010. At the time of overtopping the lake reached a
length of 21 km and had an estimated volume of 450 x 10
6
m
3
. Fig 5a and 4b shows the true-
color image of the Attabad landslide dam and lake captured by Advanced Land Imager
(ALI) on NASAs Earth Observing-1 (EO-1) satellite, on 16
th
March 2010 and 23
rd
August
2010 respectively. About 25,000 people were cut off from the downstream population and
are still suffering at the moment of writing.
Landslide
scar
Attabad
Ainabad
Shishkat
Gulmit
Pasu
Lake
1 km
16
th
March 2010
Location of
1858
landslide
(a)
Landslide
scar
Attabad
Ainabad
Shishkat
Gulmit
Pasu
Lake
1 km
23
rd
August 2010
Location of
1858
landslide
(b)


225
4.2 Post overtopping situation
The Attabad landslide dam survived the much feared failure upon overtopping. The dam
has emerged as a permanent geological feature that has caused a 23 km stretch of the
Karakorum highway and about 232 houses to be inundated (NDMA) and isolated the
communities at its upstream reaches. Moreover the longevity of the dam is coming under
question. Erosion of the spillway occurred at the time of overtopping that continued for
several days which later on was followed by a slow down-cutting of the channel. The
channel erosion was stopped by the presence of large boulders embedded in the channel
bed, which prevented the catastrophic failure, while slowed the drainage of the lake. The
dam survived the high discharge of glacier melting of 2010. Blasting option was considered

Fig. 6. Contour map of Hattian landside, dammed lakes with their catchments and
downstream areas.

Fig. 7. Front view of the Hattian landslide mass, showing the downstream face erosion.
Spillway
Eroded portion from
downstream face
Hattian Dam
Karli
catchment
Tang
catchment
Landslide
source
area
Karli Lake
Tang Lake
Jhelum River
Karli
River
Activated
Balakot-
Bagh fault
Hattian Bala

226
to widen the spillway in late June, 2010 which was conducted at small scale and was
unsuccessful (Pamir Times). As the water level of the lake dropped, attempts were made to
lower the level of the spillway in January 2011, which was eventually unsuccessful.
Tang Lake
Filling
Tang Lake
Breaching
Piping in front of
Tang Lake
2005 Kashmir
Earthquake,
Landslide trigger
Karli Lake
Overtopping
Karli Lake Filling
Tang Lake
Overtopping
Erosion Gully
appeared
Excavation of
spillway
Gradual Toe
Erosion
Karli Lake
Breaching
8
-
O
c
t
-
0
5
6
-
A
p
r
-
0
6
3
-
O
c
t
-
0
6
1
-
A
p
r
-
0
7
2
8
-
S
e
p
-
0
7
2
6
-
M
a
r
-
0
8
2
2
-
S
e
p
-
0
8
2
1
-
M
a
r
-
0
9
1
7
-
S
e
p
-
0
9
1
6
-
M
a
r
-
1
0

Fig. 8. Timeline of post formation features and major events at Hattian Bala landslide
dam
Possible failures will include those due to down cutting erosion, seepage or piping
failure, earthquake induced dam collapse, landsliding into the lake, and the glacial lake
outburst flood (GLOF). The concerns are rising as the water level of the lake started
increasing in June-July 2011 because of the glacier melting. The dam is intact at the
moment of writing; however the stability of the dam is again threatened as the monsoon
of 2011 approaches.
5. Hattian Bala landslide dam
On the 8th October 2005 earthquake (epicenter 342935N, 733744E, focal depth 26
km, 7.6 Mw (USGS)) caused widespread destruction in northern areas of Pakistan and
Pakistan-administered Kashmir, killing more than 86,000 people. The earthquake triggered a
large landslide mass near the south-eastern end of the Balakot-Bagh fault, the fault that
caused the earthquake (Fig. 6). The slopes around the landslide source area have been a site
of frequent landslide activities predating the earthquake. The 65 million m
3
landslide
destroyed a village and killed about 1000 people (about 1.1% of total earthquake casualties)
making it one of the most devastating recorded historical landslide events (Dunning at el.,
2007). The landslide blocked two tributaries of the Karli River and created two lakes, Karli
Lake (large lake) with a maximum natural capacity of 86 million m
3
, and a Tang lake (small
lake) with maximum natural capacity of 5 million m
3
. Immediately the water started to be
impounded in the two lakes behind the landslide mass, and thus a threat of dam breach
flooding was eminent.


227

Fig. 9. (a) Satellite imageries of Karli River before and after breaching of Hattian dam
(Google Earth), along with the maximum flow depths measured along the Karli River after
failure of the dam (b) Profile of Hattian dam and Karli River before and after breaching of
the Hattian dam
As the water level in the lakes rose, the concern about the stability of the landslide dam
was increasing. Mitigation studies were conducted and based on the recommendations
the relevant authorities decided to excavate a spillway to reduce the maximum possible
volume of the lake. The seepage through the landslide mass, the water levels and inflows
to the lakes were continuously monitored as part of emergency response (WAPDA, 2007).
A 450 m long spillway was excavated for the Karli Lake and relatively short 130 m
spillway was excavated for the Tang Lake. The excavation of the spillway considerably
reduced the total natural storage capacity of the Karli Lake from 86 million m
3
to 62

228
million m
3
. The Tang Lake filled up in February 2006 and started overflowing. A timeline
of various events occurring at Hattian Bala landslide dam after its formation are
presented in figure 8.
The debris mass developed seepage erosion and the water level of the Tang Lake gradually
reduced by 7-8 m over a period of two months (May-June 2006) and later on reached a state
of equilibrium between inflow and seepage discharges, keeping the lake volume at
approximately 3.5 million m
3
. The Karli Lake filled up in April, 2007 and started to
overflow. The dam survived the overtopping and hence less attention was drawn towards
the monitoring of the dam as the spillway was performing well and no immediate threat
was imminent. On the basis of the comparison between the volumes of the debris mass with
the volume of the water stored in the lake the dam was thought to be stable by many
experts. Therefore, less attention was drawn towards the monitoring of the dam site.
However, after overtopping gradual erosion of the dam body started to occur due to the
overflowing and seepage water.
5.2 Breaching and breach inflected morphological changes
Gradual toe erosion was initially observed, which was then followed by a rapid head
cutting erosion of the downstream face of the dam (Fig. 7). On 9
th
February 2010, the
water of Karli Lake breached the debris dam after five days of incessant rain, and the
breaching drained about 36 million m
3
of water from the lake. According to the back
analysis conducted by Konagai, at el., (2011) the peak outflow was about 5500 m
3
/s. A
moderate 32 mm/day rainfall was estimated on the day prior to the day of breaching.
About 7.78 million m
3
of soil was estimated to have been eroded during the breaching
event. The flushed debris mass was deposited along the Karli River. Fig. 9 shows the
satellite image and profile of the Karli River before and after the breaching. A major
portion of the eroded material is deposited over the approximately 2km downstream
stretch from the dam along Karli River and less significantly beyond the stretch. The flow
along the Karli River reached the depth up to 20 m. The debris deposit was about 40 m
thick near the toe of the dam.

Fig. 10. (a) Satellite image of the junction of the Karli River with the Jhelum River before and
after the breaching of Hattian dam (Google earth)
200 m 200 m
(a) (b)


229
The dam break flood and debris destroyed one bridge near the junction of the Karli River
with the Jhelum River and damaged about 24 houses as shown in Fig. 10. However major
damage from the dam breach occurred as a landslide on the right bank of the Karli lake that
was triggered by emptying of lake water. The figure 11 shows the satellite image of the right
bank landslide after the breaching event, along with the contour map of the landslide and
longitudinal profile of the landslide. This landslide destroyed 174 houses forcing about 1000
residents to evacuate.

Fig. 11. (a) Satellite image of the Hattian landslide dam, Karli lake and the right bank
landslide after the breaching event, along with the contour map of the landslide, (b) the
longitudinal profile of the right bank slide. Modified from Konagai at el., 2011
The Tang Lake breached during the monsoon rains (July-August) of 2010. In the breaching
process, the water from Tang Lake eroded the north-eastern end of the landslide lobe,
which had been covered with boulders segregated up at the time of deposit, and
approximately 15 m depth of the lake water was drained leaving 1.9 million m
3
water in
the lake (Fig. 12).

230

Fig. 12. Tang Lake after breaching (Photo taken on 6
th
Dec-2010). From Konagai at el., 2011
6. Conclusion
A review of emergency responses and mitigation strategies adopted for the landslide dams
formed in the recent earthquakes shows the problems being faced in the risk management of
the hazards posed by the landslide dams. Excavation of the spillway seems to be the most
common and effective mitigation measure. Meanwhile the time seems to be the most crucial
factor during the emergency response, as the rise of water level inundates more area at its
upstream and increase the lake volume. Higashi Takezawa landslide dam presented an
example where a combination of different mitigation techniques such as pumps, siphons,
diversion channels, and spillway, were possible for the relatively small landslide mass
which was easily accessible. For the large landslide masses at remote locations, spillway
remains as the most suitable mitigation measure.
It is also observed that the failure potential of landslide dammed lakes is not a well
understood phenomenon. The Attabad landslide dam which was expected to fail upon
overtopping due to the material composition of the dam body survived and continues to
pose threats. On the other hand the Hattian landslide dam that seemed to be relatively
stable and expected to remain intact, after showing some signs of instability failed as a result
of moderate rainfall. The risk assessment for landslide dam mostly considers the expected
inundation from dam breach, which might not be sufficient. From the case study of Hattian
landslide it is observed that the damage resulting from the slope instability as a result of
drawdown can be equally severe as the damage resulting from dam breach flood.
The available response time varies from case to case, as in the case of Hattian dam it took
over one year for the lake to fill, while Tangjiashan landslide dammed lake provided less
then a month for the response before the lake filled. The consideration of the recent
landslide dams and their emergency treatment leads to the conclusion that for potentially
large landslide dams, the early identification of possible situation is essential to organize
appropriate preventive actions in due time. A detailed knowledge of the past events like
those mentioned above will always be helpful determining the optimum actions to be
adopted considering the fact that the significant bases in this field are lacking.
7. Acknowledgment
This paper summarizes one of the outcomes of the MEXT Research Project, Scientific
surveys for long-lasting geotechnical problems caused by large earthquakes and their
15m
Approx.
Breach channel
Tang Lake


231
implementations for rational rehabilitation strategies, Konagai K. Leader of the project,
2008 Grant-in-aid for scientific research (A) No. 20254003, Ministry of Education, Culture,
Sports, Science and Technology (MEXT).
8. References
Bonnard C. (2006). Technical and Human aspects of historic rockslide dammed lakes and landslide
dam breaches, Italian journal of engineering geology and environment, Special issue
1 doi: 10.4408/IJEGE.2006-01.S-03
Costa J. E. and Schuster R. L. (1988). The formation and failure of natural dams, GSA Bulletin;
July 1988; v. 100; no. 7; p. 1054-1068; DOI: 10.1130/0016-
7606(1988)100<1054:TFAFON>2.3.CO;2
Dunning S. A., Mitchell W. A., Rosser N. J., Petley D. N. (2007). The Hattian Bala rock
avalanche and associated landslides triggered by the Kashmir Earthquake of 8th October
2005, Eng Geol 93(34):130144
Ermini L., Casagli N. (2003). Prediction of the behaviour of landslide dams using a
geomorphological dimensionless index, Earth Surface Processes and Landforms,
Volume 28, Issue 1, pages 3147, January 2003 DOI: 10.1002/esp.424
Geological Survey of Pakistan (GPS). (2010). Post disaster threats in Attabad and adjacent areas,
Hunza valley, Gilgit-Baltistan, Information release no 893, April 2010
Konagai K., Ahsan S. (2011). Partial breaching of Hattian Bala landslide dam formed in the 8th
October 2005 Kashmir earthquake, Pakistan, Landslides Journal (in review)
Liu N., Chen Z., Zhang J., Lin W., Chen W. and Xu W. (2010). Draining the Tangjiashan Barrier
Lake , DOI: 10.1061/ ASCE HY.1943-7900.0000241, Journal of Hydraulic
Engineering, Vol. 136, No. 11, November 1, 2010
Pamir Times. Voices of the Mountain Communities, community news and views blog of Gilgit
Baltistan. http://pamirtimes.net/
Petley D. (2010). The landslide at Attabad in Hunza, Gilgit/Baltistan: current situation and
hazard management needs. Report prepared for Focus Humanitarian Assistance,
Pakistan, based upon a rapid field assessment on 26th February 4th March
2010
Qiang X., Xuan-Mei F., Run-Qiu H., Cee V. (2009). Westen Landslide dams triggered by the
Wenchuan Earthquake, Sichuan Province, south west China. Bull Eng Geol Environ
(2009) 68:373386 DOI 10.1007/s10064-009-0214-1
Randall W. Jibson and Edwin L. Harp (2011). Field Reconnaissance Report of Landslides
Triggered by the January 12, 2010, Haiti Earthquake Open-File Report 2011102 U.S.
Department of the Interior U.S. Geological Survey
Sassa K. (2005). Landslide disasters triggered by the 2004 Mid-Niigata Prefecture earthquake in
Japan Landslides, (2005) 2: 135142. DOI: 10.1007/s10346-005-0054-4
Takeji Kokusho and Tadashi Hara (2005): Photographs of Landslides triggered in the 2004
Chuets Earthquake, Earthquake damage in active-folding areas: Creation of a
comprehensive data archive and suggestions for its application to remedial
measures for civil-infrastructure systems (Project leader: Kazuo Konagai), Research
and Development program for Resolving Critical Issues, Special Coordination
Funds for Promoting Science and Technology, Ministry of Education, Culture,
Sports, Science and Technology, 2005-2007.

232
WAPDA Report. (2007). Study of Hattian Ballah landslide, Potential hazards of land sliding and
mitigation measures at Hattian Ballah and other earthquake hit areas, National
Engineering Services Pakistan limited (NESPAK) and Geological Survey of
Pakistan (GSP), Water and Power Development Authority (WAPDA), 2007,
Government of Pakistan
10
Rate-Dependent Nonlinear Seismic Response
Analysis of Concrete Arch Dam
Xiao Shiyun
Faculty of Infrastructure Engineering, Dalian University of Technology, Dalian,
China
1. Introduction
In civil engineering, all kinds of concrete structures inevitably encounter some form of
dynamic loading during their lifetime. For example, bridges and tall buildings encounter
wind loading, dams suffer from hydrodynamic pressure loading, ocean platforms encounter
the impact of ocean waves, and all kinds of structures may suffer from earthquake loading.
Because of their unpredictability and destructive capacity, these kinds of loadings always
become important factors in controlling structural design.
The concrete is a typical rate-dependent material: its strength, stiffness, and ductility (or
brittleness) are affected by loading rates. Researches on the rate dependency of concrete
started in 1917 by the Abrams dynamic compressive experiment (Abrams

1917). Jones

(1936)
investigated the relationship between compressive strengths of concrete and loading rates.
Their experiments gave the conclusion that the compressive strengths of concrete increased
with loading rates. Numerous tests have been carried out to investigate the response of
concrete to rapid loading. Watstein

(1953) observed that the compressive strengths of concrete
increased an average of 80 percent when the strain rate increased from the static loading rate
10
-6
s
-1
to 10s
-1
. Based on the results of his experiments, Norris (1959) proposed an empirical
formula and predicted that the compressive strengths were increased up to 33%, 24%, and 17%
greater than the static strengths when the strain rates were 3s
-1
, 0.3s
-1
, and 0.1s
-1
, respectively.
Atchley

(1967) reported that the dynamic compressive strength increased from 25% to 38%.
Experimental results from Hughes

(1972) illustrated that the compressive strengths of concrete
increased 90% more than the static strength on the impact loading.
Although researchers are not in complete agreement with which strain rates cause the
increase in strength to be significant, it is generally accepted that a definite increase in the
uniaxial compressive strength of concrete correlates with the increase of strain rates.
However, confusion also has arisen in regard to the increase in magnitude of dynamic
strengths. Some experimental results (Abrams

1917; Jones

1936; Watstein

1953; Rush 1960;
Atchley 1967; Spooner 1972; Hughes 1972; Sparks 1973; Dilger 1984) showed that an increase
of 30 percent more than the static strength of concrete, and even up to 80 percent, is possible.
Others, such as Moore

(1934), Evans

(1942) and Dhir

(1972), indicated that the increase in the
strength of concrete was less than 20 percent and was not influenced by the rate of loading.
Bischoff

(1991) reviewed and analyzed the dynamic compressive experiments of concrete
and deduced that the confusion about the increased magnitude of dynamic strengths arose

234
because many factors, such as concrete quality, aggregate, age, curing and moisture
conditions, influenced the behavior during the rapid loading.
The reported dynamic tensile tests of concrete in literature are more difficult to perform and
the results are few. Birkimer

(1971) conducted two sets of dynamic tensile tests using plain
concrete cylinders. In the first set, the dynamic strength at the strain rate of 20s
-1
was
between 17.2 MPa and 22.1 MPa, whereas the static tensile strength was 3.4 MPa at the
quasi-static strain rate of 0.5710
-6
s
-1
. In the second set, the concrete dynamic strength was
between 15.4 MPa and 27.6 MPa. Zielinski

(1981) studied the behavior of concrete subjected
to the uniaxial impact tensile loading and found that the ratios of impact and static tensile
strengths were between 1.33 and 2.34 for various concrete mixes. Oh

(1987) presented a
realistic nonlinear stress-strain model that could describe the dynamic tensile behavior of
concrete. An equation was proposed to predict the increase of tensile strengths resulting
from an increase of strain rate. Tedesco

(1991) conducted the direct tension tests of plain
concrete specimens on a split-Hopkinson pressure bar to investigate the effects of increasing
strain rate on the tensile strength of concrete. Rossi

(1994) made an experimental study of
rate effects on the behaviors of concrete under tensile stress to investigate the effect of the
water/cement ratio on the tensile strength enhancement. In addition, an analysis of the
physical mechanisms was developed to investigate how the Stefan effect, the cracking
process, and the inertia forces participated together in the dynamic behavior of a specimen
subjected to a uniaxial tensile test

(Rossi, 1996). Cadoni

(2001) studied the effect of strain rate
on the tensile behavior of concrete at different relative humidity levels. Malvar

(1998)
reviewed the extant data characterizing the effects of strain rate on the tensile strength of
concrete and compared the DIF formulation with that recommended by the European CEB.
Finally, an alternative formulation was proposed based on the experimental data.
Many high arch dams have been built and will be built in areas of China with high seismic
activity. Some of them will reach 300 meters in height. For researchers and engineers, the
significant concern is paid on the safety of these structures against earthquake shocks.
During the past two or three decades, many sophisticated computer programs are
developed and used for numerical analysis of the arch dams. Our ability to analyze
mathematical models of dam structures subjected to earthquake ground motions has been
improved dramatically. Nevertheless, the current design practice in the seismic design of
arch dams has to be based on the linear elastic assumption. The key property that
determines the capacity of arch dams to withstand earthquakes is the tensile strength of
concrete. However, the design criterion for the tensile stress remained a problem at issue. A
widely accepted standard has not been available. The conventional design practice accounts
for the rate sensitivity by means of drastic simplifying assumptions. That is, in all cases, the
allowable stresses of an arch dam under earthquake load are increased by, such as a Chinese
Standard (2001), 30% of the value specified for static case. Similarly, the dynamic modulus
of elasticity is assigned 30% higher than its static value. Raphael (1984) carried out the
dynamic test of concrete from dam cores and reported an average dynamic-static splitting
tensile strength ratio of 1.45, and an average dynamic-static compressive strength ratio of
1.31 for the same loading rate ranges. Harris (2000) performed laboratory tests on concrete
cores drilled from dams and tested at strain rates that simulated dynamic and static loading
conditions. Results indicated that dynamic-static strength ratios were greater than those for
both the tensile and compressive strength tests. Thereby, it is improper that the same
increments of strengths and elastic modulus of concrete at different strain rates are adopted
during the process of analyzing the seismic response of dams.

Rate-Dependent Nonlinear Seismic Response Analysis of Concrete Arch Dam

235
Few researchers considered the effect of strain rates on dynamic responses of arch dams
because there was a lack of rate-dependent dynamic constitutive models of concrete.
Cervera (1996) developed a rate-dependent isotropic damage model for the numerical
analysis of concrete dams. The application of the proposed model to the seismic analysis of a
large gravity concrete dam showed that the tensile peak strength of concrete could be
increased up to 50 percent for the range of strain rates that appear in a structural safety
analysis of a dam subjected to severe seismic actions. Lee (1998) developed a plastic-damage
model for the concrete subjected to cyclic loading using concepts of fracture-energy-based
damage and stiffness degradation. The rate-dependent regularization was used to guarantee
a unique converged solution for softening regions. No effect for the rate-dependency on the
stress distribution has been involved. Chen (2004) proposed a rate-dependent damage
constitutive model for massive concrete by introducing rate-dependant plastic damage
variables as internal variables. The nonlinear seismic responses of arch dams were
computed using this model and the results were compared with the results given by the
corresponding rate-independent damage model. It showed that the distribution of strain
rates not only influenced the vibration modes of dam but also had significant effects on the
dynamic damage of arch dams. Li (2005) analyzed the seismic response of a high arch dam,
in which a rate-dependent damage constitutive model of concrete was considered and the
nonlinear contact of joints was simulated by direct stiffness method based on the Lagrange
multiplier. The study showed that the nonlinear concrete model had great effects on the
dynamic opening of the contraction joints caused by the nonlinear softening and cracking.
Bai (2006) established a rate-dependent damage constitutive model for simulating the
mechanical behaviors of concrete by introducing the effect of strain rate into the damage
tensor. The model was applied to analyze the seismic overload response of a typical
concrete gravity dam. Results indicated that the distribution of strain rate caused by seismic
loading varied at the dam surface and significantly affected the dynamic response.
The effect of strain rates on dynamic behaviors of concrete is an important aspect in the
evaluation of the seismic responses of concrete structures. To evaluate the seismic behaviors
of concrete structures, the dynamic experiment and the dynamic constitutive model of
concrete are necessary. The main objective of this study is, based on the results of the
dynamic uniaxial tensile and compressive experiments on the concrete, to establish the
dynamic constitutive model of concrete and study the effect of strain rates on dynamic
responses of concrete dams.
2. Dynamic experiments of concrete
2.1 Dynamic uniaxial tensile experiment of concrete
2.1.1 Tensile specimen
A concrete mix with proportions, by weight, of cement: water: gravel: sand content =
1.00:0.75:4.09:2.56 was used in the study. The employed cement is 425 Portland cement,
the fine aggregate is general river sand, and the coarse aggregate is crushed rock.
Specimens were cast in steel moulds and cured in moisture condition for 7 days, then they
were naturally cured at 203 Celsius degree temperature in the laboratory. Fifty
dumbbell-shaped specimens were cast for the tensile experiment, as shown in Fig.1. The
specimens of dumbbell shape ensure that the specimens were destroyed at the middle of
specimen in tension.

236

Fig. 1. Tensile specimen
2.1.2 Tensile test loading system and measuring system
The tensile dynamic test was carried out using the 1000-kN servo fatigue testing machine at
the State Key Laboratory of Coastal and Offshore, Dalian University of Technology. During
the process of experiments, the loading sign is sent by the control center, and then it is
transferred to the servo fatigue testing machine. The magnitude and frequencies of loading
are controlled by the control center.
During the tensile test, the specimen was adhered to two steel plates by the constructional
glue. The bottom steel plate was fixed to the base with bolts and the upper steel plate was
connected to the load cell with the load transducer, as shown in Fig. 2. In order to increase
the stiffness of the loading system, a frame was formed by four steel bars connecting the
load cell to the base with bolts.

Fig. 2. Tensile test setup


237
The data acquisition processor is 32-channel. Vertical and lateral strains of specimens were
measured by four pairs of decussate strain gauges adhered on the four sides of specimens.
Vertical and lateral displacements were measured by two opposite Linear Variable
Displacement Transducers (LVDT) fixed to two opposite sides of the specimen. The load
was measured by the load transducer fixed to the specimen. All measured signals were
transmitted to the data acquisition and processing system of the computer through a
specially allocated amplifier.
2.1.3 Analysis of the tensile experimental results
2.1.3.1 Stain-rate influence on uniaxial strength of concrete
The dynamic tensile strengths of 15 specimens under different strain rates of 10
-5
s
-1
, 10
-4
s
-1
,
10
-3
s
-1
and 10
-2
s
-1
are shown in table 1. It shows that the uniaxial tensile strengths of concrete
increase with the increasing of strain rates. Compared to the quasi-static tensile strength of
concrete at the strain rate of 10
-5
s
-1
, the dynamic tensile strengths of concrete at strain rates of
10
-4
s
-1
, 10
-3
s
-1
and 10
-2
s
-1
increase 6%, 10% and 18%, respectively.

Strain rate
/ s
-1

Tensile Strengths / MPa
10
-5
10
-4
10
-3
10
-2

1 1.388 1.435 1.549 1.639
2 1.582 1.480 1.502 1.753
3 1.469 1.604 1.696 1.731
4 1.355 1.606 1.629
average 1.449 1.531 1.594 1.696
Table 1. Dynamic strength of concrete in tension
According to the references, the increases in strengths follow a linear-logarithmic relationship
with the increases in strain rates. By test results the linear-logarithmic relationship between the
tensile strength enhancement with the strain rate enhancement, is given by
( ) 1.0 0.057log
t ts t ts
f f c c = + (1)
Fig.3 shows the relationship between the dynamic tensile strength and the static tensile
strength of concrete at different strain rates of 10
-5
s
-1
, 10
-4
s
-1
, 10
-3
s
-1
and 10
-2
s
-1
.
2.1.3.2 Stain-rate influence on elastic modulus
The stress-strain curves of concrete in tension at different strain-rate loading are illustrated
in Fig.4. It is clear that during different strain rate loading the slope of curves is linear at the
beginning of loading, indicating that the initial tangent modulus of concrete is independent
of strain rate.

238

Fig. 3. The relationship between the dynamic tensile strength and the static tensile strength

Fig. 4. The stress-strain curves of concrete in tension


239
2.1.3.3 Stain-rate influence on critical strain
Table 2 illustrates the critical strain of concrete in tension obtained from the present test.
Table 2 implies that in the range of strain rate from 10
-5
s
-1
to 10
-2
s
-1
, the effect of strain rate
on the critical strain value of concrete in tension is little, if any.

Strain rate 10
-5
s
-1
10
-4
s
-1
10
-3
s
-1
10
-2
s
-1

1 106 106 122 102
2 127 130 98 114
3 113 112 108 123
4 102 114 110
Average 112 116 110 113
Table 2. The critical strain of concrete in tension
2.1.3.4 Stain-rate influence on Poissons ratio
Table 3 shows the change in Poissons ratio of concrete in tension and compression. It is
found that the maximum of Poissons ratio is 0.20 but the minimum is 0.13. And it is
concluded that Poissons ratio isnt obviously dependent on the loading rate according to
the average of results.

Strain rate 10
-5
s
-1
10
-4
s
-1
10
-3
s
-1
10
-2
s
-1

1 0.16 0.17 0.14 0.15
2 0.14 0.16 0.14 0.19
3 0.19 0.18 0.15 0.13
4 0.14 0.13 0.16
Average 0.16 0.16 0.15 0.16
Table 3. The Poissons ratio of concrete in tension
2.2 Dynamic uniaxial compressive experiment of concrete
2.2.1 Compressive specimen
Similar to the tensile specimen, the concrete mix with proportions, by weight, of cement:
water: gravel: sand content was still 1.00:0.75:4.09:2.56. The employed cement is 425 Portland
cement, the fine aggregate is general river sand, and the coarse aggregate is crushed rock.
Specimens were cast in steel moulds and cured in moisture condition for 7 days, then they
were naturally cured at 203 Celsius degree temperature in the laboratory. Fifty cuboid
specimens with 100100300mm were cast for the compressive experiment, as shown in Fig.5.

240

Fig. 5. Compressive specimen
2.2.2 Compressive test loading system and measuring system
The compressive dynamic test was also carried out using the 1000-kN servo fatigue testing
machine at the State Key Laboratory of Coastal and Offshore, Dalian University of
Technology. As for the compressive test, the cuboid specimen was placed vertically on the
circular steel plate connecting the base with the load transducer, as shown in Fig. 6. The
measure system is same as the tensile test.

Fig. 6. Compressive test setup


241
2.2.3 Analysis of the compressive experimental results
2.2.3.1 Stain-rate influence on uniaxial strength of concrete
Table 4 gives the dynamic compressive strengths of concrete at different strain rates. It can
be concluded that the uniaxial compressive strengths of concrete increase with the
increasing of strain rate. Compared to the quasi-static compressive strength of concrete at
the strain rate of 10
-5
s
-1
, the dynamic compressive strengths of concrete at strain rate of 10
-4
s
-
1
, 10
-3
s
-1
, 10
-2
s
-1
and 10
-1
s
-1
increase 4.8%, 9.0%, 12.0% and 15.6%, respectively.

Strain rate
/ s
-1

Compressive strengths / MPa
10
-5
10
-4
10
-3
10
-2
10
-1

1 21.89 22.09 24.63 24.04 26.19
2 22.03 24.40 23.18 25.97 24.94
3 20.67 21.72 22.60 22.70 26.72
4 23.35 23.93 25.44 25.76 23.76
average 22.00 23.03 23.96 24.62 25.40
Table 4. Dynamic strengths of concrete in compression
Similar to the dynamic tensile strength, the linear-logarithmic relationship between the
compressive strength enhancement with the strain rate enhancement, is also given by
( ) 1.0 0.040log
c cs c cs
f f c c = + (2)
Fig.7 gives the relationship between the dynamic compressive strength and the static
compressive strengths of concrete.

Fig. 7. The relationship between the dynamic and static compressive strength

242
2.2.3.2 Stain-rate influence on elastic modulus
The stress-strain curves of concrete in compression at different strain-rate loading are
plotted in Fig.8. The initial tangent modulus of concrete in compression slightly increased as
the strain rate increased. The initial tangent moduli of concrete at strain rate of 10
-4
s
-1
, 10
-3
s
-1
,
10
-2
s
-1
and 10
-1
s
-1
increase to 1.310
4
MPa, 1.3810
4
MPa, 1.4810
4
MPa and 1.6010
4
MPa from
1.2310
4
MPa at strain rate of 10
-5
s
-1
, respectively.

Fig. 8. The stress-strain curves of concrete in compression
2.2.3.3 Stain-rate influence on critical strain
Table 5 gives the results of the critical compressive strain of concrete obtained from the
present test. It shows that a slight decrease in the critical compressive strain value was
observed as the strain rate was increased.

Strain rate 10
-5
s
-1
10
-4
s
-1
10
-3
s
-1
10
-2
s
-1
10
-1
s
-1

1 2524 2164 2480 2382 2030
2 2713 2153 2244 2172 2114
3 2180 2633 2083 2231 2234
4 2353 2304 2137 2003 1965
Average 2443 2314 2236 2197 2086
Table 5. The critical strain of concrete in compression
2.2.3.4 Stain-rate influence on Poissons ratio
Table 6 shows the change in Poissons ratio of concrete in tension and compression. It is
found that the maximum of Poissons ratio is 0.20 but the minimum is 0.13. And it is
concluded that Poissons ratio isnt obviously dependent on the loading rate according to
the average of results.


243
Strain rate 10
-5
s
-1
10
-4
s
-1
10
-3
s
-1
10
-2
s
-1
10
-1
s
-1

1 0.18 0.14 0.20 0.14 0.15
2 0.13 0.19 0.18 0.18 0.17
3 0.15 0.16 0.14 0.17 0.17
4 0.17 0.15 0.17 0.20 0.15
Average 0.16 0.16 0.17 0.17 0.16
Table 6. The Poissons ratio of concrete in compression
3. Rate-dependent constitutive model of concrete
3.1 Consistency viscoplastic model theory
The consistency viscoplastic model can be seen as an extension of the classic elasto-plastic
model to account for the rate-dependent behavior of materials. This model, which uses the
Von Mises yield surface, was applied by Wang

(1997) to analyze metal. With this method,
the consistency viscoplastic Hoffman model of concrete was modified by Winnicki

(2001). In
this model, during viscoplastic flow, the actual stress state must remain on the yield surface,
and the consistency condition is imposed.
The viscoplastic yield function can be expressed as
( , , ) 0
ij
F o k k = for 0 >
(3)
In uniaxial tension and compression, equation (3) can be expressed as

( , , )
( , , )
c ij c c
t ij t t
F F
F F
o k k
o k k
=
`
=

)
(4)
It is difficult to establish biaxial or triaxial constitution relations because of the lack of biaxial
dynamic experiment results for the concrete. For simplicity, the biaxial dynamic behavior of
concrete is assumed to be the same as the uniaxial dynamic behavior with an increasing
factor
bc
K for the strength such that:
( , , )
bc bc ij c c
f K f o k k = (5)
On an arbitrary stress state, it is assumed that:
( )
c c ij
k o k = ( )
t t ij
k o k = (6)
Functions ( )
t ij
o and ( )
c ij
o should be chosen in such a way that, for loading processes
with dominant tensile stress states, ( ) 1
t ij
o = and ( ) 0
c ij
o = ; similarly, for loading
processes with dominant compressive stress states ( ) 0
t ij
o = and ( ) 1
c ij
o = , and for
loading processes with tensile-compressive stress states 0 ( ) 1
t ij
o < < and 0 ( ) 1
c ij
o < < .
These functions must satisfy the condition of ( ) ( ) 1
t ij c ij
o o + = . Consequently, weight
functions ( )
t ij
o and ( )
c ij
o can be achieved expediently.
Using such conditions, the yield function can be expressed as

244
( , , ) ( , , , , ) 0
ij ij c c t t
F F o k k o k k k k = = (7)
At the same time, the viscoplastic consistency condition must be satisfied so that:

0
ij c c t t
ij c c t t
F F F F F
d d d d d o k k k k
o k k k k
c c c c c
+ + + + =
c c c c c

(8)
The effects of
t
k and
t
k on
t
f are assumed to be independent and the instantaneous tensile
strength
t
f is formulated in a very general way as follows

( ) ( )
t ts t t t t
f f H R k k =
(9)
Similarly, the instantaneous compressive strength is computed as
( ) ( )
c cs c c c c
f f H R k k = (10)
where ( )
t t
H k , ( )
t t
R k , ( )
c c
H k and ( )
c c
R k are the assumption functions achieved by
experiment. Here, k is adopted as 2 /3
vp vp VP
ij ij
k c c c = = .
3.2 Consistency viscoplastic William-Warnke three-parameter model of concrete
(xiao, 2010)
In this study, the yield surface function is assumed to be the same as the failure surface
function of concrete so that the yield function of the William-Warnke three-parameter model
can be expressed as

1 1
( , , ) 0
( )
m m m m c
F f
r
o t u o t
u
= + =
(11)
where two parameters and ( ) r u are chosen by three conditions: 1) uniaxial tensile
strength
t
f , 2) uniaxial compressive strength
c
f , and 3) biaxial compressive
strength
bc bc c
f k f = .
Defining
ij
ij
F
m
o
c
=
c
, one can achieve:

2
1 1 1 1
3 5
m m
ij
ij m ij m ij ij
m
ij ij
m ij
F F F F r
m
r
r
s
r r
o t
o o o t o o
t
o
t o
c c c c c c c
= = + +
c c c c c c c
c
= +
c
(12)
where
1 ij ij
I o o = c c and
2 ij ij
s J o = c c .
For the sake of simplicity, defining
1
3
A
= ,
1
5
m
B
rt
= and
2
m
C
r
t
= and substituting them
into equation (11), one obtains:


245

ij ij ij ij ij
ij ij ij
F r r
m A Bs C A Bs C
u
o o
o o u o
c c c c
= = + + = + +
c c c c
(13)
Defining
u
r
v
= and taking the partial derivative with respect to the stress tensor
ij
o , one
obtains:

2
( )
du dv
v u
u v r
d d
D
v
u u
u u

c c
= = =
c c
(14)
where
1
2 2 2 2 2 2
2
2 ( )cos (2 )[4( )cos 5 4 ]
c c t c t c c t t c t
u r r r r r r r r r r r u u = + +

2 2 2 2
4( )cos ( 2 )
c t c t
v r r r r u = +
2 2
2 2
1
2 2 2 2
2
4 (2 )( )sin cos
2 ( )sin
[4( )cos 5 4 ]
c t c t c
c t c
c t t c t
r r r r r du
r r r
d
r r r r r
u u
u
u
u

= +
+

2 2
8( )sin cos
t c
dv
r r
d
u u
u
=

According to the William-Warnke three-parameter model, the relationship between u and
ij
o is given as follows:

3
3
2
2
3 3
cos3
2
J
J
u =
(15)
By taking the partial derivative of equation (11) with respect to the stress tensor
ij
o , the
following expression can be obtained as:

3 3 3 2 2
5 3
2 2
2 3
2 2
3
5 3
2 2
2 2
3 3 3 1
4sin3 2sin3
3 3 3 1
4sin3 2sin3
ij ij ij ij ij
ij ij
J J J J J
J J
J J
J
s t
J J
u u u
o o o u o u o
u u
c c c c c c c
= + = +
c c c c c c c
=
(16)
where
3
ij
ij
J
t
o
c
=
c
and
ij
t has the behavior such as:
(1) 0
ij ij
t o = (2)
3
3
ij ij
t s J = (3)
2
2
2
3
ij ij
t t J =

246
Simply, defining
3
5
2
2
3 3
4sin3
J
E
J
u
= and
3
2
2
3 1
2sin3
U
J
u
= , equation (16) can be expressed
as

ij ij
ij
E s U t
u
o
c
= +
c
(17)
Substituting equations (10) and (13) into equation (9),
ij
m is simplified as:

( )
( )
ij ij ij ij ij
ij
ij ij ij
ij ij ij
F
m A B s CD E s U t
A B CDE s CDU t
s t
o
o
o
oo |
c
= = + + +
c
= + + +
= + +
(18)
where A o = , B CDE | = + and CDU = .
In equation (5),
c
F
k
c
c
,
c
F
k
c
c
,
t
F
k
c
c
and
t
F
k
c
c
could be expressed as
( )
c c
c c c c c c
f f F F F F r
f f r f
k k k
c c c c c c c c
= = +
c c c c c c c c

( )
c c
c c c c c c
f f F F F F r
f f r f
k k k
c c c c c c c c
= = +
c c c c c c c c

( )
t t
t t t t t t
f f F F F F r
f f r f
k k k
c c c c c c c c
= = +
c c c c c c c c

( )
t t
t t t t t t
f f F F F F r
f f r f
k k k
c c c c c c c c
= = +
c c c c c c c c

where the curves of
c
c
f
k
c
c
,
c
c
f
k
c
c
,
t
t
f
k
c
c
and
t
t
f
k
c
c
can be achieved by uniaxial compressive and
tensile tests of concrete.
Based on the associated plastic flow rule, the viscoplastic strain is defined as

vp
ij ij
ij
F
d d d m c
o
c
= =
c
(19)
The invariable can be expressed as

2 2
( )
3 3
vp vp
ij ij ij ij ij
d d d m m d g d k c c o = = = (20)


247
Using these functions, the consistency equation (4) can be expressed as

0
ij ij
m d hd sd o + + =
(21)
where
( ) ( ) ( ) ( )
c c c c c t t t t t
F
h a h R a h R k k k k
c
= = +
c

( ) ( ) ( ) ( )
c c c c c t t t t t
F
s a H r a H r k k k k
c
= = +
c

In which
( ) ( )
t ts t ij ij
t
F
a f g
f
o o
c
=
c
( ) ( )
c cs c ij ij
c
F
a f g
f
o o
c
=
c

( )
( )
t t
t t
t
H
h
k
k
k
c
=
c

( )
( )
c c
c c
c
H
h
k
k
k
c
=
c

( )
( )
t t
t t
t
R
r
k
k
k
c
=
c

( )
( )
c c
c c
c
R
r
k
k
k
c
=
c

Dynamic tensile and compressive tests were carried out to investigate the effect of strain
rates on the dynamic tensile and compressive behaviors of concrete

(Xiao, 2008). Test results
indicate that the tensile and compressive strengths of concrete increase with the increase of
the loading rate. The initial tangential modulus and the critical strain of concrete in tension
are independent of strain rate, but those in compression slightly increased with the strain
rate. Poissons ratio of concrete in both tension and compression is not obviously dependent
on loading rate.
Based on the experimental data, the functions ( )
t t
H k and ( )
t t
R k are given in Fig. 9(a) and
Fig. 9(b), where the plotted curves are the fitting curves for the later calculation.

(a) Function of ( )
t t
H k (b) Function of ( )
t t
R k
Fig. 9. Function of ( )
t t
H k and ( )
t t
R k

248
Similarly, the functions of ( )
c c
H k and ( )
c c
R k are given in Fig. 10(a) and Fig. 10(b).

(a) Function of ( )
c c
H k (b) Function of ( )
c c
R k
Fig. 10. Function of ( )
c c
H k and ( )
c c
R k
3.3 Euler return mapping algorithm
At the time t , the stress
t
ij
o , the invariable
t
k and the rate of invariable
t
k should satisfy
the yield condition:

( , , ) 0
t t t
ij
F o k k =
(22)
At the time t t A + , the stress can be written as

( , )
ep t t t t
ij ij ij ij ij kl ijkl
D
A
o o Ao o o A Ac
+
= + = +
(23)
where ( , )
ep
ij ijkl
D o A is the tangent module of the consistency viscoplastic model.
Also, the stress
t t
ij
A
o
+
, the invariable
t t A
k
+
and the rate of invariable
t t A
k
+
at the time t t A +
should satisfy the yield condition:

( , , ) 0
t t t t t t
ij
F
A A A
o k k
+ + +
=
(24)
At the short time increment, the assumption is an approximation for
as

t t
t
A
A
+
=
(25)
Thus, the internal parameter
t t A
k
+
and its rate
t t A
k
+
can be expressed by
( ) ( )
t t
t t t t t t t t
ij ij
g g
t
A
A A A A
A
k o o
A
+
+ + + +
= =
(26)


249
( )
t t t t t t t
ij
g
A A A
k k A o
+ + +
= + (27)
Consequently, the yield condition at the time t t A + can be formulated using the above
expression as in the classic rate-independent plasticity as follows:
( , ) 0
t t t t
ij
F
A A
o A
+ +
= (28)
During the th k iteration at the time t t A + the stress
k
ij
o and plastic multiplier
k
A might
not satisfy the yield condition:
( , ) 0
k k k
ij
F o A = (29)
Generally, the stress
k
ij
o does not reflect the real stress
t t
ij
A
o
+
at the end of the given time
t t A + . So, the residual stress can be expressed as
( , ) ( , )
ep k k k k t t k t
ij ij ij ij ij ij ij kl ijkl
r D
A
o A o o o o o A Ac
+
= = (30)
During the ( 1)th k + iteration at the time t t A + the yield function value
1 k
F
+
and the
residual stress
1 k
ij
r
+
can be achieved from the truncated Taylor's series expansion of the
yield F and the residual stress
ij
r about position k , and then set to zero:

1
0
k k
k k k k
ij
ij
F F
F F oo oA
o A
+
c c
= + + =
c c
(31)

1
0
k k
k k k k
ij ij ij
ij
r r
r r oo oA
o A
+
c c
= + + =
c c
(32)
Equations (31) and (32) represent a set of linear equations for the iterative stress update
k
ij
oo
and the iterative viscoplastic multiplier update
k
oA .
Defining
| |
k k
ij
k k
ij
F F
r r
A o
A o
(
c c
(
c c
(
=
(
c c
(
(
c c

A

{ }
k
k
ij
F
r

=
`

)
b

equations (31) and (32) can be expressed as

250
| | { }
k
k
ij
oA
oo

=
`

)
A b
(33)
where
1
k k k k k k
k
F F F d F F s
h
d t t
A A A A
c c c c c
= + = + = +
c c c c c

ij ij ij ij
ij
F
m s t oo |
o
c
= = + +
c

( )
ep
kl ijkl
D
r
Ac
A A
c
c
=
c c

( )
ij
ep
ijkl pq ijpq
kl kl
r
I D Ac
o o
c
c
=
c c

where
ijkl
I was the identity tensor of the fourth order.
Taking into account the symmetry of the stress tensor, the iterative stress update
k
ij
oo and
the iterative viscoplastic multiplier update
k
oA are achieved by solving the set of linear
equations. The iteration process continues until the norms F and
ij
r reaches to reasonably
small ammount. The final values of A and
ij
Ao at the end of the time step are obtained by
a summation process:

1
N
k
k
A oA
=
=
(34)

1
N
k
ij ij
k
Ao oo
=
=
(35)
where N is the total number of iterations.
3.4 Tangent module of the consistency viscoplastic model
According Hookes law, the stress change can be written as

( )
ijkl kl ijkl kl ijkl kl ijkl kl kl
vp vp e e e e e
ij
d D d D d d D d D d o c c c c c = = =
(36)
Substituting the equation (36) into the equation (8), the consistency viscoplastic condition is
expressed by
( ) 0
ijkl kl ijkl kl
vp e e
ij
F F F
D d D d d d c c k k
o k k
c c c
+ + =
c c c

(37)


251
According to the flow law, the viscoplastic strain change is given by the following:

vp
ij
ij
Q
d d c
o
c
=
c
(38)
Substituting it into equation (37), one obtains

( ) ( ) ( ) 0
ijkl ijkl
e e
ij
ij ij ij
F F Q F F
D d d D d d c k k
o o o k k
c c c c c
+ + =
c c c c c

(39)
Therefore, the viscoplastic multiplier is expressed as

( )
1 1
( ) ( )
ijkl
ijkl
e
ij
ij
e
ij kl
F
D d
d
F Q F F
D d d
d d
c
o
k k
o o k k
c
c
=
c c c c

c c c c

(40)
Substituting it into equation (36), the following formula is obtained as

1 1
1
vp e e
ij ijkl kl ijkl kl
e e
ijkl kl ijkl
kl
e
mnpq pq
e e mn
ijkl kl ijkl
e
kl
mnpq
mn pq
e e
ijkl mnpq kp lq
e mn kl
ijkl kl
e
mnpq
mn pq
d D d D d
Q
D d D d
F
D d
Q
D d D
F Q F F
D d d
d d
F Q
D D
D d
F Q F
D
d
o c c
c
o
c
o
c
o
k k
o o k k
o o
o o
c
o o
=
c
=
c
c
c c
=
c c c c
c

c c c c
c c
c c
=
c c c
c c
1
( )
kl
vp ep e
ijkl kl kl ijkl ijkl
d
F
d d
d
D D d D d
c
k k
k k
c
c
c c
= + =
(41)
where
ep
ijkl
D is the tangent module of the consistency viscoplastic model.

