Sunteți pe pagina 1din 127

guidelines for

design of dams
for earthquake
-lu #_,
AUSTRALIAN NATIONAL
COMMITTEE ON
LARGE DAMS
GUIDELINES FOR
DESIGN OF DAMS
FOR EARTHQUAKE
AUGUST 1998
AUSTRALIAN NATIONAL
COMMITTEE ON
LARGE DAMS
AUSTRALIAN NATIONAL COMMITTEE ON LARGE DAMS
GUIDELINES FOR DESIGN OF
DAMS FOR EARTHQUAKE
AUGUST 1998
IMPORTANT DISCLAIMER
"ANCOLD and its Members,and the Convenor,Members and Assistants of the
Working Group which developed these Guidelines do not accept responsibility
for the consequences of any action taken or omitted to be taken by any person,
whether a purchaser of this publication or not,as a consequence of anything
contained in or omitted from this publication. No persons should act on the basis
of anything contained in this publication without taking appropriate professional
advice in relation to the particular circumstances".
ANCOLD Guidelines for Design of Dams for Earthquake
TABLE OF CONTENTS
Page No.
FOREWORD
ANCOLD WORKING GROUP MEMBERSHIP LIST
INTRODUCTION 1
EARTHQUAKES AND THEIR CHARACTERISTICS 2
2.1 Earthquake Mechanisms and Terminology 2
2.2Earthquake Ground Motion2
2.3Surface Rupture3
2.4Magnitude and Intensity 3
2.5Changes to Seismic Waves Near the Ground Surface4
2.6Attenuation and Amplification of Ground Motion4
2.7Reservoir Induced Seismicity 5
EARTHQUAKE HAZARD IN AUSTRALIA 6
3.1 General 6
3.2Mechanism of Earthquakes8
3.3Earthquake Depths9
3.4Evaluation of Seismic Hazard10
3.5Attenuation11
3.6Maximum Credible Earthquake Magnitude11
3.7Estimates of Ground Motion and Response Spectra at a Site12
3.8 Earthquake Hazard Maps12
SELECTION OF DESIGN EARTHQUAKE 16
4.1 Definitions16
4.2Selection of the Design Earthquake20
4.3Selection of the Operating Basis Earthquake (OBE)32
4.4Concurrent Load Combinations32
4.5Earthquakes Induced by the Reservoir33
4.6Response Spectra and Accelerograms *33
DESIGN OF EMBANKMENT DAMS AND ANALYSIS OF
/LIQUEFACTION 33
5.1 Effect of Earthquake on Embankment Dams33
5.2General ("Defensive") Design Principles for Embankment Dams34
5.3Liquefaction of Dam Embankments and Foundations36
ANCOLD Guidelines for Design ofDams for Earthquake
6. SEISMIC STABILITY ANALYSIS OF EMBANKMENTS 57
6.1 Preamble57
6.2Pseudo-Static Analysis57
6.3Simplified Methods of Deformation Analysis59
6.4Post Liquefaction Stability and Deformation Analysis63
6.5Numerical Methods65
6.6Proposed Guidelines67
7. ANALYSIS AND DESIGN OF CONCRETE DAMS 69
7.1 Past Performance of Concrete Dams in Earthquakes69
7.2Defensive Design Measures70
7.3Analysis Methods71
7.4Design Earthquake and Hydrodynamic Loads82
7.5Design Criteria83
7.6Dynamic Material Properties86
8. APPURTENANT STRUCTURES 87
8.1 Introduction87
8.2Performance Requirements87
8.3Intake Towers89
REFERENCES
APPENDIX A
TERMS OF REFERENCE
APPENDIX B
TYPICAL EASTERN AUSTRALIAN PEAK
GROUND ACCELERATION VS AEP
RESPONSE SPECTRUM FOR 1 in 1000AEP
MODIFIED MERCALLI SCALE
APPENDIX C
EXTRACTS FROM CANADIAN DAM SAFETY
GUIDELINES
APPENDIX D
ADDITIONAL INFORMATION ON ACCEPTABLE
RISKS
ANCOLD Guidelines for Design ofDams for Earthquake
FOREWORD
Even in the matter of earthquakes,Australia can be considered the "Lucky Country" in not being on the
edge of major tectonic plates. Neighbouring countries like New Zealand and Indonesia are renowned for
their volcanoes and frequent earthquakes. Australia is relatively earthquake free by comparison and
earthquakes were seldom considered in early dam designs.
Certainly there were some zones of known activity such as the Adelaide Hills and the Western
Australian wheat belt,and whilst major damage had occurred,it was not on the same scale as in other
countries.
In 1979,the Standards Association of Australia produced the "Earthquake Code" AS2121. It showed
zones of seismic activity and recommended methods of determining loads on building structures. The
development of this code was based largely on statistics of historic earthquakes,for which there were
relatively short term records.
However,several major earthquakes subsequently occurred in areas indicated by the code as having
negligible earthquake risk,the most notable being the 1989 earthquake at Newcastle (Magnitude 5.6) in
which 12 people died and the Tennant Creek Earthquake in 1988 (Magnitude 6.8). This led to the
introduction of a new earthquake code (AS1170.4-1993)which included data from more widespread and
reliable seismographs and furthermore considered the all important geological situations.
In parallel with these developments,analytical methods used by dam engineers were improving beyond
the simplistic application of a horizontal force equating to seismic acceleration. Improvements were
based on the observed fact that earth dams subjected to earthquakes had slumped vertically rather than
fail by slipping of a face as indicated by the simplistic analyses.
Methods of analysing slumping were developed,and further supplemented by sophisticated finite
element analyses which,by utilising modem computer power,give an ability to undertake rigorous
analyses of dams where necessary.
This ANCOLD Guideline brings together improved appraisals of the earthquake loadings that a dam
may suffer and then describes appropriate methods for analysis and evaluation. Whilst specific to the
Australian considerations,the majority of this guideline could be applied to dam structures throughout
the world. The ICOLD Bulletins No. 46(1983),No. 52(1986),No. 62(1988) and No. 72(1989) are
parallel documents in this regard,although not including recent advances.
This guideline is a major contribution to dam engineering and the voluntary work by the ANCOLD
subcommittee has unselfishly provided their experience to the dam building community and indeed the
wider community. Our appreciation goes to Prof Fell and his team for producing this valuable guideline.
This guideline is not a design code,and dam designers must continue to apply their own considerations,
judgements and professional skills when designing dams to resist earthquakes. As time goes on there
will no doubt be improved data and design tools to help the designer and it is intended that this guideline
will be updated as circumstances dictate. ANCOLD welcomes contributions to discussion on this
guideline which will assist with future revisions.
/
JOHN PHILLIPS
Chairman,ANCOLD
ANCOLD Guidelines for Design of Dams for Earthquake
MEMBERSHIP OF THE ANCOLD WORKING GROUP FOR
GUIDELINES FOR THE DESIGN OF DAMS FOR EARTHQUAKE
4
Robin Fell School of Civil Engineering,University of New South Wales
Gamini Adikari Snowy Mountains Engineering Corporation,Victoria
John Bosler Snowy Mountains Engineering Corporation,Cooma,New South Wales
Brian Cooper Dams and Civil Section,Public Works and Services Department,NSW
Peter Foster Works Consultancy Services,Power Engineering,New Zealand
Gary Gibson Seismology Research Centre,RMIT,Melbourne
Sergio Giudici Hydro-Electric Commission,Hobart
Nasser Khalili School of Civil Engineering,University of New South Wales
Ian Landon-Jones
Dams Safety Group,Sydney Water,New South Wales
Kevin McCue Australian Seismological Centre,Canberra
Len McDonald Dams and Civil Section,Public Works and Services Department,NSW
Brian Shannon
Water Resources,Department of Primary Industries,Queensland
David Stapledon Geotechnical Consultant,Adelaide,South Australia
John Waters
Geo-Eng Pty Ltd,Perth,Western Australia
Ron Wyburn Halcrow Water Power,Victoria
ANCOLD Guidelines for Design of Dams for Earthquake
1. INTRODUCTION
Public awareness of the potential for damage
and loss of life in Australia from earthquakes
was highlighted by the Newcastle earthquake in
December 1989. This was a Magnitude 5.6
(M5.6) event,and because of its proximity to
Newcastle,and local ground conditions,caused
approximately $1 billion damage.
Dam engineers in Australia have been
conscious of earthquakes for many years,but it
was the earthquakes at Tennant Creek in 1988,
which were M6.3,M6.4,M6.7,with a total fault
scarp length of 32km which raised the question
most acutely as to whether dams in Australia
could be subject to large earthquakes,and if so,
could they withstand them without resultant
loss of the facilities and lives,property,and
environmental values downstream. Other large
earthquakes in the M6 to M7range had
occurred in Australia,the most notable being in
Meckering in 1968 (M6.9),but the Tennant
Creek event was critical because it occurred in
an area which had previously been regarded as
virtually free of earthquakes.
Recent assessments of earthquake ground
motions for some large Australian dams have
been based on the assumption that the
maximum credible earthquake is M7.5,which is
large by any standards. The seismologists
involved in these studies indicate that on the
available evidence,such earthquakes,ie. M7.5,
could occur anywhere in Australia.
In general,it is not possible to identify active
faults which might cause such earthquakes. For
example,there had been no movement on the
Tennant Creek fault for more than 200,000
years (Crone and Machette,1992),so the
question arises,can it occur at,or close to any
damsite? Peak ground accelerations close to a
M7.5 earthquake can be very high.
The past performance of dams in earthquake
h&s been very good,with few dams suffering
major damage. Where this has occurred,it has
been due to liquefaction in the dam or the
foundation. Very few of these dams have
breached and released a flood wave. However,
several might have breached,if the reservoir
level had been higher at the time of the
earthquake. Seed (1979),USCOLD (1992),
ICOLD (1986),NSWDSC (1993) and Hinks
and Gosschalk(1993) give some details.
The approach taken by seismologists in
Australia is to use statistical analysis to predict
the frequencies of recurrence of ground
motions. This typically results in 1 in 1000
AEP peak ground accelerations of 0.15g,1 in
10,000AEP s0.35g,and 1 in 100,000AEP
~0.5g. These are large loadings and it is likely
that assessment of many of the existing dams in
Australia for such loads could indicate some
deficiencies to either the dam or appurtenant
structures. New dams would also need (less)
expensive additional design features to cope
with earthquake.
In recognition of the need to provide some
guidance to dam engineers and owners in
Australia,ANCOLD established a Working
Group to prepare Guidelines for the Design of
Dams for Earthquake. The Working Group was
established in September 1993,and took over
from an earlier ANCOLD Working Group
preparing Guidelines on Seismic Analysis and
Design of Embankment Dams. The Terms of
Reference for the Working Group are in
Appendix A.
These guidelines are to cover all types of dams,
including tailings dams,and apply to existing
and new dams. They cover the selection of the
design earthquake,analysis and design of
embankment and concrete dams,and
appurtenant structures.
The guidelines are not meant to be used as a
design code,and of necessity,do not include
complete details of all the analysis and design
methods which are recommended. The area is
rapidly evolving,and those involved in the
analysis and design of dams for earthquake
should refer to the references given,and to
more recent publications so as to be fully
informed. In some situations it will be
necessary to seek specialist advice.
