Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
,
,
-
-
!
"
# " $% $&
$% $' ( ') * '
Proceedings of International Workshop on Seismic Requalification of
Geotechnical Structures (SRGS), New Delhi, 17th December 2012
Seismic Requalification of
Geotechnical Structures
(Draft version)
Edited by
Subhamoy Bhattacharya
University of Surrey and University of Bristol, UK
A.Murali Krishna
Indian Institute of Technology Guwahati, India
Workshop Organisers:
Organising Committee:
Patrons
Director
IIT-Delhi
Prof. K. S. Rao,
President, Indian Geotechnical Society,
Indian Institute of Technology Delhi, India.
Conveners
Dr A. Murali Krishna
IIT Guwahati, India
Prof. Subhamoy Bhattacharya
University of Bristol & University of Surrey, UK
Prof. Chandan Ghosh
Honorary Secretary, Indian Geotechnical Society
NIDM, New Delhi
Advisory Committee:
Prof. Deepankar Choudhury, IIT Mumbai, India
Dr. Barnali Ghosh, Mott MacDonald Limited, London, UK
Dr. Tiziana Rossetto, University College London, UK
Dr. Ayothiraman, IIT Delhi, India
Mr. Jai Bhagawan, CRRI, Chairman, IGS Delhi chapter,
Mr. Ravisundaram, Cengrs Geotechnica Pvt. Ltd., New Delhi, India
Mr. Manish Gupta, CSMRS, New Delhi India
Prof. MR Madhav, Former Vice President, ISSMGE, JNTU / IIT Hyderabad.
Prof. Madhavi Latha, IISc Bangalore
Prof. SC Dutta, Indian Institute of Technology Bhubaneswar, India
Prof. Bhoominathan, IIT Madras
Prof. BK Maheshwari, IIT Roorkee
Dr. Suresh R Dash, Indian Institute of Technology Bhubaneswar, India
Dr. Rajib Sarkar, University of Bristol, UK
UKIERI International Workshop on Seismic Requalification of Geotechnical Structures (SRGS)
17th December 2012, Delhi.
Preface
This workshop is the first of its kind as is a part of UKIERI [UK India
Education and Research Initiative project number UKUTP201100296] targeted
on the theme 'Seismic Requalification of Geotechnical Structures' where by
design and construction of pile foundations and retaining walls are considered.
The project started in April 2012 and the duration is 24 months making the end
date as March 2014. We aim to conduct a series of workshop to facilitate the
dissemination of the knowledge for the profession mainly in the two areas.
1. What are the ground response for such a scenario earthquake i.e. will be
ground liquefy or will there be site amplification? Will there be lateral
spreading of the ground? [See Section C of the proceedings]
2. If liquefaction is a problem, how long will the ground remain liquefied
and how much time will it take to liquefy?
3. For estimating the time to liquefaction and how long the soil will stay
liquefied, we need to generate synthetic ground motion. [See Section C of
the proceedings]
4. What are the plausible failure mechanisms of the pile foundation? [See
Section B of the proceedings]
5. Estimate the probability of failure [Probabilistic approach] or determine
the factor of safety against a particular failure mechanism.
6. If the hospital is unsafe - can it be retrofitted? What are the options? –
ground improvement, add extra piles?
We are happy to coordinate the workshop which delivers talks on diverse topics
related to the theme, addressing the above questions. The proceedings is aimed
at collating the necessary information and provide a comprehensive guide for
such purpose. We thank all the contributing speakers for their time and effort to
sharing their technical experience and expertise. The paper titled "Probabilistic
buckling analysis of axially loaded piles in liquefiable soils" provided in the
proceedings gives a method to carry out such assessment.
It must be mentioned that this is a draft copy. Through this workshop, we aim to
gather feedback, collate the information and bring out a book or a special issue
of the IGS journal in due course. These papers will be peer reviewed. We would
like to record our appreciation to various organisations, in particular Indian
Geotechnical Society, National Institute for Disaster Management, IIT Delhi
and of course IIT Guwahati, Universities of Bristol and Surrey (UK).
Table of Contents
SECTION A:
1. Seismic Requalification of Geotechnical Structures:
This article provides some background behind the project
2. Engineers versus nature
This is essentially the genesis behind the project!
3. Liquefaction of soil in the Tokyo Bay area from the 2011 Tohoku (Japan) earthquake,
Soil Dynamics and Earthquake Engineering, Vol 31, 2011, pp 1628-1628.
This provides the lessons learnt from the magnitude 9.0 earthquake.
In the final proceedings, we aim to collate information from many earthquakes.
SECTION B:
This section aims to provide a comprehensive review of the latest understanding on the
behavior of piles and retaining walls under earthquake loading. The following papers are
collated:
Piles: There are three papers
4. Soil-Structure Interaction Considerations in the Seismic Analysis of Pile Foundations
5. Lateral Dynamic Response and Stiffness of Full Scale Single Piles
6. Seismic design of piles in liquefiable soils
Retaining wall: There are two papers
7. Seismic behaviour of earth retaining structures and design issues
8. A solution to seismic response of cantilever retaining wall with generalized backfill
soil
Section C
This section collates the ingredients necessary for back-analysis and seismic requalification.
They are:
9. Seismic Hazard Assessment and Microzonation
10. Generation of synthetic ground motion for site response analysis
11. Assessment of liquefaction potential at the site
12. Geotechnical and geophysical testing required for site characterization
13. Methodology for assessment of pile foundations
Section D
This section provides some guidelines for seismic strengthening and retrofitting
SECTION A
UKIERI International Workshop on Seismic Requalification of Geotechnical Structures (SRGS)
17th December 2012, Delhi.
ABSTRACT:
The most recent earthquake of Japan (see Bhattacharya et al, 2011 for details of the 2011
Tohoku earthquake) demonstrated the need of seismic requalification of the important
structures (hospitals, transportation system, power plants etc which were built earlier) in the
light of revised understanding of the behaviour of built environment under earthquake
loading. Extent of earthquake damage to build environment (structures) depends not only on
the details of the structure but also on the quality of the ground on which the structure is built.
It is well established that geotechnical aspects i.e. soil can be a major contributor to damages
to the built environment. It is therefore important to consider various geotechnical structures.
This workshop is focused on requalification of piles and retaining walls. This workshop has
the following aims:
(a) Narrate the background behind the workshop and the project
(b) Provide a comprehensive overview of the up-to-date understanding of the behaviour of
piles and retaining walls in seismic areas
(c) Describe the methodology and ingredients necessary for seismic re-qualification studies
(d) Present the various method of seismic strengthening and retrofitting
India is currently experiencing a rapid development in infrastructure and many parts of the
country lie in a high seismicity zone. Figure 1 shows a predicted PGA motion for South Asia
obtained from the GSHAP studies. Figure 2 shows the seismic hazard map of India and
adjoining regions for 10% probability of exceedance in 50 years having a contour interval
0.05g based on GSHAP. Details of the studies can be found in Bhatia et al (1997). As will be
shown in this chapter that the subject of earthquake geotechnical engineering is new and
evolving. This calls for re-examining the structures and foundations designed and built earlier
in the light of the new understanding.
It is recognised that there are not yet enough books available in the subject to bridge the gap
1
A-2 SRGS, 17th December 2012, Delhi
Figure 1: Predicted PGA for South Asia, after Bhatia et al, 1997
Figure 2: Seismic hazard map of India and adjoining regions for 10% probability of exceedance in
50 years. (Contour interval 0.05g) after Bhatia et al, 1997
Seismic requalification of Geotechnical Structures A-3
The structural failures and loss of life in recent earthquakes have shown the shortcomings of
current design methodologies and construction practices. Post earthquake reconnaissance
investigations have led to improvements in engineering analysis, design and construction
practices. A brief historical development of earthquake engineering practice illustrating how
earthquake engineers have learned from failures in the past is outlined in Table 1 following
and adapted from Bhattacharya et al (2011). Table 1 shows that earthquake geotechnical
engineering subject is very new. Soil liquefaction has been identified in 1964 and the
knowledge of soil amplification was noticed and came into practice after the 1985 Mexico
earthquake. Figure 1 shows two contrasting photographs of damages to buildings observed
during the 2011 Bhuj earthquake. Figure 3(a) shows a building collapse where structural
inadequacies was responsible. On the other hand, the failure in Figure 3(b) is due to the
geotechnical aspects i.e. soil is a major contributor.
Table 1: Development of Earthquake Geotechnical Engineering Practice
1964 Niigata earthquake Soil can also be a major contributor of damage. Soil liquefaction studies started.
(Japan)
1971 San Fernando Bridges collapsed, dams failed causing flood. Liquefaction studies intensified. Bridge retrofit
earthquake (USA) Soil effects observed. studies started.
1994 Northridge earthquake Steel connections failed in bridges. Importance of ductility in construction realised.
(USA)
1995 Kobe earthquake Massive foundation failure. Soil effects were the Downward movement of a slope (lateral spreading) is
(Japan) main cause of failure. said to be one of the main causes. JRA (1996) code
Kawashima (2000) modified (based on lateral spreading mechanism) for
design of bridges.
1999 Chi-Chi earthquake Many bridges collapsed as they were located Special care must be undertaken while designing
(Taiwan) close to the faults important structure in the vicinity of the plate
boundaries and faults
1999 Koceli earthquake Damage to Bolu tunnel due to fault movement. Research on fault induced failures intensified.
(Turkey) Damage to buildings and bridges. However,
buildings conforming the design codes
performed well
2001 Bhuj earthquake Large scale destruction. Good performance of Large diameter piles concrete performed better than
(India) some new jetties of the Kandla port. Tilting of small diameter piles. Steel piles filled in concrete also
the Kandla Tower building without any damage performed better.
2004 Sumatra earthquake Destruction to built environment due to Research on Tsunami warning started
earthquake and giant tsunami waves
2011 Tohoku earthquake Complete collapse of pile-supported buildings in
coastal areas due to tsunami
A-4 SRGS, 17th December 2012, Delhi
Figure 3 Damages caused during 2001 Bhuj earthquake; (a) Failure due to structural
inadequacies; (b) Damages due to soil effects
A devastating earthquake of moment magnitude Mw9.0 struck the Tohoku and Kanto regions
of Japan on 11th March at 2:46PM. The earthquake caused great economic loss, loss of lives
and tremendous damage to structures and infrastructures. This is the largest earthquake ever
recorded in Japan and one of the five most powerful earthquakes in the world since modern
record-keeping began in 1900. According to the Japan Meteorological Association (JMA), the
epicentre was located about 150km off-shore from Sendai (largest city in the Tohoku region)
and at a depth of about 24km (Figure 4). Details of various aspects of the earthquake can be
found in Bhattacharya et al (2011). Intense tremor was felt across coastal areas of the Tohoku
and Kanto regions (6 to 7 magnitude based on JMA seismic intensity scale). Extensive
damage was caused by massive tsunami in many cities and towns along the coast. At many
locations (e.g. Natori, Oofunato and Onagawa), tsunami heights exceeded 10m, and sea walls
and other coastal defence systems failed to prevent the disaster. The earthquake and its
associated effects (i.e. tsunami) also initiated the crisis of the Fukushima Dai-ichi nuclear
power plants. In Tokyo, which is about 350-400km southwest from the epicentre, relatively
intense tremor of JMA scale 5 was registered. This earthquake has triggered and induced
many large aftershocks with Mw greater than 6.0, which further caused additional damage to
structures that survived during the mainshock, see Table 2 for some details. It is expected that
the increased seismicity will be relatively high in the next several months to a few years and
impose additional threat to residents in the affected regions. Figures 5 and 6 shows some
damages caused by the earthquake.
The paper by Bhattacharya et al (2011) is added in this section for ready reference
Seismic requalification of Geotechnical Structures A-5
Figure 4: Map of the Tohoku and Kanto regions and recorded ground motion data at three K-
NET stations
Figure 5: (a) Devastation of a Coastal City during the 2011 Japan Earthquake; (b) Damage to a
pile-supported building in Onagawa
Immediately following the earthquake, a field investigation was carried out along the coastline
of Tokyo Metropolitan Government and Chiba Prefecture. The area of survey was limited to
Urayasu, ShinKiba, Chiba, Kinshicho, Haneda airport, Tokyo Disneyland and Higashisuna
A-6 SRGS, 17th December 2012, Delhi
along the river Naka (see Figure 7a). No apparent failure and collapse of large superstructures
were observed, and therefore, the damage was dominated by soil liquefaction. Surface
liquefaction was observed at many locations. The streets and car parks were filled up with
solid mixture of brown and/or grey colour fine particles materials and water. Clearly they
erupted from underneath the ground and are mostly fill material and/or young alluvium
(Figure 7b).
The lessons learnt from the 2011 Tohoku earthquake can be summarized as follows:
Ground improvement helps to mitigate liquefaction, for further details see
Bhattacharya et al (2011)
Research is necessary for designing piled buildings in tsunami prone areas. This is due
to the fact that large number of pile-supported buildings collapsed, see Figure 5(b)
Retaining walls on fill land suffered damage
There is a need to re-assess the safety of critical infrastructure
2.1 Figure 8 shows the photograph of retrofitting work in progress at the old Tokyo
subway station. The load bearing columns are encapsulated by a steel jacket.
Section A provides the background behind the workshop. This section also reviews the
observed geotechnical damages from the past earthquakes. The following papers are appended
in this section:
S. Bhattacharya, M. Hyodo, K. Goda, T. Tazoh, C.A. Taylor (2011): Liquefaction of soil in the Tokyo Bay area
from the 2011 Tohoku (Japan) earthquake, Soil Dynamics and Earthquake Engineering, Vol 31, 2011, pp 1628-
1628.
Section C collates the ingredients necessary for back-analysis and seismic requalification.
They are:
(i) Seismic Hazard Assessment and Microzonation: This will provide us with the
scenerio earthquake
(ii) Generation of synthetic ground motion for site response analysis
(iii) Assessment of liquefaction potential at the site
(iv) Geotechnical and geophysical testing required for site characterization
(v) Methodology for assessment of pile foundations
REFERENCES
S. Bhattacharya, M. Hyodo, K. Goda, T. Tazoh, C.A. Taylor (2011): Liquefaction of soil in the Tokyo Bay area
from the 2011 Tohoku (Japan) earthquake, Soil Dynamics and Earthquake Engineering, Vol 31, 2011, pp
1618-1628
Bhatia S C , H.K. Gupta, M. Ravi Kumar, N.P. Rao, G.R. Chitrakar, P. Zhang and Z-X Yang, 1997. Seismic
hazard map of GSHAP Area XII comprising eastern Himalaya, Northeastern India, Burmese arc, South
China and adjoining regions, Abstract, 29th General assembly of IASPEI, August 18-28, 1997,
Thessaloniki, Greece, p387.
I was on the 9th Floor of a 20-storey
Tokyo hotel when the 9.0 magnitude
earthquake hit Japan earlier this year.
This was one of the five most powerful
earthquakes in the world since modern
record keeping began in 1900, with an
unusually long duration of around three
minutes. At 2.46pm I started to feel the
building shaking. At first I thought it
might be a mild one, but soon the violent
shaking started. Based on my watch and
crude measurements, I realised that this
was a low frequency vibration - my guess
was 0.5 to 1 cycles per second, which
indicated that the epicentre was not
very close. I peeped through the window
and the swaying movement of a crane
nearby confirmed that this was big.
I could hear the partition walls creaking,
Bristol Engineering
but the connections were so strong and
well-designed that nothing happened.
Even the photographs on the wall did
not fall off. I felt quite safe throughout,
Engineers
and continuously thanked Japanese
engineering! Had this happened in
the developing world, there is a strong
LynchPin
likelihood that the hotel would have
versus Nature
collapsed, as plenty of earthquakes in the
last decade have demonstrated.
.
The lift was switched off and public
announcements in Japanese began. With
the help of a Japanese student I met in
the corridor, I grasped that the message Dr Subhamoy Bhattacharya was in Tokyo
was to stay in my room. After about 30
minutes, I was led out through a stairwell when the earthquake struck in March this year 17
with solid concrete walls to the lobby. The
lobby was the safest place to be as there up. I also went to the Oimachi train bay area. I had heard that extensive
were so many tall buildings in the vicinity, station but found it deserted and all the soil liquefaction was observed in the
any of which could in theory have collapsed. trains were cancelled. reclaimed area of Tokyo bay area.
Within an hour, the lifts were working, and
things got back to normal: power, hot water, I was expecting to meet Professor I also reflected, as a civil engineer, on
and internet all continued working during Masayuki Hyodo from Yamaguchi the fact that my hotel building and many
the strong shaking! University at 8pm. I was unable to ring other buildings in Tokyo showed that it
him as the entire mobile network was is possible to design even non-structural
Still, the hotel manager advised us not to down so I went out for dinner in a nearby elements (e.g. partition walls) so that they
go our hotel rooms as they were expecting restaurant, where I found the staff and don’t fall apart. I looked carefully at my
more aftershocks. The images of the customers incredibly calm. On the street, room and it was obvious that a lot
disastrous tsunami were being broadcast people were composed, and the hotel of thought went into the design
on television and I was desperate to receptionists, police and traffic wardens and construction. I regularly visit
contact my family to let them know I was were working with a smile - my respect Japan for academic collaboration not Suby Bhattacharya
safe. My flight to London was cancelled, for the Japanese people increased to only with universities but also with observes an area of
soil liquefaction in
my UK mobile phones were not working, an even higher level than before. In the private companies. Japan has five the Tokyo bay area
I could not connect to Skype, and I could railway station at Oimachi all the trains big construction companies: Shimizu,
The 20-storey hotel
not go my hotel room. So I thought a and the subway were out of service while Obayashi, Kajima, Taisei and Takenaka, where Suby was
good way to communicate would be via railway tracks, bridges and tunnels were all of which have large research divisions. staying when the
earthquake hit
Facebook. I’m not a regular user, but I being checked (this affected around nine This is one of the reasons why the
started providing commentary on the hour- million people), but, extraordinarily, there country coped so well with this gigantic
by-hour development. One of my friends was no feeling of chaos. earthquake: buildings are built to the
rang my wife and gave her the news that latest state-of-the-art understanding of
I was in the earthquake but OK. Aftershocks continued throughout the earthquake engineering.
evening and night: I returned to my
While we were in the lobby, we hotel at 11pm and finally met Professor In my view, it would be an understatement
experienced a number of powerful Hyodo – he had walked for seven hours to say that lots of lives were saved by
aftershocks and I started to panic a little to be there! After our late meeting I Japanese engineering. The tsunami was a
after it emerged that there could be food went to bed, sleeping fully clothed, with terrible tragedy, but the earthquake itself
and water shortages. I went to a nearby my wallet and passport in my pocket, so could also have caused far greater loss of
convenience store and bought enough that I could rush out of the hotel should life. The fact that buildings like the one
food and water for a week. By this time, the need arise. On the following day, I was in, and thousands of others,
the stores were rationing supplies, but Professor Hyodo and I went for a field remained intact shows just how important
the manager kindly allowed me to stock reconnaissance survey around the Tokyo good earthquake engineering is.
Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628
Technical Note
a r t i c l e i n f o a b s t r a c t
Article history: Immediately following the 11th March 2011 Mw 9.0 Tohoku (Japan) earthquake, a field investigation
Received 14 May 2011 was carried out around the Tokyo Bay area. This paper provides first-hand observations (before or just
Received in revised form at the onset of repair) of widespread liquefaction and the associated effects. Observations related to
15 June 2011
uplift of manholes, settlement of ground, performance of buildings and bridges and the effects of
Accepted 16 June 2011
ground improvements are also presented. Recorded ground motions near the Tokyo Bay area were
Available online 7 July 2011
analysed to understand their key characteristics (large amplitude and long duration). Lessons learnt are
also presented.
Crown Copyright & 2011 Published by Elsevier Ltd. All rights reserved.
0267-7261/$ - see front matter Crown Copyright & 2011 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.soildyn.2011.06.006
S. Bhattacharya et al. / Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628 1619
Kanto area
34oN
200 s
Fig. 1. Map of the Tohoku and Kanto regions and recorded ground motion data at three K-NET stations.
Higashisuna
Shinkiba
CHB009 (K-NET)
Chiba
Tokyo Disneyland
Haneda Airport
CHB014 (K-NET)
KNG002 (K-NET)
Fig. 2. Locations of survey around the Tokyo Bay area (photo courtesy Google Maps).
destruction (e.g. 1978 Mw 7.7 earthquake). One distinct difference of The entire rupture process involved a large fault plane with
the 2011 Tohoku earthquake from previous ones is that the rupture 400–500 km in length by 100–200 km in width [10,19]. The fault
zone was significantly larger and involved multiple nearby rupture rupture models estimated by Shao et al. [19] (UCSB model) and by
zones, such as off-shore Fukushima Prefecture and off-shore Ibaraki the Geospatial Information Authority of Japan (2011) (GSI model)
Prefecture to south, and off-shore Iwate Prefecture to north (which are illustrated in Fig. 1 (note: many other fault plane models are
were usually thought to rupture individually, rather than simulta- available). These models were derived by focusing on macro
neously). In other words, the 2011 Tohoku earthquake resulted from features of observed ground deformation and long-period ground
coupled co-seismic rupture of the above-mentioned major fault motions. In particular, the GSI model approximates the overall
segments; such simultaneous ruptures have not happened since the rupture process as a sequence of two sub-processes, initiated in the
869 Jogan earthquake [16], which caused massive tsunami along the northern plane and propagated to the southern plane. Although
coastal areas of the Tohoku region. the overall extent of the estimated fault rupture area differs
1620 S. Bhattacharya et al. / Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628
significantly, depending on the analysis methods and the considered the rupture source to the site, whereas they tend to be smaller in
frequency ranges for earthquake source inversion, locations for large peak and stretched over longer duration if the rupture direction is
slips are generally similar among different approaches (i.e. an area opposite of the source-to-site orientation.
around the GSI’s rupture plane 1 in Fig. 1). A significant amount of The complex rupture process described above can be illustrated
permanent ground deformation of up to about 30 m to the east [10] by inspecting recorded acceleration time-series data (both NS and
occurred along the trench line off-shore of the Tohoku region (i.e. EW components) at three K-NET stations MYG008, FKS013 and
eastern boundary of the GSI/UCSB fault planes in Fig. 1) and this was IBR007 (see Fig. 1). At the MYG008 station, two clear phases of
mainly responsible for the generation of massive tsunami. seismic wave arrivals can be seen, whereas at the FKS013 and
The macro-rupture process based on preliminary analyses IBR007 stations, a single phase of seismic wave arrivals is featured.
conducted by various research groups generally indicates that In particular, for the IBR007 station, the acceleration amplitude or
the rupture initiated at the earthquake epicentre and propagated strong motion part is more concentrated than the other two
southwards, rupturing the segments off-shore Fukishima and stations. This is due to the combined effects of the complex rupture
Ibaraki Prefectures [10]. Moreover, a closer investigation into process with multiple strong motion generation sources and the
strong ground motion generation in the short-period range by source-to-site directivity; the rupture is initiated at the epicentre
Irikura and Kurahashi [12] suggests that the initiated rupture also and triggering of rupture moved towards south and west. By
propagated towards deeper segments along the dip and triggered considering a typical rupture propagation speed (which is about
several small patches of fault rupture with high stress drops in 80% of the seismic wave propagation speed at deeper locations),
areas between the UCSB’s western boundary and the GSI’s arrivals of the triggered seismic waves at different parts of the fault
western boundary shown in Fig. 1. (Note: high stress drops tend to be more concentrated in time, if a recording station is
usually produce intense short-period ground motions, and are located towards the direction of fault rupture process (e.g. IBR007),
typical characteristics of deep inslab earthquakes.) In particular, whereas they become more spread out if a recording station is in the
the source model by Irikura and Kurahashi [12] includes four opposite of the rupture propagation direction (e.g. MYG008). It is
small rupture segments (see grey patches shown in Fig. 1): two also noted that the observed values of peak ground acceleration
segments are placed in the west of the GSI’s northern plane (i.e. (PGA) at these K-NET stations are large, exceeding the PGA of
off-shore Miyagi Prefecture), while the other two segments are 1000 cm/s2 (at some K-NET stations, MYG004 and MYG012,
placed in the west of the GSI’s southern plane (i.e. one is off-shore recorded PGAs exceeded 2500 cm/s2). Large observed PGAs at
Fukushima Prefecture and the other is off-shore Ibaraki Prefec- several stations in Miyagi, Fukushima and Ibaraki Prefectures are
ture). These four segments are closer to the coastal areas of the partly due to proximity to localised strong ground motion genera-
Tohoku region and are at deeper locations; generated seismic tion areas along the coastline [12].
waves travel through a so-called high-Q region [15], experiencing
less attenuation of seismic waves over distance, and thus results 3. Field survey around Tokyo Bay area
in larger ground motions. Hence, the rupture and wave propaga-
tion process of the Tohoku earthquake was rather complex, Immediately following the earthquake, a field investigation was
because multiple sub-ruptures were responsible for different carried out along the coastline of Tokyo Metropolitan Government
episodes of the overall rupture process (shallow versus deep) and Chiba Prefecture. The survey area was limited to Urayasu,
and their effects on generated ground motions differ, depending ShinKiba, Chiba, Kinshicho, Haneda airport, Tokyo Disneyland and
on the frequency range of interest (high versus low). Another Higashisuna along the river Naka (see Fig. 2). No apparent failures or
important consideration in understanding a relationship between collapses of large superstructures were observed. The damage was
complex rupture process and observed ground motions is the dominated by soil liquefaction. Surface liquefaction was observed at
directivity effect. Observed ground motion time-series data tend many locations. The streets and car parks were filled up with a solid
to be greater in peak and more concentrated in duration if the mixture of brown and/or grey colour fine particle materials and
rupture propagation direction coincides with the orientation from water. Clearly these materials erupted from underneath the ground
Fig. 3. Map of the Kanto region and recorded ground motion data at four K-NET stations.
S. Bhattacharya et al. / Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628 1621
and were mostly fill materials and/or young alluvium. In the ground motion characteristics (i.e. soft soil, relatively large ground
following, observed ground motions and field investigations of motion amplitude and long duration) are the necessary conditions for
liquefaction and ground settlement around the Tokyo Bay area are liquefaction triggering [18,22]. Moreover, to inspect the frequency
summarised. content of the observed ground motion records and elastic seismic
demand characteristics, power spectral densities of two horizontal
components of the four records are shown in Fig. 4(a)–(d) Average
3.1. Observed ground motions around the Tokyo Bay area
5%-damped response spectra of the four records are shown in
Fig. 4(e). The dominant period ranges of the recorded ground motions
Typical recorded ground acceleration time-series data at four
at sites around the Tokyo Bay area were around 0.75–1.25 s, and
K-NET stations in the Tokyo Bay area are presented in Fig. 3. The
high-frequency content of these records was relatively low.
selected stations are at soft soils (typically, average shear wave
velocity in the upper 30 m VS30 is less than 200 m/s, corresponding
to the NEHRP site class D/E boundary or E). These sites are susceptible 3.2. Observed liquefaction in parks and streets
to soil liquefaction. The recorded ground motions at the four stations
are similar and have a PGA of about 100–200 cm/s2. Note that the Fig. 5 shows the condition of the 10-lane road to Tokyo Disney-
duration of the strong ground motion part (or significant cyclic load land and parking as noted on the 12th March morning before the
reversals) is rather long, because of the large size of the event. These repair work commenced. At one location the road surface buckled,
Fig. 4. Power spectral density of ground motion data (a–d) and 5%-damped acceleration response spectra of ground motion data (e) at four K-NET stations in the Tokyo Bay area.
1622 S. Bhattacharya et al. / Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628
Fig. 7. Observed liquefaction in the park (left) and parking area of Disneyland (right).
and other surface liquefied soil can be seen at the surface, while in material. The type of ejected materials varied from location to
the parking area widespread liquefaction was observed. Fig. 6 shows location and seemed to depend on the dates when the land
the brittle failure of a thin ‘‘black top’’ asphalt slab of the road to reclamation or ground improvement was carried out. The boiled
Tokyo Disneyland. Clearly a compressive force was developed on the material also differed from place to place and was highly
earth surface causing the weak pavement slab to buckle upwards. dependant on the specific gravity of the solid grains constituting
The possible causes may be due to a combination of vertical up- the soil. Liquefaction caused settlement, and the amount varied
down motion, dynamic surface wave propagation effects or accu- from location to location, depending on the type of ground
mulated permanent compressive strains in the surface soils. These improvement carried out in the locations. Generally, the parks
observations of buckled pavements were not very common in the incurred liquefaction disturbance more than the built up areas.
damaged area surveyed (i.e. localised), which may hint that these More discussion of ground improvement practices is given later in
zones were those where the construction of the road or compaction the paper. Fig. 8 shows the enormous quantity of ejected sand in
of the ground was less than perfect. the main roads. It mainly erupted from the edges where the road
Widespread liquefaction was observed in certain areas with met the footpath or the cycle path. It must also be mentioned that
plenty of evidence of sand boils. Fig. 7 shows photographs of sand as the duration of the earthquake was rather long (approximately
boils on the morning of 12th March. While Fig. 7(left) shows the 100–200 s, see Fig. 3), the soil was subjected to a significant
sand boils in the Takasu Park, Fig. 7(right) shows the liquefaction number of load reversals beyond that normally expected. This
in the paved car park of the Disneyland amusement park. The must have contributed to the unusually high amounts of liquefied
liquefied and ejected soil consisted of different types of materials material ejected to the ground surface. Fig. 8 shows the repair
ranging from pure sand (brown colour) with small fines content work being carried out on one of the main roads where a vast
to grey silty sand, and also in some locations dredged recycled amount of ejected liquefiable material was on the road surface.
S. Bhattacharya et al. / Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628 1623
Fig. 8. Repair works of the road along with ejected material at the surface of the road.
2 Max Mw = 8.5
Sandy silt Max Mw = 8.0
4 Max Mw = 9.0
6
8
Depth (m)
18 Sandy silt
20
Soil type 0 10 20 30 0.0 0.5 1.0
N count Liquefaction
probability
Fig. 9. Typical soil profile in Urayasu (for location see Fig. 2).
For illustration, a simplified stress-based liquefaction trigger- Gaps were formed between sidewalks of buildings and the building
ing analysis [18,22] was carried out by considering a typical soil itself, causing noticeable relative settlement. Fig. 10 shows a
profile in Urayasu, where widespread liquefaction and sand boils photograph illustrating the settlement of the footpath in a multi-
were observed. Specifically, a probabilistic model of liquefaction storied building in Urayasu. It was observed that the relative
initiation, developed by Cetin et al. [7], was adopted. The results settlement between the footpath and the building was variable
are shown in Fig. 9. The calculated probability of liquefaction over due to the different types of foundations. In some buildings minor
depth up to 20 m is presented in the third sub-figure in Fig. 9. tilting associated with differential settlement was observed. Minor
Three analysis cases were conducted by considering the different tilting also limited the damage to the superstructure. Fig. 11 shows
moment magnitude values, because the developed model has not some light weight structures where significant tilt can be noted.
been validated/calibrated against a mega thrust Mw 9.0 subduc- Many of these light weight structures were supported on small
tion event. The analysis results indicate that soil liquefaction is footings connected through grade beams. The relative settlement
highly likely to be triggered at shallow depths (up to 6 m) and varied between 5 and 15 cm.
medium depths (12–16 m), and the effects of the moment
magnitude are not so significant at these depths. The main reason 3.4. Liquefaction induced damage to flyovers and elevated highways
for the high liquefaction potential for this profile is the low N
count (less than 5). It is noted that the liquefaction triggering is Many of the elevated highways, flyovers and bridges were
not necessarily a reliable indicator of the extent of soil liquefac- surveyed. Damage to superstructures or piers of these structures
tion and settlement; more detailed analysis is needed (which is was not observed, which can be considered as a success in
beyond the scope of this study). comparison with the performance of Hanshin expressway during
the 1995 Kobe earthquake. However, it should be noted that
3.3. Liquefaction induced damage to light structures, walkways, ground motion amplitudes in the Tokyo area were much lower
traffic signal posts and sign posts than those observed in Kobe, but the duration of the strong
motion in 2011 Tohoku earthquake was significantly longer than
In some locations, liquefaction caused damage to light struc- the Kobe motion. The downtime was minimal and these struc-
tures, such as traffic signal posts, lamp posts and electrical poles, tures became operational within a few hours after inspection.
foot paths, walkways and plinth protection structures of buildings. This allowed the survey team to travel by train to the site almost
1624 S. Bhattacharya et al. / Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628
Fig. 10. (a) Gap forming between a building and a sidewalk of a building; (b) local settlement depression of a hotel; (c) settlement of surrounding ground next to a
bridge pier.
Fig. 11. Differential settlement of light structures: (a) Tilting of vending machines due to subsurface liquefaction; (b) tilting of signal posts; (c) severe differential
settlement of a transformer box near Maihama Station.
immediately. Fig. 12 shows the liquefaction induced damage surrounding ground and the upward buoyancy force generated
around the foundations of the piers for the elevated flyovers in by liquefaction.
Maihama. The damage was due to uneven settlement of the
surrounding ground relative to the pier; the relative settlement
varied between 10 and 30 cm. The foundations of these elevated 4. Discussion
highways were piles.
The failures (structural and/or geotechnical) in past earth-
quakes have shown the shortcomings of design methodologies
3.5. Uplift of manholes and construction practices. Post earthquake reconnaissance inves-
tigations have led to the identification of limitations of engineer-
As was observed in previous earthquakes (such as 2004 ing analysis, design and construction practices, as outlined in
Niigata Ken-Chuetsu earthquake), many manholes popped out Table 1. This section of the paper summarises the key observa-
of the ground, causing damage to the water and sewerage tions from the present field investigations and discusses the
pipelines. Fig. 13 shows a photograph of an uplifted manhole. A lessons learnt. The observations are also compared and contrasted
plausible reason is a combination of the settlement of the with the previous earthquakes.
S. Bhattacharya et al. / Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628 1625
Table 1
Historical development of earthquake engineering practice.
1908 Reggio Messina 120,000 fatalities. A committee of nine practising engineers Base shear equation evolved i.e. the lateral force exerted on
earthquake (Italy) and five professors were appointed by Italian government to the structure is some percentage of the dead weight of the
study the failures and to set design guidelines. structure, (typically 5–15%).
1923 Kanto earthquake Destruction of bridges, buildings. Foundations settled, tilted Seismic coefficient method (equivalent static force method
(Japan) and moved. using a seismic coefficient of 0.1–0.3) was first
incorporated in design of highway bridges in Japan (MI
1927).
1933 Long Beach Destruction of buildings specially school buildings. UBC (1927) revised. This is the first earthquake for which
earthquake (USA) acceleration records were obtained from the recently
developed strong motion accelerograph.
1964 Niigata earthquake Soil can also be a major contributor of damage. Soil liquefaction studies started.
(Japan)
1971 San Fernando Bridges collapsed, dams failed causing flood. More soil effects Liquefaction studies intensified. Bridge retrofit studies
earthquake (USA) were observed. started.
1994 Northridge Steel connections failed in bridges. Importance of ductility in construction realised. Significant
earthquake (USA) damage potential due to near-fault motions was
recognised.