1 1
ep vp e
ijkl ijkl ijkl
e e
ijkl mnpq kp lq
e mn kl
ijkl
e
mnpq
mn pq
D D D
F Q
D D
D
F Q F F
D d d
d d
o o
o o
k k
o o k k
= +
c c
c c
=
c c c c

c c c c

(42)
Therefore,

252

1 1
e e
ijkl mnpq kp lq
vp
mn kl
ijkl
e
mnpq
mn pq
F Q
D D
D
F Q F F
D d d
d d
o o
o o
k k
o o k k
c c
c c
=
c c c c

c c c c

(43)
The yield function of the classic William-Warnke three-parameter model is expressed as

1 1
( , , ) 0
( )
m m m m c
F f
r
o t u o t
u
= + =
(44)
The divergence of the yield function is written as
ij ij ij
ij
F
s t oo |
o
c
= + +
c
(45)
To the relative flow criteria, the flow law is the same as the yield function of the model:

( , , ) ( , , )
ij ij
Q F o k k o k k =
(46)
Then, the tangent module of the consistency viscoplastic model can be written as follows

1 1
( ) ( )
( ) ( )
(3 2 2 )(
e e
ijkl mnpq kp lq
vp
mn kl
ijkl
e
mnpq
mn pq
e e
ijkl mn mn mn mnpq kl kl kl kp lq
e
mn mn mn mnpq pq pq pq
ij ij ij
F Q
D D
D
F Q F F
D d d
d d
D s t D s t
s
s t D s t h
t
K G s G t
o o
o o
k k
o o k k
oo | oo | o o
oo | oo |
A
oo |
c c
c c
=
c c c c

c c c c
+ + + +
=
+ + + +
+ +
=
2 2 2 2
2 2 3
3 2 2 )
4
9 4 12
3
kl kl kl
K G s G t
s
K G J G J G J h
t
oo |
o | |
A
+ +
+ + +
(47)
Defining
2 2 2 2
0 2 2 3
4
9 4 12
3
s
P K G J G J G J h
t
o | |
A
= + + +

1 0
3 P K P o =

2 0
2 P G P | =

3 0
2 P G P =

Therefore, the equation (47) can be expressed as


253

1 2 3 1 2 3
( )( )
vp
ij ij ij kl kl kl ijkl
D P P s P t P P s P t o o = + + + + (48)
The tangent module of the consistency viscoplastic model is obtained as

1 2 3 1 2 3
( )( )
ep vp e
ijkl ijkl ijkl
e
ijkl ij ij ij kl kl kl
D D D
D P P s P t P P s P t o o
= +
= + + + +
(49)
3.5 Comparisons with experimental data
The stress-strain curves of concrete for the uniaxial tension at the strain rate 10
-3
/s, shown in
Fig. 11(a), are calculated and compared with experimental results. Fig. 11(b) shows the
stress-strain curves of concrete for the uniaxial tension at the strain rate 10
-5
s
-1
, 10
-4
s
-1
, 10
-3
s
-1

and 10
-2
s
-1
nd the comparisons with the experimental data. Similarly, the stress-strain curves
of concrete for the uniaxial compression at the strain rate 10
-2
s
-1
, shown in Fig. 12(a), are
calculated and compared with the experimental results. Fig. 12(b) illustrates the stress-strain
curves of concrete for the uniaxial compression at the strain rate 110
-5
s
-1
, 10
-4
s
-1
, 10
-3
s
-1
, 10
-2
s
-1

and 10
-1
s
-1
and the comparisons with the experimental data.

(a) strain rate 10
-3
/s (b) different strain rates

Fig. 11. Model and tensile test results

254

(a) strain rate 10
-2
/s (b) different strain rates
Fig. 12. Model and compressive test results
The consistency viscoplastic model is modified from the classic William-Warnke three-
parameter model of concrete and has the advantages and disadvantages of the William-
Warnke three-parameter model. It may directly simulate the dynamic behaviors of concrete
and it is simple and easy to calculate. Thus, the proposed model is good for analyzing the
dynamic responses of concrete structures.
3.6 Numerical example
In order to study the effect of strain rate on the dynamic response of concrete structures, the
dynamic response of a simple-supporting beam with dimensions 8m1m1m is analyzed
with this model. Fig. 13 shows the discretized beam and calculated elements adopting three-
dimension eight-node equivalent parameter elements. An impact loading is imposed on the
midpoint of beam and Fig. 14 depicts the loading history. Dynamic response is analyzed
with the ADNFEM program compiled by the authors. The material properties are as
follows: the elastic modulus of concrete is 1.610
4
MPa, the Poissons ratio is 0.17, the mass
density is 2.410
3
kg/m
3
, the static compressive strength is 22MPa and the static tensile
strength is 2.2MPa.

Fig. 13. Structure and meshes of the beam


255

Fig. 14. Time-dependent curve of loading
Fig. 15 illustrates the displacement of midpoint A with time. Three curves shown in the
figure denote the vertical displacement of the beam at model I (linear elastic model), model
II (rate-dependent William-Warnke three-parameter model), and model III (rate-
independent William-Warnke three-parameter model), respectively. Fig. 15 shows that, at
the beginning of loading, the beam is at the elastic state and three displacement curves are
the same. When the stress of the beam reaches to the initial yield stress, the displacement
curves of three models separate, and when time is 0.1 second, the loadings of three models
reach to their maximums, but the displacements of the three models do not reach to their
maximums at the same time. The vertical displacement of model I reaches to its maximum
1.50mm at time 0.114 second, but that of model II reaches to its maximum 1.62mm at time
0.110 second, and that of model III reaches to its maximum 1.90mm at time 0.117 second. It
is clearly shown that displacement of the beam changes greatly after considering the effect
of strain rate. The displacement of the beam with the model II decreases with 14.7 percent
compared with that of the beam with model III but increases with 8.0 percent compared
with that of the beam with model I.

Fig. 15. Displacement of point A

256
At the same time, the stresses of the beam with the three models differ greatly from
each other. Table 7 lists the maximums of the tensile and compressive principal stress
of the beam with the three models. It can be seen clearly that principal stresses of the
beam with different models vary greatly. The tensile principal stress of the beam with
model II increases with 13.10 percent compared with that of the beam with model III,
but the compressive principal stress decreases with 11.6 percent. Compared with model
I, the tensile principal stress of the beam with model II decreases with 16.80 percent, but
the compressive principal stress increases with 19.1 percent. Similarly, the stress
distribution of the beam changes greatly. Fig. 16(a), (b) and (c) show the stress
distribution of the beam with the three models respectively when the displacement is
maximum. The tensile stress distribution figure is shown above and the compressive
stress distribution figure is illustrated below. It can be seen clearly from these figures
that the stress magnitudes and distributions of the beam change greatly with the
different models. Consequently, it can be seen that the dynamic response of the concrete
beam, the displacement, and the stress magnitude and distribution, change greatly after
considering the effect of strain rate.

model
The first principal stress of point B
MPa
The third principal stress of point A
MPa
model I 1.97 -2.04
model II 1.64 -2.43
model III 1.45 -2.75
Table 7. Maximal principal stress of beam
4. Seismic response of arch dam
4.1 Model and parameters of arch dam
In order to illustrate the effect of the rate dependency on the dynamic structural response, a
278m high arch dam in China subjected to earthquake excitation is analyzed by the
proposed model. The dam and the foundation are discretized into 450 and 1,040 three-
dimensional isoparametric 8-node elements, respectively. Fig. 17 shows the discretized dam-
foundation system.
The material properties are as follows: for the dam body, the elastic module is 2.410
4
MPa,
the Poissons ratio is 0.17, the density is 2.410
3
kg/m
3
, the static compressive strength is
30MPa, and the static tensile strength is 3MPa; for the foundation rock, the elastic module is
1.610
4
MPa, the Poissons ratio is 0.25, and the density is 2.010
3
kg/m
3
. The five lowest
vibration frequencies of the dam in care of full reservoir are:
1
f =0.997Hz,
2
f =1.004Hz
,
3
f
=1.450Hz,
4
f =1.497Hz and
5
f =1.542Hz. An assumption of massless foundation is
introduced to simplify the dam-foundation interaction analysis, although more rigorous
interaction effects can be included.


257

(a) model I

(b) model II

(c) model III

Fig. 16. Distribution of principal stress of beam, Pa

258

Fig. 17. Geometry and mesh of arch dam

Fig. 18. Time history of earthquake input
Three-dimensional earthquake waves are used as the input. The design earthquake
acceleration peak is 0.321 gal . Fig. 18 shows the typical artificial unitary acceleragram that met
the requirements of the Chinese Specifications for Seismic Design of Hydraulic Structures.
4.2 Design of the ADNFEM Program
The ADNFEM (Arch Dam Nonlinear Finite Element Method) program compiled by the
authors is used to calculate the dynamic response of arch dams in the rate-dependent
constitutive model. The program is validated by ANSYS in the rate-independent model
when the strain rate was zero in the rate-dependent constitutive model.
The dynamic response of arch dams includes two parts: static analysis and seismic response
analysis. The static analysis of arch dams is carried out to calculate the initial stress and
strain state of the seismic response of arch dams. In order to form the Rayleigh damping
matrix, the model of arch dams is calculated during the process of seismic response analysis.
The processes of static analysis and seismic response analysis are listed as Table 8 and
Table 9.


259
1. Input initial data and form one-dimension bandwidth memory vector and constrain
condition vector;
2. Input the constrain condition;
3. Set the loading step 1 I = ;
4. Form the Ith static load vector f ;
5. Judge elastic-plasticity state of every element and calculate the elastic-plasticity
impact factor;
6. Form the tangent stiffness matrix of structures K ;
7. Calculate the elastic-plasticity predicted state;
8. Set the iterative 1 J = of the Ith loading step;
9. Solve for the iterative displacement
1
o

= u K f ;
10. Update incremental displacements A A o = + u u u ;
11. For every Gaussian integration point:
a. Compute the incremental strains A
b. Compute the elastic-plastically predicted stress state
ep ep
t
A = + D
c. If
ep
violates the yield criterion, perform the return mapping for the
consistency model
12. Check for convergence, if not, JJ+1, go to step 9;
13. Update the loading step II+1, go to step 4;
14. Output the static analysis results.
Table 8. Summary of static analysis

1. Form the mass matrix M and initial stiffness matrix K ;
2. Calculate the first two frequencies
1
e and
2
e with the subspace iterative method;
Input structural damping ratio and calculate the Rayleigh damping factor
1 2
1
1 2
2e e
o
e e
=
+
,
2
1 2
2
o
e e
=
+
;
3. Form the damping matrix
1 2
o o = + C M K ;
4. Input the earthquake wave acceleration
g
a ;
5. Initialize the parameters of the Newmark- method
N
| and
N
;
6. Set the loading step 1 I = ;
7. Form the Ith earthquake load vector
t t t t
g g
A A + +
= f Ma ;
8. Initialize the displacements, velocities, and accelerations;
0 A = u , 0 o = u ;
0
1 1
( 1)
2
t t t t
N N
t
A
| A |
+
= a v a ,
0
(1 ) (1 )
2
N N
t t t t
N N
t
A

A
| |
+
= + v v a

260
9. Set the iterative 1 J = of the Ith loading step;
10. Update velocities and accelerations
1
2
1
J J
t t t t
N
t
A A
o
| A

+ +
= + a u a ,
1
1
J J
t t t t
N
t
A A
o
| A

+ +
= + v u v ;
11. Judge elastic-plasticity state of every element and calculate the elastic-plasticity
impact factor;
12. Form the tangent stiffness matrix of structures K ;
13. Compute the equivalent stiffness matrix and equivalent force vector;
2
1
N
N N
t t

| A | A
= + + K K M C ;
1 1
( ) { ( 1) }
2 2
t J J J J N N
g t t t t t t t t
N N N N
t
A A A A

| A | | |
+ + + +
= + + + + f f M v a C v a ;
14. Solve for the iterative displacement
1
o

= u K f ;
15. Update incremental displacements A A o = + u u u ;
16. For every Gaussian integration point:
a. Compute the incremental strains A
b. Compute the elastic-plastically predicted stress state
ep ep
t
A = + D
c. If
ep
violates the yield criterion, perform the return mapping for the consistency
model
17. Check for convergence, if not, JJ+1, go to step 11;
18. Update the loading step II+1, go to step 8;
19. Output the seismic response analysis results.
Table 9. Summary of dynamic analysis
4.3 Seismic response of arch dam
4.3.1 Stresses in arch dams
The dynamic response analyses of the arch dam are performed with three models: model I
(linear elastic model), model II (rate-dependent William-Warnke three-parameter model),
and model III (rate-independent William-Warnke three-parameter model). The maximum
values of the first and the third principle stresses in the dam are shown in Table 10. Fig. 19
shows the distributions of the third principle stresses obtained from the three models. It is
seen that, in all three cases, the maximum compressive stress is the same and appeared at
the bottom of the upstream face; the material remaines working in the elastic range. While,
for the maximum tensile stress there is the marked difference among the calculated results
of the three models. Owing to the plasticity of concrete, the maximal values of the first
principal stresses of model II and III decrease with 37.7% and 44.5%, respectively, compared
with model I. Because the dynamic tensile strength of concrete increases with the increase of
strain rates, the maximal values of the first principal stresses of model II, taking into account
the effect of strain rates, increase with 12.2% compared with model III. Fig. 20, Fig. 21 and
Fig. 22 show the distributions of the first principle stress obtained from the three models.


261
model
The first principal stress
MPa
The third principal stress
MPa
model I 5.17 -12.50
model II 3.22 -12.50
model III 2.87 -12.50
Table 10. Maximal principal stress of arch dam

(a) upstream face (b) downstream face
Fig. 19. Distribution of the third principle stress (model I, II, and III), Mpa

Fig. 20. Distribution of the first principle stress (model I), Mpa

Fig. 21. Distribution of the first principle stresses (model II), Mpa

262

Fig. 22. Distribution of the first principle stresses (model III), MPa
4.3.2 Strain and strain rate of arch dam
The maximal equivalent strain of concrete in three cases are the same because the
compressive strain of concrete plays a more important role in the equivalent strain of
concrete in the dominant compressive stress states, although the tensile strain is important
in the dominant tensile stress states, but the values are smaller than those for the
compressive strain. Fig. 23 shows the distributions of the equivalent strain of the arch dam
from the three models. It is clear that the maximal equivalent strain is 4.7510
-4
, and it
appears at the bottom of the downstream face. Similarly, the maximal equivalent strain rates
in the three cases are the same, and Fig. 24 shows the distributions of the equivalent strain
rates of the arch dam from the three models. The maximal equivalent strain rate is up to
3.4710
-2
s
-1
and it also appears at the bottom of the downstream face.

Fig. 23. Distribution of the equivalent strain (model I, II, and III)

Fig. 24. Distribution of the equivalent strain rate (model I, II, and III), s
-1


263
4.3.3 Plastic strain and plastic strain rate of arch dam
The equivalent viscoplastic strain of concrete appears only on the tensile zones of the arch
dam. Fig. 25 and Fig. 26 show the distributions of the maximal equivalent viscoplastic
strains obtained from model II and model III, respectively. It is shown that the maximal
equivalent viscoplastic strains appear on the bottom of the upstream face and that the strain
rates has little effect. Fig.27 and Fig.28 show the distributions of the maximal equivalent
viscoplastic strain rates obtained from model II and model III, respectively. Similarly, the
maximal equivalent viscoplastic strain rate appears on the bottom of the upstream face but it
decreases with 17.5% after taking into account the effect of strain rates.


Fig. 25. Distribution of the equivalent viscoplastic strain (model II)


Fig. 26. Distribution of the equivalent viscoplastic strain (model III)

264


Fig. 27. Distribution of the equivalent viscoplastic strain rate (model II), s
-1


Fig. 28. Distribution of the equivalent viscoplastic strain


265
5. Conclusions
Based on the numerical results obtained, the following conclusions are drawn:
1. Results indicate that the tensile and compressive strengths of concrete increase with
increase of the rate of loading. The initial tangent modulus and the critical strain of
concrete are independent of strain rate in tension but slightly increased as the strain rate
increased in compression. Poissons ratio of concrete isnt obviously dependent on the
loading rate both in tension and in compression.
2. Comparisons between the models and experimental data show that the consistency
model may simulate directly the uniaxial dynamic behaviors of concrete. The dynamic
responses of a simple-supporting beam show that dynamic responses of concrete beam,
the displacement, the stress magnitude and distribution, change greatly after
considering the effect of strain rate.
3. In all three cases, the maximum values of the compressive stress are the same and the
concrete remains working in the elastic range. The maximal values of the first
principal stresses of the arch dam, taking into account the effect of strain rates,
increase with 12.2% because the dynamic tensile strength of concrete increases with
the increase of strain rates. There are no effects of strain rates on the maximal
equivalent strain and the maximal equivalent strain rate of the arch dam. The strain
rates have little effect on the maximal equivalent viscoplastic strain, while the
maximal equivalent viscoplastic strain rate decreases with 17.5% after taking into
account the effect of strain rates.
6. Acknowledgements
This study was founded by the National Science Foundation of China under grant
No.50408030 and No. 51121005, the College Innovation Group of Liaoning Province under
grant No: 2008T223 and the Fundamental Research Funds for the Central Universities under
grant No: DUT10LK29.
7. Symbol notation
Symbol Description
t
f

dynamic tensile strength of concrete
ts
f

static tensile strength of concrete
c
f

dynamic compressive strength of concrete
ts
f

static compressive strength of concrete
t
c

dynamic tensile strain rate
ts
c

quasi-static tensile strain rate
c
c

dynamic compressive strain rate
cs
c

quasi-static compressive strain rate
E elastic modulus of concrete
K volume module of concrete
G shear module of concrete

Poissons ratio of concrete

266
ij
o
the stress tensor
ij
s
the deviatoric stress tensor
ij
r
the residual stress tensor
m
o

the average stress
m
t

the average shear stress
ij
c
the strain tensor
vp
ij
c the viscoplastic strain tensor
k internal variable
k rate of internal variable
the viscoplastic multiplier
t

weight function of tensile invariable
c

weight function of compressive invariable
F yield function
Q plastic flow function
1
I

the first invariant of stress tensor
2
J

the second invariant of deviatoric stress tensor
3
J

the third invariant of deviatoric stress tensor
e
ijkl
D
elastic stiffness of concrete
vp
ijkl
D viscoplastic stiffness of concrete
8. References
Abrams DA (1917), Effect of Rate of Application of Load on the Compressive Strength of
Concrete, ASTM Journal, 17: 364-377.
Atchley BL and Furr HL (1967), Strength and Energy-absorption Capabilities of Plain
Concrete under Dynamic and Static Loading, ACI Journal, 64: 745-756.
Bai WF, Chen JY and Zhong H (2006), Seismic Overload Response Analysis of Concrete
Gravity Dam Based on Strain Rate-dependent Damage Model, Journal of Hydraulic
Engineering, 37(7): 820-826. (in Chinese)
Birkimer DL and Lindemann R (1971), Dynamic Tensile Strength of Concrete Materials,
ACI Journal Proceedings, 1: 47-49.
Bischoff PH and Perry SH (1991), Compressive Behaviour of Concrete at High Strain
Rates, Materials and Structures, 24(144): 425-450.
Cadoni E, Labibes K and Albertini C (2001), Strain-rate Effect on the Tensile Behaviour of
Concrete at Different Relative Humidity Levels, Materials and Structures, 34(235):
21-26.
Cervera M, Oliver J and Manzoli O (1996), A Rate-dependent Isotropic Damage Model for
the Seismic Analysis of Concrete Dams, Earthquake Engineering and Structural
Dynamics, 25: 987-1010.


267
Chen JY, Hu ZQ and Lin G (2004), Seismic Analysis of Arch Dam with Joints Based on a
New Strain-rate-dependant Plastic Damage Model, Chinese Journal of Computational
Mechanics, 21(1): 45-49. (in Chinese)
Chinese Standard (2001), Specifications for Seismic Design of Hydraulic Structures, Beijing:
Chinese Electric Power Press. (in Chinese)
Dhir RK and Sangha CM (1972), A Study of Relationships between Time, Strength,
Deformation and Fracture of Plain Concrete, Magazine of Concrete Research, 24(81):
197-208.
Dilger WH, Koch R and Kowalczyk R (1984), Ductility of Plain and Confined Concrete
under Different Strain Rates, ACI Journal, 81(1): 73-81.
Evans RH (1942), Effect of Rate of Loading on the Mechanical Properties of Some
Materials, Journal of Institute of Civil Engineers, 18(7-8): 96-306.
Harris DW, Mohorovic CE and Dolen TP (2000), Dynamic Properties of Mass Concrete
Obtained from Dam Cores, ACI Material Journal, 97(3): 290-296.
Hughes BP and Gregory R (1972), Concrete Subjected to High Rates of Loading in
Compression, Magazine of Concrete Research, 24(78): 25-36.
Jones PG and Richart FE (1936), The Effect of Testing Speed on Strength and Elastic
Properties of Concrete, ASTM Journal, 36: 380-392.
Lee J and Fenves GL (1998), A Plastic-damage Concrete Model for Earthquake Analysis of
Dams, Earthquake Engineering and Structural Dynamics, 27(9): 937-956.
Li J, Chen JY and Bai WF (2005), Application of Strain Rate-dependent Damage Model in
Analysis of Dynamic Opening of High Arch Dam Transverse Joints, Journal of
Hydraulic Engineering, 36(7): 870-875. (in Chinese)
Malvar LJ and Ross CA (1998), Review of Strain Rate Effects for Concrete in Tension, ACI
Materials Journal, 95(6): 735-739.
Moore OL (1934), Report of Working Committee of Committee C-1 on Plastic Mortar Tests
for Portland Cement, Proceedings of the American Society for Testing Materials, 34:
322.
Norris CH, Hansen RJ, Holley MJ, Biggs JM, Namyet S and Minami JK (1959), Structural
Design for Dynamic Loads, New York: McGraw-Hill Book Co. Inc.
Oh BH (1987), Behavior of Concrete under Dynamic Tensile Loads, ACI Materials Journal,
84(1): 8-13.
Raphael JM (1984), Tensile Strength of Concrete, ACI Journal, 87(17): 158-165.
Rossi P and Toutlemonde F (1996), Effect of Loading Rate on the Tensile Behaviour of
Concrete: Description of the Physical Mechanisms, Materials and Structures,
29(186): 116-118.
Rossi P, Van M and Jan GM (1994), Effect of Loading Rate on the Strength of Concrete
Subjected to Uniaxial Tension, Materials and Structures, 27(169): 260-264.
Rush H (1960), Research towards a General Flexural Theory for Structural Concrete, ACI
Journal, 57(1): 1-28.
Spark PR and Menzies JB (1973), The Effect of Rate Loading upon the Static and Fatigue
Strengths of Plain Concrete in Compression, Magazine of Concrete Research, 25(83):
83-80.

268
Spooner DC (1972), Stress-stress-time Relationships for Concrete, Magazine of Concrete
Research, 24(81): 197-208.
Tedesco JW, Ross CA, McGill PB and ONeil BP (1991), Numerical Analysis of High Strain-
rate Concrete Direct Tension Tests, Computers and Structures, 40(2): 313-327.
Wang W (1997), Stationary and Propagative Instabilities in Metals a Computational Point of
View, PhD dissertation, TU Delft, Netherlands.
Watstein D (1953), Effect of Straining Rate on the Compressive Strength and Elastic
Properties of Concrete, ACI Journal, 49: 729-744.
Winnicke A, Pearce CJ and Bicanic N (2001), Viscoplastic Hoffman Consistency Model for
Concrete, Computers and Structures, 79(1): 7-19.
Xiao SY, Li HN and Lin G (2008), Dynamic Behaviour and Constitutive Model of Concrete
at Different Strain Rates, Magazine of Concrete Research, 60(4): 271-278.
Xiao SY, Li HN and Lin G (2010), 3-D Consistency Dynamic Constitutive Model of
Concrete, Earthquake Engineering and Engineering Vibration, 9(2): 233- 246.
Zielinski AJ, Reinhardt HW and Krmeling HA (1981), Experiments on Concrete under
Uniaxial Impact Tensile Loading, Materials and Structures, 14(80): 103-112.
11
Seismic Potential Improvement of
Road Embankment
Ken-ichi Tokida
Osaka University
Japan
1. Introduction
Roads or highways are very important and fundamental infrastructures for the activities of
life and production, and their safety are required to be secured against the various disasters
such as rainfalls and earthquakes etc. The road networks are composed of many types of
attached structures such as bridges, embankments, slopes, tunnels etc. During earthquakes,
the abovementioned structures have been damaged in the past earthquakes and will be
possible to be damaged in the near future.
For reducing the damage potential of the road networks, the seismic design and/or counter-
measures for bridges have been conducted prior to other structures in the last about 40 years
in Japan because bridges are considered to be difficult to retrieve when they will be damaged
by earthquakes. On the other hand, road embankments have not been reinforced at all against
seismic disasters because it is believed that embankments can be retrieved easily.
However, severe damages of road embankments have been occurred and it has been
clarified that the performance of road networks can be lost severely for a long time, for
example, 2weeks by the damages of road embankments because it needs a long time to
retrieve numbers of damaged embankments and the ones located at the mountainous area.
As mentioned above, road embankments are also necessary to be reinforced against seismic
ground motion, in other words, the seismic potential of road embankments should be
improved against seismic ground motion during future earthquakes.
However the performance-based design will be very reasonable and effective for improving
the seismic design of road embankments in the future, the design concepts have not been
clarified yet and necessary to be established as soon as possible, similarly with bridges.
In this chapter, based on the typical damages of road embankments induced by recent
earthquakes in Japan, the macroscopic assessment method for seismic potential of road
embankments and the new concepts for the performance-based design of road
embankments from the engineering view points are proposed and their applications are
discussed through large-scaled field tests, centrifuge tests and analytical simulations. The
following concerns summarizes the results of the recent research and development initiated
by Osaka University in Japan from 2004 to the present through Tokida et al. (2007), Tokida
et al. (2008), Tatta et al. (2009), Tokida et al.(2009) and Tokida et al. (2010).
2. Damage of road embankment in past earthquakes in Japan
During the recent decade in Japan, severe damages of many road embankments occurred during
the earthquakes such as the 2004 Mid-Niigata Prefecture Earthquake, the 2007 Noto Peninsula
Earthquake, the 2009 Suruga Bay Earthquake and the 2011 Off the Pacific Coast of Tohoku
Earthquake. The typical characteristics of damages of road embankments and the several
important lessons obtained can be indicated as follows from the engineering view points.
2.1 The 2004 Mid-Niigata prefecture earthquake
The 2004 Mid-Niigata Prefecture Earthquake is an inland earthquake and occurred at the
mountainous area in October 23, 2004. The middle-class magnitude of this earthquake is 6.8
and the shallow depth of epicenter is 13 km.
Fig. 1 shows the time history just after the main shock occurrence for opening Kanetsu
Highway and Hokuriku Highway which were stopped mainly by the damages relating to
not the bridges but the embankments to the emergency vehicles such as ambulances etc. and
the public cars. It can be seen it took about 13 days to open the highways for the personal
vehicles under the speed regulation. As mentioned above, we can realize the importance of
seismic design and reinforcement for road embankments in the future.

Fig. 1. Recovery of traffic performance after stopping caused by earthquake
Item 2 on the status of drainage system can be selected because the damage risk relates to
the level of moisture in the embankment. For example, the assessment rank and each
relating score are assumed as follows.
The typical examples of severe or large-scaled damages and slight or small-scaled damages
of road embankments can be shown as Photo. 1 and Photo. 2, respectively. The former
indicates three examples, i.e. (a) very severely damaged site reported by Higashi Nippon
Highway Co. Ltd. where all 4 lanes were slid perfectly, (b) severely damaged site reported
by Hokuriku Regional Bureau of Ministry of Land and Transportation where all 2 lanes
were slid perfectly and (c) a little severely damaged site also reported by Hokuriku Regional
Bureau where almost 2 lanes were slid partially and each traffic performance was lost
perfectly. While, the latter indicates two examples, i.e. (a) partially slid site also reported by
Hokuriku Regional Bureau where the sliding occurred only at the shoulder of the
embankment and was retrieved easily with large-scaled sand bags as shown in Photo. 2, and
(b) cracked site reported by Higashi Nippon Highway Co. Ltd. where only the shoulder was
cracked a little and dump trucks can run without urgent retrieval, then their loss of traffic
Lapse Time after Stop
100h 34day
2
60
50
40
Regulation of 50 km/h
19h 13day
Only
Permitted
Vehicle
One Lane
in Each
Direction
Only
Emergency
Vehicle
B
u
m
p

a
t

C
r
e
s
t

c
m
Stopped
Personal Vehicle
Full four
Lanes
Regulation
of 80 km/h
0
0
Full Four
Lanes

Seismic Potential Improvement of Road Embankment 271
performance were small and the road could be used just after the earthquake by small-scale
and short-time restoration or not.

(a) 4 lanes slid perfectly (b) 2 lanes slid perfectly (c) Almost 2 lanes slid
Photo 1. Large-scaled damage of road embankment without traffic performance

(a) Partial sliding and urgent retrieval with sand bags (b) Slight crack at shoulder
Photo 2. Small-scaled damage of road embankment with a certain traffic performance
The above two types of damage levels of road embankments give us the important lessons
to consider the fundamental concepts for the performance-based design of road
embankments in the future which are shown in Chapter 3.
2.2 The 2007 Noto peninsula earthquake
The 2007 Noto Peninsula Earthquake is one type of inland earthquakes and occurred near
the coast of Noto Peninsula in March 25, 2007. The middle-class magnitude of this
earthquake is 6.9 and the shallow depth of epicenter is 11 km.
Photo.3 reported by Ishikawa Prefecture shows four examples of the slid road embankments
which locate at Noto Toll Road at the mountainous area of Noto Peninsula. These photos indicate
(a) severe damaged site where 2 lanes were slid perfectly, (b) severe damaged site where almost 2
lanes were slid, (c) severe damaged site where slight 2 lanes were slid and (d) severe damaged
site where one side of 2 lanes slid and each traffic performance was lost perfectly.
The severe sliding damages such as Photo. 3 which resulted in the perfect stopping of traffic
performance occurred at 13 sites located along the road of about 48.2 km. The site shown in
Photo. 3(c) is selected to apply the analytical method named Newmark Method to simulate
the sliding plane and its displacement in Chapter 5.
In Photo. 4 reported by Ishikawa Prefecture, Photo. 4(a) shows the severe damaged on ramp
close to the toll gate. Most damaged embankments were retrieved urgently and
permanently with use of the geotextile which can be shown in Photos. 4(b) and 4(c),
respectively. Noto Toll Road was opened to the public urgently in about one month later

after the earthquake and the permanent retrieval of all damaged embankments was
completed in about five months later after the urgent retrieval.

(a) 2 lanes slid perfectly (b) Almost 2 lanes slid perfectly

(c) A little more 1 lane slid (d) 1 lane slid/1 lane remain
Photo 3. Large-scaled damage of road embankment without traffic performance

(a) Sliding of entrance lane (b) Urgent retrieval (c) Permanent retrieval
Photo 4. Damage and seismic retrieval with geotextile
It can be indicated that the seismic design with the seismic coefficient of 0.2 was applied to
reinforce the retrieved embankments strongly more than that before the earthquake for the
future earthquakes. Such seismic retrieval for the road embankments at Noto Toll Road may
be the first case in Japan and can be referred for other embankments in the future.
2.3 The 2009 Suruga bay earthquake
The 2009 Suruga Bay Earthquake is a plate-slid earthquake and occurred at Suruga bay
neighboring to Pacific Ocean in August 11, 2009. The middle-class magnitude of this
earthquake is 6.6 and the depth of epicenter is 20 km.
In this earthquake, only one embankment slid at Tohmei Highway which is the most
important trunk road between Tokyo and Nagoya as shown in Photo. 5 reported by Naka
Nippon Highway Co. Ltd. However the sliding occurred partially at the shoulder and slope
as shown in Photo. 5(a) and (b), the traffic performance was lost for about 5 days and the

economical influences and so on were very large for Japan. However the urgent and
extended retrieval was executed, it took 5 days to recover the limited traffic performance for
the personal vehicles as shown in Photo. 5(c). The permanent retrieval ended on June 10,
2010, i.e. in about ten months and the traffic performance was recovered perfectly.

(a) Overview of sliding (b) Shoulder and slope slid (c) Urgent retrieval
Photo 5. Partially damaged site without traffic performance and urgent retrieval
We can get the important lesson from this experience that the only one damaged site warns
the more severe damages of embankments by the very large earthquakes named Tokai
Earthquake and/or Tokai & Nankai Earthquake whose occurrence has been indicated
around this site in the near future in Japan.
2.4 The 2011 off the pacific coast of Tohoku earthquake
The 2011 Off the Pacific Coast of Tohoku Earthquake is a typical plate-slid earthquake and
occurred about 130 km off the main island of Japan in March 11, 2011. The very large-class
magnitude of the earthquake is 9.0 and the depth of epicenter is 24 km. The very wide area
where post earthquakes occurred and/or are occurring at present, i.e. the scale of moved
plate by the main-shock and after-shocks is assumed to be 200 km multiplied by 500 km.
The outline on the charactoristics of the earthquake and the observed ground motion can be
referred the prompt reports by the Japan Meteorological Agency (JMA) and the National
Research Institute for Earth Science and Disaster Prevention (NIED), respectively.
However the damages by tsunami occurred widely and severely, other damages of civil
engineering structures such as road bridges, road embankments, river dykes and so on were
not so remarkable because the epicenter of the main-shock is rather far and the damages
occurred along the coasts may be washed away by tsunami attacks. The most severely
damaged embankment of all ones at Tohoku Highway and Joban Highway which are the
main trunk routes to connect Tokyo with Tohoku Region, where the earthquake occurred
can be introduced as shown in Photo. 6 reported by Kanto Branch Office of Higashi Nippon
Highway Co. Ltd. As shown in Photo. 6(a), the one side of 2 lanes in one direction slid.
However the thickness of the soft ground beneath the embankment body is about 3.6m, the
type of the failure is a toe sliding failure and a little small failure. The work flow of the
urgent retrieval for this embankment can be shown as Photo. 6(b), 6(c) and 6(d),
respectively. As shown in Photo. 7 under urgent retrieval work, the height and the damaged
length of the embankment are about 8m and 150m, respectively. The traffic performance for
the personal vehicles was recovered urgently on March 21, 2011. Then it takes about 10 days
to complete the urgent retrieval and open the highway to the public.

(a) Crushed (b) Sliding occurrence
(b) Urgent retrieval
with gravel and soil
(c) Pavement
Photo 6. One side of two lanes damaged and urgent retrieval

Photo 7. Under urgent retrieving
Some lessons concerning to road embankments against tsunami attacks can be obtained as
follows. Photo. 8 is a newly constructed river dyke along the channel which locates in
Miyagi Prefecture facing the Pacific Ocean where the earthquake occurred. As shown in the
Photo. 8, the river dyke which height and width at the crest are about 4m and 6m,
respectively is paved and can be used as a road. Then the lessons obtained from the
damages of this dyke can be applied for the future design against tsunami disasters. Photo.
8(a) shows the condition of the front side of dyke slope against tsunami attacks which came
from the left side of the Photo. 8(a). It can be seen that almost all of the front sides were not
damaged, i.e. the front side may be strong against tsunami attacks which height is assumed
to be about 10m around this site and then the overflown water depth at the dyke crest can
be estimated to be about 4m. Photo. 8(b) shows the condition at the dyke crest where the
asphalt pavements are stripped partially, but the road function was not lost so much. Then it
can be indicated that the asphalt pavement at the crest is very effective against the
overflowed water by tsunami. Photo. 8(c) shows the conditions of the back side of dyke
slope washed away by tsunami. It can be known that the back side slope is eroded near the
surface and the toe ground is eroded severely and trenched deeply. However the back side
slope is fallen down partially, the minimum stability of the dyke can be maintained to keep
the performance of the dyke and/or road. However the above lessons may be limited to this
dyke, they become very important hints for the design against tsunami attacks.

(a) Front slope non-damaged (b) Crest stripped partially (c) Back slope and toe eroded
Photo 8. Dyke overflown by tsunami


(a) Opposite side from sea (b) Sea side
Photo 9. Comparison of damages at both sides of highway embankment
Photo. 9 shows another example from which we can get some lessons on the performance of
the road embankments against tsunami attacks. Photos. 9(a) and 9(b) show the landscapes
after tsunami attacks at the opposite side and the front side of the road embankment against
tsunami attacks, respectively. However the road embankment was expected to perform the
barrier function against tsunami attacks, the performance may be small as shown in Photo.
9(a) because the box culvert and the elevated bridge permit tsunami flow to go through.
From this example, it can be known that the road embankment may be effective as a barrier
against tsunami when the open space such as box culvert etc. can be constructed carefully
considering tsunami attacks.
3. Macroscopic assessment for seismic potential of road embankments
It is necessary to select the objective road embankments with high seismic potential for the
further detail investigations and the economical countermeasures against future
earthquakes. Based on the damages in the past earthquakes as shown in Chapter 2, the
assessment items and ranks for elemental functions traversing damage of embankments
caused by the sliding failure can be obtained and a macroscopic assessment method to
estimate the seismic potential of the road embankments can be proposed as follows.
3.1 Factors related to damage potential
Macroscopic assessment methods of seismic and disaster prevention functions are discussed
and proposed by Tokida et al. (2007) based on the on-site investigations as for the damages
of expressways and national highways in the 2004 Mid Niigata Prefecture Earthquake and
the 2007 Noto Peninsula Earthquake.
Typical 8 items such as Item 1 to Item 8 to assess the structural vulnerability of each road
embankment are selected and the estimation weight between each item and their detail
ranks and their added scores to estimate the risk potential quantitatively can be
recommended as follows.
Item 1 on the topographical conditions below embankments can be selected because the
damage risk relates to the original terrain condition under the embankment such as ridge
slope or stream. For example, the assessment rank and each relating score are assumed as
shown in Fig. 2.
Rank-1 with score = 0 in case of embankment located on large ridge slope
Rank-2 with score = 1 in case of embankment located on small ridge slope
Rank-3 with score = 2 in case of embankment located on slope in parallel
Rank-4 with score = 4 in case of embankment located on small stream
Rank-5 with score = 5 in case of embankment located on large stream
The relative weight for Item 1 between 8 items is assumed to be 2, then the maximum value
of the score S1 for Item 1 is 10 multiplying 2 by 5.

Fig. 2. Identification of topographical ranks for Item 1
Rank-1 with score = 0 in case of embankment where the whole methods of A to E are
applied.
Rank-2 with score = 1 in case of embankment where both A and B methods are applied.
Rank-3 with score = 2 in case of embankment where the C method is applied.
Rank-4 with score = 4 in case of embankment where both D and E methods are applied.
Rank-5 with score = 5 in case of embankment where no drain method is applied.
The above drain methods of A to E are identified as follows.
A is a method to drain the underground water from the mountain side with pipes or
ditches laid on the base ground of embankments.
B is a method to set the horizontal or vertical drain materials laid inside of embankments.
C is a method to drain the surface water from the mountain side with pipes and ditches
laid inside of the embankments.
D is a method to drain the crest surface water with side ditches.
E is a method to drain the slope surface water by side ditch.
Item 3 on the soil properties and quality of construction can be selected because the damage
risk relates to the embankment soil properties or the quality of workmanship used during
construction, such as compaction.
Rank-1 with score = 0 in case of embankment where the embankment material is
enough well, and good compaction or improvement is done.
Rank-2 with score = 3 in case of embankment where the embankment material is
normal and construction is normal.
Rank-3 with score = 5 in case of embankment where the embankment material is not
good and the construction condition is not good.
Rank-1 Rank-5 Rank-2 Rank-5 Rank-3 Rank-4
Embankment
Contour Line of
Base Ground

Item 4 on the traverse structure of embankment can be selected because the damage risk
relates to the traverse structure of the embankment, including the incline of the foundation
ground, the configuration of the embankment, as shown in Fig. 3.
Rank-1 with score = 0 in case of embankment on flat base ground.
Rank-2 with score = 1 in case of embankment on ups and downs base or ridge slope or
one side cut and other side fill at the inclination base ground.
Rank-3 with score = 2 in case of embankment on one side fill at the inclination base
ground where the embankment edge is flat.
Rank-4 with score = 3 in case of embankment on one side fill at the inclination base
ground where the embankment edge slopes.

Fig. 3. Identification of traverse structures for Item 4
Rank-5 with score = 4 in case of embankment on both side fills at the inclination base
ground where the embankment edge is flat.
Rank-6 with score = 5 in case of embankment on both side fills at the inclination base
ground where the embankment edge slopes.
of rthe score S4 for Item 4 is 10 multiplying 2 by 5.
Item 5 on the height of embankment can be selected because the occurrence and scale of
damage risk relate to the embankment height as follows.
Rank-1 with score = 0 in case of embankment which height is 4 m or less.
Rank-2 with score = 2 in case of embankment which height is 8 m or less exceeding 4 m.
Rank-3 with score = 4 in case of embankment which height is 20 m or less exceeding 8 m.
Rank-4 with score = 5 in case of embankment which height is 20 m excess.
Item 6 on the inclination of embankment slope can be selected because the stability of
embankment relates to the slope inclination as follows.
Rank-1
Rank-2
Rank-2
Rank-3
Rank-4
Rank-5
Rank-6
Rank-1 with score = 0 in case of embankment which slope inclination is below the
standard one.
Rank-2 with score = 5 in case of embankment which slope inclination exceeds the
standard one.
Item 7 on the conditions reinforcing the embankment can be selected because the seismic
and/or a certain improvements are effective to reduce damage of embankments as follows.
Rank-1 with score = 0 in case of embankment reinforced based on the seismic design fully.
Rank-2 with score = 2 in case of embankment where the retaining wall is applied.
Rank-3 with score = 4 in case of embankment where the attached structures such as
gabions are set.
Rank-4 with score = 5 in case of embankment where the countermeasures are not done.
Item 8 on the transformation history and/or present state of embankments can be selected
because the past and/or present transformation indicates the sign and warning for the
future damage occurrence. Four sub-items are selected to judge the risk of embankments
considering the past and/or present conditions as follows.
Sub-Item-1 on the existence level of water spring on the slope and/or toe of embankment can
be selected because the water conditions inside of embankments are most important. Each
score of 0 to 5 are set considering the level of water spring and the relative weight is assumed
to be 2, then the maximum value of the score S8-1 for Sub-Item 1 is 10 multiplying 2 by 5.
Sub-Item-2 on the transformation level of embankment crest such as cracks can be selected
because the transformation is one of signs to be failed. Each score of 0 to 5 are set
considering the level of transformation and the relative weight is assumed to be 0.6, then the
maximum value of the score S8-2 for Sub-Item 2 is 3 multiplying 0.6 by 5.
Sub-Item-3 on the transformation level of embankment slope such as cracks can be selected
because the transformation is one of signs to be failed. Each score of 0 to 5 are set
considering the level of transformation and the relative weight is assumed to be 0.4, then the
maximum value of the score S8-3 for Sub-Item 3 is 2 multiplying 0.4 by 5.
Sub-Item-4 on the present performance of drainage system can be selected because the
conditions of drainage system relate to the water conditions inside of embankments. Each score
of 0 to 5 are set considering the level of transformation and the relative weight is assumed to be
2, then the maximum value of the score S8-4 for Sub-Item 4 is 10 multiplying 2 by 5.
The maximum value of the score S8 for Item 8 added each score of the above four sub-items
gives 25. As a summary, the total score named SRP added from Item 1 to Item 8 which
means the score of risk potential can be obtained in the Eq. (1) and scores and weight for
each item and rank are summarized in Table 1.

SRP=S1+S2+S3+S4+S5+S6+S7+S8
(1)


Table 1. Relation between item, rank, weight and scores
The value of SRP gives 0 to 100 and is named the evaluation value of seismic risk potential
of an embankment at a certain site, i.e. means that when the value is larger, the damage
potential can be indicated to become larger. Basing on the proposed assessment method
with use of SRP, the seismic risk potential of a road embankment can be estimated easily,
macroscopically and practically.
3.2 Application of macroscopic assessment method
The example of application of the macroscopic assessment method with the value of SRP
can be obtained as shown in Fig. 4 where 13 damaged and 19 non-damaged embankments
in the 2007 Noto Peninsula Earthquake are selected and compared each other.
The average score and standard deviation are 60.7 and 8.2, respectively on the 19 non-
damaged embankments. Then the score range of is from 52.5 to 68.9. The average score
and standard deviation are 75.2 and 4.8, respectively on the 13 damaged embankments and
then the score range of is from 70.4 to 80.0.