ANCOLD Guidelines for Design ofDams for Earthquake
1
2. EARTHQUAKES AND
THEIR CHARACTERISTICS
2.1 Earthquake Mechanisms and
Terminology
An earthquake is the motion that is produced
when stress within the earth builds up over a
long period of time until it eventually exceeds
the strength of the rock,which then fails and a
break along a fault is produced. It may take
tens,hundreds or thousands of years for the
stress to build up in a particular area,and it is
then released in a few seconds. Part of the
energy is transmitted away as seismic waves
and part of it as heat.
The fault displacement in a particular
earthquake may vary from centimetres up to a
few metres in a great earthquake. Once
ruptured,the fault is a weakness which is more
likely to fail in future earthquakes,so a large
total displacement may build up from many
earthquakes over a long period of time. This
may eventually measure kilometres for thrust
faults produced by compression,or hundreds of
kilometres for horizontal strike-slip faults such
as the San Andreas.
The point on the fault surface where a
displacement commences is called the
hypocentre or focus,and the earthquake
epicentre is the point on the ground surface
vertically above the hypocentre. The
displacement usually propagates along the fault
in one direction from the hypocentre,but
sometimes it propagates in both directions.
Energy release is near but not exactly at the
hypocentre.
The hypocentral distance from an earthquake to
a point is the three dimensional slant distance
from the hypocentre to the point,while the
epicentral distance is the horizontal distance
from the epicentre to the point.
2.2Earthquake Ground Motion
Earthquake ground vibration is recorded by a
seismograph or a seismogram. Most modem
seismographs record three components ol
motion: east-west,north-south and vertical.
The rupture time for small earthquakes is a
fraction of a second,for earthquakes of
magnitude 5.0it is about a second,and for large
earthquakes may be up to tens of seconds.
However the radiated seismic waves travel al
different velocities,and are reflected and
refracted over many travel paths,so the total
duration of vibrations at a site persists longei
than the rupture time,and shows an exponentia
decay.
Several types of seismic wave are radiated from
an earthquake. Body waves travel in three
dimensions through the earth,while surface
waves travel over the two dimensional surface
like ripples on a pond. There are two types oi
body wave (P and S waves),and two types oi
surface wave (Rayleigh and Love waves).
Primary or P waves are ordinary sound waves
travelling through the earth. They are
compressional waves with particle motion
parallel to the direction of propagation.
Secondary or S waves are shear waves,with
particle motion at right angles to the direction of
propagation. The amplitude of S waves from an
earthquake is usually larger than that of the P
waves.
P waves travel through rock faster than S
waves,so they always arrive at a seismograph
before the S wave.
The frequency content of earthquake ground
motion covers a wide range of frequencies up to
a few tens of hertz(cycles per second). Most
engineering studies consider motion between
about 0.2 and 25 Hz.
The amplitude,duration and frequency content
of earthquake ground motion at a site depend on
many factors,including the magnitude of the
earthquake,the distance from the earthquake to
the site,and local site conditions.
The larger the earthquake magnitude,the
greater the amplitude (by definition a factor of
ten for each magnitude unit),the longer the
2
ANCOLD Guidelines for Design of Dams for Earthquake
duration of motion,and the greater the
proportion of seismic energy at lower
frequencies. A small earthquake has low
amplitude (unless it is very close),short
duration,and has only high frequencies.
The smaller the distance from an earthquake to
the site,the higher the amplitude. The duration
is not strongly affected by distance. High
frequencies are attenuated by absorption within
the ground more quickly than low frequencies,
so at greater distances the proportion of seismic
vibration energy at high frequencies will
decrease.
2.3Surface Rupture
Surface rupture is a relatively rare phenomenon
which occurs when a fault break reaches the
ground surface. It may produce a vertical or
horizontal offset (or both) with a displacement
of millimetres to a few metres,and a length
from metres to tens of kilometres.
Because rock near the surface is relatively
weak,few earthquake hypocentres occur in the
top one or two kilometres. It is common for
surface sedimentary rocks to be folded in
response to faulting at depth,giving a
monocline and scarp at the surface,but without
a surface fault.
Most earthquakes,especially most larger
earthquakes,occur on existing faults. This is
because faults are weaker than surrounding
unbroken rock,and are much more likely to fail
again when stress rebuilds.
A site will have surface rupture potential if an
existing fault is found which has been active in
the recent geological past (perhaps the past few
million years). This will be quite rare,and
possibly be difficult to establish. It will usually
be easier to show that a site with simple surface
geology has no faulting history,than to show
that a site with complex geology has suffered
recent faulting.
2.4Magnitude and Intensity
Earthquakes vary enormously in size. In 1935
Richter defined a magnitude scale to indicate
the size of an earthquake. For the Richter local
magnitude scale,ML,the logarithm of the peak
ground displacement is taken and an empirical
correction depending on the distance from
earthquake to seismograph is subtracted. The
resulting values are averaged for all the
seismographs that have recorded the
earthquake.
Other magnitude scales have been defined,
including moment magnitude,and while not
exactly the same as the Richter local magnitudes,
they give similar values that can range from 0.0
to over 9.0. For each unit of magnitude there is
a tenfold increase in ground displacement,and a
thirtyfold increase in seismic energy release.
Another measure of earthquake size is the fault
area,or the area of the fault surface which is
ruptured. The fault area ruptured in an
earthquake depends on the magnitude and stress
drop in the earthquake. For a given magnitude,
a higher stress drop will give a smaller rupture
area. Typically,a magnitude 4.0earthquake
ruptures a fault area of about 1 square
kilometre,magnitude 5.0about 10square
kilometres,and magnitude 6.0about 100square
kilometres (perhaps 10by 10kilometres).
Earthquake Intensity is a measure of the effect
of the seismic waves at the surface,and is
normally given on the Modified Mercalli
Intensity scale,a copy of which is attached in
Appendix B. This is an arbitrary scale defined
by the effects observed (whether sleeping
people were woken,trees shaken,etc) and on
the amount of damage caused. Normally the
maximum intensity occurs near the epicentre of
the earthquake,and intensity then decreases
with distance. However,this may be affected
by the orientation of the earthquake rupture,or
by local ground conditions such as topography
or surface sediments.
The earthquake recurrence or seismicity
(seismic activity) of an area must take the range
of earthquake sizes into account. There are
many more small earthquakes than large. In
ANCOLD Guidelines for Design of Dams for Earthquake
3
most places around the earth there are about ten
times as many earthquakes exceeding
magnitude 3.0than there are exceeding
magnitude 4.0,and ten times as many again
exceeding magnitude 2.0. In seismicity studies,
the logarithm of this factor is called the b value,
so a value of 1.0is typical. The b value may be
1.3or higher if there are many small
earthquakes,or 0.7or lower if there are few
small earthquakes.
2.5 Changes to Seismic Waves Near
the Ground Surface
The energy in seismic waves depends upon
their amplitude and the physical properties of
the material through which they are passing.
When waves pass from high stiffness material
(eg. rock at depth) into lower stiffness material
(eg. near-surface rock,or sediments) they are
reflected towards the vertical and their
amplitude increases. Their amplitude also
increases as they approach the earth's (free)
surface,at which they are reflected. The nature
and extent of free surface amplification varies
with topography,even in fresh,strong rock.
Changes in soil thickness above an irregular
bedrock surface can give complex surface
amplification that varies with earthquake wave
duration.
Resonance in the surface sediments causes
amplification at particular frequencies,
especially at the natural frequency of the
sediments. This depends on the thickness and
elastic properties of the sediments. Earthquake
motion recorded on hard rock includes all
frequencies up to a value that depends on
magnitude,while that recorded on soft
sediments is usually dominated by the resonant
frequency.
In surface sediments,high frequency vibrations
are attenuated much more with distance than
low frequencies. If sediments are very thick,
much of the high frequency motion will be lost
and peak surface accelerations will be low,even
if resonance has amplified motion at the low
resonant frequency.
2.6 Attenuation and Amplification of
Ground Motion
Earthquake ground motion attenuates with
increasing distance from the source due to
radiation and hysteretic damping. High
frequency motion is attenuated more quickly
with distance than lower frequency motion.
For estimates of peak ground acceleration,
attenuation is allowed for by using an
attenuation function of the form
a =b,eb2Mir*3
where a =acceleration
R =focal distance
M =Magnitude
b^jbj are constants,which
vary considerably over the
world.
Some earthquake hazard studies use the Esteva
and Rosenblueth (1969) attenuation functions,
which give peak ground velocity (mm/s),peak
ground acceleration (mm/s2) and Modified
Mercalli Intensity (IMM) at an epicentral distance
x kilometres from an earthquake at depth z
kilometres with local magnitude M. The
equations are:
R
=
Vx2 +z2 +400
Vpeak
= 160e10M R"17
Speak
= 20000e08 mR20
Imm
=
loge(2980e15MR-23)
Because of the 400term in the expression for R,
corresponding to a minimum R of 20
kilometres,these equations give low values of
ground motion at distances closer than a few
kilometres.
These relations were determined using
Califomian data,and should only be used with
magnitudes determined using a compatible
function. If the magnitudes computed for
seismographs at different distances vary,then
the attenuation function is invalid for the area.
4
ANCOLD Guidelines for Design ofDams for Earthquake
In selecting attenuation relationships care is
needed,and attention paid to the mechanism of
the source earthquake,eg. whether shallow
intraplate,or deep crustal boundary
earthquakes.
Weak surface materials absorb seismic energy
rather than transmit it unchanged,thus tending
to reduce amplitudes at the surface. The
amount of attenuation depends on the properties
of the materials,and especially on their
thickness.
Near-surface layers will vibrate preferentially at
their own natural frequencies,depending on
their thickness and elastic properties. The
earthquake motion at the natural frequencies of
the near-surface layers is amplified,while
motion at other frequencies may be little
affected or even attenuated. The amplification
effect can be especially pronounced for deep
soft sediments such as those underlying Mexico
City,but in deep,stiff sediments subject to high
frequency earthquake,attenuation may result.
Dams (like all other structures) have natural
frequencies of their own depending on their
mass and stiffness,usually in the range from
about 0.5 hertzto about 5 hertzfor embankment
dams and 2 hertzto 20hertzfor concrete
gravity dams.
2.7Reservoir Induced Seismicity
Reservoirs may induce seismicity by two
mechanisms. Either the weight of the water
may change the stress field under the reservoir,
or the increased ground water pore pressure
may decrease the stress required to cause an
earthquake. In either case,reservoir induced
seismicity (RIS) will only occur if relatively
high stresses already exist in the area. If the
stress has been relieved by a recent large
earthquake,say in the last few hundred years
for low seismicity areas like Australia,then RIS
is unlikely to occur.
/
RIS events initially usually occur at shallow
depth under or immediately alongside a
reservoir. As years pass after first filling,and
groundwater pore pressure increases permeate
to greater depths and distances,the events may
occur further from the reservoir. This occurs at
a rate of something like one kilometre per year.
RIS is experienced under new reservoirs,
usually starting within a few months or years of
commencement of filling,and usually not
lasting for more than about twenty years. Once
the stress field and the pore pressure fields
under a reservoir have stabilised,then the
probability of future earthquakes reverts to a
value similar to that which would have existed
if the reservoir had not been built. Most of the
earthquake energy does not come from the
reservoir,but from normal tectonic processes.
The reservoir simply acts as a trigger.