1995 Kobe earthquake Massive foundation failure. Soil effects were the main cause Downward movement of a slope (lateral spreading) is said
(Japan) of failure. to be one of the main causes of foundation failure. JRA
(1996) code modified (based on lateral spreading
mechanism) for design of bridge piles.
1999 Chi-Chi earthquake Many bridges collapsed as they were located close to the Special care must be undertaken while designing
(Taiwan) faults. important structure in the vicinity of the plate boundaries
and faults
1999 Koceli earthquake Damage to Bolu tunnel due to fault movement. Damage to Research on fault induced failures intensified.
(Turkey) buildings and bridges. However, buildings conforming the
design codes performed well.
2001 Bhuj earthquake Large-scale destruction. Good performance of some new Large diameter piles concrete performed better than small
(India) jetties of the Kandla port. Tilting of the Kandla Tower diameter piles. Steel piles filled in concrete also performed
building without any damage. better.
2004 Sumatra earthquake Destruction to built environment due to earthquake and New research focused on tsunami warning systems.
and tsunami giant tsunami waves.
4.1. Ground response and large number of load reversals contribute significantly to the
liquefaction triggering [22].
Fig. 14 compares ground motion time-series data at two
recording sites: CHB008 and CHBH14 (see Figs. 2 and 3 for 4.2. Effects of ground improvement
locations). CHBH14 station is a part of the KiK-NET, and both
the surface and borehole data (recording at the bedrock) are It was clear from the survey that the severity of liquefaction
available. Fig. 14(c) shows the bedrock motion recorded at the varied enormously at different locations. There was a construc-
CHBH14 site, which can be considered to be the same as the site tion site in Hanamigawa-Ku (Chiba) with a stark contrast. The
CHB008 for all practical analysis purposes. Fig. 14(a) and (b), on part of the site where the ground was improved did not liquefy at
the other hand, shows the effect of wave propagation on soil all, while the remaining part liquefied. Sand compaction piles
layering at two locations. The comparison of surface-to-borehole (SCP) were used at the site for ground improvement. Another
ground motion data at these locations demonstrates the soil contrasting example was the excellent performance of Haneda
amplification in the top soft soil layers. At these locations, the airport, which was built on reclaimed land using various ground
ground motion was amplified by more than 4 times. Fig. 14(a) and improvement techniques. There was practically no downtime in
(b) also demonstrates the long-duration cyclic loading with rela- Haneda airport, and flights were operational from the very next
tively large amplitudes at CHB008 and CHBH14. Both amplitudes day of the earthquake.
1626 S. Bhattacharya et al. / Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628
Fig. 14. Amplification of ground motion as evidenced from the recordings at two stations; (a) ground motion as recorded at the surface at CHB008 location (K-NET);
(b) ground motion as recorded at the surface at CHBH14 location (KiK-NET); (c) bedrock motion recorded at the CHBH14 site.
Various methods are available for ground improvement in Japan, increases the relative density of the soil and thus strengthens the
such as the sand drain method and the sand compaction pile method. susceptibility against liquefaction. Following the ground improve-
ment, proper foundations (pile or mat) are put in place. For the
ongoing construction site in Hanamigawa-Ku referred to earlier, the
4.2.1. Sand drain method
sand compaction piles were 700 mm in diameter and extended to a
In this method, vertical drains either of coarse soil or permeable
depth of 16–20 m. Sand compaction piles were used to treat the
synthetic variety are installed through liquefiable soils to alleviate
Haneda airport. The latter experienced the earthquake without
the liquefaction problems. The principle is to speed the dissipation
suffering significant liquefaction damage. A good discussion of the
of the excess pore water pressure before reaching full liquefaction.
methodology of SCP can be found in Akiyoshi et al. [1].
This method was used in the Haneda airport underneath some of
the runways. A good discussion on the vertical drains and the
4.3. Good performance of the Haneda airport using the various
associated methodology can be found in Yasuda et al. [21], Orense
methods of ground improvement
et al. [17], Adalier et al. [2] and Brennan and Madabhushi [6].
good fill materials are very important. The fill materials used at The load carrying capacity of the pile ranges between 2356 and
sites that performed very well during the earthquake consisted of 15,009 kN. Each column was supported on one of these large
sands, stones and light weight cement treated dredged soils with diameter piles. The superstructure is a steel frame encased within
and without air. It has been estimated that 6 million m3 of cement reinforced concrete and the interior is light wood partition wall.
treated dredged soils were used during the construction of The long side of the building runs along the river Naka (Naka-
Haneda airport. They are considered to be very light and strong gawa). Lateral spreading was not observed along the river banks
recycled materials. Cement treated dredged soil without air is due to good performance of river bank protection work.
heavier than water but lighter than ordinary soils. On the other Table 2 lists the diameter of piles of ten pile-supported
hand, cement treated dredged soil with air, known as ‘‘Super Geo structures in liquefiable soils and their observed performance
Material (SGM)’’, is very light and almost the same weight as following a strong earthquake. The case records were taken from
water. These soils were used as land reclamation materials behind the performance of structures following the 1964 Niigata earth-
the sea wall, underneath the runway and taxiway. The earthquake quake, the 1995 Kobe earthquake and the 2001 Bhuj earthquake.
showed that these soils are very effective in reducing consolida- It may be noted that the 1960’s design concept of few small
tion/settlement and avoiding ground failure of sea walls. diameter piles (typically 0.3–0.6 m) were gradually replaced by a
few large diameter piles in the 1990s. The poor performance in
this context signifies formation of plastic hinges in the pile and
4.5. A case study: good performance of piled-supported 15-storey cracking. Further details of these case studies and failure modes
building can be found in Dash et al. [8,9], Bhattacharya and Madabhushi
[3] and Bhattacharya et al. [4,5]. In most of these cases it has been
Based on the preliminary investigation and experience from estimated that soil liquefied up to 7–10 m of depth. It is quite
the past earthquakes, it was apparent that the soil remained clear from the table that small diameter piles performed poorly.
liquefied for some time. A survey was carried out in a modern Bhattacharya et al. [4] carried out a comprehensive study of 15
(2005) 15-storey condominium (mansion) in Higashisuna in the case histories of piled foundation performance based on buckling
Sunamachi area on the banks of Naka River (Nakagawa). The analysis parameters. Essentially, the study showed that a pile is
condominium has 236 apartments and runs parallel to the Tokyo laterally unsupported in the liquefiable zone due to the removal
Metropolitan Area highway. There was no damage to the super- of lateral confinement by the soil owing to liquefaction. The
structure, see Fig. 15 for photograph taken after the earthquake slenderness ratio, Leff/rmin, of the pile in the region that could
and the plan of the building. It may be noted that the building is L become unsupported is used to classify pile foundation perfor-
shaped but there is no structural connection between the two mance. Leff is the Euler’s effective length of an equivalent pin
sections. ended strut and rmin is the minimum radius of gyration of the pile
The building is supported on concrete piles having 1–2 m section. The study showed that a line representing a slenderness
diameter and is 53–54 m long. The diameter of the pile at the base ratio (Leff/rmin) of 50 could distinguish between poor and good pile
ranges from 1.4 to 3.5 m to have an enhanced end-bearing. performance. This line is of some significance in structural
Fig. 15. Pile-supported condominium in Sunamachi area (photo and plan view).
Table 2
Case histories of pile foundation performance during past earthquakes.
NFCH pile-supported building during 1964 Niigata earthquake, see [11] 0.35 m dia RCC hollow Poor
Pile-supported Yachiyo Bridge during 1964 Niigata earthquake, see [11] 0.3 m dia RCC Poor
Pile-supported N.H.K building during 1964 Niigata earthquake [11] 0.35 m dia RCC Poor
Pile-supported Showa Bridge during the 1964 Niigata earthquake [5] 0.6 m dia Steel Poor
Pile-supported LPG tank 101 during 1995 Kobe earthquake [13] 1.1 m dia RCC Good
Pile-supported LPG tank 106 and 107 during 1995 Kobe earthquake [13] 0.3 m dia RCC hollow Poor
Pile-supported 14 storey building during 1995 Kobe earthquake [20] 2.5 m dia RCC Good
Pile-supported Kandla Port Tower during 2001 Bhuj earthquake [8] 0.4 m dia RCC Poor
Pile-supported old jetty in Kandla port during the 2001 Bhuj earthquake [3] 0.5 m dia RCC Poor
Pile-supported new jetty in Kandla port during the 2001 Bhuj earthquake [3] 1.0 m RCC filled steel tube Good
1628 S. Bhattacharya et al. / Soil Dynamics and Earthquake Engineering 31 (2011) 1618–1628
Piles
UKIERI International Workshop on Seismic Requalification of Geotechnical Structures (SRGS)
17th December 2012, Delhi
ABSTRACT:
It is utmost important that lifeline infrastructures (such as bridges, hospitals, power plants,
dams etc.) are safe and functional during earthquakes as damage or collapse of these
structures may have far reaching implications. A lifeline’s failure may hamper relief and
rescue operations required just after an earthquake and secondly its indirect economical
losses may be very severe. Therefore, safety of these structures during earthquakes is vital.
Soil-Structure Interaction (SSI) is one of the design issues, which is often overlooked and
even in some cases ignored. The pile foundations are used mostly when the soil conditions
are poor or loads are heavy. For the soft soil conditions, the effect of SSI is significant.
Therefore, it is necessary to consider SSI for design of pile foundations particularly during
seismic excitations. Failures of many structures occurred during the 1989 Loma Prieta and
1994 Northridge, California earthquakes and the 1995 Kobe, Japan earthquake due to SSI or
a related issue. Many jetties had failed in Andaman and Nicobar islands due to 2004 Sumatra
earthquake and ensuing tsunamis. It is because of this recent experience that the importance
of SSI on dynamic response of structures during earthquakes has been fully realized. General
belief that the SSI effects are always beneficial for the design of foundations is not correct.
This chapter addresses the effects of dynamic SSI for the analysis and design of pile
foundations for earthquake loads. Thus, seismic analysis of soil-pile-structure interaction is
discussed. Further recent advances in this area have also been discussed. The article has been
divided in following major headings:
1. Importance of Soil-Pile-Structure Interaction (SPSI)
2. Objective and Scope of Study
3. Steps of SSI and Effects of SSI
4. Different Approaches for SPSI Analyses
5. Recent Advances in SPSI
6. Summary and Conclusions
There are numerous examples (Seed et al. 1990 and 1992) which clearly demonstrated that
structures including buildings, bridges, dams, offshore structures, pile foundations damaged
in past earthquakes due to ignorance of soil-structure interaction in design. In the Mexico city
earthquake, the time period of 10-12 storey buildings founded on soft clay were increased
from 1.0 to 1.5 seconds bringing it closer to resonance conditions. Stewart et al. (1999 a, b)
investigated the strong motion data from recent earthquakes to evaluate the effects of SSI on
structural response for a range of site and structural conditions. Ghosh and Lubkowski (2007)
presented a review on modeling dynamic SSI under seismic loads.
1
SRGS, 17th December, Delhi
The role of SSI on the failure of the Hanshin Expressway in 1995 Kobe Earthquake was very
destructive (Fig. 1), 630 m length of the Hanshin expressway collapsed and overturned.
According to Gazetas and Mylonakis (1998) the effect of SSI in the response of the bridge
was to increase the natural period of vibration and thus increase the ductility demand on the
pier by twice due to the increased base shear forces. Figure 2 shows another examples where
cap of pier of a bridge is damaged in 2001 Bhuj earthquake due to SSI effects.
From above examples as well as many other case histories on damages in past earthquakes, it
was realized that there is a need to rationally incorporate soil-structure interaction in the
design of structures. Gazetas and Mylonakis (1998) illustrated that ignoring SSI has lead to
2
SSI for Seismic Analysis of Pile Foundations
Soil stratum is a semi-infinite medium and being unbounded domain, its modeling requires
considerable amount of efforts. However, appropriate modeling of unbounded domain is
necessary for any SSI analysis. Further the dynamic behavior of soil is much dependent on
frequency. Therefore the methodology used for the analysis shall be able to capture this
characteristic. Beside frequency dependence, the behavior of soil under strong ground
excitation is nonlinear which needs to be properly modeled. If soil medium is saturated loose
sand then it may be subjected to liquefaction. SSI analysis with liquefaction is very complex.
First, fundamental concept of SSI is discussed followed by different steps involved i.e. free
field response, kinematic interaction and inertial interaction. This is followed by explanation
of direct and substructure methods of analysis. For SSI analyses there are a variety of
approaches which may be broadly categorized in 2 groups i.e. simplified models and rigorous
models. Modeling of unbounded soil domain i.e. modeling of boundary is discussed in detail.
Next effect of soil nonlinearity on seismic response is discussed. Finally a practical example
on soil-pile interaction explaining the various effects of SSI is discussed.
External
Seismic Dynamic
Loading Loading
Structure
Soil-Stratum
Fictitious
Boundary
The basic principle of dynamic soil-structure interaction is explained as follows (Wolf 1985):
Soil is a semi-infinite medium, and a major problem in dynamic soil-structure interaction is
the modeling of the unbounded soil domain. For dynamic loading, a structure (it may be
foundation or superstructure) always interacts with surrounding soil and it is not adequate to
analyze the structure independently. If seismic loading is applied to the soil region around the
3
SRGS, 17th December, Delhi
structure, then one has to model this region along with the structure. In the case of a static
loading, a fictitious boundary can be included at a sufficient distance from the structure,
where the response is diminished from a practical point of view. However, for dynamic
loading this procedure cannot be used. The fictitious boundary would reflect waves
originating from the vibrating structure back into the discretized soil region, instead of letting
them pass through and propagate through infinity (Fig. 3)
A semi-infinite soil medium should be modeled in such a way as to satisfy the radiation
condition i.e. only waves can radiate outside the considered region and they cannot reflect
back inside this boundary. Thus this boundary represents an energy sink where only outgoing
waves can occur and this boundary should be modeled adequately. In sequel, the fundamental
objective of the analysis of soil-structure interaction is that dynamic response of the structure
as well as of soil is to be calculated, taking into account the radiation of energy of the waves
propagating into the soil region not included in the model (Fig. 3).
For the structure on the rock: The horizontal motion can be applied directly to the base of
the structure. The input acceleration resulting in the applied horizontal inertial loads will be
constant over the height of the structure. For a given control motion, the seismic response of
the structure depends only on the properties of the structure.
For the structure founded on soft soil: The motion of the base of the structure will be
different from the control motion in the control point because of the coupling of the structure-
soil system. The soil affects the dynamic response of the structure in three ways:
(a) It modifies the free-field motion i.e. the motion of the site in the absence of the structure
and of any excavation. In general, the motion is amplified, but not always (depending on
its frequency content), thus resulting in horizontal displacements that increase towards
the free surface of the site.
(b) Excavating and inserting the rigid base (say foundation) into the site will modify the
motion. The rigid base will experience some average displacements and a rocking
component. This kind of interaction between soil and base in the absence of
superstructure is called kinematic interaction.
(c) In the presence of structure, the inertial loads applied to the structure will lead to an
overturning moment and a transverse shear. This will cause deformation in the soil and
thus once again modify the motion at the base. This part of the analysis is referred as
inertial interaction.
4
SSI for Seismic Analysis of Pile Foundations
It is reasonable to assume that the soil-structure interaction increases the more flexible the
soil is and the stiffer the structure is. On the other hand, it will be negligible for a flexible
structure founded on firm soil.
It is obvious from the many opposing effects that it is, in general, impossible to determine a
priori whether the interaction effects will increase the seismic response. If, however, the first
effect discussed above (the change of the seismic-input motion) is neglected when the
interaction analysis is performed, the response will in many circumstances be smaller than the
fixed-base response obtained from an analysis that neglects all interaction effects. For this
approximate interaction analysis, the control motion is directly used as input motion in the
final dynamic system.
In general, SSI leads to smaller accelerations and stresses in the structure and thereby smaller
forces in the structure. However, there are numerous documented case histories where the
perceived beneficial effects of SSI have lead to oversimplification in the design leading to
unsafe design in foundations and superstructure (Mylonakis and Gazetas, 2000).
5
SRGS, 17th December, Delhi
The American seismic code for nuclear structures (ASCE 4-1998) indicates that fixed base
conditions can be assumed to apply when Vs > 1100 m/s. This condition is generally satisfied
in weak rocks. Dowrick (1987) indicated that the fixed base conditions can be assumed for
structures when Vs > 20 fh. Where, h and f are the height and fundamental fixed based
frequency of the structure, respectively.
The continuous models can fully take into account the effect of radiation damping as well as
the effect of added mass of soil but it is difficult to effectively model material nonlinearity
and loss of bond between soil and pile in such models. On the other hand, the reverse is true
in the case of discrete models (such as shown in Fig. 5). Finite element techniques overcome
the limitation of both of these but for pile groups it may require enormous computation and
prove to be very expensive which may not be justified from the accuracy obtained in the end
results. Various steps of second approach are discussed in detail.
6
SSI for Seismic Analysis of Pile Foundations
7
SRGS, 17th December, Delhi
For step 2, in frequency domain the dynamic impedances (K*) at a particular frequency
may be defined as follows
(a) For viscous damping: K * K st M 2 iC (1a)
Where k1 and k2 are real and imaginary dimensionless coefficients of impedance respectively.
k1 represents the spring stiffness including the effect of inertia while k2 represents damping
force. The values of Kst, k1 and k2 can also be found using simple hand calculations as
proposed by few researchers, Gazetas (1991) and Wolf (1988).
Simplified Approaches
Some of the simplified approaches where analysis can be performed using hand calculations
are presented. These are used only for linear analysis. These are approaches are usually used
for interaction between a rigid circular foundation within a homogeneous soil. One of such
simple approach as presented by Gazetas (1991) is discussed here. According to this, Eq. 2
can be written as:
K * K k iC (3)
Where K is static stiffness, k spring stiffness and C dashpot coefficient. The values of these
can be found from tables and charts presented by Gazetas (1991). Pai (2011) presented a
review on SSI analysis using simplified models.
Significant advancement has been made in SPSI studies in last two decades. With the
advances in computer technology now it is possible to model and analyze SPSI problems
more rigorously. Most of these advances are on two major and very important issues. First,
the soil being an unbounded domain, its precise modeling is a complex issue. Second, the
behavior of soil during strong excitation is highly nonlinear and even for saturated loose
sands the liquefaction may occur. Considering nonlinearity of soil and liquefaction, further
complicate problems. Recent studies performed to deal with these issues are discussed. The
author and his Ph.D. students (Emani and Sarkar) worked extensively in the area of SSI/SPSI.
Maheshwari (2003) used Kelvin Elements to model the boundary and HiSS soil model to deal
with nonlinearity. Emani (2008) carried out nonlinear dynamic SSI analysis using CIFECM
for boundary and hybrid methods for computation. Sarkar (2009) considered liquefaction for
SPSI analysis. A number of research publications from these works are listed in references.
Maheshwari et al. (2005) used Kelvin elements for boundary. Emani and Maheshwari (2009)
used Consistent Infinitesimal Finite Element Cell Method (CIFECM) to model the boundary
for SSI problems. Emani (2008) proposed and demonstrated a hybrid framework of analysis.
8
SSI for Seismic Analysis of Pile Foundations
To account for material nonlinearity of soil, HiSS model (Desai 2001), Fig. 6, is employed.
Two types of SSI studies are done: 1) pile Cap-Soil-Pile Interactions (CSPI) under elasto-
dynamic conditions, and 2) Pile-Soil-Pile Interaction (PSPI) under inelastic dynamic
conditions. Maheshwari and Sarkar (2011), Sarkar and Maheshwari (2012) investigated the
three-dimensional soil-pile behaviour under dynamic condition for the soil with liquefaction.
30
25
g = 0.047 Ultimate yield surface
b =0
n = 2.4
20
a1 = 0.0034
√(J2D), kPa
h1 = 0.78
15 n = 0.42 Phase change line
10
a=0.04
5
0.06 0.05
0.08 0.07
0
0 20 40 60 80 100 120
J1 (k Pa)
For an adequate analysis and design of a SPS system involving nonlinear response, both the
superstructure and foundation should be considered at every time step. Also to closely
simulate the expected seismic foundation input, so that the kinematic interaction can be
analyzed properly, the model should be capable of accepting uneven support excitations.
Kinematic response thus found may differ considerably from the free-field motion. This
nonuniformity of the support motion may have a significant effect on the failure mechanism
of the structure as well as that of the piles and should be properly taken into account.
Fig. 7 shows the complete soil-pile-structure system. Foundation includes a 2*2 pile group
with a rigid massless pile cap on the top. Piles are square in cross section, made of concrete
and resting on bedrock. Superstructure shown is a pier bent of a bridge. For simplicity, it was
assumed that seismic excitation is due to vertically propagating shear waves. To solve the
system a successive-coupling incremental scheme is used (Maheshwari et al. 2005).
9
SRGS, 17th December, Delhi
0.75 m 0.75 m
Structur
e 6
m
Pile
Cap
Pile
Soil s
10
0.5 0.5 m
m 2 m
m
15 m
Fig. 7 Soil-pile-structure systems considered in
the analyses
Full three-dimensional finite element model for SPS system were considered. However, for
economy in computation it was desirable that degrees of freedom are reduced. Thus taking
into account the symmetry and antisymmetry, only one fourth of the actual model was
considered. To properly satisfy the radiation conditions, Kelvin elements are used at the
boundary. Results of analysis are presented without going in detail of computation technique.
Though analyses were performed in time-domain and also for a real earthquake, however
here only results of harmonic excitation are presented. For wide applicability, all results are
presented in dimensionless form.
Free-Field Response: Fig. 8 shows the effect of soil amplification on the input bedrock
motion. It shall be noted that amplitude of input bedrock motion at all frequency considered
was unity. However, amplitude of free-field response (at ground level) is significantly
changed suggesting that input bedrock motion cannot be directly used in place of free-field
response. Also it can be noted that effect of soil on the free-field response is frequency
dependent.
Effect of Soil-Pile Interaction (Single Pile): Due to presence of pile in soil, free field
response is further modified and this is shown in Fig. 9. Where the transfer function shown is
the ratio of pile head response to the free-field response at a particular frequency. It can be
observed that due to interaction effect between soil and pile, peak value of free-field response
is almost doubled of the free-field response. Thus effect of interaction is significant.
10
SSI for Seismic Analysis of Pile Foundations
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Dimensionless frequency (a0 = *d/Vs)
2.5
Transfer Function U p/Ug
1.5
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Dimensionless Frequency (a0 = *d/Vs)
Effect of Pile-Soil-Pile Interaction (Group effect): Fig. 10 shows the real and imaginary
part of the dynamic stiffness at different frequencies of excitations for different pile spacing.
It can be observed for different spacings, peak value of group stiffness is different and occurs
at different frequencies. For close spacing (s/d = 2) pile group acts like a block however at
higher spacing interaction effect is significant (Kaynia and Kausel 1982).
11
SRGS, 17th December, Delhi
1.5
kxxG/(N*kxxS(a0=0))
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
a0 = *d/Vs
4
cxxG/(N*kxxS(a0=0))
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
a0 = *d/Vs
12
SSI for Seismic Analysis of Pile Foundations
response is calculated from the kinematic interaction of the pile of the seismic wave through
the soil. This procedure ignores the feedback from the structure (the inertial interaction).
16
Amplification w.r.t. input bedrock motion
14
12
10
6
without interaction (w/o-int)
4
with interaction (w-int)
2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Period (s)
The results are shown in Fig. 11 in terms of amplifications of response with respect to the
input bedrock motion. The free-field response is also shown. It can be observed from Fig. 11
that the kinematic seismic response of the pile head (without inertial interaction) follows the
free-field ground motion. The response of the pile foundation increases slightly due to the
inertial interaction but the natural period of the foundation subsystem increases from about
0.22 sec to 0.4 sec. The increase in the natural period may be attributed to increased soil
nonlinearity due to the inertial interaction. It is also noted that the kinematic interaction
amplifies the input motion at the base of the structure for short period (T < 0.3 sec)
excitations. On the other hand, the kinematic interaction has no effect on the structure input
motion for long period excitations (T > 0.4 sec).
Fig. 11 also shows that the fundamental period of the structure increases and the structure
peak response decreases significantly due to the inertial interaction. For the range of
parameters considered in the study, the soil-pile-structure interaction elongates the period of
the structure and tends to decrease the peak response.
It can be observed from the discussion presented in this chapter that the soil-pile-structure
interaction is a very important issue to be considered in the design of structures. Ignoring this
effect may not only lead to an uneconomical design but also it may be unsafe. It is not always
13
SRGS, 17th December, Delhi
the case that considering SPSI would lead to conservative design, rather it may provide an
economical as well as rational design. Effects of SSI are not always detrimental but may be
favorable too. Some of the key conclusions from advanced studies in SPSI are as follows
The results from inelastic dynamic analysis of pile groups showed significant reduction,
caused by soil plasticity, in both lateral stiffness and damping of the pile groups.
The effect of soil plasticity is not significant on the kinematic interaction effects in the
considered pile group model.
Both the real and imaginary parts of the dynamic stiffness reduce significantly due to
nonlinearity of the soil medium. For the pile groups, the group effect also reduces
significantly due to nonlinearity of the soil medium.
Consideration of pore water pressure further reduces the dynamic stiffness significantly.
The seismic response of the soil-pile systems for work hardening soil nonlinearity is
greater than those with the soil nonlinearity without work hardening.
Due to incorporation of pore water pressure generation in the soil medium, the seismic
response increases significantly. Since both translation and rotation increases
significantly, for design of piles in liquefying soil medium rotation needs to be taken into
account which is not a common practice.
From the investigation for real earthquake excitations, it is observed that the effect is very
much dependent on PGA and frequency characteristics of input motion. The results
obtained by real earthquake time history are consistent with the harmonic loading results.
The translational KIF of the soil-pile system increases significantly for the softer soil with
pore pressure.
References
14
SSI for Seismic Analysis of Pile Foundations
8. Gazetas G., Fan K., Tazoh T., Shimizu K., Kavvadas M., and Makris N. (1992),
“Seismic Pile-Group-Structure Interaction”, in Piles Under Dynamic Loads, Edited by
Shamsher Prakash, Geotechnical Special Publication No. 34, ASCE, New York.
9. Gazetas, G. & Mylonakis, G. (1998). “Seismic soil structure interaction: new evidence
and emerging issues.” Geotechnical Special publication No. 75, 1119-1174.
10. Ghosh B. and Lubkowski Z. (2007), “Modeling Dynamic Soil-Structure Interaction
under Seismic Loads”, Chapter 7 in Design of Foundation in Seismic Areas: Principles
and Applications, Editor, Bhattacharya S., NICEE, IIT Kanpur, pp. 353-386.
11. IS: 1893-2002 – Part 1: Criteria for Earthquake Resistant Design of Structures: General
Provisions and Buildings, Bureau of Indian Standards, New Delhi.
12. Kaynia, A.M. and Kausel, E. (1982). “Dynamic behavior of pile groups”, Proceedings
of 2nd Int. Conf. on Numerical Methods in Offshore Piling, Austin, Texas, 509-532.
13. Liyanapathirana, D.S. and Poulos, H.G. (2005). “Seismic lateral response of piles in
liquefying soil.” Journal of Geotechnical and Geoenvironmental Engineering, ASCE,
131(12), 1466-1478.
14. Luco, J.E. (1980), “Linear soil structure interaction in Seismic Safety Margins Research
program.” U.S. Nuclear Regulatory Commission, Washington D.C.
15. Maheshwari B.K. (1997) “Soil-Structure-Interaction on the Structures with Pile
Foundations-A Three Dimensional Nonlinear Dynamic Analysis of Pile Foundations”,
Ph.D. Dissertation, Dept. of Civil Engineering, Saitama University, Japan Sept. 1997.
16. Maheshwari B.K. (2003), “Three-dimensional finite element nonlinear dynamic
analyses for soil-pile-structure interaction in the time domain”, Research report
submitted to Mid America Earthquake Center (NSF), Dept. of Civil Engineering,
Washington University, St. Louis, Missouri, March 2003.
17. Maheshwari B.K., Truman K.Z., El Naggar M.H., Gould P.L. (2004a). “Three-
dimensional Finite Element Nonlinear Dynamic Analysis of Pile Groups for Lateral
Transient and Seismic Excitations”, Canadian Geotechnical J. , Vol. 41, pp. 118-133.
18. Maheshwari B.K., Truman K.Z., El Naggar M.H., Gould P.L. (2004b). “3-D Nonlinear
Analysis for Seismic Soil-Pile-Structure Interaction”, Soil Dynamics and Earthquake
Engineering, (Elsevier), Vol. 24, No. 4, pp. 345-358.
19. Maheshwari B.K., Truman K.Z., Gould P.L., El Naggar M.H. (2005), “Three-
Dimensional Nonlinear Seismic Analysis of Single Piles using FEM: Effects of
Plasticity of Soil”, Int. Journal of Geomechanics, ASCE, Vol. 5, No. 1, March 2005.
20. Maheshwari B.K., Watanabe H. (2006) “Nonlinear Dynamic Behavior of Pile
Foundations: Effects of Separation at the Soil-Pile Interface”, Soils and Foundations,
Japanese Geotechnical Society, Paper No. 3234, Vol. 46, No. 4, pp. 437-448.
21. Maheshwari, B.K., Nath, U.K. and Ramasamy, G. (2008). “Influence of liquefaction on
pile-soil interaction in vertical vibration.” ISET Journal of Earthquake Technology,
45(1), 1-13.
22. Maheshwari B.K. and Watanabe H. (2009). “Seismic Analysis of Pile Foundations
using Simplified Approaches”, International Journal of Geotechnical Eng., Vol. 3, Issue
3, pp. 387-404.
15
SRGS, 17th December, Delhi
16
SSI for Seismic Analysis of Pile Foundations
39. Uzuoka, R., Sento, N., Kazama, M., Zhang, F., Yashima, A. and Oka, F. (2007).
“Three-dimensional numerical simulation of earthquake damage to group-piles in a
liquefied ground.” Soil Dynamics and Earthquake Engineering, 27, 395–413.
40. Veletsos A. S., Prasad A. M. and Wu W. H. (1997). "Transfer functions for rigid
rectangular foundations", J. of Earthquake Engg & Structural Dyn; Vol. 26, No. 5, pp.
5-17.
41. Wolf, J.P. (1985), Dynamic Soil-Structure-Interaction, Prentice-Hall, Inc., Englewood
Cliffs, New Jersey.
42. Wolf, J.P. (1988), Soil-Structure-Interaction in the Time Domain, Prentice-Hall, Inc.,
Englewood Cliffs, New Jersey.
43. Wolf J. P. (1994), “Foundation vibration analysis using simple physical models”. PTR
Prentice-Hall, Englewood Cliffs, NJ, USA.
44. Wolf, J. P., and Song, C. (1996). Finite element modeling of unbounded domain, John
Wiley & Sons Ltd, Chichester.
17
UKIERI International Workshop on Seismic Requalification of Geotechnical Structures (SRGS)
17th December 2012, Delhi.
ABSTRACT:
Dynamic behavior of piles is strongly affected by the variation in the soil stiffness with depth
and the interaction between the pile and the soil. The determination of dynamic characteristics
of soil-pile system by full-scale lateral dynamic pile load testing is an important aspect in the
design of pile foundations subjected to dynamic loads. There is a need for experimental
research, in particular large-scale field tests, to validate existing linear/nonlinear models,
which are used to predict the dynamic stiffness and damping of the soil-pile system. This
paper presents the results of field lateral dynamic load tests conducted at four different sites in
India and the measured dynamic constants of the soil-pile system. Dynamic analyses of single
piles using Improved Novak’s Method were performed to study the effect of week zone
around the pile shaft on the lateral stiffness of the piles and the validity of simple models in
predicting the lateral stiffness of piles.
Keywords: Full scale pile tests, Dynamic Lateral Loading, Lateral pile stiffness
1. INTRODUCTION
Stiffness and damping of single piles can be computed using various techniques. Most of the
modern techniques consider propagation of elastic waves through the soil, which modifies the
stiffness of the piles and make it frequency dependent. Piles in a homogenous soil subjected to
small amplitudes vibrations are compatible with the assumptions of linearity (Novak & Sheta,
1982). However, field experiments indicate that the assumption of soil homogeneity is not
suitable for practical applications, as it results in overestimation of pile stiffness and damping.
This is mainly due to the variation in the stiffness of the soil with depth. The analysis and
design of piles subjected to dynamic lateral loads is complex due to the degradation of
stiffness of the soil near the ground surface, compounded by the soil-pile separation.
Additionally, the stiffness in the lateral direction is very low in comparison to the vertical
stiffness and thus the lateral capacity/stiffness of the pile governs the design in most cases.
The lateral capacity, stiffness and the bending behaviour of the pile are mainly dependent
upon the characteristics of the top soil layers within a few meters below ground level, which
mainly consists of weak deposits such as soft clay or loose sand, therefore the degree of
nonlinearity that prevails during the lateral loading on piles is severe. Hence, evaluation of
lateral stiffness of a single pile/pile group under dynamic lateral loading taking into account
the strong nonlinearity and soil gapping becomes a crucial step in the satisfactory design and
performance of pile-supported foundations subjected to dynamic loads.
available to confirm the reliability of these linear and nonlinear models. The limited, but
major full-scale field testing carried out on piles embedded in clay and sandy clay sites by
various authors (Novak 1985, Blaney and O'Neill 1989, El-Marsafawi et al. 1992, Puri and
Prakash 1992, Crouse et al. 1993, Anandarajah et al. 2001, Boominathan et al. 2002, Pak et
al. 2003, Boominathan and Ayothiraman 2006) clearly demonstrate that the performance of
existing linear and nonlinear models are highly dependent on in-situ soil nonlinearity and
dynamic loading characteristics. Hence, the dynamic characteristics of soil-pile system should
be accurately assessed by conducting in-situ full-scale dynamic tests on piles in order to have
a better understanding of nonlinear response of piles. This paper presents the results of full-
scale field dynamic lateral pile load tests carried out at four sites in India
(Chennai,Hazira,Kakinada and Pradip).The field test results are also compared with the
semi-analytical Novak approach.The effect of weak zone on lateral pile stiffness is studied by
carrying out dynamic analysis of piles using a computer code DynaN.
2. SECTION HEADING
Steady state forced lateral vibration tests were conducted on driven cast-in situ piles (Site-I:
Chennai, Tamil Nadu; Site-II: Hazira, Gujrat; Site-III: Paradip, Orrisa) and precast driven
piles (Site-IV: Kakinada, Andhra Pradesh) installed at four different sites in India as per the
procedure recommended by Indian Standard code of practice IS: 9716 (see Fig. 1). The soil
and pile properties at the test sites, test setup for forced lateral vibration tests and the test
procedure are discussed below.
Figure 1. Lateral forced vibration test (after Boominathan and Ayothiraman, 2006).