Fig. 4. Scores of seismic risk potential between non-damaged and damaged embankments in
the 2007 Noto Peninsula Earthquake
Based on the results as shown in Fig. 4, the following lessons can be drawn to use the
evaluation value of seismic risk potential practically.
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20 25 30
S
c
o
r
e
s

o
f

S
i
e
i
s
m
i
c

r
R
i
s
k

P
o
t
e
n
t
i
a
l
Position of Embankment (km)
Non-Damaged Embankment
Damaged Embankment
0 SRP < 60 : Seismic Potential of road embankment is very low or low.
60 SRP < 70 : Seismic Potential of road embankment is a little high or high.
70 SRP 100 : Seismic potential of road embankment is high or very high.
4. Performance-based design concept for road embankment
Recently the new design concept, i.e., performance-based design is necessary for reasonable
and economical design of road embankments against severe earthquakes. In this Chapter,
four design concepts considering the seismic performance of road embankments are
proposed, and the criterion to judge the damage level of settlements or sliding
displacements is introduced.
4.1 Concepts based on performance-based design
Comparing the severe damage of road embankment shown in Photo. 1(b) with the slight
damage shown in Photo. 2(a), it can be hinted to control the position of the sliding plane
occurred not to reach to the traffic lane, then to keep the traffic performance at a certain level
just after an earthquake. This fundamental design concept was proposed by Tokida et al.
(2007) and named Sliding Failure Control Method. The control of the sliding failure aims
to change the positon of sliding plane and/or to decrease the sliding displacement occurred
during earthquakes not to disturb the traffic performance perfectly. Furthermore, as shown
in Fig. 5, four specific design concepts such as Crest Reinforcement Structure, Barrier
Reinforcement Structure, Artificial Reinforcement Structure and Toe Reinforcement
Structure can be proposed to achieve the fundamental design concept. The performance of
each reinforcement structure is as follows.
Crest Reinforcement Structure aims to reinforce the crest of embankment partially with use of
geotexstile etc. and not to occur the sliding plane and decrease the sliding displacement on the
top of the embankment, then maintain the traffic performance enough just after an earthquake.
Toe Reinforcement Structure aims to reinforce the toe of embankment to control the sliding
plane and decrease the sliding displacement. Artificial Reinforcement Structure aims to control
the sliding plane by improving the embankment body stiffly as shown in Fig. 5 to control the
sliding plane not to reach the traffic lane. Barrier Reinforcement Structure aims to control the
position of the sliding plane with use of sheet piles etc. and not to occur at the traffic lane, then
maintain the traffic performance enough just after an earthquake.

Fig. 5. Typical concepts proposed for performance-based design of road embankment
As mentioned above, these four reinforcing structures are constructed partially and then
economically.

4.2 Performance level for seismic design of road embankment
For the performance-based design of road embankments, the performance level as the
purpose of the seismic design is necessary to be set. Because the performance level for road
embankments has not been clarified in Japan, Tokida et al. (2007) conducted the driving
tests to investigate the relation between the step height and the driving velocity through the
steps to know the limit height of steps occurred at the pavement surface during earthquakes
as shown in Fig. 6(a).
The test results are summarized in Fig. 6(b) which shows the limit height of steps able to drive
through without repair, relating to some conditions of the small or large vehicle size, the
emergency or personal vehicle and the velocity of vehicles driving through the steps such as
almost stop (0 to 10 km/h), slow down (15 to 20km/h) and normal (for example, 60 to 80km/h).
Based on the driving test results such as in Fig.6(b), the performance level considering traffic
performance just after an earthquake can be proposed as shown Table 2. The performance
levels are classified into four types as follows.

(a) Driving test (b) Relations between step and driving velocity
Fig. 6. Driving tests and results

Table 2. Performance level considering traffic performance just after earthquake
15 20km
Emergency
Vehcle
Personal
Vefcle
Small SisedVehcle Large SisedVehcle
0 10km 0
10km 1520km
Normal
S
t
e
p

o
r

S
e
t
t
l
e
m
e
n
t
(
c
m
Normal Slow Down Slow Down

Almost
Stop
Almost
Stop
Personal
Vefcle
Emergency
Vehcle
Level 1: The traffic performance can be maintained normally just after an earthquake,
because the steps occurred are smaller than 2 to 3 cm.
Level 2: The traffic performance is decreased a little just an earthquake, but can be recovered
rapidly and easily, because the steps occurred are smaller than 20 to 25 cm where the
vehicles can be go through the steps as shown in Fig. 6 (a).
Level 3: The traffic performance is decreased just an earthquake and difficult a little to
recover soon, because the steps occurred are larger than 20 to 25 cm and smaller than 50cm.
Level 4: The traffic performance is decreased just an earthquake and it takes long time to
recover, because the steps occurred are larger than 50 cm.
As shown above, Table 2 can be applied to set the objective performance level of the sliding
failure of road embankments in the design of both new embankments and existing ones in
the future.
5. Analytical estimation of sliding failure of road embankment
It is necessary to estimate the damage level of road embankments for the performance-based
design. In this chapter, the typical analytical method named Newmark Method (Tateyama et
al., 1998) modified the Original Newmark Method (Newmark, 1965) for calculating the
sliding displacement of road embankments during earthquakes is shown and its application
is discussed based on the damaged examples of road embankments in the past earthquakes
shown in Chapter 2.
Fig. 7 is the example appling the Newmark Method to the embankment damaged during the
Noto Peninsula Earthquake in 2007. The ground motion inputted at the bottom of the
embankment model shown in Fig. 7(b) is revised the maximum acceleration from the
ground acceleration of 580gal in the East-West direction shown in Fig. 7(a) which was
observed at Togi City set by National Research Institute for Earth Science and Disaster
Prevention (NIED) to the one of 500gal. As known in Fig. 7(b), the position of the sliding
plane measured after the earthquake is almost similar to the analytical one. Furthermore, the
measured sliding displacement at the crest of the embankment is 10.9m and similar to the
9.7m measured.
In this case study, the sliding failure such as the position of the sliding plane and the sliding
displacement estimated with use of the Newmark Method is well estimated analytically.

(a) Observed E-W ground motion at Togi (b) Measured and analytical sliding failure
Fig. 7. Example of application of Newmark Method
-600
-400
-200
0
200
400
600
0 5 10 15
Time(sec)
a
c
c
e
l
e
r
a
t
i
o
n
(
g
a
l
)
time (sec)
a
c
c
e
l
e
r
a
t
i
o
n

(
g
a
l
)

6. Performance-based countermeasures for road embankment
Several seismic countermeasures are proposed concretely and their applications are
discussed. In the case of the Crest Reinforcement Structure, the crest of road embankments
is reinforced with use of geosynthetics and its application is verified based on the static
large-scaled model test, the dynamic centrifuge test and the static and dynamic analyses
quantitatively and/or qualitatively. In case of the Toe Reinforcement Structure, the toe of
road embankments is reinforced by large-scaled gabions and its application is verified based
on analyses qualitatively. In case of the Artificial Base-ground Structure, the body of road
embankments is improved partially to construct the artificial base ground where the sliding
plane is assumed not to be occurred and its application is verified based on analyses
qualitatively. In case of the Barrier Reinforcement Structure, the barrier structures such as
sheet piles and underground walls may be constructed practically and its application is
verified based on the centrifuge test quantitatively.
6.1 Crest reinforcement structure with geosynthetics
Fig. 8 shows the example of the Crest Reinforcement Structure with use of the geosynthetics.
This structure can be applied to both existing and newly constructed embankments. In the
case of the existing embankment, the sheets of geosynthetics are installed at the overall crest
of the embankment as shown in Fig. 8 based on the excavation and the re-pavement, or
around the shoulder of the embankment partially. The reinforcing work is necessary to be
paid attention to reduce the effects on the traffic performance as little as possible. On the
other hand, in the case of newly construction, the sheets of geosynthetics can be easily set
inside of the embankment. The practical applications of the crest reinforcement structure
with use of geosynthetics have been discussed and clarified as follows.

(a) Reinforced with geosynthetics (b) Geosynthetics
Fig. 8. Example of crest reinforcement structure
6.1.1 Static large-scaled model test
The large-scaled embankment model is constructed in extension of 3.6m, depth of 1.8m and
height of 4.1m, which is surrounded by both wooden panels and sandbags shown Fig. 9(a).
As shown in Fig. 9(b), the model is divided into 4 reinforcement patterns. In each pattern,
the overburden surface layer with the thickness of 50cm are set at the crest of the
embankment model. Pattern 1 is the fundamental model without reinforcement. In cases of
Pattern 2 and Pattern 3, 5 layers with the thickness of 30cm in length of 1m and 2m,
respectively are reinforced with geosynthetics connected by metal fittings each other. In case
of Pattern 4, the embankment body is lapped with the geosynthetics which is tied with

Region rei nforced by Geosynthetics
Body of Embankment
ropes. The soil properties are shown in Table 3 and the fine aggregate without fine-grained
fraction of the wet density of
t
=18kN/m
3
which tends to occur a collapse easily. The static
failure of the reinforced embankment model is planned to be occurred by removing the
panels and sandbags.

(a) View of embankment
model
(b) 4 patterns of reinforced embankment models (unit:nm)
Fig. 9. Large-scalled embankment model in field test
The collapse mode of each reinforced embankment model is illustrated in Fig. 10(a). In case
of Pattern 1, the large sliding failure occurred at the whole area of the embankment body
beneath the surface layers. In case of Patterns 2 and 3, each deformation of the embankment
body is smaller than Pattern 1. Moreover, the deformation of the embankment model in case
of Pattern 3 with a longer reinforcement than that of Pattern 2 became smaller than that of
Pattern 2. The difference of each damage level can be known in Fig. 10(b) which shows the
side plane conditions and the surface condition of the crest of the embankment models. In
case of Pattern 4, the whole reinforced part slid down severely because of the weight of the
reinforced body.

Table 3. Properties of test material


(a) Deformation of models (unit: mm) (b) Difference of Settlement Level

(c) Comparison of settlement at shoulder
Fig. 10. Comparison of damage level between different reinforcement patterns
The test results on the settlements at the shoulders of each embankment model can be
summarized in Fig. 10(c) where the settlements of Pattern 1, Pattern 2, Pattern 3 and Pattern
4 are 1.7 m, 1.3 m, 0.3m and 1.9m, respectively. Therefore, it can be indicated that Pattern 3
where the horizontal length of the geosynthetics is set deeper than the estimated sliding
plane is most effective to keep the shape of the embankment during an earthquake.
6.1.2 Dynamic centrifuge model test
Applying the steel container which is 900mm wide, 300mm thick and 280mm high as
shown in Fig. 11, several centrifuge tests are carried out under the acceleration of 30g. A
half-sized embankment on the solid base ground is made using the DL clay and silicon oil
to adjust the water content to 5% in height of 290 mm which corresponds to the actual
height of 8.7m. The density of total fill is 1.52g/cm and the shear wave velocity of
embankment model measured is 158m/sec in 1g and 185m/sec in 30g. As shown in Fig.
11, the horizontal and vertical accelerometers are installed in three sites to measure the
input acceleration on the bottom of soil container and response acceleration of
embankment model near the toe of slope in height of 90mm and under the edge of slope
in height of 230mm from the bottom. To measure the final residual deformation of model,
several colored sand columns are installed into the central cross section of model from the
surface in the interval of 60mm and several markers put at the surface of the embankment
in the interval of 50mm. Furthermore, two thin plates with the thickness of 0.3mm on

Patt ern 1
Patt ern 4
Pat tern 3
Pattern 2
3600 3600
5
0
0

5
0
0

1
7
0
0

1
3
0
0

3
3
0
0

4
1
0
0

1
9
0
0

4
1
0
0
3
0
0

1
8
0
0

1
8
0
0

1
8
0
0

1
8
0
0

1
8
0
0

1
8
0
0

3600 3600
700 800
1400 1200
800
1000
2000
3
0
0

5
0
0

5
0
0

S
e
l
t
t
l
e
m
e
n
t
(
m
)
P
a
t
t
e
r
n

1

P
a
t
t
e
r
n

2

P
a
t
t
e
r
n

3

P
a
t
t
e
r
n

4

2.0
1.0
0.0
which the strain gauges are pasted on both sides are placed in the upper and central parts
of embankment to investigate the failure mechanism.
Fig. 12(a) shows the cross section cut at the center of the embankment model which is not
reinforced and slid largely by the inputted motion with the maximum acceleration of 488gal.
Fig. 12(b) compares the initial and slid shapes of the embankment model drawn by markers
located at the surface and the colored sands columns. Comparing the positions of markers
before and after dynamic motions in Fig. 12(b), it can be seen that however the embankment
body around the shoulder is disturbed according to the sliding, the sliding body is driven
over all because the displacement of markers at the shoulder of 89mm and the one along the
slope of 87mm are almost same. On the other hand, the vertical settlement at the shoulder is
65mm which is converted to 1.95m at the field. As mentioned above, it can be clarified that
its possible to reproduce the dynamic sliding plane artificially by the centrifuge test.

(a) Cross sectrion (unit: mm) (b) Plane section (unit:mm)
Fig. 11. Embankment model and measurement for dynamic centrifuge test

Estimated
slip surface by the shape
of colored sands

Before Motion
After Motion
unit : mm

(a) Sliding plane occurred (b) Before and after dynamic motion
Fig. 12. Sliding failure in case of non-reinforced model
As shown in Fig. 13, four types of embankment models which are aimed the Crest
Reinforcement Structure reinforced with changing the numbers of reinforced layers and the
installed length of geosynthetics. The embankment modei in Case 1 is reinforced at the
whole surface of crest with two layers by three long geosynthetics sheets which are fixed
with the side wall. The one in Case 2 is reinforced with one layer by two long geosynthetics
sheets and two layers with two short one which are set more deeply than the estimated
sliding plane occurred in the case of nonreinforced embankment model shown in Fig. 12.

sand sheet
plate

plate beads t o measure
the displ acement
of surface of
ground

bender element
accel erometer
st rain gauge
laser displacement
sensor
col or sand

These two cases are supposed for the newly constructed embankment or the existing one to
reinforce wholly. The ones in Case 3 and Case 4 are reinforced with three layers by four
short geosynthetics sheets and two layers with three short ones, respectively. These two
cases are supposed to apply to the existing embankment not to disturb the traffic activity
under construction works.
In the above four cases, the thickness of both long and short reinforced layer is set to be
30mm which scale is 0.9m in the field. To ensure the total performance of reinforced zone,
geosynthetics sheets are connected with each other at the surface of slope.

(a) Case1 (unit: mm)

(b) Case2 (unit: mm)
geosynthetics
geosynthetics
metal
fittings
metal
fittings
fixing
fixing
Estimated slip surface
without reinforcement

(

(c) Case3 (unit:mm)

(d) Case4 (unit:mm)
geosynthetics
geosynthetics
fixing
fixing

Fig. 13. Embankment Model Reinforced with Geosynthetics
Fig. 14 compares the slid conditions between four cases measured at the central cross section
after inputting the sinusoidal motion with the maximum acceleration of about 500gal. It can
be indicated qualitatively as follows based on the results of Fig. 14.
Case 1: The deformation on the crest of embankment is small and the sliding occurs at the
area lower than the reinforced zone including the lower side of reinforced zone. The volume
of sliding area is larger than that of Case 2.
Case 2: The deformation on the crest of embankment is small extremely and the sliding
occurs at only the area lower than the reinforced zone which is not affected by the sliding.
The effectiveness of the reinforcement can be known clearly in Photo. 10
Case 3: The deformation on the crest of embankment is very small and the sliding occurs at
only the area lower than the reinforced zone which is not affected by the sliding.
Case 4: The deformation on the crest of embankment is occurred a little and the sliding
occurs not only at the area lower than the reinforced zone but also at the boundary of
reinforced zone which is reached to the surface of the crest.

(a) Case1 (unit:mm)

(b) Case2 (unit:mm)
geosynthetics
metal
fittings
geosynthetics
metal
fittings

(c) Case3 (unit:mm)

(d) Case4 (unit:mm)
geosynthetics
geosynthetics

Fig. 14. Sliding failure in case of reinforced model with geosynthetics

(a) Sliding at cross section (b) Condition at crest
Photo 10. Overview of deformation of reinforced embankment model in Case 2
The residual displacements measured at the shoulder and upper and lower slope of the four
reinforced models comparing with the non-reinforced one in Fig. 12 are summarized in Fig.
15. Compared the reinforced models with the one without reinforcement, both horizontal
and vertical residual displacements at the shoulder of the embankment model are decreased
greatly in the whole reinforced models as shown in Fig. 15(a). Furthermore, the residual
displacements along the upper-side slope are also decreased greatly as shown in Fig. 15(b).
However the residual displacements along the lower-side slope are not so smaller than that
without reinforcement, it is not an important problem because the slid zone below the
reinforced zone doesnt affect directly on the traffic performance will be repaired easily later.
As mentioned above, it can be indicated that the Crest Reinforcement Method with
geosynthetics sheets is very effective for both newly constructed embankment and existing
ones to get minimum traffic performance just after earthquakes. Some lessons can be
obtained that the number of reinforced layers is very important for both long sheets and
short sheets and the short sheets should be set deeper than the estimated sliding plane and
almost fixed at the inside end.


(a) At shoulder (b) At lower or upper slope
Fig. 15. Comparison of Residual Displacement
6.1.3 Analytical estimation
The example of analytical application of the Crest Reinforcement Structure with
geosynthetics should be referred to Chapter 6.3.
6.1.4 Practical application in field
The Crest Reinforcement Structure has been applied at a certain site where the road
embankment approaching to the other construction site is constructed temporally as shown
in Fig. 16(a). The maximum height of the embankment is 20m and the slope ratio of that is
1:1.8. The soil materials used for the embankment body are classified into the silty clay with
high natural water content of 50% and the average cohesion is 12.8kN/m
2
.as shown in Fig.
16(b). Therefore, because the embankment body is necessary to be improved to achieve the
sliding safety factor of 1.2 in general conditions, the embankment body is improved to the
shear strength of 74kN/m
2
with cements. Furthermore, the Crest Reinforcement Structure
with geosynthetics as shown in Fig. 16(a) is designed additionally.
As shown in Fig. 16(a), one reinforced layer with two long sheets of 9m and two reinforced
layers with two short sheets of 2m which are similar to the Case 2 in the centrifuge test in
Fig. 13. The thickness of each layer is 0.5m. However the seismic design hasnt been applied
in this case, its very important that this case is the first trial to apply the Crest
Reinforcement Structure in Japan as shown in Fig. 16(b) and can be obtained important
lessons to construct the structures in the field.

(a) Design condition for cross section (b) Completed embankment
Fig. 16. Application of crest reinforcement structure to actual road
6.2 Toe reinforcement structure
In Japan, the groundwater in road embankments causes severe damage of road
embankments due to earthquakes, and a lot of embankments to be damaged have been
0
20
40
60
80
100
1 2 3 4
(
m
m
)
r
e
s
i
d
u
a
l

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)

without
reinforcement
Case1 Case2 Case3 Case4
horizontal
vertical
0
20
40
60
80
100
1 2 3 4
(
m
m
)
r
e
s
i
d
u
a
l

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)

without
reinforcement
Case1 Case2 Case3 Case4
upper
lower
Wide Reinforced Area 9 m0.5 m : 1 layer
Narrow Reinforced Area
2 m0.5 m 2 layers
reported in recent years. A large-scale gabion is given as a typical structure of the Toe
Reinforcement Structure in this Chapter. As for the large-scale gabion, the merits to apply it
are low-cost, good construction workability, and there are a lot of actual results in the field
as the disaster restoration method for the purpose of drainage as shown in Photo.11 (a).

(a) Disaster restoration (b) Moved gabions and occurred sliding
Photo 11. Application of large-scaled gabion at present
However the gabions were set only for drainage in advance, they could not work against the
seismic motion as shown in Fig. 11(b). Then it is necessary and effective to apply the large-
scale gabions for not only drainage but also reinforcement of the embankment.
For the reinforcing design of gabions, friction factors are necessary to be estimated. As for
the friction factors between the base ground and the superstructure, Table 4 summarizes the
static friction factors applied to the erosion cntrol dam, the retaining wall and the river wall
in Japan based on Erosion Control & Land Slide Technical Center (2009), Japan Road
Association (1999) and Japan River Association (1997), respectively. However Table 4 can be
applied for the road embankment design, the drag force tests are tried as follows to
investigate both static and dynamic friction factors for road embankments in detail.

Table 4. Friction coefficients applied to several structures in Japan
6.2.1 Friction coefficient of large-scaled gabion
The field tests using full-scaled gabions were carried out and the effects on friction factors
by the mesh-structure, the ground surface condition and the weight and/or load of gabions

are discussed. Photo. 12 shows the three types of the conditions at the ground surface where
the gabions are placed: a mat-placed ground, a sandy ground and a crushed-stone-layered
ground, respectively. The sheet of a mat is permeable and possible to be set at the bottom of
gabions to prevent fine soil particles to be flown out from the beneath the ground.

(a) Diamond-shaped(b) Grid-shaped (c) Ground surface condition to place gabions
Photo 12. Structure of gabion and ground surface condition

Table 5. Static and dynamic friction coefficient

Photo 13. Typical method to drag gabion and measure drag-Load
Furthermore, three types of the weight and/or load condition are considered. The weight at
Stage 1: 7.487kN are the same with that of the practical gabion and each weight at Stage 2:
15.249kN and at Stage 3: 21.619kN is about twice and three times the weight of the practical
one, respectively. Photo.13 shows the typical method to drag a gabion and to measure the
drag-load where the steel wire connected to the gabion is pulled by the winch which is set
on the right side in Photo. 13.
Both the static friction coefficients and dynamic ones can be summarized in Table 5 from
which the findings obtained are summarized as follows.
1. The static friction coefficients used in a present design differ depending on the structure
of the large-scale gabion or the foundation ground condition. The static friction
coefficients for all conditions range 0.510.79 and the ones of the diamond-shaped

Gabi on
L oad Cell
Drag
Gabi on
L oad Cell
Drag
Mat-placed Crashed stone
Sandy soil
wire-mesh and the grid-shaped one at the crushed stone and sandy soil ground range
0.610.74 and 0.770.79, respectively. Furthermore, it can be found that the ones at
mat-placed ground are generally smaller than that at crushed-stone and sandy soil one.
This indicates that its necessary to pay attention to use a sheet of mat at the bottom of
gabions.
2. The static friction coefficients obtained in these tests are same with or more than that in
the cases of gravel layers and sand layers shown in Table 4.
3. On the other hand, the dynamic ones range 0.480.74. The relations of dynamic friction
coefficients between mat-placed ground and crushed-stone ground or sandy soil
ground are similar with that of static friction coefficients as mentioned above.
6.2.2 Seismic potential improvement with gabion
The centrifuge test is conducted on the effects by the Toe Reinforcement Structure with a
rigid body shown in Fig. 17(a). The test conditions and measurement system are same with
the former tests in Fig. 11. The toe structure set on the bottom of the container is not fixed.
However the friction coefficient between the rigid body and the surface of the bottom is
measured to be 0.78 in 30g, the rigid body isnt moved by dynamic motion with the
maximum acceleration of 485 gal. Fig. 17(b) shows the displacements of the embankment
model after shaking. It can be seen that the sliding plane is shallower and each displacement
measured is smaller than that without reinforcement in Fig. 12. Furthermore, it can be
indicated that it isnt necessary to fix the toe structure on the ground surface and its enough
to keep the high friction coefficient not to be slid. As the rigid body, the retaining wall and
the large-scaled gabion etc. can be considered as the practical toe reinforcement structures.

(a) Toe reinforced with rigid body (b) Residual displacement after shaking
Fig. 17. Centrifuge test on toe reinforcement structure
6.3 Artificial base-ground structure
The practical examples on the application of the Artificial Base-ground Structure and the
Crest Reinforcement Structure with geosynthetics are introduced as follows. Fig. 18 shows a
certain site same with Fig. 16 where the road embankment indicated by Layer 4 shown in
Fig. 18(a) approaching to the other construction site is constructed temporally. Fig. 18(a)
shows the soil conditions in each layer, i.e. Layer 1 and Layer 2 are natural rigid base
grounds, and Layer 3 is a little soft clayey deposit. The groundwater table is shown as the
blue line. The soil conditions of each layer are estimated in Fig. 18(b) based on the soil
investigation.
Sliding Plane 1
Sliding Plane 2
Before Motion
After Motion
unitmm
Before Motion
Toe Block


(a) Cross section (b) Conditions of soil profile
Fig. 18. Analytical model of in-situ field shown in Fig. 16
Four analytical cases of the reinforced embankment model are prepared as shown in Fig. 19.
Table 6 shows the design conditions for the geosynthetics to reinforce the crest of the
embankment and/or the improved strength of the embankment body as the artificial base-
ground where the sliding plane dont pass through.

Table 6. Reinforcement conditions
The seismic ground motion measured at Togi City during the Noto Peninsula Earthquake in
2007 which is same with the one in Fig. 7(a) is applied with the maximum acceleration of
508gal converted at the base ground.
The analytical results on the sliding displacement and reach length identified as shown in
Fig. 19(a) are shown in Fig. 19. These results can be summarized as follows.
Case 1: However the reach length of 12.8m is large, the sliding displacement of 0.16m can be
reduced to be small. The sliding displacement of 0.16m can be judged to be Level 2
performance identified in Table 2.
Case 2: However the sliding displacement of 0.69m is very large, the reach length of 1.7m
can be reduced to be small. Because the reach length of 1.7 is located around the shoulder of
the embankmentt, Level 2 performance identified in Table 2 can be expected.
Case 3: However the reach length of 23.4m is very large, the sliding displacement of 0.08m
can be reduced to be very small. The sliding displacement of 0.08m can be judged to be
Level 2 performance identified in Table 2.
Case 4: The sliding displacement of 0.23m is small a little, the reach length of 2.4m can be
also reduced to be small a little. Both the sliding displacement and the reach length can
show Level 2 performance identified in Table 2.

0 20 40 60 80
40
0 2 4 60 80
0
20
1
2
4
3

1,2,3,4 No. of Layer
0 20 40 60 80 (m)
(m)
40
20
0
The construction costs corresponding to the unit length in the longitudinal direction of the
embankment among these four cases can be ordered in Case 4, Case 1, Case 2 and Case 3
from the low cost to the high one. Because the cost ratio of Case 3 against Case 4 is about
1.13, Case 4 can be indicated to be the best reinforcement plan based on the seismic
performance-based design.

(a) Case 1: General improved ground (b) Case 2: Artificial base-ground

(c) Case 3: General improved ground and
geosynthetics
(d) Case 4: Artificial base-ground and
geosynthetics
Fig. 19. Sliding characteristics according to reinforcement structures
6.4 Barrier reinforcement structure
The centrifuge test is conducted on the effects by Barrier Reinforcement Structure with a
rigid plate shown in Photo. 14. The test conditions and measurement system are same with
the former tests in Fig. 11. The under edge of the barrier structure is set to be rotated and not
to be moved horizontally, i.e. assumed to be a hinge. The upper edge of the one is free and
not to be connected with the side wall of the container.

Photo 14. Centrifuge test on the Barrier Reinforcement Structure

It can be known that however the rigid plate is inclined a little toward the slope, the sliding
plane can be controlled toward the shoulder of the crest where the traffic performance isnt
affected severely. If the upper edge is fixed, the effectiveness of the barrier may become
larger. The Barrier Reinforcement Structure can be assumed to be effective to decrease the
residual displacement at the crest of road embankments. As the rigid plate, the sheet piles
and the reinforced concrete wall etc. can be considered as the practical barrier reinforcement
structures.
7. Conclusion
Lessons obtained in this chapter can be summarized as follows.
1. Based on the damages of road embankments induced by such typical earthquakes as
the 2004 Mid-Niigata Prefecture Earthquake, the 2007 Noto Peninsula Earthquake, the
2009 Suruga Bay Earthquake and the 2011 off Pacific coast of Tohoku Earthquake,
several important lessons for the performance-based design of road embankments can
be obtained from the engineering view points.
2. The macroscopic seismic assessment method proposed can be applied to estimate the
seismic risk potential of the existing road embankments for the future detail
investigation and seismic reinforcement plan.
3. The performance level proposed for the performance-based design of road
embankments can be applied to set the objective performance level of the sliding failure
of road embankments in the design of both new embankments and existing ones in the
future.
4. The practical performance-based design concepts such as Crest Reinforcement
Structure, Barrier Reinforcement Structure, Artificial Reinforcement Structure and
Toe Reinforcement Structure can be proposed and their applications are discussed
and clarified qualitatively and/or quantitatively.
8. Acknowledgment
The research and development on the economical seismic reinforcement technology
considering the performance of the road embankments conducted in 2007 to 2009 have been
supported by the Committee of New Highway Technology presided by Ministry of Land,
Infrastructure and Transport. The author would like to express grateful acknowledgement
to the Committee members and the cooperative researchers: Dr. Akinori Nakahira of CTI
Engineering Co. Ltd., Dr. Kenji Hayashi of Forest Engineering Inc., Dr. Naoki Tatta of
Maedakosen Co. Ltd., Dr. J. Jang of Shimizu Corporation and Mr. Yusuke Egawa of CTI
Engineering Co. Ltd..
9. References
Erosion Control & Land Slide Technical Center (2009). Design Guideline for Erosion Control
Steel Structure, pp.32 (in Japanese) Japan Meteorological Agency (JMA).
http://jma.go.jp/jma/press/1103/11g/kaisetsu201103112200_pdf (in Japanese)
Japan River Association (1997). Technical Code -Design II- for River & Erosion Control Structure,
pp.40 (in Japanese)
Japan Road Association (1999). Design Criteria for Retaining Wall, p.21 (in Japanese)
National Research Institute for Earth Science and Disaster Prevention (NIED).
http://www.bosai.go.jp/kyoshin/topics/html20110311144626/main_20110311144
626.html (in Japanese)
Newmark, N. M. (1965). Effects of earthquakes on dams and embankments, Geotechnique,
Vo.15, No.2, pp.139-159
Tateyama, M., Tatsuoka, F., Koseki, J. & Horii, K. (1998). Research on Seismic Design
Method of Embankment, Report of Railway research Institute, Vo.12, No.4, pp.7-12 (in
Japanese)
Tatta, N., Jang, J., Tokida, K., Oda, K. & Nakahira, A. (2009). Experimental Study on Seismic
Reinforcement Method at Crest of Road Embankment by Geosynthetics, Proceedings
of The International Symposium on Earthquake Geotechnical Case Histories for
Performance-Based Design, Is-Tokyo, pp.963-970, Tsukuba, Japan, June 15-18
Tatta, N., Jan, J., Tokida, K., Oda, K., Nakahira, A. (2010). Seismic Reinforcement Method at
Crest of Road Embankment by Geosynthetics, Proceedings of The 9th International
Conference on Geosynthetics, IGC2010, pp.1611-1618, Guaruj&aacute, Brazil, May 23-
27
Tokida, K., Oda, K., Nabeshima, Y. & Egawa, Y. (2007). Damage Level of Road Infra-
structure and Road Traffic Performance in The Mid Niigata Prefecture Earthquake
of 2004, Structural Engineering/Earthquake Engineering, JSCE, Vol. 24, No. 1, pp.51-61.
Tokida, K., Oda, K. & Nakahira, A. (2008). Research and Development on Economical
Seismic reinforcing Technology for Road Embankment Based on Performance-
based Concepts, Report of Research and Development for Improving Highway Policy,
Committee of New Highway Technology of Ministry of Land, Infrastructure and
Transport, No.17-4, pp.1-47 (in Japanese)
Tokida, K., Oda, K., Hayashi, K. & Nakahira, A. (2009). Macroscopic Aseismic Assessment
Method of Road Embankment, Proceedings of The International Symposium on
Earthquake Geotechnical Case Histories for Performance-Based Design, Is-Tokyo,
pp.1633-1638, Tsukuba, Japan, June 15-18
Tokida, K., Chen, W.Z, Nakahira, A., Echigawa, H., Minamimoto, M., Hayashi, K. (2010).
Field Test on Friction Factor between Gabion and Ground, Proceedings of The 4th
Japan-China Geotechnical Symposium, pp.141-147, Okinawa, Japan, April 12-14
12
Seismic Response Analysis and Protection
of Underground Monumental Structures
The Catacombs of Kom EL-Shoqafa,
Alexandria, Egypt
Sayed Hemeda
Conservation Department, Faculty of Archaeology, Cairo University
Egypt
1. Introduction
Alexandria is located in eastern part of the Mediterranean Basin (Northern Egypt) and it is a
place of great historical and religious interest. Numerous Catacombs and cemeteries for
Greek-roman were erected in Greek-roman and Christian era has been found. They
represent actually a large complex of an underground necropolis. The aim of the present
study is the investigation and documentation of the existing stability conditions of the site of
Catacombs in order to define the instability problems to interpret the pathology and to
propose the best retrofitting procedure.
On 28 September 1900 the ground on the Hill of potsherds (Kom El- Shoqafa) spontaneously
opened, and a donkey disappeared into the crevasse. The unfortunate beast had inadvertently
discovered one of Alexandrias most important archaeological sites, the principal hypogeum of
a funerary complex dating from the end of the first century of the Christian era and still in use
at the beginning of the fourth (Empereur,2003). It had been known for some time that this area
held antique tombs, since the hill has being extensively quarried to provide building materials
for a fast expanding modern Alexandria. Much had already been destroyed, though certain
archaeologists of the late 19th century had been able to record other tombs that were
subsequently to disappear. These reports have descriptions and drawings, which show that
the complex that can visit today was part of a vast necropolis, traces of which must still exist
under the foundations of the neighboring buildings.
The method of construction of these underground monuments and accesses was mainly
rock cutting and carving. The Ancient Egyptian did have a large experience in cutting and
digging in rocks. They used simple hand tools and with the experience they got in treatment
of rock either soft (e.g. limestone and sandstone) or hard (e.g. granite, basalt, quartzite), they
brought up the large number of monuments that we have discovered until now or that
which may be revealed in the future. It is hard to think nowadays how they did such works
with their simple toolscomparing them with the resources we have now available either
mechanical, electrical or other forms of developed equipment. The tools available at that
timeas already mentioneddid not exceed those made from hard stone, wood, copper,
iron and bronze. We can hardly imagine that all these monuments, pyramids, temples and
underground tombs were constructed by hand and by these simple tools.

298
The stability conditions of the historical monuments are of crucial interest, especially in
regions like the Mediterranean Basin and particularly Alexandria, Egypt, where the
seismotectonic and weathering regime is active and the geological structure is complex
.Phenomena like settlement and slope movements as well as earthquakes and tectonic
activity contribute to the damage of the historical buildings. The ground water activity is
also an important factor, especially in cases underground monuments the environmental
factor is also necessary to be taken in mind, when different protection measures are decided
to apply (hemeda, et al, 2007).
The earthquakes affected a large variety of structural systems. The severity of damage is a
function of the structural type, quality of work manship, material, and local soil conditions.
Observations of the damaged areas close to the epicenter of the Dahshour earthquake (Cairo,
1992) indicated that they can be rated VII on the Modified Mercalli intensity (MMI) scale with
an estimated peak ground acceleration of about 10% g, (Badawi and Mourad, 1994).
The damages which occurred can be classified as:
- STRUCTURAL; resulting in damage to load carrying components. Foundations,
Bearing walls, columns, etc.
- NON STRUCTURAL; producing damage to architectural or functional components,
such as partitions, suspended ceiling parapets.
Structural damage
Old load bearing unreinforced Masonry walls, constructed of solid blocks, had a large
number of failures and collapse. The extent of damage varied from minor cracks to complete
collapse of walls.
- Failures were due to inplane cyclic shear cracking. Out-of -plan e instability, and impact
with adjacent buildings (pounding)
- Masonry walls that were acting as a shear walls showed flexural failures between
windows, combined with some diagonal shear cracking.
- In spite of damage to wall- bearing structures, recently constructed reinforced concrete
buildings did not suffer severe damage, except the complete collapse of a residential
building in Cairo.
Non structural damage
- - Different types of cracks, shear and bending were observed in partitions.
- - Poorly built parapets were either dislocated or collapsed due to relative motion.
- - Collapsed many masonry parapets and facades
Geotechnical effects of the earthquake
The soil in the area (Cairo and its districts) is silty to fine-sandy soil that makes up the valley
of the Nile. When this type of soil is water-saturated and strongly shaken, it is highly
susceptible to the liquefaction phenomenon.
- The soil can quickly convert from solid material into liquid which has no strength to
support structures that may rest its surface on when subjected to prolonged shaking.
- The potential for widespread liquefaction is very high on the Nile flood plain, especially
in irrigated fields. A highway west of the Nile sunk as much as105m because the
agricultural land adjacent to the pavement had liquefied over a few hectares
Seismic Response Analysis and Protection of Underground
Monumental Structures The Catacombs of Kom EL-Shoqafa, Alexandria, Egypt

299
- When the strength of the soil beneath the structure vanished (due to the excess pore
pressure) during the earthquake, the structures settled as the liquefied soil escaped
latterly into the field.
- Also, many zones of liquefaction in the agricultural fields were reported when revealed
on the ground by sand boils.
The instability problems and depreciation phenomena of subterranean monuments in
Alexandria is not likely to be dominated by gravity fall or sliding on structural features,
other factors such as excessively high rock stress, creep effect, poor geotechnical properties
of rock structures, weathering and /or swelling rock and excessive groundwater pressure or
flow, seismic loading as well as utter lack of preservation become important and can be
evaluated by means of a classification of rock quality
The aim of the analysis carried out in this research is to investigate the safety margins of the
underground monuments, under their present conditions, against unfavorable
environmental (i.e. weathering and high underground water table), utter lack of
preservation, geotechnical and extreme seismic conditions.
Underground structures safety analysis is performed using the finite element (FE) method.
The research presents a comprehensive study for the underground monuments safety
analysis. The safety analysis includes not only a failure analysis but the effect of weathering
specially the underground water on the differential settlement will be investigated. The
commercial FE package Plaxis (Karstunen et al, 2006) is used for conducting stress, as well
as settlement analysis. PLAXIS is a finite element program developed for numerical analysis
of geotechnical and underground structures (plaxis manual, 2002).
To compute the deformation of these underground monuments as realistically as possible,
an advanced nonlinear elastoplastic material model needs to be utilized in PLAXIS which is
capable of utilizing such advanced material models. MohrsCoulomb model is used for
deformation and consolidation analysis in this study. The consolidation analysis is
performed using PLAXIS utilizing Biots consolidation theory in 2D (Biot, 1941) and the
nonlinear material behavior is taken into account as mentioned before.
Also in this research, we attempt to construct and analyze a three-dimensional (3D) finite
element model (FEM) of the central rotunda in catacombs of Kom El-Shoqafa with its six
supporting rock piers excavated in sandy oolitic limestone deposit, using the FLAC 3D code.
For the seismic analysis, we have modeled the complex catacomb assuming an equivalent
plane strain model and applying the Plaxis b.v. 8 with different seismic scenarios,
corresponding to the seismotectonic features of Alexandria. Advanced soil-rock elastoplastic
modeling has been used. Extensive time domain parametric analysis were performed in
order to examine the response of the catacombs subjected to seismic motions with different
amplitudes of ground motion and different frequency content. (Kalamata in Greece, 1986,
Erzincan in Turkey, 1992, Aqaba in Egypt, 1995). The analysis takes into account the
complex behavior of the structure with the aim to determine the threshold peak ground
acceleration (PGA) and the corresponding developed stresses, which should remain lower
than the actual strength of different elements composing the catacombs.
2. The objectives of this study
The Catacombs in Alexandria, Egypt from the Greek-Roman era represent cultural heritage
of outstanding universal values. They suffer weathering aging as well as multiple

300
geotechnical and earthquake problems. A pilot study has been carried out for on the
Catacombs of Kom El-Shoqafa in order (a) to define the pathology and the causes (b) to
investigates the safety assessment of the structure under static and seismic loads. (c) To
assess the global risk due to several factors and (d) to define the appropriate retrofitting

Fig. 1. Plan view and the two main cross- sections of Catacombs of Kom El-Shoqafa. (a)
cross-sections and (b) plan view.

301


302

Fig. 2. Present state of the catacombs of Kom El-Shoqafa..
techniques to be applied. In the paper a general outline of the various tests, surveys and
analyses is presented, highlighting the most important issues related to the static and
seismic stability of underground structures like Catacombs in particularly unfavorable
geotechnical and environmental conditions.

303
Parameters Symbol and unit
Rock
material
Brick
(old restoration)
Rock unit weight
unsat
kN
/m
3
18 17
Rock unit weight
sat
kN
/m
3
22 20
Youngs modulus E
ref
kN/m
2
2.270 x10
6
1.350 x10
6

Shear modulus G
ref
kN/m
2
8.867 x10
5
5.400 x10
5

Oedometer modulus E
oed
kN/m
2
2.902 x10
6
1.620 x10
6

Poissons ratio
(nu)
0.28 0.25
Cohesion c
ref
kN/m
2
500 450
Friction angle
O
35 31
Shear velocity Vs m/s 715 557.9
Longitudinal
wave velocity
Vp m/s 1293 9663
Uniaxial
compressive strength
UCS kN/m
2
2400 1100
Bending strength y kN/m
2
560 200
Shear strength
f
kN/m
2
364 -
Dilatancy (o) 1
o
0
Table 1. Material properties of rock and other construction materials used for the stability
analysis of the Catacombs of Kom El-Shoqafa.
3. Preliminary 2D static analysis
In the initial static analysis, the excavation is modeled by assuming non-linear soil / rock
behavior and the Mohr coulomb failure criterion. The following parameters are used: =36
o
,
c=500 kN/m
2
, E=2.270E+06 KN/m
2
, =0.28, Vs=715 m/sec for the rock material, and =31
o
,
c=500 kN/m
2
, E=1.350E+06 KN/m
2
, =0.25, Vs=550 m/sec for the modern brick support
walls and piers. Figure 3 presents the total displacements of the soft rock excavations of
catacombs of Kom El-Shoqafa, for two typical cross sections, 1 and 2, for the first and second
floor, respectively. The results from the preliminary static analysis indicate that the ground
displacements were small (of the order of a few millimetres; 1.13 mm); some rock pillars are
under relatively high compression stresses. The calculated effective peak principal
compressive stresses on supporting rock pillar 1 is about 1.42 MPa.

304

Fig. 3. Total vertical static displacements at the catacombs of Kom El-Shoqafa for the first
and second floor.

305
4. Preliminary 3D static analysis
An attempt has been made to construct a three-dimensional (3D) numerical model for the
central Rotunda in the catacombs of Kom El-Shoqafa with its six supporting rock pillars
excavated in sandy oolitic limestone deposit. The objective of the 3D analyses is to evaluate
the stress state in the pillars taking into account the 3D geometry. The 3D effects issue is
considered on a basic engineering approach in the subsequent sections. The various
simulations described herein are conducted using the FLAC 3D code (Itasca, 2007).

Fig. 4. Contour lines of vertical effective stresses
yy
through the rotunda. The maximum
effective compressive stresses on pillar_1 = -1.74*10
3
( ) kN/m
2
.

306
The results from the 3D static analysis indicate that the ground displacements above the
rotunda (catacombs) small. The maximum total (vertical) displacements range between 2.6
mm and 3 mm in the whole domain. The peak horizontal displacement is about 1.0 mm.
Some rock piers are under relatively high compression stresses. The calculated peak
effective principal vertical compressive stresses on supporting rock pillar 1 is 1.7410
3
KN/m
2
and the calculated peak effective principal tensile stress is about 200 KN/m
2
. The
factor of safety of the rock pillar 1 is 1.47, (note that the acceptable safety factor for the
underground structures is > 1.6 in static state ). Also the overstress state is beyond the elastic
regime (limit of domain) (Hemeda, 2008).

Fig. 5. Results from the 3D analysis indicating that the maximum ground displacements
above the rotunda are relatively small (of the order of 2.6 mm), and the maximum
horizontal displacement is about 1 mm.