In areas with horizontal tectonic compression
and reverse faulting,like Australia,filling a
reservoir should increase the vertical minimum
principal stress and reduce the chance of an
earthquake under the reservoir. This has been
called reservoir induced a seismicity. However,
in some cases earthquakes could then be
induced by later releasing water from the
reservoir. Alternatively the change in stress
during filling could induce earthquakes beside
the reservoir rather than under it,although this
stress change is less pronounced.
It has been suggested that filling a reservoir will
cause compression under it,increasing the pore
pressure of the existing groundwater,and so
tend to induce earthquakes even in areas of
horizontal compression. Stress change induced
seismicity,either direct or through this indirect
mechanism,should occur soon after filling. It
may then cause seasonal variations in
seismicity,sometimes lagging a few weeks or
months behind water level.
Pore pressure induced seismicity is normally
delayed,and may occur years after filling. Pore
pressure increases always tend to induce events.
If there is a major fault near the reservoir,RIS
can produce earthquakes exceeding magnitude
6.0(Xinfengjiang,China,1962,M6.1; Koyna,
India,1967,M6.3). Such events will only occur
if the fault is already under high stress. A
number of Australian reservoirs have triggered
earthquakes exceeding magnitude 5.0
(Eucumbene,1959,M5.0; Warragamba,1973,
M5.0; Thomson,1996,M5.2).
ANCOLD Guidelines for Design of Dams for Earthquake
5
MPP*8-
It is more common for a reservoir to trigger a
large number of small shallow earthquakes,
especially if the underlying rock consists of
jointed crystalline rock like granite (Talbingo,
1973to 1975; Thomson,1986 to 1995). These
events possibly occur on joints rather than
established faults,so are limited in size,and
only give magnitudes up to 3or 4. There is no
hazard from such low magnitude reservoir
induced earthquakes,even if they occur
regularly. Their shallow depth means that they
may often be felt or heard.
RIS has been observed for over one hundred
reservoirs throughout the world,and small
shallow induced events have probably occurred
under many others. A relatively high
proportion of reservoirs with RIS seismograph
networks do record such activity. A high
proportion of RIS examples occur in intraplate
areas,with above average rates in China,
Australia,Africa and India.
It is not easy to predict whether a reservoir will
experience RIS because the stress and strength
at earthquake depths are not easily measured.
For the same reason,prediction of normal
tectonic earthquakes has been unsuccessful in
most parts of the world.
It seems that RIS with many small events is
more likely to occur in intraplate areas with
near surface crystalline rocks like granite,rather
than sedimentary rocks. A larger magnitude
RIS event can only occur if there is an existing
fault of sufficient dimension that is late in its
earthquake cycle (the stress is already
approaching the strength of the fault).
3. EARTHQUAKE HAZARD
IN AUSTRALIA
The understanding of the hazard imposed by
earthquakes in Australia is critical to selection
and application of design earthquakes for dams.
Hence,a relatively detailed discussion on the
topic follows. This is largely taken from Gibson
(1994).
3.1 General
The Australian continent is within a tector
plate shared with Southern India,so all of
earthquakes are intraplate. The pla
boundaries to the north and east are among t
most active on the earth. Possibly as a result
this,Australia is one of the most actr
intraplate areas on the earth. Despite this,tf
hazard is quite low when compared with acth
interplate areas.
Most people in Australia can expect to feel
earthquake about every five or ten year
although many of these may not be recognisf
as an earthquake. Most Australian earthquake
that are reported are heard with a noise like
distant quarry blast or thunder,with possibly
slight vibration being felt.
Only a proportion of earthquakes that are fel
perhaps about one in twenty,will cause som
damage in their epicentral area. If they occur i
an inhabited area,most earthquakes larger tha
about magnitude 4.0will cause some damage.
By contrast,in an active interplate area lik
New Britain or Bougainville in Papua Nev
Guinea,earthquakes are felt very often,c
average every week or two. These are normal
felt rather than heard,with any sounds being thf
reaction of a building to the vibration rathe
than the earthquake itself.
PNC
havf
A very small proportion of these
earthquakes,perhaps about 1 in 500,
caused any damage in their epicentral area,an<
earthquakes smaller than magnitude 6.0rarelj
cause damage.
There are a number of factors that influence the
estimation of earthquake hazard in Australia.
One is the short duration of documented history,
with a little over 200years of data about
Sydney and considerably less for most of the
continent.
Another factor is the large area of the country
relative to the size of the population.
Seismographs are distributed relatively
sparsely,limiting the accuracy of earthquakej
locations.
ANCOLD Guidelines for Design of Dams for Earthquake
As was the case over most of the earth,
seismograph coverage of local earthquakes in
Australia was only established in about 1960
following the International Geophysical Year.
There should be some link between population
and seismograph density with more
instrumentation in the populated southeast.
However,coverage is still far from complete.
The Australian National Seismograph Network,
operated by the Australian Seismological
Centre at Australian Geological Survey
Organisation (AGSO),aims to locate all
earthquakes in Australia larger than magnitude
ML3.0. Local seismograph networks have
non-uniform coverage,often with good
coverage of large dams and poor coverage of
rqajor cities.
Figure 1 shows the location of Australian
earthquakes with magnitude exceeding ML4in
the period 1850to 1993(note that many
earthquakes prior to 1960will not be shown).
Table 1. Some Large Australian Earthquakes (Gibson,1994).
No. Date Place Mag Imax
Damage,A$(1990) Where Given
1 1892-01-26 Flinders Island,TAS 6.9 7+ Offshore epicentre,little damage. Felt
throughout Tasmania and Eastern Victoria
2 1897-05-10 Beachport,SA 6.7 8 Damage in Kingston,Robe and Beachport,
liquefaction
31902-09-19 Warooka,SA 6.0 8 At Warooka chimneys fell,walls partially
demolished,few buildings without damage
4 1903-07-14 Warrnambool,VIC 5.37 Extensive minor damage and liquefaction at
Warrnambool. Followed similar event April
5 1918-06-06
Bundaberg,QLD 6.0 6 Epicentre offshore,minor damage in
Rockhampton
6 1941-04-29 Meeberrie,WA 7.2 8 Isolated area. Damage to remote farm
houses included cracked walls,burst tanks
7 1954-02-28 Adelaide,SA 5.4 8 Widespread minor damage. $71m.
Epicentral area was rural,but is now urban
8 1961-05-21 Robertson,NSW 5.6 7 Damage in the Moss Vale,Robertson and
Bowral area. $3.4m
9 1968-10-14
Meckering,WA 6.8 9 Most buildings in Meckering destroyed.
$29m. 32km surface rupture
10 1969-06-20 Boolarra,VIC 5.6 6 Cracked walls and fallen chimneys in
epicentral area
11 1973-03-09
Burragorang,NSW 5.0 6 Minor damage in Picton,Bowral and
Wollongong. $2.3m
12 1979-06-02 Cadoux,WA 6.2 9 Many buildings at Cadoux destroyed,most
others damaged. $9m. Only one injury.
Surface rupture extended over 15km
131986-03-30 Marryat Creek,SA 5.9 8 Epicentre in remote area. Cracked walls in
nearest homestead,13kmsurfacerupture
14 1988-01-22 Tennant Creek,NT 6.8 9 Epicentre in remote area. $1.2m,mainly to
damaged gas pipeline,35km surface rupture
15 1989-05-28 Uluru,NT 5.7 6 Minor damage at Uluru National Park
(Ayers Rock). Epicentre west of Mt Olga
/16 1989-12-27 Newcastle,NSW 5.6 8 Thirteen fatalities,$1500m + damage.
Widespread minor to moderate damage
17 1992-09-30 Amhem Land,NT 5.1 5 Epicentre offshore. No significant damage
ANCOLD Guidelines for Design ofDams for Earthquake
1
3.2 Mechanism of Earthquakes
Almost all Australian earthquakes have
mechanisms with the maximum principal stress
near horizontal. Most have the minimum
principal stress near vertical,producing revet
or thrust faults. There may be some strikes
movement,especially if the failure is on an {
fault that is oriented at an angle to the princij
stress.
113 120 123
s u
130
roi-
133 140 143 150
10
15
20
40
no <&> 00
o-Ofyo0
a a O O
0^ O
O009 &>
o
aO
q.
*6
90o
. Q
If
s>
O do
a
o
*
8 o
Q
<?
o 00
o
oCO
o
o
a!?
H
Australian Earthquakes to 1994 Magnitudes: *405 o
Figure 1. Australian earthquakes with magnitude exceeding ML4.0since 1850(Gibson,1994).
Earthquakes on reverse faults usually give a
high stress drop,where the seismic energy
comes from a small source volume. High stress
drop earthquakes radiate a greater than average
proportion of their energy in higher frequencies.
Higher frequency motion implies higher
accelerations for a given energy release or
earthquake magnitude,but not necessarily
fewer cycles,than for similar magnitude
earthquakes in,say,West Coast USA.
Compression giving reverse and thrust faul
produces surface uplift. Therefore,earthquakf
are most likely to occur in areas whei
mountains are developing,and less likely und<
large sedimentary basins. They may b
expected under the Eastern Highlands an
Flinders Ranges,but are less likely under th
Murray Basin,Great Artesian Basin and tl
Eucla Basin.
8
ANCOLD Guidelines for Design of Dams for Earthquake
It may be argued that dams are particularly
likely to be built near to an active fault. This is
because dams are built in valleys which have
recently been eroded. Erosion requires uplift,
and faulting is the geological process that gives
most pronounced uplift. In places where
horizontal compression produces reverse
faulting,the fault will inherently dip back under
the upthrown block,and under the dam.
Unlike at plate boundaries,there are no large
dominant active faults in Australia. Instead,
earthquakes are distributed over many smaller
faults. Most of these have poor surface outcrop
and have not yet been identified. Large
earthquakes,eg. Tennant Creek (1988),have
occurred on faults which were not previously
apparent on the surface. Crone and Machette
(1992) describe evidence that there had been no
movement on the fault at Tennant Creek for
200,000years.
Based on the available information,it would
appear that the differences between Australian
earthquake ground motion and those from
earthquakes in USA,China etc,on which many
of the dam design methods are based,are not
sufficiently great as to invalidate the design
methods. This is an aspect which will need to
be further assessed as more ground motion data
for Australian earthquakes is gathered.
3.3Earthquake Depths
Usually,earthquake depths can be precisely
determined only if the distance to the nearest
seismograph is not greater than about twice the
earthquake depth,or about 10kilometres. The
depths of most Australian earthquakes are
poorly constrained,because of their relatively
shallow depths and the low density of
seismographs.
If it is not possible to constrain an earthquake
depth using seismogram data,an arbitrary
"normal" depth may be used. The value of the
normal depth may or may not be realistic,and
typical values used in Australia include 0,5,10
or 33kilometres.
Table 1 shows some large Australian
earthquakes and a description of the damage.
Eastern Australian earthquakes are usually at
depths between 1 and 20km,while those in
Central and Western Australia may be a little
deeper. Australia's deepest known earthquake
occurred at 39km depth offshore from Arnhem
Land in 1992 (McCue and Michael-Leiba,
1993).
In Eastern Australia,earthquakes at depths of
less than 5km are regarded as shallow,and
greater than 15km are regarded as deep. The
Newcastle earthquake of December 1989 was at
a depth of about 12km,and a number of
earthquakes about Katoomba were at about 18m
depth.