The soil profile and its properties determined/estimated by various tests/methods, pile
dimensions and its properties for four sites are described herein.
corresponds to a dynamic Young’s modulus (Ep) of 37000 MPa. The soil profile of Site-I and
its properties obtained from the borehole data are presented in Table 1. It can be observed
from Table 1, that the Site-I is characterized by four layers with a 5m thick very soft silty clay
layer at the top. The shear wave velocity of different layers up to a depth of 12.5 m in the site
was determined from MASW test and the shear wave velocity of soil layers below 12.5 m
depth was estimated using Eqn.1. based on the average SPT-N value (Imai, 1977). The
maximum dynamic shear modulus required for determining soil pile stiffness is determined
based on the shear wave velocity using Eqn. 2 (Prakash and Puri, 1987) and the values are
reported in Table 2.1.
Vs = 91 N0.337 (1)
Yellow
18.0-20.2 2.2 >100 430# 2100 388.3
Silty Clay
*
based on MASW test; #based on SPT ‘N’ value
wave velocity. The values of shear wave velocity and shear modulus of various layers are
summarized in Table 2.2.
The forced lateral vibration test was carried out by applying a steady state sinusoidal force
with a mechanical oscillator of 5-tonne capacity. Details on the test procedure can be found at
Ayothiraman et.al., (2012). More information about interpretation of test data can be obtained
from Boominathan and Ayothiraman (2006). The displacement amplitude of vibration, Ax,
was computed from the measured acceleration using Eqn. 3.
Lateral Dynamic Response and Stiffness of Full Scale Single Piles 5
ax
Ax (3)
4 2 f 2
The computed values of amplitude of displacement corresponding to the pile cut-off level at
each frequency for different eccentricities of the oscillator were plotted as frequency response
curves. Typical frequency response curves obtained for Site-I to Site-III for piles having an
L/D ratio of about 30 are presented in Fig. 2 to 4. The frequency response curve for a pile with
L/D ratio of 122 at Site-IV is presented in Fig. 5. Although, all the piles tested are long piles
(L/D > 10), the difference in the L/D ratio between the first three sites and the fourth site will
provide an insight into the significance of the stiffness of the soil at the top 1/3rd of the pile
length. It can be noticed from the Fig. 2 that the resonant frequency of soil-pile system
measured from experimental data for Site-I ranges from 10.5 to 14 Hz, i.e., it reduces from 14
Hz to 10.5 Hz as the magnitude of the dynamic force increases indicating a non-linear
response of the soil-pile system due to the degradation of soil stiffness. This observation is
consistent for almost all test piles at Site-I. The observed nonlinearity could be due to the
presence of a 5 m thick very soft silty clay layer at the top with zero SPT-N value. It is
observed from Fig. 3 that the resonant frequency of soil-pile system measured from field test
data for Site-II ranges from 13 Hz to 14 Hz, i.e., the resonant frequency remains practically
same irrespective of different magnitude of the dynamic force levels. This indicates that the
degree of nonlinearity for Site-II is less, which is due to presence of relatively medium stiff to
stiff clay layers within 8.5 m depth from the ground surface. At Site-III the frequency
decreases from about 16 Hz to 12 Hz, similar to the decrease observed at Site-I. Although, the
stiffness variation along the length of the pile and the L/D ratio of piles in Site-II and III are
similar, their behaviour tends to be different. The frequency response of pile with L/D ratio of
about 120 in Site-IV shows a decrease in frequency from 15 Hz to 11 Hz with increase in the
applied force similar to Site I & III. It can be inferred that lateral dynamic response of single
piles is practically same beyond certain length of piles.
Figure 6. Force vs. Static amplitude plot for Figure 7. Force vs. Static amplitude plot for
Site-I Site-II
Lateral Dynamic Response and Stiffness of Full Scale Single Piles 7
Figure 8. Force vs. Static amplitude plot for Figure 9. Force vs. Static amplitude plot for
Site-III Site-IV
The dynamic stiffness of the soil-pile system is typically obtained from a plot of dynamic
force versus equivalent static amplitude using the procedure described in Boominathan and
Ayothiraman (2006) and as per IS 9716 (1981). The dynamic lateral stiffness of the piles is
also obtained using a semi-analytical Novak’s approach (Novak and El-Sharunby, 1983). The
pile is assumed to be pinned, and the stiffness is evaluated for two conditions: assuming the
soil profile to be homogenous and parabolic. The variation of static amplitude with dynamic
force measured from field tests are shown in Fig. 6 to 9. The lateral stiffness of soil-pile
obtained from the slope of the force vs static amplitude plots (Fig. 6 to 9) and Novak's
approach for the sites are presented in Table 2.5.
The site conditions, and in particular, the properties of the top soil layers greatly govern the
degree of non-linearity and the dynamic lateral stiffness of the soil-pile system. To understand
the effect of weak zone in the top 1/3rd pile length on the stiffness, dynamic analysis were
8 SRGS, 17th December 2012, Delhi
Both theoretical and experimental studies have shown that the dynamic response of the piles
is very sensitive to the properties of the soil in the vicinity of the pile shaft (Han and Novak,
1988). A rigorous approach to the nonlinearity of soil-pile system is extremely difficult and
therefore approximate theories have to be used. Novak and Sheta (1980) proposed including a
cylindrical annulus of softer soil (an inner weak zone or so called boundary zone) around the
pile in a plane strain analysis. The ideal model of boundary zone should have properties
smoothly approaching those of the outer zone to alleviates wave reflections from the interface.
The model of non-reflective interface assumed that the boundary zone has a non-zero mass
and a smooth variation into the outer zone by introducing a parabolic variation function,
which may be best fit with use of experimental data. In the present analysis, DynaN is used to
determine the stiffness of the soil-pile system considering the mass in the boundary zone.
(a) Site – I
(b) Site – II
Top layers of the soil were assumed to have developed week zones due to driving of piles.
The weak zone effect is modeled by reducing the shear modulus ratio by 0.5% and increase in
material damping of the layer by 0.5%.Weak zone shear modulus is the ratio of shear modulus
in the disturbed zone around the pile to the shear modulus of soil present outside the
disturbance zone. Weak zone thickness ratio is the distance of the disturbed zone from the
outer diameter of the pile to the radius of the pile. A zone of 1.25 times the radius was
assumed to be weak zone in the top layers. The soil and pile properties were adopted as per
the field results. The pile is assumed to have a fixed connection with the pile-cap. The
dynamic amplitude vs frequency plots for Site I and II are presented in Fig. 10.
It can be observed from Fig. 10 that the presence of weak zone shifts the predominant
frequency to higher frequency range by about 5 Hz, but cannot be precisely attributed to the
weak zone because of approximations involved in Novak’s approach. However, the lateral
extent of the weak zone around the pile shaft has negligible effect on the frequency. The force
vs static amplitude plots are presented in Fig. 11. It can be observed from Fig. 11(a & b) that
the stiffness of the soil-pile system for both the sites decreases due to presence of weak zone.
The stiffness of the soil-pile system for Site - I and II are 7MN/m and 20 MN/m. The decrease
in the stiffness, due to the presence of the weak zone in the top soil layers, is significant for
the very stiff soil site (Site – II).
Based on the lateral dynamic pile loads tests carried out on full scale single piles at four
different sites in India, it is found that the site conditions, and in particular, the properties of
the top soil layers greatly governs degree of nonlinearity and the dynamic lateral stiffness of
the soil-pile system. It is found that the piles in medium stiff to stiff clay have about 6 times
higher lateral stiffness compared to the piles in very soft clays. The presence of weak zone
around the pile shaft significantly decreases the stiffness of the soil-pile system, especially in
the case of stiff-soil sites. The variation in the lateral extent of the weak zone around the pile
shaft has negligible effect on the stiffness degradation.
REFERENCES
Anandarajah, A., Zhang, J., Gnanaranjan, G. and Ealy, C. (2001). Back-calculation of Winkler
foundation parameters for dynamic analysis of piles from field-test data. Proc. of NSF
International Workshop on Earthquake Simulation in Geotechnical Engineering, Case
Western Reserve University, Ohio, 1–10.
Assareh, M.A. and Asgarian, B. (2008). Nonlinear behavior of single piles in jacket type
offshore platforms using incremental dynamic analysis. American Journal of Applied
Sciences. 5:12, 1793–1803.
Ayothirman, R., Boominathan, A., and Krishna Kumar, S. (2012). Lateral Dynamic Response
of Full Scale Single Piles ─ Case Studies in India. The Role of Full-Scale Testing in
Foundation Design, Edited by M.H. Hussein, R.D. Holtz, K.R. Massarsch, and G.E.
Likins, ASCE, GSP 227,710-721.
Badoni, D. and Makris, N. (1996). Nonlinear response of single piles under lateral inertial and
seismic loads. Soil Dynamics and Earthquake Engineering. 15:1, 29-43.
Blaney, G. and O'Neill, M.W. (1989). Dynamic lateral response of a pile group in clay.
Geotechnical Testing Journal, ASTM, 12:1, 22–29.
10 SRGS, 17th December 2012, Delhi
Abstract
Lands around the coasts are often reclaimed, contain fill materials/loose fills and therefore
potentially liquefiable under moderate to strong shaking. Collapse and/or severe damage to
pile-supported structures are still observed in liquefiable soils after most major earthquakes
e.g. the 2011 Japan earthquake, 2010 Chile earthquake, the 2008 Wenchuan earthquake, the
2004 Sumatra earthquake, the 2001 Bhuj earthquake, the 1995 Kobe earthquake. Poor
performance of pile foundations remains a great concern to the earthquake engineering
community. A comprehensive overview of pile damages observed in the recent earthquakes
is presented in this paper. This paper also critically reviews all the established pile failure
theories. Deficiencies of the existing pile failure theories are highlighted. Various analyses
procedures for piles in liquefiable soils are discussed.
Keywords
Pile failure Liquefaction Earthquake
Y. Huang et al. (eds.), New Frontiers in Engineering Geology and the Environment, 31
DOI: 10.1007/978-3-642-31671-5_3, Ó Springer-Verlag Berlin Heidelberg 2013
32 S. Bhattacharya et al.
was repaired with a steel beam covering a length of 45 m in failures occurred immediately or during the earthquake. (c)
the north side of the bridge and maintaining the same Effects of liquefaction mainly in the river bridges; (d)
foundations made of wood piles. Lateral spreading could Inadequate structural consideration: Poor beam column
induce strength degradation and bending failure in the stiffness, irregular shape causing stiffness irregularity
bridge. However, this explanation doesn’t necessarily (bridges curved in plan); (e) Active Fault movement and
explain the failure of spans in the middle extension of the surface fracture induced by active fault.
bridge. Figure 3b shows Llacolen bridge damage during the
same Chile earthquake.
2.4 Bhuj (India) Earthquake
2.3 Wenchuan Earthquake (China) in 2008 The Bhuj earthquake (magnitude, Mw = 7.7) that struck the
Kutch area in Gujarat at 8.46 AM (IST) on January 26, 2001
Wenchuan earthquake was of the major earthquake that was the most damaging earthquake in India in the last
struck China. In this earthquake several hundred bridges 50 years. The epicenter of the quake was located at 23.4°N,
collapsed. Figure 4 shows the collapse of one of the spans 70.28°E and at a depth of 25 km, which is to the north of
of the MPZ (Miaoziping bridge in Chengdu). The reasons Bacchau town (Fig. 5). This earthquake has caused exten-
for collapse of so many bridges can be summarized as sive damage to the life and property. Figures 6 shows
follows: (a) Site effects: many of the bridges that collapsed failure of a pile-supported Kandla Port tower. This tower
were near the epicenter of the earthquake due to strong near was located in laterally spreading ground. This tower tilted
field motion. The recorded horizontal PGA is near about 1 g possibly due to the interaction between axial load induced
and the recorded vertical near field PGA in Wolong Station settlement owing to liquefaction and bending due to lateral
is 948.1 gal i.e. 0.95 g; (b) Geological hazards due to spreading forces. Details of the failure can be found in Dash
earthquake i.e. mainly due to landslides, rockfall. These et al. (2009a, b).
Fig. 4 a Photograph of the MPZ bridge following the 2008 Wenchuan earthquake b Schematic diagram of the bridge
0.3 m
(a) (b) 10 m
2m
Before earthquake
After earthquake
20 m
0.3 m
Boundary wall
10 m soft clay
18 m Arabian Sea
12 m sandy soil
0.45m
Brown clay
Fig. 6 a Kandla Port tower after the 2001 Bhuj earthquake b Plausible settlement mechanism of failure showing the tilting the Tower assuming
there is no structural failure of piles
1. The failure of foundations not only occurred in laterally between liquefied and non-liquefied layers. Often cracks
spreading grounds but also was observed in level were observed at the pile head. More details can be
grounds where no lateral spreading would be anticipated. found in Bhattacharya & Madabhushi (2008). The
This has been demonstrated through the building dam- authors showed that plastic hinges also formed not only
ages in Fig. 7. The failure patterns of the buildings in at the boundaries of the liquefiable and non-liquefiable
level ground and in laterally spreading ground are sim- layers but also at various depths. There are few cases
ilar i.e. the buildings tilt considerably, see for example where plastic hinges formed at the middle of the lique-
Fig. 7. In most cases, the superstructure remains fied layer, see for example Fig. 8b. It has been reported,
undamaged. Therefore, by considering the building Yoshida et al. (2005) that this damage was observed
failure alone, it is difficult to ascertain the type of ground when the structure was very close to the quay wall which
i.e. whether lateral spreading ground or level ground. moved towards the sea significantly.
2. Figure 8 shows several examples of pile damage along 3. Horizontal displacement of the piles were measured in
which the soil profiles and the damage pattern. Cracks some cases of pile failure, see for example Yoshida and
were observed near the bottom and top boundaries Hamada (1990), Hamada & O’Rourke (1992). In some
36 S. Bhattacharya et al.
Depth(m)
Sand
4 4 10
5 5
6 6 15 No.1
No.2
7 7
8 8 20
SPT N-value
(b) 0102030
GL- m
4,500 ∼ 6,500
7 9
≈11,000
φ≈10∼15°
8 10
9 12
10 8
2,000 ∼3,000
11 19
12 23
400
13 19
14 25
1,000 ∼2,000
Unit: mm
(c)
cases, the displacement of the piles was found to agree on 0.5 m diameter RCC piles suffered cracking during
with the horizontal displacement of the ground. the 2001 Bhuj earthquake. In contrast, in the same
4. It has been observed that large diameter piles per- area, the 1.0 m diameter RCC piles on the newer jet-
formed well during the earthquake. For example, the ties and the concrete-filled-steel tubular piles performed
older jetties of the Kandla port which were supported well.
Seismic Design of Piles in Liquefiable Soils 37
Load
Load
Lateral soil
spring Pu Pu
yu Displacement yu Displacement
Load
Load
Pu Pu
Pile
yu Displacement yu Displacement
Pu Pu
Load
Load
K
yu Displacement yu Displacement
End bearing
spring
(a) (b)
Initial Effective Overburden Pressure where r0 v0 is the initial value of effective overburden
pressure at that level; and k2, m and n are experimental
Initiate Accumulated Irrecoverable constants for a given sand. The current shear modulus and
Volumetric Strain (εv) bulk modulus is modified progressively for changing
effective stresses in each time interval as follows (Martin &
Calculate Octahedral Shear Strain Seed 1979; Arduino et al. 2002)
0:5 0:5
r0v r0v
Check whether No G ¼ Gmax ; K ¼ Kmax ð8Þ
RETURN r0vo r0vo
Strain Reversal
Yes where Gmax and Kmax are initial shear and bulk modulus
Calculate Nondimensional Constants respectively. For the cases where due to rise in pore water
C1 and C2 for Byrne Model pressure, effective stress becomes zero (complete liquefac-
tion), to maintain stability in the calculations, shear mod-
Calculate Increment of Volume ulus equal to 3 % of initial shear modulus (Gmax) is
Decrease (Δεv) and Elastic Rebound generally retained. The algorithm for pore water pressure
Modulus of Sand Skeleton (Er)
generation is presented below in Figs. 16 and 17. This
algorithm is capable to consider pore pressure generation
Calculate Pore Pressure Increment with advanced plastic soil models.
(Δu)
Acknowledgments The authors acknowledge the funding received
Update Effective Overburden from National Science Foundation (China) and Royal Society (U.K).
Pressure, Stress and Irrecoverable This work was supported by the National Natural Science Foundation
Volumetric Strain (εv) of China through the Grant No 41111130205.
Bhattacharya, S., Adhikari, S., & Alexander, N. A. (2009). A of Structural and Construction Engineering, AIJ (Architectural
simplified method for unified buckling and free vibration analysis Institute of Japan) 617, 169–175.
of pile-supported structures in seismically liquefiable soils. Soil Knappett, J. A., Madabhushi, S. P. G. (2005). Modelling of Lique-
Dynamics and Earthquake Engineering, 29, 1220–1235. faction-induced Instability in pile groups. In: Boulanger and
Bhattacharya, S., Hyodo, M., Goda, K., Tazoh, T., & Taylor, C. A. Tokimatsu (Eds.), ASCE Geotechnical special publication No
(2011). Liquefaction of soil in the Tokyo Bay Area from the 2011 145 on seismic performance and simulation of pile foundations in
Tohoku (Japan) Earthquake. Soil Dynamics and Earthquake liquefied and laterally spreading ground (pp. 255–267).
Engineering, 31, 1618–1628. Lin, S. S., Tseng, Y. J., Chiang C. C., & Huang, C. L. (2005). Damage
BTL Committee. (2000). Study on liquefaction and lateral spreading in of piles caused by lateral spreading—Back study of three cases. In:
the 1995 Hyogoken-Nambu earthquake. Building Research Report R.W. Boulanger and K. Tokimatsu (Eds.), ASCE Geotechnical
No 138, Building Research Institute, Ministry of Construction, special publication no 145 on seismic performance and simulation
Japan (in Japanese). of pile foundations in liquefied and laterally spreading ground
Byrne, P. M. (1991). A cyclic shear-volume coupling and pore (pp. 121–133).
pressure model for sand. Proceedings of the 2nd International Liu, L., & Dobry, R. (1995). Effect of liquefaction on lateral response
Conference on Recent Advances in Geotechnical Earthquake of piles by centrifuge model tests. NCEER report to FHWA.
Engineering and Soil Dynamics. St Louis, Report 1.24 (pp. 47–56). NCEER Bulletin, 9(1), 7–11.
Dash, S. R. (2010). Lateral pile-soil interaction in liquefiable soils. Maheshwari, B. K., & Sarkar, R. (2011). Seismic behaviour of soil-
PhD thesis, University of Oxford, UK. pile-structure interaction in liquefiable soils: A parametric study.
Dash, S. R., Bhattacharya, S., Blakeborough, A., & Hyodo, M. (2008). International Journal of Geomechanics, ASCE, 11(4), 335–347.
P-Y curve to model lateral response of pile foundations in liquefied Martin, P.P., & Seed, H. B. (1979). Simplified procedure for effective
soils. Paper number 04-01-0089, Proceedings of the 14th World stress analysis of ground response. Journal of Geotechnical
Conference on Earthquake Engineering, October 12–17, 2008, Engineering Division, ASCE 105(GT6), 739–758.
Beijing. Martin, G. R., Finn, W. D. L., & Seed, H. B. (1975). Fundamentals of
Dash, S. R., Bhattacharya, S., & Blakeborough, A. (2009a). A study of liquefaction under cyclic loading. Journal of Geotechnical Engi-
p-y curves for liquefiable soils. Proceedings of the International neering Division, ASCE 101(GT5), 423–438.
Conference on Performance-based Design in Earthquake Geo- Novak, M., Sheta, M., El-Hifnawy, L., El-Marsafawi, H., & Ramadan,
technical Engineering. IS- Tokyo 2009, CRC Press (Balkema) O. (1990). DYNA3: A computer program for calculation of
Taylor & Francis Group Book and CD-ROM ISBN 978-0-415- foundation response to dynamic loads. London: Geotechnical
55614-9. Research Centre University of Western Ontario.
Dash, S. R., Govindaraju, L., & Bhattacharya, S. (2009b). A case study of Rollins, K. M., Gerber, T. M., Lane, J. D., & Ashford, S. A. (2005).
damages of the Kandla Port and Customs Office tower supported on a Lateral resistance of a full-scale pile group in liquefied sand.
mat-pile foundation in liquefied soils under the 2001 Bhuj earth- Journal of Geotechnical and Geoenvironmental Engineering,
quake. Soil Dynamics and Earthquake Engineering, 29(2), 333–346. ASCE, 131, 115–125.
Dash, S. R., Bhattacharya, S., & Blakeborough, A. (2010). Bending- RTRI. (1999). Design standard for railway facilities-seismic design.
buckling interaction as a failure mechanism of piles in liquefiable Tokyo: Railway Technical Research Institute.
soils. Soil Dynamics and Earthquake Engineering, 30, 32–39. Sarkar, R., & Maheshwari, B. K. (2012). Effects of separation on the
Fan, K., Gazetas, G., Kaynia, A. M., Kausel, E., & Shahid, A. (1991). behaviour of soil-pile interaction in liquefiable soils. International
Kinematic seismic response of single piles and pile groups. Journal Journal of Geomechanics, ASCE, 12(1), 1–13.
of Geotechnical Engineering ASCE, 117(12), 1860–1879. Shanker, K., Basudhar, P. K., & Patra, N. R. (2007). Buckling of piles
Finn, W. D. L. (1982). Dynamic response analysis of saturated sands. under liquefied soil condition. Geotechnical and Geological
In: G. N. Pande, & O. C. Zienkiewicz (Eds.), Soil mechanics— Engineering, 25(3), 303–313.
transient and cyclic loads. NewYork: John Wiley & Sons Ltd. Takahashi, A., Kuwano, Y., & Yano, A. (2002) Lateral resistance of
Gobindraju, L., & Bhattacharya, S. (2012). Site-specific earthquake buried cylinder in liquefied sand. Proceedings of the International
response study for hazard assessment in Kolkata city, India. Conference on physical modelling in geotechnics, ICPMG-02, St.
Natural Hazards, 61(3), 943–965 John’s, Newfoundland, Canada.
Hamada, M., & O’Rourke, T. D. (Eds.). (1992). Case studies of Tokimatsu, K., Oh-oka, H., Satake, K., Shamoto, Y., & Asaka, Y.
liquefaction and lifeline performance during past earthquakes (vol. (1998). Effects of lateral ground movements on failure patterns of
1). Japanese case studies, Technical Report NCEER-92-0001. piles in the 1995 Hyogoken-Nambu earthquake. Proceedings of a
Ishihara, K. (1997). Terzaghi oration: Geotechnical aspects of the 1995 speciality conference, geotechnical earthquake engineering and
Kobe earthquake. Proceedings of ICSMFE, (pp. 2047–2073). soil dynamics III, ASCE Geotechnical Special publication No 75
Hamburg. (pp. 1175–1186).
JRA. (2002, 1996). Japanese Road Association. Specification for Wu, G., & Finn, W. D. L. (1997a). Dynamic elastic analysis of pile
Highway Bridges, Part V, Seismic Design. foundations using finite element method in the frequency domain.
Kawamura, S., Nishizawa, T., & Wada, H. (1985).Damage to piles due Canadian Geotechnical Journal, 34, 34–43.
to liquefaction found by excavation twenty years after earthquake. Wu, G., & Finn, W. D. L. (1997b). Dynamic nonlinear analysis of pile
Nikkei Architecture (pp. 130–134). foundations using finite element method in the time domain.
Kaynia, A. M., & Kausel, E. (1982), Dynamic stiffness and seismic Canadian Geotechnical Journal, 34, 44–52.
response of pile groups. Research Report R82-03, Order No. 718, Yasuda, S., Terauchi, T., Morimoto, M., Erken, A., & Yoshida, N.
Cambridge, Massachusetts. (1998). Post liquefaction behavior of several sand. Proceedings of
Kimura, Y., & Tokimatsu, K. (2007). Buckling stress of slender pile the 11th European Conference on Earthquake Engineering,
with lateral displacement at the pile head in liquefied soils. Journal Balkema, Rotterdam.
44 S. Bhattacharya et al.
Yoshida, N., & Hamada, M. (1990). Damage to foundation piles and Yoshida, N., Towahata, I., Yasuda, S., & Kanatani, M. (2005).
deformation pattern of ground due to liquefaction-induced perma- Discussion to the paper by Bhattacharya et al. (2004) on ‘‘An
nent ground deformation. Proceedings of 3rd Japan-US Workshop alternative mechanism of pile failure in liquefiable deposits during
on Earthquake resistant Design of Lifeline Facilities and Coun- earthquakes’’. Geotechnique 55(3), 259–263.
termeasures for Soil Liquefaction (pp. 147–161).
SECTION B:
Retaining Walls
UKIERI International Workshop on Seismic Requalification of Geotechnical Structures (SRGS)
17th December 2012, Delhi.
ABSTRACT
Knowledge of seismic earth pressure is very important for design of earth retaining structures
in earthquake prone areas. Many researchers have developed several analytical and numerical
methods for the estimation of seismic earth pressure to design retaining walls in earthquake
prone areas. Also some experimental studies are available to undestand the seismic behaviour
of such earth retaining structures. Most popular analytical design methods pseudo-static and
pseudo-dynamic for seismic earth pressure calculation. The critical study of all these methods
shows the advantage of pseudo-dynamic method over the conventional pseudo-static method
with validation through dynamic centrifuge test results. In this paper an overview of the
available procedures for seismic analysis and design of earth retaining structures and
geosynthetic reinforced retaining structures has been explained. Its also highlights the salient
features of the methods which are currently used in practice for seismic design of retaining
structures. Also, Indian and important international design codes for the seismic design of
retaining structures are explained in brief. In summary, this paper mentions some seismic
requalification techniques for designing earth retaining structures in earthquake prone areas.
Keywords: seismic analysis, pseudo-static & dynamic, retaining structures, design code.
1. INTRODUCTION
Knowledge of seismic earth pressure is very important for design of retaining structures in the
earthquake prone areas. Many researchers have developed methods for design of retaining
structures in the seismic areas. However pseudo-static and pseudo-dynamic methods are most
popularly used methods in Geotechnical Engineering practice. Mononobe and Okabe used the
first explicit application of pseudo-static approach for retaining wall in 1926 and 1929. But
the details of the pseudo-static approach for slope stability analysis were described by
Terzaghi (1950) (see Kramer, 1996). This paper explains these methods for the analysis and
design of the earth retaining structures along with reinfirced retaining structures in the
seismic active area. Advantages and disadvantages of these methods are summerized in
conclusions.
In 1920s, the seismic stability of earth structures was analysed by a pseudo-static approach in
which the effects of an earthquake forces were represented by constant horizontal and(or)
1
2 SRGS, 17th December 2012, Delhi
vertical accelerations. The pseudo-static method is the oldest method to design facilities
against earthquake effects. The first explicit application of the pseudo-static approach to the
analysis of seismic of seismic slope stability has been attributed to Mononobe and Okabe in
1929 (see Kramer, 1996). But the details of the pseudo-static approach for slope stability
analysis were described by Terzaghi (1950) (see Kramer, 1996). Their method statically
applies a force to a designed facility (pseudo-static or quasi-static method). The magnitude of
this force is specified to be K*W in which K is called the seismic coefficient and W is the
weight of the facility. This is the first method to design the structure against additional
destabilizing earthquake forces. Since this method works easily with static calculation, it is
still widely used.
Weight W
Acceleration, A
Figure 1 (a). Example of elementary Figure 1 (b). D’Alembert’s principle of
seismic slope stability analysis mechanics
(Towahata, 2008) (Towahata, 2008)
The value of K depends on the region, seismic activity, and importance of facilities, local
geology and soil conditions. Different countries have different values of K. Today, the value
of K in Japan is 0.15-0.2 or greater (Towahata, 2008). Various researchers like Saran and
Prakash (1968), Madhav and Kameswara Rao (1969), Saran and Gupta (2003) and many
others had applied pseudo-static method for computation of seismic earth pressure behind
retaining walls.
Richards and Elms (1979) proposed a method for seismic design of gravity retaining walls
which is based on permanent wall displacements. The effect of wall inertia was of the same
order as that of the soil dynamic pressure computed by the Mononobe-Okabe (See Kramer,
1996) analysis. Authors also mentioned that, in contrast to their insignificant effect on the soil
dynamic pressure, the vertical acceleration and the angle of wall friction between the soil and
the wall were important parameters in calculating the wall inertia effect. The authors also
suggested new design approach based on the initial choice of a specified limiting wall
displacement. This procedure computed the weight of the wall required to prevent the motion
greater than that specified. It achieved this by combining an incremental relative displacement
model with the Mononobe-Okabe analysis. Figure 2 shows the forces considered in the
Seismic Behaviour of Earth Retaining Structures and Design Issues 3
analysis. Richards and Elms (1979) gave expression for permissible displacement as given by
Eqn. (3).
Considering the equilibrium of the wall,
1 (1)
Ww = γ H 2 (1 − kv ) K ae CIE
2
where,
(2)
cos( β + δ ) − sin( β + δ ) tan φb
CIE =
(1 − kv )(tan φb − tan θ )
where,
Ww = Self-weight of the retaining wall,
φb = Friction angle between the wall and the soil on which it is resting,
kh and kv are seismic acceleration coefficients in horizontal and vertical directions,
Kae is seismic active earth pressure coefficient,
is unit weight of soil.
Figure 2. Seismic forces on gravity retaining wall as considered by Richards and Elms (1979)
Richards and Elms (1979) gave the expression for permanent block displacement as in Eqn. 3,
2 3
vmax amax (3)
d pem = 0.087
a 4y
where, vmax is the peak ground velocity, amax is the peak ground acceleration, ay is the yield
acceleration for the wall-backfill system.
Choudhury and Subba Rao (2002) presented a detailed study for the estimation of passive
earth resistance under seismic conditions by adopting the pseudo-static approach for negative
wall friction case. The application of this case is for anchor uplift capacity. Simple limit
equilibrium approach was used and the failure surface was considered to comprise of the arc
of a log spiral. The model considered to estimate seismic passive resistance by Choudhury and
Subba Rao (2002) is shown in Figure 3 (a). It assumes failure plane shape as log spiral for
passive earth resistance with negative wall friction case. Seismic passive resistance is given
by Eqn. 4.
4 SRGS, 17th December 2012, Delhi
Figure 3 (a). Seismic passive resistance for negative Figure 3 (b). Typical seismic passive
wall friction case as considered by Choudhury and earth pressure coefficient as per
Subba Rao (2002) Choudhury and Subba Rao (2002) for
= = 0, /φ = -0.5
1 2 1
Ppd = 2cHK pcd +qHK pqd + H Kp d (4)
2 cos
where Ppd is seismic passive resistance, Kp d, Kpqd, and Kpcd are the seismic passive earth
pressure coefficients corresponding to the individual minimum values of Pp d, Ppqd, and Ppcd,
respectively. Figure 3 (b) explains typical results. It can be observed from Figure 3 (b) that
the seismic passive earth pressure coefficients always decrease with an increase in the vertical
seismic acceleration, but the horizontal seismic acceleration results in either an increase or a
decrease in the earth pressure coefficients, depending on the combinations of , , and /φ. In
the presence of seismic forces, ground slope variation from negative to zero increases Kp d
values, and a further increase in ground slope is found to decrease Kp d values. An increase in
the magnitude of results in lower seismic passive earth pressure coefficients.
Choudhury et al. (2004) presented a comprehensive review for different methods to calculate
seismic earth pressures and their point of applications. With the example of numerical
example the merit of a displacement based analysis over force-based analysis by considering a
permissible displacement of the wall was explained. Also, authors suggested modification of
IS code for seismic design of retaining wall. According to authors, for seismic design of earth
dam currently pseudo-static methods are used, which excludes the deformation criteria. So,
authors suggest that, factored shear strength and pore pressure values must be used in the
design of earth dam. Figure 4 (a) shows model considered by Choudhury et al. (2004).
Choudhury et al. (2004) had given expression for the point of application of seismic passive
resistance by Eqn. 5.
where,
2π Vs 2π V p (6)
λ= and ζ =
ω ω
excitation frequency, Vs and Vp are shear and primary wave velocities respectively, H is
height of retaining wall.
Figure 4 (a). Model considered by Figure 4 (b). Typical results for point of
Choudhury et al. (2004) application for seismic passive earth
pressure by Choudhury et al. (2004)
Figure 4 (b) shows typical results obtained by Choudhury et al. (2004). It can be seen from
Figure 4 (b), that the point of application of total passive resistance need not act always at
0.33H from the base of the wall with vertical height H, but it can vary. Hence, Choudhury et
al. (2004) concluded that the distribution of seismic passive earth pressure along the depth is
nonlinear in most of the cases.
Subba Rao and Choudhury (2005) presented a pseudo-static limit equilibrium analysis for the
computation of the seismic passive earth resistance coefficients by using a composite failure
surface for positive wall friction angle. Unlike previous researches, which were mainly for
sands, a generalized solution was proposed in which consideration of the cohesive backfill
was made. Figure 5 (a) shows the model considered by Subba Rao and Choudhury (2005).
The total seismic passive force Ppd on the retaining wall of height H becomes,
1 1 (7)
Ppd = 2cHK pcd + qHK pqd + γ H 2 K pγ d
2 cos δ
where Ppd is seismic passive resistance, Kp d, Kpqd, and Kpcd are the seismic passive earth
pressure coefficients corresponding to the individual minimum values of Pp d, Ppqd, and Ppcd,
respectively.
Figure 5 (b) shows typical results obtained by Subba Rao and Choudhury (2005). It can be
seen from Figure 5 (b) that the Kp d values are decreasing with an increase in kh value for a
given kv value.
6 SRGS, 17th December 2012, Delhi
Figure 5 (a). Composite Failure Surface and forces Figure 5 (b). Typical results by Subba
on the system considered by Subba Rao and Rao and Choudhury (2005) for = 0,
Choudhury (2005) = 0, /φ = 0.5
Also considered in the analysis were the effect of surcharge and self weight of the backfill
soil. Design charts were proposed for the direct computation of seismic passive earth
resistance coefficients. Using the limit equilibrium method of analysis with a pseudo-static
approach for seismic forces, comprehensive results of seismic passive earth pressure
coefficients with respect to own weight, surcharge, and cohesion components have been
generated considering composite failure surfaces. The application of this case is for bearing
capacity of foundations.
The pseudo-static method was used by Choudhury and Ahmad (2007) to analyze the stability
of waterfront retaining wall. Authors presented a study in which the effect of earthquakes
along with the hydrodynamic pressure including inertial forces on such a retaining wall has
been observed. The hydrodynamic pressure has been calculated using Westergaard’s (1933)
approach, while the earth pressure has been calculated using Mononobe-Okabe’s pseudo-
static analysis. A free body diagram of the wall showing different forces coming onto it from
soil, water, and due to seismicity along with their respective points of applications is shown in
Figure 6 (a). Figure 6 (b) explains the typical result for factor of safety in overturning mode.
The equations for factor of safety given in terms of non dimensional parameters are as shown
in Eqn. 8 and Eqn. 9.