307
5. Seismic response analysis
In the present study, we have selected three reference earthquakes to be used as input motions
to the time-history analysis of the monumental structure studied, namely: (i) Aqaba, Egypt,
1995, (ii) Erzincan, Turkey, 1992 and (iii) Kalamata, Greece, 1986. The time histories (Figure 6)
of these earthquakes representing different seismotectonic settings and frequency content,
were scaled to three peak ground acceleration values equal to 0.08g, 0.16g, and 0.24g (g is
acceleration due to gravity), respectively. The design acceleration in Alexandria according to
the Egypt seismic code is 0.08g (ECP-203). The structures have been analyzed under single
horizontal ground acceleration and not under the three ground acceleration.
5.1 Seismicity of Egypt
Egypt lies at the north east of the African tectonic plate whose borders are the Red Sea and
the Jordan Valley. Another major tectonic feature is the African Rift Valley which extends
from the Red Sea down through Ethiopia towards South and East Africa. The contemporary
geological setting of the region is mainly governed by active extensional tectonics taking
place within the Gulf of Suez Rift. For over 20 years, geologically young faulting has been
known to exist both along the margins and within the Gulf of Suez (Said, 1990).
In the 15th century, an Egyptian scholar listed more than 30 earthquake events that happened
from AD 796 to 1500. The Dahshour earthquake was the latest in a long history of earthquakes
that occasionally cause damage to buildings in Cairo and surrounding areas in northeastern
Egypt. On 7 August 1847, an earthquake with an estimated magnitude of 6.8 occurred about
100 km southwest of Cairo, near E1-Fayoum. Three thousand houses and 42 mosques were
destroyed in E1- Fayoum and damage was extensive in Cairo (Kebeasy et al., 1976).
Two recent earthquakes occurred within the last three decades. The first, having a magnitude
7.0, occurred in the northern part of the Red Sea on 31 March 1969. It caused landslides, rock
falls, and fissures, in addition to damage to a power station and some hotels in a city
nearby (Maamoun et al., 1984). The second earthquake occurred on 14 November 1981, in
Aswan, 1000 km south of Cairo, along the Nile River. It measured 5.6 on the Richter scale, in
an area that was considered a seismic. Some scientists attribute this earthquake to the
construction of a dam along the river, resulting in an artificial reservoir, approximately 300 km
long, and having a maximum capacity of 164000 million cubic meters (Bolt, 1988).
5.2 Seismicity of Alexandria
Alexandria during his long history has suffered important seismic damages from near and
distant sources earthquakes. As the under-water archeological remains in Abou Kir bay
strongly support, either local or remote earthquakes (El-Sayed, 2004) destroyed the city.
Seismogenic zones such as the Red Sea, Gulf of Aqaba-Dead Sea Hellenic Arc, Suez-Cairo-
Alexandria, Eastern-Mediterranean-Cairo-Faiyoum and the Egyptian coastal area may all
affect the city. However, the seismic hazard of the city has not been fully defined.
Alexandria is located approximately 300 to 600 km from three known active plate
boundaries, namely: the Red-Sea, the Gulf of Aqaba and the Hellenic Arc (Mckenzie 1970,
1972; Sestini, 1984; Mesherf, 1990). The interaction among these three plate boundaries
created major fault zones in Egypt as: (1) Eastern- Mediterranean Cairo Faiyoum fault zone
(Neev, 1975; Sestini, 1984; Mesherf, 1990), (2) Suez-Cairo-Alexandria fault zone (Kebeasy,

308
1990). These fault zones are very close to the city of Alexandria. Moreover, deep seismic
sounding reveals that there are minor faults such as Rosetta fault which trend a few
kilometers from Alexandria city (Hussein & Abd Allah, 2001).
Because of this complex tectonic setting, many earthquakes occurred approximately
Alexandria, both in recent and historical time (Ambraseys et al., 1994; Maamoun et al., 1984).
The spatial distribution of the earthquakes epicenters shows that there are areas of very
intense (e.g., plate boundaries) and others of low (e.g., offshore area) activities. For those of
intense seismicity, there are a considerable number of focal mechanisms that allow us to
understand its geodynamic behaviors (Mckenzie, 1970; and Rotstein and Kafka, 1982; Sestini,
1984; CMT database). Controversy, the number and quality of the focal mechanisms available
for those of low seismic activity (like the Egyptian coastal zone) are not enough to have the
clear understanding for the tectonic setting (Sestini, 1984; Kebeasy, 1990; Mesherf 1990).
Because of this complex stress regime, damaging earthquakes had occurred in the vicinity of
Alexandria (Ambraseys et al., 1998; Kebeasy, 1990; Maamoun et al., 1984). Some of these
damaging earthquakes are apparently missing. As an example, the damage in Menouthis
and Herakleion is more likely caused by an earthquake (which is not known yet). This is
supported by: (1) the collapsed columns are falling down in the same direction NE-SW, (2)
the presence of coins and jewelry suggest a sudden collapse, (3) the sharp sand grains in the
bottom of Abou kir Bay reflecting active tectonic environment, and (4) the Cairo- Alexandria
fault system (NE-SW) is passing by the area and recently generated frequent moderate
earthquakes (GEOTIMES, 2000). This does not exclude the possibilities of land subsidence as
suggested in the STANFORD REPORT (2000).
5.3 Seismic input
In the present study, we have selected three reference earthquakes. (i) Aqaba, Egypt, 1995
(ii) Erzincan, Turkey, 1992 and (iii) Kalamata, Greece, 1986. The time histories (Figure.6) of
these earthquakes representing different seismtectonic settings and frequency content were
scaled to three peak ground acceleration values equal to 0.08g, 0.16g,and 0.24g respectively.
And they are used as input motions at the bedrock. The design acceleration in Alexandria
according to the Egypt code is 0.08g.
We believe that with the advances in computational methods it is now possible to predict
with reasonable accuracy the seismic demands on these geometrically complex monuments.
Specially, computer modeling and simulations are very useful tools for identifying regions
of stress concentration where only non-invasive techniques are allowed. Accurate
quantification of stresses are also useful for understanding the direction of cracks
propagation and for quantifying the seismic demands on whatever new materials may be
introduced in the retrofit program. Three earthquakes were chosen:
Aqaba: 22/11/1995, M=7.1, Ml=6.2: Station Eilat. Distance (km): Closest to fault rupture (93.8).
Erzincan:13/3/1992, Mw=6.9, Ms=6.8.Station:95 Erzincan, Rrup=2km, Re=1km.
Kalamata:13/9/1986.17:24:35, Ms=5.8 .Mw=5.9, Ml=5.5. Station: Old telecommunication
building, Re=10 km.
22 November 1995, an major earthquake (Aqaba earthquake) with a magnitude of 6.2 on the
local scale ML, and a moment magnitude of MW D 7:1, (PDE bulletin, 1996), struck the

309
shorelines cities of the Gulf of Aqaba, such as Aqaba (Jordan), Eilat (Israel), Hagel (Saudi
Arabia) and Nuweiba (Egypt), Damage occurred in many parts northeastern Egypt as far as
Cairo. This major event was followed by 2089 earthquakes ranging in magnitude from 2 to
5.5 on local magnitude (ML) recorded and/or analyzed by the Jordan Seismological
Observatory (JSO bulletin, 1998). This seismic swarm activity began on 22 November 1995
and continues at least until the end of December 1997, (EID AL TARAZI, 2000).
The Kalamata earthquake was recorded on hard ground at a distance of about 9 km from
the epicenter and its magnitude was Ms=6.2. The record samples the near field strong
motion that caused considerable damage to the buildings of the city of Kalamata. The
duration of the strong motion is about 6 sec and the maximum accelerations are 0.24g in the
N-S direction and 0.27g in the E-W direction. The corresponding peak velocities are 32.0 and
23.5 cm/s, respectively.
The 13 March 1992 Erzincan earthquake, M=6.8, occurred in the eastern half of the Erzincan
basin. The largest aftershock took place near Plmr on 15 March 1992. No clear surface
breaks were observed, although teleseismic studies suggested that it was a strike-slip
earthquake striking parallel to the North Anatolian fault, with a focus of approximately 102
km depth, 30 km rupture length, 95 cm of slip, and a 1.161026 dyn.cm seismic moment.
The aftershock distribution concentrated at an area of the intersection between the North
Anatolian fault and the Ovacik fault. These results indicate that the previously suggested
seismic gap along the North Anatolian fault, east of Erzincan, remains unruptured (AYKUT
et al 1993).
The criterion for this choice was their different frequency content, as they will give
information about the response of these structures in different period ranges Figure (6)
shows one of the horizontal components of acceleration for each record. The records were
retrieved from PEER and ESMD online database.
In order to estimate the threshold PGA values to collapse, a set of parametric analysis was
carried out. These structures were subjected to increasing level of horizontal accelerations.
Figures (7) through (13) depict the main results of the analysis. In case of the Aqaba
earthquake, it is clear that a great part of seismic energy is dissipated by the upper parts of
catacombs (ground surface) even for small values of PGA. Kalamata and Erzincan input
motions give much lower displacements values (Arias intensity and total duration will
explain the difference in responses from the three ground accelerations). The maximum
horizontal displacement at the top of catacombs for Aqaba earthquake at PGA = 0.24g
earthquake scenario was u
x
=

7.95 cm, while the peak vertical effective principal stress was
4190 kN/m
2
. For the Erzincan and Kalamata earthquakes, the respective values were 3400
kN/m
2
and

3580 kN/m
2
, respectively. Moreover, the maximum horizontal displacement at
the top of the catacombs was 2.34 cm and 2.01 cm for Erzincan and Kalamata earthquakes
respectively. The maximum vertical displacement at the top of catacombs, are 3 mm, 2.6
mm, and 5.3 mm for Kalamata, Erzincan, and Aqaba earthquakes, respectively.
Given the value of the static strength estimated in the laboratory (UCS=2.5 MPa), the seismic
analysis of the catacombs complex proves that the supporting rock piers and columns,
which are the most vulnerable parts of the whole complex, are rather safe for PGA values
lower than 0.24g in case of the Kalamata and Erzincan earthquakes and PGA =0.12g for the
Aqaba seismic scenario (see Figure 13a, b).

310

Fig. 6. Seismic excitations at the bedrock for the reference ground motions (acceleration-time
history). (a) Aqaba, 1995 (b) Erzincan 1992, and (c) Kalamata, 1986 earthquakes.

311
For larger earthquakes, which are most likely to happen in the region, the seismic stability of
the catacombs is not satisfied and it is necessary to proceed to specific retrofitting works to
upgrade their seismic performance. The maximum relative horizontal displacements of the
top and the base of the rock piers are of the order of 3 to 5 mm (Figure 10). Considering that
the induced seismic ground deformations are better correlated with the intensity of
damages in underground structures, the seismic design of the catacombs must be based on
these kinematic forces.


312

-4
-3
-2
-1
0
1
2
3
0 3 6 9 12 15
Dynamic time {s}
a
c
c
e
l
e
r
a
t
i
o
n

{
m
/
s
2
}
Top of Catacombs
Bottom of Catacombs
Bottom of mesh

-2
-1
0
1
2
3
0 4 8 12 16 20
Dynamic time {s}
a
c
c
e
l
e
r
a
t
i
o
n

{
m
/
s
2
}
Top of Catacombs
Bottom of Catacombs
Bottom of mesh

-3
-2
-1
0
1
2
3
4
0 4 8 12 16 20 24 28 32 36 40
Dynamic time {s}
a
c
c
e
l
e
r
a
t
i
o
n

{
m
/
s
2
}
Top of Catacombs
Bottom of Catacombs
Bottom of mesh


313

-0.024
-0.016
-0.008
0
0.008
0.016
0.024
0.032
0 3 6 9 12 15
Dynamic time {s}
D
i
s
p
l
a
c
e
m
e
n
t

{
m
}
Top of Catacombs
Bottom of Catacombs
Bottom of mesh

-0.048
-0.032
-0.016
0
0.016
0.032
0 4 8 12 16 20
Dynamic time {s}
D
i
s
p
l
a
c
e
m
e
n
t

{
m
}
Top of Catacombs
Bottom of Catacombs
Bottom of mesh

-0.12
-0.08
-0.04
0
0.04
0.08
0.12
0 4 8 12 16 20 24 28 32 36 40
Dynamic time {s}
D
i
s
p
l
a
c
e
m
e
n
t

{
m
}
Top of Catacombs
Bottom of Catacombs
Bottom of mesh

Fig. 7. Deformed meshes and peak total displacement, and acceleration, horizontal
displacements-time histories, (a) Kalamata , (b) Erzincan , and (c) Aqaba eartquakes. Input
motion PGA = 0.24g. Catacombs of Kom El-Shoqafa, a typical cross section 1, first floor.

314
-4500
-4000
-3500
-3000
-2500
-2000
-1500
-1000
-500
0
0 3 6 9 12 15
Dynamic time {s}
y
y
{
}
Top of rock pier_1
Middle of rock pier_1
Base of rock pier_1
UCS of rock material

-200
-100
0
100
200
-2.E-04 0.E+00 2.E-04 4.E-04 6.E-04
Shear strain %
S
h
e
a
r

s
t
r
e
s
s

s
i
g

/
-
x
y

{

K
N
/
m
2
}
Top of rock pier_1
Base of rock pier_1

a)

315
-4500
-4000
-3500
-3000
-2500
-2000
-1500
-1000
-500
0
0 5 10 15 20
Dynamic time {s}
y
y
{
}
Top of rock pier_1
Base of rock pier_1

-200
-100
0
100
200
-3.E-04 -2.E-04 -1.E-04 0.E+00 1.E-04 2.E-04
Shear strain %
S
h
e
a
r

s
t
r
e
s
s

s
i
g

/
-
x
y

{

K
N
/
m
2
}
Top of rock pier_1
Base of rock pier_1

b)
E
f
f
e
c
t
i
v
e

s
t
r
e
s
s
e
s

y
y

{
k
N
/
m
2
}

316

-4500
-4000
-3500
-3000
-2500
-2000
-1500
-1000
-500
0
0 5 10 15 20 25 30 35 40
Dynamic time {s}
y
y
{
}
Top of rock pier_1
Base of rock pier_1

-200
-100
0
100
200
-1.E-03 -8.E-04 -4.E-04 0.E+00 4.E-04 8.E-04 1.E-03
Shear strain %
S
h
e
a
r

s
t
r
e
s
s

s
i
g

/
-
x
y

{

K
N
/
m
2
}
Top of rock pier_1
Base of rock pier_1

c)
Fig. 8. Effective vertical compressive stresses
/
YY
- time histories and shear stress versus
shear strain for the most critical rock pier 1 (Figure 9 at the right side of rotunda). (a)
Kalamata (b) Erzincan (c) Aqaba earthquakes. PGA value = 0.24g.

317

Fig. 9. The typical cross sections 1and 2, which have been used in the seismic analysis.

318
0
5
10
15
20
25
H
o
r
i
z
o
n
t
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
m
}
1 2 3 4
Number of rock pier
top of rock piers
Bottom of rock piers

0
0.5
1
1.5
2
2.5
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
m
}
1 2 3 4
Number of rock pier
Top of rock pier
Bottom of rock pier

28
28.5
29
29.5
30
30.5
31
H
o
r
i
z
o
n
t
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
m
}
1 2 3 4
Number of rock pier
Top of rock pier
Bottom of rock pier


319
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
m
}
1 2 3 4
Number of rock pier
Top of rock pier
Bottom of rock pier

74
75
76
77
78
79
80
81
82
83
H
o
r
i
z
o
n
t
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
m
}
1 2 3 4
Number of rock pier
top of rock piers
Bottom of rock piers

0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
m
}
1 2 3 4
Number of rock pier
Top of rock pier
Bottom of rock pier

Fig. 10. Differential horizontal and vertical displacements on the top and base of the rock
piers in the catacomb complex (a) Kalamata at time (t) =4.60 sec, (b) Erzincan at (t) = 7.09
sec, (c) Aqaba at time (t) = 23.22 sec. PGA = 0.24g.

320


321

-0.0005
0
0.0005
0.001
0.0015
0 3 6 9 12 15
Time{s}
Top of Catacombs
"Bottom of Catacombs"
"Bottom of mesh"

-0.0005
0
0.0005
0.001
0.0015
0.002
0 4 8 12 16 20
Time {s}
Top of Catacombs
Bottom of Catacombs
Bottom of mesh

-0.001
0
0.001
0.002
0.003
0.004
0 5 10 15 20 25 30 35 40
Time {s}
Top of Catacombs
Bottom of Catacombs
Bottom of mesh

V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
}
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
}
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
}

322

-80
-40
0
40
80
-2.E-04 -1.E-04 -5.E-05 0.E+00 5.E-05 1.E-04 2.E-04
S
h
e
a
r

s
t
r
e
s
s

s
i
g

/
-
x
y

{

K
N
/
m
2
}
Top of rock column_1
Middle of rock column_1
Base of rock column_1

-80
-40
0
40
80
-6.E-04 -4.E-04 -2.E-04 0.E+00 2.E-04 4.E-04
Shear strain %
S
h
e
a
r

s
t
r
e
s
s

s
i
g

/
-
x
y

{

K
N
/
m
2
}

-80
-40
0
40
80
-1.E-04 -5.E-05 0.E+00 5.E-05 1.E-04 2.E-04
Shear strain %
S
h
e
a
r

s
t
r
e
s
s

s
i
g

/
-
x
y

{

K
N
/
m
2
}

Fig. 11. Deformed meshes, vertical displacement-time history, shear stresses-shear strain, (a)
Kalamata , (b) Erzincan , and (c) Aqaba earthquakes. Input motion PGA = 0.24g. Catacombs
of Kom El-Shoqafa, a typical cross section 2, second floor, the royal tomb.

323

0
0.02
0.04
0.06
0.08
0.1
0.08 0.16 0.24
PGA Value {g}
M
a
x
.

H
o
r
i
z
o
n
t
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
}

Aqaba RQ
Erzincan RQ
Kalamata RQ

0
0.001
0.002
0.003
0.004
0.005
0.006
0.08 0.16 0.24
PGA Value {g}
M
a
x
.

V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

{
m
}

Aqaba RQ
Erzincan RQ
Kalamata RQ

Fig. 12. (a) Maximum horizontal displacements, (b) Maximum vertical displacements on the
top of Catacombs, for Aqaba, Erzincan, and Kalamata earthquakes, scaled to several values
of PGA.

324

Fig. 13. (a) Maximum vertical effective compressive stresses
/
yy, (b) Maximum effective
shear stresses
/
xy on the base of rock pier_1, for Aqaba, Erzincan, and Kalamata
earthquakes, scaled to several values of PGA.
6. Polymer anti-seismic piling to protect the catacombs against strong
earthquakes
We employed a newly developed polymer seismic isolation method, which has been
employed successfully to retrofitting the Nakagawa underground station in Yokohama city
(Japan), to protect the underground structures of catacombs of Kom El-Shoqafa against
strong earthquakes with PGA>0.24g. The polymer seismic isolation method outlined in
Figure 14 presents certain advantages compared to other conventional anti-seismic methods
(such as the steel jacket method).This method is apriority suitable for the seismic protection

325
of the underground monuments, because the application of this technique is non-destructive
and do not involve any change in the original materials of the monuments; in addition we
do not employ new materials or constructions to the monumental initial structure.

Catacombs wall

Fig. 14. The polymer seismic isolation method to protect the catacombs against strong
earthquakes, (conceptual diagram).

326
The polymer seismic isolation method' involves the construction of polymer walls on both
sides of underground structure in order to reduce the seismic actions transmitted from the
surrounding ground to the structure. The stiffness of the polymer material should be about
1/10 to 1/100 that of the surrounding ground. This method is not intended to prevent or
control the seismic ground deformation itself, but to isolate structures from seismic forces
transmitted from the surrounding ground (Hemeda, 2008).
-0.0005
0
0.0005
0.001
0.0015
0.002
0.0025
0.003
0 3 6 9 12 15
Dynamic time {sec}
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t
s

{
m
}
Catacombs not equipped with
polymer seismic isolator
Catacombs protected from
seismic activity

-600
-400
-200
0
200
400
600
0 3 6 9 12 15
Dynamic time {sec}
E
f
f
e
c
t
i
v
e

s
h
e
a
r

s
t
r
e
s
s
e
s

/
x
y

{
k
N
/
m
2
}
Catacombs not equipped with polymer
seismic isolator
Catacombs protected from seismic
activity
Shear strength of rock
material.experimental

a)

327
-600
-400
-200
0
200
400
600
0 4 8 12 16 20
Dynamic time {sec}
E
f
f
e
c
t
i
v
e

s
h
e
a
r

s
t
r
e
s
s
e
s

/
x
y

{
k
N
/
m
2
}
Catacombs not equipped with polymer
seismic isolator
Catacombs protected from seismic
activity

-0.0005
0
0.0005
0.001
0.0015
0.002
0.0025
0.003
0 4 8 12 16 20
Dynamic time {sec}
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t
s

{
m
}
seismic activity

b)

328

-600
-400
-200
0
200
400
600
0 10 20 30 40
Dynamic time {sec}
E
f
f
e
c
t
i
v
e

s
h
e
a
r

s
t
r
e
s
s
e
s

/
x
y

{
k
N
/
m
2
}
Catacombs not equipped
with polymer seismic isolator
seismic activity

-0.002
0
0.002
0.004
0.006
0.008
0 10 20 30 40
Dynamic time {sec}
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t
s

{
m
}
seismic activity

c)
Fig. 15. Vertical displacements, Effective shear stresses
/
xy
- time histories on the base of
rock pier 1 before and after the installation of polymer anti-seismic slurry walls. (Figure 9 at
the right side of rotunda). (a) Kalamata (b) Erzincan (c) Aqaba earthquakes. PGA = 0.24g.

329

a)

b)
Fig. 16. Maximum effective shear stresses sig/xy on the base of rock pier_1, (a) Initial model
(b) After the installation of polymer anti-seismic piling. For Aqaba, Erzincan, and Kalamata
earthquakes, scaled to several values of PGA.
With the installation of polymer anti-seismic slurry walls, the shear forces on the rock
pillars, which are the most vulnerable parts inside the catacombs, reduce by up to 50% for
the three earthquakes scenarios, (see Figure 15). The effective shear stresses on the pillars are
also reduced considerably: in particular from 410 kN/m
2
to 250 kN/m
2
in the case of

330
Kalamata earthquake for PGA = 0.24g, from 400 kN/m
2
to 280 kN/m
2
in the case of
Erzincan earthquake, and from 430 kN/m
2
to 300 kN/m
2
in the cases of Aqaba earthquake.
The decrease of the computed acceleration values was also obvious in all the three
earthquakes scenarios; for example the horizontal acceleration on the top of the catacombs
decreases from 3 m/s
2
to 2 m/s
2
in the case of Aqaba earthquake and from 2.6 m/s
2
to 2.3
m/s
2
in the case of Erzincan earthquake. The horizontal displacements on the top of
catacombs decrease from 24 mm to 20 mm in the case of Kalamata earthquake and from 25
mm to 21 mm in the case of Erzincan earthquake, while in the case of Aqaba earthquake, for
PGA =0.24g, the displacement reduction is lower; from 80 mm to 78 mm. The vertical
displacements at the top of catacombs decrease from 2.8 mm to 1.0 mm in the case of
Kalamata earthquake, from 2.4 mm to 1.5 mm in the case of Erzincan earthquake, and from
6 mm to 2.5 mm in the case of Aqaba earthquake.
From the above short presentation and discussion it is obvious, that the seismic stability of
the catacombs has been upgraded after the installation of the polymer anti-seismic slurry
walls in the perimeter of an underground monument. The relative deformations are reduced
considerably and the developed seismic shear forces on the sidewalls can be easily
controlled within acceptable safety margins even for major earthquakes.
It is expected that the above-mentioned polymer seismic isolation method may contribute
effectively to the improvement of the seismic safety margin of underground monuments,
without employing other retrofitting techniques, which may modify the architectural and
archeological principles of the preserved monuments.
7. Conclusions
Considering all other affecting factors (aging, weathering, multiple geotechnical and seismic
instability problems and the specific geometry of the complex it has been shown that the
low rock strength affects seriously the safety of the catacombs both under static and seismic
loading conditions.
The results from the 2D-3D static analysis indicate that the ground displacements above the
catacombs are small (the maximum total vertical displacements are of the order of 2.6 mm to
3 mm in the whole domain, and the peak horizontal displacement is 1.0 mm). Some rock
piers exhibit relatively high compression stresses. The calculated peak effective principal
vertical compressive stress on the pillar 1 is -1.74* 10
3
KN/m
2
. The calculated peak
effective principal tensile stress is 200 KN/m
2
and the factor of safety of the rock pillar 1 is
1.47, which is not adequate, where the acceptable safety factor must be > 1.6. Also the
overstress state is beyond the elastic limit. It is so then damage does occur. You can read
more on damage assessment and future work that can be done using damage indices
(journal of structural engineering 137(3): 456-567: damage-based design earthquake loads for
sdof inelastic structures).
The seismic analysis of these underground monumental structures for three seismic
scenarios of different PGA values, proved that for PGA > 0.10g, which is a rather low value
considering the seismic activity and the past seismic history of the city, there are some
critical supporting parts of these catacombs structures (i.e. rock piers and columns) that are
not safe, and in general, the catacombs need considerable strengthening.

331
We also presented some preliminary results applying a polymer seismic isolation method
which has been proved to contribute effectively in reducing the induced inertial forces on
this type of underground monuments.
8. Acknowledgment
The author wish to express his deep acknowledgement to Professor Pitilakis.K, Dr Bakasis. I,
Civil Engineering Department, AUTH, Greece for his support in performing part of the
experimental and numerical analysis.
9. References
Biot, M.A. (1941). General theory of three- dimensional consolidation. International Journal of
Applied Physics Vol.12, pp155164.
Badawi, H. S. (1991). High earthquake risk buildings in Egypt. International. Sympos.
Geophysical Hazards in Developing Countries and their Environmental Impacts,
Perugia, Italy, August 1991.
Badawi, H. S and Mourad,Sh. (1994) .Observations from the 12 October 1992 Dahshour
Earthquake in Egypt. Natural Hazards 10: 261-274, 1994.
Bolt, B. A. (1988).Earthquakes. W. H. Freeman, New York, pp. 147-150.
El-Sayed, A & Korrat, I & Hussein, M. (2004). Seismicity and seismic hazard in Alexandria
(Egypt) and its surrounding. International Journal of pure appl.geophys.Vol.161 (2004),
pp1003-1019.
Hemeda, S., Pitilakis, K., Bandis, S., Papayianni, I., and Gamal M.(2007). The Underground
monuments in Alexandria, Egypt. Proc. 4
th
International Conference on
Earthquake Geotechnical Engineering, Thessaloniki, Greece, 25-27 June, pp. 715-
738.
Hemeda, S. (2008). An integrated approach for the pathology assessment and protection of
underground monuments in seismic regions. Application to some Greek-Roman
monuments in Alexandria, Egypt, PhD Thesis, Aristotle University of Thessaloniki,
Greece.
Pitilakis, K., Hemeda, S., and Bandis, S. (2009). Geotechnical investigation and seismic
analysis of underground monuments in Alexandria, Egypt. 17
th
International
Conference on Soil Mechanics & Geotechnical Engineering, Alexandria, Egypt,
October 5-9.
ITASCA Consulting Group. (2007). Inc, 3 Dimensional distinct element code (3D DEC).
Karstunen, M & Wiltrafsky, C & Krenn, H & Scharinger, F. (2006). Modeling the behavior of
an embankment on soft clay with different constitutive models. Vol. 30, Issue 10, 25
August, pp 953982.
Kebeasy, R. M. and Albert, R. N. (1976). Investigation of the seismo-tectonic nature of the
Middle East region, Helwan Institute of Astronomy and Geophysics Bulletin, No.
126.
Maamoun, M., Megahed, A., and Allam, A. (1984). Seismicity of Egypt, Helwan Institute
of Astronomy and Geophysics Bulletin, Vol. IV, Ser. B, pp. 109-162.
Ministry of Development, New Communities, Housing and Utilities: 1989, Egyptian Code of
Practice for Design and Execution of Reinforced Concrete Constructions.

332
Moustafa,A. (2009). Egypt seismicity and national seismic network, Proceedings of the
thirteenth Ain Shams Conference on geotechnical and Structural Engineering, 27-29
December 2009, Cairo, pp 30-39 .
PLAXIS Manual. (2002). Finite element code for soil and rock analysis published and
distributed by AA Balkema Publishers, Nederlands Comput. Geotech. 32(5):326
339.
Said, R. (ed.). (1990). The Geology of Egypt, Balkema, Rotterdam.
13
Seismic Protection of Monolithic Objects of
Art Using a Constrained Oscillating Base
Alessandro Contento and Angelo Di Egidio
Department of Structural, Hydraulic and Geotechnical Engineering,
University of LAquila
Italy
1. Introduction
The model of rigid block is well known in literature. In the past, several papers analyzed
the behaviour of rigid blocks under different kind of excitations because many monolithic
objects of art, such as statues, obelisks and fountains, subject to earthquake excitation, can be
modelled as rigid blocks. In [Shenton & Jones, 1991] a general bi-dimensional formulation of
the rigid block has been obtained and rocking and slide-rock approximated conditions have
been written. More recently this model has been used to describe the behaviour of
monolithic bodies subject to base excitations as a one-sine pulse excitation in [Zhang &
Makris, 2001; Makris & Black, 2004, Kounadis 2010] and earthquake excitation in [Agbabian
et al., 1988; Pompei et al, 1998; Taniguchi, 2002]. Almost all the papers on rigid blocks
subject to base excitation focus their attention on symmetric rigid bodies. Only a few papers
concern non-symmetric rigid bodies that, usually, represent objects of art better than
symmetric rigid blocks. In [Boroscheck & Romo, 2004] the influence of the eccentricity of
the centre of mass on the motion of the system has been studied. In [Purvance, 2005;
Purvance et al., 2008] an analytical and experimental estimation of overturning events
under seismic excitations has been carried out, both for symmetric and non-symmetric rigid
bodies. In particular, in [Zhang & Makris, 2001] for a one-sine pulse excitation and in
[Purvance, 2005; Purvance et al., 2008] for seismic excitation, the existence of survival
regions that lie above the PGA (Peak Ground Acceleration) associated with the first
overturning occurrence have been shown. In recent years, methods to reduce the effects of
seismic excitation on art objects have been studied in some papers. In [Fujita et al, 2008 ] a
critical excitation problem for a rigid block subjected to horizontal and vertical simultaneous
base inputs is considered. In [Vestroni & Di Cinto, 2000] a base isolation system has been
used to protect statues from seismic effects. The work of art has been modeled through an
equivalent elastic beam. In [Cali & Marletta, 2003] the same problem has been analyzed,
but the art object has been modeled as a symmetric rigid block simply supported on an
oscillating base connected to the ground by a visco-elastic device. The sliding of the body is
prevented by special seismic restraints. These analyses have shown the effectiveness of the
isolation system and the role of many parameters. To make things more realistic, in
[Contento & Di Egidio, 2009], the model presented in [Cali & Marletta, 2003] has been
enriched considering also the eccentricity of the centre of mass of the rigid body and the
presence of security stops, able to prevent the breaking of the isolation device by limiting

334
the displacement of the oscillating base to a maximum safety value. More recently, in [Di
Egidio & Contento, 2009; Di Egidio & Contento, 2010], they have introduced sliding effects,
that make the model able to carry out more complicated motions like slide-rocking, and
sliding constrains, that prevent the rigid body from falling off the base. Alternatively, the
possibility for the rigid block to be partially removed from the oscillating base has been
considered. Security stops to avoid damaging the base isolation system are also considered.
The behaviour of the whole system is studied under two types of excitations: impulsive and
seismic. Exact nonlinear equations of motion are written for the different phases of motion:
full-contact, sliding, rocking, slide-rocking; transition phase conditions are obtained by
generalizing to the case under analysis those obtained in [Taniguchi, 2002; Cali & Marletta,
2003; Contento & Di Egidio, 2009]. To describe the motion when the body is outside the
oscillating support, original equations of motion, describing the rocking and the slide-
rocking motions of the rigid block around a different point from one of the corners of its
base, are obtained. The influence of the friction coefficient, of the eccentricity, of the security
stops and of other parameters are analyzed by performing an extensive parametric analysis
via a direct numerical integration of the equations of motion. Comparison between results
obtained for isolated rigid body and non-isolated rigid body are also carried out to show the
effectiveness of base isolation with respect to the absence of this passive control system.
Another possibility, not considered here, is the use of three-dimensional models of rigid
blocks, mostly circular based. They are used to study the sloshing in circular shaped tanks
[Taniguchi, 2004], the wobbling motion [Stefanou et al., 2011] and the motion of a disk of
finite thickness on a planar surface [Koh & Mustafa, 1990; Batista 2006].
2. Description of the object considered and model hypotheses
The two-dimensional model presented is that of an isolated rigid body, where the isolation
system is an oscillating base connected to the ground by a linear visco-elastic device. The
geometric dimensions of the rigid body are taken with ratios similar to those of real works
of art and its characteristics are reported in Fig.1(a). With respect to the traditional models
presented in literature, here, also the eccentricity of the centre of mass of the rigid body is
considered: being C and O the centres of mass of the body and of the base respectively and
M the middle point of the base, the eccentricity e= b
1
-b is positive when the vertical
projection of the centre of mass C is shifted on the right side of M. In Fig.1(b) parameters
characterising the base isolation are shown, where c and k are the damping and the stiffness
of the linear visco-elastic device representing the base isolation system.
The rigid body here considered can undergo different kinds of motion: (a) full-contact
motion where the rigid body remains in contact with the isolated base when it oscillates, (b)
sliding motion where the rigid body slides on the oscillating base, (c) rocking motion where
the rigid body rocks around one of the two bottom corners without sliding and (d) slide-
rocking motion where a combination of rocking and sliding motion occurs (Fig.2). Three
Lagrangian parameters are used to describe the different phases of motion (Fig.2): u(t) that is
the translation of the oscillating base, x(t) that is the translation of the body due to slide
with respect to the isolated base and (t) that is the rotation of the body around one of the
bottom corners and can be positive or negative if the body rocks around the left or the right
corner respectively. Quantities u
g
and v
g
, shown in Fig.1(b), are the horizontal and vertical
ground displacements respectively. Since there are no constraints the rigid block can come
partially away from the base so that the rocking motion can occur not only around one of

Seismic Protection of Monolithic Objects of Art Using a Constrained Oscillating Base

335
the corners of the rigid body but also around one of the corners of the isolating base. For the
same reason one of the possible collapse condition of motion is the body falling off the base.
The other is the overturning of the body. In the model, frictional forces are expressed by
using the Coulomb description and the friction coefficient has been varied in the different
analyses carried out.
b b
u x
x x
x
u
max
max
b
1
b
2
c
h
k
r
1
r
1
r
1 r
2
r
2
r
2
O
O
C C
C
M M M
q q
cr1 cr2
e>0
1 1
1 2 2
2
(a) (b) (c)
u
v
G
G

Fig. 1. Mechanical system: (a) geometrical parameters of the rigid body; (b) parameters of
the base isolation system without security stops for the sliding motion; (c) security stops for
the sliding motion.
(c)
u>0 u>0
u>0 u>0
x>0 x>0
q>0 q>0
(d) (a) (b)
M M
M M
O O O O

Fig. 2. Phases of motion: (a) full-contact; (b) sliding; (c) rocking; (d) slide-rocking.
3. General formulation
In this section the equations of motion for the different phases of motion, the transition
conditions among them and the impact conditions are derived.
3.1 Equations of motion
Exact nonlinear equations of motion are written by using a Lagrangian approach for the
different phases of motion.
3.1.1 Full-contact motion
The full-contact motion equation is the linear equation of a single degree of freedom system
that represents the equilibrium of all forces acting on the system in the horizontal direction:

336

( )( ) 0 + + + + =
b G
m m u u cu ku
(1)
where m and m
b
are the mass of the rigid block and the mass of the isolated base,
respectively.
3.1.2 Sliding motion
During a pure sliding motion kinetic energy T and potential energy V of the system can be
written respectively:

2 2 2
1 1 1
; ( )
2 2 2
= + =
C B O O G
T mu m u V k u u
(2)
where
c
u and
o
u are the positions of the centres of the body C and of the base O expressed
as functions of two Lagrangian parameters (Fig.3(a)):

( ) ( ) ( ) ( ), ( ) ( ) ( ) = + + = +
C G O G
u t u t u t x t u t u t u t
(3)
Lagranges equations of motion are derived as follows:

( ) 0
( )( ) 0
+ + + =
+ + + + + =

g
b g
m u u x F
m m u u mx c u ku
(4)
where F is the Coulomb kinetic friction force given by:

( ) ( ) = +
k g
F sign x m g v
(5)
and where
k
is the kinetic friction coefficient and g the gravity acceleration.
3.1.3 Rocking motion
For the rocking motion, due to the fact that the model is non-symmetric, two different sets of
equations of motion are necessary, depending on which corner the motion occurs. When the
relative displacement

[ , ] e x x x , the base of the body is inside the oscillating support (see
Fig.1(b)). In this case the positions of the centres of the rigid body C and of the base O, when
a rocking around the corner 1 occurs, can be obtained by referring to Fig.3(b):

1 1
1
( ) ( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
' = + +
' = +
= +
C G
C G
O G
u t u t u t b b
v t v t h h
u t u t u t
(6)
Kinetic energy, potential energy and vitual work oW of the non-conservative forces then read:

2 2 2 2
2
1
1 1 1
( ) ;
2 2 2
1
( );
2
( ) .
0
o o
= + + +
' = +
=
C C b O C
T m u v m u I
V ku mg h h
W c u t u
(7)


337
Lagranges equations of motion read:

2
1 1
1 1 1
( )( ) 0
( ) ( ) 0
0 0
0
' ' + + + + =
' ' + + + =

b G
G G
m m u u cu ku m h m b
I m u u h m g v b
(8)
where

1 1 1 1

cos( ) sin( ); sin( ) cos( ) 0 0 0 0 ' ' = = + b b h h b h (9)
are the horizontal and vertical distances between centre of mass and corner 1 (that is the
centre of rotation) in the actual position, after rotation (see Fig.3(b)), and where
1 1
= b b (see
Fig.1(a)). In Eqs.(8)
2
1 1
:= +
C
I I mr is the polar inertia around corner 1;
C
I is the polar inertia
with respect to centre C of the body. The first equation of motion of Eqs.(8) represents the
equilibrium of the forces acting on the body together with the base in the horizontal
direction, while the second represents the equilibrium of the moments around the corner on
which the rotation occurs.
h h
u u u
h
h
b
b
b
b
1
1
1
1
1
1
(b) (c) (a)
C
C C
q -q
cr1
r
1
q
q
O O
O
M 1
M
M
x x
x<0
y
x
C
q -q
cr1
q
cr1
1
r
C

Fig. 3. Geometrical description of the different phases of motion: (a) sliding; (b) (slide-)
rocking around corner 1; (c) (slide-)rocking where the left corner of the body is outside the
oscillating base.
When the variable
[ , ] e x x x , the base of the rigid body is outside the oscillating support. In
this case the centre of rotation of the body is located in one of the corners of the oscillating
support (see Figs. 3(c)). The positions of the centres of the rigid body C and the base O,
when a rocking around the left corner of the support occurs (
> x x ), are equal to Eqs.(6)

where, however,
1 1
= + b b y with
= + y x x ( 0 < x not depending on time, see Fig.3(c)). By

using Eqs.(7) it is possible to obtain Lagranges equations of motion equal to Eqs.(8). In this
case
1
I is the polar inertia around the left corner of the support;
1
' b and
1
' h represent the
horizontal and vertical distances between the centre of mass and the left corner of the
oscillating support (and not the left corner of the rigid body as in the previous case, see
Fig.3(c)). They are still given by Eqs.(9). The equations of motion of rocking around the right
corner of the base or of the support can be obtained in a similar way (cases
[ , ] e x x x and
> x x respectively).

338
3.1.4 Slide-rocking motion
Also for the slide-rocking motion, due to the non-symmetry of the rigid body, two different
sets of equations of motion are necessary, depending on which corner the rotation occurs.
When during the motion the relative displacement

[ , ] e x x x , by evaluating kinetic and
potential energies and taking into account the virtual work of the generalized forces given
by Eq.(7), Lagranges equations of motion around the left corner can be found and read:

2
1 1 1
2
1 1
1 1 1
( ) 0
( )( ) 0
( ) ( ) 0
0 0
0 0
0
' ' + + + =
' ' + + + + + =
' ' + + + + =

G
b G
G G
m u u x m h m b F
m m u u mx cu ku m h m b
I m u u x h m g v b
(10)
where
1
' b and
1
' h , given by Eqs.(9), are the horizontal and vertical distances between the
centre of mass and corner 1 (that is the centre of rotation) in the actual position, after
rotation (see Fig.3(b)), and where
1 1
= b b . Eq.(10
1
)

represents the equilibrium in the
horizontal directions of all the forces acting on the rigid body (Fig.4(b)), while Eqs.(10
2
) and
(10
3
) have the same meaning of Eqs.(8). The Coulomb kinetic friction force
1
F is given by:

1 1
( ) =
k
F sign x Y (11)
where the vertical reaction
1
Y can be obtained by evaluating the vertical component of the
total force acting on the rigid body (Fig.4(b)):

2
1 1 1
( ) 0 0 ' ' = + +

G
Y m g v mb mh
(12)
The equations of motion of rocking around corner 2, in the case in which the relative
displacement

[ , ] e x x x can be obtained similarly.
When the variable

[ , ], e x x x the centre of rotation of the body is located in one of the
corners of the oscillating support (see Fig.3(c)). It is possible to obtain Lagranges equations
of motion for the left corner:

2
1 1
2
1 1
1 1 1 1
( )cos( ) ( )sin( ) 0
( )( ) cos( ) 2 sin( ) 0
( ) ( ) 2 0
0 0 0 0
0 0 0 0 0
0 0
+ + + + + =
' ' + + + + + =
' ' + + + + =

G G
b G
G G
m u u mx m g v m h m b F
m m u u mx cu ku m h m b mx
I m u u h mxh m g v b mx b

(13)

In this case
1
I is the polar inertia around the left corner of the support;
1
' b and
1
' h represent
the horizontal and vertical distances between the centre of mass C and the left corner of the
oscillating support (and not the left corner of the rigid body as in the previous case, see Fig.3
(c)). They are still given by Eqs.(9). It is interesting to observe that the first of Eqs.(13)
represents the equilibrium of the forces acting during the slide-rocking motion along the
slide direction parallel to the base of the body, when it is in a rotated configuration. The
interpretation of the other two equations of motion do not change with respect to the
previous case. A representation of the forces acting on the system during the slide-rocking
motion around the left corner of the support is reported in Fig.4(c). The kinetic friction force


339
1
F is still given by Eq.(11) where
1
Y represents the total force orthogonal to the inclined
slide direction and it can be obtained by the forces acting on the rigid body (Fig.4(c)):

2
1 1
( )cos( ) ( )sin( ) 2 0 0 0 0 0 = + + + +

G G
Y m g v m u u mx mb mh
(14)
It is interesting to observe that in this case the Coriolis force ( 2 ) 0

mx appears in the system
due to the fact that the centre of rotation of the rigid body changes its position with respect
the body during the motion.
(b)
q
Y
1
( )
G
m u u x + +
.. .. ..
(c)
q
( )
G
m g v +
..
.
2
1
mrq
..
1
mrq
..
C
I J
1
1
F
F
F
1
1
Y
1
( )
G
m g v +
..
.. ..
( )
G
m u u +
..
mx
.
' 2
1
mr q
.
.
2mxq
..
'
1
mr q
q
cr1
q
cr1
ku ku
ku
cu cu
. .
cu
.
O
( )
b G
m u u +
.. ..
( )
b G
m u u +
.. ..
..
C
I J
(a)
2
1
( )
G
m g v +
..
( )
G
m u u x + +
.. .. ..
C
( )
b G
m u u +
.. ..

Fig. 4. Forces acting on rigid body: (a) during a sliding motion; (b) during (slide-)rocking
around corner 1; (c) during (slide-)rocking where the left corner of the body is outside the
oscillating base.
3.2 Transition and termination conditions
Transition phase conditions are obtained by the equilibrium among forces or moments
acting on the rigid body during the motion. In the following only the case related to motions
around the left corner is presented while transition and termination conditions evaluated for
a motion occurring around the right corner can be found similarly.
3.2.1 Sliding motion
A pure sliding phase can take place starting from the rest or from the full-contact phase. By
referring to the forces acting on the system during a full-contact phase shown in Fig.4(a)
with 0 = x , a sliding phase can occur when the static frictional force ( ) = +
s s G
F m g v is
lesser than, or equal to, the absolute value of the horizontal inertial force ( ) +
G
m u u . This
condition reads:

( ) + s +
s G G
g v u u
(15)
where
s
is the static friction coefficient. Equation (15) refers to starting condition from the
rest when 0 u = at 0 t = .
It is assumed that the termination condition of a sliding phase corresponds to the vanishing
of the sliding velocity 0 x = .