All of these Australian earthquakes are very
shallow compared with those in plate boundary
areas,where earthquakes can occur as deep as
700km,and depths of less than 70km are
regarded as shallow,and greater than 300km as
deep.
It has been suggested that small earthquakes are
more likely to be shallow,while larger
earthquakes can occur at all depths.
Shallow earthquakes often have many
aftershocks,some with magnitudes approaching
that of the main shock. Deep earthquakes
usually have few aftershocks,and these are
often much smaller than the main shock.
Because of their shallow depth,small
Australian earthquakes are often felt or heard.
Magnitude 1.0events can be felt to a distance of
about one kilometre and magnitude 2.0to about
four kilometres. These are slant distances,so
only very shallow events of these magnitudes
will be felt.
Because the short travel distances do not give
very much attenuation of high frequency
vibrations,and because a high stress drop gives
a high proportion of high frequency vibration
from the earthquake source anyway (see
Section 3.7),these events have enough energy
hi the audio frequency range to allow them to
be heard. For many such small earthquakes the
sound heard is more significant to an observer
than the vibrations felt.
ANCOLD Guidelines for Design of Dams for Earthquake 9
; /
For similar reasons,moderate magnitude
earthquakes in Australia produce motion with
strong peak ground accelerations,and can
produce significant damage. The Newcastle
earthquake was only of magnitude MLS.6,but
caused extensive damage.
3.4Evaluation of Seismic Hazard
The estimation of ground motion requires the
following seismicity information about the
surrounding area:
the rate of occurrence of earthquakes
relative proportion of small to large events
(b value)
maximum earthquake size expected
(maximum credible magnitude)
the spatial distribution of earthquake
epicentres including delineation of faults.
The seismicity can be evaluated by a modified
Richter relation (Gibson,1994)
log,o(P) =-log10[ 10( 10"bM-10"bMmax] -log10(No)
P is the return period in years for an
earthquake of magnitude M or greater.
N0is the rate of occurrence of earthquakes,
given as the number of earthquakes of
magnitude zero or greater per year per
unit volume or per unit area. An area
of 100x 100kilometres is commonly
used. This must be converted to per
square kilometre or per cubic
kilometre for ground motion
calibrations.
b is the Richter b value,which gives the
relative number of small earthquakes
to large. It is the logarithm to the base
10of the ratio of the number of events
exceeding M+l to the number
exceeding M. A value of 1.0would
correspond to ten times as many
earthquakes exceeding magnitude M
as would exceed magnitude M+l.
Mmiix is the magnitude of the maximum
credible earthquake for the area. It is
the magnitude of an earthquake with
infinite return period. Because of t
low probability of very lar
earthquakes,Mmax does not critica
affect ground motion recurren
estimates for return periods up
hundreds of years,especially when t
b value is high. However,it is mo
important for low AEP events such
may be important for design of hi|
hazard dams. The maximum credit
magnitude causes the magnitin
recurrence plot to flatten out ai
asmyptote to that value.
The seismicity parameters N0and b
determined using available earthquake data
In most places there are insufficiei
earthquake data to determine Mmax fro
historical records,but values can be estimate
by considering the tectonic situation and loc,
fault dimensions. The b value is much moi
critical than N0,and a small adjustment to
will give a large change in N0.
Hazard evaluation depends on tfi
extrapolation of data from smaller earthquake
to larger magnitudes. The lower the b valui
the greater the resulting hazard estimate,
catalogue with missing small earthquakes wi
give an invalid low b value,and an estimate
hazard that will be too high. A catalogue wit
smaller quarry blasts incorrectly identified a
earthquakes will give a high b value,and ai
extrapolation that is non-conservative.
In Southeastern Australia the b value varie
from about 0.75 in Western Victoria to a littl
over 1.0south of Sydney,then down to 0.90ii
Northeast New South Wales. Tto
corresponding N0value is about 0.002 event
larger than MLO.O per cubic kilometre per year,
These methods are commonly used b;
seismologists to estimate design earthquakes ii
Australia. Their advice is that they an
reasonably reliable,when combined witl
suitable attenuation functions,and can be use<
to estimate ground motion for low AEP events-
10
ANCOLD Guidelines for Design of Dams for Earthquake
3.5Attenuation
The accelerograph and seismograph data
currently available are not sufficient to allow
determination of good local attenuation
functions in Australia,especially spectral
attenuation functions.
If a valid attenuation function is being used,
magnitudes computed from seismograph data
will not vary with distance. When the Richter
attenuation function from California is used in
Southeast Australia there is no significant
variation with distance. If it is used in Central
or Western Australia,magnitudes increase with
distance. This means that attenuation in
Southeast Australia is similar to that in
California (a little above the world average),
while in Central and Western Australia it is well
below the world average.
It follows that the Esteva and Rosenblueth
equation is likely to be reasonable for
estimating acceleration in Eastern Australia,but
will underestimate peak ground accelerations
within 20km focal distance of the earthquake.
The Gaull et al (1990) equation should be used
to estimate velocity.
Gaull et al (1990) have used the variation of
Modified Mercalli Intensity with distance to
estimate attenuation functions for Western,
Southeastern and Northeastern Australia. It is
suggested that these estimated attenuation
functions may be used to estimate peak
acceleration and velocity in these regions. It is
pointed out that care is needed in all these
calculations because the earthquake magnitudes
have been estimated from assumed attenuation
functions,so one needs to be consistent. Where
it is critical,it is clear that the advice of
seismologists should be sought on attenuation,
as well as other aspects of earthquake loadings.
3.6Maximum Credible Earthquake
Magnitude
/
In view of the impracticabilityof identifying the
faults in Australia on which major earthquakes
may occur,it is necessary to use probabilistic
methods to estimate expected ground motion
versus AEP. For this,it is necessary first to
assess the magnitude of the maximum credible
earthquake.
If the earthquake catalogue only covers a short
period compared with the required return
period,then a large area must be considered
when estimating Mmax,perhaps as large as the
whole of Australia or even including other
intraplate areas over the earth.
Table 1 lists some of the largest earthquakes
which have occurred in Australia. It will be
seen that there are several in the range
M6.5-M7.
To give some appreciation of the likely
maximum credible magnitude,seismologists
have considered the credible length and depths
of fault which may rupture to cause an
earthquake.
There is an approximate relationship between
the rupture area of a fault and the magnitude of
the earthquake (Gibson,1994).
Magnitude Fault Area km2
4 1
5 10
6 100
7 1000
If earthquakes are constrained to shallow
depths,as is the case in Australia,then a large
earthquake will require a very long fault. For
example,if the down-dip distance of a
magnitude 7earthquake is 20km,then the strike
length will be 50km. A magnitude 8 earthquake
similarly constrained would give an unlikely
rupture length of hundreds of kilometres.
Very few intraplate earthquakes are larger than
magnitude 7.5. A series of very large intraplate
earthquakes in the New Madrid area of
Missouri,USA,in 1811 to 1812,apparently had
magnitudes exceeding 8.0(Johnston and
Shedlock,1992). Other authors believe that the
New Madrid earthquakes were smaller
(Everhden,1975,M6.9; Gomberg,1992,M7.3;
Gibson,1994).
There is a general consensus among Australian
seismologists that the maximum credible
ANCOLD Guidelines for Design of Dams for Earthquake 11
magnitude of an earthquake in Australia is
around M7.5. This is based on consideration of
the depth at which earthquakes occur and the
sensible length of fault which will rupture. A
M7.5 earthquake could occur anywhere in the
country.
Given the shallow seismogenic depths in
Eastern Australia,the value of 7.5 for Mmax is
possibly conservatively high. For return
periods to 1000years,decreasing it to 7.2 or 7.3
would have negligible effect on ground motion
recurrence estimates. For long return period
motion,the lower value could be justified,
especially if there is little earthquake and
geological evidence for large nearby faults. If
the greater depth of Central and Western
Australian earthquakes is verified,the value of
7.5 may not be conservative. Gibson (1994)
indicates that there is some evidence that
earthquake activity is cyclical,occurring in
clusters,perhaps about every hundred years for
active areas,or at much longer intervals for
inactive areas. Clusters are affected by activity
in neighbouring areas,so at any particular place
the activity is cyclical rather than periodic. The
fact that say a M6.5 earthquake has occurred
does not preclude a M7.() or M7.5 in the
foreseeable future (ie. <100year). In fact,there
is some evidence that large earthquakes may be
preceded by smaller ones.
Given that since 1968 Australia has experienced
four earthquakes of magnitude 6.2 to 6.8,all of
which produced surface rupture,from 13km to
35km in length,it does not seem "incredible"
that we could experience an M7.5 event.
It is worth noting that when one considers the
10"3AEP earthquake for say Sydney,one may
be expecting a M6 or M6.5 earthquake. By
comparison a 10"3AEP earthquake for San
Francisco will be say M8 (Committee on Safety
Criteria for Dams (1985) suggest M8 has 150
year return period for San Andreas fault).
However,when one considers the 10"5 AEP
event in Sydney it is likely to be around M7.5,
and San Francisco will have risen to say M8.5.
Hence,the differential becomes less at the
lower probability events.
3.7Estimates of Ground Motion and
Response Spectra at a Site
Estimates of ground motion at a site are made
by combining the earthquake occurrence
estimates,with a suitable attenuation function.
Response spectra are estimated using data fr<
overseas earthquakes,modified to
Australian conditions. Data from Austral!
earthquakes are less reliable because of t
scarcity of data. Typical peak grou
accelerations vs AEP and response spectra t
given in Appendix B,but an individi
assessment should be made for each dam.;
should be noted that these will usually be f
bedrock conditions,unless otherwise specifiec
The installation of digital accelerographs If
shown that Australia experiences high pe
ground accelerations. An ML4.9 aftershock
the Tennant Creek earthquakes was recorded
a distance of 9km with a peak grow
acceleration of 0.53g,at high frequency. Mc
accelerograms recorded in Australia to da
show these high accelerations. These hi;
accelerations occur in high frequency peai
which have little impact on the response of
structure. Accelerations with higher th,
expected values are also being recorded in oth
intraplate areas such as Canada.
For other than for very preliminary studie
estimates of ground motion and respom
spectra should be made by experienc<
seismologists.
3.8 Earthquake Hazard Maps
(a) History of Development
Earthquake hazard maps of Australia ha\
evolved considerably over the past eightee
years,in line with improved seismograp
coverage,more and better earthquake data,an
a better understanding of the geologic!
processes producing earthquakes. They may t
useful in defining design earthquakes,at lea
for high AEP events.
Figures 2 to 5 are examples of these map
showing the reducing emphasis on particuk
past earthquakes and on the seismograp
distributed with time.
McEwin,Underwood and Denham (1976
produced a series of maps showing the expecte
50year return period ground acceleration:
velocities and intensities. The map giving th
50year return period peak ground velocities i
shown as Figure 2. Data were taken from th
period 1960to 1972. Only six seismograph
were operating in Australia before 1960,an
12 AN COLD Guidelines for Design of Dams for Earthquake
" K
P they were of little value for the study of smaller
If local earthquakes.
|f The maps produced were significantly affected
I by the distribution of seismographs at the time,
and by the locations of the large earthquakes
that occurred during the period.