1 2 1
γ w ( h / H ) + µ (1 − kv )(b / H )γ c + K ae γ sin δ
FS slidingr =
2 2 (8)
1 2 7 2 1
(h / H ) + k γ w ( h / H ) + K ae γ cos δ + k h (b / H )γ c
2 12 h 2
1 2 1
w (h / H) + (1 - kv )(b / H) c + K ae sin
2 2 (9)
FS slidingf =
1 2 7 2 1
(h / H) + kh w (h / H) + K ae cos + kh (b / H) c
2 6 2
Seismic Behaviour of Earth Retaining Structures and Design Issues 7
where, Fsslidingr is factor of safety against sliding for restrained water case and FSslidingf is factor
of safety against sliding for free water case. Depending on the hydraulic conductivity (k) of
the soil, two different cases namely restrained water case (k~10-3 cm/sec) and free water case
(k~very high) may arise for the generation of the hydrodynamic pressure in the backfill soil
(Kramer, 1996).
Figure 6 (a). Free body diagram of the wall Figure 6 (b). Factor of safety in overturning
subjected to different forces as considered mode for different (h/H) values as obtained by
by Choudhury and Ahmad (2007) Choudhury and Ahmad (2007
Shukla et al. (2009) have described the derivation of an analytical expression for the total
active force on the retaining wall for c- soil backfill considering both the horizontal and
vertical seismic coefficients using pseudo-static method. Authors have found that their results
are in agreement with previous researcher’s work.
Pseudo-static method is still in use because it’s simplicity and the factor of safety can be
calculated by the same way as the conventional static stress calculation. No advanced or
complicated analysis is necessary. It made a great contribution to the improvement of seismic
safety. Still it suffers some serious disadvantages, which are:
1. The real seismic force is cyclic, changing in direction with time, and its time duration is
limited. In contrast, the seismic coefficient method applies a force in a static manner. This
seismic force overestimates the risk of earthquake failure.
2. Representation of the complex, transient, dynamic effects of earthquake shaking by single
constant unidirectional Pseudo-static acceleration is vey crude.
3. At the time of 1994 Northridge earthquake near Los Angeles, the maximum horizontal
acceleration of 1.8 G or possibly 1.9 G was recorded at Tarzana site. Within ten meters
from the accelerometer here, a small hut did not suffer damage (Towahata, 2008). Was
this structure well designed against a horizontal static force as intense as 1.8 times its
weight? This explains the uncertainties involved in the detemination or predifined values
of “K”.
4. The relation between K and the maximum ground acceleration is not clear. 1.9 G
acceleration does not mean K = 1.9
Besides these shortcomings, pseudo-static method is widely used because of its simplicity.
Pseudo-static method is simple and straight forward. Also it uses limit state equilibrium
analysis which is routinely conducted by Geotechnical Engineers. Also, the computations are
easy to understand and perform. However, that the accuracy of the pseudo-static approach is
8 SRGS, 17th December 2012, Delhi
governed by accuracy with which the simple pseudo-static inertial forces represent the
complex dynamic inertial forces that actually exist in an earthquake.
To improve the pseudo-static method an attempt was made by Steedman and Zeng (1990).
They considered phase difference due to finite shear and primary waves propagation behind a
retaining wall. This method also considers the effects of horizontal and vertical seismic
accelerations as time, frequency and amplification dependent parameters for seismic design of
retaining structures. A vertical rigid retaining wall supporting a particular value of soil friction
angle (φ) and a particular value of seismic horizontal acceleration (khg, where g is the
acceleration due to gravity) has been considered.
Choudhury & Nimbalkar (2005) modified this method by considering the effect of vertical
seismic acceleration along with horizontal seismic acceleration for passive case. Authors also
studied the effects of a wide range of parameters like wall friction angle, soil friction angle,
shear wave velocity, primary wave velocity and horizontal and vertical seismic accelerations
on seismic active earth pressure. They also considered that the shear modulus (G) is constant
with depth of a retaining wall throughout the backfill and only the phase and not the
magnitude of accelerations are varying along the depth of the wall.
The major conclusions of Choudhury & Nimbalkar (2005) was that the pseudo-dynamic
method gives more realistic non-linear seismic active earth pressure distribution behind the
retaining wall as compared to the Mononobe–Okabe method using pseudo-static approach.
Choudhury & Nimbalkar (2006) extended their work for seismic active earth pressure
determination. Figure 7 (a) shows the model considered by Choudhury and Nimbalkar (2005)
for passive case, where as figure 8 (a) shows model by Choudhury and Nimbalkar (2006) for
active case.
Figure 7 (a). Model retaining wall considered Figure 7 (b). Typical variation for passive
by Choudhury and Nimbalkar (2005) for earth pressure distribution by Choudhury and
computation of pseudo-dynamic passive earth Nimbalkar (2005)
pressure φ = 30°, δ = φ/2, H/ = 0.3, H/ = 0.16
Seismic Behaviour of Earth Retaining Structures and Design Issues 9
Figure 8 (a). Model retaining wall considered Figure 8 (b). Typical variation for active
by Choudhury and Nimbalkar (2006) for earth pressure distribution by
computation of pseudo-dynamic active earth Choudhury and Nimbalkar (2006)
pressure kv = 0.5kh, φ = 30°, δ = φ/2, H/ = 0.3, H/ =
0.16
Figure 7 (b) and Figure 8 (b) show typical results by Choudhury and Nimbalkar (2005) for
passive case and by Choudhury and Nimbalkar (2006) for active case respectively. It can be
seen from Figure 7 (b) and Figure 8 (b) that that the seismic active as well as passive earth
pressure distribution varies non-linearly with the dimensionless depth parameter (z/H).
Again, in a similar way, Nimbalkar and Choudhury (2008) developed the theory to estimate
the seismic passive earth pressure by using the pseudo-dynamic approach by considering soil
amplification. In this work, Nimbalkar and Choudhury (2008) considered the amplification of
the soil by the formulae as given by equation (10) and (11).
H−z H−z
a h ( z, t ) = 1 + ( f a − 1) k h g sin ω t − (10)
H Vs
H−z H−z
av ( z, t ) = 1 + ( f a − 1) k v g sin ω t − (11)
H V ps
where Vs and Vp are shear and primary wave velocities respectively, fa is the amplification
factor, H is the height of retaining wall in meter, t is time in second
H
Qha t = m a ( z )a h ( z , t )dz (12)
0
It was observed that effect of the soil parameters are more pronounced on the passive state of
seismic earth pressure compared to that for the active state.
Choudhury and Nimbalkar (2007) have presented pseudo-dynamic method to compute the
rotational displacements of rigid retaining wall supporting cohesionless backfill under seismic
loading for the passive earth pressure condition. The proposed method considers time, phase
difference and effect of amplification in shear and primary waves propagating through both
the backfill and the retaining The influence of ground motion characteristics on rotational
displacement of the wall is evaluated. The rotational displacement of the wall increases
substantially with increase in amplification of both shear and primary waves, time of input
motion, period of lateral shaking and decreases with increase in soil friction angle, wall
10 SRGS, 17th December 2012, Delhi
friction angle. Authors have consider rigid vertical gravity wall, ABKL of height H and width
bw, supporting horizontal cohesionless backfill as shown Figure 9 (a). Figure 9 (b) shows the
variation of rotational displacement ( ) with horizontal seismic acceleration coefficient (kh)
for different values of time history of input motion (t). From the plot, it is seen that the
rotational displacement ( ) increases with an increase in time of input motion. From Figure 9
(b) it is also clear that for kh = 0:3, rotational displacement ( ) corresponding to t = 0.3s is
176.61% and 441.18% smaller than that corresponding to t = 0.4s and 0.5s respectively.
Nimbalkar and Choudhury (2007) have used a rigid vertical retaining wall under passive earth
pressure conditions with cohesionless backfill material for sliding stability. Authors have
expressed results in graphical form show the non-linear variations of the soil passive
resistance factor (FT), wall inertia factor (FI) and combined dynamic factor (Fw) with respect
to the horizontal seismic acceleration coefficient (kh), required for the design of the wall.
Figure 10 (a) shows the model considered by Nimbalkar and Choudhury (2007) for
determining stability against sliding for passive case. Figure 10 (b) shows the variation of the
combined dynamic factor Fw with different values of the period of lateral shaking (T) for φ =
30°, = φ/2, kv = 0.5kh, f = 1.2. From the plot it is seen that for kh = 0.1, the combined
dynamic factor, Fw decreases by 11.32% when T changes from 0.2 to 0.3s, Fw decreases by
13.12% when T changes from 0.3 to 0.4 s and Fw decreases by 15.56% when T changes from
0.4 to 0.5s.
Figure 11 (a) shows a rigid vertical gravity wall, ABKL of height H and width bw, supporting
horizontal cohesionless backfill as considered by Nimbalkar and Choudhury (2008).
Seismic Behaviour of Earth Retaining Structures and Design Issues 11
Figure 10 (a). Details of forces acting on Figure 10 (b). Combined dynamic factor versus
the soil wedge and the wall for sliding kh( = φ /2=15°, f = 1.2, kv=0.5kh) or sliding
case by Nimbalkar and Choudhury (2007) stability by Nimbalkar and Choudhury (2007)
Figure 11 (a). Details of forces acting on the Figure 11 (b). Typical variation of soil thrust
soil wedge and the wall for active case factor FT, wall inertia factor FI and combined
Nimbalkar and Choudhury (2008) dynamic factor for = φ/2 by Nimbalkar and
Choudhury (2008)
The shear wave velocity through the backfill soil, Vss = (Gs/ s)1/22, where, s is the density of
the backfill material and primary wave velocity, Vps = (Gs(2 – 2 s)/ s(1 – 2 s))1/2, where s is
the Poisson’s ratio of the backfill are assumed to act within the soil media due to earthquake
loading. Similarly, the shear wave velocity through the wall, Vsw = (Gw/ w)1/2 where, w is the
density of the wall material and primary wave velocity Vpw = (Gw(2 – 2 w)/ w(1 – 2 w))1/2
where, w is the Poisson’s ratio of the wall material, are assumed to act within the retaining
wall due to earthquake loading.
In Figure 11 (b), the typical variation of design factors viz. soil thrust factor FT, wall inertia
factor FI and combined dynamic factor are presented. From Figure 11 (b), it is clear that the
presence of seismic forces, either with kh or kv, induces reduction in seismic stability of
retaining wall leading to the higher values of the design factors which in turn gives the
12 SRGS, 17th December 2012, Delhi
requirement of higher weight of the wall to maintain equilibrium against sliding under seismic
conditions.
Ahmad and Choudhury (2008) analyzed the reinforced waterfront retaining wall by using
pseudo-dynamic approach. Figure 12 (a) shows typical reinforced wall considered in the
analysis. Figure 12 (b) shows typical results for a set of given input data. Ahmad and
Choudhury (2008) explained the stability of geotextile reinforced waterfront retaining wall
when subjected to both earthquake and tsunami. Authors considered both linear and polylinear
failure surfaces for the analysis. An increase in the required geosynthetic reinforcement force
with an increase in the horizontal seismic acceleration coefficient (kh) was observed. The
effect of tsunami was considered along with seismic and hydrodynamic forces by Ahmad and
Choudhury (2008).
Figure 12 (a). Typical waterfront reinforced Figure 12 (b). Typical results of required
retaining wall subjected to earthquake and reinforcement strength for various seismic
tsunami (Ahmad and Choudhury, 2008) accelerations (Ahmad and Choudhury, 2008)
Ghosh (2008) presented study on seismic active earth pressure behind a non-vertical
cantilever retaining wall using pseudo-dynamic analysis. The effects of soil friction angle,
wall inclination, wall friction angle, amplification of vibration, and horizontal and vertical
earthquake acceleration on the active earth pressure are studied. The method adopted in this
study predicts non-linear variation of active earth pressure along the wall. Rigid, non vertical
retaining wall of height H, with dry, cohesionless, horizontal backfill had been considered by
Ghosh (2008) as shown in Figure 13 (a). The values predicted by Mononobe-Okabe method
are slightly higher than the values obtained by the method proposed by Ghosh (2008). The
difference is because of the pseudo-static method considered by Monobe-Okabe does not
consider the effect of time and phase. It can be seen from Figure 13 (b) that magnitude of
earth pressure increases continuously with an increase in magnitude of horizontal seismic
acceleration coefficient ( h).
Choudhury and Ahmad (2008) presented study of a waterfront retaining wall which, retaining
a partially submerged backfill, subjected to seismic forces using pseudo-dynamic approach.
Authors have also highlighted effect of soil friction angle ( ) on the factor of safety in sliding
mode. It was observed that with increase in the value of from 25 to 35°, there is an increase
in the factor of safety in the sliding mode by 50%.
Basha and Babu (2010) have presented use of pseudo dynamic method to compute the
rotational displacements of gravity retaining walls under passive condition when subjected to
seismic loads. Authors have combined the concept of Newmark sliding block method for
computing the rotational displacements under seismic condition by using the limit equilibrium
Seismic Behaviour of Earth Retaining Structures and Design Issues 13
Figure 13 (a) Model considered by Ghosh Figure 13 (b) Variation of active pressure
(2008) by pseudo-dynamic method for active coefficient Kae with h for φ = 300, H/ =
case 0.3, H/ = 0.16, fa = 1 by Ghosh (2008)
Figure 14 (a). Gravity wall with composite failure Figure 14 (b). Influence of kh and φ on
mechanism consists of logarithmic spiral and rotational displacement ( ) of gravity
planar failure surfaces as considered by Basha and retaining wall by Basha and Babu (2010)
Babu (2010)
Figure 14 (b) explains rotational displacements ( ) for different soil friction angles (φ) and
horizontal seismic acceleration (kh). The results demonstrate that the amount of tilt of the wall
is significantly affected by the increment of soil friction angles (φ) which tend to improve the
available passive resistance. When kh=0.45 and the φ value increases from 20° to 35°, the
rotational displacement ( ) reduces from 30° to 1°. Thus, the major factor which controls the
amount of rotation of wall during an earthquake is the soil friction angle. Basha and Babu
(2010) concluded that assumption of planar failure mechanism for rough soil-wall interfaces
significantly overestimates the threshold seismic accelerations for rotation and underestimates
the rotational displacements.
Ghosh and Sharma (2010) presented the pseudo-dynamic analysis for calculating active
seismic earth pressure for non-vertical retaining wall supporting c- backfill. This work is an
extension of Choudhury and Nimbalkar (2006), who have performed the analysis for
14 SRGS, 17th December 2012, Delhi
sohesionless soils only. Considering a planar rupture surface, the effect of wide range of
parameters like inclination of retaining wall, wall friction and soil friction angle, shear wave
and compression wave velocity, horizontal and vertical seismic coefficients are taken into
account to evaluate the seismic active force.
Bellezza et al. (2012) claim that a more rational pseudo-dynamic approach has been
developed for fully submerged soil under the assumption that a restrained or free water
condition exists within the backfill. The results obtained demonstrate that the analysis
proposed by authors is consistent with the widely used pseudo-static approach. Authors have
also extended their study to study the effect of amplification phenomena. Within a range of
practical interest of soil and seismic input parameters, the acting point of the seismic active
thrust has been found to be very close to a height of H/3 from the base of the wall by authors.
5. OTHER METHODS
Karpurapu and Bathurst (1995) described finite element models that were used to simulate the
behaviour of two carefully constructed and monitored large-scale geosynthetic reinforced soil
retaining walls. The walls were constructed using a dense sand fill and layers of extensible
polymeric (geosynthetic) reinforcement attached to two very different facing treatments. The
model walls were taken to collapse using a series of uniform surcharge loads applied at the
sand fill surface. The numerical simulation work reported in this paper was performed using a
finite element program and ancillary utilities (GEOFEM) developed by the authors.
Ho and Rowe (1996) presented a detailed finite element study for assessing the effect of
different geometric parameters like length of geosynthetic reinforcement, its distribution
along the height of the wall and the number of layers on the overall stability of a typical
reinforced soil retaining wall. A firm rigid base beneath the structure was adopted. It was
observed that the most important parameter affecting the results in terms of stability of the
wall was the ratio of geosynthetic reinforcement length and the height of the wall. The
number of layers of the geosynthetic reinforcement did not seem to the affect the results to
significant extent.
Zeng (1998) described the behaviour of gravity quay walls under earthquake loading using
data from three centrifuge tests. Major damage included lateral possible to calculate the angle
of failure wedge behind are displacement, vertical settlement, and tilting of the gravity
retaining wall. Using a pseudo-static approach, the wedge angle wall, and ground settlement
in the backfill, influence of pore pressure on the wedge angle has also been studied. Figure 15
(a) explains the experimental setup. Figure 15 (a) shows model for used centrifuge models,
the first has dry loose backfill. Figure 15 (b) shows Profile of the Gravity wall before and after
earthquake by Zeng (1998). The gravity wall was observed to settle 0.16 m and moved out by
0.54 m without noticeable tilting. There was a substantial settlement of up to 0.68 m in the
backfill. The area of significant ground settlement extended as far as 11.2 m away from the
wall.
For a wall with saturated backfill, excess pore pressure generated in the soil increases the
horizontal thrust on a wall, deteriorates the stiffness and strength of soil (which can lead to
amplification and phase shift of vibration), and can initiate failure of the structure. The effect
is strongly pronounced when excess pore pressure is high. In such cases the results of limiting
pore water pressure plays a very important role in the design method adopted. Also, they have
performed some studies on fundamental frequency of retaining walls.
Seismic Behaviour of Earth Retaining Structures and Design Issues 15
Many other researchers developed methods for analysis and design of retaining structures in
seismic areas. Shahgholi et al. (2001) used horizontal slice method of analysis for
investigating the seismic stability of reinforced soil walls. Authors considered simplified
formulation of the horizontal slice method by considering the vertical equilibrium for
individual slices together with the overall horizontal equilibrium for the whole wedge, but no
moment equilibrium was taken into the analysis which reduced the number of equations and
unknown from 4N to 2N+1. The failure surface was assumed to be polylinear. Authors had
shown that the results were in close agreement with the results produced using a log-spiral
failure surface.
Atik and Sitar (2010) have conducted an experimental and analytical program was designed
and conducted to evaluate the magnitude and distribution of seismically induced lateral earth
pressures on cantilever retaining structures with dry medium dense sand backfill. Authors
concluded that, designing cantilever retaining walls for maximum dynamic earth pressure
increment and maximum wall inertia, as is the current practice, is overly conservative and
does not reflect the true seismic response of the wall-backfill system. The relationship
between the seismic earth pressure increment coefficien at the time of maximum overall wall
moment and peak ground acceleration obtained from our experiments suggests that seismic
earth pressures on cantilever retaining walls can be neglected at accelerations below 0.4 g.
Figure 16 (a) shows the experimental setup while Figure 16 (b) shows typical results. Figure
17 (b) shows results for total dynamic moment generated in model for motion of Loma Prieta
and Kobe earthquake respectively.
Figure 16 (a). model used for centrifuge test by Atik and Figure 16 (b). typical results
Sitar (2010) as given by Atik and Sitar
(2010)
16 SRGS, 17th December 2012, Delhi
6. CODAL PROVISIONS
IS 1893 - Part 5 (1984), provides information regarding earthquake resistant design for
retaining wall for active and passive case. In seismic design of earth dam, currently pseudo-
static methods are used based on Mononobe-Okabe method of analysis (Seee Kramer, 1996),
which excludes the deformation criteria. As this is a force based method, IS code does not
specify permissible displacements. The static component of the total pressure shall be applied
at an elevation h/3 above the base of the wall. It arbitrarily mentions that the point of
application of the dynamic increment shall be assumed to be at mid-height of the wall, for
active case. For passive case, the static component of the total pressure can be applied at an
elevation (h/3) above the base of the wall. The point of application of the dynamic decrement
shall be assumed to be at an elevation 0.66 h above the base of the wall. It also, mentions the
effect of saturation on lateral earth pressure. Figure 17 shows the wall considered in IS 1893-
Part 5 (1984).
Figure 17. Earth pressure due to earthquake on retaining wall (IS 1893-Part 5, 1984)
Pa = (1/2) W. H2 Ca
(13)
where Ca is given by,
2
(1 ± α v ) cos 2 (φ − λ − α ) 1 (14)
Ca = ×
cos λ cos 2 α cos(δ + λ + α ) 1
sin(φ + δ ) sin(φ − i − δ ) 2
1+
cos(α − i ) cos(δ + α + λ )
vvertical seismic coefficient - its direction being taken consistently throughout the stability
analysis of wall and equal to (1/2) h, where h horizontal seismic coefficient.
Seismic Behaviour of Earth Retaining Structures and Design Issues 17
(1 + α v ) cos 2 (δ + α − λ ) 1 (17)
Cp = ×
cos λ cos 2 α cos(δ − α + λ ) 1
sin(φ + δ )sin(φ + i − δ ) 2
1+
cos(α − i ) cos(δ − α + λ )
where φ is soil friction angle, friction angle for soil and wall, is defined by Eqn. 15, as
shown in Figure 17.
Eurocode 8 (2003) explains the design of structures for earthquake resistance, wherein part 5
explains the procedure for foundations, retaining structures and geotechnical aspects. The
seismic calculations for Eurocode 8 (2003) are also based on pseudo-static method based on
approach given by Richards and Elms (1979). Hence, the permissible displacements for
translation and rocking mode are considered in the analysis. Design seismic inertia forces FH
and FV acting on the ground mass, for the horizontal and vertical directions respectively, in
pseudo-static analyses shall be taken as (Clause 4.1.3.3),
FH = 0,5 * * S * W
FV = ±0,5FH , if the ratio avg/ag is greater than 0.6
(18)
FV = ±0,33FH , if the ratio avg/ag is not greater than 0.6
where,
is the ratio of the design ground acceleration on type A ground, ag, to the
acceleration of gravity g;
avg is the design ground acceleration in the vertical direction
ag is the design ground acceleration for type A ground (also, topographic amplification factor
for ag shall be taken into account according to clause 4.1.3.2 (2);
S is the soil parameter of EN 1998-1:2004, 3.2.2.2
W is the weight of the sliding mass.
Eurocode 8 (2003) highlights guidelines to take into account values of kh and kv in absence of
any study. Eurocode Clause 7.3.2.2 explains guidelines for seismic action as,
(l) For the purpose of the pseudo-static analysis, the seismic action shall be represented by a
set of horizontal and vertical static forces equal to the product of the gravity forces and a
seismic coefficient.
18 SRGS, 17th December 2012, Delhi
(2) The vertical seismic action shall be considered as acting upward or downward so as to
produce the most unfavourable effect.
(3) The intensity of such equivalent seismic forces depends, for a given seismic zone, on the
amount of permanent displacement which is both acceptable and actually permitted by the
adopted structural solution.
(4) In the absence of specific studies, the horizontal (kh) and vertical (kv) seismic coefficients
affecting all the masses shall be taken as:
S
kh = α ,
r
kv = ±0.5kh , if a vg /a g is larger than 0.6
kv = ± 0.33kh , otherwise
where, kh and kv are seismic horizontal and vertical coefficients, ratio of the design ground
acceleration on type A ground, ag, to the acceleration of gravity g, avg is design ground
acceleration in the vertical direction, ag is design ground acceleration on type A ground. This
assumes point of application of dynamic increment at mid height of the wall. Hydro-dynamic
force for saturated backfill is assumed to act at 0.4H from the base of the wall with height H.
IBC (2006) categorizes EARTHQUAKE LOADS into various categories. It categories the
sites into categories namely A, B, C, D, E, F based on soil profile, shear wave velocity,
standard penetration values and undrained shear strength values. Based on that, the design
seismic category should be selected. It mentions that Retaining walls shall be designed to
ensure stability against overturning, sliding, excessive foundation pressure and water uplift.
Retaining walls shall be designed for a safety factor of 1.5 against lateral sliding and
overturning.
7. FIELD OBSERVATIONS
Tatsuoka et al. (1996) explained various case histories of reinforced wall performances which
were constructed at various locations in Japan. It was observed that the ‘Tanata wall’
performed well in the event of seismic activity. Even though the intensity was severe at the
site, this wall was observed to perform better compared to the other type of unreinforced walls
explained as indicated by the authors. The total length of the wall was 305m, and the greatest
height was 6.2m. Figure 18 (a) shows the typical wall cross section while Figure 18 (b) shows
the collapse of the wooden houses after the earthquake, whereas the reinforced soil-wall is
standing in its position. This wall deformed and moved slightly in lateral direction. Hence, it
was mentioned that the geotextile reinforced soil-retaining walls performed better when they
were exposed to severe seismic loadings.
Sitar et al. (2012) have studies number of case history data and the experimental work to show
that the current methods are based on conservative design procedure in region where PGA
exceeds 0.4g. Authors also commented based on experimental data that the seismic pressure
distribution for moderate size retaining structures, on the order of 6-7 m high, is triangular
increasing with depth. Authors also commented that there is no significant increase in seismic
earth pressure between unbraced and braced structures with fixed base, while the loads on free
standing cantilever structures are substantially lower owing to their ability to translate and
rotate. Authors also mention a well documented case history of 1971 San Fernando
Seismic Behaviour of Earth Retaining Structures and Design Issues 19
earthquake. Clough and Fragaszy (1977) showed that the concrete cantilever structures, well
designed and detailed for static loading, performed without any sign of distress at
accelerations upto 0.4g. This is consistent with the conclusions mentioned by Seed and
Whitman (1970).
Figure 18 (a). Cross-section of geogrid Figure 18 (b). Failure of the collapsed wooden
reinforced soil-wall (Tanata wall) houses in comparison to reinforced soil-wall
(Tatsuoka et al., 1996) under earthquake condition (Tatsuoka et al.,
1996)
Prakash (1981) identified three questions that need to be answered when designing a retaining
wall for seismic loads.
1. What is the magnitude of total (seismic plus dynamic) earth pressure on the wall?
2. Where is the point of application of the resultant?
3. How much has the structure been displaced?
To answer these questions, many researchers have performed numerous experimental studies
in the past and their results are very much a in connection with their experimental design.
Either shaking table or the geotechnical centrifuge is used to approach the problem in a cost
effective way. Structures should be designed to resist additional seismic forces by using any
of the methods explained in this paper.
General guidelines given in Eurocode (1998) for retaining wall design are:
1. The choice of the structural type shall be based on normal service conditions, following
the general principles of EN 1997-1:2004, Section 9.
2. Proper attention shall be given to the fact that conformity to the additional seismic
requirements may lead to adjustment and, occasionally, to a more appropriate choice of
structural type.
3. The backfill material behind the structure shall be carefully graded and compacted in situ,
so as to achieve as much continuity as possible with the existing soil mass.
4. Drainage systems behind the structure shall be capable of absorbing transient and
permanent movements without impairment of their functions.
5. Particularly in the case of cohesionless soils containing water, the drainage shall be
effective to well below the potential failure surface behind the structures.
6. It shall be ensured that the supported soil has an enhanced safety margin against
liquefaction under the design earthquake.
It mentions that, the point of application of the force due to the dynamic earth pressures shall
be taken to lie at mid-height of the wall, in the absence of a more detailed study taking into
account the relative stiffness, the type of movements and the relative mass of the retaining
structure. Eurocode also highlights on Stability of foundation soil. It mentions that overall
stability and local soil failure be checked.
9. CONCLUSIONS
1. Calculation of seismic earth pressure is very important for design of retaining structures
in the earthquake prone areas.
2. Pseudo-static method was the first attempt for analyzing structures in seismic areas.
Though it has serious limitations, it is widely used till date because of it’s simplicity.
3. Pseudo-dynamic method considers the time dependant nature of the earthquake force.
Hence has more accuracy as compared with pseudo-dynamic methods.
Seismic Behaviour of Earth Retaining Structures and Design Issues 21
REFERENCES
Indrajit Chowdhury1
1
Head of the department
Civil and Structural Engineering
Petrofac International Limited
Sharjah, United Arab Emirates
Indrajit.Chowdary@petrofac.com
ABSTRACT
This paper presents a unified formula for determination of seismic response of cantilever
type retaining walls either to be constructed or already built and that which needs to be
seismically qualified. Considering the time period of the soil and wall that remains the
fundamental dynamic property, modal analysis technique is applied to estimate the
amplitude and finally dynamic moments and shears in the wall. The analysis also takes
into consideration the extreme case when the backfilled soil gets liquefied and also
subjected to overburden load. Results are presented for various types of soils and
comparison is also carried with the conventional Mononobe and Okabe’s method.
Keywords: Shape function, Modal analysis, Time period, Liquefaction, Moment, Shear.
INTRODUCTION
Retaining walls play an important role in a post earthquake scenario to retain the backfill
in a number of industrial and infrastructure projects. A number of researchers have
worked on seismic response of such retaining walls around the world, like Mononobe
(1929), Okabe (1924), Seed and Whitman (1970) Whitman et al (1990, 1991) to name
some of the pioneering few. However all these researches are based on the assumption
that the wall is gravity type where it has got an extremely high stiffness, such that the
seismic excitation is restricted to soil part only. Though it is a fact that with advent of
reinforced concrete technology retaining walls have undergone a significant change in its
character, where, in today’s world it would be most improbable that a gravity wall will be
deployed for retaining back fills even for heavy bridge girders.
Figure-1 Gravity retaining walls and reinforced concrete retaining wall used to retain
soil.
Shown in figure-1 are retaining walls used earlier and the RCC retaining walls that are
used presently. It is apparent that present day walls are far too flexible and basic
assumption deployed by previous researchers that the wall is infinitely stiff- cannot be
justified for these modern day retaining walls. Thus the pseudo static approach
considered till date for determination of dynamic pressure under seismic load (usually
based on Mononobe & Okabe’s(M-O) method) may not be justified now. It is apparent
that the present day constructed walls do have a finite stiffness vis-a vis time period that
will surely influence the dynamic response of these walls under earthquake force.
The M-O method that was considered for a cohesion less dry backfill(c=0) has also been
examined by a number of researchers in India and abroad like, Das and Puri (1996),
Ghosh et al (2010,2008, 2007) Saran et al ( 1968, 2003), Choudhury et al
(2002,2004,2006) to name some of the noted works. However, most of them are a
variation of M-O method in one form or other, trying to incorporate other soil conditions
like c- soil, or using logarithmic spiral curves etc within the M-O frame work. Recent
research carried out in USA by Ostadan & White (1997), Ostadan(2004) has also shown
that M-O based methods significantly under predict the dynamic pressure under seismic
loads, to the extent that Nuclear regulatory board of USA has now stopped using any of
the M-O based methods for determining earth pressure for any of their structures.
Recently, Chowdhury et al (2011, 2012) has proposed a method based on modal response
technique for estimating the seismic response for such cantilever and counterfort
retaining walls. A simplified version of the same is presented herein that is quite
amenable to design office practice as well as estimating the response of already
constructed walls.
PROPOSED METHOD
To start with we take the simplest of the case as shown below
q kN/m
A D
B 45 / 2
Figure-2 A Cantilever retaining wall with general c- backfill and overburden
The soil profile under active case is at incipient failure when the failure line makes
angle = tan(45+/2) as shown in the above figure.
Since soil profile is already under failed condition under static load, it will not induce
any stiffness to the overall dynamic response but will only contribute to the inertial
effect.
Since the cantilever wall is relatively thin ,mass contribution of the wall itself may be
ignored compared to that of the soil. The wall thus contributes only to stiffness of the
overall soil-structure system
The retaining wall is fixed at base and foundation compliance has been ignored for
the present analysis.
It will be observed that assumptions made above are identical to what Mononobe or
Steedman and Zeng (1990) has assumed in their analysis. Based on above assumptions the
analysis is carried out as elaborated hereafter.
Dynamic response of wall with c- backfill and over burden load q
For a general c-soil the active earth pressure is expressed [Murthy (1984)] as
pa s z cot 2 2c cot (1)
where, = 45+/2 , = angle of friction s= Weight density of soil c= cohesion
Considering the overburden pressure the active earth pressure can be expressed as
The equation of equilibrium for the wall considered as a fixed base cantilever beam can
be expressed as
d 4u
EI 4
s .z. cot2 2c cot q cot2 (3)
dz
On successive integration of equation (3), and imposing the boundary conditions
d 3u
1) At z=0 0 4a)
dz3
d 2u
2) At z=0 2 0 4b)
dz
du
3) At z=H 0 4c)
dz
4) At z=H u 0 4d)
We finally have
s H 5 cot 2 5 5 4 4 q cot 2 H 4 4 4
u 1 1 (5)
30 EI 4 4 3 3 8EI 3 3
s H 5 cot 2 5
5 4 4 4 4
u 1 (6)
30 EI 4 4 3 3 3 3
15H c 15q
In equation (6), as mentioned above, tan , , H c 2c / s , z / H
4H 4 s H
Thus the maximum static deflection is obtained ( 0 ) as
s H 5 cot 2
u static 1 (7)
30 EI
ustatic
T 2 (8)
g
We finally have
sH 5
TA 3.97 cot 1 (9)
Et 3 g
For modal analysis, maximum amplitude (Sd) is expressed as (Clough 1984)
S
S d a2 (10)
Here Sa= Spectral acceleration and 2 / T , natural frequency of the wall.
Sa
S d (11)
2
n
m i i
Here Modal mass participation factor and is expressed as i 1
n
m
2
i i
i 1
= A code factor expressed as ZI/2R where Z= Zone factor I= Importance factor and R=
Response reduction factor.
Thus based on equation (15) the dynamic amplitude of the wall can be expressed as
Sa 2
u T f ( ) (12)
4 2
S z3 z2 q
M z s cot 2 a 1 H c tan (14)
g 6 2 s
S z2 q
Vz s cot 2 a 1 z H c tan (15)
g 2 s
Equation (13) to (15) gives the displacement, moment and shear under seismic loading
for the most general condition of soil.
When there is no overburden i.e., 0 , the formulas converges to the case of c-soil
only. When again, 0 , the equations converges to the case of pure cohesion less
soil(c=0).
Finally an interesting case is observed, when, 0 , 0 and even
=0(i.e. cot 2 1) , the values obtained in Equations (13),(14), (15) are not zero and
finite and converges to a hydrostatic case. Thus if we take density of soil in this particular
case (sat-w) the reflects a case when the soil is in a liquefied state
1
5 5 4 4 4 4
4
4
1
3
3
3
3
d
10
2
(16)
5
5 4
4 4
4
0 4 4 1 3 3 3 3 d
Equation (16) on integration can be explicitly expressed as
5 1
42 9 (17)
16 2
315
2
32
315
295
2772
25
448
Considering the effect of vertical acceleration as Sv =Sa/2 as per IS-1893 (2002) we have
Here FR= Friction force along the surface BD that resists the motion.
W= Weight of the soil body ABD (W=m.g)
45 / 2 for active case
Internal angle of friction of soil @ tan
Considering the resistive friction force FR as the damping force Equation (20) can be
expressed as
C.v W (sin cos ) (21)
Here C= damping of the system, velocity v=Sa/ where Sa is the seismic acceleration and
the natural frequency @ k / m
Equation (21) on simple algebraic manipulation can be expressed as
C sin cos
(22)
Cc Sa
2
g
Here Cc= Critical damping =2 k / m
Considering
C
the damping ratio we have
Cc
sin cos
A (23)
Sa
2
g
The above gives the damping ratio of the soil in active case. For passive case
sin p cos p
p , where p 45 / 2 (24)
Sa
2
g
For a conservative estimate we should consider to be minimum- but having a finite
rational value. This is valid either when the numerator in Equation (23) and (24) is
minimum or the denominator is maximum.