340
3.2.2 Rocking motion
The pure rocking phase can take place from the rest or the full-contact phase. A rocking
phase can occur when the resisting moment
R
M due to the vertical accelerations ( )
G
m g v +
of the body is smaller than the overturning moment
O
M due to inertial force ( )
G
m u u + (see
Fig.4(a) with 0 x = ). Considering the moments around the left corner (of the body or of the
support), the following equation can be obtained:

1
( )
G G
b
u u g v
h
+ > + (16)
where
1 1
b b = when the relative displacement

[ , ] x x x e during the full-contact motion,
(see Fig.3(b)). On the contrary if the relative displacement
x x > , then
1 1
b b y = + with
y x x = + ( 0 x < , see Fig.3(c)). Equation (16) refers to a starting condition from the rest when
0 u = at 0 t = (and 0 x = ). There is a favoured direction for the beginning of the rocking
phase associated with the sign of the total acceleration
G
u u + . In some cases when
1
0 b = or
2
0 b = , which are the cases where the centre of mass is placed on the vertical projection of
the centre of rotation of the body, an unstable equilibrium position manifests itself.
No particular conditions are assumed to describe the termination condition of a rocking
phase. This means that the rocking phase finishes when the energy associated to this phase
is completely dissipated.
3.2.3 Slide-rocking motion
A slide-rocking motion can take place following the rest condition, the full-contact phase, the
pure sliding phase or the pure rocking phase. Depending on which phase it starts, different
starting conditions must be considered. First the starting condition of a rocking phase during a
sliding motion will be analyzed. By referring to forces acting on the system during a sliding
phase shown in Fig.4(a), a rocking phase can occur when the resisting moment
R
M due to the
vertical accelerations ( )
G
m g v + of the body is smaller than the overturning moment
O
M due
to inertial force ( )
G
m u u x + + . Considering the moments with respect to the left corner (of the
body or of the support), the following equation can be obtained:

1
( )
G G
b
u u x g v
h
+ + > +
(17)
where, if the displacement

[ , ] x x x e , then
1 1
b b = (se Fig.3(b)); if the relative displacement
x x > , then
1 1
b b y = + with
y x x = + ( 0 x < , see Fig.3(c)). There is a favoured direction for

the beginning of the rocking phase also in this case. The starting condition of a sliding phase
during a rocking motion can be evaluated by referring to the forces acting on the rigid body
during a rocking motion shown in Fig.4(b) with 0 x = . When the body is rocking around the
left corner (of the body or of the support), the starting condition for a slide can be easily
written as:

1 1 s
Y X s
(18)


341
where
1
X and
1
Y are the reactions of the base as in Fig.4(b),(c). If the displacement
[ , ] x x x e ,
force
1
Y is given by Eq.(12), while
1
X can be evaluated by computing the total horizontal
force acting on the rigid body. By referring to forces in Fig.4(b) (with 0 x = ), it reads:

2
1 1 1
( )
G
X m u u mh mb 0 0 ' ' = +

(19)
When
x x > , the force

1
Y is given by Eq.(14) with 0 x = , while
1
X can be evaluated by
computing the total force acting on the rigid body along the sliding direction. By referring to
forces in Fig.4(c) (with 0 x x = = ), it reads:

2
1 1
( )cos( ) ( )sin( )
G G
X m u u m g v mh mb 0 0 0 0 = + + +

(20)
Finally the starting condition of a slide-rocking phase from the full-contact phase is
analyzed. A slide-rocking motion can take place from this phase if pure sliding and rocking
conditions occur simultaneously. In particular a slide-rocking around the left corner (of the
body or of the support) occurs when Eq.(15) and Eq.(16) are simultaneously proven. A slide-
rocking phase takes place directly from the rest if the same previous equations, with 0 u =
at 0 t = , are simultaneously proven (
1 1
b b = ).
Also in this case, the termination condition of the sliding motion is taken to be 0 x = . When
this condition is satisfied only the rocking motion remains active. The termination of the
rocking motion is associated with the total dissipation of the energy associated to this phase;
in this case only the sliding motion remains.
3.3 Impact conditions
Both during pure rocking motion and during slide-rocking motion an impact among the
rigid block and the isolated base occurs when the angle 0 approaches zero. No bouncing
phenomenon is taken into account. Post impact quantities can be found, assuming that the
impact happens instantly and the body position remains unchanged, imposing the
conservation of the angular momentum and the conservation of the linear momentum
along the horizontal direction. Referring to an impact that occurs when the object
approaches the base by rocking around the left corner (of the body or of the support), the
angular momentum after the impact
2
I mhu 0
+ +
+
has to be equal to the angular momentum

before the impact
1 1 2 2
2 ( ) I mr b sin mhu 0 o 0

+

. This condition reads:

1 1 1 2
2 ( ) I mr bsin mhu I mhu 0 o 0 0
+ +
+ = +

(21)
where
2 2 2
1 1
r b h = + . The conservation of the linear momentum is needed to relate u

+
and
u
under the assumption that the impact (that happens instantly by hypothesis) does not
affect the sliding motion:
( ) ( )
b b
mh m u x m u mh m u x m u 0 0
+ + +
+ + + = + + +

(22)
If the relative displacement

[ , ] x x x e , then
1 1
b b = (see Fig.3(b)); if the relative

displacement
x x > , then
1 1
b b y = + with
y x x = + ( 0 x < , see Fig.3(c)). Taking Eq.(21) and


342
Eq.(22) into account, which means considering an impact that happens when the body is
approaching the base by rocking around the left corner and successively rocking around the
right corner, for a rectangular body, the following post-impact velocities are obtained:

2 2
1 1 1
2 2
2
( ) 2 sin( )( )
( )
b b
b
m h I m m mbr m m
m h I m m
o
0 0
+
+ + +
=
+

(23)
and

1 2 1 1
2 2
2
( 2 sin( )
( )
b
hm I I bmr
u u
h m I m m
o
0
+

= +
+

(24)
The maximum value of the coefficient of restitution that allows rocking motion of a block on
an isolated base then reads:

2 2
2 2
1 1 1
1
2 2
2
( ) 2 sin( )( )
( )
b b
b
m h I m m mbr m m
r
m h I m m
o 0
0
+
| | | |
+ + +
= =
| |
| |
+
\ . \ .
(25)
In Fig.5 some values of the restitution coefficient are shown. The dashed and dotted lines
represent r
1
obtained for 0.3 e b = and 0.3 e b = respectively, while the solid line represents
1
( 0) r r e = = as in [Vassiliou & Makris, 2011]. In the analyses the restitution coefficient has
been taken accordingly with Eq.(25).
l
2
1
r
J
J
+
-

=

&
&
0 e=
0.3 e b = -
0.3 e b =

Fig. 5. Restitution coefficient for different values of the eccentricity.
4. Description of the excitations
Two types of base excitations are considered in the following: impulsive excitation and
seismic excitation. Results are obtained in an analytical or in a numerical way depending on
the excitation considered. Several geometrical and mechanical parameters are introduced to
characterise the behaviour of the system. In particular:

b
k
m m
e =
+
;
2
T
t
e
= ;
2( )
b
c
m m
e
=
+
(26)


343
where e is the frequency, T is the period and is the damping ratio of the isolated system
in the full contact phase, that are well known quantities. Other quantities are introduced to
characterize the system:

e
b
c =
;
h
b
=
;
b
m
m
o =
(27)
where c is the eccentricity ratio with respect to the base of the rigid body, is the
slenderness of the body and o is the mass ratio.
4.1 Impulsive excitation
An horizontal impulsive ground excitation has been considered. The solution of the full-
contact, Eq.(1), under an impulsive ground acceleration I, can be found in closed form and
is well known, so, in this case, most of the results have been found analytically. Since the
system exhibits a symmetry in the rocking motion, conditions where there is a rocking
motion have been evaluated only referring to a rocking around corner 1 and for and
eccentricity 0 c < . Results are exposed trough maps describing the criteria for the different
phases of motion. These maps, firstly obtained for non isolated, symmetric rigid body
subjects to a horizontal ground acceleration in [Shenton, 1996], were successively extended
to vertical ground acceleration in [Tung, 2007]. Here they are extended to base isolated non-
symmetric rigid body subjects to a horizontal ground impulsive excitation and comparison
between not isolated and isolated systems is carried out.
4.2 Seismic excitation
In this case the system has been excited with two different Italian registered seismic ground
motions. Also for this kind of excitation only horizontal components of the seismic source
(a)
(b)
a/g
a/g
T ( ) s
0 0.5 1 1.5 2 2.5 3 3.5
Calitri
Lenght = 35.019 s
0 0.5 1 1.5 2 2.5 3 3.5
T ( ) s
Brienza
Brienza
a = 2.2 m s /
max
2
a = 1.2 m s /
max
2
Lenght = 34.012 s
t ( ) s
t ( ) s
a ( ) m/s
2
a ( ) m/s
2
0
0
5
5
10
10
15
15
20
20
25
25
30
30
35
35
-2.2
-1.2
-1.1
-0.6
0
0
1.1
0.6
2.2
1.2
0
0.1
0.2
0.3
0.4
0.5
0
0.2
0.4
0.6
0.8
1.0

Fig. 6. Time-history and response spectrum of the Italian seismic accelerations used in the
analysis: (a) Brienza; (b) Calitri.

344
( ) ( )
g
u t f t = are considered, where ( ) f t is the recorded seismic acceleration and is a
variable coefficient used to scale the maximum amplitude of the seismic accelerations (PGA:
Peak Ground Acceleration). In Fig.6 the seismic inputs used and the time-histories with their
elastic response spectrums are shown. The Brienza seismic source (Fig.6(a)) is used to analyze
the behaviour of the isolated rigid block and its dependence on several parameters, especially
on the friction coefficient; the Calitri seismic source (Fig.6(b)) is used to evaluate the influence
of the spectral characteristic of the earthquake in the dynamics of the isolated system.
4.3 Description of the simulation
Results are obtained via a direct numerical integration of the equations of motion using a
Runge-Kutta fourth order scheme. Special care is devoted to the choice of the integration
time step and appropriate procedures are followed to iteratively identify the transitions
among the different phases. Extensive parametric analyses are conducted to evaluate the
influence of the friction coefficient, the eccentricity and the slenderness as well as the period
of the isolated base on the behavior of the system. Analyses are performed by increasing the
PGA or the impulsive excitation to find the first occurrence of rocking, sliding and falling or
overturning of the body. Maps of behavior are produced to compare the results obtained
with similar systems.
5. Description of the results
Results have been obtained for two different cases, the first one where only the rocking
motion is allowed and a second one where the body is able to slip as well.
5.1 Introduction of safety devices
Two types of security stops have been introduced in the model considered and their effects
on the system investigated. The first is able to prevent isolation device breakage by limiting
the displacement of the oscillating base to a maximum safety value, the others are
introduced to prevent the rigid body from falling off the base.
5.1.1 Security stops for the base isolation
This kind of security stops is introduced to prevent the breaking of isolation devices that
generally cannot support displacements greater than a boundary value
max
u (see Fig.1(b)).
During the motion, if the velocity u is different from zero, an impact among oscillating base
and security stops occurs when:

max
u u = (28)
To evaluate the condition after an impact, it is assumed that: the impact happens instantly
and the body position remains unchanged; during the impact, a fraction of the horizontal
momentum is dissipated and after the impact there is an inversion in the direction of
motion. The horizontal momentum after the impact ( )
H b
Q m m u
+ +
= + becomes a fraction
i
q
of the horizontal momentum before the impact ( )
H b
Q m m u

= + . By letting
H H
Q Q q
+
= , it
is possible to obtain the post-impact velocity of the body:


345

i
u u q
+
= (29)
where
i
q is the restitution coefficient that can assume values in the range [0, 1].
5.1.2 Security stops for the slip of the body
Special sliding constrains that prevent the rigid body from falling off the base have been
introduced in the model as an alternative to the possibility of the rigid block coming
partially away from the base (Fig.1(c)). An impact among the body and the security stops,
during a slide or a slide-rocking motion occurs when:

x x = (30)
if the velocity x is different from zero. After the impact it is supposed that the horizontal
momentum due to slide motion mx
completely vanishes since it is partially dissipated and

partially transferred to the horizontal momentum of the oscillating base. In particular:
( ) ( )
b s b
m m u mx m m u q
+
+ + = + (31)
from which it is possible to obtain the post-impact velocity of the base:

s
b
m
u u x
m m
q
+
= +
+

(32)
where
s
q is the restitution coefficient that can assume values in the range [0, 1], under the
assumptions that the impact happens instantly.
5.2 Pure rocking motion
Analyses are performed to evaluate the first time at which rocking and overturning of the
body occur. This evaluation is done for several values of parameters c , T and . The
analyses here performed do not permit to obtain the so-called survival regions that lies
above the first overturning occurrence as in [Zhang & Makris, 2001; Purvance, 2005;
Purvance et al., 2008].
5.2.1 Impulsive excitation
The rocking condition around corner 1 can be obtained by equating the maximum
overturning moment
max
mh u , due to the inertia forces, with the resisting moment
( ) mg b e + , due to gravity, in which
max
u is the absolute maximum acceleration generated
by an impulse. This equation can be written, taking into account Eq.(27):

max
(1 )
g
u c
= +
(33)
The solution of the full-contact equation (1), under an impulsive ground acceleration I , is
well known and reads:

346

2
2
1
( ) sin( 1 )
1
t
I
u t e t
e
e
e

=
(34)
By solving a maximum problem, it is possible to obtain the absolute maximum acceleration:

max
2 ( )
( )
I a
u I a
T
t
e = = (35)
where the function ( ) a , which is always positive, reads:

2 2
2
2
1 1 4
Arctan
3 4
1
( ) 1 for 0
1
( ) for 0
2
a
a e

| |

|
\ .
= =
= < s
(36)
Starting rocking condition can be obtained by Eq.(16) (with 0
g g
u v = = ), taking into account
Eq.(35):

1
: (1 )
2 ( )
R
T g
I
a
c
t
= + C (37)
In the following, with solid lines overturning curves are indicated, while with dashed lines
rocking curves are indicated; moreover thin lines refer to an isolated rigid block, while thick
lines refer to a non-isolated rigid block. In Fig.7(a) minimum rocking and overturning
impulses versus the period of the isolated base T are plotted for a fixed damping . In grey
regions a rocking phase occurs. Below these regions only a full-contact phase is possible,
while above an overturning of the rigid block occurs. It is observed that when T is increased
the object rocks or overturns with higher impulse amplitude, however, the distance
between the rocking and the overturning curves decreases. It is also possible to observe that
for increased eccentricities the behaviour of the system worsens since smaller impulses are
able to cause the rocking or the overturning. Also the amplitude of the rocking regions
becomes smaller when the eccentricity is increased. In Fig.7(b) rocking and overturning
impulses versus the eccentricity are shown both for isolated and non-isolated systems.
Referring to the isolated system, also in this case, the grey area indicates the rocking phase
region. The best behaviour of the system is obtained when 0 c = (symmetric body); the
presence of an eccentricity reduces the performances since a smaller impulse is required to
cause rocking or overturning. When 1 c = (the centre of mass is located on a vertical side
of the body), since no resisting moment is present, the system is in an unstable equilibrium
position. By comparing rocking and overturning curves for isolated and non-isolated
systems, it is possible to observe the efficiency of the base isolation. It has to be highlighted
that the rocking curve for non-isolated system is coincident with the horizontal axis (the
minimum rocking impulse is always zero) because, given that an impulse corresponds to an
initial velocity, a non-isolated body has always to start its motion with an angular velocity
(0) 0 0 =
. Finally in Fig.7(c) rocking and overturning impulses versus the slenderness are
shown both for isolated and non-isolated systems. Curves are plotted starting by the value


347
of the slenderness 1 / 2 = above which no bouncing phenomena occur. For a fixed
eccentricity an increase of the slenderness causes an evident worsening of the behaviour. For
higher values of the eccentricity both isolated and non-isolated systems reduce their
performances. Grey region refers to rocking phase for 0.5 c = , both for isolated and non
isolated structure; an increasing of the eccentricity also causes a reduction in the dimensions
of the rocking region.
5.2.2 Seismic excitation
In Fig.8 comparison between the behaviour of an isolated rigid body with and without
security stops for the oscillating base is shown under Buia earthquake (security stops for the
sliding are not considered here). In the vertical left side of the graphs the scale factor is
reported, while in the vertical right side the PGA is reported.
In particular in Fig.8(a) rocking and overturning curves versus eccentricity c are drawn in
the case of absence of security stops (thick lines) and the case in which these safety devices
are present (thin lines). Dotted thick line refers to impact events: it gives the scaling factor
(or PGA) at which impacts on security stops occur for each value of eccentricity . c It is
possible to observe that after an impact rocking and overturning happens for a lower value
of the scaling factor with respect to the case in which security stops are not considered.
Time-histories of ( ) u t and ( ) t 0 show better what happens when an impact occurs
(Fig.8(b),(c)). Also in these figures thick lines denote the absence of security stops while thin
lines denote the presence of this safety device. For the case labelled with H in Fig.8(a), when
the displacement of the oscillating base ( ) u t reaches security stops (
max
( ) u t u = ) one or
more successive impacts can occur, as shown in Fig.8(b). The angle of rocking ( ) t 0 (Fig.8(c))
in general increases with respect to the case in which security stops are not considered and
causes the overturning of the rigid body for lower values of the scaling factor. Quantity
r
t refers to the time at which rocking phase starts, that is the instant at which Eq.(16) is
satisfied.
In the following, when safety devices are introduced, these characteristics are always
considered: 0.70, q =
max
0.20 u = . In Fig.9 comparison between isolated and non isolated
system is shown to better evaluate the efficiency of the isolation system under Buia
earthquake for a fixed damping and in presence of security stops. In particular in Fig.9(a)
rocking and overturning curves versus the period T are plotted for several eccentricity c .
Dots on -axis refer to the behaviour of the non isolated system. It is possible to observe that
increasing the level of the protection (that is increasing period T) scaling factors at which
rocking occurs increase with respect to the non isolated system, instead overturning scaling
factors do not monotonically grow with T due to the high nonlinearity of the system. For
example the minimum of the overturning curve that occurs in the range (0.5, 1.0) T e is
related to the high contents of spectral energy of the Buia earthquake in the same range. Out
of this range, the isolated system requires higher scaling factors to cause the overturning of
the rigid body with respect the non isolated system. The behaviour of the system
significantly decays when the rigid body has an eccentricity different to zero. The grey area
in the graph refers to the region in which a rocking phase occurs for 0.5 c = . Below this

348
0 1 2 3 4 5
0
1.0
2.0
3.0
I ( ) m/s
e= 0
T ( ) s
e= +0.5
e= +0.25
(a)
1 2 3 4 5 6 7
1
2
3
4
5
I ( ) m/s
e= 0
e= +0.5
(c)
-1 -0.5 0 0.5 1
0
0.50
1.00
1.50
I ( ) m/s
e (b)
0
l

Fig. 7. Dependence on eccentricity c : (a) Minimum absolute impulse versus period T
( 3 = ); (b) Minimum absolute impulse versus eccentricity c ( 3, = 2 T s = ); (c) Minimum
absolute impulse versus slenderness ( 2 T s = ); ( 0.3 , b m =
3
2000 / , kg m = 0.05 = ).
region a full-contact phase takes place; above this region an overturning of the rigid body
occurs. In Fig.9(b) rocking and overturning curves versus the eccentricity c are plotted. Thick
lines refer to the non isolated system, while thin lines to isolated system. Rocking and
overturning curves for isolated system are the same shown in Fig.8(a). It is possible to note
that the best behaviour of the system is obtained when 0 c = (symmetric body); the presence
of an eccentricity reduces the performances since a smaller impulse is required to cause
rocking or overturning. By comparing rocking and overturning curves for isolated and non-
isolated systems, it is possible to observe the efficiency of the base isolation. Also in this case
grey areas indicate the rocking phase regions. Below these areas there is no rocking, above
them there is a critical region in which overturning is not guaranteed everywhere because of
the existence of survival regions that lie above the curve associated with the first overturning
occurrence. Finally in Fig.9(c) rocking and overturning curves versus the slenderness , for a
fixed eccentricity ( 0.5 c = ), are plotted. Also in this figure thick lines refer to the non isolated
system, while thin lines to isolated system. Also in this case curves are plotted starting by the
value of the slenderness 1 / 2 = . By increasing the slenderness both the non isolated
and the isolated systems generally show a decay of the performances, but it is possible to
observe also in this case the greater efficiency of the base isolated system.
The thick dotted line is the impact curve. Also in this case grey areas indicate the rocking phase
regions. It is useful to summarize the effects due to the presence of security stops. In Fig.8(a) it is
possible to observe that a range of eccentricity ( 0.75 c < ), in which impact events have the
greater effects, exists. Into this range, when an impact occurs before the rocking in the system
without security stops, the impact and the rocking curves become practically coincident. This
means that the impact is the principal cause of rocking. Instead, when an impact occurs after the
rocking, the impact and the overturning curves become coincident. This means that the impact


349
is the main cause of overturning. This limit value of the absolute eccentricity seems to be
possessed by every systems and it depends on geometrical and mechanical characteristics of the
system itself. Also above a slenderness limit value ( 3) > it happens that impact curve and
overturning curve coincide, as shown in Fig.9(c). This slenderness limit value also depends on
geometrical and mechanical characteristics of the system.
-1 -0.5 0 0.5 1
0
0
3.25 7.5
6.50 15.0
9.75 22.5
13.00 30.0
e
g
t ( ) s
t ( ) s
t
r
u ( ) m
u
u
max
max
q ( ) rad
0 4 8 12 16 20
-0.3
0
0.3
0 4 8 12
-0.02
-0.01
0
0.01
0.02
(a)
(b)
(c)
H
PGA
( ) m/s
2

Fig. 8. Effects of the security stops under Buia earthquake: (a) Seismic scale factor versus
eccentricity c ( 0.3 , b m = 3, =
3
2000 / , kg m = 0.20, = 0.70, q =
max
0.20, u = 2 T s = );
(b) Time history of displacements ( ) u t for the case labelled with H in Fig.8a
( 0.3, 6.5 c = = ); (c) Time history of rotation ( ) t 0 for the case labelled with H .
-1 -0.5 0 0.5 1
0 0
4 9.2
8 18.4
12 27.6
e
(b)
g
1 2 3 4 5
0
10
20
30
l
e= 0.5
(c)
0 1 2 3 4 5
0
7.5
15.0
22.5
30.0
e= 0
T ( ) s
e= 0.5
(a)
g
g
0
0
3.25
3.25
6.50
6.50
9.75
9.75
13.00
13.00
PGA
( ) m/s
2
PGA
( ) m/s
2
PGA
( ) m/s
2

Fig. 9. Seismic analysis under Buia earthquake: (a) Seismic scale factor versus period T
( = 3 ); (b) Seismic scale factor versus eccentricity c ( 3, = T = 2 s ); (c) Seismic scale factor
versus slenderness (T = 2 s ); ( 0.3 , b m =
3
2000 / kg m = , = 0.20 ).
The last analysis reported here points out the effects of the spectral characteristics of the
earthquake. A simplified analysis is conducted by using Brienza and Calitri Italian
registered earthquakes (Fig.6(b),(c)) with comparable lengths both normalized by assuming
an acceleration peak equal to g. In Fig.10(a) rocking and overturning curves, obtained by

350
evaluating for which slenderness the body rocks or overturns by varying the eccentricity
c , are plotted. It is clear that under Brienza earthquake the system shows a better behaviour
since rocking and overturning happen for higher slenderness with respect to the Calitri
earthquake. Grey areas indicate the rocking phase regions. Below these areas there is full-
contact, above them there is overturning. In Fig.10(b) rocking (dashed lines) and
overturning (solid lines) curves, obtained by evaluating the level of protection (that is the
period T) to prevent rocking or overturning of the body by varying the eccentricity c , are
plotted. It is evident that the Brienza earthquake requires a lower level of protection with
respect to the Calitri earthquake, to prevent rocking or overturning. For Brienza input in the
range [ , ]
A B
c c c e , base isolation ( 0 T = ) is not necessary to prevent overturning. Also in this
case grey areas indicate the rocking phase regions but on the contrary, over these areas there
is full-contact, below them there is a critical region in which overturning is not guaranteed
everywhere because of the existence of survival regions. The isolated system shows a better
behaviour for earthquake with a more narrow spectrum as found in [Purvance et al., 2008].
-1 -0.5 0 0.5 1
0
3.5
7
10.5
14
-0.9 -0.45 0 0.45 0.9
0
A B
3.5
7
e e
e e
(b) (a)
l
T ( ) s
Calitri
Calitri
Brienza
Brienza

Fig. 10. Comparison between normalized Brienza and Calitri earthquakes (PGA=g): (a)
slenderness versus eccentricity c ( 0.3 , b m = 3, = 2 T s = ) and (b) period T versus
eccentricity c ( 0.3 , b m = 3, =
3
2000 / kg m = , 0.20 = ).
5.3 Slide-rocking motion
As for the rocking motion, analyses are performed to evaluate the first time at which slide,
rocking or slide-rocking and overturning or falling of the body occur. Results are found for
several values of parameters c , T and .
5.3.1 Impulsive excitation
A horizontal impulsive ground excitation is considered individually in the following. Maps
describing the criteria for the different phases of motion are obtained for base isolated non-
symmetric rigid body subjects to a horizontal ground impulsive excitation and comparison
between not isolated and isolated systems is carried out. These criteria maps are behaviour
maps that divides the parameter plane (
s
I ) in different regions where the motion starts
with a different phase of motion. Conditions where there is a rocking motion will be
evaluated only referring to a rocking around corner 1 and 0 c < . This is possible since, as
already pointed out, the system exhibits a symmetry in the rocking motion. Considering
sliding condition given by Eq.(15) (with 0
g g
u v = = ) and taking into account Eq.(35) and
Eq.(37), it is possible to obtain the starting condition of the sliding motion:


351
:
2 ( )
S s
T g
I
a
t
= C (38)
Starting rocking condition is still expressed by Eq.(37). By equating Eqs.(37, 38) it is possible
to obtain the curve where, during rocking motion, the starting condition of sliding motion
also becomes true:

1
: (1 )
SR s
c
= + C
(39)
When (1 ) /
s
c > + only the rocking motion is allowed (case considered in section 5.2). In
order to find the condition at which a slide-rocking motion occurs starting from a pure
rocking phase Eq.(18) must be taken into account. At the beginning of a rocking motion it
can be considered that 0 0 0 = =
and, by referring to Fig.3(a),

1 1 1
, b b h h ' ' = = . From Eq.(8
2
)
(with 0
g g
u v = = ) it is possible to obtain the maximum angular acceleration
max
0
:

1
max
1 1
2 ( ) ma h mgb
I
T I I
t
0 =
(40)
where use of Eq.(35) is done. The curve along which a sliding motion occurs starting to a
rocking phase can be obtained by Eq.(18), by taking into account Eq.(12), Eq.(19) and Eq.(40).
It reads:

2 2
(3 6 3 ) ( ) 3 (1 )
:
[3 (1 ) ( ) 4 ]
RS s
I a g
I a g
c c e c
c e
+ + + + +
=
+ +
C
(41)
where use of Eq.(27) is done. Previous curves divide the parameter plane in several regions
as shown in Fig.11(b)-(f). It is possible to observe that, when a base isolation is applied to the
rigid body, a full-contact region appears.
Full contact
Full contact
Slide
Slide
Slide
Slide
Slide-Rock
Slide-Rock
Slide-Rock
Slide-Rock
Rock
Rock
Rock
0
s s
s
s
s
6
1
m
I
0
6
1
m
I I
I
0
6
1
m
I
0 6
1
m
0
6
1
m
0
6
1
I
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
SR
SR
SR
SR
SR
SR
RS
RS
RS
RS
RS
S
S
S
S
S
R
R
R
R
R
Slide
Slide
Rock
Rock
Slide-Rock
Slide-Rock
Full
contact Full
contact
Full
contact
(a)
(b) (d) (f)
(c)
T s = 0
T s = 0.1
T s = 1.0
T s = 2.0
(e)
( ) m/s
( ) m/s
( ) m/s ( ) m/s
( ) m/s
( ) m/s

Fig. 11. Criteria maps: (a) not isolated system ( 3, = 0, c = 2 T s = ); (b) isolated system
( 3, = 0, c = 0.1 T s = ); (c) isolated system ( 3, = 0, c = 1 T s = ); (d) isolated system
( 3, = 0, c = 2 T s = ); (e) isolated system ( 5, = 0, c = 2 T s = ); (f) isolated system
( 3, = 0.5, c = 2 T s = ); ( 0.20 = ).

352
By considering the rocking motion particularly dangerous for the body, it is possible to
observe that by increasing the period T of the oscillating base, the regions where there is
rocking (grey regions) decrease (Fig.11(b)-(d)). By referring to the case where = 2 T s
(Fig.11(d)), both an increase of the slenderness (Fig.11(e)) and of the eccentricity c
(Fig.11(f)) cause a deterioration in the performance to the system, since a reduction of the
full-contact region and an increase of the rocking regions are observed.
5.3.2 Seismic excitation
In the numerical simulations performed here, 0.8
k s
= is always used as in [Shenton &
Jones, 1991]; according to results found in [Cali & Marletta, 2003] and [Contento & Di
Egidio, 2009], in this paper, it is always assumed 0.2. = Where the security stop collision is
concerned,
max
0.2 u m = ,
0.1 x m = and the restitution coefficients

,
0.7
i s
q = are always
chosen. In the following figures, these graphic conventions are always used: dashed curves
are related to sliding motion and in particular heavy dashed curves are the curves at which
a sliding motion begins, while thick dashed curves are the collapse curves due to the
precipitation of the rigid block from the oscillating base; solid curves are related to rocking
motion and in particular heavy solid curves are the curves at which rocking begins, while
thick solid curves are the collapse curves due to overturning of the rigid body. Finally heavy
dotted curves refer to the reaching of a slide
x x = above which the body is partially out of

the oscillating base, thick dotted curves refer to a collision with the slide security stops and
thick dash-dot curves refer to a collision with the base isolation security stops when they are
considered.
0 0.2 0.4 0.6 0.8 1
m
s
0 0.2 0.4 0.6 0.8 1
m
s
0 0.2 0.4 0.6 0.8 1
m
s
0 0.2 0.4 0.6 0.8 1
m
s
0
0.88
1.76
2.64
3.52
4.40
PGA
[ ] g
0
0.88
1.76
2.64
3.52
4.40
PGA
[ ] g
0
0.88
1.76
2.64
3.52
4.40
PGA
[ ] g
0
0.88
1.76
2.64
3.52
4.40
PGA
[ ] g
A
A
A B
B
A
(a)
(c)
(d)
(b)
Rocking
Impact 1
Impact 2
Falling
Sliding
Out base
Overturning
CR
CR
CR
CR
R
R
R
R
S-R
S-R
S-R
S-R
S
S
S
S
FC
FC
FC
FC

Fig. 12. Seismic analysis under Brienza earthquake: (a) PGA versus static friction coefficient
s
- no security stops; (b) PGA versus static friction coefficient
s
- only base stops; (c) PGA
versus static friction coefficient
s
- only sliding stops; (d) PGA versus static friction
coefficient
s
- both base and sliding stops; ( 3, = 0, c = 2 T s = ).


353
Results of the first analysis are shown in Fig.12 where the PGA associated with the first
occurrence of sliding, rocking, falling or overturning is plotted versus the static friction
coefficient
s
, under Brienza earthquake for a body with slenderness 3 = and eccentricity
0 c = . In each graph of Fig.13 it is possible to observe the existence of five different regions:
a region S where only sliding motion can occur, a region S-R where rocking or slide-rocking
motion can occur, a region R of pure rocking motions, a region FC where during the motion
the body and the base remain in full contact and a critical region CR. Here, as for the rocking
case, the survival regions could exist. At the labelled point A a collision between the sliding
curve and the overturning one manifests itself. While the sliding curve and the critical
falling curve strongly depend on the friction coefficient
s
, it is very interesting to note that
the rocking and the overturning curve are practically independent from
s
above the point
A. As observed in [Di Egidio & Contento, 2009] in this region, where the behaviour of the
system does not depend on the friction, only a pure rocking motion is possible. Differently
from the results obtained in [Di Egidio & Contento, 2009] where the body could not exit
from the oscillating base, here, also for small values of the static friction coefficient, a
rocking motion manifests itself. This is possible since, when the body is out the base (this
happens over the curve labelled Out base), it can rock easily because the resisting moment
becomes smaller and smaller as the body comes out the base (see Figs. 12(a),(b)). The effects
of the presence of the security stops on the behaviour of the system are very interesting.
Figure 12(a) refers to the absence of these stops. When security stops on oscillating base are
considered (Fig.12(b)) a worsening of the performance of the system in the R region can be
observed, because rocking and overturning can occur for smaller values of the PGA. On the
contrary, around the conjunction between the falling and the overturning curves it is
possible to observe an improvement of the performance of the system. It is useful to note for
next comments, that the impact on the security stops (above the curve labelled Impact 1)
happens before a rocking motion manifests itself. The introduction of security stops on the
sliding motion (Fig.12(c), curve labelled Impact 2) change the falling collapse events in
overturning collapse events. Since below the point B rocking and overturing curves
coincide, rocking and overturning events happen for the same PGA and no S-R region
exists. However the presence of these kind of security stops, do not change significantly the
values of PGA at which a collapse can occur. Finally in Fig.12(d) results obtained from the
contemporary presence of the two kind of security stops are shown.
The effects of the eccentricity and specially of the slenderness on the performance of the
system are shown in Fig.13. In Figs. 13(a)-(d) the results obtained for a body with
slenderness 3 = and eccentricity 0.35 c = are reported. First of all the worsening of the
behaviour of the system due to an increasing of the eccentricity with respect to the results
shown in Fig.12 must be observed, since smaller PGA are now able to cause falling and
overturning collapse events. Another interesting aspect is related to the presence of security
stops on the oscillating base (Fig.13(b)). Due to the increased eccentricity, the rocking starts
before a collapse on security stops happens. This fact makes the system less sensitive to the
presence of this kind of security stops compared to the case analyzed previously and shown
in Fig.12. The presence of the security stops on the sliding motion (Fig.13(c)) or the
contemporary presence of the two kind of security stops (Fig.13(d)) have, in this case, a

354
small impact on the performance of the system. On the contrary, in Figs. 13(e)-(h), results
obtained for a lesser slenderness 2 = (and the same eccentricity 0.35 c = ) are shown. First
of all it must be observed that a decreasing of the slenderness causes a reduction of the
region R where a pure rocking motion manifest itself, since the sliding motion becomes
possible for higher values of the friction coefficient. By comparing results obtained without
security stops (Fig.13(e)) and with only the security stops on the oscillating base (Fig.13(f)), it
is possible to observe the great positive influence of this type of stops specially on the
overturning collapse condition. The great sensitivity to the presence of these security stops
is related to the fact that the impact on them happens before the rocking motion starts as
observed also in the previous case.
0 0.2 0.4 0.6 0.8 1
m
s
0
0.88
1.76
2.64
3.52
4.40
PGA
[ ] g
0
0.88
1.76
2.64
3.52
4.40
PGA
[ ] g
0
0.88
1.76
2.64
3.52
4.40
PGA
[ ] g
0
0.88
1.76
2.64
3.52
4.40
PGA
[ ] g
(a)
(b)
(c)
(d)
CR
CR
CR
CR
S-R
S-R
S-R
S-R
S
S
S
S
FC
FC
FC
FC
A
A
A
A
B
B
R
R
R
R
0 0.2 0.4 0.6 0.8 1
m
s
0
2.20
4.40
6.60
PGA
[ ] g
0
2.20
4.40
6.60
PGA
[ ] g
(e)
(f)
(g)
(h)
CR
CR
CR
CR
CR
S-R
S-R
S-R
S-R
S
S
S
S
FC
FC
FC
FC
A
A
A
A
B
B
B
R
R
R
R
0
2.20
4.40
6.60
PGA
[ ] g
0
2.20
4.40
6.60
PGA
[ ] g
*
*
*
*
I2
I1
I3
I4

Fig. 13. Seismic analysis under Brienza earthquake: (a)-(d) PGA versus static friction
coefficient
s
( 3 = ); (a) no security stops; (b) only base stops; (c) only sliding stops; (d)
both base and sliding stops. (e)-(h) PGA versus static friction coefficient
s
( 2 = ); (e) no
security stops; (f) only base stops; (g) only sliding stops; (h) both base and sliding stops;
( 0.35, c = 2 T s = ).


355
Unlike the case with 3 = (shown in Fig.12(b)), where security stops on an oscillating base
cause a worsening of the performance of the system, for smaller slenderness they improve
the behaviour of the system. The presence of security stops on sliding motion have a small
impact on the behaviour of the system as shown in Fig.13(g). Finally the contemporary
presence of the two types of security stops causes a general improvement of the
performance of the system compared to their total absence. To conclude, the comparison
between the case with 3 = (Figs. 13(a)-(d)) and 2 = (Figs. 13(e)-(h)) shows that the
performances of the two system are similar as to the falling collapse conditions (when no
security stops on sliding are considered), while in the overturning conditions the behaviour
of the system with lesser slenderness is much better than the other one.
In Fig.14 some time-histories of different cases contained in different regions are shown. A case
of pure slide motion in the presence of security stops is shown in Fig.14(a) for a point labelled
with I1 in Fig.13(d). It is possible to observe the impossibility of the system to overcome the
threshold value
x . When no sliding security stops are considered, the body can exit from the
base as shown in Fig.14(b) for a point labelled I2 in Fig.13(f). In Fig.14(c) a case contained in the
S-R region in presence of only security stops on the oscillation base is shown (point labelled I3
in Fig.13(f)). It is possible to observe the impacts of the oscillating base on the stops from the
time-history of the displacement u that never exceed the threshold value
max
u . From the
time-histories of x and 0 two different slide-rocking motions can be observed. Below the
time
SR
t , as the body is inside the support, rocking or slide-rocking motion around the corners
of the of the body manifest themselves; above the time
SR
t , as the body is outside the support,
rocking and slide-rocking motion happen around the corners of the support. In particular, in
this last case the rocking motion happens around the left corner of the support since
x x >
and the right corner of the body. In Fig.14(d) the time-histories of a pure rocking motion are
finally shown (point labelled I4 in Fig.13(f)).
(b) (a)
u [ ] m
x [ ] m
q[rad]
-u -u
u u
max max
max max
(c) (d)
-0.25
0
0.25
-0.25
0
0.25
-0.25
0
0.25
-0.25
0
0.25
0 5 10 15 20 25 30 35
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
0 5 10 15 20 25 30 35
t
t
SR
SR
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
0 5 10 15 20 25 30 35
-0.1
0
0.25
-0.03
0
0.06
-0.3
0
0.15
-0.006
0
0.006
-0.15
0
0.15
x
x
x
-x
-x
-x
t [ ] s t [ ] s t [ ] s t [ ] s

Fig. 14. Time-histories under Brienza earthquake: (a) system marked with I1 in Fig.13(d); (b)
system marked with I2 in Fig.13(f); (c) system marked with I3 in Fig.13(f); (d) system
marked with I4 in Fig.13(f).

356
In Figs. 15 results obtained for a body with slenderness 1 = and eccentricity 0.35 c = are
shown. It is very interesting to observe that in this case, the collapse of the system, also
when no security stops are considered (Fig.15(a)), is related only to the overturning of the
body. This fact can be explained by observing the great attitude of a body with less
slenderness to slide instead of rocking. Since these kinds of bodies easily exit from the limits
of the support by sliding and can easily reach large sliding displacement x, when a rocking
motion starts, a overturning condition follows due to the small value of the resisting
moment. When the two different security stops are considered a great improvement of the
performance of the system is observed. This fact confirms the positive effects of the stops to
system with a smaller slenderness. Above the points labelled with C rocking and sliding
curves coincide. Finally it can be noted that no pure rocking region exists.
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
0
4.40
8.80
13.20
17.60
0
4.40
8.80
13.20
17.60
m
s m
s
PGA
[ ] g
(a) (b)
C
C
B
B
CR
CR
R
S-R
S-R
S
S
FC
FC

Fig. 15. Seismic analysis under Brienza earthquake, PGA vs static friction coefficient
s
:(a)
no security stops; (b) base and sliding stops; ( 1, = 0.35, c = 2 T s = ).
6. Conclusion
A model of a non-symmetric rigid body, where the centre of gravity is not equally distant
from the two base corners can be used to investigate the behaviour of a real base isolated art
object. The base under the rigid body is connected to the ground by a linear visco-elastic
device representing the passive control system. Exact nonlinear equations of motion have
been written by using a Lagrangian approach and transition conditions have also been
derived. In the case in which no sliding security stops are considered original equations of
motion describing the slide-rocking motion of the rigid block when it is partially outside the
support have been obtained. For the impulsive excitation exact and approximated (for
damped systems) results have been obtained in closed form while seismic excitation has
been considered by using two Italian registered earthquakes. Two different kinds of collapse
condition have been considered: the fall from the oscillating base of the rigid block and the
overturning of the body. The analysis has been conducted with the aim of pointing out the
effects of the friction coefficient, the influence of the slenderness and of the eccentricity of
the body and the analysis confirms that base isolation can be more effective for rigid bodies
with geometrical parameters similar to those of real works of art. Security stops have been
considered one kind able to prevent isolation device breakage by limiting the displacement
of the oscillating base to a maximum safety value, another introduced to prevent the rigid
body from falling off the base. For wider bodies security stops turn out to have a positive
influence on the performance of the system while above a certain value of slenderness they
can worsen the behaviour of the system.