The Standards Association of Australia (1979)
produced the SAA Earthquake Code
AS2121-1979. This included a map showing
four zones of earthquake hazard based on the
expected recurrence of peak ground velocity,
Figure 3. It was produced using data up to
1976,and also considering historical earthquake
activity from 1897. It was felt that data were
complete for all earthquakes exceeding
magnitude ML4.0from 1969.
The map was still dominated by the larger
earthquakes that had occurred between 1960
and 1976,and was inevitably influenced by the
seismograph distribution. However,it did
recognise new areas of activity in Queensland
and Central Australia.
Where uncertainties were very high,zones were
deliberately given simplified rectangular
boundaries.
Most of Australia was in zone zero,but this did
not imply that earthquakes would not occur in
these areas. It was stated that significant
shaking may arise in any part of this zone in the
future.
Gaull,Michael-Leiba and Rynn (1990) used
programs based on the Comell-McGuire
method to produce plots of ground motion with
a 10% chance of being exceeded in a fifty year
period. These included peak ground velocity,
peak ground acceleration (Figure 4),and
Modified Mercalli Intensity. The data used
covered the period from 1859 to 1987.
The maps showed larger areas susceptible to
earthquakes,but with higher values for those
areas which had experienced a large earthquake
within the past 100years. They show a
reducing emphasis on particular past
earthquakes,and on the seismograph
distribution with time.
Between 1986 and 1989,several earthquakes
exceeding magnitude ML5.0occurred in areas
that had little or no previous known activity.
These included Maryatt Creek in northern
South Australia (1986),Nhill in Victoria
(1987),Tennant Creek in the Northern Territory
(1988),Uluru in the Northern Territory (1989),
Newcastle in New South Wales (1989).
Figure 2. 50year return period peak ground velocity (McEwin,Underwood and Denham,1976).
ANCOLD Guidelines for Design of Dams for Earthquake
13
.Tv;;
130
Figure3AS2121-1979-Earthquake zoning.
Figure 4. Peak ground velocity (mm/sec) with a 10% chance of being exceeded in a 50year perioi
(Gaulletal,1990).
14
ANCOLD Guidelines for Design of Dams for Earthquake
Figure 5. Acceleration coefficient map for AS 1170.4-1993.
(b) The 1993Standards Australia
Earthquake Code
A Standards Australia Committee began
working on a new Code for minimum loads on
structures (Standards Australia,1993) in late
1989. The previous Earthquake Code had been
completed over ten years before,and there was
insufficient time for the earthquake hazard to be
computed using either spectral attenuation
functions or a new computer based GIS
program.
It was decided that hazard was to be represented
by a single number varying with location,so it
was not possible to distinguish between high
frequency vibration hazard (with acceleration
dominant) and low frequency vibration hazard
(with velocity or displacement dominant).
To avoid the problems of major increments at
zone boundaries it was decided to produce a
contour map,and to define standard values for
use in major cities.
The hazard estimate is based on the velocity
recurrence map of Gaull et al (1990).
The map is based on the best available
Australian attenuation functions,derived from
isoseismal maps. The numerical values on the
velocity map were consistent with those
obtained from other studies. There is better
correlation of peak ground velocity with
intensity and damage than with peak ground
acceleration. There is also considerable
uncertainty with acceleration attenuation
functions for Australia,and the higher values
being measured would need to be adjusted for
short duration structural response.
The map was based on a Comell-McGuire
source zone configuration. This consisted of
polygons each with the equivalent of constant
N0,b and Mmax values. The polygons used had
sharp boundaries rather than fuzzy boundaries,
producing some angular patterns in the
contours. A number of additional earthquakes
had occurred.
It was decided to use the velocity map as a base
map and to smooth the contours compared to
the earlier code,according to the collective
experience and knowledge of the committee.
ANCOLD Guidelines for Design of Dams for Earthquake 15
The smoothing was done after consideration of
the gross geology and tectonics of Australia,
new information on historical earthquakes,and
the earthquakes that had occurred since the map
was originally drawn,eg. particularly the 1988
Tennant Creek and 1989 Newcastle earthquake
events.
The resulting acceleration coefficient map
(Figure 5) was smoother than the original,in
recognition of the uncertainties involved. A
number of contours are plotted to facilitate
interpolation,and should not be taken as an
indication of precision.
The map was converted from peak ground
velocity to an "acceleration coefficient" for use
in the code. The acceleration coefficient is a
dimensionless quantity numerically similar to
peak ground accelerations that would have been
computed using traditional attenuation
functions. These values are lower than peak
accelerations actually being measured from
Australian earthquakes,but may be thought of
as being an "effective acceleration" after
consideration of the short duration of the strong
motion.
The map gives the broad variation in earthquake
ground vibration hazard over the country. The
effects of site response due to weak surface
materials or topography are allowed for in an
additional site factor,in the code.
The map does not allow for any effects of local
faults,and does not cover hazards other than
ground vibration,such as surface rupture or the
potential for landslides or tsunamis. Site
response and these other hazards vary over
small distances,so could be represented in local
earthquake zoning maps,perhaps to a scale of
about 1:100,000.
4. SELECTION OF DESIGN
EARTHQUAKE
4.1 Definitions
Dam engineers,in keeping with engineers who
design other structures,have commonly
adopted a deterministic or standards based
approach to design,with two levels of design
earthquake motion; one for serviceability,
known as the Operating Basis Earthqua
(OBE) and the other for the condition whi
severe damage is expected,but uncontrolj
release of the storage water is to be prevent!
known as the Maximum Design Earthqua
(MDE). The working group has decided,
adopt slightly modified versions of {
International Commission on Large Dai
(ICOLD) definitions (ICOLD,1989) for OI
and MDE,as follows:
Maximum Design Earthquake -the ME
will produce the maximum level of groui
motion for which the dam should
designed or analysed. It will be required
least that the impounding capacity of t
dam be maintained when subjected to th
seismic load.
Operating Basis Earthquake -the 01
will produce a level of ground motif
which will cause only minor and acceptab
damage at the damsite. The dai
appurtenant structures and equipme
should remain functional and damage fro
the occurrence of earthquake shaking n
exceeding the OBE should be easi
repairable.
When considering the MDE,the concept
Maximum Credible Earthquake (MCE) is oft
used in the assessment. The MCE is defined a!
Maximum Credible Earthquake -t
MCE is the largest reasonably conceival
earthquake along a recognised fault
within a geographically defined tectoi
province,under the known or presum
tectonic framework.
ICOLD (1989) allow definition of MCE
terms of either magnitude or intensity at t
dam site. The Working Group are of the vi(
that this can be confusing,and recommend i
of MCE in terms of magnitude only.
In some situations,where two or more causati
faults are identified,each may have an MCE.
The MCE which is critical for the design of <
dam is known as the Controlling MaxiiW
Credible Earthquake (CMCE).
16 ANCOLD Guidelines for Design of Dams for Earthquake
The MDE is expressed in terms relevant to the
design of the dam,which may include peak
ground acceleration or velocity,earthquake
magnitude and distance or response spectra.
Where causative faults can be identified the
MDE may often be determined from the MCE
(or CMCE). Where such faults cannot be
identified,the MDE can be determined by
probabilisticmethods.
It is recommended that the probability that a
particular earthquake magnitude or ground
motion (eg. peak ground acceleration or
velocity) will be exceeded in any year be
expressed as the Annual Exceedence
Probability (AEP),and expressed as a 1 in Y
terminology,eg. AEP of 1 in 1000.
Probabilities can be expressed also in terms of
the life of the structure or some other period. In
this case the probability is expressed as the Y
year EP,for example,the 50year Exceedence
Probability may be 1 in 100.
The AEP and Y-year EP are related by the
formula
Y-year EP=l-(l-AEP)y
Table 2 lists exceedence probabilities for
periods of 50to 500years calculated from
AEPs varying from 1 in 100,000to 1 in 100.
Table 2. Relationship between Annual Exceedence Probability (AEP) and Exceedence Probability
(Y-year EP) for Different Periods.
AEP Y-year EP for Different Periods
Y =50Years Y =100Years Y =200Years Y =500Years
1 in 100,000 1 in 2000 1 in 1000 1 in 500 1 in 200
1 in 10,000 1 in 200 1 in 100 1 in 50 1 in 20
1 in 1,000 1 in 20 1 in 10 1 in 5 1 in 2.5
1 in 100 1 in 2.5 1 in 1.6 1 in 1.2 1 in 1.01
Table 2 shows that the AEP of an earthquake
can be a misleading measure of general
security,suggesting a greater degree of safety
than actually exists over a long period of years.
For example,if a dam is safe for an earthquake
which has an AEP of 1 in 100,and if one is
interested in the safety of downstream residents
over a period of say 50years,then the 50-year
EP of that earthquake is 1 in 2.5,ie. there is a
40% chance of that earthquake occurring in any
50year period.
Thus,in assessments of safety issues,a better
perspective of relative safety is often obtained
from consideration of earthquake exceedence
probabilities over a period of years which
relates to the life of the structure or the period
of occupancy,rather than from consideration of
the AEP.
As will be discussed later,the design of dams
for earthquake may be done using risk
assessment methods. In this guideline,the
following definitions relating to risk assessment
are adopted from the ANCOLD Guidelines on
Risk Assessment (ANCOLD,1994).
Hazard:
That which has the potential for creating
adverse consequences. Hazard is rated
according to the scale of the adverse
consequences.
Hazard Category:
A scale of adverse consequences caused by
dam failure.
Risk:
The likelihood or probability of adverse
consequences; the downside of a gamble.
ANCOLD Guidelines for Design of Dams for Earthquake 17
Acceptable Risk:
That level of risk that is sufficiently low
that society is comfortable with it. Society
does not generally consider expenditure in
further reducing such risks justifiable.
Risk Analysis:
A policy analysis tool that uses a
knowledge base consisting of scientific and
science policy information to aid in
decision making.
Risk Assessment:
The total process making use of risk
analysis and which embraces a
consideration of all costs and benefits
including the identification of risks,the
estimation of their likelihood of occurrence
and the evaluation of the social
acceptability of the risk. Risk assessment
includes the processes that lead to
decisions.
NOTE: The definitions above are to be
reviewedif/whenANCOLD agrees toany change
Risk Management:
The reduction or control of risk to an
acceptable level.
Socially Acceptable Risk:
That low level of risk that society finds
tolerable so that expenditure would not
normally be directed to its reduction.
Individual Risk Criterion:
The socially acceptable level of risk to a
particular individual.
Societal Risk Criterion:
The socially acceptable level of risk in
terms of events that impact on society at a
community,regional or national level.
The definitions of failure,accident,
deterioration and incident are taken from
ICOLD(1983):
Failure:
Two categories of failure are listed by
ICOLD(1983):
Failure Type 1 (Fl): j
A major failure involving
complete abandonment of the dam.
Failure Type 2 (F2): j
A failure which at the time may hi
been severe,but yet has permitted
extent of damage to be successfi
repaired,and the dam brought into
again.
For the purposes of this guideline,the te
failure will refer to both Type 1 and Typ
failures.
Accident:
Three categories of accidents are listed
ICOLD(1983):
Accident Type 1 (Al):
An accident to a dam which has b
in use for some time,but which
been prevented from becoming
failure by immediate remei
measures,including possibly draw
down the water.