For numerator to be minimum, it must be zero which gives =90 which is impossible to
achieve. Thus the condition is denominator is maximum. In other words Sa/g is to be
maximum.
For instance maximum value of Sa/g as per IS 1893(2002) is 2.5.Applying this value we
have
sin cos
A (25)
5
For sandy soil value usually varies from 15 degree for very loose sand to 40 degree for
very dense sand. Considering the above values, variation in damping ratio, for active and
passive cases are as shown hereafter.
Com parison of dam ping ratio for active and passive
case
0.25
Damping ratio
0.2 Damping ratio
0.15 active
0.1 Damping ratio
0.05 passive
0
15 20 25 30 35 40
phi
It will be observed that variation of damping ratio with respect to friction angle is not
widely varying thus an estimated value of 15% in active case and 20% for passive case
would cater for soil with almost all levels of in-situ compaction. The damping ratio
values also looks quite reasonable and matches the data that are usually considered from
experience for practical seismic design of such walls based on FEM.
For c- soil the damping ratios can be expressed as
cH cos ec
sin cos
A W and (26)
5
cH cos ec p
sin p cos p
p W (27)
5
for general c- soil (Figure-1) W=0.5sH2cotand for csoil with overburden
W=0.5sH2cotqcotReplace by p for passive case
The 5% material damping of wall may be added to above arrive at the design damping
ratio.
The above chart depicts the values of moments and shears without the effect of vertical
acceleration. If vertical acceleration is taken into cognizance, moments an shears shown
above are to be multiplied by a factor 1.5.
600.00
500.00
c-phi s oil
400.00 phi s oil
Moment(kN.m)
c-s oil
300.00
c-phi
200.00 +overburden
c+overburden
100.00 phi+overburden
Liquified s oil+q
0.00
0 0.6 1.2 1.8 2.4 3 3.6 4.2 4.8 5.4 6
-100.00
Depth(m)
250
150
c-s oil
100 c-phi +overburden
c+overburden
50
phi+overburden
0 Liquified s oil+q
0 0.6 1.2 1.8 2.4 3 3.6 4.2 4.8 5.4 6
-50
Depth(m)
70
60
50
pressure(kN/m2)
40 Static +Dynamic(proposed)
30 Static +Dynamic(M-O)
20
10
0
0 0.6 1.2 1.8 2.4 3 3.6 4.2 4.8 5.4 6
z(m)
Figure-9: Variation of Dynamic pressure by proposed method and M-O method for
pure sandy soil(c=0) with no overburden.
Figure 7 and 8 shows the variation of moments and shears for various type of soils along
the depth of the wall.It is observed surcharge load in proximity of the wall heavily
influences the dynamic response and has an amplifying effect.Thus while designing the
retaining walls engineer should carefully consider its effect on the overall dynamic
response.
The most dangerous case is when the soil undergoes liquefaction including
oberburden.Though due to liquefaction the overburden structure impinging the load
might collapse but could generate an instantaneous case when the impinging overburden
shoots up the moment aand shear significantly on the wall that could render either a
structural faliure of the wall or worse induuce an instabilty by either topling or sliding of
the wall – where the failure can be far more damaging.
For general case of sandy soil(c=0) with no overburden, static plus dynamic pressure by
the proposed method is compared with M-O method ( static plus dynamic) vide figure(9)
and as predicted at the outset, it is ovserved that the conventional M-O method
significantly under predicts the dynamic response.
CONLUSION AND REMARKS
A compreshensive analytical solution based on modal analysis is proposed herein that
takes into cognizance the effect of time period of the wall, the effect of which, has been
mostly ignored by the previous researchers.It also gives solution to almost all types of
soil and loading condtions that may be expected in a real field design- including
liquefaction whose effect on the wall surely needs more research.
Finally a retrospective comment on the work. The solution is fundumental and almost on
the brink of being trivial in approach.It is quite surprising to the author that nobody has
approached the problem the way it is proposed herein. It is possibly because of the
perinial aversion of the structural engineers to deal anything with soil parameters ( except
perhaps the bearing capacity of foundation and that too a unique number !!), and on the
other hand, pathogenic aversion of classical soil mecahnists,so as not to consider the
fourth order differntial equation of beam (considering it as strucutral engineering- as
such none of their business) has possibly left this fundamental problem unattended, for
last eighty years or more.
REFERENCE
1. Choudhury D & Subba Rao K.S. (2002); "Displacement - Related Active Earth
Pressure", International Conference on Advances in Civil Engineering (ACE-
2002), January 3 - 5, 2002, IIT Kharagpur, India, Vol.2., pp. 1038-1046.
2. Choudhury D, Sitharam T.G.& Subba Rao K.S.(2004); "Seismic design of earth
retaining structures and foundations", Current Science, (ISSN: 0011-3891, IF:
0.694/2003) India, Vol. 87, No. 10: pp. 1417-1425.
3. Choudhury D & Chatterjee S (2006); "Displacement - based seismic active earth
pressure on rigid retaining walls", Electronic Journal of Geotechnical
Engineering, (ISSN: 1089-3032), USA, Vol. 11, Bundle C, paper No. 0660.
4. Chowdhury I & Dasgupta S.P.(2002) “Dynamic Analysis of RCC Retaining wall
under Earthquake Loading”- ; Electronic Journal of Geo-technical Engineering
Vol-8C 2003.
5. Chowdhury I and Dasgupta S.P. (2011) An analytical solution to dynamic
response of cantilever retaining walls with generalized backfill Electronic Journal
of Geotechnical Engineering, USA, Vol. 16, Bundle C.
6. Chowdhury I, Singh J and Dasgupta S.P. (2012) Seismic response of counterfort
retaining walls with generalized backfill Electronic Journal of Geotechnical
Engineering, USA, Vol. 16, Bundle R.
7. Clough R.W.(1984) “Dynamics of Structures” M’cgrawhill Publications New
York USA.
8. Das B.M. & Puri V.K.(1996) “Static and dynamic active earth pressure”,
Geotechnical and Geological engineering Vol-14, pp-353-356.
9. Ghosh S and Saran S(2007) “Pseudo static Analysis of Rigid Retaining wall for
Dynamic Active Earth Pressure” Cenem B.E.College Kolkata India
10. Ghosh S, Dey G.N.,& Datta B.N.(2010) “Pseudo static Analysis of Rigid
Retaining wall for Dynamic Active Earth Pressure” 12th International Conference
of International Association for Computer Methods and Advances in
Geomechanics.
11. Ghosh S and Pal J(2010) “Extension of Mononobe-Okabe expression for active
earth force on retaining wall backfilled with c- soil”14th Symposium on
Earthquake Engineering, Indian Institute of Technology Roorkee India Vol-1 pp
522-530.
12. IS-1893(2002) – Code for Earthquake resistant design of Structures; Bureau of
Indian Standard Institution, New Delhi, India.
13. Mononobe N & Matsuo H (1929), “On the determination of earth pressure during
earthquakes”, Proc. World Engineering Congress, Tokyo, Vol. 9, Paper 388.
14. Murthy V.N.S. (1984) “Soil Mechanics and Foundation Engineering “ Sai Kripa
Publication Bangalore India.
15. Okabe S (1924), “General theory of earth pressures and seismic stability of
retaining wall and dam”, J. Japanese Society of Civil Engineers, Vol. 12, No. 1.
16. Ostadan F and W. H. White (1997), “Lateral seismic soil pressure, An updated
approach”, Bechtel Technical Group Report Los Angles USA
17. Ostadan F (2004), “Seismic soil pressure on building walls-An Updated
approach”, 11th International Conference on Soil Dynamics and Earthquake
Engineering. University of California, Berkeley, January.
18. Saran S & Prakash S (1968) “Dimensionless Parameters for Static and dynamic
earth pressures behind retaining walls”, Indian Geotechnical Journal Vol. (72(3)
pp 295-310.
19. Saran S and Gupta R.P.(2003) “Seismic Earth Pressure behind retaining walls”
Indian Geotechnical Journal Vol. 33(3) pp195-213.
20. Seed H.B. & Whitman R.V. (1970), “Design of earth retaining structures for
seismic loads”, ASCE Specialty Conference on Lateral Stress in Ground and
design of Earth Retaining Structures, June.
21. Steedman R.S. & Zeng X (1990), “The Seismic response of Waterfront Retaining
walls”, Proceedings on Specialty Conference on design performance of Earth
Retaining Structures, Special Technical Publication 25 Cornell University Ithaca
New York pp 897-910.
22. Whitman R.V. (1991), “Seismic design of Earth Retaining structures”,
Proceedings 2nd International conference on Recent advances in Geotechnical
Earthquake Engineering and Soil Dynamics, St Louis USA, March 11-15.
SECTION C
Section C collates the ingredients necessary for back-analysis and seismic requalification.
They are:
(i) Seismic Hazard Assessment and Microzonation: This will provide us with the
scenario earthquake
ABSTRACT
Seismic hazards that are associated with design of foundations are presented in this
paper. A review of standard methods for quantifying various seismic hazards has been
discussed. Lastly, seismic hazard specific to the Indian seismic settings has been
summarised.
Key words: Earthquake, Seismic Hazards, Faulting, Landslide, Liquefaction
INTRODUCTION
Different seismic hazards are quantified differently. A designer would like to know the following
from this analysis:
Peak Ground Acceleration (PGA) i.e. the amplitude of vibration
Earthquake induced ground displacement
Duration of the earthquake: This is important from the point of view of how long a soil
will remain liquefied.
Frequency content of the ground motion: The predominant frequency content of most
earthquakes lies in the range of 0.5 to 10 Hz.
Area
Source M=?
M=? PGA = ?
Fault PGA = ?
(Line Source)
Step-1: Finding the nearest fault and Step-2: Calculate the maximum earthquake
area source of earthquake. that could trigger from each source.
Peak Acceleration
Magnitude M
PGA = ? or PGV = ?
Distance
Step-3: Select the controlling earthquake for the Step-4: Specify the peak ground acceleration or velocity
site i.e. which earthquake has most severe effect or other measure that describe the earthquake effect for
(generally peak ground motion) at site. the site under consideration.
Area Source
Line Source
Site
Area Source
Line Source
Earthquake Magnitude, M
Step-1: Identify the source of potential Step-2: Evaluate the seismicity of earthquake
earthquake and represent graphically. sources from recurrence relationship.
M-1
M-2 Increasing
M-3 Magnitude, M
Probability of Exceedence
Ground Motion Level
Step-3: Develop the ground motion Step-4: By integration of models, produce the
attenuation models seismic hazard curves which express the
probability of exceeding various levels of
ground motion within some time period
PGA (Peak Ground Acceleration) at a site will vary with the magnitude of the earthquake, its
depth of focus, distance from the point to the epicenter or closest point on the rupturing fault plane.
Various such empirical relationships exists to evaluate PGA, see for example Ambraseys et al
(2005a, 2005b). Other sources can be found in section 3.4.3 in Kramer (1996). However, the
problem lies mainly in low seismicity areas and where the past earthquake records are not available.
Another source for obtaining ground shaking characteristics of a particular location is the GSHAP
map, which is described below.
PGA from GSHAP Studies
The GSHAP [Global Seismic Hazard Assessment Program] is a major project of the
UN/International Decade of Natural Disaster Reduction, was conducted in the 1992-1998 period
with the goal of improving global standards in seismic hazard assessment. The GSHAP provides
Global Seismic Hazard Map that has been compiled by joining the regional maps produced for
different GSHAP regions and test areas. It depicts the global seismic hazard as peak ground
acceleration (PGA) with a 10% chance of exceedance in 50 years, corresponding to a return period
of 475 years. The basic elements of modern probabilistic seismic hazard assessment that are used in
GSHAP are (http://www.seismo.ethz.ch/GSHAP/):
Earthquake Catalogues and Data bases: the compilation of a uniform database and
catalogue of seismicity for the historical (pre-1900), early-instrumental (1900-1964) and
instrumental periods (1964-today).
Earthquake Source Characterization: the creation of a master seismic source model to
describe the spatial-temporal distribution of earthquakes, using evidence from earthquake
catalogues, seismotectonics, paleoseismology, geomorphology, mapping of active faults,
geodetic estimates of crustal deformation, remote sensing and geodynamic models.
Strong Seismic Ground Motion: the evaluation of ground shaking as function of
earthquake size and distance, taking into account propagation effects in different tectonic
and structural environments and using direct measures of the damage caused by the
earthquake (the seismic intensity) and instrumental values of ground motions.
Computation of Seismic Hazard: the computation of the probability of occurrence of
ground shaking in a given time period, to produce maps of seismic hazard and related
uncertainties at appropriate scales.
Figure 3 shows the predicted PGA motion for South Asia obtained from the GSHAP. Figure 4
shows the seismic hazard map of India and adjoining regions for 10% probability of exceedance in
50 years having a contour interval 0.05g based on GSHAP. Details of the studies can be found in
Bhatia et al (1997). Table 2 lists the PGA for few cities in India based on GSHAP studies. In the
same table, the seismic zone factor is also given for comparison. Figure 5 shows the seismic
zonation of India based on IS1893 (2002). It must be mentioned that GSHAP maps are based on the
tectonic map of India and is shown in Figure 6. Details of the study and its basis can be found in
Bhatia (1997).
Figure 3: Predicted PGA for South Asia, after Bhatia et al, 1997 (see Plate-2 for coloured map)
Figure 4: Seismic hazard map of India and adjoining regions for 10% probability of exceedance in 50
years. (Contour interval 0.05g) after Bhatia et al, 1997 (see Plate-3 for coloured map)
Table 2: PGA of few cities of India based on GSHAP hazard map
Peak Ground Acceleration Seismic Zone as per Zone factor ‘Z’ as per
Location
(g) from GSHAP map IS 1893 (part-1):2002 IS 1893 (part 1):2002
Ahmadabad 0.15 - 0.20 III 0.16
Bhuj 0.15 – 0.20 V 0.36
Chennai 0.00 – 0.05 III 0.16
Dehradun 0.25 – 0.30 III 0.16
Delhi 0.00 – 0.05 IV 0.24
Gowhati 0.30 – 0.35 V 0.36
Hyderabad 0.00 – 0.05 II 0.10
Kolkatta 0.05 - 0.10 III 0.16
Lucknow 0.00 – 0.05 III 0.16
Mumbai 0.05 – 0.10 III 0.16
Patna 0.05 – 0.10 IV 0.24
Srinagar 0.30 – 0.35 V 0.36
Fault Movement
The fault movement during a seismic event is basically characterized by the amount of ground
displacement. There are many established empirical relationships for estimating the fault displacements (δf)
according to earthquake moment magnitude (M). Wells and coppersmith (1994) proposed the following
relationships based on the pervious and recent earthquake data for various fault types.
For strike slip fault: log f 6.32 0.90M (2)
For a poorly known fault or blind fault: log f 4.80 0.69M (5)
Landslide
While quantifying the ground movement due to landslide, all the potentially unstable slopes should be
identified and the nature of the movement must be ascertained. Many empirical methods are often used to
estimate the amount of landslide movement, for details see Wilson and Keefer (1983) and Jibson and Keefer
(1993).
Liquefaction
The occurrence of liquefaction for a particular intensity of seismic event at a given site can be found out
from liquefaction hazard map. The liquefaction hazard map for India is not available. So the site specific
studies have to be carried out to evaluate the liquefaction potential of the site.
The duration of strong earthquake ground motion is one of the main parameters characterizing this natural
hazard. This is important because the amount of cumulative damage that structures may incur increases with
number of cycles of loading. Seismologists have proposed several definitions to quantify the duration of
strong motion. Most widely accepted definitions are: (a) Significant duration; (b) Bracketed duration.
1. Significant duration is defined as the time (t) required to build up from 5% to 95% of the integral
acceleration (a) record a 2 (t) dt ; see Trifunac and Brady (1975).
2. Bracketed duration is defined as the time elapsed between the first and last occurrence of the
earthquake time record above a certain value of 0.03 g [Ambraseys and Sarma (1967)] or 0.05 g
[Bolt (1969)].
An empirical correlation is given by Dobry et al (1978) to estimate the significant duration (Ts) based on
the magnitude (M) of the earthquake, see Equation (5). The correlation was developed for Western United
States with earthquake magnitude ranging from 4.5 to 7.6.
log (Ts ) 0.423M 1.83 (5)
The response of foundations to earthquake motion also depends on the frequency of the dynamic loads.
The frequency content describes clearly how the amplitude of ground motion is distributed among different
frequencies. The frequency content of a ground motion can be obtained by transforming the ground motion
from time domain to a frequency domain through a FFT [Fast Fourier Transform].
CONCLUSION
The available methods for seismic hazard assessment are described with special reference to the
seismicity of India. The basic principles behind the deterministic and probabilistic assessment of PGA (Peak
Ground Acceleration) are illustrated. An alternative approach to evaluate the PGA based on GSHAP studies
has been discussed. Current state-of-the-art methods to quantify seismic hazard has also been presented.
REFERENCES
Ambraseys, N.N., Douglas, J., Sarma, S.K., and Smit, P.M. (2005a), “Equations for the estimation of strong
ground motions from shallow crustal earthquakes using data from Europe and the Middle East: Horizontal
Peak Ground Acceleration and Spectral Acceleration”, Bulletin of Earthquake Engineering (2005)
Volume 3, pp 1-53. © Springer.
Ambraseys, N.N., Douglas, J., Sarma, S.K., and Smit, P.M. (2005b), “Equations for the estimation of strong
ground motions from shallow crustal earthquakes using data from Europe and the Middle East: Vertical
Peak Ground Acceleration and Spectral Acceleration”, Bulletin of Earthquake Engineering (2005)
Volume 3, pp 55-73. © Springer.
Ambraseys, N.N. (1995), “The prediction of earthquake peak ground acceleration in Europe”, Earthquake
Engineering and Structural Dynamics, Vol. 24, pp 467-490s.
Ambraseys, N.N and Sarma, S.K. (1967), “Response of Earth Dams to Strong Earthquakes”, Geotechnique,
Vol. 17, No 3, pp 181-213, Sept 1967.
Bhatia S C , H.K. Gupta, M. Ravi Kumar, N.P. Rao, G.R. Chitrakar, P. Zhang and Z-X Yang, 1997. Seismic
hazard map of GSHAP Area XII comprising eastern Himalaya, Northeastern India, Burmese arc, South
China and adjoining regions, Abstract, 29th General assembly of IASPEI, August 18-28, 1997,
Thessaloniki, Greece, p387.
Bolt, B.A. (1969), “Duration of strong motion”, Proceedings of the 4 th World Conference on Earthquake
Engineering, Santiago, Chile, pp 1304-1315.
Dobry, R., Idriss, I.M., and Ng, E., (1978): “Duration Characteristics of Horizontal Components of Strong-
Motion Earthquake Records”, Bulletin of the Seismological Society of America; October 1978; v. 68; no.
5; p. 1487-1520.
Jibson, R.W., and Keefer, D.K. (1993), “Analysis of the seismic origin of landslides: examples from the New
Madrid seismic zone”, Geological Society of America Bulletin, April, Vol.105, pp.2521-536.
Wells, D.L. and Copersmith, K.J. (1994), “New empirical relationships among magnitude, rupture length,
rupture width, rupture area, and surface displacement”, Bulletin of Seismological Society of America,
August, Vol.4, pp.974-1002.
Wilson, R.C. and Keefer, D.K. (1983), “Dynamic analysis of a slope failure from the 6 August 1979 Coyote
Lake, California Earthquake”, Bulletin of Seismological Society of America, April, Vol. 23, pp.863-877.
Kramer, S.L. (1996), Geotechnical Earthquake Engineering, Prentice Hall International Series Publication.
Trifunac, M. D., and A. G. Brady (1975). On the correlation of seismic intensity scales with the peaks of
recorded ground motion, 65, 139-162.
UKIERI International Workshop on Seismic Requalification of Geotechnical Structures (SRGS)
17th December 2012, Delhi.
ABSTRACT:
The use of spectral-matching software, such as RSPMatch, allows the engineer to alter seed ground
motions in the time domain to provide a very close match of a target response spectrum. Compared to
the use of real recorded ground motions, a smaller number of spectrally-matched records can reliably
be used, as the variability is less. A set of criteria for selecting seed motions for RSPMatch is
presented, in which initial spectral shape is considered to be of primary importance, and other ground
motion characteristics are used to filter the initial database. Rather than using earthquake magnitude
bounds as a proxy for ground motion duration, as recommended by other authors, we suggest filtering
explicitly on duration. For structures with a long fundamental period, the frequency content of the
input motion is also very important, to avoid excessive noise at periods of interest in the original seed
record, and excessive manipulation of the record in RSPMatch.
Keywords: RSPMatch, ground motion selection, spectral matching, ground motion duration
1. INTRODUCTION
Engineers need to select/develop site-specific ground motions for designing foundations for
important structures which require non linear time history analysis. Numerous design codes
and research publications provide guidance for the selection of seismic ground motions for
use in this kind of analysis. (Bommer and Acevedo, 2004). Mostly these documents are based
on the use of a number of real recorded ground motions, scaled by a constant factor over the
full duration to be approximately consistent with the expected seismic hazard at the site.
Seismological characteristics of the records, such as earthquake magnitude and epicentral
distance, are usually considered in the selection of real records, as they are likely to influence
the shape of the response spectrum, the energy content and duration of ground shaking, and
therefore the expected demand on structures. This subject has received great deal of attention
from structural engineers (as evident from the vast array of publications) but with the
emphasis on performance based design for geotechnical structures it is likely to become
equally important for geotechnical engineers. A typical foundation design process in seismic
areas is shown in Figure 1. After identifying the expected seismic hazard at the site, selection
of appropriate earthquake records forms an important step in the process. The parts of the
design methodology where these records are used are highlighted in the design chart along
with some typical design tools (softwares). In selecting these records the only data available to
the engineers through their seismologist colleagues comprises of the earthquake source
mechanisms in the vicinity of the region (location of the fault and the likely rupture
characteristics) and in some cases few recorded data of past earthquakes in that region. For the
cases, where some recorded data is available – manipulation of these records can generate the
required site–specific motion to the satisfaction of the all stake holders.
1
2 SRGS, 17th December 2012, Delhi
However, where these records are unavailable, artificial ground motions has to be generated
for analysis and design purposes. In these cases, the concerned engineers are left with two
pointers to generate an input motion: (a) Country-specific or region-specific code of practice
which will provide a response spectrum for different type of soils [typically soft soil, stiff soil
and rock]; (b) Expected peak bed rock acceleration at the location. This is indeed the problem
with most developing countries which are prone to seismic hazard as sufficient number of
recorded motion is not available. India, for example has this problem. Several methods are
used in practice for the generation of these artificial ground motions. These approaches
include a) Direct use of strong motion recorded elsewhere having similar expected earthquake
magnitude at the site under consideration; b) Scaling of strong motion records to the expected
P.B.R.A (Peak Bed Rock Acceleration) expected at the location; c) Code-specified spectrum–
compatible ground motion. This paper presents a comparative study on the performance of
these three approaches for a typical site where response spectra and the expected peak bed
rock acceleration is compared with the code specified values. Acceleration time-series as
input motions have been generated by each of the three approaches and equivalent linear
analysis has been carried out by performing site response analysis. This analysis specifically
evaluates the effects of the local soil deposit to the expected ground motion. There are three
main empirical methods used to estimate local site effects.
This analysis produced ground motion design parameters at the surface of layered soil. These
results are evaluated and compared. The results indicate that the use of spectrum–compatible
ground motion provides least variations in the response parameters. It will be concluded that
this third approach provides uniformity amongst the designers and would avoid unnecessary
litigations.
Identification of Geo -
Hazards
(at the site)
Results
(Foundation displacements)
(Stability)
(SSI)
Figure 1: Flowchart showing Seismic Design Methodology for Foundations in seismic areas.
Selection of appropriate time histories for non linear seismic design of foundations 3
The damages caused by earthquakes in India have been instrumental in increasing earthquake
risk awareness and preparedness especially in major cities that lie in the high seismic hazard
zones. In this section seismic site response analyses will be performed for a typical soil profile
from Kolkatta using different methods for selecting the earthquake ground input motion.
The intent of the Indian Seismic Design Code (IS-1893(2002)), as with most seismic codes, is
to protect life under a rare seismic event, by preventing collapse of the structure or its parts
and maintaining structural integrity and residual load capacity. The rare seismic event is
equivalent to the ‘life safety’ objective for an earthquake event having a 10% chance of being
exceeded in 50 years (i.e. a return period of 475 years). This performance objective means the
structural and particularly the non-structural elements within a structure might be significantly
damaged. Though the level of damage may in theory be reparable, it might not be
economically viable to do so.
MCE (Maximum Credible Earthquake) is the most severe earthquake effects considered by
this standard. DBE (Design Basic Earthquake) is the earthquake which can reasonably be
expected to occur at least once during the design life of the structure. The design approach
adopted in this standard is to ensure that structures resist moderate earthquake (DBE) without
significant structural damage though some non-structural damage may occur, and aims that
structures withstand a major earthquake (MCE) without collapse. However, the code does not
give a definition on the risk level (the return period) associated with the MCE. The MCE, as
defined in the current (2009) edition of the International Building Code (IBC), as
corresponding to return period of 2,500 years. The design basic earthquake is taken as two
thirds of the MCE value.
Kolkata City lies over the Bengal Basin at the boundary of the seismic zones III and IV of the
seismic zonation map of India (IS 1893:2002). In the past the city has seen far and near source
earthquakes due to associated Himalayan tectonics. The most damaging earthquake so far has
been the 1897 earthquake which caused widespread damage in the city of Kolkata. Several
faults have been identified in this region out of which many show evidence of movement.
There are earthquake fault lines barely 100 km from Kolkata (Murty, 2006). However in the
last couple of years the city has seen rapid growth of population as well as construction of
major infrastructure projects within the ever expanding city boundaries. Seismic
microzonation of fast expanding mega city of Kolkata falling in seismic zone IV and V is
essential for retrofitting old and vulnerable buildings and for earthquake resistant design of
new constructions and installations.
In trying to understand the seismic hazard of the city it is worth remembering that the
earthquake database in India is still incomplete, especially with regards to earthquakes prior to
the historical period (before 1800 A.D.). According to Global Seismic Hazard Assessment
Programme (GSHAP) data, the state of West Bengal falls in a region of low seismic hazard in
the south-west that rises steadily towards the east and the north of the state. Historically, parts
of this state have experienced seismic activity in the M5.0-6.0 range. A probabilistic seismic
hazard assessment (PSHA) was carried out by Bhatia et al. (1999) under the GSHAP for the
Indian region. The ground-motion predictive equation of Joyner and Boore (1981) were used
as no reliable estimates of attenuation values were available for the region at the time.
4 SRGS, 17th December 2012, Delhi
The peak ground acceleration can be estimated by using Ground Motion Predictive Equations
(GMPE) which has been derived for particular type of earthquake mechanism and tectonic
regime. However no such reliable attenuation equation exists for Kolkata. According to
seismic hazard map of India, the city of Kolkata, lies in moderate seismic zone (Zone III) with
a zone factor 0.16. In this area, no seismic recording stations were established and
consequently no records of strong ground motion are available.
Mohanty et al. (2007) carried out a quasi-probabilistic seismic hazard study for Kolkata City.
The authors defined five seismic zones based on distribution of seismicity and the tectonic
setting of the region. A contour map with different PGA levels was published by the study as
shown in Figure 2 for a return period of 475 Year. The results of Mohanty et al. (2007)
indicate that the expected PGA values across the city vary from 0.10 to 0.34g, which are
larger than those suggested by Bhatia et al. (1999). As seen in this figure, a significant part of
the city can experience PGA levels between 0.25 and 0.34g. The authors indicate that the
expected PGA levels can increase as a result of the alluvium deposit, which could amplify the
seismic energy. The design ground motion parameter for this study area has been obtained
from (GSHAP) map which is based on 10% probability of exceedance in 50 years. Hence the
corresponding maximum peak ground acceleration (PGA) of 1.6 m/s2 (0.163 g) has been
selected for Kolkata.
Figure 2: Maximum PGA distribution over Kolkata City Region (Mohanty et al., 2007)
The target bedrock spectra are obtained from IS 1893 (part I): 2002 based on the desired PGA
(0.163g) at 5% damping for rock or hard soil. This spectrum is based on the assumption that
the bedrock velocity is atleast 760m/s. In absence of any detailed Unified Hazard Response
Selection of appropriate time histories for non linear seismic design of foundations 5
In general ground motion records should be selected to represent the potential earthquake
motions at the site. This is usually done by fitting the motions to a target response spectrum
and making sure that the records' parameters such as magnitude, distance, site conditions,
source mechanism, directivity and other effects correspond to the parameters of the
earthquake predicted by a seismic hazard analysis. These records can be either historical or
artificially generated. The appendix of ISO 19901-2 states that a minimum of four records
should be used for analysis, and that each record should be selected according to the following
criteria:
“Given the magnitude and distance of events dominating ELE ground motions, the earthquake
records for time history analysis can be selected from a catalogue of historical events. Each
earthquake record consists of three sets of tri-axial time histories representing two orthogonal
horizontal components and one vertical component of motion. In selecting earthquake records,
the tectonic setting (e.g. faulting style) and the site conditions (e.g. hardness of underlying
rock) of the historical records should be matched with those of the structure's site. The
information/ guidance that is not included in the code requirements are the following:
Duration of required input records when spectral matching is used.
Scaling requirements for separate components of the ground motion where spectral
matching is used.
Range of acceptable magnitudes and distances for seed selection.
An in house Arup program (Grant et al. 2008) selects the seed ground motion based on
spectral matching to minimize the artificial manipulations to the seed records from the PEER
database. Additionally this selection can be made from the PEER website.A set of criteria for
selecting seed motions for RSPMatch is presented in which initial spectral shape is considered
to be of primary importance, and other ground motion characteristics are used to filter the
initial database. Rather than using earthquake magnitude bounds as a proxy for ground motion
duration, as recommended by other authors, filtering explicitly on duration of the earthquake
motion is suggested. This program has been used to select the strong ground motion based on
specifying certain filters which are appropriate for Kolkata. In addition to this another
earthquake was also selected having magnitude (M) in the range of 6.0 to 6.9 recorded at rock
sites (site class ‘A’ as per USGS classification for which shear wave velocity, Vs > 750 m/s)
based on PGA matching without using the program. Table 1 illustrates these two selected
earthquake records.
Closet to fault
26.8 84.1
rupture (km)
Scaled and matched
PGA, magnitude, to spectra shape
Method of selection
recording station using Arup
Programme
There are three basic options available to the engineer for the selection of acceleration time
history records for use in non-linear dynamic analysis. These are described in the following
sections.
Artificial acceleration time histories may be generated by using programs such as SIMQKE
(Gasparini and Vanmarcke, 1979). The program can be used to generate a power spectral
density function from the smoothed response spectrum. This can then be used to derive
sinusoidal motions with random phase angles and amplitudes. An iterative procedure can be
set-up to improve the match with the target spectrum by calculating the ratio between the
target and the actual response ordinates at selected periods, and to adjust the power spectral
density function accordingly.
Although this approach is very attractive, due to the results being completely compatible with
the elastic design response spectrum, it is now generally accepted that the use of this method,
particularly for non-linear analysis, is problematic. This is because it is usually found that
artificial records generally tend to have an excessive number of cycles which consequently
contain unrealistically high energy content. Additionally, it has been reported in literature that
for UHRS, which are strongly influenced by more than one source of seismicity, spectrum
compatible artificial records will tend to be particularly unreasonable. Therefore, it is advised
that artificially generated time history records are generally avoided.
The synthetic Accelerograms are generated from seismological source models and account for
path and seismic effects. It is usually necessary to consult a seismologist in order to define
many parameters required to characterise the earthquake source in order to generate synthetic
Accelerograms. Additionally, the determination of these source parameters carries a high
degree of uncertainty and it is therefore advised that in absence of expert judgement, this
method is usually avoided.
The last remaining option available to engineers is the use of real Accelerograms, which by
definition, are free of all the problems mentioned in the previous sections. Real
Accelerograms from previous earthquakes have become widely available in recent years.
However, care is required in selection and scaling of these earthquake strong ground motion
records for a particular site..
These selected ground motions can be used in different ways for further analyses. These
approaches include a) Direct use of strong motion recorded elsewhere having similar expected
earthquake magnitude at the site under consideration; b) Scaling of strong motion records to
the expected P.B.R.A (Peak Bed Rock Acceleration) expected at the location; c) Code-
specified spectrum–compatible ground motion. All these methods have their advantages and
disadvantages as discussed below
Traditionally seismic hazard at a site for design purposes has been represented by design
spectra, thus all seismic design codes and guidelines require scaling of selected ground motion
time histories so that they match or exceed the controlling design spectrum within a period
range of interest. Several methods of scaling time histories have been proposed. These include
frequency-domain methods where the frequency content of the recorded ground motions is
manipulated in order to obtain a match. Regardless of the method (frequency-domain or time-
domain), in virtually all of the existing approaches, the processes of selecting earthquake
ground motions and their scaling to match the design spectrum are separate and distinct. In
other words, first one or more time histories are selected and then appropriate scaling
mechanisms for spectrum matching are applied. The scaling methods depend on the end use
of the motions and should be done with caution so as not to distort the records to the point that
they are no longer realistic.
The scaling of the time histories to match the target design hazard spectra (hereafter referred
to as the target spectrum) was carried out using the software RSPMatch2005. This program
performs a time domain modification of an acceleration time history to make it compatible
with a user specified target spectrum. The methodology is based on that proposed by
Lilhanand and Tseng. The modification of the time history can be performed with a variety of
different modification models (wavelets). The wavelet adjustment preserves the long period
non-stationary phasing of the original time history. Depending on the specific nature of the
input time history, in terms of initial frequency content and duration, the ease with which the
program is able to match the record to the target spectrum varies. This variability and effort
required is reduced through careful seed selection procedure and initial scaling of the records,
particularly over the long period as discussed before. Figure 3 shows the target spectrum and
the matched spectra for one of the selected records.
8 SRGS, 17th December 2012, Delhi
6
Target Spectra (IS-1893:2002,
Rock or Hard soil
5 Chi Chi Taiwan Original Spectra
Spectral Acceleration (m/s/s)
Scaled
RSP matched Spectra
4
0
0.01 0.1 1 10
Period (s)
It is noted that the use of spectral matching software to a certain extent reduces the reliance on
fault type, location of earthquake, magnitude and distance in so far as the spectral matching
procedure will match the seed record to the target spectrum which incorporates all of these
aspects within it. Thus the spectral matching procedure should start with the above criteria as
a guide in order to provide the closest matching time histories to the target spectrum. This is
so that artificial manipulation of the record is kept to a minimum, providing a final record that
is as close to a real earthquake in terms of frequency content as possible. In the next section
these selected records will be used for site response analysis.