357
7. References
Agbabian M.S., Masri F.S., Nigbor R.L. & Ginel W.S. (1988). Seismic damage mitigation
concepts for art objects in museum, Proceeding of the 9th World Conference on
Earthquake Engineering, No.7, pp. 235-240, Tokyo-Kyoto, Japan, 1988
Batista M. (2006). Steady motion of a rigid disc of finite thickness on a horizontal plane.
Journal of Nonlinear Mechanics, Vol.41, pp.850859
Boroscheck R.L. & Romo D. (2004). Overturning criteria for non-anchored non-symmetric
rigid bodies, Proceeding of the 13th World Conference on Earthquake Engineering,
Vancouver, B.C., Canada, August 1-6, 2004
Cali I. & Marletta M. Passive control of the seismic response of art objects. Engineering
Structures, Vol.25, pp. 1009-1018
Contento A. & Di Egidio A. (2009). Investigations into Benefits of Base Isolation for Non-
Symmetric Rigid Blocks. Earthquake Engineering and Structural Dynamics, Vol.38, pp.
849-866
Di Egidio A. & Contento A. (2009). Base Isolation of Sliding-Rocking Non-Symmetyric Rigid
Blocks Subjected to Impulsive and Seismic Excitations. Engineering Structures,
Vol.31, pp. 2723-2734
Di Egidio A. & Contento A. (2010). Seismic response of a non-symmetric rigid block on a
constrained oscillating base. Engineering Structures, Vol.32, pp. 3028-3039
Fujita K., Yoshitomi S., Tsuji M. & Takewaki I. (2008). Critical cross-correlation function of
horizontal and vertical ground motions for uplift of rigid block. Engineering
Structures, Vol.30, No.5, pp.1199-1213
Koh A.S. & Mustafa G. (1990). Free rocking of cylindrical structures. Journal of Engineering
Mechanics, Vol.116, pp.3454
Kounadis A.N. (2010). On the overturning instability of a rectangular rigid block under
ground excitation. The Open Mechanics Journal, Vol.4, pp. 4357
Makris N. & Black C.J. (2004). Dimensional analysis of bilinear oscillators under pulse-type
Excitations. Journal of Engineering Mechanics (ASCE), Vol.130, No.9, pp. 1019-1031
Pompei A., Scalia A. & Sumbatyan M. (1998). Dynamics of rigid block due to horizontal
ground motion. Journal of Engineering Mechanics, Vol.124, No.7, pp. 713-717
Purvance M.D. (2005). Overturning of slender blocks: numerical investigation and
application to precariously balanced rocks in southern California. Ph.D.
Dissertation, University of Nevada, Reno, Available from
http://www.seismo.unr.edu/PrecRock/gradresearch.html
Purvance M.D., Anooshehpoor A. & Brune J.N. (2008). Freestanding block overturning
fragilities: Numerical simulation and experimental validation. Earthquake
Engineering and Structural Dynamics, Vol.37, No.5, pp. 791-808
Shenton H.W. III. (1996). Criteria for initiation of slide, rock, and slide-rock rigid-body
modes. Journal of Engineering Mechanics ASCE, Vol.122, No.7, pp. 690-693
Shenton H.W. & Jones N.P. (1991). Base excitation of rigid bodies. I: Formulation. Journal of
Engineering Mechanics, Vol.117, No.10, pp. 2286-306
Stefanou I., Vardoulakis I. & Mavraganis A. (2011). Dynamic motion of a conical frustum
over a rough horizontal plane. International Journal of Non Linear Mechanics, Vol.46,
pp.114124

358
Taniguchi T. (2002). Non-linear response analyses of rectangular rigid bodies subjected to
horizontal and vertical ground motion. Earthquake Engineering and Structural
Dynamics, Vol.31, pp. 1481-1500
Taniguchi T. (2004). Rocking behavior of unanchored flat-bottom cylindrical shell tanks
under action of horizontal base excitation. Engineering Structures, Vol.26, pp.415
426
Tung CC. (2007). Initiation of motion of a free-standing body to base excitation. Earthquake
Engineering and Structural Dynamics, Vol.36, pp. 1431-1439
Vassiliou M.F. & Makris N. (2011). Analysis of the rocking response of rigid blocks standing
free on a seismically isolated base. Earthquake Engineering and Structural Dynamics,
DOI: 10.1002/eqe.1124
Vestroni F. & Di Cinto S. (2000). Base isolation for seismic protection of statues, Proceeding of
the 12th World Conference on Earthquake Engineering, New Zeland, 2000.
Zhang J. & Makris N. (2001). Rocking Response of Free-Standing Blocks Under Cycloidal
Pulses. Journal of Engineering Mechanics, Vol.127, No.5, pp. 473-483
14
Application of a Highly Reduced One-
Dimensional Spring-Dashpot System
to Inelastic SSI Systems Subjected
to Earthquake Ground Motions
Masato Saitoh
Saitama University, Shimo-Okubo Sakura-Ku Saitama, Saitama
Japan
1. Introduction
Recent devastating earthquakes in many countries, particularly the 2010 Haiti earthquake
(e.g., Eberhard et al., 2010) and the 2011 off the Pacific coast of Tohoku earthquake (e.g.,
Takewaki et al., 2011) in Japan, have caused severe damages to buildings and structures.
The past and the recent seismic events have led us to attempt to improve technologies used
in practical applications for evaluating the dynamic behavior of structural systems subjected
to earthquake ground motions. To accomplish this, it is considered that an appropriate
representation of soil-structure interaction (SSI) effects may be of great importance in
earthquake engineering.
SSI has been studied since the late 19th century. Since then, developments in SSI science
over the years have resulted in the latest SSI technologies (Kausel, 2010). In recent years, the
effects of SSI have been taken into consideration in various practical numerical
computations. For analyzing the dynamic response of SSI systems, a substructure method is
often used for performing more efficient computations with lesser degrees of freedom
(DOFs) rather than more comprehensive models with extremely large number of DOFs. In a
substructure method, impedance functions (IFs) are generally used to represent the dynamic
stiffness and damping of soil-foundation systems. Most IFs of soil-foundation systems
exhibit various frequency-dependent characteristics; they usually occur as a result of the
reflection and refraction of traveling waves originating from the foundations. Numerous
studies associated with the frequency-dependent characteristics of IFs have been conducted
over the past several years. IFs exhibit the following typical frequency-dependent
characteristics: (a) slight oscillation shown in soil reaction and surface rigid foundations or
embedded rigid foundations (Baranov, 1967; Beredugo & Novak, 1972; Novak, 1974; Novak
et al., 1978; Veletsos & Dotson, 1988; Gazetas, 1991; Saitoh & Watanabe, 2004; Tileylioglu et
al., 2011); (b) multiple oscillations typically exhibited in pile groups (Kaynia & Kausel, 1982;
Dobry & Gazetas, 1988; Makris & Gazetas, 1993; Mylonakis & Gazetas, 1998); and (c) a cut-
off frequency below which damping is negligible and above which damping increases
rapidly (Novak & Nogami, 1977; Nogami & Novak, 1977; Kausel & Roesset, 1975; Elsabee &

360
Morray, 1977; Takemiya & Yamada, 1981). These studies have indicated that considering
frequency dependency in the response analysis of structures can yield more accurate
calculations in SSI systems.
In contrast to the findings of many of the latest studies of various foundations under diverse
conditions and in recent numerical computations, frequency dependency in SSI systems has
been recognized as a hindrance. In accordance with the recent performance-based seismic
design of structures, nonlinearity in structural members, such as cracking, yielding, and
collapse, needs to be considered in computations in order to predict the inelastic response of
structural systems subjected to the expected earthquake waves. In calculating the response,
step-by-step numerical procedures in the time domain are usually used with constitutive
models, because the nonlinearity of structural members strongly depends upon the stress
pass being integrated stepwise. Therefore, a conventional method for considering frequency
dependency in the frequency domain cannot be applied. This has been one of the most
important problems in structural engineering and geotechnical engineering over the past
few decades.
Recently, various methods that are ready to use in practice have been proposed to overcome
this issue, when frequency dependency in IFs and nonlinearity in structures are
simultaneously taken into account. Basically, there are two main streams of thought in
dealing with this problem. One is to use a method for transforming IFs into impulse
responses in the time domain. The origins of this method date back to Wolf & Obernhuber,
1985, who developed a numerical method in the time domain in which the impulse response
obtained from dynamic soil stiffness by using inverse Fourier transform was applied to the
response analysis. Since this method was first developed in the 1980s, many transform
methods in the time domain have been proposed and improved to overcome difficulties in
various frequency dependencies in IFs (Wolf & Motosaka, 1989; Meek, 1990; Motosaka &
Nagano, 1992; Hayashi & Katsukura, 1990). Nakamura (2006a; 2006b; 2008a; 2008b) has
developed various sophisticated transform methods that can deal with strong frequency
dependency in IFs, non-causal impedance with large hysteretic damping, and soil
nonlinearity.
The other method for dealing with the SSI problem is the use of a lumped parameter model
(LPM), which is considered to be a very powerful tool for solving this problem. In general,
an LPM consists of springs, dashpots, and masses having frequency-independent
coefficients. A particular combination of these elements can simulate a frequency-dependent
impedance characteristic. The advantage of LPMs is that they can be easily incorporated into
a conventional numerical analysis in the time domain, even under nonlinear conditions of
superstructures. Time stepping methods that have conventionally been applied to structural
analyses, such as the central difference method, Newmarks method, and Wilsons
methodemployed as conventional integration schemescan be used with LPMs, while
time-domain transform methods usually need a specific scheme to incorporate the impulse
response into the response analysis.
From the viewpoint of construction schemes in LPMs, the existing LPMs can be
categorized into three types: a) semi-empirical LPMs, b) systematic LPMs, and c) modal
LPMs.
Application of a Highly Reduced One-Dimensional Spring-Dashpot
System to Inelastic SSI Systems Subjected to Earthquake Ground Motions

361

p
u
p
u
p
u
p
u
p
u
(a)
(c)
(b)
(d)
(e)
(f)
(g)
p
u
p
u
Gyro-mass

Fig. 1. Semi-empirical lumped parameter models proposed in previous studies [(a) Meek &
Veletsos, 1974; Wolf & Somaini, 1986; (b) Nogami & Konagai, 1986; 1988; (c) de Barros &
Luco, 1990; (d) Jean et al., 1990; (e) Wu & Chen, 2001; (f) Taherzadeh et al., 2009; and (g)
Saitoh, 2011a].
In semi-empirical LPMs, mechanical elements such as springs, dashpots, and masses, which
have frequency-independent coefficients, are arranged appropriately, depending on how
significantly an optimal fit of the IFs obtained using LPMs with the corresponding target
(exact) IFs can be achieved using specific arrangements. The values of the elements are
usually determined by minimizing the errors from the target IFs. Meek & Veletsos, 1974,
initially presented an LPM that represents the lateral and rocking impedance characteristics
for a truncated semi-infinite cone that can be considered as an analog of an elastic half space.
In an LPM, a mass has an additional DOF that is not directly attached to the foundation
node but rather is connected to it through a dashpot, as shown in Fig. 1a. Wolf & Somaini,
1986, extended the above model to a rigid disk, an embedded cylinder, a rectangle, and a
strip, and specified the coefficients of the elements comprising the foundations. Nogami &
Konagai, 1986; 1988, represented the subgrade reaction of soil surrounding single piles by
using three Voigt models connected in series (in the axial direction) and those with a mass
(in the flexural direction), as shown in Fig. 1b. Alternative LPMs (de Barros & Luco, 1990

362
(Fig. 1c); Jean et al., 1990 (Fig.1d)) were proposed for a better fit to the impedance functions
than the precedent LPMs. For a more concise usage of LPMs in practical applications, the
coefficients of simple LPMs have been specified in tables for rigid foundations on the
surface and with embedment in all translational and rotational motions (Wolf & Paronesso,
1992; Wolf, 1997). Wu & Chen, 2001, proposed an LPM consisting of a set of units in which
each unit consists of a mass connected to a spring and a dashpot arranged in parallel, as
shown in Fig. 1e. Wu & Chen, 2002, adopted the LPM to a simple SSI analysis for seismic
excitations. Recently, other forms of LPMs have been proposed by researchers (Taherzadeh
et al., 2009, (Fig. 1f); Khodabakhshi et al., 2011).
As described in Wolf, 1994; Wu & Chen, 2002; and Wu & Lee, 2002, however, the inclusion
of any mass in an LPM becomes a drawback when an LPM is straightforwardly used in
seismic excitations, that is, the driving forces at the foundation node induced by a
foundation input motion differ from the corresponding forces by applying the same input
motion at the bottom end (the far end) of an LPM because of the mass. This problem does
not occur when an LPM consists of springs and dashpots. However, the lack of mass in an
LPM tends to show difficulty in an appropriate fit to the exact impedance functions, and
thus, a substantial increase in the number of elements and DOFs is necessary.
To overcome this issue, Saitoh, 2007, proposed an LPM that uses a gyro-mass element
instead of an ordinary mass. The gyro-mass element generates a reaction force proportional
to the relative acceleration of the nodes between which it is placed. The use of the gyro-mass
element does not influence the driving forces acting at the foundation because the gyro-
mass element does not generate any inertia force. An LPM with a gyro-mass element
(GLPM) accomplishes a rapid change in frequency in IFs with a small number of elements
without an ordinary mass. In Saitohs study, two types of LPMsType I and Type II models
are presented for simulating IFs that have cut-off frequencies and frequency-dependent
oscillations, respectively. In Saitoh, 2011a, the accuracy in the GLPMs was verified using an
example of 2 4 pile groups embedded in a layered soil medium, supporting a 1DOF
system having inelasticity in the structural member when subjected to ground motions. In
his study, a more generalized GLPM, called the Type III model was proposed for a more
appropriate fit (Fig. 1g).
In the development of semi-empirical LPMs, attempts have been made to obtain an easier
and more systematic determination of the values for the coefficients in LPMs. A systematic
procedure was first proposed by Wolf, 1991a; 1991b. In the procedure, the dynamic-stiffness
coefficient is approximated as a ratio of two polynominals, which is then formulated as a
partial-fraction expansion whose each term is represented by a discrete model comprising
parallel-form LPMs, as shown in Fig. 2a. Wu & Lee, 2002, proposed series-form LPMs (Fig.
2b) formulated with a ratio of two polynominals approximating the flexibility functions
(compliance functions) instead of using IFs. Moreover, Wu & Lee, 2004, alternatively
proposed nested LPMs (Fig. 2c) based on a continued-fraction expansion instead of a
partial-fraction expansion. The advantage of nested LPMs is that the configuration is
independent of the soil-foundation systems being dealt with, whereas the previous series-
form LPMs, which use the partial-fraction expansion, depend upon them. Another new
systematic LPM (Fig. 2d) based on the continued-fraction expansion was proposed by Zhao
& Du, 2008.

363

p
u
p
u
p
u
p
u
(a)
(c)
(d)
(e)
p
u
(b)

Fig. 2. Systematic lumped parameter models [(a) Wolf, 1991a; 1991b; (b) Wu & Lee, 2002; (c)
Wu & Lee, 2004; and (d) Zhao & Du, 2008] and a modal LPM [(e) Saitoh, 2010a] proposed in
previous studies.
The aforementioned LPMs need to approximate the target IFs by using specific functions
such as the ratio of two polynominals in the case of systematic LPMs. Recently, two new
transform methods for constructing an exact LPM from the original systems have been
developed in the field of computational mechanics (Saitoh, 2010a; 2010b). In Saitoh, 2010a,
an LPM consisting of units arranged in seriesin which each unit consists of a spring, a
dashpot, and a gyro-mass element arranged in parallel (Fig. 2e)is formulated, i.e.,
calculated from a closed-form solution based on a modal expansion. In Saitohs study, an
LPM that represents the impedance characteristics at the extremity of a uniform, isotropic,
and homogeneous rod supported by continuously distributed springs and dashpots (the
KelvinVoigt model is assumed as a viscoelastic medium) was proposed. This method can
be applied to systems where the conventional modal expansion is available for solving the
differential equation of motion. In a later study (Saitoh, 2010b), a transformation procedure
based on a conventional complex modal analysis was proposed in which the impedance

364
function in general linearly elastic systems with non-classical damping is transformed into
an exact one-dimensional spring-dashpot system (1DSD) comprising units arranged in
series. Each unit, which is directly related to each vibrating mode of the original system, is a
parallel system consisting of a spring, a dashpot, and a unit having a spring and a dashpot
arranged in series. The properties of the elements comprising the 1DSDs are automatically
determined through the proposed procedure by using complex modal quantities. The
advantage of 1DSDs is that 1DSD transformation offers compatibility with the merit of
complex modal analysis, that is, a large number of units associated with high modes beyond
the target frequency region can be removed from the 1DSDs as an approximate expression
of impedance functions. Accordingly, a marked decrease in the computational domain size
and time with the use of the 1DSDs can be achieved. Extremely complicated frequency
dependency in impedance functions over a wide frequency range tends to appear in diverse
technological applications, as exhibited in Saitoh, 2010b. In such cases, the approximation of
the target IFs by using specific functions in the previous LPMs may encounter a certain
limitation in terms of accuracy and may not be accomplished sufficiently. On the other
hand, the 1DSDs transform procedure provides an exact LPM at the initial step, that is, we
can adjust the number of DOFs (the number of units) in the reduced LPM by achieving a
balance with the accuracy from the exact LPM. A transformation procedure for general
linearly elastic systems with classical damping was also proposed in his recent study
(Saitoh, 2011b).
Therefore, the main aim of this study is to verify the applicability of the transformation
method of 1DSDs to SSI problems influenced by seismic excitations. This study deals with
an application example of a four-story shear building supported by a shallow foundation
embedded in layered soil resting on a rigid bedrock. The soil-foundation system is modeled
using two-dimensional isoparametric finite elements, as shown in Fig. 4. In most previous
studies, nonlinearity in structural members was not considered when verifying the
performance of the proposed LPMs. In this study, therefore, the Clough model (Clough &
Johnson, 1966), which has a bilinear skeleton curve, is applied to each inter-story in the
building to compare the relationship between force and inter-storey drift obtained with
1DSDs with that obtained with the original finite element (FE) model.
2. Methodology for transforming structural systems with non-classical
damping into reduced 1DSDs
This section presents an overview of the method for transforming structural systems into a
1DSD. The configuration of a 1DSD is shown in Fig. 3. In general, the impedance function
( )
IJ
S e is defined as the ratio of the dynamic force
J
P applied at an arbitrarily selected DOF
(denoted as J-th DOF) and the response displacement
I
u at the same or another DOF
(denoted as I-th DOF). Here e is the excitation frequency. The properties of the elements
comprising a 1DSD are derived from a proposed procedure based on complex modal
analysis. In accordance with the detailed description in Saitoh, 2010b, the properties of the
elements are evaluated using the stiffness matrix, the damping matrix, the mass matrix, and
complex eigenvalues and eigenvectors. Once these values are obtained by a one-time
complex modal calculation, the properties are automatically determined from the
mathematical formula presented in Saitohs study.

365

J
p
I
u
2 T
c
2
k
2 T
k
2
c
1 T
c
1
k
1 T
k
1
c
N T
c
'
N
k
'
N T
k
'
N
c
'
j T
c
j T
k
Unit for Over-Damped Mode Unit for Under-Damped Mode

Fig. 3. One-dimensional lumped parameter model with spring and dashpot elements for
simulating the impedance function ( )
IJ J I
S p u e = in general structural systems [Saitoh,
2010b].
In this method, the equations of motion of general linearly elastic structural systems
comprising N DOFs are dealt with and are expressed by the following form:
| |{ } | |{ } | |{ } { } M u C u K u p + + =
(1)
where
| | M ,
| | C , and
| | K are the mass matrix, damping matrix, and stiffness matrix,
respectively, of the original structural systems. Each matrix has order N N . { } u and { } p
are the response displacements and the external forces at the nodes, respectively, and each
vector has order N . The dots denote partial derivatives with respect to time t . In this
study, the damping matrix | | C is assumed to be based on non-classical damping.
In complex modal analysis, the following 2N first-order equations are considered instead
of the N second-order equations of Eq. 1:
| |{ } | |{ } { } R z S z f + =
(2)
where
| |
| | | |
| | | | 0
C M
R
M
(
=
(

, | |
| | | |
| | | |
0
0
K
S
M
(
=
(

, { }
{ }
{ }
u
z
u

=
`

)
, { }
{ }
{ } 0
p
f

=
`

)
.
The complex eigenvalues and eigenvectors can be obtained according to the conventional
complex modal procedure. Each complex eigenvalue
n
is known to have an eigenvalue
n
, which is the complex conjugate of
n
; the corresponding vector { }
n
has a vector { }
n
,
whose components are complex conjugates of those of { }
n
. The eigenvectors are assembled
compactly into a matrix using diagonal matrices | | O and ( O

comprising the eigenvalues
n
and
n
, respectively, as
| |
| | | |
| || | | |

(
+ = (
( O O
(

(3a)
where

366
| | { } { } { }
1 2 N
( =

(3b)
| | { } { } { }
1 2 N
( =

(3c)
| | | |
n
diag O = , 1, 2, , n N = (3d)

n
diag ( ( O =

, 1, 2, , n N = (3e)
The matrix | | + is called the modal matrix. In general, | | | || |
T
R + + becomes a diagonal
matrix owing to the orthogonality relationships. Here the upper N components of the matrix
are denoted as
n
o , while the lower N components are the complex conjugates of
n
o ,
denoted as
n
o .
At the end of the mathematical derivation in Saitohs study, the properties of the elements
can be determined using the following formula:

( )
2 2
2
n dn
Tn
n n n dn
k
G R
o e
o e
+
=
(4)

1
2
Tn
n
c
G
= (5)

( )
( )
2 2 2
2
2
n n dn
n
n n n n dn
G R
k
G G R
e
o e
+
=
(6)

( )
( )
2 2 2
2
2
n n dn
n
n n n n dn
G R
c
G G R
e
o e
+
=
(7)
where
nI nJ
n n
n
G iR

o
+ =
and
nI nJ
n n
n
G iR

o
=
.
Here
nI
and
nJ
are the components of the n -th eigenvector at the I-th and J-th DOFs,
respectively;
nI
and
nJ
are the complex conjugates of the components
nI
and
nJ
,
respectively.
n
o is the n -th modal decay rate and
dn
e is the n -th damped natural circular
frequency defined as

n n dn
i o e = +
(8)

n n dn
i o e =
(9)
In practice, over-damped modes often appear. In this case, eigenvalues
n
are real and
negative. In Saitoh, 2010b, it was mathematically derived that the impedance function
associated with over-damped modes is expressed as a KelvinVoigt unit comprising spring
Tn
k and dashpot
Tn
c , as shown in Fig. 3.

367

n
Tn
n
k
G
o
= (10)

1
Tn
n
c
G
= (11)
Note that over-damped modes generally appear with even numbers 2m in 2N modes, so
the total unit number N changes to ( ) N N m ' = + when over-damped modes exist.
3. Application to soil-shallow foundation-structure system
3.1 FE model studied
This is the first application of a 1DSD to SSI problems. In principle, the dynamic response of
the original structural systems, which have even inelasticity in their superstructures, can be
correctly simulated using 1DSDs. However, the accuracy of the transfer functions and the
time histories of the dynamic response of the structural system comprising 1DSDs have
never been verified. In this section, therefore, a soil-shallow foundation system interacting
with a multiple-DOF system with inelasticity is used for verification.
The overall system is shown in Fig. 4. A shallow foundation 10 m wide, 50 m long, and 2 m
deep is embedded in layered soil up to the middle of the foundation. The elastic modulus of
the foundation is assumed to be rigid, imposing unique displacements
f
u and
f
u at the
center of gravity in the horizontal and rotational directions, respectively. The mass and the
mass moment of inertia of the foundation are 1000
f
m = t and 8500
f
J = tm
2
, respectively.
A four-story building supported by the foundation is represented by a 4-DOF system. The
properties of the system are shown in the figure. In this study, the inelasticity in each story
E3=200000kN/m
2

w2=17kN/m
3

E1=52000kN/m
2

Foundation
Building
50m
20m
w1=16kN/m
3
1=0.45
1=0.14
E2=68000kN/m
2

2=0.40
2=0.18
w3=18kN/m
3
3=0.35
3=0.15

Fig. 4. Two-dimensional FE model for a soil-shallow foundation system supporting a four-
story building. The unit weight, modulus of elasticity, Poissons ratio, and damping ratio of
the i-th soil layer are denoted as w
i
, E
i
,
i
, and
i
, respectively.

368
is taken into account. The soil-foundation system is modeled using conventional two-
dimensional rectangular isoparametric elements (Weaver et al., 1990), in which each element
has eight DOFs. The soil strata are composed of three soil layers resting on a rigid bedrock.
The bottom of the layered soil is fixed in the vertical and lateral directions, whereas a
viscous boundary proposed by Lysmer and Kuhlemeyer, 1969, is applied to the sidewalls of
the soil as a fictitious boundary that dissipates energy toward an infinite region of soil. The
moduli of elasticity and the damping ratios of the soil shown in the figure are assumed to
approximately account for the appreciable levels of strain during ground shaking. The total
number of nodes comprising the isoparametric elements is 560, whereas the degrees of
freedom minus the fixed degrees of freedom are 1,050. The thickness of the elements is the
same as the length of the foundation (50 m) under a plane-strain condition.
3.2 Equations of motion of FE model with structures
In this model, the soil-foundation system consists of conventional isotropic elements,
whereas the structural system comprising the superstructure and the mass of the foundation
is discretized by springs, dashpots, and masses. Therefore, the global mass matrix, stiffness
matrix, and damping matrix in the equations of motion of the total system are obtained by
superimposing local matrices in both equilibrium equations. The details of the equations are
described as follows.
The equations of motion of the soil-foundation system are expressed as
| |{ } | |{ } | |{ } { }
G G G G G G G
M u C u K u f + + =
(12)
where | |
G
M , | |
G
C , and | |
G
K are the mass matrix, damping matrix, and stiffness matrix,
respectively, of the soil-foundation system. { }
G
u and { }
G
f are the response displacements
and the forces at the DOFs in the system, respectively. In this study, the damping matrix for
each element is constructed on the basis of
| | | |
e l e
C K | = (13)
where

1
2
l
l
,
|
e
= (14)
where | |
e
C and | |
e
K are the local damping matrix and stiffness matrix, respectively, of the
isoparametric element; the parameter
l
, is the damping ratio given in each soil layer at the
fundamental natural frequency of the soil-foundation system; and
1
e is the fundamental
natural circular frequency of the undamped system. The damping ratio of each soil layer
used in this model is shown in Fig. 4. In the FE model, the elastic modulus of the elements in
the foundation is appreciably higher than that of the soil for realizing rigid body movement
of the foundation. In addition, the mass density of the elements is negligible because the
mass and the mass moment of inertia of the foundation are taken into account in the model
of the building.

369
The equilibrium equation of the 4-DOF system connected with the rigid mass of the
foundation is expressed as follows:
| |
{ }
| |
{ }
| |
{ } { } 0
0
0
s s s
I f I f I f
f f f
u u u
M u C u K u
u u u

+ + =
` ` ` `

)

) ) )

(15)
where
| |
1 1 1
2 2 2
3 3 3
4 4 4
4
2
1 1 2 2 3 3 4 4
1
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0 0
0
I
f
f i i
i
m m H
m m H
m m H
m m H
M
m
m H m H m H m H J m H
=
(
(
(
(
(
=
(
(
(
(
+ (
(

(16)
| |
1 1
1 1 2 2
2 2 3 3
3 3 4 4
4 4
0 0 0 0
0 0 0
0 0 0
0 0 0
0 0 0 0
0 0 0 0 0 0
I
c c
c c c c
c c c c
C
c c c c
c c
(
(
+
(
(
+
= (
+
(
(
(
(

(17)
| |
1 1
1 1 2 2
2 2 3 3
3 3 4 4
4 4
0 0 0 0
0 0 0
0 0 0
0 0 0
0 0 0 0
0 0 0 0 0 0
I
k k
k k k k
k k k k
K
k k k k
k k
(
(
+
(
(
+
(
=
(
+
(
(
(
(

(18)
where { } { }
( ) 1 2 3 4
T
s
u u u u u = are the absolute displacements of the DOFs in the
superstructure, excluding the displacements related to the rotational response of the
foundation, and
i
c and
i
k are the damping coefficient and the initial stiffness of each story
in the superstructure, respectively. Here the suffix i is numbered from top to bottom. The
damping matrix is constructed as Rayleigh damping by using the mass matrix (Eq. 16) and
the stiffness matrix (Eq. 18) with a first and second modal damping constant of 0.05.
In fact, the DOF of the foundation in the rotational direction is incompatible with the DOFs
in the FE model as no rotational DOF is explicitly considered at each node in the FE model.
Therefore, the rotational displacement
f
u in the rigid mass of the foundation is transformed

370
in the FE model based on the geometrical relationship using the relative displacement at the
center of the foundation to the displacement of the successive node located at the bottom of
the foundation. In addition, the rotational moment is transmitted to the corresponding
nodes in the FE model using an equivalent couple.
3.3 Transforming the original system into a 1DSD
According to the procedure shown in Section 3.1, the FE model is to be transformed into an
equivalent 1DSD. Complex modal analysis is performed to obtain the fundamental
quantities by which the properties of the elements in the 1DSD are determined. Complex
modal analysis has commonly been used for estimating modal components. In this study, a
conventional modal analysis (Foss, 1958) is used. First, the impedance function in the
horizontal direction is considered. In this case, although movement of the foundation in the
rotational direction may be restrained more accurately to obtain a net horizontal impedance
function (by extracting the interactive components), it was not restrained in this study
because the foundation is embedded in the soil only a little in the model. Thus, no difference
was observed in either of the impedance functions.
The results of complex modal analysis show that the number of over-damped modes is
2,026, whereas that of under-damped modes is 74 (the total number of modes is 2,100).
Therefore, the 1DSD consists of 2,026 KelvinVoigt units and 37 standard units. The results
show that the lowest natural frequency is 2.11 Hz, whereas the highest frequency is
4,760,000 Hz. As described above, a significant advantage of a 1DSD is that the units
comprising a 1DSD are associated with the vibration modes of the original structural
system. Therefore, a small set of units associated with modes from the lowest order can
appropriately express the dynamic characteristics of structural systems without using all
units. Recently, Saitoh, 2011a, studied the influence of frequency dependency in pile-group
impedance functions upon the elastic and inelastic responses of superstructures. The results
indicate that the important frequency range is the dominant frequency of foundation input
motions that excite the inertial structural systems. Figure 5 shows the time-history response
acceleration at the ground surface calculated using the conventional one-dimensional wave
propagation theory with the soil properties shown in Fig. 4. An observed earthquake record,
2004 Ojiya EW, is applied to the bottom soil layer. This response acceleration is used in the
following calculations as the foundation input motion in this study, which indicates that, for
simplicity, no adjustment for the kinematic interaction effects is conducted. The figure
shows that the foundation input motion contains a wide range of frequency components
showing the dominant frequency at around 1.4 Hz. The amplitude of acceleration ranges
from 0 to 10 Hz. Therefore, this frequency range is considered to be the target frequency
range in this study.
The modal analysis results show that 1,110 vibrating modes appear in this frequency range.
In fact, many units associated with these modes contain a relatively large spring constant
Tn
k than other units. These units can appreciably be removed provided the impedance
functions of the reduced 1DSD are in sufficient agreement with those of the original system.
In this study, only 35 units (12 under-damped modes and 23 over-damped modes) remain
after applying a threshold to the spring constant
Tn
k of
10
1.0 10 kN/m, above which the
corresponding unit is removed. This threshold is adjusted on the basis of both the number of

371
units remaining suitable for effective computation and the accuracy in the approximation of
impedance functions. The properties of the reduced 1DSD are shown in Table 1. Figure 6
shows a comparison of the impedance functions
hh
K obtained from the reduced 1DSD with
those obtained with the FE model. The results indicate that the impedance functions
obtained from the 1DSD agree closely with those obtained with the FE model within the
target frequency range.
0 2 4 6 8 10
0
5
10
15
20
25
Frequency (Hz)
F
o
u
r
i
e
r

A
m
p
l
i
t
u
d
e

(
m
/
s
2
-
s
)
0 10 20 30 40
-20
-10
0
10
20
A
c
c
.
(
m
/
s
2
)
Time (s)

Fig. 5. Fourier amplitude and time history of the foundation input motion.
The impedance functions
rr
K in the rotational direction are obtained along similar lines to
the impedance functions in the horizontal direction. To obtain the relationship between
rotational moment and displacement in the FE model, the center of the foundation is fixed in
the horizontal and vertical directions and lateral force is applied to the successive node at
the bottom of the foundation. The results of the complex modal analysis show that the
number of over-damped modes is 2,018, while that of under-damped modes is 78 (total
number of modes is 2,096). Therefore, the 1DSD consists of 2,018 KelvinVoigt units and 39
standard units. In this direction, 44 units (27 under-damped modes and 17 over-damped
modes) comprise the reduced 1DSD after applying a threshold to the spring constant
Tn
k in
the rotational direction of
10
7.0 10 kNm/rad. The properties of the reduced 1DSD are
shown in Table 2. The resultant impedance functions in the rotational direction at the center
of the foundation are shown in Fig. 6. Although a slight difference in the impedance
functions appear in a high-frequency region when compared with those with the FE model,
fairly close agreement can be observed over the target frequency range, on the whole.
0 1 2 3 4 5 6 7 8 9 10
0
2
4
6
8
Real Part (1DSD)
Imaginary Part (1DSD)
(x10
7
)
Real Part (FEmodel)
Imaginary Part (FEmodel)
K

h
h

(
k
N
/
m
)
Frequency (Hz)
0 1 2 3 4 5 6 7 8 9 10
0
2
4
6 Real Part (1DSD with residual stiffness)
Imaginary Part (1DSD with residual stiffness)
(x10
9
)
K

r
r

(
k
N
m
/
r
a
d
)
Frequency (Hz)
(a) (b)

Fig. 6. Impedance functions of a soil-shallow foundation system using 1DSDs and 1DSDs
with residual stiffness [(a)
hh
K in the horizontal direction and (b)
rr
K in the rotational
direction]. Results obtained from the original FE model are shown for comparison.

372
In Saitoh, 2010b, a mechanical element associated with residual stiffness
IJ
R was proposed
to improve the accuracy of the reduced 1DSD. Residual stiffness has often been applied to
approximate expressions of structural systems in conventional modal analysis. The residual
stiffness representing the stiffness of high-frequency modes is expressed as

1 2 1
1 1 1 1 1
N
IJ nIJ l IJ l IJ N IJ n l
R K K K K
'
' + + = +
= = + + +

(19)
where
For under-damped modes:
( )
2 2
2
n dn
nIJ
n n dn n
K
G R
o e
o e
+
=
(20)
For over-damped modes:
n
nIJ
n
K
G
o
= (21)
where l is the maximum mode number considered in the 1DSDs without residual stiffness.
The residual stiffness
IJ
R can be incorporated into the 1DSDs as a mechanical element
arranged in series with the 1DSDs. The residual stiffnesses in both horizontal and rotational

Mode (n) 1 2 3 4 5 6
kn -4.5692E+08 3.7591E+08 1.0618E+09 -5.1042E+08 -1.6506E+10 0.0000E+00
cn -6.6604E+06 2.1711E+07 7.9324E+07 -3.9891E+07 2.6594E+07 0.0000E+00
kTn 1.9143E+07 -4.2683E+08 -1.4162E+09 6.8531E+08 4.7634E+07 8.2805E+09
cTn 7.4465E+06 -1.1586E+07 -2.0610E+07 7.6590E+06 -2.5143E+07 2.0113E+08
7 8 9 10 11 12
kn 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
cn 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
kTn 1.9166E+09 2.9441E+09 6.5445E+09 8.8819E+09 1.9163E+09 6.5327E+09
cTn 4.6386E+07 7.1043E+07 1.5783E+08 2.1356E+08 4.5797E+07 1.5576E+08
13 14 15 16 17 18
kn 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
cn 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
kTn 1.1882E+09 5.3841E+09 4.9079E+09 1.7577E+09 8.5567E+09 4.9203E+09
cTn 2.8224E+07 1.2766E+08 1.1612E+08 4.1497E+07 2.0073E+08 1.1436E+08
19 20 21 22 23 24
kn 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
cn 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
kTn 7.8980E+09 7.1101E+08 6.3409E+09 3.0631E+09 2.2748E+09 1.7152E+09
cTn 1.7967E+08 1.6147E+07 1.4262E+08 6.8550E+07 5.0782E+07 3.7956E+07
25 26 27 28 29 30
kn 0.0000E+00 0.0000E+00 0.0000E+00 -1.8100E+10 0.0000E+00 -3.0252E+08
cn 0.0000E+00 0.0000E+00 0.0000E+00 3.9047E+08 0.0000E+00 -3.5556E+06
kTn 5.6713E+09 2.0277E+08 5.6121E+09 5.6507E+09 4.3924E+09 2.4506E+08
cTn 1.2154E+08 4.2003E+06 1.1373E+08 -1.0463E+08 8.4557E+07 7.5807E+06
31 32 33 34 35 Resi. Stif.
kn -9.4744E+09 -3.4928E+09 -4.8323E+09 -6.2970E+09 -1.1486E+10 0.0000E+00
cn 5.7840E+07 4.4528E+07 -2.1066E+07 3.2994E+07 -7.5059E+07 0.0000E+00
kTn 6.6702E+08 6.8297E+08 6.2645E+08 4.3951E+08 7.6216E+09 -6.6866E+08
cTn -3.5218E+07 -1.7234E+07 3.5920E+07 -1.8667E+07 2.2991E+08 0.0000E+00
*units: kN/m for kn and kTn; and kNsec/m for cn and cTn

Table 1. Properties of mechanical elements in reduced 1DSDs in the horizontal direction

373
directions are presented in Table 1 and Table 2, respectively. In Fig. 6, the impedance
functions of the residual 1DSDs with the residual stiffness are plotted. An appreciable
improvement can be seen in the high-frequency region when incorporating the residual
stiffness in both directions.
3.4 Dynamic response of structural system in frequency domain
In this section, the dynamic response of the structural system computed by using the
reduced 1DSDs in the frequency domain is verified by comparing it with the dynamic
response obtained with the original FE model. The complete structural system using the
1DSDs in both horizontal and rotational directions is shown in Fig. 7. The properties of the
superstructure are summarized in Table 3. The reduced 1DSDs obtained above are
connected with each DOF in the foundation, as shown in the figure. The equations of motion
of the structural system can be easily constructed using Eq. 15 with conventional spring-
dashpot matrices (details are described in Saitoh, 2010b) expressing the reduced 1DSDs. The
resultant equilibrium equations of the total system can be formulated as

Mode (n) 1 2 3 4 5 6
kn -8.7367E+10 -3.2511E+10 5.9885E+11 -8.9724E+10 -2.0509E+12 -8.2557E+10
cn -8.6002E+08 -3.2912E+08 -5.3876E+09 7.6112E+08 -2.2854E+09 8.3561E+08
kTn 6.3069E+09 3.8708E+09 -3.4692E+10 5.3955E+09 4.5267E+09 8.6083E+09
cTn 9.9889E+08 4.1663E+08 3.9650E+09 -5.4771E+08 2.4152E+09 -4.9286E+08
7 8 9 10 11 12
kn 0.0000E+00 -6.2547E+09 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
cn 0.0000E+00 -2.1604E+08 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
kTn 2.9016E+10 1.0142E+10 3.3727E+10 4.2553E+10 1.7739E+10 4.1057E+10
cTn 6.8559E+08 1.5672E+08 7.7893E+08 9.5936E+08 3.7922E+08 8.7675E+08
13 14 15 16 17 18
kn 0.0000E+00 -5.5850E+10 0.0000E+00 0.0000E+00 -1.1840E+12 3.6129E+09
cn 0.0000E+00 9.3275E+08 0.0000E+00 0.0000E+00 3.3843E+09 1.5195E+08
kTn 1.9758E+10 1.3439E+10 5.1907E+10 1.5146E+10 2.1723E+10 -6.6498E+09
cTn 4.0089E+08 -3.1988E+08 1.0150E+09 2.9275E+08 -2.7039E+09 -5.0781E+07
19 20 21 22 23 24
kn -1.0096E+09 -8.9520E+09 -3.2909E+09 -1.1530E+12 3.8877E+10 -7.9236E+08
cn -2.2987E+07 -1.0163E+08 -3.8159E+07 -3.6097E+09 6.5584E+09 -7.6230E+06
kTn 2.4822E+09 1.2331E+10 4.6047E+09 6.8295E+10 -4.5729E+10 2.1613E+09
cTn 3.2532E+07 3.0091E+08 1.0301E+08 5.3063E+09 -5.3463E+07 3.8351E+07
25 26 27 28 29 30
kn 0.0000E+00 0.0000E+00 0.0000E+00 -3.5501E+09 0.0000E+00 -1.8681E+09
cn 0.0000E+00 0.0000E+00 0.0000E+00 -6.3417E+06 0.0000E+00 -2.5699E+06
kTn -4.6466E+10 -3.5291E+10 -4.6258E+10 2.0957E+09 -4.9346E+10 2.2527E+09
cTn -4.6411E+08 -2.1706E+08 -1.9382E+08 1.5135E+07 -1.6920E+08 9.2640E+06
31 32 33 34 35 36
kn 0.0000E+00 0.0000E+00 -1.4155E+09 -1.7451E+09 8.1243E+06 -2.3990E+06
cn 0.0000E+00 0.0000E+00 -3.9743E+06 -2.7057E+06 1.5958E+04 -4.7181E+03
kTn -2.9367E+10 -1.0747E+10 1.3208E+10 9.9997E+09 -1.9771E+10 5.1191E+09
cTn -6.6288E+07 -2.3854E+07 2.3000E+07 2.5153E+07 -3.9248E+07 9.6705E+06
37 38 39 40 41 42
kn 0.0000E+00 -3.4634E+08 -2.6621E+09 -7.3699E+06 -1.6447E+07 -3.2997E+08
cn 0.0000E+00 -5.8881E+05 -3.8641E+06 -1.3444E+04 -2.8127E+04 -5.2243E+05
kTn 1.2511E+10 4.5378E+10 3.0996E+10 6.1565E+10 3.8787E+09 3.1127E+10
cTn 2.3189E+07 8.9426E+07 7.0953E+07 1.1015E+08 7.0887E+06 5.8484E+07
43 44 Resi. Stif.
kn 0.0000E+00 0.0000E+00 0.0000E+00
cn 0.0000E+00 0.0000E+00 0.0000E+00
kTn -5.8039E+10 -3.1792E+10 -1.0587E+10
cTn -9.5838E+07 -5.0192E+07 0.0000E+00

*units: kNm/rad for kn and kTn; and kNm sec/rad for cn and cTn
Table 2. Properties of mechanical elements in reduced 1DSDs in the rotational direction

374
| |{ } | |{ } | |{ } { } 0
T T T
M u C u K u + + = (22)
where
{ }
1 2 1 2
T
s f i m f i n
u u u u u u u u u u u u
(
=

(23)
where the mass matrix | |
T
M , the damping matrix | |
T
C , and the stiffness matrix | |
T
K are
the resultant matrices formed by superimposing the partial matrices.
i
u and
i
u are the
displacements at the DOFs in the reduced 1DSDs. The maximum DOFs in both directions
are represented by m ( 47 = ) and n ( 71 = ), respectively. In order to estimate the response
of the structural system, the absolute displacements are expressed by the sum of the
displacements of the inertial response and the input motion, as follows:

{ } { }
{ }
g
u U U = +
(24)
where

{ }
1 2 1 2
0 0
T
s f i f i
U U U U U U
(
= O O O O

(25)

f
J
1DSD in
Rotational
Direction
g m
U u = 0 =
n
u
Footing
4
u
f
u
4
k
4
c
f
m
f
u
4
H
Disp.
F
o
r
c
e

i
k
i
k o
1DSD in
Horizontal
Direction
3
u
3
k
3
c
3
H
2
u
2
k
2
c
2
H
1
u
1
k
1
c
1
H
Inelasticity in Inter-story

Fig. 7. Numerical model representing complete structural system.
Story No. Units 1 2 3 4
Mass m
i
t 750 750 750 750
Stiffness k
i
kN/m 2000000 2000000 2000000 2000000
Height H
i
m 12 9 6 3
Yield Strength p
i
kN 15000 30000 33000 44000

Table 3. Properties of four-story building

375

{ }
T
g g g g g g g g g g g g
U U U U U U U
(
= O O O O O

(26)
where
g
U and
g
O are the foundation input motions to the inertial system in the horizontal
and rotational directions, respectively. In this study, the foundation input motion in the
rotational direction
g
O is neglected as the amplitude is negligible because of the shallow
embedment of the foundation in the soil. Substituting Eq. 24 into Eq. 22 leads to
| |{ } | |{ } | |{ } | |{ }
T T T T g
M U C U K U M U + + =

(27)
The comparison is performed using the transfer functions (TFs) of the dynamic responses of
the superstructure with respect to the foundation input motion defined as
( )
1 1 sa g f g
T U U H U = + + O

for absolute acceleration. The TFs of the footing are also
computed. They are defined as
( )
ha f g g
T U U U = +

and
1 ra f g
T H U = O

for the absolute
acceleration associated with the horizontal and rotational motions, respectively.
Figure 8 shows the real part and the imaginary part of the transfer functions of the structural
systems defined above. The figure indicates that the transfer functions obtained with the
reduced 1DSDs are compatible with those obtained with the original FE model. This implies
that 1DSDs with almost one-tenth of the DOFs in the original system can correctly represent
the impedance functions in the target frequency region. Here the transfer functions obtained
using the 1DSDs with residual stiffness are not presented as negligibly small differences that
appeared in the impedance functions.

0 2 4 6 8 10
-4
-2
0
2
4
6
Tha
Tra
Tsa
FE model
1DSD (Tsa)
1DSD (Tha)
1DSD (Tra)
Frequency (Hz)
T
r
a
n
s
f
e
r

F
u
n
c
t
i
o
n
s

o
f

A
c
c
.

(
r
e
a
l

p
a
r
t
)
0 2 4 6 8 10
-8
-6
-4
-2
0
2
4
6
Tha
Tra
Tsa
Frequency (Hz)
T
r
a
n
s
f
e
r

F
u
n
c
t
i
o
n
s

o
f

A
c
c
.