Accident Type 2 (A2):
An accident to a dam which has bi
observed during the initial filling
the reservoir and which has bj
prevented from becoming a failure!
immediate remedial measu
including possibly drawing down!
water. i
Accident Type 3(A3): i
An accident to a dam du|
construction,ie. by settlement
foundations,slumping of side slo
etc which have been noted before
water was impounded,and where
essential remedial measures have b
carried out,and the reservoir sa:
filled thereafter.
Deterioration:
Any faulty behaviour,from the poin
view of both safety and performance,ei
during construction or after being!
service,including failure cases.
Incident:
Either a failure or an accident requi
major repair.
18 ANCOLD Guidelines for Design ofDams for Earthquake
Breach:
A failure resulting in rapid release of water
from the dam.
The ANCOLD (1994) definition of risk is not
explicitly linked to the consequences,and is
expressed in terms of probability.
In some situations it is necessary to assess the
total annual consequence of failure,and it is
useful to assess the total risk (either loss of life,
or economic consequences),where total risk is
defined as:
Total Risk: A measure of
the probability and severity of
an adverse effect to health,
property or the environment.
Total risk is estimated by the
mathematical expectation of
the consequences of an adverse
event occurring.
Total Risk is calculated from
Total Risk =(probability of event) x
(consequencesof that event)
which has the dimension
consequence event consequence
x
unit time unit time event
This definition is adapted from that used by the
Canadian Dam Safety Association (1994) for
"risk". It will be noted that the total risk can be
posed by a number of causative events,eg.
flood,earthquake,and other,eg. piping failures,
slope instability,in which case the total risk is a
summation from the causative events.
For the purposes of this guideline,the hazard
categories are defined as shown in Table 3.
These are taken from ANCOLD (1986) and are
under review by ANCOLD. The revised
guideline will include assessment of the
environmental consequences of dam failure and
is likely to have more hazard categories.
Table 3. ANCOLD Dam Hazard Categories (ANCOLD,1986).
High Significant Low
Loss of identifiable life expected
because of community or other
significant developments
downstream.
No loss of life expected,but the
possibility recognised. No urban
development,and no more than a
small number of habitable
structures downstream.
No loss of life expected.
Excessive economic loss such as
serious damage to communities,
industrial,commercial or
agricultural land or facilities,
important utilities,the dam itself or
other storages downstream.
Appreciable economic loss,such
as damage to limited land areas,
secondary roads,minor railways,
relatively important public
utilities,the dam itself or other
storages downstream.
Minimal economic loss,such
as farm buildings; limited
damage to agricultural land,
minor roads etc.
Repairs to dam not practicable. Dam
essential for services.
Repairs to dam practicable or
alternative sources of water/power
supply available.
Repairs to dam practicable.
Indirect losses not significant.
Npte that items refer to incremental losses and
effects due to dam failure. Hence,an
assessment of the damage and loss of life due to
the earthquake with and without breaching of
the dam may be made.
ANCOLD Guidelines for Design ofDams for Earthquake 19
h !
4.2 Selection of the Design
Earthquake
4.2.1 General Approach
There are two general approaches to selection
of the "design earthquake".
(a) The Deterministic,Prescribed
Approach
This is the method which has been most widely
used to date. The design earthquake load
(MDE or OBE) is determined by consideration
of the hazard category of the dam,and is
selected as an earthquake with a given AEP.
For example,a significant hazard category dam
might be designed for a 1 in 1000AEP
Table 4. Methods Used by Dam Authorities to Estimate Maximum Design Earthquake (MDE)
earthquake either on an identified fault,or f
an AEP vs ground motion graph determinec
the seismotechnic province provided for
dam. For high hazard dams,the MCE ma;
adopted to determine the MDE,but
requires a knowledge of the fault(s) on w]
the MCE may occur. This is
impracticable in Australia.
Table 4summarizes the practice of a numbs
organisations.
Organisation Method Reference
ICOLD a)Normally equal to MCE obtained
deterministically,or "50% probability or
higher of not being exceeded in a large
number of years.
b)If no hazard to life (ie. low hazard dam),
use less than MCE based on economics.
ICOLD (1989)
Federal Emergency Management
Agency (of USA) (FEMA)
Generally equated to controllingMCE. FEMA (1985)
US Bureau of Reclamation
(USER)
MCE,or where there are no specific faults,
an AEP of 1 in50,000(1).
USER (1989)
National Research Council (of
USA)
MCE for high hazard dams and less than
MCE for low hazard dams.
National Researc
Council (1985)
EC Hydro,Canada pre 1993 MCE for high hazard,0.5 MCE for
significant hazard,1 in 200AEP for low
hazard.
Salmon and von
(1993)
EC Hydro,Canada post 1993 Risk basedapproach,seeNotes(2). EC Hydro (1993]
Canadian Dam Safety
Association (CDSA)
MCE or 1 in 5000to 1 in 10,000AEP for
very high consequence; 0.75 to 1.0MCE,or
1 in 1000to 1 in 5000AEP for high
consequence,1 in 100to 1 in 1000AEP for
low consequence,seeNotes(3).
CDSA (1994)
New South Wales Dams Safety
Committee (Australia
For high hazard the most severe seismic,
loading that could reasonably occur in the
area; for significant hazard dams AEP 1 in
1000,or base on risk analysis.
NSWDSC(1993
NOTES:
(1) USER do not specify hazard
category,but it is reasonable to
assume they refer to high hazard
dams.
20 AN COLD Guidelines for Design of Dams for Earthquake
(2)For dams with greater than 100
potential premature deaths,
maximum tolerable AEP is 1 in
100,000and MCE should be
adopted. For other dams,the annual
probability of dam failure with
potential to cause premature fatalities
should be less than that resulting in
0.001 fatalities per year of dam
operation,except that no identified
individual should have more than
0.0001 probability of fatality per year
of dam operation.
(3)The boundary between very high and
high consequence categories is an
incremental loss of life of 100
persons. Low consequence is,
roughly,equivalent to significant
hazard,very low to low hazard in
Table 3.
(b) The Risk Based Approach
In this method,the adequacy of the dam is
considered by calculating the probability of
breaching of the dam and the expected
incremental loss of life due to the dam
breaching. The acceptability of the dam is then
tested by comparison of the calculated values
with societal and individual risk criteria.
In this approach,the concept of MDE is not
applicable,since one must consider the risk
arising from the full range of possible
earthquake events.
There are two problems with each of these
methods:
the deterministic/prescribed approach fails
to account for the fact that the conditional
probability of breaching of the dam,given
that the MDE (or other selected earthquake)
occurs,is not 1,and may be very different
for different types of dams and foundation
conditions. Hence,the adoption of a
probability of the initiating event,ie. the
/earthquake,may result in a very wide range
of probabilities of dam breaching. This is
quite different to the case for flooding and
embankment dams,because the probability
that a dam fails on overtopping is high,but
the probability that a dam breaches due to an
earthquake is low (based on historic data).
This appears to be recognised to some extent
at least in the USBR (1989)
recommendation of an AEP for MDE of 1 in
50,000for high hazard category dams,
although they do not justify the figure in
these terms. It also appears to be recognised
in the recommended AEPs in Canadian Dam
Safety Association(1994)
the risk based approach is a relatively new
one,and the methods for assessing the
conditional probabilities of breaching are
still being developed,and can be time
consuming and costly. It is also difficult to
justify only considering the risk due to
earthquake,since logically one should sum
the risk from flood,earthquake and other
causes before comparing with acceptable
risk criteria. This means that studies should
consider flood and other causes,increasing
the complexity and cost of the study.
There is some evidence of a trend towards a risk
based approach to earthquake design. The most
detailed discussion of this is given in Salmon
and von Hehn (1993) who describe the BC
Hydro approach at that time. BC Hydro (1993)
give more details of these methods,which are
being used by that organisation.
Finn (1993) indicates that in California there is
a gradual shift from use of MCE to a Safety
Evaluation Earthquake (SEE). He indicates that
in selecting this,consideration is given both to
the probability of occurrence and the
consequences of failure.
(c) Recommended Approach
Based on consideration of all of the factors,the
Working Group has concluded that for high and
significant hazard dams,the design for
earthquake should where practicable be based
on risk assessment methods. These are detailed
in section 4.2.2. For low hazard category dams,
a deterministic approach using ,a MDE
equivalent to an AEP of 1 in 200may be
adopted,but the decision will probably be based
on economic considerations.
It is realised that many will be more
comfortable with an AEP for the MDE related
ANCOLD Guidelines for Design of Dams for Earthquake 21
directly to the hazard category of the dam. The
information in Table 4may be used if this
approach is followed. In particular,reference
should be made to the Canadian Dam Safety
Association (1994) criteria which are
reproduced in Appendix C. However,it must
be recognised that there is an implied
probability of breaching and probability of loss
of life in these values. In reality these
probabilities are very variable,eg. possibly 1 to
3orders of magnitude different for PEB (the
probability of breaching) of a homogeneous
earthfill dam compared to a concrete gravity
dam. The potential for loss of life in the event
of breaching is also very site dependent.
Where causative faults are identified in the
vicinity of the dam,the MDE from these faults
should be determined. For very high hazard
category dams,the more critical of the ground
motion obtained from the MCE on these faults,
and probabilistic approaches should be adopted.
For significant and high hazard category dams,
an AEP should be assigned to the CMCE
determined from geological and
geomorphological assessment of the causative
faults,and the MDE obtained this way
compared to that obtained from the probabilistic
approach,with the larger being adopted.
4.2.2 Risk Assessment Method for Design
for Earthquake
The steps involved in this process are:
(i)determine the AEP of earthquake ground
motion (PE) over the range of earthquake
events which may affect the dam. Table 5
gives an example for AEP vs ground
acceleration
(ii)determine the conditional probability (Pbc)
that for each of the ground motion ranges
(eg. 0.125g to 0.175g in Table 5) the dam
will breach. In assessing this conditional
probability all modes of failure should be
considered and the probabilities combined,
making allowance for interdependence and
mutual exclusivity or otherwise (eg. for
embankment dams,slope instability,
piping,liquefaction/instability,and for
concrete gravity dams,overturning,j
sliding see section 4.4.2 (PB)
(iii)assess the probability of failure for ej
range of ground motion by multiplying)
AEP with PBc,ie. PB =PE x PBC ;
TableS
(iv)sum the probabilities to give the ovl
annual probability of failure due
earthquake. To this,should be added th
annual probability of breaching due
flooding,and other causes (eg. pip
instability) making allowance !
dependence,independence and mu
exclusivity of the events. The Work
Group recognises,however,that in m
cases,these studies will be d
independently,or the risk from earthqu
may far exceed those due to flood or of
causes,and the probability of failure dw
earthquake alone will be carried forwan
the next step.
(v)determine the incremental loss of
expected for the dam breach using
method detailed in section 4.2.5. It sho
be noted that there may be differ
estimates of the loss of life for
different earthquake ranges,and
earthquake,flood and other causes
(vi)determine the average individual risk
the Population at Risk (PAR) from (v) i
the PAR,and the risk to the individ
most at risk
(vii)assess the acceptability of risk using
criteria detailed in section 4.2.3.