During an earthquake, stress waves which are generated at some depth are modified as they
travel through the various soil layers. The soil acts as a filter, amplifying energy at some
frequencies and attenuating energy at others. The effect of this is often calculated by
performing a site response analysis. In the present study the parameters of interest in the
seismic response analysis involves the fundamental frequency of the soil layers, amplification
factor of ground motion parameters and response spectra.. The site response analyses
undertaken for this study project was carried out using the Arup in-house program Oasys
SIREN. The program operates in the time domain enabling it to model non-linear soil
properties with hysteretic damping. The inputs to the program are soil unit weight, small
strain shear modulus, modulus degradation curves, bedrock properties and input time
histories.
In the present study, geotechnical bore hole data from more than 100 locations were obtained
from various construction organizations and research institutions in and around Kolkata
(Govindraju and Bhattarcharya 2008). Important locations from where the geotechnical data
was collected were categorized as Type 1, Type 2, Type 3 and Type 4. Type 2 is most critical
Selection of appropriate time histories for non linear seismic design of foundations 9
from understanding the seismic soil behavior for new constructions as it covers the reclaimed
areas. Figures 4 illustrates typical soil profiles and the corresponding average Standard
Penetration (SPT) ‘N’ values for Type-2 soil profile.
SPT- N SPT - N
0 20 40 60 80 100 0 10 20 30 40 50 60
0 0
0.9 m
10
Silty clay with grass roots (CH)
20
15
9.8 m
Loose brownish gay sandy silt mixed with mica (ML)
Depth (m)
Depth (m) 30
Medium dense light gray fine sand mixed with mica and 20
Medium dense / dense light gray sand mixed with mica and
varying percentages
50 of fines (SP-SM) 30
35
60
40
36.0 m
SPT - N
SPT - N
10 20 30 40 50 60
Figure 4. Generalized soil profile
0 and SPT ‘N’ variation for category
0 10 20 30 40 50 60 70 80 0
Type 2 (Govindraju et
al. 2008) Type 3
5
Type 4
The input parameter for the 20model such as shear wave velocity
15 of each soil layer has been
Depth (m)
Depth (m)
obtained from the empirical relation proposed by Japanese Road Association (Lee, 1992). The
relationship between shear modulus and strain amplitude is20 typically characterised by a
normalised modulus reduction curve.
30
25
It has been well established in40 the literature that shear modulus30values reduce with increasing
shear strains and, that the rate of degradation is dependant on the material type, depth, OCR
(clays) and PI (clays). Therefore, shear modulus degradation curves are required to model the
soil at increased strains due to earthquake motions, which reduce the soil’s shear modulus and
hence the shear-wave velocities. There are a number of shear modulus degradation curves
which have been published for different soils in the literature. For the present site, the clayey
silt has been modelled by a shear modulus degradation curve defined for silt, by Pappin et al.
(1989) and the silty clay has been modelled by the degradation curve defined for clay with PI
of 20%, by Vucetic & Dorby (1991). These curves are shown in Figure 5.
10 SRGS, 17th December 2012, Delhi
1.1
1
G/Go Degradation Curves
0.9
0.8
0.7
G/G0
0.6
0.5
0.4
0.3
0
1.00E-04 1.00E-03 1.00E-02 1.00E-01 1.00E+00 1.00E+01
Shear Strain %
Figure 5: Shear modulus degradation curves used in the analysis of site response
Four types of time histories were used for the analyses. Figure 6 shows the surface response
spectrum obtained by using an unscaled record and a spectrum compatible time history. The
natural frequency of the soil layer is in the range of 0.3-0.4s and peak spectral accelerations
are also observed at this period. Spectral accelerations for the unscaled motions are much
higher than the spectral accelerations seen for the spectrum compatible motions. The
frequency content of the surface response is also different.
Surface Spectrum
8 Northridge_Unscaled_RSPmatched
Spectral Acceleration (m/s/s)
Northridge_Unscaled
7
0
0.01 0.1 1 10
Period (s)
6
Chi Chi (Spectrum Compatible)
5 Northridge_Unscaled_RSPmatched
Spectral accleration (m/s/s)
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Period (s)
Figure 7: Comparison of surface response spectrum for different types of earthquake input
motion
Figure 7 compares the surface spectral accelerations obtained by using two earthquake
motions which are spectrum compatible. The Chi Chi earthquake motion has been scaled
initially and then matched with the target bedrock spectrum. The Northridge earthquake
motion has not been scaled but has been spectrally matched with the target spectrum. Several
comments can be made from this figure
• Scaling does not make a significant difference to the final result if the input motion is
spectrally matched to the target spectrum.
• The seed records which may have different frequency content in the recording tend to
give the same surface spectral acceleration if they have been matched to the same
target spectrum. In this case it may not be necessary to use three input motion as
specified in the codes.
• If the seismic design was done according to IS-1893, the soil would be classified as
Medium soil (From N values) and surface spectrum shown in Figure 7 for medium soil
would be used. However as the results indicate that the spectral acceleration at low and
intermediate periods will be underestimated by approximately 1.5 times. This is also a
consequence of not including the S (soil factor) in the design spectrum. This factor is
used in most codes (ex Eurocode) to account for site response effects.
The results indicate that the use of spectrum–compatible ground motion provides least
variations in the response parameters. However it should also be kept in mind that these
spectrum compatible motions can induce additional displacements and the records need to be
baseline corrected before they are used for further analysis. The best results – in terms of
successful and timely convergence to a solution, and minimal adjustment of the seed record –
are obtained when the initial seed has a response spectrum which provides a good initial fit of
the target spectrum.
12 SRGS, 17th December 2012, Delhi
It will be concluded that this third approach provides uniformity amongst the designers and
would avoid unnecessary litigations. Additional comments that can be made from this study
are
• Ground motion amplification with respect to a reference site condition can be
estimated using empirical amplification factors based on a site classification system
which is used by many building codes. (ex EN1998-1). Some codes (ex. UBC 1993)
uses site categories classified based on the average S-wave velocity over the top 30 m
(Vs30).
• If ground motions of the required intensity are available for the site, these should be
used directly since they incorporate local site effects. This option is available for those
areas of the world where earthquake recorded motions are readily available (Example
Japan, Some parts of USA etc).
• Peak ground motion parameters or spectral values may be estimated using attenuation
relations that include a term dependent on site conditions. This refers specifically to
the Next Generation of Attenuation Relations (NGA) which has developed improved
ground motion predictive equations for shallow crustal earthquake in Western United
States and other similar tectonic areas.
ACKNOWLEDGEMENT
The first author would like to thank several of her ex Arup colleagues who made significant
contributions to the work presented here: Matthew Free, Ziggy Lubkowski, Damian Grant, Yasir
Khoker.
REFERENCES
ABSTRACT
The occurrence of liquefaction has been long recognised as a major hazard during
earthquakes, and recent earthquakes including that in Gujarat in 2001 continue to exhibit
liquefaction behaviour. This paper explains the phenomenon of liquefaction and the possible
effects. The identification of liquefiable soils based on site investigations is introduced. A
brief review on the liquefaction of fine grained soils is presented based on different criteria.
The effect of development of water films in layered soil profiles and the consequences due to
liquefaction is discussed. Based on this a case study of damages of the Kandla Port and
Customs Office tower supported in layered ground that liquefied during the 2001 Bhuj
earthquake is presented.
1. INTRODUCTION
Soil liquefaction is a complex phenomenon that can exhibit itself in several different ways in the field.
During many earthquakes liquefaction has been deemed responsible for significant damage to
buildings and other infrastructures. The occurrence of liquefaction has been studied extensively over
the past three decades and substantial advances have been made in understanding the development and
effects of this phenomenon. These advances have led to a series of practical procedures for evaluating
the potential for occurrence and for estimating its effects. These developed procedures, however, are
almost entirely empirical in nature and, as such, are difficult to apply to the problem of performance
prediction for individual structures.
Liquefaction was viewed in two different ways. One way correspond to the condition at which the
effective stress reaches a value equal to zero, while the other correspond 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. In the field, significant lateral deformations can be caused by
either of these phenomena. The deformations produced by flow liquefaction are usually referred to as
flow slides, and those produced by cyclic mobility as lateral spreads. But it is frequently impossible to
distinguish between the two in the field.
2. EFFECTS OF LIQUEFACTION
The phenomena of liquefaction can alter ground motions in terms of amplitude, frequency content, and
duration. Liquefaction can lead to ground failure, either through flow liquefaction or lateral spreading.
Further, liquefaction can affect the performance of buildings and bridges in a number of different
ways.
1
2 SRGS, 17th December 2012, Delhi
Local site conditions such as soft sediments of considerable depth strongly affect the
amplitude of ground motions (Tsuda et al., 2006). The thickness, stiffness, and damping
characteristics of the various soil layers that underlie a particular site control the relative
amplification or de-amplification of various components of a bedrock motion. Stiff soil
deposits tend to amplify the higher-frequency components of a bedrock motion, while soft soil
deposits amplify low-frequency motions.
In general, the development of excess pore water pressure and consequent reduction in
effective stress will lead to the softening (i.e. reduction in stiffness) of a liquefiable soil
deposit. Therefore, a soil layer that may amplify relatively high-frequency components of the
early portion of a bedrock motion will tend to amplify successively lower frequency
components as the motion proceeds. Ground motions recorded at the surface of liquefiable
soil deposits showed a pronounced reduction in high-frequency amplitude and increase in
low-frequency amplitude following initial liquefaction (Fig.1). This change in frequency
content corresponds to the dramatic reduction in stiffness and strength that accompanies initial
liquefaction. Here the dramatic change in frequency content after initiation of liquefaction at 6
to 7 seconds can be noticed. This change in frequency content corresponds to the reduction in
stiffness and strength due to initial liquefaction.
Figure 1. Time history of ground surface acceleration from Niigata, Japan (Kramer, 2001)
Soil stiffness is usually considered to be a function of the square root of effective stress (see,
e.g., Hardin and Drnevich, 1972). This has the associated effect of reducing the natural
frequency of the soil layer. It is therefore important that the site response analysis for
liquefiable conditions allows for the generation of pore water pressures.
natural = 0.59. Guessing a viscous damping coefficient of about 15%, this corresponds to an
amplitude amplification of about 1.5 times according to Figure 2. For the liquefying soil,
normalised frequency is 2 Hz / 1.9 Hz = 1.1. The same curve in Figure 2 now suggests an
amplification of about 3 times, illustrating how the site response has been somewhat altered
by the pore pressure generation.
Ignored from the above analysis are the effects of shear strain. In the liquefying soil then
somewhat larger shear strains would be expected, which in themselves will reduce shear
modulus yet further, and also cause an increase in the degree of damping (selected somewhat
arbitrarily above).
3 e2 1 2K 0
(1)
Gmax 3230 v
1 e 3
G
Vs (2)
Vs
f0 (3)
4h
As excess pore pressure increases, so the lateral pressure from the soil increases. Retaining
structures such as walls and quays come under increased load for a long duration (hours or
even days) that may be more damaging than the transient lateral accelerations of the
earthquake itself. For example, back analysis and model testing of failed quay walls in Kobe
showed that this had indeed happened (Inagaki et al., 1996).
4 SRGS, 17th December 2012, Delhi
Excess pore pressures are limited to the value of the initial vertical effective stress.
Liquefaction conditions therefore represent an excess pore pressure profile as shown in Figure
3, that is, with an upwards hydraulic gradient of γ’/γ, the critical hydraulic gradient. Water
rushes upwards rapidly and piping can occur. More common is the phenomenon of sand
boiling. This fast moving fluid carries sand particles upwards (possible, as the hydraulic
gradient is critical) and out across the ground surface. If the surface is a different material to
the liquefiable sand, clear volcano-like structures are seen as shown schematically and
photographically in Figure 6. Sand boils are commonly used as indicators of damage,
e.g. Juang et al. (2005).
Figure 3. Schematic illustration of sand boiling. Photo: National Information Service for
Earthquake Engineering, University of California, Berkeley.
The final liquefaction-related issue described in this paper is that of lateral spreading. Shallow
slopes (3° - 5°) such as riverbanks can move downslope as shown in Figure 4. As a
consequence, the objects such as piles in the ground can be subjected to lateral forces
(Abdoun et al., 2003, Dobry et al., 2003, Haigh and Madabhushi, 2002a). There is still some
debate over the magnitude and influence of such forces. Existing code provision is given by
the Japanese, although Haigh and Madabhushi suggests that transient forces due to soil
dilation could be somewhat larger. Lateral spreading is diagnosed by the characteristic lateral
cracks shown in Figure 4, which occur when more brittle materials are present at the surface.
Several approaches are available for estimation of permanent deformations associated with
lateral spreading. The lateral spreading phenomenon is a complex one, and it has proven to be
extremely difficult to make accurate prior predictions of permanent deformations using
analytical/numerical procedures alone. As a result, currently available procedures are
empirically based.
Liquefaction of Soils – Effects and Assessment 5
Figure 4. Schematic illustration of lateral spreading. Photo: lateral spreading at Navlaki port,
January 2001, S.P.G. Madabhushi.
Liquefaction can cause the failure of foundation systems by a variety of mechanisms. Both shallow
and deep foundations can be damaged by soil liquefaction. The most visible of these shallow
foundation failure mechanisms is through the loss of bearing capacity associated with loose saturated
soils with low residual strength. By this mechanism, the earthquake shaking can trigger flow
liquefaction and dramatic bearing failures of the type shown in Figures 5.
Liquefaction can also have a significant impact on pile foundations. Liquefaction and lateral spreading
in Niigata caused failure of pile foundations of the Showa bridge (Figure 6). Liquefaction-induced
failure of deep foundations has been observed in many other earthquakes also.
Figure 5. Example failures associated with liquefaction strength loss. a) Taiwan, 1999, photo Uzarski
and Arnold (2001); b) Turkey, 1999, photo Youd et al. (2000); c) Niigata, 1964, photo National
6 SRGS, 17th December 2012, Delhi
Figure 6. Failure of Showa River Bridge in Niigata, Japan, due to lateral spreading.
A number of investigations have been reported correlating the liquefaction potential of a site
with in situ tests. The most used methods for determining liquefiability are the in-situ
Standard Penetration Test (SPT) (Seed and Idriss, 1971) and Cone Penetration Test (CPT)
(Robertson and Campanella, 1985). Of these, the SPT presents a dynamic load to the soil and
the CPT a continuous push, so therefore the SPT should be more likely to pick out a
liquefiable soil.
The original work for SPTs was by Seed and Idriss (1971), and perhaps the most up-to-date
version was presented by Idriss and Boulanger (2004). The method works by estimating the
expected resistance of the soil, the “cyclic resistance ratio” or CRR, and the expected load on
the soil, the “cyclic stress ratio” or CSR. The ratio between these is the factor of safety against
liquefaction. CSR is determined by Equation 4, amax is the maximum acceleration expected
during shaking and g is the acceleration due to gravity. rd is a depth reduction factor to
account for soil not being rigid, and may be given as 1 – 0.015z where z is depth in metres.
MSF is the magnitude scaling factor to account for earthquakes of magnitude other than the
7.5 for which the method was originally derived, and given by Equation 5 for earthquake
magnitude M.
eq v 0 amax rd
CSR 0.65 (4)
v 0 v 0 g MSF
M
MSF 6.9 exp 0.058 (5)
4
CRR is determined graphically. The chart in Figure 7 presents curves for a given corrected
SPT blow count, gives the cyclic resistance ratio for clean sands (less than 5% fines content).
Marked on the figure are some case histories where liquefaction has (filled markers) or has
not been (open markers) observed, from the extensive data set used to derive the curve. The
curve may be seen as a border between liquefiable (above the line) and non-liquefiable (below
the line) soils, and constantly evolves with the data set. In truth, this is a very rudimentary
method and highly unsatisfactory for a number of reasons, but it remains the current best
liquefaction assessment tool available to the engineer at site. Engineers should aim to support
the results from this “simplified procedure” by personal inspection of soil samples and use of
judgment.
Liquefaction of Soils – Effects and Assessment 7
Note that different curves are sometimes presented for different fines contents. Expectation at
time of production was that fines, being non-liquefiable, would reduce the likelihood of
liquefaction. More recent work (e.g. Thevanayagam and Mohan, 2000, Thevanayagam and
Martin, 2002) has suggested that this is not the case and that a small fines content can reduce
the stability of a soil and hence increase its liquefaction potential. Therefore such curves
should be used with caution when significant fines content is present. In particular, the so-
called Chinese criteria should be avoided (Polito, 2001).
CRR may be further multiplied by factors Kα and Kσ to account for initial static shear and
normal effective stresses respectively. These are most recently considered by Boulanger
(2003a, 2003b). The effect of such parameters is small compared to the accuracy of the
method as a whole, and they are therefore omitted for brevity. The SPT forms the basis for the
liquefaction hazard evaluation in several codes.
Figure 7. Chart for determining CRR for clean sands (Idriss and Boulanger, 2004).
The CPT-based liquefaction correlation was reevaluated by Idriss and Boulanger (2003) using
case history data compiled by Shibata and Teparaksa (1988), Kayen et al (1992), Boulanger et
al (1995, 1997), Starkand Olson (1995), Suzuki et al (1997), and Moss (2003).The work of
Moss (2003) was particularly valuable in providing the most comprehensive compilation of
field data and associated interpretations. This re-evaluation of the CPT-based procedures
incorporated adjustments and parameter revisions. The chart in figure 8 presents curves for a
given corrected CPT tip resistance.
8 SRGS, 17th December 2012, Delhi
The shear wave velocity can be measured in situ employing different geophysical techniques.
Andrus and Stokoe (2000) have proposed a method to evaluate liquefaction potential of a site
containing clean sands, silty sands and sandy silts based shear wave velocities as shown in
figure 9. Here, vs1 represents the corrected shear wave velocity.
Figure 8. CPT-based case histories and recommended relation for clean sands with relations
proposed by others (Idriss and Boulanger, 2004).
Figure 9. Chart for determining CRR for clean sand, silty sand and
sandy silt (Andrus and Stokoe, 2000).
Liquefaction of Soils – Effects and Assessment 9
The studies used to derive the initial framework of liquefaction behaviour have been largely
based around the results of laboratory element tests. These typically contain uniformly-graded
sands under uniaxial cyclic shear. In addition to grading variability, natural soil deposits may
be layered in profile, and real earthquakes are not uniaxial. This section aims to introduce
some of the work performed highlighting the specific effects of these more complex (though
perhaps more realistic) situations.
Recent case histories have shown that fine grained soils especially silts are also prone to liquefaction
in certain situations, see for example liquefaction of fine-grained soils in Adapazari, Turkey during the
1999 Kocaeli earthquake, Bray et al (2004). These fine-grained soils typically consisted more than
about 70% fines (i.e. finer than the 200 sieve) with plasticity index ranging between 0 and 25. Recent
investigations, Bray et al (2004) show that it is not the amount of “clay-size” particles in the
fine grained soil but it is the amount of “clay minerals” that best indicate a soil’s susceptibility
to liquefaction. Several criteria were established by various researchers in the recent past
based on the data obtained from the sites where liquefaction was observed and not observed after
earthquakes. Table 1 illustrates criteria that have been established to identify the potential of
liquefaction of such soils.
Table 4.1. Criteria for Liquefaction of Fine Grained Soils
Clay
Further
Content Grain size Ratio Soil Cyclic
Criteria WL (%) PI (%) Liquefaction Studies
(<0.002mm) (< 5mm) w/WL Group Mobility
Required
%
Chines
< 15% < 35% > 0.9 YES
Wang (1979)
CH, CL
NO
Youd (1998) SC, GC
< 35% < 7% YES
Polito (2001) 25% < WL < 35% 7% < PI < 10% YES
The stratified soils most of interest are those where a potentially liquefiable layer is beneath a
layer of low permeability such as clay or a pavement. Lateral spreading and sand boiling
introduced above become problematic issues if these layers are near the surface. In addition,
at any depth it is possible that pore fluid dissipating out of a liquefied soil may be stopped
from moving upwards by such layers, and a thin layer of water may be formed separating the
two layers. Figure 10 shows this developing by plotting effective stress against depth on the
left. The right hand side of Figure 13 shows the evolution of settlement of various layers with
time, beginning at the start of shaking and progressing until some significant time after
shaking stops. Three distinct phases of dissipation are shown in Figure 10; further explanation
10 SRGS, 17th December 2012, Delhi
of these is given by Brennan and Madabhushi (2005). This water film obviously has the
potential to form a sliding surface.
This section illustrate a case study of failure of a pile-supported building possibly due to the
interaction between axial-load induced settlement owing to liquefaction and bending due to
lateral spreading forces. The building is the Port and Customs Office Tower of Kandla Port,
which is supported on a piled raft and that tilted during the 2001 Bhuj earthquake. A thorough
geotechnical study of the site has been carried out and the foundation system is analysed
considering the soil–pile interaction, effect of foundation mat and nonlinear behaviour of the
soil. This main intention of this study is to investigate the plausible causes of the failure of the
building.
The Bhuj earthquake that struck the Kutch area in Gujarat at 8.46am (IST) on January 26,
2001, with a magnitude (Mw) of 7.7 was one of the major earthquakes in India that caused
extensive damage to the built environment. Along with other cities of Gujarat, Kandla also
experienced significant damage. Kandla, located at the mouth of little Rann of Kutch on the
south eastern coast, is about 50km from the epicenter of the 2001 Bhuj earthquake. Many pile-
supported buildings, warehouses and cargo berths in the Kandla area were damaged during
the earthquake.
A 22m high six-floor building called the Port and Customs Office Tower (which will be
referred as “Building” in rest of the paper) located in the Kandla port area very close to the
waterfront. This pile-supported building leaned about 30cm at its top and separated from its
adjacent building. Figure 11 shows the location map of the building in Kandla port precinct
along with the tower building. The building was founded on 32 short cast-in-place concrete
piles. Each pile was 18m long and 0.4m in diameter. The piles were passing through 10m of
clayey crust and then terminated in a sandy soil layer. The loads from superstructure are
transferred to the piles through the foundation mat, which eventually acts like a pile cap.
Major structural details of the building are shown in figure 12. A thorough geotechnical study
Liquefaction of Soils – Effects and Assessment 11
of the site has been carried out. The foundation system was analyzed considering the soil-pile
interaction, effect of foundation mat and the nonlinear behaviour of the soil.
The Port of Kandla is built on natural ground comprising of recent unconsolidated deposits of
inter bedded clays, silts and sands. The water table is about 1.2 - 3.0m below the ground level.
Figure 13 shows the borehole profile (taken from EERI, 2002) of the natural ground near to
the building. Post earthquake observations at the site indicated that the ground in the vicinity
of the tower settled by about 30cm (one foot), resulting in the settlement of the floating mat
floor of the building. There were evidences of extensive liquefaction with ejection of sand
through ground cracks in the vicinity of the building. Post earthquake observations also reveal
lateral spreading in many places in the Kandla port area.
Figure 11. Location map of Port and Customs tower in Kandla Port
5.3 Liquefaction potential of sandy soil and cyclic failure potential of clayey soil
For the soil at building site, the potential for liquefaction of sandy soils has been evaluated
based on the method recommended by Idriss and Boulanger (2004). Further, cyclic failure in
clays has been evaluated based on the new procedure proposed by Boulanger and Idriss
(2005). The results of these analyses give factor of safety (FOS) against liquefaction potential
and cyclic failures with depth as presented in Figure 13. It is evident from the figure 13 that
most part of the clay layer except the top 2m undergoes cyclic failure that may result in
ground deformation and cracking. Furthermore, the entire sandy stratum between 10m to 22m
is likely to have experienced liquefaction which may result in ground settlement and flow
12 SRGS, 17th December 2012, Delhi
failure. The above analysis is confined to free field conditions without considering the vertical
stresses from the building loads. If vertical stresses due to the building are considered, then
even the top 2m depth of clay will experience cyclic failure resulting in substantial
deformation and cracking. The above analysis substantiates the observations that liquefied
fine sand ejected to the ground surface through the ground cracks in the top clay layer.
Figure 12: Major details of the Kandla Port and Customs Office tower
Figure 13: Typical soil profile in the vicinity of tower of the Port and Customs office
(near transit shed and parking place of Berth III; PL, plastic limit; LL, liquid limit; N, SPT
blow count; and , total unit weight of soil)
The amount of seismic settlement of the soil deposit after liquefaction at the building site is
calculated using two different methods; a) Method-1as suggested by Tokimatsu and Seed
(1987), and b) Method-2 as suggested by Ishihara and Yoshimine (1992). The analyses predict
Liquefaction of Soils – Effects and Assessment 13
Figure 14. Plausible settlement mechanism of failure showing the tilting the tower
6. CONCLUDING REMARKS
Liquefaction of collapsible soils such as saturated loose sands/silty sands is a major issue
when constructing in seismically-active regions. Determination of whether any given soil is
liquefiable or not, and the potential consequences of a liquefaction event in any given soil
layer, are very much a matter for engineering judgement rather than clear-cut simple
problems. This is particularly pertinent as real sites commonly deviate from the idealisation of
the laboratory; issues around the effects of soils with fines content, with stratifications etc.
There are several gaps in the existing literature and no definite guidelines are available to
ascertain liquefaction susceptibility of fine grained soils. Hence, further studies are required to
understand the liquefaction behavior of fine grained soils.
REFERENCES
Abdoun, T., Dobry, R., O’Rourke, T.D. and Goh, S.H. (2003). Pile response to lateral
spreads: centrifuge modeling. Journal of. Geotechnical. and Geoenvironmental
Engineering., ASCE. 129(10): 869-878.
Adalier K. and Elgamel, A.W. (2001). Seismic response of dense and loose sand columns.
Proc. Fourth Int. Conf. on Recent Advances in Geotechnical Earthquake Engineering and
Soil Dynamics, San Diego, CA, USA. Paper no. 4.19.
Boulanger, R.W. (2003a). Relating Kα to relative state parameter index. Journal of
Geotechnical and Geoenvironmental Engineering, ASCE. 129(8): 770-773.
14 SRGS, 17th December 2012, Delhi
Boulanger, R.W. (2003b). High overburden stress effects in liquefaction analyses. Journal of
Geotechnical and Geoenvironmental Engineering, ASCE. 129(12): 1071-1082.
Bray, J.D., Rodolfo, B.S., Riemer, M. and Durgunoglu, H.T. (2004). Liquefaction
susceptibility of fine-grained soils, Proc. 11th Int. Conf. on Soil Dynamics and
Earthquake Engineering/3rd Int. Conf. on Earthquake Geotechnical Engineering,
Berkeley, CA, USA (D. Doolin, A. Kammerer, T. Nogami, R.B. Seed and I. Towhata,
eds.), 655-675.
Bray, J.D. and Travasarou, T. (2007). Simplified procedure for estimating earthquake induced
deviatoric slope displacements. Journal of Geotechnical and Geoenvironmental Engineering,
381-392.
Brennan, A.J. and Madabhushi, S.P.G. (2005). Liquefaction and drainage in stratified soil
Journal of Geotechnical and Geoenvironmental Engineering, ASCE. 131(7): 876-885.
Castro, G. (1975). Liquefaction and cyclic mobility of saturated sands. Journal of
Geotechnical Engineering, ASCE. 101(6): 551-569.
Dobry, R., Abdoun, T., O’Rourke, T.D. and Goh, S.H. (2003). “Single piles in lateral spreads: field
bending moment evaluation”. Journal of Geotechnical and Geoenvironmental Engineering.,
ASCE. 129(10): 879-889.
Haigh, S.K. and Madabhushi, S.P.G. (2002a). Centrifuge modelling of lateral spreading past
pile foundations. Proc. Int. Conf. on Physical Modelling in Geotechnics, St Johns, NF,
Canada (R. Phillips, P.J. Guo and R. Popescu, eds.). pub. Balkema, Rotterdam. 471-475.
Hardin, B.O. and Drnevich, V.P. (1972). Shear modulus and damping in soils: design
equations and curves. Journal of Soil Mechanics and Foundations Div., ASCE. 98(7):
667-692.
Idriss, I.M. and Boulanger, R.W. (2004). Semi-empirical procedures for evaluating
liquefaction potential during earthquakes. Proc. 11th Int. Conf. on Soil Dynamics and
Earthquake Engineering/3rd Int. Conf. on Earthquake Geotechnical Engineering,
Berkeley, CA, USA (D. Doolin, A. Kammerer, T. Nogami, R.B. Seed and I. Towhata,
eds.).1: 32-67.
Inagaki, K., Iai, S., Sugano, T., Yamakaki, H. and Inatomi, T. (1996). Performance of caisson
type quay walls at Kobe port”. Soils and Foundations special edition January 1996, on
geotechnical aspects of the January 17th 1995 Hyogoken-Nambu earthquake, JSSMFE.
119-136.
Ishihara, K and Yoshimine, M. (1992). Evaluation of settlement in sand deposits following
liquefaction during earthquakes. Soils and Foundations, 32(1), 173-188.
Juang, C.H., Yuan, H., Li, D.K., Yang, S.H. and Christopher, R.A. (2005). Estimating
severity of liquefaction-induced damage near foundation. Soil Dynamics and Earthquake
Engineering, Elsevier Science. 25: 403-411.
Kramer, S.L. (1996): Geotechnical Earthquake Engineering, Pearson.
Kulasingam, R., Malvick, E.J., Boulanger, R.W. and Kutter, B.L. (2004).Strength loss and
localisation at silt interfaces in slopes of liquefied sands. Journal of Geotechnical and
Geoenvironmental Engineering, ASCE 130(11): 1192-1202.
Polito, C.P. (2001). Plasticity based liquefaction criteria. Proc. Fourth Int. Conf. on Recent
Advances in Geotechnical Earthquake Engineering and Soil Dynamics, San Diego, CA,
USA. Paper no. 1.33.
Robertson, P.K. and Campanella, R.G. (1985). Liquefaction potential of sands using the CPT. Journal
of Geotechnical Engineering, ASCE. 111(3): 384-403.
Tokimatsu, K. and Seed, H.B. (1987). Evaluation of settlements in sand due to earthquake
shaking. Journal of Geotechnical Engineering, ASCE, 23(4): 56-74
Tsuda, Kenichi, Jamison Steidl, Ralph Archuleta, and Dominic Assimaki (2006). Site-
response estimation for the 2003 Miyagi-Oki earthquake sequence considering nonlinear
site response. Bulletin of the Seismological Society of America, 96(4A):1474–1482, doi:
10.1785/0120050160.
UKIERI International Workshop on Seismic Requalification of Geotechnical Structures (SRGS)
17th December 2012, Delhi.
ABSTRACT:
Earthquake is disaster natural hazard where damage not only depends on physical status of
the structures but also by the geotechnical materials around the structures. Even though
science has attained considerable advancement, it is almost impossible to predict earthquake
precisely until now, but is possible to precisely predict possible hazards in a given location.
Only way to minimise disaster due to earthquake is to be prepared for earthquake for any
earthquake hazard in the region. Most common seismic hazards are ground motion, ground
failure/displacement, site effects (amplification), induced effects ( liquefaction, landslide and
Tsunami), fire and flood. Most of these hazards were reported due to the inherent problem
in the geotechnical aspect around the structures. It is very easy to minimise/avoid these
hazards in the newly constructed structures, whereas it is a quite challenging task in the
existing structures. Proper re-qualification of geotechnical structures can help to find the best
suitable technology to retrofit the existing structure, so that disaster due to possible future
earthquake hazards can be minimised. This paper presents an overview of the Geotechnical
and Geophysical Tests for re- qualification of geotechnical structures with case studies using
these tests. Case study of rock depth mapping, soil profiling, identification of filled soil,
locating weak zones in historic foundation layers, dames and dikes are presented.
Keywords: Earthquake Hazards, Geotechnical and Geophysical Testing, GPR Testing Soi
layers and weak zones.
1. INSTRUCTIONS
Large earthquakes have caused massive loss of lives and extensive physical destruction
throughout the world (Armenia, 1988; Iran, 1990; US, 1994; Japan, 1995; Turkey, 1999;
Taiwan, 1999, India 2001, Sumatra 2004, Pakistan, 2005, China 2008, Haiti 2010, Japan
2011). India has been facing threat from earthquakes since ancient times. In India, the recent
destructive earthquakes are Killari (1993), Jabalpur (1997), Bhuj (2001), Sumatra (2004),
Indo-Pakistan (2005) and Sikkim (2011). Very preliminary process of reducing the effects of
earthquake hazard is to assess the factors responsible to cause damages. The widely used site
characterization methods for earthquake geotechnical studies are geological methods,
geotechnical methods and geophysical methods. The popular geotechnical methods are
standard penetration test (SPT), dilatometer meter test (DMT), pressure meter test and
seismic cone penetration test (SCPT), among these SPT is widely used. SPTs are more
suitable as it is correlated with dynamic properties and liquefaction resistance of the soil.
Many geophysical methods are attempted for seismic site characterization, but widely used
methods are Spectral Analysis of Surface Waves (SASW) and Multichannel Analysis of
Surface Waves (MASW). SASW and MASW are surface wave methods widely used for
many civil and earth science applications. These methods are widely used to estimate shear
1
2 SRGS, 17th December 2012, Delhi
wave velocity of the subsurface layer and thereby site response parameters such as
amplification and predominant frequency of soil can be estimated. Even though SPT and
surface wave methods can provide dynamic properties of soil, they are time consuming and
expensive. These tests have limited capability to measure the spatial variation of these
properties in the site. Ground Penetrating Radar (GPR) can be used to map spatial variability
and to study hidden problem below ground. This paper presents brief summary of SPT,
MASW and GPR and application of these tests for the seismic re-qualification of
geotechnical structures.
The Standard Penetration Test (SPT) is one of the oldest, most popular, and commonly used
in situ test for exploration in soil mechanics and foundation engineering because the
equipment and test procedure are simple. Standard Penetration Test is a widely used in-situ
test in a borehole to evaluate geotechnical properties of soil. A ‘Standard Penetration Test’
commonly known as the ‘SPT’, is carried out in a borehole, by driving a standard ‘split
spoon’ sampler using repeated blows of a 63.5 kg hammer falling through 762 mm. The
hammer is operated at the top of the borehole, and is connected to the split spoon sampler by
rods. The split spoon sampler is lowered to the bottom of the hole, and is then driven to a
depth of 450 mm in three 150 mm intervals and the blows are counted for each 150 mm
penetration. The penetration resistance (N) is the number of blows required to drive the split
spoon for the last 300 mm of penetration. The penetration resistance during the first 150 mm
of penetration is ignored, because the soil is considered to have been disturbed. SPT bore log
data includes the location of wells, SPT test results, ground water level, grain size, Liquid
limit, Plastic limit and strength of the soil and rock. SPT N values are useful for seismic site
characterization, site response, and liquefaction studies towards seismic microzonation. In
most cases the specific site response analysis, shear wave velocity, and shear modulus
(Gmax) of layers are estimated using relationships based on the SPT N values (Anbazhagan
et al., 2012 a and b).