(
i
m
a
g
i
n
a
r
y

p
a
r
t
)
(a) (b)

Fig. 8. Comparisons of the transfer functions of the structural system computed using
1DSDs without residual stiffness and the original FE model [(a) real part and (b) imaginary
part].
3.5 Dynamic response of structural system in time domain
3.5.1 Elastic response of structural system
In the previous section, it was shown that a structural system with reduced 1DSDs can
correctly simulate the transfer functions in the real part and the imaginary part in the
frequency domain. This implies that the time-history response of a structural system having
linearly elastic members can be appropriately calculated using 1DSDs. In this section, the

376
time-history response of a structural system with 1DSDs when subjected to foundation
input motion is computed and compared with that obtained with the original FE model. An
attempt is then made to compute the time-history response of the structural system with
inelasticity in the superstructure.
0 10 20 30 40
-40
-20
0
20
40
Superstructure (Top)
1DSD
FEmodel
A
c
c
.
(
m
/
s
2
)
Time (s)

0 10 20 30 40
-40
-20
0
20
40
Foundation (Horizontal)
1DSD
FEmodel
A
c
c
.
(
m
/
s
2
)
Time (s)

0 10 20 30 40
-1.0
-0.5
0.0
0.5
1.0
Foundation (Rotational)
1DSD
FEmodel
A
c
c
.
(
m
/
s
2
)
Time (s)

Fig. 9. Time-history response accelerations of superstructure and footing in the horizontal
and rotational directions when subjected to ground motion associated with 2004 Ojiya EW
by using 1DSDs and the original FE model.
The time-history analysis is performed on the basis of Eq. 27 by using Newmarks |
method ( 1 4 | = ) as a numerical integration scheme in which the time interval t A is 0.001
s. The acceleration response at the ground surface shown in Fig. 5 is applied as the
foundation input motion. Here the response accelerations defined in the previous section
(the same definitions as in the transfer functions) are calculated. Figure 9 shows the time-
history responses of the acceleration of the structural systems using reduced 1DSDs without
residual stiffness. The figure shows that the time histories of the 1DSDs are compatible with
those of the original FE model.
Note that a structural system using 1DSDs with residual stiffness cannot be obtained
correctly because the dynamic response of the system is not converged (the response
increases oscillatory with time). This implies that the structural system becomes unstable,
which is attributed to the residual stiffness components. The unit for residual stiffness in the
horizontal and rotational directions comprises only a negative stiffness, as shown in Table 1
and Table 2. This negative stiffness could be a drawback to stabilizing the dynamic response

377
of structures, although the units can achieve better agreement with the impedance functions
of the original systems in the frequency domain.
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03
-30000
-20000
-10000
0
10000
20000
30000
4th Inter-story
1DSD
FEmodel
S
t
r
e
n
g
t
h

(
k
N
)
Interstory Drift (m)
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03
-60000
-40000
-20000
0
20000
40000
60000
3rd Inter-story
S
t
r
e
n
g
t
h

(
k
N
)

-0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08
-60000
-40000
-20000
0
20000
40000
60000
2nd Inter-story
S
t
r
e
n
g
t
h

(
k
N
)
-0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08
-60000
-40000
-20000
0
20000
40000
60000
1st Inter-story
S
t
r
e
n
g
t
h

(
k
N
)

Fig. 10. Relationship between shear force and inter-story drift in each story when subjected
to ground motion associated with 2004 Ojiya EW by using 1DSDs and the original FE
model.
3.5.2 Inelastic response of structural system
Comparisons are extended to the structural system to allow inelasticity in the
superstructure. The inelasticity of the superstructure is represented by the Clough model
(Clough & Johnson, 1966), which is generally used to model reinforced concrete members.
The spring of the superstructure has a bilinear skeleton curve in which the ratio of tangent
stiffness to initial stiffness is assumed to be 0.1, as shown in Fig. 7. The yield strength
i
p in
each story is presented in Table 3. In this study, the modified NewtonRaphson method is
applied to calculate the nonlinear response of the system.
Figure 10 shows the relationship between shear force and inter-story drift in each story.
Figure 11 shows the time-history responses of the displacement of the foundation
f
u and
1 f
H u in the horizontal and rotational directions, respectively. The results indicate that
although a slight difference appears in the relationship between shear force and inter-story
drift in the first inter-storey, the inelastic responses obtained with the 1DSDs show
sufficiently close agreement with those obtained from the original FE model. From a
practical viewpoint, a dominant advantage is that the computational time can be

378
considerably reduced by transforming the original system into 1DSDs. According to a rough
measurement using the authors PC (CPU 3.40 GHz, RAM 3.00 GB), the inelastic responses
shown above were obtained in about 11,965 s with the original FE model, while those with
1DSDs were obtained in about 68 s.
0 10 20 30 40
-0.02
-0.01
0.00
0.01
0.02
Foundation (Horizontal)
1DSD
FEmodel
D
i
s
p
.
(
m
)
Time (sec)
0 10 20 30 40
-0.04
-0.02
0.00
0.02
0.04
Foundation (Rotational)
1DSD
FEmodel
D
i
s
p
.
(
m
)
Time (sec)

Fig. 11. Time-history response displacements of the foundation in the horizontal and
rotational directions when subjected to ground motion associated with 2004 Ojiya EW by
using 1DSDs and the original FE model.
4. Conclusion
This study demonstrates the transformation procedure using a 1DSD for a multi-story
building supported on a shallow foundation embedded in layered soil. In accordance with
the conventional modal concept, the reduced 1DSDs in both horizontal and rotational
directions are constructed with a small number of units associated with important modes in
the target frequency region in this application example. The impedance functions obtained
with the 1DSDs correctly simulate the impedance functions of the original FE model. The
transfer functions of the structural systems in the frequency domain using the 1DSD show
fairly good agreement with those obtained with the FE model. The time-history responses of
structures in both linearly elastic and inelastic cases can be simulated using the 1DSDs. The
results indicate that a significant decrease in the structural system with the 1DSD lead to a
marked decrease in the computational time taken for the results. Therefore, it may be
concluded that 1DSD transformation is effective and efficient for numerical computations in
SSI problems influenced by ground motions.
5. References
Baranov, V. A. (1967). On the calculation of excited vibrations of an embedded foundation
(in Russian), Voprosy Dynamiki I Prochnocti, Polytechnical Institute of Riga, No. 14,
pp. 195-209

379
Beredugo, Y. O. & Novak, M. (1972). Coupled horizontal and rocking vibration of
embedded footings, Can. Geotech. J., 9 (4), pp. 477-497
de Barros, F. C. P. & Luco, J. E. (1990). Discrete models for vertical vibrations of surface and
embedded foundations, Earthquake Engineering and Structural Dynamics, 19, pp. 289
303
Clough, R. W. & Johnson, S. B. (1966). Effect of stiffness degradation on earthquake ductility
requirements, The second Earthquake Engrg. Sym., Proc., pp. 227-232
Dobry, R. & Gazetas, G. (1988). Simple method for dynamic stiffness and damping of
floating pile groups, Geotechnique, 38 (4), pp. 557-674
Elsabee, F. & Morray, J. P. (1977). Dynamic behavior of embedded foundation, Research
Report, R77-33, Dep. of Civil Engrg., MIT. Cambridge, Mass.
Foss, K. A. (1958). Co-ordinates which uncouple the equations of motion of damped linear
dynamic systems, Journal of Applied Mechanics, 57, pp. 361364
Gazetas, G. (1991). Formulas and charts for impedances of surface and embedded
foundations, J. Geotech. Engrg. ASCE, 117(9), pp. 1363-1381
Hayashi, Y. & Katsukura, Y. (1990). Effective time domain soilstructure interaction analysis
based on FFT algorithm with causality condition, Earthquake Engineering and
Structural Dynamics, 19, pp. 693-708
Jean, W. Y., Lin, T. W. & Penzien, J. (1990). System parameters of soil foundations for time
domain dynamic analysis. Earthquake Engineering and Structural Dynamics, 19, pp.
541-553
Kausel, E. (2010). Early history of soil-structure interaction, Soil Dynamics and Earthquake
Engineering, 30, pp. 822-832
Kausel, E. & Roesset, J. M. (1975). Dynamic Stiffness of Circular Foundations, J. Engrg. Mech.
Div., ASCE, Vol. 101, No. EM6, pp. 770-1731
Kaynia, A. & Kausel, E. (1982). Dynamic stiffness and seismic response of pile groups,
Research Report, R82-03, Dept. of Civil Engrg., MIT, Cambridge, Mass.
Khodabakhshi, P., Jahankhah, H., & Ghannad, M. A. (2011). A discrete model for response
estimation of soil-structure systems with embedded foundations, Earthquake
Engineering and Engineering Vibration, 10 (2), pp. 263-276
Lysmer, J. & Kuhlemeyer, R. L. (1969). Finite dynamic model for infinite media, J. Eng. Mech.
Div., ASCE, 95, pp. 859-877
Makris, N. & Gazetas, G. (1993). Displacement phase differences in a harmonically
oscillating pile, Geotechnique, 43 (1), pp. 135-150
Meek, J. W. & Veletsos, A. S. (1974). Simple models for foundations in lateral and
rocking motion, Proc., 5th World Conf. on Earthquake Engrg., 2, Rome, Italy, pp.
2610-2613
Meek, J. W. (1990). Recursive analysis of dynamic phenomena in civil engineering,
Bautechnik, 67, pp. 205210, (in German)
Motosaka, M. & Nagano, M. (1992). Recursive evaluation of convolution integral in
nonlinear soilstructure interaction analysis and its applications, Journal of
Structural and Construction Engineering (AIJ), 436, pp. 71-80, (in Japanese)
Mylonakis, G. & Gazetas, G. (1998). Vertical vibration and additional distress of grouped
piles in layered soil, Soils and Foundations, JGS, 38 (1), pp. 1-14

380
Nakamura, N. (2006a). A Practical method to transform frequency dependent impedance to
time domain, Earthquake Engineering and Structural Dynamics, 35(2), pp. 217-234
Nakamura, N. (2006b). Improved methods to transform frequency dependent complex
stiffness to time domain, Earthquake Engineering and Structural Dynamics, 35(8), pp.
1037-1050
Nakamura, N. (2008a). Transform Methods for Frequency Dependent Complex Stiffness to
Time Domain Using Real or Imaginary Data Only, Earthquake Engineering and
Structural Dynamics, 37(4), pp. 495-515
Nakamura, N. (2008b). Time history Response Analysis Considering Dynamic Stiffness with
Both Frequency and Strain Dependencies, J. of Eng. Mechanics, ASCE, 134(4), pp.
530-541
Nogami, T. & Novak, M. (1977). Resistance of soil to a horizontally vibrating pile, Earthquake
Engineering and Structural Dynamics, 5, pp. 249-261
Nogami, T. & Konagai, K. (1986). Time domain axial response of dynamically loaded single
piles, J. Engrg. Mech., ASCE, 112(11), pp. 1241-1252
Nogami, T. & Konagai, K. (1988). Time domain flexural response of dynamically loaded
single piles, J. Engrg. Mech., ASCE, 114(9), pp. 1512-1525
Novak, M. (1974). Dynamic stiffness and damping of piles, Can. Geotech. J., 11, pp. 574-
598.
Novak, M. & Nogami, T. (1977). Soil-pile interaction in horizontal vibration, Earthquake
Novak, M., Nogami, T., Konagai, K., & Aboul-Ella, F. (1978). Dynamic soil reactions for
plane strain case, J. Engrg. Mech., ASCE, 104(4), pp. 953-959
Saitoh, M. & Watanabe, H. (2004). Effects of flexibility on rocking impedance of deeply
embedded foundation, Journal of Geotechnical and Geoenvironmental Engineering,
ASCE, 130(4), pp. 438-445
Saitoh, M. (2007). Simple model of frequency-dependent impedance functions in soil-
structure interaction using frequency-independent elements, Journal of Engineering
Mechanics, ASCE, 133(10), pp. 1101-1114
Saitoh, M. (2010a). Lumped parameter models representing impedance functions at the
interface of a rod on a viscoelastic medium, Journal of Sound and Vibration, 330, pp.
2062-2072
Saitoh, M. (2010b). Equivalent One-Dimensional Spring-Dashpot System Representing
Impedance Functions of Structural Systems with Non-Classical Damping, CMES:
Computer Modeling in Engineering & Sciences, 67(3), pp. 211-238
Saitoh, M. (2011a). On the performance of lumped parameter models with gyro-mass
elements for the impedance function of a pile-group supporting a single-degree-of-
freedom system, Earthquake Engineering and Structural Dynamics, DOI: 10.1002/
eqe.1147
Saitoh, M. (2011b). A one-dimensional lumped parameter model representing impedance
functions in general structural systems with proportional damping, International
Journal for Numerical Methods in Engineering, DOI: 10.1002/nme.3323

381
Taherzadef, R., Clouteau, D., & Cottereau, R. (2009). Simple formulas for the dynamic
stiffness of pile groups, Earthquake Engineering and Structural Dynamics, 38(15), pp.
1665-1685
Takemiya, H. & Yamada, Y. (1981). Layered soil-pile-structure interaction, Earthquake
Takewaki, I., Murakami, S., Fujita, K., Yoshitomi, S., Tsuji, M. (2011). The 2011 off the Pacific
coast of Tohoku earthquake and response of high-rise buildings under long-period
ground motions, Soil Dynamics and Earthquake Engineering, doi:10.1016/j. soildyn.
2011. 06. 001
Tileylioglu, S., Stewart, J. P., & Nigbor, R. L. (2011). Dynamic stiffness of a shallow
foundation from forced vibration of a field test structure, Journal of Geotechnical and
Geoenvironmental Engineering, Vol. 137 (4), pp. 344-353
Eberhard, M. O., Baldridge, S., Marshall, J., Mooney, W., G. J., Rix (2010). The Mw 7.0 Haiti
Earthquake of January 12, 2010, USGS/EERI Advance Reconnaissance Team: Team
Report V.1.1
Veletsos, A. S. & Dotson, K. W. (1988). Horizontal impedances for radially inhomogeneous
viscoelastic soil layers, Earthquake Engineering and Structural Dynamics, 16, pp. 947-
966
Weaver, J. W., Timoshenko, S. P., & Young, D. H. (1990). Vibration Problems in Engineering,
Fifth Edition, Wiley Interscience
Wolf, J. P. (1994). Foundation Vibration Analysis Using Simple Physical Models. Englewood
Cliffs, NJ, Prentice-Hall
Wolf, J. P. & Obernhuber, P. (1985). Nonlinear soil-structure-interaction analysis using
dynamic stiffness or flexibility of soil in the time domain, Earthquake Engineering
and Structural Dynamics, 13, pp. 195-212
Wolf, J. P. & Motosaka, M. (1989). Recursive evaluation of interaction force of unbounded
soil in the time domain, Earthquake Engineering and Structural Dynamics, 18, pp. 345-
363
Wolf, J. P. & Somaini, D. R. (1986). Approximate dynamic model of embedded foundation in
time domain, Earthquake Engineering and Structural Dynamics, 14, pp. 683-703
Wolf, J. P. & Paronesso, A. (1992). Lumped-parameter model for a rigid cylindrical
foundation embedded in a soil layer on rigid rock, Earthquake Engineering and
Structural Dynamics, 21, pp. 1021-1038
Wolf, J. P. (1997). Spring-dashpot-mass models for foundation vibrations. Earthquake
Wolf, J. P. (1991a). Consistent lumped-parameter models for unbounded soil: physical
representation, Earthquake Engineering and Structural Dynamics, 20(1), pp. 11-32
Wolf, J. P. (1991b). Consistent lumped-parameter models for unbounded soil: frequency-
independent stiffness, damping and mass matrices, Earthquake Engineering and
Structural Dynamics, 20(1), pp. 33-41
Wu, W. H. & Chen, C. Y. (2001). Simple lumped-parameter models of foundation using
mass-spring-dashpot oscillators, J Chin Inst Eng, 24(6), pp. 681-97
Wu, H. W. & Chen, C. Y. (2002). Simplified Soil-structure Interaction Analysis Using
Efficient Lumped-parameter Models for Soils, Soils and Foundations, 42, pp. 41-52

382
Wu, W. H. & Lee, W. H. (2002). Systematic lumped-parameter models for foundations based
on polynomial-fraction approximation. Earthquake Engineering and Structural
Dynamics, 31, pp. 1383-1412
Wu, H. W. & Lee, H. W. (2004). Nested Lumped-parameter Models for Foundation
Vibrations, Earthquake Engineering and Structural Dynamics, 33, pp. 1051-1058
Zhao, M. & Du, X. (2008). High-order lumped-parameter model for foundation based on
continued fraction, Beijing, 14WCEE
15
Numerical Prediction of Fire Whirlwind
Outbreak and Scale Effect of
Whirlwind Behavior
Seigo Sakai
Yokohama National University
Japan
1. Introduction
Our Japanese, especially the residents in east area of Japan, have experienced a large
earthquake on March 11, 2011, i.e. East Japan great earthquake disaster (Takewaki et al.,
2011). There were a lot of fires in the northeast area of Japan, for example in Kesen-numa
City. Despite of a number of town area fires, a fire whirlwind was never observed in this
disaster. However, fire whirlwind is still one of the concerned accidents in the earthquake
(Hough & Bilham; 2005).
When a large-scale wide area fire such as a town area fire or a forest fire occurs, there can be
a strong rotating flow to be called fire whirlwind. A fire whirlwind is a tornado that
includes flames, hot winds and sparks. The fire whirlwind is regarded as the worst case
which we should avoid at the time of a large-scale fire, because the whirlwind itself is
critical and scatters sparks widely to promote spread of a fire.
As a fire occurs, a flame makes an ascending current of air and uses up neighboring oxygen.
Furthermore, to collect oxygen from a wide area, there is a current of air against the flame,
resulting in big natural convection in the fire current of air. When the wind from a certain
specified direction blows in this fire current of air, homogeneity of suction of air with an
ascending current of air collapses. Then, a vortex is easy to come to occur, the fire current of
air becomes a fire whirlwind that is an ascending current of air accompanied with rotating.
Fire whirlwind may be pushed away by wind downstream, or may move in search of
oxygen.
Aiming at a property and elucidation of an outbreak factor of fire whirlwind as examples of
the pasts for a lesson, investigation and a reproduction experiment of the outbreak situation
(Graham, 1955; Emmons & Ying, 1967; Byram & Martin, 1970; Haines & Updike, 1971;
Martin et al., 1976; Muraszew et al., 1979; Emori & Saito, 1982; Satoh & Yang, 1996; Hayashi
et al., 2003; Liu, 2005; Kuwana et al., 2007; Liu et al., 2007; Kuwana et al., 2008; Chuah et al.,
2011), numerical analysis are performed till now (Satoh & Yang, 1997; Battaglia et al., 2000a;
Battaglia et al., 2000b; Snegirev et al., 2004; Hassan et al., 2005; Chuah et al., 2007; Grishin,
2007; Grishin et al., 2009). Though various factors are thought about outbreak of a fire
whirlwind, such as climatic condition or existence of underground flammable gas, it is hard

384
to say that property and outbreak mechanism of a fire whirlwind are to be elucidated
enough.
Therefore, in this study, it is firstly aimed to get basic knowledge to elucidate outbreak
mechanism of fire whirlwind and the property, by evaluating influence between a natural
convection generated by the fire and the wind blowing to fire whirlwind on the ground
utilizing numerical analysis. It is secondly examined whether a relationship exists between a
real phenomenon and the phenomenon in reduction models, performing the numerical
analysis of a fire whirlwind with respect to scale effect.
Three dimensional analyses are performed to investigate the thermal and flow fields by
using an analytical software FLUENT 6.3. Natural convection is caused from a plane source
of constant heat flux or constant temperature in the flat ground large enough. It is observed
and evaluated whether a fire whirlwind occurs or not in a constant wind of air. It is also
observed and evaluated how the whirlwind behaves in case that the whirlwind occurs.
Then, it is analyzed that those swirling flow in original scale, 1/10 scale, 1/50 scale, 1/100
scale from the original brake out to vanish. As an analytical condition, parameter calculation
is repeated to get the velocity of a parallel flow which is the easiest to occur the swirling
flow for each reduction model, and then scale effect is discussed by comparing the velocity
of the natural convection, the velocity of the parallel flow, the center pressure of the
whirlwind and the continuance time of the swirling flow. For making of analysis models, a
representative example of the fire whirlwind that occurred at Tokyo in the Great Kanto
Earthquake (1923) is referred, and three types of heat source model (L-character model,
random model and C-character model) are constructed.
2. Nomenclature
a Absorption coefficient of air
1 2 3
C ,C ,C
c c c
Constants for standard k- c turbulent model
p.j
c Specific heat of species of j at constant pressure
E Energy
F
External force vector

G Irradiance
b
G Turbulent kinetic energy production due to buoyancy force
k
G Turbulent kinetic energy production due to gradient of time-averaged
velocity
g
Gravitational acceleration vector

h Sensible heat enthalpy
j
h Sensible heat enthalpy of chemical species j
I Unit tensor
j
J
Diffusive flux vector of chemical species j

k Turbulent kinetic energy
eff
k Effective thermal conductivity
p Pressure

Numerical Prediction of Fire Whirlwind Outbreak and Scale Effect of Whirlwind Behavior

385
r
q Radiative heat flux
h
S Heat source including volumetric heat generation
k
S ,S
c
Source terms for k and c
T Temperature
ref
T Reference temperature
t Time
v
Velocity vector
j
Y Mass fraction of chemical species j
M
Y Expansion dissipation term for k
Greek symbols
c Turbulent energy dissipation rate
Viscosity of air
t
Turbulent viscosity of air
Density of air
o Stefan-Boltzmann constant
k
,
c
o o Turbulent Prandtl numbers for k and c
t Shear stress tensor
eff t Effective shear stress
3. Governing equations
In this analysis, chemical reaction and combustion phenomenon are not dealt with, and
driving force for convective heat transfer is due to density change of air from the heat
source. Therefore, governing equations of the phenomenon are as follows (Fluent Inc/Flient
Asia Pacific, 2006).
Mass conservation equation
( v) 0
t
c
+ V =
c

. (1)
Navier-Stokes equation

( )
( v) ( vv) p g F
t
c
+ V = V + V t + +
c

, (2)
where

T
2
( v v ) vI
3
(
t = V + V V
(

, (3)
and the second term of right side shows effect of volume expansion.
Energy equation

386
{ }
( ) { }
eff
eff j j h
( E) v( E p) k T h J v S
t
c
+ V + = V V + t +
c

. (4)
The first, second and third term of right side show energy transfers by conduction, diffusion
of chemical species, and viscous dissipation, respectively. In equation (4),

2
p v
E h
2
= +
, (5)
and given for ideal gas as

j j
j
h Yh =

. (6)
j
h is given as follows.

ref
T
j p. j
T
h c dT =
}
. (7)
where
ref
T is set to 300K in this analysis.
Equation of Turbulent model
Turbulent flow is treated in the analysis, and the standard k-c model is employed.

t
i k b M k
i j k j
k
( k) ( ku ) G G Y S
t x x x
(
| | c c c c
+ = + + + c + (
|
c c c o c
(
\ .

, (8)

2
t
i l k 3 b 2
i j j
( ) ( u ) C (G C G ) C S
t x x x k k
c c c c
c
(
| | c c c cc c c
c + c = + + + + (
|
c c c o c
(
\ .

. (9)
Radiation model
P-1 radiative heat exchange model (Siegel & Howell, 2002) is also employed to simulate the
radiative heat transfer from the heat source at high temperature. The equation of P-1 model
is as follows.

4
r
q aG 4a T V = o . (10)
Numerical analysis is performed three-dimensionally with FLUENT6.3.
4. Analysis model
A calculation domain is taken enough widely to avoid the influence of natural convection
due to heat source. Depth and width are 2000m and height is 500m. Heat source is modeled
by a combination of 16 squares (200m200m) at the center of bottom in the domain. Each
square is given a constant value of heat flux individually, and various profiles of large-scale
fires are easily constructed. While heat flux of house fire is about 1.0 MW/m
2
, simulation is
performed at 0.03MW/m
2
of heat flux. Fig. 1 shows the calculation domain for analysis.


387

Fig. 1. Calculation domain for analysis
Three models are considered in the analysis. One is based on the large-scale fire at the Great
Kanto Earthquake (1923 in Japan), and the profile of the heat source at the bottom is L-
character type (Model 1). The second one is the profile in which the heat source is
randomly dispersed at the bottom (Model 2). The third one is the profile of the heat source
at the bottom is C-character type (Model 3). Fig. 2 shows the profiles of the heat source.

(a) L-character type (Model 1) (b) random type (Model 2) (c) C-character type (Model3)
Fig. 2. Profiles of heat source at the bottom
Initially, air is uniformly enclosed in the domain at atmospheric pressure and temperature
of 300K. Bottom of the domain assumes adiabatic except 16 squares of modeled heat source.
Four sides and top of the domain is free for inflow and outflow.
5. Results and discussions
5.1 Outbreak of fire whirlwind
In this section, it is aimed to get basic knowledge to elucidate outbreak mechanism of fire
whirlwind and the property, by evaluating influence between a natural convection
generated by the fire and the wind blowing to fire whirlwind on the ground utilizing
numerical analysis (Sakai & Watanabe, 2007).
Horizontal wind
2000m
500m
2000m
200m
200m

388
For 5 minutes from calculation start, natural convection is developed in the domain. Then,
horizontal air flow is introduced from one side of the domain at 4, 6 and 8m/s of the
velocity. Wind on the ground is modeled by the air flow. Calculation is continued for
several minutes from the introduction of the air flow.
5.1.1 Model 1
Figs. 3 and 4 are the sample of streamline distribution after 5 minutes and that after 30
minutes for the wind of 4m/s. These streamlines are drawn from the pressure surface
whose pressure is lower by 30 Pa from the atmospheric pressure, and are colored by
velocity magnitude. Number of generated fire whirlwind and behavior of the whirlwind
(movement and extinction) are altered dependent on the velocity of wind. It is
summarized in the Fig. 5.

Fig. 3. Distribution of streamlines after 5 minutes for the wind of 4m/s

Velocity [m/s]
Velocity [m/s]


389

(a) wind of 4m/s (b) wind of 6m/s

(c) wind of 8m/s
Fig. 5. Summary of generation and extinction of fire whirlwind in Model 1
Red arrows in the figure show the distance from the generation point for the duration time
of the whirlwind. In the case of wind of 4m/s, three whirlwinds are generated, and duration
time is longer than the other 2 cases. In the case of wind of 6m/s, two whirlwinds are
generated, and duration time is shorter than the case of 4m/s. In the case of wind of 8m/s,
just one whirlwind is generated, and duration time is the shortest in 3 cases.
Generation points of the whirlwind are located at the tips and corner of L character. Fig. 6
shows a numerical result of streamlines in natural convective heat transfer at 5 minutes (no
wind). The points where the natural convective flow develops become origins of the fire
whirlwind.
The number of the generated whirlwind is decreased as the velocity of wind is increased.
This is because too fast wind blows the ascending current of air by natural convection to the
downstream. Figs. 7 and 8 are side views of streamlines for the wind of 4m/s in Model 1.
The former is streamlines after 18.6 minutes, and the latter is that after 26.6 minutes. As time
passes, it is shown that the ascending current of air by natural convection is blown to the
downstream by the wind.
generated at 3min.
extinct at 10min.
generated at 3min.
after moving, extinct at 4min.
generated at 3min.
extinct at 15min.
generated at 3min.
generated at 17min.
keep until 40min.
generated at 4min.
extinct at 6min.

390

Fig. 6. Distribution of streamlines after 5 minutes in natural convective heat transfer

Fig. 7. Side view of streamlines after 18.6 minutes for the wind of 4m/s in Model 1

Fig. 8. Side view of streamlines after 26.6 minutes for the wind of 4m/s in Model 1
Velocity [m/s]
Velocity [m/s]
Velocity [m/s]


391
5.1.2 Model 2
Figs. 9 and 10 are the samples of streamline distribution after 15 minutes and that after 35
minutes for the wind of 4m/s. These streamlines are also drawn from the pressure surface
whose pressure is lower by 30 Pa from the atmospheric pressure, and are colored by velocity
magnitude.

In contrast with the results in the L character type, number of generated fire whirlwind is
almost the same, i.e. two, even if the velocity of wind is altered. Behavior of the whirlwind
(movement and extinction) is also almost the same in the both cases of wind velocity 6 and
8m/s. In the case of wind velocity 4m/s, a whirlwind generated after 12 minutes around
two heat sources surfaces moves and once weakened because it is apart from the heat
source. Then, it is strengthened again as it returns on the heat source. The point that we
should pay attention to is that two fire whirlwinds in the case of 4m/s (one is generated
after 12 minutes and remains until the end of simulation, and the other is generated at 35
minutes and extinct at 39 minutes) move to the upstream.
Generation points of the whirlwind are located at the downstream heat source, and the
number of the generated whirlwind is almost the same though the velocity of wind is
varied. In the random type of heat source surfaces, there is a lack of heat source in the
upstream. Therefore, interaction between the ascending current of air by natural convection
and wind is disturbed (See Fig. 11). Then, the interacted flow arrives at the downstream heat
source, and easily generates the fire whirlwind because heat is supplied.
It is summarized in the Fig. 12. Duration time of the whirlwind becomes shorter as the
velocity of wind is increased.
Velocity [m/s]

392


Fig. 11. Upper view of streamlines after 18 minutes for the wind of 4m/s in Model 2.
5.1.3 Summary
As a result of numerical analysis, interaction between the ascending current of air by natural
convection and wind results in a fire whirlwind. The whirlwind sometimes moves and
extinct. Generation points of the whirlwind are located at the tips and corner in the model of
L character type. The points where the natural convective flow develops becomes origins
of the fire whirlwind. Generation points of the whirlwind in the model of random type are
located at the downstream heat source, and the number of the generated whirlwind is
almost the same though the velocity of wind is varied. This is because the ascending current
of air by natural convection is partially disturbed by the wind.
Velocity [m/s]
Velocity [m/s]


393

(a) wind of 4m/s

(b) wind of 6m/s (c) wind of 8m/s
Fig. 12. Summary of generation and extinction of fire whirlwind in Model 2
5.2 Scale effect analysis
In this section, performing the numerical analysis of fire whirlwind with respect to scale
effect, it is examined whether a relationship exists between a real phenomenon and the
phenomenon in the reduction model with taking into account radiative heat transfer (Sakai
& Miyagi, 2010).
Preceding the parametric calculation study, grid sensitivity for numerical result is
evaluated. Residuals of continuity, velocity components, energy, turbulent energy and
dissipation ratio of turbulent energy are derived and those absolute criteria are set to 10
-3

generated at 4min.
generated at 5min.
generated at 12min.
generated at 12min.
during moving, weakened, then strengthened at the other source.
generated at 35min.
generated at 5min.
generated at 6min.

394
except 10
-6
for energy to judge the convergence. For all scale ratios, grid number is altered
and grid sensitivity is checked. Even if the coarse mesh number of 570700 cells and 1442164
faces is utilized for calculation, those residuals are well converged within 1000 iterations for
each time step. Therefore, mesh number of 570700 cells and 1442164 faces is employed to
attain the precision and calculation speed.
5.2.1 Effect of radiative heat exchange
In our previous study, radiative heat exchange was not considered to simulate the
whirlwind, because we had to simplify the real phenomenon of the whirlwind into the
fundamental thermal and flow fields as much as possible (Sakai and Watanabe, 2007).
However, radiative heat exchange plays an important role in the heat transfer at the higher
temperature field. Therefore, the P-1 model is utilized to simulate the radiative heat transfer
from the heat source at high temperature.
Fig. 13 shows numerical results of streamlines in natural convective heat transfer at 5
minutes (no horizontal wind) for the L-character heat source in the original scale; (a) is the
case of no radiative heat exchange considered, and (b) is the case of radiative heat exchange
considered. Colors of streamlines show the temperature of the upward flow.

(a) No radiative heat exchange considered (b) Radiative heat exchange considered
Fig. 13. Distribution of streamlines after 5 minutes in natural convective heat transfer
In the case of no radiative heat exchange considered, temperature decreases within a few
decade meters from the heat source. Therefore, air above the heat source plays an adiabatic
role to prevent the heat transfer to the upper air layer. On the other hand, in the case of
radiative exchange considered, heat from the heat source flows more to the upper air layer,
resulting in more developed upward flow. In the case of no radiative exchange considered,
the maximum upward velocity of natural convection and the appropriate velocity of
horizontal wind to generate swirling flow are 23.5 m/s and 4 m/s, respectively. In the case
of radiative exchange considered, the maximum upward velocity of natural convection and
the appropriate velocity of horizontal wind are increased to 65.2 m/s and 11 m/s,
respectively, due to the effective heat transfer. It is clarified that the radiative heat exchange
must be considered to precisely simulate the fire whirlwind. The following numerical
results are taking into account the radiative heat exchange.
Temperature [K]


395
5.2.2 Generation and extinction of swirling flow in Model 1
Fig. 14 shows samples of pressure field at the bottom surface and streamline distributions
for the original scale simulation; (a) is at 2 minutes since the introduction of horizontal wind,
(b) is at 5.5 minutes since the introduction of horizontal wind, and (c) is at 7.5 minutes since
the introduction of horizontal wind. These streamlines are drawn from the bottom surface,
and are colored by velocity magnitude.
From Fig. 14 (a), two whirlwinds are generated and generation points of the whirlwind are
located at the tip and the corner of L character. At 5.5 minutes since the introduction of
horizontal wind, a whirlwind generated at the tip is extinguished, and the other whirlwind
is flown to the downstream (See Fig. 14 (b)). The remaining whirlwind is the strongest at
this time, and the maximum center pressure of the whirlwind is lower than the atmospheric
pressure by 1,061Pa. At 7.5 minutes since the introduction of horizontal wind, the remaining
whirlwind is divided into two whirlwinds (See Fig. 14 (c)), and is going to be extinguished
at 10 minutes since the introduction of horizontal wind.

Fig. 14. Pressure field at the bottom surface and streamline distributions since the
introduction of horizontal wind in Model 1
Numerical calculation is performed not only for the case of the original size, but also for
1/10 scale, 1/50 scale and 1/100 scale. In the original size calculation, natural convection is
developed in the domain after 5 minutes. The smaller the scale is, the shorter the time
necessary to develop the natural convection is. The times necessary to develop the natural
convection are 90 seconds for 1/10 scale model, 45 seconds for 1/50 scale model, and 30
seconds for 1/100 scale model. This is because dynamics between the original size and
scaled models are said to be similar if the rate of the scale ratio to the square of velocity (or
time) are the same, in which Froude numbers are the same.
After defining the appropriate velocity of horizontal wind to generate swirling flow,
behavior of the whirlwind (movement and extinction) is observed for all the cases. Number
of generated fire whirlwind and behavior of the whirlwind are almost the same for all the
cases. Therefore, characteristic physical quantities of whirlwind are discussed by comparing
with scale ratio and square root of scale ratio in the followings.
(a) 2 minutes (b) 5.5 minutes (c) 7.5 minutes
Pressure [Pa]
Velocity [m/s]

396
5.2.2.1 Pressure of whirlwind
Fig. 15 shows relation among the scale ratio, the square root of scale ratio, and pressure at
the bottom of natural convection after each period necessary to develop the natural
convection. Taken the original size as a standard, the pressure at the bottom of natural
convection is more coincident with the scale ratio rather than the square root of scale ratio.

Fig. 15. Relation among scale ratio, square root of scale ratio, and pressure at the bottom of
natural convection in Model 1
Fig. 16 shows relation among the scale ratio, the square root of scale ratio, and the maximum
center pressure of whirlwind, which is lower than the atmospheric pressure. (The times of
the strongest whirlwind generation after the introduction of the horizontal wind, i.e. the
times when the center pressure of whirlwind is maximum, are different for all cases, and
discussed in the next subsubsection.) The maximum center pressure of whirlwind is also
more coincident with the scale ratio rather than the square root of scale ratio. Therefore,
scale effect is recognized, in which pressure of whirlwind is coincident with the scale ratio.

Fig. 16. Relation among scale ratio, square root of scale ratio, and maximum center pressure
of whirlwind in Model 1


397
5.2.2.2 Time variation of whirlwind
Figs. 17 and 18 show relation among the scale ratio, the square root of scale ratio, and times
of whirlwind generation and extinction since introduction of horizontal wind. Taken the
original size as a standard, the times of whirlwind generation and extinction since
introduction of horizontal wind are more coincident with the square root of scale ratio
rather than the scale ratio. Duration time of whirlwind is derived from the times of
whirlwind generation and extinction since introduction of horizontal wind. As a matter of
course, the duration time is more coincident with the square root of scale ratio rather than
the scale ratio.

Fig. 17. Relation among scale ratio, square root of scale ratio, and time of whirlwind
generation since introduction of horizontal wind in Model 1

extinction since introduction of horizontal wind in Model 1

398
Fig. 19 shows relation among the scale ratio, the square root of scale ratio, and time of the
strongest whirlwind generation since introduction of horizontal wind. The time of the
strongest whirlwind generation is also more coincident with the square root of scale ratio
rather than the scale ratio. Therefore, scale effect is recognized, in which time variation of
whirlwind is coincident with the square root of scale ratio.

Fig. 19. Relation among scale ratio, square root of scale ratio, and time of the strongest
whirlwind generation since introduction of horizontal wind in Model 1
5.2.2.3 Velocity of horizontal wind
Fig. 20 shows relation among the scale ratio, the square root of scale ratio, and appropriate
velocity of horizontal wind to generate swirling flow. The appropriate velocity of horizontal
wind to generate swirling flow is also more coincident with the square root of scale ratio
rather than the scale ratio. Therefore, scale effect is recognized, in which velocity of
horizontal wind is coincident with the square root of scale ratio.

Fig. 20. Relation among scale ratio, square root of scale ratio, and appropriate velocity of
horizontal wind to generate swirling flow in Model 1


399
5.2.3 Generation and extinction of swirling flow in Model 3
Fig. 21 shows samples of pressure field at the bottom surface and streamline distributions
for the original scale simulation; (a) is at 2 minutes since the introduction of horizontal wind,
(b) is at 4.5 minutes since the introduction of horizontal wind, and (c) is at 5.5 minutes since
the introduction of horizontal wind. These streamlines are drawn from the bottom surface,
and are colored by velocity magnitude.

Fig. 21. Pressure field at the bottom surface and streamline distributions since the
introduction of horizontal wind in Model 3
From Fig. 21 (a), two whirlwinds are generated and generation points of the whirlwind are
located at the corners of C character. The whirlwinds are flown to the downstream, and
are the strongest at this time (See Fig. 21 (b)). The maximum center pressure of the
whirlwinds is lower than the atmospheric pressure by 970Pa. Then, the whirlwinds
approach each other regardless of horizontal wind direction (See Fig. 21 (c)), and is
remaining at 10 minutes since the introduction of horizontal wind.
Numerical calculation is performed not only for the case of the original size, but also for
1/10 scale, 1/50 scale and 1/100 scale. In the original size calculation, natural convection is
developed in the domain after 5 minutes. The smaller the scale is, the shorter the time
necessary to develop the natural convection is. The times necessary to develop the natural
convection are 30 seconds for 1/10 scale model, 15 seconds for 1/50 scale model, and 9
seconds for 1/100 scale model. These values are smaller than those of L-character heat
source. This is because the wider the area of heat source is, the more natural convection
develops.
Introducing the appropriate velocity of horizontal wind to generate swirling flow in Model
1 as a velocity of horizontal wind, behavior of the whirlwind (movement and extinction) is
observed for all the cases. While number of generated fire whirlwind is the same for all the
cases, whirlwinds are not extinguished, and are still remained in the cases of the original
size and 1/50 scale model. This is because the appropriate velocity of horizontal wind to
generate swirling flow in Model 1 is larger or smaller compared with the appropriate
velocity of horizontal wind to generate swirling flow in Model 3. Relation between
configuration of heat source and horizontal wind velocity is one of the important roles to
simulate generation and extinction of fire whirlwind. Despite of the difference in behavior of
(a) 2 minutes (b) 4.5 minutes (c) 5.5 minutes
Pressure [Pa]
Velocity [m/s]

400
the whirlwind, characteristic physical quantities of whirlwind are discussed by comparing
with scale ratio and square root of scale ratio as well as in the case of Model 1.
5.2.3.1 Pressure of whirlwind
Fig. 22 shows relation among the scale ratio, the square root of scale ratio, and pressure at the
bottom of natural convection after each period necessary to develop the natural convection.

Fig. 22. Relation among scale ratio, square root of scale ratio, and pressure at the bottom of
natural convection in Model 3
Taken the original size as a standard, the pressure at the bottom of natural convection is
more coincident with the scale ratio rather than the square root of scale ratio.
Fig. 23 shows relation among the scale ratio, the square root of scale ratio, and the maximum
center pressure of whirlwind, which is lower than the atmospheric pressure. The maximum
center pressure of whirlwind is also more coincident with the scale ratio rather than the
square root of scale ratio. Therefore, scale effect may be recognized, in which pressure of
whirlwind is coincident with the scale ratio.

Fig. 23. Relation among scale ratio, square root of scale ratio, and maximum center pressure
of whirlwind in Model 3


401
5.2.3.2 Time variation of whirlwind
Figs. 24 and 25 show relation among the scale ratio, the square root of scale ratio, and times
of whirlwind generation and the strongest whirlwind generation since introduction of
horizontal wind. Here is no discussion for time of whirlwind extinction, because whirlwinds
are not extinguished, and are still remained in the cases of the original size and 1/50 scale
model. The times of whirlwind generation and extinction since introduction of horizontal
wind are more coincident with the square root of scale ratio rather than the scale ratio.
Therefore, scale effect may be recognized, in which time variation of whirlwind is coincident
with the square root of scale ratio.

generation since introduction of horizontal wind in Model 3

extinction since introduction of horizontal wind in Model 1

402
5.2.4 Summary
In this section, performing the numerical analysis of fire whirlwind with respect to scale
effect, it is examined whether a relationship exists between a real phenomenon and the
phenomenon in the reduction model with taking into account radiative heat transfer. It is
clarified that the radiative heat exchange must be considered to precisely simulate the fire
whirlwind, because the radiative heat transfer encourages the maximum upward velocity of
natural convection and the appropriate velocity of horizontal wind due to the effective heat
transfer. Scale effect is recognized in the L-character heat source model, in which pressure of
whirlwind is coincident with the scale ratio, time variation of whirlwind and velocity of
horizontal wind are coincident with the square root of scale ratio.
6. Conclusions
In this study, it is firstly aimed to get basic knowledge to elucidate outbreak mechanism of
fire whirlwind and the property, by evaluating influence between a natural convection
generated by the fire and the wind blowing to fire whirlwind on the ground utilizing
numerical analysis.
- Interaction between the ascending current of air by natural convection and wind results
in a fire whirlwind. The whirlwind sometimes moves and extinct. Generation points of
the whirlwind are located at the tips and corner in the model of L character type. The
points where the natural convective flow develops becomes origins of the fire
whirlwind.
- Generation points of the whirlwind in the model of random type are located at the
downstream heat source, and the number of the generated whirlwind is almost the
same though the velocity of wind is varied.
It is secondly examined whether a relationship exists between a real phenomenon and the
phenomenon in reduction models, performing the numerical analysis of a fire whirlwind
with respect to scale effect.
- It is clarified that the radiative heat exchange must be considered to precisely simulate
the fire whirlwind, because the radiative heat transfer encourages the maximum
upward velocity of natural convection and the appropriate velocity of horizontal wind
due to the effective heat transfer.
- Scale effect is recognized in the L-character heat source model, in which pressure of
whirlwind is coincident with the scale ratio, time variation of whirlwind and velocity of
horizontal wind are coincident with the square root of scale ratio.
7. References
Battaglia, F.; McGrattan, K. B.; Rehm, R. G. & Baum, H. R. (2000b). Simulating Fire Whirls,
Combustion Theory and Modelling, Vol.4, No.3, pp.123-138
Battaglia, F.; Rehm, R. G. & Baum, H. R. (2000a). Fluid Mechanics of Fire Whirls: An Inviscid
Model, Physics of Fluids, Vol.12, No.11, pp.2859-2867
Byram, G. M. & Martin, R. E. (1970). The Modeling of Fire Whirlwinds, Forest Science, Vol.16,
No.4, pp.386-399


403
Chuah, K. H. & Kushida, G. (2007). The prediction of flame heights and flame shapes of
small fire whirls, Proceedings of the Combustion Institute, Vol. 31, pp.25992606
Chuah, K. H.; Kuwana, K.; Saito, K. & Williams, F. A. (2011). Inclined fire whirls, Proceedings
of the Combuation Institute, Vol.33, pp.2417-2424
Emmons, H. W. & Ying, S. J. (1967) The fire whirl, Proceedings of the Combustion Institute,
Vol.11, pp. 475488
Emori, R. I. & Saito, K. (1982). Model experiment of hazardous forest fire whirl, Fire
Technology, Vol.18, No.4, pp.319-327
Fluent Inc./Fluent Asia Pacific (2006). FLUENT 6.3 Users Guide, Fluent Incorporated
Graham, H. E. (1955). Fire whirlwinds, Bulletin American Meteorological Society, Vol.36, pp.
99103
Grishin, A. M. (2007). Effect of the Interaction between Fire Tornadoes on Their Propagation,
Doklady Physics, Vol.52, No.10, pp. 521-522
Grishin, A. M.; Matvienko, O. V. & Rudi, Y. A. (2009). Mathematical Modeling of Gas
Combustion in A Twisted Jet and of The Formation of A Fiery Whirlwind, Journal of
Engineering Physics and Thermophysics, Vol.82, No.5, pp.906-913
Haines, D. A. & Updike, G. H. (1971). Fire Whirlwind Formation over Flat Terrain, USDA
Forest Service Research Paper, NC-71
Hassan, M. I.; Kuwana, K.; Saito, K. & Wang, F. (2005). Flow Structure Of A Fixed-frame
Type Firewhirl, Proceedings of the Eighth International Symposium, International
Association for Fire Safety Science, Beijing, China, September 18-23, pp.951-962
Hayashi, Y.; Ohmiya, Y.; Iwami, T. & Saga, T. (2003). Experimental Study on Fire and Plume
Properties Using BRIs Fire Wind Tunnel Facility, International Journal for Fire
Science and Technology, Vol.22, pp.17-35
Hough, S. E. & Bilham, R. G. (2005). After the Earth Quakes: Elastic Rebound on an Urban
Planet, Oxford University Press USA
Kuwana, K.; Sekimoto,

K.; Saito, K. & Williams, F. A. (2007). Can We Predict the Occurrence
of Extreme Fire Whirls?, AIAA JOURNAL, Vol.45, pp.16-19
Kuwana, K.; Sekimoto,

K.; Saito, K. & Williams, F. A. (2008). Scaling fire whirls, Fire Safety
Journal, Vol.43, No.4, pp.252-257
Liu, N. A. (2005). Experimental and Theoretical Investigation on Fire Interactions and the
Induced Firewhirls in Square Fire Arrays, Proceedings of Fifth NRIFD International
Symposium on Forest Fires, Tokyo, pp.293-301
Liu, N. A.; Liu, Q.; Deng, Z. H.; Satoh, K. & Zhu, J. P. (2007). Burn-out Time Data Analysis
on Interaction Effects among Multiple Fires in Fire Arrays, Proceedings of the
Combustion Institute, Vol.31, pp.2589-2597
Martin, R. E.; Pendleton, D. W. & Burgess, W. (1976). Effect of Fire Whirlwind Formation on
Solid Fuel Burning Rates, Fire Technology, Vol.12, No.1, pp.33-40
Muraszew, A.; Fedele, J. B. & Kuby W.C. (1979). The fire whirl phenomenon, Combustion and
Flame, Vol. 34, pp.29-45
Sakai, S. & Miyagi, N. (2010). Numerical Study of Fire Whirlwind Taking into Account
Radiative Heat Transfer, IOP Conference Series: Materials Science and Engineering,
Vol.10, 012031
Sakai, S. & Watanabe, Y. (2007). Numerical Study of Interaction between Natural
Convection Flow and horizontal wind, Proceedings of the FEDSM2007, 5th Joint

404
ASME/JSME Fluids Engineering Conference, FEDSM2007-37212, San Diego,
California, USA, July 30-August 2, 2007
Satoh, K. & Yang, K. T. (1996). Experimental Observations of Swirling Fires, Proceedings of
ASME Heat Transfer Division, HTD-Vol.335, No.4, pp.393-400
Satoh, K. & Yang, K. T. (1997). Simulations of swirling fires controlled by channeled self-
generated entrainment flows [A], Fire Safety Sci., Proceedings of the 5th International
Symposium[C], pp.201-212
Siegel, R. & Howell, J. (2002). Thermal Radiation Heat Transfer, 4
th
Edition, Taylor & Fransis
Snegirev, A. Y.; Marsden, J. A.; Francis, J. & Makhviladze, G. M. (2004). Numerical studies
and experimental observations of whirling flames, International Journal of Heat and
Mass Transfer, Vol.47, pp.25232539
Takewaki, I.; Murakami, S.; Fujita, K.; Yoshitomi, S. & Tsuji, M. (2011). The 2011 off the
Pacific coast of Tohoku earthquake and response of high-rise buildings under long-
period ground motions, Soil Dynamics and Earthquake Engineering, in press
0
The Vibration of a Layered Rotating
Planet and Bryans Effect
Michael Y. Shatalov
1
, Stephan V. Joubert
2
and Charlotta E. Coetzee
2
1
CSIR (Pretoria) and Tshwane University of Technology
2
Tshwane University of Technology
South Africa
1. Introduction
Because, among other seismological observations, it is important to be able to predict the
location of the vibrating pattern of an earthquake, in this chapter we take the rst tentative
steps towards including "Bryans effect" in a mathematical model for the seismic vibration
pattern of a layered slowly rotating vibrating planet or moon.
Historically, G.H. Bryan observed (in his article (Bryan, 1890)) that when a vibrating structure
is subjected to an inertial rotation, the vibrating pattern rotates within the structure at a rate
proportional to (but in general not equal to) the inertial angular rate. This effect has come to
be known as " Bryans effect". It is interesting to note that (Rayleigh, 1894) mentions Bryans
effect in 233, but thereafter investigations of Bryans effect appear to have lain dormant
for about 75 years, reappearing in connection with a resonator gyroscope in 1965 in the
small Delco Wakeeld, MA, USA R&D facility, according to (Rozelle, 2009). Rozelle states
that the resonator gyroscope "has been utilized in many applications over its developmental
lifetime: aircraft navigation, strategic missile navigation, underground borehole navigation,
communication satellite stabilization, precision pointing, and in deep space missions". The
effect may be useful in understanding the dynamics of pulsating stars in astrophysics and this
was mentioned in the paper (Shatalov et al., 2009) in which the theoretical background on this
chapter was discussed in general terms. For the constant of proportionality, Bryan made the
following calculation for a body consisting of a ring or cylinder:
BF =
Angular rate of the vibration pattern
Inertial angular rate of the vibrating body
. (1)
This constant of proportionality BF has come to be known as "Bryans factor". The
authors ( Zhuravlev & Klimov, 1988) investigated Bryans effect for an elastic, isotropic,
spherically symmetric body, rotating in three-dimensional space. Among other results, they
demonstrated that Bryans factor depends on the vibration mode. Their concise formulation
is given in general terms without computational detail or assumptions on the magnitude of
rotation or illustrative examples. In (Shatalov et al., 2009) "slow rotation" (explained below)
was assumed for spherical bodies consisting of concentric layers of elastic and/or acoustic
media. In that paper, some detail was supplied for computations and an illustrative example
was presented. They did not assume a "thin shell theory", as in (Loveday & Rogers, 1998),
16
2 Will-be-set-by-IN-TECH
where Bryans effect is considered in a thin cylindrical shell for both high and low rotational
rates. Consequently the model described by (Shatalov et al., 2009) is readily adaptable to the
structure of an ideal planet or moon that is rotating slowly.
In this chapter we study Bryans effect applied to a rotating planet or moon that may have
several solid and liquid layers that are assumed to be rotating at the same rate. The rotation
rate is assumed to be small when compared to the lowest eigenvalue of vibration =
2
T
where T is the period of oscillation of the vibration pattern for a natural mode of vibration. As
a rst tentative approach, we discuss isotropic solid spherical layers that may be isotropically
damped as well as isotropic inviscid spherical liquid layers. If damping is present in the media
we expect to encounter, as a rst approach, we assume that it will be isotropic and be light in
the sense that the "damping factor" will be substantially smaller than the lowest eigenvalue
of the system. Using Rayleighs dissipation function (Goldstein, et al.), we demonstrate that
light, isotropic, viscous damping does not inuence Bryans effect or Bryans factor (Equation
(1)). Hence, in the sequel, we assume that the body is subjected to nondecaying vibrations
in one of its natural modes. The introduction of "impurities" such as prestress, mass-stiffness
imperfections and anisotropic damping effects into the calculations is important for real-life
situations (as opposed to ideal situations with at most isotropic, viscous damping) and has
been earmarked by us for further study. Indeed, we have recently published a spherical model
with varying mass density (Shatalov et al., 2011) that indicates that the rotation rate of the
vibrating pattern is nonlinear (as opposed to the linear rate for the ideal state discussed below)
and that this nonlinearity manifests itself as a "capture effect" whereby the rotation angle of
the vibration pattern varies periodically.
2. Preliminaries
Consider a coordinate system Oxyz and a composite spherical body, with its centre at the
origin O, consisting of concentric solid and or acoustic layers. Let N be the number of
concentric spherical media in the system and a
i
and a
i1
the inner and outer radii of the
i
th
layer respectively, i = 1, , N. We convert to spherical coordinates Or as depicted in
Figure ??, where we have adopted the notation of (Spiegel, 1967).
Consider the position of rest P(r, , ) of a vibrating particle in the i
th
layer where a
i
r
a
i1
. Let r be the unit vector in the direction of increasing r. Hence the position vector of the
point P(r, , ) is r = r r. Consider the usual unit vectors