The total risk to property is calculated ii
similar way,except that PBC is replaced b;
vulnerability term,ie. the proportion of
element at risk,eg. house,bridge,which is 1
as a result of the earthquake (Vbc)-Vbc is in
range 0for zero damage to 1.0for total loi
Then total risk =2 PE x PBC x (value of elemf
summed over the range of earthquake ev<
and the elements at risk.
22 ANCOLD Guidelines for Design ofDams for Earthquake
TableS.
Acceleration Annual
Probability
Conditional(1)
Probability of (Pbc)
P (2)
r B
<0.075g 0.874 0.0005 0.0004
0.075gto 0.125g 0.100 0.005 0.0005
0.125gto 0.175g 0.015 0.05 0.0007
0.175g to 0.225g 0.007 0.1 0.0007
0.225g to 0.39 0.003 0.3 0.0009
>0.3g 0.001 0.5 0.0005
TOTAL 1.000 0.0037
(1) Given the earthquake occurs(2) PB =annual probability x PBC
4.2.3Acceptable Risk
The calculated risks are compared to the
acceptable risks. What is acceptable will
depend on many factors,including social,
political and economic,and the decision may be
made by persons other than the person carrying
out the risk assessment.
It is recommended that the calculated risk be
related to the acceptable risks outlined in the
ANCOLD Guidelines on Risk Assessment
(ANCOLD,1994),which is due for revision in
1998. The relevant ANCOLD (1994)
recommendationsare:
Societal Risk:
Ensure that new dams,and dams being
upgraded,satisfy the Societal Risk
criterion given by the Objective curve
of Figure 6.
Ensure that existing dams satisfy the
Societal Risk criterion given by the
Limit curve of Figure 6,but carefully
consider the ALARP (As Low As
Reasonably Practicable) principle.
Individual Risk:
/For new dams and upgrading of
existing dams,ensure that the average
risk of death to particular members of
the public from dam failure does not
exceed 10"6 per exposed person per
annum. Do not subject any person,
being a member of the public,ie. the
individual most at risk,to a risk greater
than 10"5 per annum.
For existing dams,individual risks up
to 10times those for new dams could
be tolerable subject to application of the
ALARP principle.
Dam owners are to assess the
acceptable risks for workers at and below
their dam sites.
ANCOLD (1994,1996),recommends that a dam
should comply with both individual risk and
societal risk criteria.
ANCOLD (1994) also indicates that in risk
assessment,if the cost per life saved is less than
A$500,000(1990values) the investment can be
considered as certainly worthwhile.
Appendix D presents some additional
information relating to acceptable and actual
risks.
As discussed above,the probability which
should be used in Figure 6 is for all modes of
failure -earthquake plus flood plus other
causes. In the event that only earthquake is
being considered,allowance should be made for
flood and other causes,eg. one might assign
l/3rd of the acceptable probability of failure
(breaching) to earthquake,but consideration
should be given to the relative probability of
failure under flood,earthquake and other static
loading.
ANCOLD Guidelines for Design ofDams for Earthquake 23
1010410J
N,number of fatalities
due to dam failure
Figure 6. ANCOLD amended interim societal risk criteria (ANCOLD,1996).
4.2.4Estimation of the Probability of Dam
Breaching from Flood,Earthquake and
Other Causes
(a) Worldwide Experience of Behaviour
of Dams
It is common to assume that embankment dams
will fail when they are overtopped by flooding;
and concrete gravity dams will fail when the
IFF level is reached. The IFF is based on
assumed uplift pressures,foundation strengths,
etc,or on the flood level of record. In both
cases,there is an implicit assumption that the
probability of breaching PB is 1.0. Many
embankment dams,particularly those with steel
mesh reinforcement of the downstream zone
(for floods during construction),would
withstand some overtopping without'
Many concrete gravity dams won
flood levels higher than the IFF dep
how conservativelythe IFF had been
Hence,in reality,PB is less than 1.0f<
Table 6 shows the detail of reasons
of embankment dams,other than
appurtenant works (which indue
overtopping).
24 ANCOLD Guidelines for Design of Dams for Earthquake
Table 6. Most Frequent Classification of Deterioration and Dam Failure in Embankment Dams
(ICOLD,1983).
Classification Percentage of Total Ratio of Failures to
Classification Deterioration
Deterioration Failure As a Percentage
Percolation (foundation) 27 26 9
Percolation (dam) 25 38 15
Slope protection 19 <1 NA
Differential movement 17 31 18
Internal erosion (dam) 17 49 29
Downstream slips 11 16 15
Internal erosion (foundation) 11 17 15
Compaction 8 12 14
Upstream slips 8 5 6
Bonding between concrete structures 7 6 38
and embankment
Deformation and land subsidence 5 7 14
Watertight systems 4 <1 NA
Earthquakes 4 <1 NA
Drainage system and filters 4 <1 NA
Shear strength 4 5 13
NOTES:
(1)NA =Not available.
(2)'Percolation' is often related to lack of seepage control measures.
(3)Deformation is often associated with percolation,internal erosion,shear strength. Land
subsidenceis a minor cause.
(4)Total number of dams9900
Total number of deteriorations (a)432 (4.4%)
Total number of failures (a)43(0.43%)
(a) Not including appurtenant works,which total 39 failures. These account for most overtopping
failures due to inadequate spillways.
As discussed ICOLD (1983) and reproduced in
Fell,MacGregor and Stapledon (1992),less
than 1% of dam failures are due to earthquake,
and approximately 50% of failures are due to
causes other than flood overtopping and
earthquake.
Given that 0.83% of all embankment dams
surveyed by ICOLD failed,the probability of
failure can be calculated to be as shown in
Table?.
Similar calculations have been done to produce
Tables 8 and 9 for concrete dams (excluding
masonry dams),and Table 10for appurtenant
works.
/
ANCOLD Guidelines for Design of Dams for Earthquake 25
I
:
Table 7. Average Probability of Failure of Embankment Dams Over the Life of the Dam,
on ICOLD (1983).
Cause of Failure % of Failures Probability
of Failure <
Flood overtopping and failure of other appurtenant works 48% 1 in 250!
Slope instability 8% 1 in 1500
Internal erosion and poorly controlled seepage in embankment 28% 1 in 425 |
Internal erosion and poorly controlled seepage in foundation 12% 1 in 10001
Other 4% 1 in 3000;
Total 100% 1 in 120]
NOTE:
(1) Combined probability of internal erosion failure 1 in 300
Table 8. Most Frequent Classification of Deterioration and Dam Failure in Concrete Di
(ICOLD,1983).
Percentage of Total Ratio of Failures to
Classification Classification Deterioration
Deterioration Failure As a Percentage
Inadequacy of site investigation 7 14 60
Shear strength foundations 9 14 43
Percolation foundation 28 5 5
Internal erosion (foundation) 21 29 37
Grout curtain etc 11 10 25
Drainage in foundation 11 5 12
Uplift affecting foundation 12 5 11
Shape of dam and valley (arch) 4 5 33
Shape of dam and valley (gravity) 5 10 40
Tensilestresses indam 11 5 9
NOTES:
(1)Total number of dams4250
Total number of deteriorations (a)251 (5.9%)
Total number of failures (a)12(0.3%)
(a) Not including appurtenant works,which total 8 failures
(2)Does not include masonry dams.
Table 9. Most Frequent Classification of Deterioration and Dam Failure in Concrete Dams (
the Life of the Dam (ICOLD,1983).
Cause of Failure % of Failures Probability
of Failure
Flood overtopping and failure of other appurtenant works 40% 1 in 525
F oundation strength 15% 1 in 1400
Foundationerosionandseepage 25% 1 in 850
Foundation uplift and drainage 5% 1 in 4000
Tensilestresses indam 3% 1 in 7000
Other 12% 1 in 1750
Total 100% 1 in 210
NOTE:
(1) Combined probability of foundation failure 1 in 475
26ANCOLD Guidelines for Design of Dams for Earthquake
Table 10. Most Frequent Classification of Deterioration and Dam Failure in Appurtenant Works
(ICOLD,1983).
Percentage of Total Ratio of Failures to
Classification Classification Deterioration
Deterioration Failure As a Percentage
Percolation information 6 13 29
Internal erosion of foundation 8 11 19
Removal of rip-rap <1 4 40
Corrosion of steel and other materials 3 4 18
Structural failure 10 7 10
Excessive flow,including spillways 24 73 44
Solid materials carried by waterflow 10 5 8
Local scouring 14 11 11
Malfunction of discharge equipment 11 9 13
Erosion by abrasion (concrete) 12 <1 NA
Erosion by cavitation 9 <1 NA
These tables are for average dams,and the
figures in Tables 7and 9 are over the life of the
dam. The average age of the dams in the
ICOLD survey was -30years,so average
annual probabilities of failure can be obtained
by dividing the figures in Tables 7and 9 by 30.
This gives the average annual probability of
failure of embankment dams as ^O.00015,or 1
in 6000,and of concrete dams 0.0001,or 1 in
10,000(both figures excluding overtopping and
other failures of appurtenant works).
The ICOLD (1983) study also assembled data
on the time of deterioration after construction.
This data is shown in Table 11.
It will be noted that a disproportionate number
of failures occur on first filling. This can be
taken account of in the overall assessment,eg.
for embankment dams,40% of failures occur
during first filling,so the average probability of
failure during first filling is 0.4x 0.43% or 1
in 500. The equivalent value for concrete dams
is alsol in 500.
Table 11. Distribution in Percentage of Deterioration an Failure Cases in Terms of Time After
Construction.
Failure Time Concrete Dams Embankment Dams
Failure Deterioration Failure Deterioration
Construction 11 7 14 11
First filling 67 11 40 25
First 5 years 6 10 11 20
After 5 years 11 30 30 26
Not available 5 42 5 18
These figures do not take account of the design
of the dam,the quality of investigation and
construction,or the fact that dam engineering
has improved since the ICOLD survey in 1975,
or the degree of monitoring and surveillance.
Logically,some types of dam are more
susceptible to failure and breaching than others.
For example,a homogeneous earth dam
constructed of dispersive soil,and showing
signs of excessive seepage,is more likely to fail
and breach than a central core earth and rockfill
dam with well designed and constructed filters.
McCann et al (1985) have made an attempt at
quantifying this,but their study does not
account for zoning or condition of the dam.
ANCOLD Guidelines for Design of Dams for Earthquake 27
1
A study by Hackney (1994) of 65 of the case
histories of embankment dams (including 15
failures)on which the ICOLD survey was
based,showed that homogeneous earthfiII and
zoned earthfilldams which have no filter &
were disproportionately represented in
incidents. Figure? shows the results.
Figure 7. Dam category vs percentage of total observed incidents (Hackney,1994).
Category CI-Homogeneous Earthfill }
Category C2-Earthfill with Toe Drain } No filters
Category C3-Zoned Earthfill }
Category C4-Earthfill with Horizontal Drain
Category C5 -Earthfill with Vertical and Horizontal Drain
Category C6 -Earth and Rockfill -Central Core
Category C7-Earth and Rockfill -Sloping Upstream Core
Category C8 -Concrete Face Rockfill
Category C9 -Earthfill with Concrete Core Wall
The sample of dams studied by Hackney (1994)
was reasonably representative of the ICOLD
(1983) study population of dams in respect to
age of dam and dam height. However,there is
not sufficient data in the ICOLD study to
determine the total number of dams in each
category,so it is not clear whether the reason
for the overrepresentation of homogeneous
earthfill and zoned earthfill dams in the
incidents in Figure 7is due to a higher
proportion of those dams being present,or
whether they are more susceptible to failure and
accidents. Given that a large proportion of the
incidents are due to internal erosion of the dam
and the foundation,and there is poor control
over these in homogeneous and zoned earthfill
dams,it seems likely that these types of dam are
more susceptible to failure and accident
others.