Most of the surface wave applications follow three fundamental steps: Acquisition -
>Dispersion Analysis -> seeking the layered-earth model (Vs, Vp, h, r, etc.). Early
pioneering work in surface waves goes back to 1950s when the steady state method was first
used by Van der Pol (1951) and Jones (1955). At that time, it was based on the fundamental-
mode (M0)-only Rayleigh wave was assumed and all other types of waves—higher modes,
body waves, etc.—were ignored. This method later evolved to more-commonly called
Continuous Surface Wave (CSW) method (Matthews et al., 1996). In early 80s, a two-
receiver approach was introduced by investigators at the University of Texas (UT), Austin,
that was based on the Fast Fourier Transform (FFT) analysis of phase spectra of surface
waves generated by using an impulsive source like the sledge hammer. It then became widely
used among geotechnical engineers and researchers. This method was called Spectral
Analysis of Surface Waves (SASW) (Heisey et al., 1982). Simultaneous multi-frequency
(not mono- frequency) generation from the impact seismic source and then separation by FFT
during the subsequent data processing stage greatly improved overall efficiency of the
method in comparison to earlier methods such as the continuous surface wave (CSW)
method. Since then, significant research has been conducted at UT-Austin (Nazarian et al.,
Geotechnical and Geophysical Testing for Seismic Re-Qualification of Geotechnical Structures 3
1983; Rix et al., 1991; Al-Hunaidi, 1992; Gucunski and Woods, 1992; Aouad, 1993; Stokoe
et al., 1994; Fonquinos, 1995; Ganji et al., 1998) and a more complete list of the publications
on SASW up to early 1990s can be found in “Annotated bibliography on SASW” by Hiltunen
and Gucunski (1994). The overall procedure of SASW is as follows (Fig. 1).
SASW was being used in many geotechnical applications, one has to know difficulties in
three main categories:
The MASW (Multichannel Analysis of Surface Waves) method came into popular use among
the geotechnical engineers. The term “MASW” originated from the publication made on
Geophysics by Park et al. (1999). The project actually started in mid-90s at the Kansas
Geological Survey (KGS) by geophysicists who had been utilizing the seismic reflection
method—long used in the oil industry to image the interior of the earth for depths of several
kilometers. Called the high-resolution reflection method, it was used to image very shallow
depths of engineering interest (e.g., 100 m or less). MASW first measures seismic surface
waves generated from various types of seismic sources—such as sledge hammer—analyzes
the propagation velocities of those surface waves, and then finally deduces shear-wave
velocity (Vs) variations below the surveyed area that is most responsible for the analyzed
propagation velocity pattern of surface waves. Shear-wave velocity (Vs) is one of the elastic
constants and closely related to Young’s modulus. Under most circumstances, Vs is a direct
indicator of the ground strength (stiffness) and is therefore commonly used to derive load-
bearing capacity. After a relatively simple procedure, final Vs information is provided in 1-
D, 2-D, and 3-D formats. The common procedure for (1-D, 2-D, and 3-D) MASW surveys
4 SRGS, 17th December 2012, Delhi
Figure 2. Common procedure for MASW surveys for 1-D, 2-D, and 3-D Vs mapping.
Spreading multiple geophones in a straight line can help to get average shear wave velocity
of the spread at mid point during 1-D MASW survey. MASW are being used to measure
shear wave velocity of subsurface layers and which are applied in the seismic site
classification and microzonation studies. Multiple record acquisition (of twelve or more
Geotechnical and Geophysical Testing for Seismic Re-Qualification of Geotechnical Structures 5
Geotechnical test of SPT and Geophysical methods of MASW and GPR are widely used for
several geotechnical subsurface explorations and forensic geotechnical investigations,
However very limited studies have used above methods for seismic re-qualification of
6 SRGS, 17th December 2012, Delhi
geotechnical structures. Next sections describe applications SPT, MASW and GPR for
seismic re-qualification of geotechnical structures for earthquake hazards like ground
motion, ground failure/displacement, site effects (amplification), induced effects (
liquefaction, landslide and Tsunami), fire and flood.
5. SEISMIC RE-QUALIFICATION
Re- qualification/re evaluation of any structures are important to understand safety aspect of
for future hazards. Also evaluation of present condition of structure using advanced
technology and sophisticated instrument which are available during construction of
structures can help to retrofit structure to perform designed objectives. Re- qualification/re
evaluation are mostly adopted for nuclear power plant (NPP) facilities, however due to
increase in earthquake geotechnical hazards in urban areas, it necessitates re- qualification/re
evaluation of geotechnical structures which are responsible for catastrophe failures. This
section provides details of each earthquake hazards, geotechnical parameters responsible for
damage and quantification of these parameters using SPT, MASW and GPR.
When an earthquake occurs, seismic waves will generate and it will radiate away from the
source and travel rapidly through the earth’s crust. When these waves reach on earth’s
surface, it will produce shaking which will remain for seconds to minutes. The strength and
duration of shaking of a particular place depends upon the size and location of the earthquake
and the characteristics of the site. All the sites near the source of large earthquake will
produce huge damage. In fact, we can consider ground shaking as the most important hazard
because all other hazards are occurring due to the ground shaking. Investigations of seismic
source below or close to structures are very important to predict expected shaking level.
Geotechnical and Geophysical Testing for Seismic Re-Qualification of Geotechnical Structures 7
Presence of seismic source in the site/area can be interpreted by looking at spatial variation
geology from the finer geology map or soil map from bore hole data. In absence of these
details, carrying out such study may take considerable time. Alternatively, GPR can be used
to extract presence of fault signature in the area. Demanet et al. (2001) contrast electrical,
electromagnetic, seismic, and ground-penetrating radar (GPR) for imaging active faults in the
Roer Graben. Meghraoui et al. (2000) employed GPR, seismic refraction, and electrical
tomography for investigating active faulting along the Bree fault of the lower Rhine graben.
Grasmueck (1996) and Se´ne´chal et al. (2000) gave examples of shallow fault detection
using GPR. Successful application of geophysical methods in urban environments is often
hindered by the overwhelming contribution of cultural noise to the geophysical measurement.
Furthermore, trenching in cities is difficult due to intensive land use and the presence of
subsurface utilities. Fault detection using GPR requires an observable fault signature in the
radar record. The subsurface geology will dictate the nature of the GPR response, if any, to a
fault. Where sedimentary deposition results in well-layered sequences, the GPR time section
is typically characterized by numerous near horizontal reflections originating from the change
in ε at layer interfaces. Shallow fault detection with GPR is then based on flexured and
terminal reflectors associated with fault movement and flexuring (e.g. Demanet et al., 2001)
or sharp discontinuities at vertically offset reflectors (Meghraoui et al., 2000; Se´ne´chal et
al., 2000). Alternative scenarios may occur if a dipping discontinuity such as a fault or
fracture is directly associated with an electrical discontinuity. Such responses have been
effectively used to identify the location and dip of fractures, which represent a strong e
contrast, in crystalline rock (Olsson et al., 1992; Grasmueck, 1996; Miwa et al., 1999; Lane et
al., 2000; Seol et al., 2001). Pilon et al. (1991) recorded similar GPR events from reverse
faults in bedrock at Meteor Crater.
Aerial photo interpretation of the Aqaba regional surficial geology suggests that a strike-slip
fault emerges from the gulf and that slip is transferred to cross faults. This geometry
constrains the location of the Aqaba fault to lie east of the cross faults and west of alluvial fan
surfaces that contain no north- to northeast-striking fault lineaments. The location of the
Aqaba fault is therefore believed to lie within a 500-m swathe that is covered by the modern
city of Aqaba. Based on the aerial photo interpretation, it appears that the Aqaba fault follows
a recent wadi that has obliterated its active fault morphology. The fault appears to trend
northeast with increasing curvature as it dies out toward the north-east. This bending to the
northeast of the eastern bounding fault of a pull-apart has been noted for other large pull-
apart basins along the Dead Sea and the Sea of Galilee (Garfunkel et al., 1981; Garfunkel,
1981). Reches (1987) showed using clay model experiments that faults initially bend away
from each other at a dilational jog in ductile materials. This geometry led Reches (1987) to
conclude that the Arabian plate was thickening by ductile deformation. The Aqaba fault
should therefore have a reverse component to its motion. Profiles were constrained to vacant
lots and sidewalks. Interference from cultural debris within vacant lots and utilities under the
lots and sidewalks was a key concern in fault location efforts. Fig. 4 shows GPR Profile,
which traversed a large vacant lot where in situ material is partially exposed.
This combination of linear events was observed on five of seven GPR profiles run in
accessible areas of Aqaba City. Two typical profiles with fault interpretation is given in
Figure 5a and b. The integrated results presented in this study provide strong evidence for
fault zone detection, although the origin of the linear events in the radar record remains
uncertain. Linear events and offset/flexured reflectors observed at the control site are well
correlated with the location of two known faults. The control site was free of surface features
and field tests suggest that the linear events are not attributable to the survey procedure. The
8 SRGS, 17th December 2012, Delhi
asymmetry in the linear events about these faults can be interpreted as two planar reflectors,
dipping in the direction of the known dip on these faults. There is a strong correlation
between the GPR signature observed on five profiles run in Aqaba City and that observed at
the control site. Subsurface excavations at locations in Aqaba City are required to absolutely
constrain the interpretation of the geophysical data. Efforts to obtain permission to excavate
within the city of Aqaba are ongoing.
Fig 4. The interpreted location of two faults based on linear events and flexure reflectors
Figure 5. The interpreted location of two faults based on linear events and flexured reflectors
The reliability of earthquake assessments depends largely on coseismic displacement and slip
rate data gathered from geological evidence of past earthquakes along active faults. Very
little is known about timing, recurrence, slip rate, rupture segment and slip per event of
earthquakes generated along active faults. The geophysical mapping suggests location of
active faults in the area. This will help to re-evaluate important buildings such as the hospital
and schools that lie close to the faults. This type of study has significant implications for
disaster management programme. A similar study will help to relocate emergency services,
military bases, and residential areas in proximity to the fault zone. By doing so, disaster due
to future hazard can be minimized.
Geotechnical and Geophysical Testing for Seismic Re-Qualification of Geotechnical Structures 9
Soil condition modifies or undergoes deformation due to the ground motion. Soil
deformation/failure during earthquake depends on ground motion level and soil strength.
Deformation induces effects such as liquefaction and land slide. Modification of the
amplitude, frequency content and duration of ground motion depends on ground motion
characteristic and geotechnical properties such as type of soil, thickness of deposits, spatial
and topographical variation and soil strength. Modification of wave characters are called as
site effects, and are primarily based on geotechnical properties of the subsurface materials.
Site effects are a combination of soil and topographical effects, which can modify (amplify
and deamplify) the characteristics of the incoming wave field. Amplification and liquefaction
are the major effects of earthquakes that cause massive damage to the infrastructures. Ground
motion level and characters can be estimated by detailed seismic hazard analysis.
Geotechnical properties of type of soil, thickness of deposits, spatial and topographical
variation and soil strength are being estimated by carrying out SPT and MASW tests.
However quantification of geotechnical properties in the urban environment using multiple
SPT and MASW testing is almost impossible. Integrated approach of using few SPT and
MASW tests and many GPR tests can help to effectively evaluate subsurface geotechnical
properties with spatial variation.
Conventionally, drilling with SPT measurement is being followed to estimate strength and
thickness up to 30 m or hard rock level to estimate site and induced effects, which is time
consuming and more expensive. In this study SPT test has been carried out to measure SPT N
values and to know subsurface soil profiles with soil sampling and in the same location
MASW test has been carried out and shear wave velocity are measured. SPT N values can be
used to measure soil strength and there by deformation/failure possibility during future
earthquake and re-qualify the site for seismic liquefaction hazard. These results help to re-
qualify site at tested location or the entire site, assuming that there is similar condition
throughout the site. Similarly, shear wave velocity measured from MASW can be used to
estimate amplification by adopting simple site classification system or carrying out detailed
site response study. Extraction of depth of rock and soil type from MASW data is limited.
Soil sample obtained from SPT test can be used to selected shear modulus and damping
curves and to find depth of input. In the study Ground Penetration Radar (GPR) test has been
carried out in the same site. The soil thickness and rock depth variation is profiled, which is
not similar to the predictions based on single SPT and MASW test results. The GPR survey
has been carried out around the structures using ground coupled 25 MHz unshielded antenna
capable of penetrating an optimum depth of 25 m. All the recorded data was processed by
removing the background noise and applying filter and time corrections. The “DC removal”
is applied to remove constant component of the signal. DC component (component at the
zero frequency) is the mean value of the waveform. The arithmetic mean value of the samples
within the specified time range is taken as a constant component value. The application of
‘background removal’ routine is necessary when the instrument noise blocks up the desired
signal. The subtraction of the mean trace determined in the window with fixed size running
along the profile from the whole set of traces has been performed. Figure 6 shows typical
radargram obtained from GPR survey.
It can be noticed from GPR radargram that soil deposit in the site is not uniform and
thickness varies from place to place. Assuming uniform thickness based on borehole or 1-D
MASW and estimating liquefaction and site amplification for re-qualification may miss lead
the results. It should be remembered here that GPR can provide limited information about
10 SRGS, 17th December 2012, Delhi
soil type and strength. Integrated approach of using SPT, MASW and GPR has shown that
seismic re-qualification is more reliable due to accurate geotechnical subsurface information
available for liquefaction assessment and fast due to GPR survey. Detailed spatial variation
not only help to understand subsurface profile but also to generate 2-D and 3-D model of
subsurface for estimating site effects and landslide possibilities for seismic re-qualification.
In this study, an attempt has been made to map the subsurface profile in filled up land using
2-D MASW survey. MASW system consisting of 24 channels Geode seismograph with 24
geophones of 4.5 Hz capacity have been used in this study. The seismic waves are created by
impulsive source of 15 pound (sledge hammer) with 300 mm x 300 mm size hammer plate
with ten shots. The captured Rayleigh wave is further analyzed using SurfSeis software and
shear wave velocity was obtained. Figure 7 shows 2-D subsurface profile from MASW
survey. It can be noted from Figure 7 that there are considerable ground layering anomaly
present in the subsurface with shear wave velocity less than 180 m/s, which might be due to
filled up soil in the area. In order to confirm this, a bore hole has been drilled and borelog is
shown in Figure 8. While drilling, presence of boulders in soil does not allow penetration of
SPT sampler tube. The 2-D shear wave velocity profile shows that up to 6.0 m to 8.0 m, shear
Geotechnical and Geophysical Testing for Seismic Re-Qualification of Geotechnical Structures 11
wave velocity distribution is irregular due to layering and heterogeneity. These filled up soil
locations has loosely packed soil with bigger stones which results in the lower velocity (<180
m/s) patches in the 2-D velocity profile. Earlier, this area was a tank bed, which is now
encroached and filled for habitation. GPR has been carried out on pit filled with coarse grain
soil and waste old concrete. Radargram from this study is shown in Figure 9. It is found that
wherever concrete is present strong reflected waves are noticed and waves are absorbed in
the soil filled area due high saturation. Similarly Anbazhagan et al. (2012) investigated the
homogeneity of the compaction quality of new soil fill below a finished floor slab using GPR.
Analysis of GPR radargram showed that electromagnetic wave characters and amplitude of
wave traces are not uniform. Change in the wave characters indicate heterogeneity in quality
of compaction layers and Anbazhagan et al (2012) related the wave amplitude with water
content and degree of compaction. GPR results were compared with drilled borehole and
measured soil strength in terms of standard penetration test (SPT) N values at distressed
location due to poor compaction.
GPR was also used to identify the week spots in geotechnical structures such as earth dams,
embankment, dikes, rock slopes and soil layers below foundations and roads. Giuiig (2004)
used GPR to investigate river dike of 5700 km in Vietnam and identified week zones and
termites net holes, which are potential to create damage of entire structures. Leucci (2006)
carried out GPR survey at Chiesa Matrice Church built in 1500 and identified voids and
fractures in the subsoil, which were used to adopt adequate restoration work to increase the
structural reliability of the foundation and structure. Satriani et al (2010) used GPR to show
the presence of the fine and woody tree roots in the sand- clay box laboratory model. GPR
techniques can help to identify the tree roots growth close to buildings and to identify if the
roots are potential causes of damage to buildings in urban area (Satriani et al., 2010). Prinzio
et al (2010) carried out GPR study on river embankments and levee systems and detected
animal burrows to avoid piping failures. GPR is particularly suitable for embankments study
because it rapidly covers a wide area and gives detailed information about the possible
presence of voids and discontinuities (Prinzio et al., 2010).
12 SRGS, 17th December 2012, Delhi
BORE LOG
Location RVCE Campus Date of commencement 09.05.2006
BH No RVCE-1 Date of completion 12.05.2006
Ground Water Table 15m
Depth Thickness Details of Sampling SPT
Below Soil Description of Strata Legend Type Depth N Value
GL(m) (m) (m)
0.0
Filled Up Soil/ Gravelly Soil
2.0 with Boulders
4.0 SPT 5.0 10+R
Filled Up Soil/ Gravelly Soil
5.0 with Boulders
6.0 6
Weathered soil SPT 7.5 8+R
7.0 with boulders
6m to 8m 2
SPT 9 R
13.0 8 m to 15 m
14.0
15.0 7 SPT 15 R
Hard Rock
20.0 From 18 m
22.0
Figure 8. Bore log from filled soil site and corresponding to Figure 7
Xu et al. (2010) showed a GPR capability to detect ferralsol in tropical and subtropical
regions and shows high value of application in detecting some hidden troubles such as caves
and settlements at low to moderate depths inside dams in Karst regions. GPR technique is
also capable of detecting carbonate rocks to certain depth and yield precise results, these
results can be applied to analysis and discovery of leakage channels inside dikes and dams
(Xu et al. 2010). Zhang et al., (2010) used GPR to detect the grout thickness behind the lining
segments of metro tunnel lines in Shanghai, China up to depth of 1 m. This GPR detection
can be used to mitigate the risks of long-term ground settlement, a critical issue of shield
tunnel construction in soft soil areas. GPR is being used for several areas of geotechnical
engineering, but way of investigation and data interpretation is unique to each case.
Optimized GPR data are useful to view geotechnical structure in detail without drilling.
Integrated approach of SPT, MASW and GPR for geotechnical structure investigation can
provide qualitative data for seismic re-qualification studies.
Geotechnical and Geophysical Testing for Seismic Re-Qualification of Geotechnical Structures 13
6. SUMMARY
This paper presented three popular test methods in geotechnical and geophysical survey
methods and case studies relevant to the seismic re-qualification of geotechnical structures.
Integrated investigation of geotechnical structures using SPT, MASW and GPR can provide
sufficient information about soil type, strength, thickness, bedrock location, heterogeneity
and weak spots. These parameters are very useful to re evaluate seismic hazards of ground
motion, ground failure/displacement, site effects (amplification), induced effects (
liquefaction, landslide and Tsunami), fire and flood. Even though this paper related several
case studies of MASW and GPR to seismic re-qualification, very limited analysis or
publications are available in this area.
REFERENCES
Heisey, J.S., Stokoe II, K.H., and Meyer, A.H. (1982). Moduli of pavement systems from
Spectral Analysis of Surface Waves, Transp. Res. Rec., v. 852, Washington D.C, p. 22-
31.
Nazarian, S., Stokoe II, K.H., and Hudson, W.R., (1983), Use of spectral analysis of surface
waves method for determination of moduli and thicknesses of pavement systems, Transp.
Res. Rec. v. 930, Washington DC, p. 38-45.
Rix, G.J., Stokoe II, K. H., and Roesset, J.M., (1991), Experimental study of factors affecting
the Spectral Analysis of Surface Waves method, Research report 1123-5, Center for
Transportation Research, The University of Texas at Austin.
14 SRGS, 17th December 2012, Delhi
Al-Hunaidi, M.O., (1992), Difficulties with phase spectrum unwrapping in spectral analysis
of surface waves nondestructive testing of pavements, Can. Geotech. J., v. 29, p. 506-
511.
Gucunski, N., and Woods, R.D., (1992), Numerical simulation of the SASW test, Soil Dyn
and Earth-quake Eng. J., v. 11, n. 4, p. 213-227.
Aouad, M.F., (1993), Evaluation of Flexible Pavements and Subgrades Using the Spectral-
Analysis-of-Surface-Waves (SASW) Method, PhD thesis, Univ. of Texas at Austin,
Texas.
Stokoe II, K.H., Wright, S.G., Bay, J.A., and Roesset, J.M., (1994), Characterization of geotechnical
sites by SASW method, in Geophysical characterization of sites. ISSMFE Technical Committee
#10, edited by R.D. Woods, Oxford Publishers, New Delhi.
Fonquinos Mera, R., (1995), Dynamic nondestructive testing of pavements, PhD thesis, Univ.
of Texas at Austin, Texas.
Ganji, V., Gucunski, N., and Nazarian, S., (1998), Automated inversion procedure for
spectral analysis of surface waves, J. Geotech. Engrg., ASCE, v. 124, n. 8, p. 757-770.
Hiltunen, D.R., and Gucunski, N., (1994), Annotated bibliography on SASW, in Geophysical
characterization of sites, ISSMFE Technical Committee #10, edited by R.D.Woods,
Oxford Publishers, New Delhi.
Sheu, J.C., Stokoe, K. H., II, and Roesset, J. M., (1988), Effect of reflected waves in SASW
testing of pavements: Transportation Res. Record, v. 1196, p. 51–61.
Hiltunen, D.R., and Woods, R.D., (1990), Variables affecting the testing of pavements by the
surface wave method, Transp. Res. Rec., v. 1260, p. 42-52.
Foti, S., (2000), Multistation methods for geotechnical characterization using surface waves:
PhD thesis, Politecnico di Torino, Italy.
Al-Hunaidi, M.O., (1992), Difficulties with phase spectrum unwrapping in spectral analysis
of surface waves nondestructive testing of pavements, Can. Geotech. J., v. 29, p. 506-
511.
Miller, R.D., Xia, J., Park, C.B., and Ivanov, J.M., (1999), Multichannel analysis of surface
waves to map bedrock, Kansas Geological Survey, The Leading Edge, December, p.
1392-1396.
Benson, R. and Glaccum, R., (1979). The application of ground-penetrating radar to soil
surveying. Final Rep. NASA, Cape Kennedy Space Cent., FL. Technos, Miami, FL.
Johnson, R.W., Glaccum, R. and Wojtasinski, R., (1980). Application of ground penetrating
radar to soil survey. Proc. Soil Crop Sci. Soc. Fla., 39: 68-72.
Doolittle, J.A. and Collins M.E. (1995). Use of soil information to determine application of
ground penetrating radar. Journal of Applied Geophysics, 33, 101-108.
Anbazhagan P, Deepu Chandran and Shashank Burman (2012). Investigation of Soil
Compaction Homogeneity in the Finished Building Using Ground Penetrating Radar.
Proceedings of 6th ASCE conference on Forensic Engineering, 773- 782.
Demanet, D., Renardy, F., Vanneste, K., Jongmans, D., Camelbeeck, T., Meghraoui, M.,
(2001). The use of geophysical prospecting for imaging active faults in the Roer Graben,
Belgium. Geophysics, 66, 78– 89.
Meghraoui, M., Camelbeeck, T., Vanneste, K., Brondeel, M., Jongmans, D., (2000). Active
faulting and paleoseismology along the Bree fault, lower Rhine graben, Belgium. J.
Geophys. Res. 105 (B6), 13809–13842.
Grasmueck, M., (1996). 3-D ground-penetrating radar applied to fracture imaging in gneiss.
Geophysics 61, 1050–1064.
Se´ne´chal, P., Pe´rroud, H., Se´ne´chal, G., (2000). Interpretation of reflection attributes in a
3-D GPR survey at Vallee d’Ossau, western Pyrenees, France. Geophysics, 65, 1435–
1445.
Geotechnical and Geophysical Testing for Seismic Re-Qualification of Geotechnical Structures 15
Olsson, O., Falk, L., Forslund, L., Lundmark, L., Sandberg, E., 1992. Borehole radar applied
to the characterisation of hydraulically conductive fracture zones in crystalline rock.
Geophys. Prospect. 40, 109– 142.
Grasmueck, M., (1996). 3-D ground-penetrating radar applied to fracture imaging in gneiss.
Geophysics 61, 1050–1064.
Miwa, T., Sato, M., Nitsuma, H., (1999). Subsurface fracture measurement with polarimetric
radar. IEEE Trans. Geosci. Remote Sens., 37 (2), 828–837.
Lane Jr., J.W., Buursink, M.L., Haeni, F.P., Versteeg, R.J., (2000). Evaluation of ground-
penetrating radar to detect free-phase hydrocarbons in fractured rocks—results of
numerical modeling and physical experiments. Ground Water, 38 (6), 929–938.
Seol, S.J., Kim, J.H., Song, Y., Chung, S.H., (2001). Finding the strike direction of fractures
using GPR. Geophys. Prospect. 49, 300– 308.
Pilon, J.A., Grieve, R.A.F., Sharpton, V.L., (1991). The subsurface character of meteor
crater, Arizona, as determined by groundprobing radar. J. Geophys. Res. 96 (E1), 15563–
15576.
Garfunkel, Z., Zak, I., Freund, R., (1981). Active faulting in the Dead Sea rift.
Tectonophysics, 80, 1 – 26.
Garfunkel, Z., (1981). Internal structure of the Dead Sea leaky transform (rift) in relation to
plate kinematics. Tectonophysics, 80, 81–108.
Leucci G, (2006). Contribution of Ground Penetrating Radar and Electrical Resistivity
Tomography to identify the cavity and fractures under the main Church in Botrugno
(Lecce, Italy), Journal of Archaeological Science, 33, 1194-1204.
Satriani A., A. Loperte, M. Proto, and M. Bavusi, (2010).Building damage caused by tree
roots: laboratory experiments of GPR and ERT surveys, Advances in Geosciences, 24,
133–137.
Prinzio M. D. , M Bittelli , A Castellarin., and P. R. Pisa (2010). Application of GPR to the
monitoring of river embankments, Journal of Applied Geophysics, 71, 53–61.
Xu X., Q. Zeng , D. Li , J. Wu , X. Wu , and J. Shen (2010). GPR detection of several
common subsurface voids inside dikes and dams, Journal of Applied Geophysics, 71, 53–
61.
Zhang F, X. Xie and H Huang, (2010), Application of ground penetrating radar in grouting
evaluation for shield tunnel construction, Tunneling and Underground Space
Technology, 25, 99–107.
Giuiig N V, (2004). Geotechnical and Engineering Applications of GPR in Vietnam,
Proceeding of Tenth International Conference 011 Ground Penetrating Radar, 21-24
June, Delft, the Netherlands, 399-401.
Matthews MC, Hope VS and Clayton CRI, (1996). The use of surface waves in the
determination of ground stiffness profiles, Proc. Instn Civ Engrs Geotech. Engg., 119,
84-95
Author's personal copy
a r t i c l e i n f o abstract
Article history: Recent research has demonstrated that axial load alone can cause a slender pile to fail by forming a
Received 5 December 2011 plastic hinge, if soil surrounding the pile liquefies in an earthquake. This failure mechanism is due to
Received in revised form buckling instability. Lateral loads from lateral spreading of the surrounding soil and/or inertia and
20 August 2012
imperfection inherent in a pile result in increased deflection, which can promote more rapid collapse.
Accepted 25 October 2012
These effects are secondary to the basic requirement so that axially loaded piles passing through
liquefiable soils should be checked against Euler’s buckling in addition to the bending mechanism of
failure, i.e. incorporation of P D effect. While fewer large diameter piles are currently being used in
modern construction practice (which performed better in liquefiable areas rather than multiple small
diameter piles), there are many pile-supported structures where buckling considerations were not
taken into account and therefore may need retrofitting. This paper develops a probabilistic tool which
can be used for assessing the likelihood of a buckling failure of existing piled foundations due to a
scenario earthquake. This tool can equally serve as a valuable decision-support tool for implementing
earthquake risk mitigation measures. A case study is presented to show the applicability of the method.
Crown Copyright & 2012 Published by Elsevier Ltd. All rights reserved.
1. Introduction yield moment capacity MY and plastic moment capacity MP) of the
member under consideration. In contrast, buckling is an unstable
Pile foundations are one of the most common forms of founda- mechanism. It occurs suddenly and drastically when the elastic
tion in seismically liquefiable areas, as it is thought to be an all-safe critical load is reached. It is the most destructive mode of failure
solution. It is routinely used to support foundations of short-to- and depends on the geometrical properties of the member, i.e.
medium span bridges and buildings typically over 4 stories. All slenderness ratio, rather than the member strength. For example,
current design methods, such as JRA (1996), NEHRP [31], IS:1893 steel pipe piles with identical length and diameter but having
[18], and Eurocode 8 [8], focused on bending strength of the pile to different yield strengths (e.g. 200 MPa, 500 MPa, and 1000 MPa) will
avoid bending failure due to lateral loads (combination of inertia buckle at similar axial loads but can resist different amounts of
and/or lateral spreading). In contrast to these conventional design bending. In other words, bending failure may be avoided by increas-
codes, which advocate bending mechanism as the main design ing the yield strength of the material, but it may not suffice to avoid
consideration, recent research showed that an axially loaded pile buckling. To prevent buckling failure, there should be a minimum pile
can be laterally unsupported in liquefied soils and is susceptible to diameter depending on the depth of liquefiable soils. Therefore,
buckling failure (see for example [2–6,21,24,27,35]). Buckling designing against bending would not automatically satisfy buckling
instability under the interaction of axial and lateral loads can be requirements. It is envisaged that there are plenty of existing pile-
more critical design consideration because of its sudden nature and supported structures that may need retrofitting.
sensitivity to imperfection (see for example [11]). More recently, This paper develops a probabilistic method to assess the pile
Bhattacharya et al. [7] included the effects of dynamics on the foundation instability due to buckling mechanism by considering
combined axial and lateral loads on a pile foundation. Essentially, a scenario earthquake. The methodology, which is illustrated in
piles in liquefied soils may be better regarded as columns carrying Figs. 1 and 2, consists of four components:
lateral loads rather than laterally loaded beams.
In design, beam bending and column buckling are approached (a) Estimation of input ground motion at a site of interest based on
differently. Bending is a stable mechanism as long as the pile remains nearby observations. This can be carried out by using a regional
elastic and secondary failure (for example local buckling) is not a ground motion prediction equation (GMPE), such as Kanno et al.
possibility. This failure mode depends on the bending strength (e.g. [23] and Zhao et al. [41] for Japanese earthquakes, together with
a spatial correlation model of seismic excitations, such as Goda
n
Corresponding author.
and Hong [12] and Goda and Atkinson [13,14].
E-mail addresses: S.Bhattacharya@surrey.ac.uk, (b) Evaluation of the extent of liquefaction depth based on soil
subhamoy.bhattacharya@gmail.com (S. Bhattacharya). characteristics and seismic excitation which will lead to the
0267-7261/$ - see front matter Crown Copyright & 2012 Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.soildyn.2012.10.004
Author's personal copy
length. The ground motion at a building site where ground failure Using the seismological models described above and a Monte
and pile foundation failure occur is rarely recorded directly. Carlo technique, samples of PGAbldg given PGArec (with information
Rather, it is routinely assumed that the observed ground motion on Mw, Rrec, VS30,rec, Rbldg, VS30,bldg, D, applicable GMPE, and spatial
at a nearby location is applicable to the building site. For instance, correlation model) can be generated by following two steps:
if PGA is obtained as 0.5g at a recording station that is away from
a building site by a short distance (e.g. less than 1 km), the 1. Calculate a value of eobs obs
rec as erec ¼ logPGArec f M w ,R ,V .
rec S30,rec
same PGA value may be adopted as long as local site conditions at 2. Generate a sample of PGAbldg from PGAbldg ¼ exp f M w ,Rbldg ,
the two sites are similar. Alternatively, one can estimate PGA V S30,bldg Þ þ ebldgjerec ¼ eobs Þ where ebldg jerec ¼ eobs is a normal random
rec rec
at the building site conditioned on the observed PGA at the variable with mean given by Eq. (4) and variance given by Eq. (5).
recording site by using a GMPE and spatial correlation model of
peak ground motions. PGA values at the recording site and the The generated samples of PGAbldg have desirable statistical
building site (due to the same earthquake), PGArec and PGAbldg, can characteristics, consistent with the seismological models and the
be expressed as observed PGA at the recording site.
Although the use of empirical GMPEs for estimating ground
logPGArec ¼ f Mw ,Rrec ,V S30,rec þ erec , ð1aÞ
motion parameters at unobserved sites, as described above, is a
valid approach, recent studies on uncertainty of the GMPEs
and suggest that for specific scenarios (i.e. similar source, path, and
site), significant reduction of the total sigma s is possible [26,29].
logPGAbldg ¼ f M w ,Rbldg ,V S30,bldg þ ebldg , ð1bÞ
For example, analysis results by Morikawa et al. [29] indicate the
reduction of s by 50–60%. Because a situation/application dis-
where f(Mw,R,VS30) is the median GMPE for PGA as functions of the cussed in this study is focused on a scenario, rather than a
moment magnitude Mw, the distance measure R (typically, closest prediction for multiple scenarios (as in probabilistic seismic
distance to fault rupture plane), and the average shear-wave velocity hazard analysis), the reduction factor of 50% for the total sigma
in the uppermost 30 m VS30 (m/s); and e is the random error that is is considered as a base case. The implication of new findings on
modeled as a normal variable with zero mean and logarithmic the assessment of pile foundation stability will be explored in
standard deviation (sigma) s. Note that values of R, Vs30, and e differ sensitivity analysis (Section 3.3).
for the recording and building sites. For Japanese earthquakes,
several GMPEs, such as Kanno et al. [23] and Zhao et al. [41], are
2.2. Assessment of the unsupported pile length DL
available. Moreover, uncertainties associated with prediction of
PGArec and PGAbldg (i.e. erec and ebldg) are correlated. The covariance
The next step is to conduct probabilistic liquefaction potential
of erec and ebldg, COVerec,ebldg(D), is given by [12]:
evaluation of a soil column (along a pile) at the building site to
COV erec , ebldg ðDÞ ¼ s21 þ r2 ðDÞs22 , ð2Þ estimate DL for a given seismic excitation level. Such assessment
is often conducted by using simplified stress-based methods of
Seed and Idriss [34] based on standard penetration test (SPT) data,
where D (km) is the separation distance between the recording and cone penetration test (CPT) data, and shear-wave velocity (VS)
building sites; s1 and s2 are the inter-event sigma and intra-event data. A comprehensive guideline for these methods can be found
sigma (note: s2 ¼ s21 þ s22), respectively; and r2(D) is the intra-event in [17,19,21,40]. Recently, probabilistic procedures for liquefac-
spatial correlation coefficient. The mathematical derivation of Eq. (2) tion initiation have been developed by considering different in-
is given in [12]. For Japanese earthquakes, the intra-event correla- situ measures for describing soil strength. Using a sophisticated
tion coefficient r2(D) is given by [13,14] Bayesian regression analysis and well-screened case studies, Cetin
et al. [9,10] developed a statistical model for calculating the
r2 ðDÞ ¼ max½1:389 exp 0:206D0:386 0:389,0, ð3Þ probability of liquefaction initiation based on SPT data, while
Moss et al. [30] developed a counterpart using CPT data. Using the
which was derived based on statistical analysis of strong ground first-order reliability method, Juang et al. [22] developed a similar
motion data. It is noteworthy that erec and ebldg are bivariate normal model based on VS data. The significance of these developed
variables with covariance COVerec,ebldg. Thus the conditional prob- models is that key uncertainties associated with the input data/
ability distribution of ebldg given that erec equals eobs
rec is the normal
parameters and the adopted models themselves are taken into
distribution with mean equal to account; the developed models can evaluate unbiased potential of
liquefaction initiation, and are useful to conduct probabilistic
COV erec , ebldg ðDÞ
m ¼ eobs
rec , ð4Þ liquefaction hazard analysis [15].
ebldg erec ¼ eobs
rec
s2 In this study, the SPT-based model developed by Cetin et al.