=
r
(in the direction of

increasing ) and

=
r
(in the direction of increasing ). Let w

i
+u
i
+v
i
(where
w
i
= w
i
r, u
i
= u
i

and v
i
= v
i

) represent the displacement from the position of rest of the
vibrating particle in the i
th
layer. For the sake of simplicity, we suppress subscripts i when no
confusion is expected. The position vector of the vibrating particle is thus
R = (r + w) r + u

+v

. (2)
Now consider an inertial coordinate system OXYZ with origin O, where initially the
X, Y, Z-axes correspond to the x, y, z-axes respectively. Let the spherical body (the Or
Oxyz system) rotate about the z-axis with respect to inertial space OXYZ with a small constant
angular rate. If

k is the unit vector in the direction of increasing z, then let the angular velocity
of the body be
=
k = ( r cos

sin ), (3)
406
Planet and Bryans Effect 3
Fig. 1. The spherical coordinate system Or and its relation to the reference frame Oxyz that
is rotating slowly about the zaxis at rate , showing a particle in the i
th
layer with position
of rest P(r, , ) and displacement from the position of rest u in the direction of increasing
angle , v in the direction of increasing angle and w in the radial direction.
where the dimensionless parameter is a measure of smallness. By the word smallness of
the angular rate of rotation we mean (as mentioned above) that this rate is substantially
smaller than the lowest eigenvalue of the system.
3. Gyroscopic effects in distributed bodies
The mathematical formulation given below (in spherical coordinates) is presented within the
framework of the linearised, three-dimensional theory of elasticity (Redwood, 1960). With
Lagranges equations (Spiegel, 1967) in mind, we formulate expressions for the (approximate)
kinetic and potential energies of the system of concentric spherical bodies. The absolute linear
velocity of the vibrating particle is
V =
dR
dt
+R =
( w v sin ) r+
( uv cos )

+
( v + (u cos + (r + w) sin ))

. (4)
407 The Vibration of a Layered Rotating Planet and Bryans Effect
Neglecting terms of O(
2
), the approximate kinetic energy of the system of concentric
spherical bodies is given by
K =
1
2
N
i=1
i
2
_
0
_
0
_
a
i1
a
i
__
u
2
i
+ v
2
i
+ w
2
i
_
+ 2[(u
i
v
i
u
i
v
i
) cos +
( v
i
(r + w
i
) v
i
w
i
) sin ]} r
2
sin drdd. (5)
The potential energy of the system of concentric spheres is
P =
1
2
N
i=1
2
_
0
_
0
_
a
i1
a
i
_
i,rr
i,rr
+
i,
i,
+
i,
i,
+
i,r
i,r
+
i,
i,
+
i,r
i,r
_
r
2
sin drdd (6)
where we use for mass density, for stress and for strain. We indicate Lams
constants (from the theory of elasticity) by
i
and
i
for the i
th
layer. In a spherical coordinate
system, with the subscript i suppressed for the sake of brevity (i.e. we write
rr
for
i,rr
etc.), a
standard reference such as (Redwood, 1960) yields stresses
rr
= (
rr
+
) + 2
rr
;
= (
rr
+
) + 2
= (
rr
+
) + 2
r
= 2
r
;
= 2
;
r
= 2
r
; (7)
and strains
rr
=
w
r
;
=
1
r
_
u
+ w
_
;
=
1
r
_
u cot +
1
sin
v
+ w
_
;
r
=
u
r
+
1
r
_
w
u
_
;
=
1
r
_
1
sin
u
+
v
v cot
_
;
r
=
v
r
+
1
r
_
1
sin
w
v
_
. (8)
Assume that we can express the magnitude of the displacements u
i,
v
i
and w
i
of the vibrating
particle P in the i
th
body as follows:
u
i
(r, , , t) = U
i
(r, ) [C(t) cos m + S(t) sin m] , (9a)
v
i
(r, , , t) = V
i
(r, ) [C(t) sin m S(t) cos m] , (9b)
w
i
(r, , , t) = W
i
(r, ) [C(t) cos m + S(t) sin m] , (9c)
where the nature of the functions C(t) and S(t) is still to be determined, U
i
(r, ), V
i
(r, ) and
W
i
(r, ) are unknown eigenfunctions of the system and m is the circumferential wave number.
Substituting Equations (9) into Equations (8), (7), (6) and (5) involves a long algebraic
calculation. A computer algebra system is handy for checking the calculation that yields:
K =
_
I
0
_

C
2
+

S
2
_
+ 2I
1
_

CS C

S)
_
_
(10)
408
and
P = I
2
_
C
2
+ S
2
_
. (11)
Here
I
0
=
1
2
N
i=1
_
0
_
a
i1
a
i
_
U
2
i
+ V
2
i
+W
2
i
_
r
2
sin drd, (12)
I
1
=
N
i=1
_
0
_
a
i1
a
i
(U
i
cos +W
i
sin ) V
i
r
2
sin drd, (13)
I
2
=
N
i=1
_
0
_
a
i1
a
i
2
i
_
m csc U
i
+ cot V
i
V
i
_
2
+
_
W
i
+ r
U
i
r
U
i
_
2
+
_
V
i
+ m csc W
i
) r
V
i
r
_
2
+
(cot U
i
+ m csc V
i
+W
i
)
_
(
i
+ 2
i
)
_
cot U
i
+
m csc V
i
+ 2 (
i
+
i
) W
_
+
i
_
U
i
+ r
W
i
r
__
+
_
W
i
+
U
i
_
_

i
(cot U
i
+ m csc V
i
) +
2 (
i
+
i
) W
i
+ (
i
+ 2
i
)
U
i
+ r
i
W
i
_
+
r
W
i
r
_
i
_
cot U
i
+ m csc V
i
+
2 W
i
+
U
i
_
+ r (
i
+ 2
i
)
W
i
r
__
sin drd. (14)
Because K = K(C, S,

C,

S) and P = P(C, S), the Lagrangian
L(C, S,

C,

S) = K P (15)
yields two equations of motion from Lagranges equations (Goldstein, et al.):
d
dt
L

C

L
C
=
F

C
, (16a)
d
dt
L

S

L
S
=
F

S
(16b)
where
F =
1
2
(c

C
2
+ s

S
2
) (17)
is Rayleighs dissipation function, c and s are viscous damping constants. We assume isotropic
damping, that is, say, c = s = D. We further assume that for the media we will encounter,
the "damping factor"
=
D
2I
0
(18)
is substantially smaller than the lowest eigenvalue of the vibrating system. Computation of
Equations (16) yields a coupled system of second-order linear ordinary differential equations
(ODE):
C + 2
S +
2
C + 2

C = 0, (19a)
S 2

C +
2
S + 2

S = 0, (19b)
where
1 =
I
1
I
0
1 (20)
and
2
=
I
2
I
0
. (21)
We now show that in Equation (20) is Bryans factor as given by Equation (1) and that in
Equation (21) is an eigenvalue for the vibrating system.
In order to interpret what the system of ODE (19) represents, combine the two equations by
considering the complex function Z = C + iS to obtain the single equation
Z + 2( i)

Z +
2
Z = 0. (22)
where, as is usual, i
2
= 1. Writing Z in polar form
Z(t) = Y(t)e
i(t)
(23)
and assuming that (t) has the linear form
(t) = at, (24)
while Y(t) decays according to
Y(t) = X(t)e
bt
, (25)
then
Z = X(t)e
(iab)t
(26)
and substituting into Equation (22), we obtain the ODE
X + 2 [(ia b) + ( i)]

X+
_
2 ( i) (ia b) + (ia b)
2
+
2
_
X = 0. (27)
If we choose a = and b = , then the coefcient of

X vanishes in Equation (27) and we
obtain the ODE
X +
2
X = 0 (28)
where
2
=
2
2
( i)
2
(29)
Neglecting O(
2
), we obtain
. (30)
Consequently Equation (28) approximates the equation of motion of a harmonic oscillator.
Equations (19) can now be viewed in the form
Z(t) =
_
e
t
X(t)
_
e
it
. (31)
410
These equations represent a "vector" in the complex plane with its size varying like a damped
harmonic oscillator and its position varying at a rate (in the rotating reference frame
Oxyz). Thus, according to Equation (1) , is Bryans factor for the system. Consequently,
neither Bryans effect nor the value of Bryans factor , depends on the inclusion of light
isotropic viscous damping in the model and we will thus neglect damping in the sequel. The
rotation of the vibrating pattern is in the direction of rotation of the system if > 0 and in
the opposite direction if < 0. Equations (28) and (30) show that is an eigenvalue of the
vibrating system.
Taking Equation (30) into account, Equation (28) has a general solution of the form
X(t) = Acos(t + ) (32)
where the amplitude A and phase angle are arbitrary constants. Consequently, neglecting
damping, from Equation (31) we obtain
C + iS = Acos(t + )e
it
. (33)
Equating real and imaginary parts we have a good approximation
C(t) = Acos t cos(t + ), (34a)
S(t) = Asin t cos(t + ) (34b)
and so initially, that is, when t 0, we nd that
C(t) Acos(t + ), (35)
S(t) 0. (36)
4. Eigenfunctions
In order to determine the nature of Bryans constant (Equation (20)) and the square of the
corresponding eigenvalue (Equation (21)), we neglected O(
2
). We are going to demonstrate
that it is also feasible to neglect the small rotation rate when we derive expressions for
the eigenfunctions U
i
(r, ), V
i
(r, ) and W
i
(r, ), i = 1, , N, as given in Equations (9).
Indeed, assume that we have a system of partial differential equations (PDE) involving the
eigenfunctions U
i
(r, ), V
i
(r, ) and W
i
(r, ). Using the small parameter and a perturbation
method Nayfeh (1973), assume that we have an asymptotic expansion
U
i
(r, ) = U
i,0
(r, ) + U
i,1
(r, ) +
2
U
i,2
(r, ) + , (37)
where the U
i,j
(r, ), j = 0, 1, 2, are independent of and U
i,0
(r, ) is that part of the
solution when = 0. A similar argument is valid for the eigenfunctions V
i
(r, ) and W
i
(r, ).
Substituting these three expansions into Equations (12) and (13) and again neglecting all term
of O(
2
), the formula for given in Equation (20) yields
=
2
N
i=1

i
_
0
_
a
i1
a
i
__
U
i,0
cos +W
i,0
sin
_
V
i,0
+O()
r
2
sin drd
N
i=1

i
_
0
_
a
i1
a
i
_
U
2
i,0
+ V
2
i,0
+W
2
i,0
+O()
_
r
2
sin drd
. (38)
Now write Bryans factor as an asymptotic expansion
=
0
+
1
+
2
2
+ (39)
where the
j
, j = 0, 1, 2, are independent of and
0
is given by Equation (38) when = 0,
that is
0
=
2
N
i=1

i
_
0
_
a
i1
a
i
__
U
i,0
cos +W
i,0
sin
_
V
i,0
r
2
sin drd
N
i=1

i
_
0
_
a
i1
a
i
_
U
2
i,0
+ V
2
i,0
+W
2
i,0
_
r
2
sin drd
. (40)
Hence, because we neglect O(
2
), Equation (39) yields
=
0
+O(
2
)
0
(41)
and so

0
. (42)
A similar result is valid for the square of the eigenvalue determined by Equation (21).
Consequently, because only the eigenfunctions of a non-rotating body need be considered
for the purposes of numerically calculating , in the sequel we will use the notation u
i
for u
i,0
,
U
i
for U
i,0
, etc. Indeed, using (Redwood, 1960) and our notation for stresses, the equations of
motion of an isotropic solid, nonrotating vibrating body in spherical coordinates are
2
u
t
2
=

rr
r
+
1
r
+
1
r sin
+
2
rr
+cot
r
r
, (43a)
2
v
t
2
=

r
r
+
1
r
+
1
r sin
+
3
r
+cot (
)
r
, (43b)
2
w
t
2
=

r
r
+
1
r
+
1
r sin
+
3
r
+2 cot
r
. (43c)
The stresses are given by Equation (7). The coupled system of second-order PDE given by
Equations (43) for the three displacement components u, v and w can be uncoupled directly,
but this leads to a sixth-order PDE. Hence, in a manner similar to that explained in (Eringen
& Suhubi, 1975), we express the displacement components of Equations (43) in terms of
derivatives of potentials = (r, , ), = (r, , ) and = (r, , ) as follows:
u =
_
1
r
_
+
(r)
r
_
+
1
a sin
_
cos (t + ) , (44a)
v =
_
1
r sin
_
+
(r)
r
_
1
a
_
cos (t + ) , (44b)
w =
_
r
+

2
(r)
r
2
+ r
2
_
cos (t + ) , (44c)
where
2
the Laplace operator in spherical coordinates, is the eigenvalue mentioned above,
is some phase angle (see Equation (32)) and a is a nonzero constant with the dimension of
length. When Equations (44) are substituted into Equations (43) and the resulting equations
of motion are uncoupled, it is found that each potential , and satises the Helmholtz
equations
2
+ k
2
1
= 0,
2
+ k
2
2
= 0,
2
+ k
2
2
= 0 (45)
412
with wave numbers
k
1
= k
1
() =
_
/( + 2), (46a)
k
2
= k
2
() =
_
/. (46b)
In order to obtain initial (at t 0) agreement with our choice of Equations (9), the appropriate
solutions to Equations (45) are
m,n
(r, , ) =
[B
1
j
n
(k
1
r) + B
2
y
n
(k
1
r)] P
m
n
(cos ) cos (m) , (47a)
m,n
(r, , ) =
[B
3
j
n
(k
2
r) + B
4
y
n
(k
2
r)] P
m
n
(cos ) cos (m) , (47b)
m,n
(r, , ) =
[B
5
j
n
(k
2
r) + B
6
y
n
(k
2
r)] P
m
n
(cos ) sin (m) , (47c)
where, as usual, j
n
(kr) =
_

2kr
J
n+1/2
(kr) and y
n
(kr) =
_

2kr
Y
n+1/2
(kr) represent the
spherical Bessel and Neumann functions respectively where J
n+1/2
and Y
n+1/2
represent the
Bessel and Neumann functions respectively, while P
m
n
is the associated Legendre polynomial.
The symbols B
1
, B
2
, , B
6
are arbitrary constants (if the body contains the centre O, then the
constants B
2
= B
4
= B
6
= 0). By substituting Equations (47) into Equations (44) we obtain
the nature of the eigenfunctions U(r, ), V(r, ) and W(r, ) as given in Equations (9).
The motion of a compressible, inviscid acoustic medium is represented by the following wave
equation:
2
p =

2
p
t
2
, (48)
where is the bulk modulus and the mass density of the acoustic medium. With the wave
number
k
3
= k
3
() =
_
/, (49)
an appropriate solution to PDE (48) is
p
m,n
(r, , , t) =
{[B
7
j
n
(k
3
r) + B
8
y
n
(k
3
r)] P
m
n
(cos ) cos (m)} cos (t + ) , (50)
where p = p
m,n
(r, , , t) is the pressure in the acoustic medium and B
7
, B
8
are arbitrary
constants. Particle displacement of the acoustic medium in the radial direction is
w =
1
2
p
r
. (51)
In the sequel, keep in mind that for j = 1, 2, 3 and layers i = 1, , N, we will write k
i,j
for k
j
in layer i, etc.
5. Boundary and continuity conditions
Observing Equations (44) it is possible to distinguish between spheroidal and torsional modes.
For the spheroidal mode we assume that
i
= 0. In this case the stress components of the
solids are
i,rr
=
_
2
i

2
i
r
2

i
k
2
i,1
i
_
+ 2
i
_
3
(r
i
)
r
3
+ k
2
i,2
(r
i
)
r
_
, (52a)
i,r
=
2
i
r
_
_
i
r

i
r
_
+
_
r
i
r
2
+

i
r
+
_
rk
2
i,2
2

1
r
_
i
__
, (52b)
i,r
=
2
i
r sin
_
_
i
r

i
r
_
+
_
r
i
r
2
+

i
r
+
_
rk
2
i,2
2

1
r
_
i
__
, (52c)
where i = 1, 2, 3, 4, 6 denotes a solid layer in the spherical structure. Indeed, here we assume
i = 1 for the outer solid layer (crust), i = 2 for the next inner layer (upper mantle-1), i = 3
for the upper mantle-2, i = 4 for the lower mantle and i = 6 for the inner solid core.
We do not use i = 5 in the list of the solids because this describes the liquid outer core.
For the investigation of the torsional modes we assume that
i
= 0 and
i
= 0 and hence the
corresponding stress components are
i,r
=

i
a sin
i
r

i
r
_
, (53a)
i,r
=
i
a
i
r

i
r
_
. (53b)
Let us model a spherical planet with solid inner core, four solid layers and one liquid
layer. Assume that the outer radius of the planet is r = a and that the boundary between
the crust and the upper mantle-1 is located at radius r = a
1
. Furthermore, assume that
the boundary between the upper mantle-1 and the upper mantle-2 is situated at radius
r = a
2
, while at r = a
3
we nd the boundary between the upper mantle-2 and the lower
mantle. All these layers are considered as isotropic solid bodies. Assume that the boundary
between the lower mantle and the liquid outer core is situated at radius r = a
4
and nally,
radius r = a
5
is the boundary between the liquid outer core and the solid inner core.
The elastic potentials and pressure that describe the spheroidal modes of the planet are
solutions given by Equations (47a), (47b) and (50):
i
(r, , ) =
_
A
4i3
j
n
_
k
i,1
r
_
+ A
4i2
y
n
_
k
i,1
r
_
P
m
n
(cos ) cos m, (54a)
i
(r, , ) =
_
A
4i1
j
n
_
k
i,2
r
_
+ A
4i
y
n
_
k
i,2
r
_
P
m
n
(cos ) cos m, (54b)
p
5
(r, , ) = [A
17
j
n
(k
5,3
r) + A
18
y
n
(k
5,3
r)] P
m
n
(cos ) cos m, (54c)
6
(r, , ) = [A
19
j
n
(k
6,1
r)] P
m
n
(cos ) cos m, (54d)
6
(r, , ) = [A
20
j
n
(k
6,2
r)] P
m
n
(cos ) cos m, (54e)
where the symbols A with various subscripts represent arbitrary constants that will be
determined from the boundary and continuity conditions described below and i = 1, 2, 3, 4.
The upper core of the planet does not have a torsional mode and is considered to be an
inviscid ideal uid. Despite the fact that the inner core is solid and hence that it could
support the torsional mode, we do not consider it when we analyse the torsional mode
because the shear stress in the inner core cannot be transformed into the outer layers
through the shear free liquid outer core. The elastic potential describing the torsional modes
of the planet is obtained from Equation (47c):
i
(r, , ) =
_
B
2i1
j
n
_
k
i,2
r
_
+ A
2i
y
n
_
k
i,2
r
_
P
m
n
(cos ) cos m, (55)
where i = 1, 2, 3, 4.
414
The boundary and continuity conditions belowexpress the balance between the radial stresses
and pressure and equality of the tangential stresses in the solid layers. They also express the
equality of the radial and tangential components between the solid layers (correspondingly,
only radial components between the solids and uid layer). The outer layer is assumed
to be free from radial and tangential stress components. Using these assumptions together
with Equations (52) we obtain the following boundary and continuity conditions for the
investigation of the spheroidal modes:
At r = a
0
= a
__
2
1

2
1
r
2

1
k
2
1,1
1
_
+ 2
1
_
3
(r
1
)
r
3
+ k
2
1,2
(r
1
)
r
__
r=a
= 0, (56)
2
1
a
_
_
1
r

1
r
_
+
_
r
1
r
2
+

1
r
+
_
rk
2
1,2
2

1
r
_
1
__
r=a
= 0. (57)
At r = a
1
__
2
1

2
1
r
2

1
k
2
1,1
1
_
+ 2
1
_
3
(r
1
)
r
3
+ k
2
1,2
(r
1
)
r
__
r=a
1
=
__
2
2

2
2
r
2

2
k
2
2,1
2
_
+ 2
2
_
3
(r
2
)
r
3
+ k
2
2,2
(r
2
)
r
__
r=a
1
, (58)
2
1
a
1
_
_
1
r

1
r
_
+
_
r
1
r
2
+

1
r
+
_
rk
2
1,2
2

1
r
_
1
__
r=a
1
=
2
2
a
1
_
_
2
r

2
r
_
+
_
r
2
r
2
+

2
r
+
_
rk
2
2,2
2

1
r
_
2
__
r=a
1
, (59)
_
1
r
+
_
2
(r
1
)
r
2
+ k
2
1,2
(r
1
)
__
r=a
1
=
_
2
r
+
_
2
(r
2
)
r
2
+ k
2
2,2
(r
2
)
__
r=a
1
, (60)
_
1
+
(r
1
)
r
_
r=a
1
=
_
2
+
(r
2
)
r
_
r=a
1
. (61)
At r = a
2
__
2
2

2
2
r
2

2
k
2
2,1
2
_
+ 2
2
_
3
(r
2
)
r
3
+ k
2
2,2
(r
2
)
r
__
r=a
2
=
__
2
3

2
3
r
2

3
k
2
3,1
3
_
+ 2
3
_
3
(r
3
)
r
3
+ k
2
3,2
(r
3
)
r
__
r=a
2
, (62)
2
2
a
2
_
_
2
r

2
r
_
+
_
r
2
r
2
+

2
r
+
_
rk
2
2,2
2

1
r
_
2
__
r=a
2
=
2
3
a
2
_
_
3
r

3
r
_
+
_
r
3
r
2
+

3
r
+
_
rk
2
3,2
2

1
r
_
3
__
r=a
2
, (63)
_
2
r
+
_
2
(r
2
)
r
2
+ k
2
2,2
(r
2
)
__
r=a
2
=
_
3
r
+
_
2
(r
3
)
r
2
+ k
2
3,2
(r
3
)
__
r=a
2
, (64)
_
2
+
(r
2
)
r
_
r=a
2
=
_
3
+
(r
3
)
r
_
r=a
2
. (65)
At r = a
3
__
2
3

2
3
r
2

3
k
2
3,1
3
_
+ 2
3
_
3
(r
3
)
r
3
+ k
2
3,2
(r
3
)
r
__
r=a
3
=
__
2
4

2
4
r
2

4
k
2
4,1
4
_
+ 2
4
_
3
(r
4
)
r
3
+ k
2
4,2
(r
4
)
r
__
r=a
3
, (66)
2
3
a
3
_
_
3
r

3
r
_
+
_
r
3
r
2
+

3
r
+
_
rk
2
3,2
2

1
r
_
3
__
r=a
3
=
2
4
a
3
_
_
4
r

4
r
_
+
_
r
4
r
2
+

4
r
+
_
rk
2
4,2
2

1
r
_
4
__
r=a
3
, (67)
_
3
r
+
_
2
(r
3
)
r
2
+ k
2
3,2
(r
3
)
__
r=a
3
=
_
4
r
+
_
2
(r
4
)
r
2
+ k
2
4,2
(r
4
)
__
r=a
3
, (68)
_
3
+
(r
3
)
r
_
r=a
3
=
_
4
+
(r
4
)
r
_
r=a
3
. (69)
At r = a
4
__
2
4

2
4
r
2

4
k
2
4,1
4
_
+ 2
4
_
3
(r
4
)
r
3
+ k
2
4,2
(r
4
)
r
__
r=a
4
= p
5
|
r=a
4
, (70)
2
4
a
4
_
_
4
r

4
r
_
+
_
r
4
r
2
+

4
r
+
_
rk
2
4,2
2

1
r
_
4
__
r=a
4
= 0, (71)
2
4
a
4
_
_
4
r

4
r
_
+
_
r
4
r
2
+

4
r
+
_
rk
2
4,2
2

1
r
_
4
__
r=a
4
=
1
2
p
5
r
r=a
4
. (72)
At r = a
5
__
2
6

2
6
r
2

6
k
2
6,1
6
_
+ 2
6
_
3
(r
6
)
r
3
+ k
2
6,2
(r
6
)
r
__
r=a
5
= p
5
|
r=a
5
, (73)
2
6
a
5
_
_
6
r

6
r
_
+
_
r
6
r
2
+

6
r
+
_
rk
2
6,2
2

1
r
_
6
__
r=a
5
= 0, (74)
2
4
a
5
_
_
6
r

6
r
_
+
_
r
6
r
2
+

6
r
+
_
rk
2
6,2
2

1
r
_
6
__
r=a
5
=
1
2
_
p
5
r
_
r=a
5
. (75)
416
The following eigenfunctions are obtained for the spheroidal modes by substituting Equations
(54) into Equations (44):
U
i
(r, ) =
A
4i3
_
1
r
j
n
_
k
i,1
r
_
_
+ A
4i2
_
1
r
y
n
_
k
i,1
r
_
_
+
A
4i1
_
(n+1)
r
j
n
_
k
i,2
r
_
k
i,2
j
n+1
_
k
i,2
r
_
_
+
A
4i
_
(n+1)
r
y
n
_
k
i,2
r
_
k
i,2
y
n+1
_
k
i,2
r
_
_
_
(n + 1) cot P
m
n
(cos ) +
nm+1
sin
P
m
n+1
(cos )
_
, (76)
V
i
(r, ) =
A
4i3
_
m
r sin
j
n
_
k
i,1
r
_
+ A
4i2
_
m
r sin
y
n
_
k
i,1
r
_
+
A
4i1
_
m
r sin
_ _
(n + 1) j
n
_
k
i,2
r
_
k
i,2
rj
n+1
_
k
i,2
r
_
+
A
4i
_
m
r sin
_ _
(n + 1) y
n
_
k
i,2
r
_
k
i,2
ry
n+1
_
k
i,2
r
_
P
m
n
(cos ) , (77)
W
i
(r, ) =
A
4i3
_
n
r
j
n
_
k
i,1
r
_
k
i,1
j
n+1
_
k
i,1
r
_
+
A
4i2
_
n
r
y
n
_
k
i,1
r
_
k
i,1
y
n+1
_
k
i,1
r
_
+
A
4i1
_
n(n+1)
r
j
n
_
k
i,2
r
_
_
+ A
4i
_
n(n+1)
r
y
n
_
k
i,2
r
_
_
P
m
n
(cos ) , (78)
U
5
(r, ) =
1
2
_
A
17
_
1
r
j
n
(k
5,3
r)
_
+ A
18
_
1
r
y
n
(k
5,3
r)
__
_
(n + 1) cot P
m
n
(cos ) +
nm+1
sin
P
m
n+1
(cos )
_
, (79)
V
5
(r, ) =
1
2
_
A
17
_
m
r sin
j
n
(k
5,3
r)
+ A
18
_
m
r sin
y
n
(k
5,3
r)
_
P
m
n
(cos ) , (80)
W
5
(r, ) =
1
A
17
_
n
r
j
n
(k
5,3
r) k
5,3
j
n+1
(k
5,3
r)
+
A
18
_
n
r
y
n
(k
5,3
r) k
5,3
y
n+1
(k
5,3
r)
P
m
n
(cos ) , (81)
U
6
(r, ) =
_
A
19
_
1
r
j
n
(k
6,1
r)
_
+ A
20
_
(n+1)
r
j
n
(k
6,2
r) k
6,2
j
n+1
(k
6,2
r)
__
_
(n + 1) cot P
m
n
(cos ) +
nm+1
sin
P
m
n+1
(cos )
_
, (82)
V
6
(r, ) =
A
19
__
m
r sin
_
j
n
(k
6,1
r)
+ A
20
_
m
r sin
_
[(n + 1) j
n
(k
6,2
r) k
6,2
rj
n+1
(k
6,2
r)]
P
m
n
(cos ) , (83)
W
6
(r, ) =
A
19
_
n
r
j
n
(k
6,1
r) k
6,1
j
n+1
(k
6,1
r)
+
A
20
_
n(n+1)
r
j
n
(k
6,2
r)
_
P
m
n
(cos ) , (84)
where i = 1, 2, 3, 4.
For the investigation of torsional modes we formulate the boundary and continuity conditions
which express the balance between the tangential stresses in the solid layers as well as the
equality of the corresponding tangential displacements on junctions of the layers. The outer
layer is assumed to be free from tangential stress components. Using these assumptions
together with Equations (53) we obtain the following boundary and continuity conditions:
At r = a
1
a
_
1
r

1
r
_
r=a
= 0. (85)
At r = a
1
1
a
1
_
1
r

1
r
_
r=a
1
=

2
a
1
_
2
r

2
r
_
r=a
1
, (86)
1
|
r=a
1
=
2
|
r=a
1
. (87)
At r = a
2
2
a
2
_
2
r

2
r
_
r=a
2
=

3
a
2
_
3
r

3
r
_
r=a
2
, (88)
2
|
r=a
2
=
3
|
r=a
2
. (89)
At r = a
3
3
a
3
_
3
r

3
r
_
r=a
3
=

4
a
3
_
4
r

4
r
_
r=a
3
, (90)
3
|
r=a
3
=
4
|
r=a
3
. (91)
At r = a
4
4
a
4
_
4
r

4
r
_
r=a
4
= 0. (92)
For the torsional modes, the eigenfunctions are as follow for i = 1, 2, 3, 4:
U
i
(r, ) =
_
B
2i1
_
m
a sin
j
n
_
k
i,2
r
_
_
+ B
2i
_
m
a sin
y
n
_
k
i,2
r
_
__
P
m
n
(cos ) , (93)
V
i
(r, ) =
_
B
2i1
_
1
a
j
n
_
k
i,2
r
_
_
+ B
2i
_
1
a
y
n
_
k
i,2
r
_
__
_
(n + 1) cot P
m
n
(cos ) +
n m + 1
sin
P
m
n+1
(cos )
_
, (94)
W
i
(r, ) = 0. (95)
418
i
( Pa)
i
( Pa)
i
_
kg/m
3
_

i
( m/s)
i
( m/s)
Crust 33.4 10
9
36.4 10
9
2.6 10
3
6.40 10
3
3.75 10
3
Upper
Mantle
-1
110 10
9
85.3 10
9
3.7 10
3
8.70 10
3
4.80 10
3
Upper
Mantle
-2
156 10
9
130 10
9
4.0 10
3
10.2 10
3
7.50 10
3
Lower
Mantle
343 10
9
238 10
9
5.0 10
3
12.8 10
3
6.90 10
3
Outer
Core
783 10
9
0 11.1 10
3
8.4 10
3
0
Inner
Core
115 10
10
206 10
9
12.9 10
3
11.0 10
3
4.00 10
3
Table 1. Physical properties of the moons layers and the phase velocities
i
and
i
of the P-
and S- waves respectively.
Consider the following ctional example: In the course of the Avatar mission it was found
that the moon Pandora has a spherical layered structure with ve concentric layers and
an inner core. It was determined that the outer radius of the moon is a = 6371 km; the
boundary between the upper layer (crust) and the upper mantle-1 is situated at radius
a
1
= 6336 km; the boundary between the upper mantle-1 and upper mantle-2 is at radius
a
2
= 5961 km; the boundary between the upper mantle-2 and lower mantle is at radius
a
3
= 5711 km; the boundary between the lower mantle and the outer core is at radius
a
4
= 3482 km; and nally, the boundary between the outer core and inner core is at
radius a
5
= 1217 km. It is assumed that the outer layers (crust, upper mantle-1, upper
mantle-2, lower mantle and the inner core) are solids and the outer core is liquid,
which does not support shear stresses (i.e. it is simulated by the model of an inviscid ideal
uid). Table 1 gives the physical properties of the moon as well as the phase velocities of the
P-wave
i
=
i
+ 2
i
i
(96)
and the S-wave
i
=
_
i
, (97)
i = 1, , 6 where
i
,
i
are Lame coefcients (keeping in mind that
5
=
5
the bulk modulus
with
5
= 0) and
i
are mass densities of the layers.
For the spheroidal mode of vibration, for each pair of wave numbers n and m, by using
Equations (54) and applying the boundary and continuity conditions Equations (56), ,(75)
we obtain a matrix equation
M
A
1
.
.
.
A
20
0
.
.
.
0
(98)
5.1 Example
where the 20 20 matrix M = M() =
_
a
i,j
()
_
i,j=1,,20
. Linear algebra tells us that
Equation (98) has a nontrivial solution if the determinant
det (M()) = 0. (99)
Fig. 2. A graph of the function g() = ln |det (M())|. The vertical "spikes" point towards
roots of the function f () = M(). It appears that "good guesses" for the rst and second
eigenvalues are
1
0.002 rad. s
1
and
2
0.004 rad. s
1
respectively.
Determining the roots of the function
f () = det (M()) (100)
numerically yields a sequence of eigenvalues
1
<
2
< (and hence periods of vibration
T
1
> T
2
> where T
j
=
2
j
). For instance, with n = m = 2, a graph of the function
g() = ln | f ()| (101)
will be similar to that given in Figure 2. The vertical "spikes" indicate roots of the function
f (). It appears that "good guesses" for the rst and second eigenvalues are
1

0.002 rad. s
1
and
2
0.004 rad. s
1
respectively. These "good guesses" are then used in the
numerical routine of a computer algebra system, producing
1
and
2
to, say, four signicant
gures of accuracy and hence yield the values T
1
=
2
1
= 68.7 min and T
2
= 24.3 min as
given in Table 2. Once a suitable number of eigenvalues have been determined, by choosing a
xed eigenvalue
k
, the matrix M(
k
) has known constant entries. Hence it is now possible
to determine a non-trivial solution to
M(
k
)
A
1
.
.
.
A
20
0
.
.
.
0
. (102)
420
n m
T
1
( min)
1
T
2
( min)
2
T
3
( min)
3
T
4
( min)
4
T
5
( min)
5
2 2
68.7
0.86
24.3
0.63
15.7
0.14
14.4
0.24
9.2
0.17
3 2
41.2
0.47
17.7
0.51
13.2
0.05
10.3
0.10
7.8
0.13
3 3
41.2
0.70
17.7
0.76
13.2
0.08
10.3
0.14
7.8
0.19
4 2
28.2
0.25
14.4
0.43
11.8
0.03
7.9
0.04
7.1
0.12
4 3
28.2
0.44
14.4
0.64
11.8
0.04
7.9
0.06
7.1
0.19
4 4
28.2
0.58
14.4
0.86
11.8
0.05
7.9
0.08
7.1
0.25
5 2
21.0
0.20
12.5
0.32
10.5
0.06
6.7
0.07
6.5
0.02
5 3
21.0
0.30
12.5
0.48
10.5
0.09
6.7
0.11
6.5
0.03
5 4
21.0
0.40
12.5
0.64
10.5
0.12
6.7
0.14
6.5
0.04
5 5
21.0
0.50
12.5
0.80
10.5
0.15
6.7
0.18
6.5
0.05
Table 2. Period of oscillation T
j
and Bryans factor
j
of the spheroidal modes of vibration for
wave numbers n and m.
Consequently the eigenfunctions given by Equations (76) to (84) are completely determined.
Keeping Equations (40) and (42) in mind, Bryans factor is calculated by using
=
2
_
0
_
6
i=1
_
a
i1
a
i
[
i
(U
i
cos +W
i
sin ) V
i
] r
2
dr
_
sin d
_
0
_
6
i=1
_
a
i1
a
i

i
_
U
2
i
+ V
2
i
+W
2
i
_
r
2
dr
_
sin d
, (103)
where a
0
= a and a
6
= 0.
For the spheroidal modes of vibration, the periods of oscillation T
j
=
2
j
as well as the
corresponding Bryans factors
j
for a pair of wave numbers m and n are given in Table 2.
Doing a similar analysis (as explained above) for the torsional modes, the eigenfunctions given
by Equations (93), , (95) as well as the eigenvalues are completely determined by applying
the boundary and continuity conditions given by Equations (85), , (92) to Equations (55).
Here Equation (103) for Bryans factor may be simplied, because the radial displacements
are zero W
i
= 0. Bryans factor for the torsional modes is calculated by using the formula
TM
=
2
_
0
_
6
i=1
_
a
i1
a
i
U
i
V
i
r
2
dr
_
sin cos d
_
0
_
6
i=1
_
a
i1
a
i

i
_
U
2
i
+ V
2
i
_
r
2
dr
_
sin d
. (104)
Values of the periods of oscillation T
j
and the corresponding Bryans factors
j
for the n and
m torsional modes are given in Table 3.
n m
T
1
( min)
1
T
2
( min)
2
T
3
( min)
3
T
4
( min)
4
T
5
( min)
5
2 2
42.4
0.33
11.9
0.33
7.2
0.33
5.0
0.33
3.7
0.33
3 2
27.2
0.17
11.0
0.17
7.0
0.17
4.9
0.17
3.7
0.17
3 3
27.2
0.25
11.0
0.25
7.0
0.25
4.9
0.25
3.7
0.25
4 2
20.7
0.10
10.0
0.10
6.8
0.10
4.8
0.10
3.6
0.10
4 3
20.7
0.15
10.0
0.15
6.8
0.15
4.8
0.15
3.6
0.15
4 4
20.7
0.20
10.0
0.20
6.8
0.20
4.8
0.20
3.6
0.20
5 2
16.9
0.07
9.1
0.07
6.5
0.07
4.7
0.07
3.5
0.07
5 3
16.9
0.10
9.1
0.10
6.5
0.10
4.7
0.10
3.5
0.10
5 4
16.9
0.13
9.1
0.13
6.5
0.13
4.7
0.13
3.5
0.13
5 5
16.9
0.17
9.1
0.17
6.5
0.17
4.7
0.17
3.5
0.17
Table 3. Period of oscillation T
j
and Bryans factor
j
of the torsional modes of vibration for
wave numbers n and m.
6. Conclusions and discussions
After deriving expressions for Bryans factor in terms of eigenvalues and eigenfunctions of
vibration we demonstrated that neither Bryans effect nor the value of Bryans factor depend
on the inclusion of light, isotropic, viscous damping in the model. Consequently we neglected
damping in the model that we presented. However, we pointed out that preliminary studies
indicate that the inclusion of "impurities" such as mass-stiffness, prestress and anisotropic
damping into the model appears to produce "capture effects" such as the rotation angle of
the vibration pattern varying periodically. Hence more realistic models (as opposed to ideal
situations) that will include these "impurities" need to be developed.
Solutions were obtained for the dynamic equations of slowly rotating, vibrating planets or
moons considered as ideal elastic solid and inviscid acoustic bodies composed of, for instance,
four solid layers, one liquid layer and a solid core. Boundary conditions were formulated for
calculating the eigenvalues and eigenfunctions of vibration and a brief discussion of how they
are numerically calculated was given. It was pointed out that for each pair of wave numbers
n and m, there is a sequence of eigenvalues
1
<
2
< and hence periods of vibration
T
1
> T
2
> (where T
j
=
2
j
).
The model derived indicates that there are two modes of vibration, namely the "spheroidal"
and "torsional" modes and Bryans effect occurs for both modes.
Table 1 provides feasible physical properties for a ctional moon and gives possible phase
velocities for both P- and S-waves.
422
Using Table 1, a numerical experiment produced Table 2, where it appears that for the
spheroidal modes, for a given polar wave number n and a given non-negative integer j, the
period of oscillation T
j
does not depend on the circumferential wave number m. However,
Bryans factor depends on both wave numbers n and m as well as the period of oscillation
T
j
for each pair n and m. Bryans factor appears to be negative in all cases and this means
that an antinode of vibration will rotate through the moons crusts in the opposite direction
to the moons rotation at a rate , where is the moons rotation rate.
Another numerical experiment produced Table 3, where it appears that for torsional modes,
for a given wave number n and a given non-negative integer j, the period of oscillation T
j
is independent of the circumferential wave number m and that it varies with n. However,
Bryans factor appears to be invariant and independent of the period of oscillation for a
given pair of wave numbers n and m, but appears to vary as each pair varies and it appears to
be positive in all cases. Consequently Bryans factors for the torsional modes depend only on
angular components of the vibrating patterns, that is, on the corresponding wave numbers n
and m, and not on radial vibrations because all radial displacements W
i
= 0. These positive
Bryans factors indicate that the antinodes of vibration will rotate through the moons crusts
in the same direction as the moons rotation at a rate , where is the moons rotation rate.
7. Acknowledgement and disclaimer
This material is based upon work supported nancially by the Tshwane University of
Technology (TUT), the Council for Scientic and Industrial Research (CSIR) of South Africa
and the National Research Foundation (NRF) of South Africa (NRF grant reference number
EV2009011 400005). Any opinions, ndings and conclusions or recommendations expressed
in this material are those of the authors and therefore the TUT, the CSIR and the NRF do not
accept any liability in regard thereto.
8. References
Bryan, G.H. (1890). On the beats in the vibrations of a revolving cylinder or bell, Proceedings of
the Cambridge Philosophical Society 7, pp. 101-111.
Eringen, A.E. & Suhubi, E.S. (1975). Elastodynamics, vol. I1. Academic Press, New York.
Goldstein, H., Poole, C. &Safko, J. (2001). Classical mechanics, 3rd ed, Addison-Wesley, Reading,
MA.
Loveday, P.W & Rogers, C.A. (1998). Free vibrations of elastically supported thin cylinders
including gyroscopic effects, Journal of Sound and Vibration 217(3), pp. 547-562.
Nayfeh, A.H. (1973). Perturbation methods, John Wiley & Sons, Inc., New York.
Rayleigh, J.W.S. (1894). The theory of sound, volume I, second edition, MacMillan & Company,
London.
Redwood, M. (1960). Mechanical waveguides, Pergamon Press, Oxford.
Rozelle, D.M. (2009). The hemispherical resonator gyro: From wineglass to the planets,
Proceedings of the 19th AAS/AIAA Space Flight Mechanics Meeting, February 8-12,
Savannah, Georgia, USA, pp. 1157-1178.
Shatalov, M.Y., Joubert, S.V., Coetzee, C.E. & Fedotov, I. (2009). Free vibration of rotating
hollow spheres containing acoustic media, Journal of Sound and Vibration, 322(4-5),
pp. 1038-1047.
Available online: doi:10.1016/j.jsv.2008.11.020, accessed July 2011.
Shatalov, M.Y., Joubert, S.V. & Coetzee, C.E. (2011). The inuence of mass imperfections on the
evolution of standing waves in slowly rotating spherical bodies, Journal of Sound and
Vibration, 330, pp. 127135.
Available online: doi:10.1016/j.jsv.2010.08.001, accessed July 2011.
Spiegel, M.R. (1967). Theoretical mechanics. McGraw-Hill, New York.
Zhuravlev, V. & Klimov, D. (1988). Applied methods in the theory of oscillations. Moscow: Nauka
(in Russian).
424

Advances in Geotechnical Earthquake Engineering - Soil Liquefaction and Seismic Safety of Dams and Monuments

Uploaded by

Copyright:

Available Formats

Advances in Geotechnical Earthquake Engineering - Soil Liquefaction and Seismic Safety of Dams and Monuments

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Advances in Geotechnical Earthquake Engineering - Soil Liquefaction and Seismic Safety of Dams and Monuments

Uploaded by

Copyright:

Available Formats

ADVANCES IN

Normal Slow Down Slow Down

> x x ), are equal to Eqs.(6)

= + y x x ( 0 < x not depending on time, see Fig.3(c)). By

b b = when the relative displacement

b b = (se Fig.3(b)); if the relative displacement

y x x = + ( 0 x < , see Fig.3(c)). There is a favoured direction for

x x > , the force

has to be equal to the angular momentum

r b h = + . The conservation of the linear momentum is needed to relate u

b b = (see Fig.3(b)); if the relative

y x x = + ( 0 x < , see Fig.3(c)). Taking Eq.(21) and

completely vanishes since it is partially dissipated and

and, by referring to Fig.3(a),

0.1 x m = and the restitution coefficients

x x = above which the body is partially out of

External force vector

Gravitational acceleration vector

Diffusive flux vector of chemical species j

(in the direction of

(in the direction of increasing ). Let w

You might also like