[9,10] is adopted for liquefaction potential assessment, because
and variance equal to
this is one of the most extensive and up-to-date probabilistic
methods. The probability of liquefaction initiation PL at a depth of
s2e jerec ¼ eobs ¼ s2 1½COV erec , ebldg ðDÞ=s2 2 , ð5Þ
bldg rec interest can be estimated as
N 1,60 ð1 þ 0:004FCÞ13:32 lnðCSReq Þ29:53 lnðM w Þ3:70 lnðsv’ =P a Þ þ 0:05FC þ 16:85
PL ¼ F , ð6Þ
2:7
noting that COVerec,ebldg(D)/s2 represents the total correlation coeffi- where F is the standard normal function; N1,60 is the corrected
cient of erec and ebldg (including both inter-event and intra-event SPT counts (but not adjusted for fines content); FC is the fines
components). content (in percentage); CSReq is the cyclic stress ratio (but not
Author's personal copy
adjusted for the moment magnitude); and s0 v is the vertical if there is a complex soil profile with alternating liquefiable and
effective stress. The model coefficients shown in Eq. (6) are non-liquefiable soil layers, more detailed analysis is required.
obtained for the case ‘‘with measurement errors’’ [9]. The correc- For instance, a non-liquefied layer with thickness less than or equal
tion of the in-situ N values to N1,60 can be done by using a set of to 5–6 times the diameter of the pile (d) (Cases 1 and 2 in Fig. 3)
correction factors, as suggested by Youd et al. [40] and Cetin et al. may be considered as ‘‘unsupported’’, if the non-liquefied layer is
[10]. The value of CSReq for a given depth d is calculated by inserted within thick liquefied layers, while a non-liquefied layer
PGA sv with thickness greater than or equal to 9–10 times the diameter
CSReq ¼ 0:65 r , ð7Þ
g sv’ d (Case 3 in Fig. 3) may be sufficient to provide partial restraint of the
lateral pile movement at the depth (note: the mentioned factors are
where PGA is the geometric mean of two horizontal components
for illustration only and the suitable value should be selected
at ground surface [40]; sv is the total vertical stress; and rd is the
non-linear stress reduction factor. Cetin et al. [9,10] conducted carefully on a case-by-case basis). This issue of partial restraint
extensive site response analyses to derive a new probabilistic depends on the relative pile-soil stiffness and the diameter of the
relation for rd, where the mean relation is given by pile, which is similar to obtaining fixity of pile for bending calcula-
tions (Appendix A). It is acknowledged that such assessment
þ 0:999M w þ 0:0525V S12
1 þ 16:258 þ23:0132:949PGA
0:201 expð0:341½minðz,20Þ þ 0:0785V S12 þ 7:586Þ involves uncertainty and subjective judgement. A rigorous assess-
mrd ¼ 23:0132:949PGA þ 0:999M w þ 0:0525V S12
1 þ 16:258 þ 0:201 expð0:341½0:0785V S12 þ 7:586Þ
ment of defining the unsupported pile length for a complex soil
0:0046 maxðz20,0Þ, ð8Þ profile requires experimental investigations, which is beyond the
scope of the current study. However, sensitivity of the analysis
and the standard deviation is given by results for different situations will be investigated in Section 3.3.
srd ¼ 0:0198 ½minðz,12Þ0:85 , ð9Þ
2.3. Assessment of the critical pile length HC
in which VS12 (m/s) is the average shear-wave velocity in the
uppermost 12 m and z (m) is the depth from ground surface.
2.3.1. Buckling failure mechanism
Using Eqs. (6)–(9), values of PL at different depths along the pile
length can be evaluated. The input parameters, PGA, Mw, N1,60, FC, This section of the paper describes the mathematical back-
rd, s0 v, and sv, are treated as a random variable, and are sampled ground behind the rationale of obtaining the critical pile depth HC.
in Monte Carlo simulation. Then, occurrence of liquefaction for a Before the onset of shaking, the static axial load Pstatic acts on each
given layer can be assessed by comparing a standard uniform pile beneath a building, assuming that each pile is equally loaded
variable U with PL; if U is less than PL, liquefaction initiation is during static condition and neglecting any eccentricity of loading.
predicted for the layer (note: this re-sampling is carried out to During an earthquake, inertial action of the superstructure
obtain a soil profile over depth indicating liquefied/non-liquefied imposes the dynamic axial load on the piles, which will increase
layers). Based on the above simulation-based procedure, one can the axial load on some piles. These piles with increased axial
obtain many realizations of a liquefaction occurrence profile over loads may be vulnerable to buckling. An estimate of the max-
the pile length. imum axial compressive load acting on a pile can be given by
The next step is to determine the unsupported pile length (DL)
Pdynamic ¼ ð1þ aÞP static , ð10Þ
based on the liquefaction profile. DL is equal to the thickness of
liquefied soil layers plus some additional length necessary for fixity where a is termed as the dynamic axial load factor [2,5] and is a
at the bottom of the liquefied soils. Typical calculations (see function of type of superstructure, height of the center of mass of
Appendix A) show that the fixity depth is about 3–5 times the the superstructure, and characteristics of the earthquake shaking
diameter of the pile. If a relatively thin non-liquefied layer is (e.g. frequency content and amplitude).
overlain and underlain by thick liquefied layers, lateral restraint of For buckling instability analysis, each pile needs to be eval-
the pile at the non-liquefied layer might not be expected, i.e. the pile uated with respect to its end conditions, i.e. fixed, pinned, or free.
is unsupported. In such a case, the thin non-liquefied layer can be Each pile in a group of identical piles will have the same buckling
ignored in determining DL and the unsupported length needs to be resistance as a single pile. If a group of piles is fixed in a stiff pile
extended until a thick non-liquefied layer is encountered. However, cap and embedded sufficiently at the tip, as in Fig. 2(b), the pile
Table 1
Values of K in Eq. (14).
Boundary condition of the pile at the top and bottom of the liquefied layer Effective length K Examples
Top Bottom
Fixed Fixed [sufficient embedment at the dense layer] Leff ¼0.5HC 0.5 Pile group with raked piles
Free to translate but restrained Pinned [insufficient embedment at the dense layer] Leff ¼2HC 2 See for example NFCH building in
against rotation—sway frame Bhattacharya [2]
Free to translate but restrained Fixed [sufficient embedment at the dense layer] Leff ¼HC 1 Most cases fall under this category,
against rotation—sway frame see for example Fig. 2
Fixed in direction but free to rotate Fixed [sufficient embedment at the dense layer] Leff ¼0.7HC 0.7 Pile group with raked piles, Improper
pile–pile cap connection
Fixed in direction but free to rotate Pinned [less embedment at the dense layer] Leff ¼HC 1 Pile groups with raked piles.
Improper pile–pile cap connection
Free i.e. unrestrained against Fixed [sufficient embedment at the dense layer] Leff ¼2HC 2 Piles in a row such as the Showa
rotation and displacement Bridge piles
group will buckle in side sway. The elastic critical load of a single
pile, Pcr, can be estimated as
p2 EI
P cr ¼ 2 ð11Þ
Lef f
where Leff is the effective length, i.e. Euler’s equivalent buckling
length of a strut pinned at both ends, EI is the bending stiffness of
the pile. Further details on the effective length of the pile can be
found in [6] (see Table 1). The pile that will be considered in the
example of the paper will have an effective length (Leff) equal to
the unsupported length (DL). The unsupported length of the pile
DL is equal to the thickness of liquefiable soil plus some additional
length necessary for fixity at the bottom of the liquefiable soil.
The applicability of the elastic critical load, as in Eq. (11), to
pile buckling failure is an important factor. Experiments show
that the actual failure load of a slender column is much lower
than that predicted by Eq. (11). Rankine [33] recognized that the
actual failure involves an interaction between elastic and plastic
modes of failure. Lateral loads and inevitable geometrical imper-
fection lead to creation of bending moments in addition to axial
loads. Bending moments have to be accompanied by stress
Fig. 4. Buckling amplification factor versus normalized axial load.
resultants that diminish the cross-sectional area available for
carrying the axial load; thus the actual failure load is likely to be
less than the elastic critical load, i.e. Pfailure oPcr. Equally, the Moreover, it would also be unwise to use a factor of safety less
growth of plastic bending zones reduces the effective elastic than 3 against the Euler load of a pile, i.e. (P/Pcr ¼0.33). Such
modulus of the section, thereby resulting in the decreased critical consideration is consistent with general design practice where
load for buckling (i.e. capacity). Furthermore, these processes feed structural engineers use a factor of safety of at least 3 against linear
each other as explained by Horne and Merchant [16]. As the elastic buckling to take into account the eccentricity of load, dete-
elastic critical load is approached, all bending effects are magni- rioration of elastic stiffness due to plastic yielding, and unavoidable
fied. Stability analysis of elastic columns, Timoshenko and Gere imperfection. The actual failure load Pfailure is therefore some factor f
[36], showed that if lateral loads in the absence of axial load (f o1) times the theoretical Euler’s buckling load given by Eq. (11).
would create a maximum lateral displacement d0 in the critical
Pf ailure ¼ fP cr , ð13Þ
mode shape of buckling, then the displacement d under the same
lateral loads but with the concurrent axial load P is given by: Based on the above discussion, it may be inferred that buckling
instability is initiated at f ¼0.35. It is noted that in reality, this
d 1
¼ ð12Þ factor depends on the axial load, imperfection of piles, and
d0 1P=Pcr residual stress in the pile due to driving. The selection of f is
The term d/d0 is the buckling amplification factor (i.e. ampli- one of the significant sources of uncertainty in determining the
fication of lateral displacement due to the presence of the axial critical pile depth HC.
load). Fig. 4 presents a graph of the buckling amplification factor
plotted against the normalized axial load P/Pcr, where P denotes 2.3.2. Determination of ‘‘critical depth’’ HC
the applied axial load. It can be observed from Fig. 4 and Eq. (12) In the limit state condition of failure, Pdynamic ¼Pfailure. For a
that if the applied load is 50% of Pcr, the amplification of lateral structure shown in Fig. 2, Leff ¼ HC in Eq. (11). In order to
deflection due to lateral loads is about 2 times. At these large generalize the boundary condition of the pile (i.e. pile head
deflections, secondary moments will be generated due to P D fixity condition with pile cap/superstructure and the fixity at
moment, leading to more deflection. It is therefore important to the interface between liquefiable and non-liquefiable layers at
remain in the linear regime and not in anyway near the asymp- deeper depths), one may write as follows:
totic region, where the buckling amplification factor increases
dramatically (e.g. P/Pcr 40.6). Lef f ¼ K HC , ð14Þ
Author's personal copy
where K is the column effective length factor (e.g. K ¼1 for free In a post-liquefaction situation, the building is supported
lateral translation but restrained against rotation-sway motion). by piles that have relatively long unsupported lengths of DL
Values of K for other boundary conditions of the pile are given in (as evaluated based on the liquefaction initiation analysis). In
Table 1. this case, the natural vibration period in a post-liquefaction stage,
With the above-mentioned assumptions, Eq. (13) can be Tpost can be calculated as
rewritten as sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
W=g
T post ¼ 2p ð18Þ
fp2 EI N p 12EI=D3L
P dynamic ¼ fP cr ¼ ð15Þ
K 2 H2C
where 12EI/(DL)3 is the lateral stiffness of each pile. Then, the
maximum axial force acting on a pile is given by
Rearranging Eq. (15) gives the estimate of the critical depth HC W=g SA T post , xpost DL þ b3 HB
Ppost ¼ ð19Þ
for a pile as follows: N p =2 B
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
fp2 EI fp2 EI where SA(Tpost,xpost) (g) is the spectral acceleration at Tpost with
HC ¼ ¼ ð16Þ damping ratio xpost (typically, 10–30% representing the damping of
K 2 Pdynamic K 2 ð1 þ aÞPstatic
liquefied soil); and b3 is the coefficient to account for the effective
height where the inertia acts in a post-liquefaction condition
Therefore, the assessment of HC is based on the calculation of the (typically 0.5). The underlying assumption of Eqs. (18) and (19) is
critical buckling load for a pile foundation surrounded by liquefied that the building is considered as a rigid mass and the pile provides
soils (from which the pile cannot receive sufficient lateral support). the primary lateral stiffness to the building. It must be mentioned
Key parameters in Eq. (16) are f, K, and a. Reasonable values of f that this approach is very simple and many uncertainties, such as
and K can be selected based on engineering judgment. effects of vertical inertia, timing of max inertia and timing of loss of
One of the critical factors is the assessment of a. This can be facili- lateral support due to liquefaction, and potential effects due to
tated by examining axial forces acting on a pile in pre-liquefaction lateral spreading, are not taken into consideration.
and post-liquefaction situations (Fig. 5). In a pre-liquefaction stage, By defining a ¼ max(Ppre/Pstatic,Ppost/Pstatic), a is given by
the plane of fixity of a building with pile foundation surrounded by 2
non-liquefied soils can be estimated by following a similar procedure
a¼ max b1 SA T pre , xpre Df ix þ b2 HB ,SA T post , xpost DL þ b3 HB
gB
as in Appendix A. It may be noted that the depth from ground surface ð20Þ
to the plane of fixity is denoted by Dfix in Fig. 5 and is typically a few
meters. By assuming that the natural vibration period of the building
The additional information needed for evaluating a is the
before liquefaction is Tpre, the maximum axial force acting on a pile
estimated spectral acceleration values at vibration periods Tpre
due to inertia can be calculated as
and Tpost with damping ratios xpre and xpost (note: Tpre and Tpost are
b W=g SA T pre , xpre Df ix þ b2 HB random variables; in particular, Tpost is significantly affected by
P pre ¼ 1 , ð17Þ liquefaction initiation analysis). One way to estimate SA(Tpre,xpre)
Np =2 B
and SA(Tpost,xpost) is to adopt a similar analysis method that is
where W is the total weight of a building (note: W¼PstaticNp); outlined in Section 2.1. For this purpose, the correlation model for
SA(Tpre,xpre) (g) is the spectral acceleration at Tpre with damping ratio spectral accelerations can be used [13], and the corresponding
xpre (typically, 2–5%); b1 is the coefficient to account for modal mass covariance matrix can be constructed. Subsequently, error terms
for the fundamental vibration mode (typically, 0.8–0.9); b2 is the for SA(Tpre,xpre) and SA(Tpost,xpost) are generated from the condi-
coefficient to account for the effective height where the inertia due to tional ground motion model (see Eqs. (1)–(5)). It is noted that
the modal mass acts (typically, 0.65–0.75); HB is the height of when SA(Tpre,xpre) and SA(Tpost,xpost) are generated at the building
building; Np is the number of piles (assuming that an equal site, the conditional simulation should be carried out by taking
number of piles are positioned in two rows); and B (m) is the realized values of PGArec and PGAbldg into account (note: it is
foundation width between the two rows of piles (note: the possible to consider spectral accelerations at the recording
foundation width B is along the direction where axial force station, in addition to PGA if ground motion time-histories at
induced by overturning moment due to lateral inertia). Tpre can the recording station are available). For adjusting damping ratios,
be estimated by using an empirical equation, such as conversion factors can be derived from response spectral analysis
Tpre ¼0.09HB/B0.5 (see [1] also adopted in IS 1893). or empirical conversion equations, such as in Eurocode 8 [8].
Based on the procedure described above, both DL and HC can be shear-wave velocities in the uppermost 12 and 30 m are estimated
assessed probabilistically. The occurrence of pile foundation failure to be about 147 and 216 m/s, respectively. The water table level was
due to buckling mechanism is indicated if HC oDL. By sampling about 2 m below ground surface. The post-earthquake investigation
DL and HC many times, probabilistic assessment of liquefaction- by Tokimatsu et al. [37] indicated that soil layers shallower than 9 m
induced pile foundation failure due to buckling mechanism can be were liquefied (based on a simplified stress method). Moreover, the
carried out. borehole logging data shown in Fig. 8 suggest that the sandy silt
layer between 12 m and 16 m, having low strength, may be
disturbed by strong ground motion; thus this layer may not offer
3. Application of pile foundation stability analysis due to much fixity to the pile.
buckling mechanism The nearby ground motion recording was obtained at the Higashi
Kobe Bridge, which is away about 0.9 km from the building site;
3.1. Case study distance to the rupture plane for the observation and building sites is
about 5.2 and 4.4 km, respectively. The site condition at the
Several buildings with pile foundation had to be demolished
due to severe liquefaction damage during the 1995 Mw6.9 Kobe
earthquake. Because of the dramatic consequences, several detailed
post-earthquake investigations were conducted to examine the
cause and failure mechanism of these cases [37]. For illustration,
one of such case studies [5,38] is focused upon. A 5-story reinforced
concrete frame building (total height HB ¼14.5 m) was located at
6 m from the quay wall on a reclaimed fill in the Higashi-Nada area
of the Kobe City; the distance from the rupture plane to the building
site was about 5 km (Fig. 6). The Kobe earthquake caused a lateral
displacement of 2 m to the quay wall towards the sea, and the
building was tilted by 31 due to lateral spreading. The schematics of
the post-earthquake investigation of the building and pile founda-
tion are shown in Fig. 7. At the building site, significant lateral
spreading was observed (about 1.0–1.5 m deformation/movement
of the ground towards the sea [37]).
The building was supported on 38 hollow pre-stressed concrete
piles (there were two pile rows separated by 7.5 m and 19 piles
were aligned in each row); the pile length was 20 m with exterior
and interior diameters of 0.4 and 0.24 m, respectively. Fig. 8 shows
variations of soil profiles and SPT N counts with depth at the
building site. The site has fill/sand layers with relatively low N
counts (e.g. 2–9 m and 12–16 m), which are susceptible to liquefac-
tion (i.e. saturated sand layers with low strength); the average Fig. 7. Post-earthquake investigation of a case study [37].
Fig. 9. Ground motion time history (a) and 5%-damped response spectra (b) at the Higashi Kobe Bridge observation site.
observation site is similar to that at the building site (typically, ground failure may be expected. Further to note, geographical
NEHRP site class D or E). The recorded acceleration time-histories at positions of the fault plane, hypocenter, and the observation
the observation site are shown in Fig. 9 [32]. The 5%-damped site (Fig. 6) are of typical ‘‘near-fault motions’’ due to forward
response spectra of the two horizontal components and their geo- directivity [28]; this can be corroborated by large response
metric mean are presented in Fig. 9, and are compared with the spectra values at long periods (Fig. 9b) and by inspecting
median GMPE by Zhao et al. [41]; this relation is used as a velocity time histories of the acceleration data where large
representative regional model throughout this study to estimate velocity pulses are clearly visible.
ground motion parameters at the building site. The comparison of In this study, key input parameters are treated as random
the calculated response spectra with the Zhao et al. relation indicates variables, rather than deterministic ones. They include: seismologi-
that the observed response spectra have less spectral content at cal parameters (Mw and PGA), soil parameters (water table level, FC,
vibration periods less than 1.0 s, while they contain rich spectral N, sv, and s0 v), and pile/structural parameters (Tpre, EI, and Pstatic).
content at vibration periods greater than 1.0 s. The responses at short The probabilistic information (i.e. mean, coefficient of variation,
vibration periods are likely to be affected by non-linear site ampli- lower and upper limits, and probability distribution type) of these
fication (i.e. de-amplification), and at such a site, liquefaction-induced parameters is summarized in Table 2. The assumed variability of
Author's personal copy
Table 2
Summary of probabilistic information of input parameters.
Parameter Mean Coefficient of variation [Lower & upper limits] Distribution type
a
This is the standard deviation.
b
It depends on the water table.
c
Dry and wet soil densities are set to 1.76 and 1.92 g/cm3.
Mw is adopted from the PEER-NGA database (http://peer.berke- To show how some of the key output parameters are distributed,
ley.edu/nga/), and the uncertainty of PGA is consistent with a probability mass functions of DL, Tpost, a, and HC are presented in
regional ground motion prediction equation [13,41]. The uncertain- Fig. 10 (based on 10,000 simulation runs). It is observed that DL have
ties of FC, N, sv, and s0 v are based on Juang et al. [22], while those of two modes, a major mode at 16 m and a minor mode around 7–9 m;
Tpre, EI, and Pstatic, listed in Table 2, are assumed to be adequate for the major mode, the intermediate sand layer (between 9 m and
(note: uncertainties of these three parameters have minor effects on 12 m in Fig. 8) is regarded as ‘‘unsupportive’’, whereas for the minor
the final results). The values of f, K, and Dfix are set to 0.35, 1.0, and mode, the intermediate layer becomes sufficiently thick to provide
2.0 (m), respectively. The coefficients b1, b2, and b3 are taken as 0.85, lateral support to the pile. This bi-modal behavior of DL (Fig. 10a)
0.67, and 0.5, respectively. The damping ratios xpre and xpost are set affects the probabilistic characteristics of Tpost, as shown in Fig. 10b; if
to 0.05 and 0.20, respectively; from preliminary analysis results, the the unsupported pile length is relatively short, Tpost falls within 0.5 s
assessment is not sensitive to these parameters. and 2.0 s, while if it is long, Tpost can be much greater. The probability
mass function of a (Fig. 10c) indicates that the impact of dynamic
axial forces can be significant (mean is about 2.4), noting that it is
3.2. Pile foundation stability due to buckling: Base case
partly influenced by large long-period ground motions at the
observation and building sites (i.e. site amplification and forward
To assess the unsupported pile length DL, thickness of liquefied
directivity). Thus it is important to take response spectral shape
soils needs to be evaluated. First, the mean and covariance of the
effects into account when assessing ground motion parameters
random error ebldg at the building site are evaluated using
(beyond PGA) at the building site. Finally, inspection of the prob-
Eqs. (3)–(5) based on the observed PGA of 0.444g at the recording
ability mass function of HC shown in Fig. 10d indicates that HC is
site (note: eobs rec ¼ 0.163, indicating that the Zhao et al. relation distributed around 8–10 m (with a single mode). The pair-wise
overestimates the observed PGA; see Fig. 9). The correlation
comparison of DL and HC (Figs. 10a and d) leads to the probability
coefficient r2(D) and total correlation COVerec,ebldg(D)/s2 are about
of pile foundation failure in buckling of about 0.895, which is
0.751 and 0.826, respectively, and the corresponding values of
relatively high. This result suggests that, if all assumed input
mebldg jerec ¼ eobs
rec
and s2e je ¼ eobs are calculated as 0.134 and 0.166, parameters and adopted assessment criteria (particularly for DL)
bldg rec rec
respectively (note: e is in natural logarithm). Using these condi- are reasonable and adequate, it is highly likely that the pile
tional statistics of PGAbldg, a simulation-based liquefaction poten- foundation for the case study fails in buckling. However, this result
tial assessment using Eqs. (6)–(9) can be carried out [15]. is sensitive to the input information and criteria, and thus calcula-
A typical outcome from such assessment is the profile of liquefied tions with different input parameters and criteria are conducted in
and non-liquefied soil layers. Then, the next critical step is to the following section.
determine the unsupported pile length DL based on a realized
liquefaction profile. 3.3. Sensitivity analysis
For the base case, it is considered that the non-liquefied soil
layers are unrestrictive, if thickness of non-liquefied soils is less To investigate the impact of key model parameters on the
than or equal to 2 m (i.e. 5d, where d is the diameter of a pile). On assessment, sensitivity analysis is conducted. The varied para-
the other hand, a non-liquefied soil layer of thickness of 3 m (i.e. meters are: (i) the reduction factor for the total sigma of a GMPE
7.5d) or more is considered to exert lateral support to a pile. It is (base case is set to 0.5), (ii) the criteria to determine DL based on
noteworthy that the assessment results are affected significantly liquefied soil profile (base case is set to a limiting thickness of
by the definitions/criteria adopted for determining DL. The set-up non-liquefied soil layers for lateral support of a pile equal to 7.5d
considered herein is also constrained by the interval of available N or 3 m), (iii) pile stiffness (EI), and (iv) the buckling amplification
values over depth (i.e. 1 m interval). factor f (base case is set to 0.35). The observations obtained from
Given a value of DL and other random parameters (e.g. EI and the sensitivity analyses can be summarized as follows:
Pstatic), values of Tpost and Ppost can be evaluated using Eqs. (18)
and (19). Then, the dynamic amplification factor a (Eq. (20)) and The increased sigma value leads to a greater probability of pile
thus the critical pile length HC (Eq. (16)) can be obtained (note: for foundation failure in buckling. Specifically, the reduction
the calculation of a, SA(Tpre,xpre) and SA(Tpost,xpost) are calculated factor of 0.4, 0.6, 0.8, and 1.0 results in the probability of
from the corresponding conditional probability distributions). buckling failure of 0.910, 0.879, 0.845, and 0.811, respectively.
Using a Monte Carlo technique, the above step is repeated many Note that the average PGA for the considered scenario is
times to generate sufficient samples of ‘‘pile foundation failure in relatively large and is likely to cause severe liquefaction at
buckling mechanism’’ (i.e. HC oDL). the building site. Therefore, large uncertainty increases the
Author's personal copy
Fig. 10. Probability mass functions of model parameters: (a) unsupported pile length DL, (b) post-liquefaction vibration period Tpost, (c) dynamic amplification factor a, and
(d) critical pile depth HC.
Fig. 11. Comparison of probability mass functions for DL and HC: (a) DL is determined based on the assumption that 5d (2 m) thickness of non-liquefied layers can provide
lateral support to a pile and (b) DL is determined based on the assumption that 10d (4 m) thickness of non-liquefied layers can provide lateral support to a pile.
chance that the PGA at the building site becomes smaller such liquefied layers is needed, the probability of buckling failure is
that no significant liquefaction profile is developed. increased to 0.943. The dramatic change of the probability of
The thickness of non-liquefied soil layers that is sufficient buckling failure is due to the significant change of the
to provide lateral support to a pile has significant impact characteristics of DL. To show this clearly, comparisons of the
(as expected). If 5d or 2 m thickness of non-liquefied layers is probability mass functions for DL and HC for the two cases are
considered to be applicable, the probability of buckling failure presented in Fig. 11. The change of the modal concentrations
is decreased to 0.572, whereas if 10d or 4 m thickness of non- for DL (i.e. from 9 m to 16 m) depends on the soil profile at the
Author's personal copy
by embedding the pile to a depth equivalent to 3–5 times the pile [23] Kanno T, Narita A, Morikawa N, Fujiwara H, Fukushima Y. A new attenuation
diameter. relation for strong ground motion in Japan based on recorded data. Bulletin
of the Seismological Society of America 2006;96:879–97.
[24] Kimura Y, Tokimatsu K. Buckling stress of slender pile with lateral displace-
References ment at the pile head in liquefied soils. Journal of Structural and Construction
Engineering, AIJ 2007;617:169–75.
[25] Ledezma C, Bray JD. Probabilistic performance-based procedure to evaluate
[1] Anderson AW, Blume JA, Degenkolb HJ, Hammill HB, Knapik EM, Marchand
pile foundations at sites with liquefaction-induced lateral displacement.
HL, et al. Lateral forces of earthquake and wind. Transactions of the ASCE
1952;117:716–80. Journal of Geotechnical and Geoenvironmental Engineering 2010;136:
[2] Bhattacharya S. Pile instability during earthquake liquefaction. PhD thesis. 464–76.
U.K: University of Cambridge; 2003. [26] Lin PS, Chiou B, Abrahamson N, Walling M, Lee CT, Cheng CT. Repeatable
[3] Bhattacharya S, Madabhushi SPG, Bolton MD. An alternative mechanism of source, site, and path effects on the standard deviation for empirical ground-
pile failure in liquefiable deposits during earthquakes. Geotechnique motion prediction models. Bulletin of the Seismological Society of America
2004;54:203–13. 2011;101:2281–95.
[4] Bhattacharya S, Madabhushi SPG, Bolton MD. Reply to the two discussions on [27] Madabhushi SPG, Knappett J, Haigh SK. Pile design in liquefiable soils.
the paper ‘‘An alternative mechanism of pile failure in liquefiable deposits Imperial College Press; 2010.
during earthquakes’’. Geotechnique 2005;3(55):259–63. [28] Mavroeidis GP, Papageorgiou AS. A mathematical representation of near-
[5] Bhattacharya S. Safety assessment of existing piled foundations in liquefiable fault ground motions. Bulletin of the Seismological Society of America
soils against buckling instability. ISET Journal of Earthquake Technology 2003;93:1099–131.
2006;43:133–47. [29] Morikawa N, Kanno T, Narita A, Fujiwara H, Okumura T, Fukushima Y, et al.
[6] Bhattacharya S, Madabhushi SPG. A critical review of methods for pile design Strong motion uncertainty determined from observed records by dense
in seismically liquefiable soils. Bulletin of Earthquake Engineering 2008;6: network in Japan. Journal of Seismology 2008;12:529–46.
407–46. [30] Moss RES, Seed RB, Kayen RE, Stewart JP, Der Kiureghian A, Cetin KO. CPT-
[7] Bhattacharya S, Adhikari S, Alexander NA. A simplified method for unified based probabilistic and deterministic assessment of in situ seismic soil
buckling and free vibration analysis of pile-supported structures in seismi- liquefaction potential. Journal of Geotechnical and Geoenvironmental Engi-
cally liquefiable soils. Soil Dynamics and Earthquake Engineering 2009;29: neering 2006;132:1032–51.
1220–35. [31] NEHRP. National Earthquake Hazards Reduction Program (NEHRP, 2000):
[8] CEN. Eurocode 8, Design of structures for earthquake resistance—Part 1:
commentary for Federal Emergency Management Agency (FEMA, USA 369)
general rules, seismic actions and rules for buildings. EN 1998-1:2004.
on seismic regulations for new buildings and other structures; 2000.
Brussels, Belgium: Comite Europeen de Normalisation; 2004.
[32] Public Works Research Institute. Strong-motion acceleration records from
[9] Cetin KO, Der Kiureghian A, Seed RB. Probabilistic model for the initiation of
Public Works in Japan (no. 21). Tsukuba, Japan: Ministry of Construction;
seismic soil liquefaction. Structural Safety 2002;24:67–82.
[10] Cetin KO, Seed RB, Der Kiureghian A, Tokimatsu K, Harder Jr. LF, Kayen RE, 1995.
et al. Standard Penetration Test-based probabilistic and deterministic [33] Rankine WJM. Useful rules and tables. London; 1866.
assessment of seismic soil liquefaction potential. Journal of Geotechnical [34] Seed HB, Idriss IM. Simplified procedure for evaluating soil liquefaction
and Geoenvironmental Engineering 2004;130:1314–40. potential. Journal of the Soil Mechanics and Foundations Division 1971;97:
[11] Dash SR, Bhattacharya S, Blakeborough A. Bending-buckling interaction as a 1249–73.
failure mechanism of piles in liquefiable soils. Soil Dynamics and Earthquake [35] Shanker K, Basudhar PK, Patra NR. Buckling of piles under liquefied soil
Engineering 2010;30:32–9. conditions. Geotechnical and Geological Engineering 2007;25:303–13.
[12] Goda K, Hong HP. Spatial correlation of peak ground motions and response [36] Timoshenko SP, Gere JM. Theory of elastic stability. New York: McGraw-Hill
spectra. Bulletin of the Seismological Society of America 2008;98:354–65. Book Company; 1961.
[13] Goda K, Atkinson GM. Probabilistic characterization of spatially correlated [37] Tokimatsu K, Ohoka H, Shamoto Y, Asaka Y. Failure and deformation modes
response spectra for earthquakes in Japan. Bulletin of the Seismological of piles due to liquefaction-induced lateral spreading in 1995 Hyogoken-
Society of America 2009;99:3003–20. Nambu earthquake. Journal of Structural and Construction Engineering, AIJ
[14] Goda K, Atkinson GM. Intra-event spatial correlation of ground-motion 1997;495:95–100.
parameters using SK-net data. Bulletin of the Seismological Society of [38] Uzuoka R, Sento N, Yashima A, Zhang F. 3-dimensional effective stress
America 2010;100:3055–67. analysis of a damaged group-pile foundation adjacent to a quay wall. Journal
[15] Goda K, Atkinson GM, Hunter JA, Crow C, Motazedian D. Probabilistic of Japan Association for Earthquake Engineering 2002;2:1–14.
liquefaction hazard analysis for four Canadian cities. Bulletin of the Seismo- [39] Valsamis AI, Bouckovalas GD, Chaloulos YK. Parametric analysis of single pile
logical Society of America 2011;101:190–201. response in laterally spreading ground. Soil Dynamics and Earthquake
[16] Horne, Merchant. The stability of frames. Pergamon; 1965. Engineering 2012;34:99–110.
[17] Idriss IM, Boulanger RW. Soil liquefaction during earthquakes. Oakland, CA: [40] Youd TL, Idriss IM, Andrus RD, Arango I, Castro G, Christian JT, et al.
Earthquake Engineering Research Institute; 2008. Liquefaction resistance of soils: summary report from the 1996 NCEER and
[18] IS 1893. Indian Standard for Seismic Design; 2001.
1998 NCEER/NSF workshops on evaluation of liquefaction resistance of soils.
[19] Kramer SL. Geotechnical earthquake engineering. Upper Saddle River, NJ:
Journal of Geotechnical and Geoenvironmental Engineering 2001;127:
Prentice-Hall; 1996.
817–33.
[20] Kramer SL, Elgamal AW. Modeling soil liquefaction hazards for performance-
[41] Zhao JX, Zhang J, Asano A, Ohno Y, Oouchi T, Takahashi T, et al. Attenuation
based earthquake engineering. PEER Report 2001/13. Berkeley, CA: Pacific
Earthquake Engineering Research Center; 2001. relations of strong ground motion in Japan using site classification based on
[21] Lin SS, Tseng Y, Chiang CC, Hung CL. Damage of piles caused by lateral predominant period. Bulletin of the Seismological Society of America
spreading—back study of three cases. In: Boulanger, Tokimatsu, editors. ASCE 2006;96:898–913.
Geotechnical Special Publication no. 145 on seismic performance and [42] API. Recommended Practice for planning designing and constructing Fixed
simulation of pile foundations in liquefied and laterally spreading ground; Offshore Platforms—Working Stress Design. USA: American Petroleum Insti-
2005. p. 121–33. tute; 2000.
[22] Juang CH, Yang SH, Yuan H. Model uncertainty of shear wave velocity-based [43] Davisson MT, Robinson KE. Bending and buckling of partially embedded
method for liquefaction potential evaluation. Journal of Geotechnical and piles. In: Proceedings of the 6th ICDMFE, vol. 2; 1965. p. 243–6.
Geoenvironmental Engineering 2005;131:1274–82.