The discussion so far has referred to dams ui
non earthquake conditions. The performam
dams under earthquake conditions has genei
been good with few dams suffering m
damage except where liquefaction has occii
in the dam or its foundation. Seed (19
ICOLD (1986),USCOLD (1992),NSW1
(1993) and Hinks and Gosschalk (1993)
some details.
However,this needs to be considered in
context that relatively few dams
experienced major earthquake loading,so
too much comfort should be taken in
performance.
1
28 ANCOLD Guidelines for Design of Dams for Earthquake
What can be expected in major earthquake is:
materials susceptible to liquefaction may
lose a significant amount of strength,and
large deformations may occur. This is
discussed at length in Section 5
embankment dams can be expected to settle
and crack. Most cracking is likely to be
longitudinal,but transverse cracking may
occur
concrete dams can be expected to crack and
may move on their foundations. This may
significantly affect the available shear
strength on the foundation,foundation uplift
pressures,uplift pressures and tensile
strengths in the dam
Hence,it is logical that the dams will be more
susceptible to failure after experiencing the
earthquake,than before. The extent to which
this affects the probability of failure and
breaching depends on:
the severity of the earthquake and damage to
the dam,for example,homogeneous earthfill
dams will have a higher probability of
failure post earthquake cracking,than central
core earth and rockfill dam,with well
designed filters and free draining rockfill; a
concrete gravity dam on strain weakening
interbedded shale and sandstone,is more
likely to experience stability problems than
one on granite with favourably oriented
joints
the detailed design of the dam
the water level at the time of the earthquake
and how quickly it can be lowered
the potential for earthquake aftershocks.
At this time no specific guidance can be given
as to how these issues can be quantified.
(b) Recommended Method
There are two approaches which can be taken:
(i)^ Event tree analysis,where the potential
mechanisms leading to failure and
breaching are identified,and probabilities
assigned to these events,leading to an
overall assessment of the probability of
breaching. The procedures for this method
are described in ANCOLD (1994). The
method requires the use of judgemental
"expert" opinion on assessing
probabilities,and is in use in Canada,eg.
Salmon (1995),Salmon and Hartford
(1995 (a),(b)) (and more recently in
Australia,eg. Prospect Dam,
(Landon-Jones et al,1995).Expert opinion
can be supported by reliability and other
analyses that help to describe the response
of the dam system to earthquake loading.
It is recommended that where practicable
event tree analysis is carried out,since this
includes assessment of the details of the
design,construction,monitoring and
surveillance of the dam.
(ii) Empirical approach. An approximate
empirical method for estimating the
probability of failure has been proposed in
Fell (1995). As pointed out in the paper,
the method is based on many assumptions
which cannot be backed up with reliable
data,and it should only be used for
preliminary assessments of existing and
new dams.
4.2.5 Estimation of the Loss of Lives if a
Dam Breaches
(a) General
The most comprehensive method for estimating
the loss of life from a dam breaching has been
to carry out a dam break analysis and use the
USER (1989) "Policy and Procedures for Dam
Safety Modification and Decision Making"
approach to estimate loss of life,given the flood
velocities,depths and other data. However,the
USER procedure has drawbacks,notably the
dichotomous nature of the critical warning time
parameter.
The paper "Predicting Loss of Life in Cases of
Dam Failure and Flash Flood",by DeKay and
McClelland (1993),provides an improved
method for estimating the potential loss of life
due to severe flooding. It is derived from the
historical record of dam failures and flash flood
cases via logistic regression. This uses the data
on which the USER (1989) method is based,as
well as some other cases,and analyses the data
ANCOLD Guidelines for Design of Dams for Earthquake 29
somewhat more rigorously than the USBR
method. Because of this,and because it gives a
single continuous equation,the DeKay and
McClelland (1993) method is recommended. A
summary of the method is included in these
guidelines,but it is important that users read
that paper in detail,because it explains the
empirical basis on which the method is
determined. In particular,users should be
aware of the wide confidence intervals applying
to the regression equations,as evidenced by the
differences between the actual and predicted
loss of life given in Table 1 of the paper.
(b) The DeKay and McClelland Method
The general equation for potential loss of life is:
LOL=
PAR
1 + 13.277(PAR )
where LOL
PAR
WT
Force
044\ (0759(WT)-3790(For<B)-2 223(WT)(Force)
)
Potential loss of life
Population at risk
Warning time (hours)
Forcefulnessof floodwaters
1 for high force
0for low force
The method is only statistically rated for a
population at risk of less than 100,000,and is
not applicable to cases of dam failure equivalent
to Vaiont Dam,where a massive landslide into
the reservoir resulted in 1209 deaths from one
village of 1345 persons alone.
In the DeKay and McClelland method the
following assumptions are made:
(i) Population at Risk (PAR) is defined
as the population at risk of "getting
their feet wet" if they took no action to
evacuate,ie. regardless of the depth or
velocity of the flood water. No-one
who is more than three hours flood
travel time below the dam should be
included in the PAR.
The PAR will usually be calculated
from a dam break analysis to determine
the area which will be inundated.
(ii)
The number of residents iq
inundated area may be estimati
one of the following methods:
1. Use the current cens
population of the town|
region only if it
completely inundated.
2. Use census data based
postal codes within
inundated area.
3. Multiply the number
homes in the area by
average number of peoj
per home for that particu|
area.
4. Conduct a house -to-hou
survey.
The assessment process needs
subdivide the numbers dependini
distance (and,hence,warning t)
from the dam.
Once the number of residents infl
area is estimated,the population atj
is found by multiplying the numbe
residents by an annual exposure fac
This is the fraction of a year tha|
typical individual would spend
home. A typical exposure time?
residents is in the range of 0.6 to 0.8|
The population at risk for facilil
other than residence (such as schod
industrial buildings,parks,shoppj
centres,etc) should be found
multiplying the estimated aver!
number of people within the facilit
an exposure factor,which may refl
the fraction of the year that the facil
is open. If the number of people andj
exposure factor cannot accurately!
found,then estimates should be ittf
using the best overall judgemetl
However,it should be noted thai
'I
some areas these "special" populati
can be very large,and need toj
estimated accurately.
Warning Time (WT). There is
single method for estimating war!
30 ANCOLD Guidelines for Design ofDams for Earthquake
time. In their study database DeKay
and McClelland (1993) used the
recorded WT (adjusted as described in
the paper). USBR( 1989) states that:
"Specifically,the time to be
estimated is the time at which the
public warning process has been
officially set in motion and the first
individuals for each PAR are being
warned to evacuate. Again,this
time is expressed as the number of
hours prior to the arrival of
flooding at the location of the
PAR".
In other words,they state that warning
time is equal to the difference between
the time the public evacuation warning
is initiated and the actual flood arrival
time for the population at risk.
Because of the difficulty in estimating
the time that the public evacuation
warning would be initiated,a
conservative estimate for dams which
have full time operating staff would be
to use the time at the beginning of dam
breach (provided the operator can issue
a warning). That is,warning time is
equal to the flood arrival time as shown
on the inundation maps. According to
DeKay and McClelland (1993):
"Such assumptions may be
reasonable for earthquake
-induced failures (and perhaps
terrorist acts),but other modes of
failure will typically allow for
longer WTs and greater
evacuation benefits. We assume
that the case in which an
evacuation is ordered at the time
of dam failure represents the
worst-case scenario and the most
reasonable baseline for assessing
the benefits of additional WT".
The actual warning time will depend on
the nature of the dam failure,whether
there is surveillance of the dam,flood
preparedness plans,and will have to be
determined on a case by case basis. For
unmanned dams in remote areas,
warning times may be less than the
flood arrival time,because the failure
may go unnoticed.
(iii) Force. DeKay and McClelland (1993)
describe Force as flooding lethality
indicated by variations in the depth and
velocity of the flood waters. In the
general equation Force is treated as a
dichotomous variable with values of
one for high force and zero for low
force.
The term high force refers to flood
waters that are likely to be deep and
swift. This would be typical in a
narrow valley or canyon area.
Substituting a Force value of one into
the general equation gives the
following expression for potential loss
of life in a high force or canyon area:
LOL=
PAR
1 +13211 {PAR044)g(2'982(',T)"379())
The term "low force" refers to flood
waters that are likely to be shallow and
slow. This would be typical in the
mature reaches of a river where there
are wide alluvial flood plains. With a
Force value equal to zero,the general
equation gives the following expression
for potential loss of life in a low force
or plain area:
LOL=
PAR
1 + 17)211 {PAR0u)e^159i}VT))
For cases where the population at risk
is located partly in a narrow valley area
and partly in a plain area,it may be
necessary to divide the PAR. However,
DeKay and McClelland (1993) point
out that because the loss-of-life
equations vary nonlinearly with PAR
and WT,this may lead to an
overestimation of the potential loss of
ANCOLD Guidelines for Design of Dams for Earthquake 31
life. They indicate that division of the
total population at risk into more and
more groups can easily lead to
significant overestimates of the
potential loss of life. DeKay and
McCleIland(1993) suggest that:
"the PAR should only be divided
when there are likely to be
significant differences in the
forcefulness of the flood waters
reaching the PAR within 3hr of
the dam. In general,no more than
two subpopulations should be
used in predicting potential LOL".
Therefore,in applying these guidelines,
it is recommended that the population
at risk may be divided into two groups,
but only when there is a significant
difference in forcefulness or warning
time. Any situations that may call for a
greater division of the population at risk
should be considered for a more
in-depth study of the potential loss of
life.
(iv) Incremental Consequences of
Failure. In earthquakes,there may be
extensive property damage and loss of
life independent of the dam failure.
Therefore,when assessing the risk due
to a possible dam failure event,it is
necessary to use the incremental
consequence of failure. The
incremental loss of life for an event is
equal to the losses that would occur
with dam failure minus the losses that
would occur without dam failure for the
same extreme event. In equation form,
this statement would look like:
INCREMENTAL LOL =
LOL WITH DAM FAILURE ~
LOL WITHOUT DAM FAILURE
For some cases,the incremental loss of
life may be equal to the loss of life with
dam failure. This,however,should be
decided when the actual procedures are
being applied to a risk assessment for a
specific dam. This also needs to be
considered when considering eco
losses,and social,cultural;
environmental losses which;
attributable to the dam failure
earthquakes,there may be ext|
property damage and loss oj
independent of dam failure.
4.3Selection of the Operating Ba
Earthquake (OBE)
The OBE should relate to the life o
structure,the economic and i
consequences of damage to the dam a1
likely to be different for different parts o
dam. Typical values are in the range AE
200to 1 in 1000. For many situation^
requirements of Australian Standard AS1 f
Minimum Design Loads on Structures,P*
Earthquake Loads,will apply. The accele
coefficients for AS 1170.4are based on ai
chance of exceedance in 50years (equival
an AEP of 1 in 475). These are mo
depending on occupancy type (importanc
the structure and likely consequence
failure).
It should be noted that care must be take
selecting OBE for appurtenant structures,w
on first assessment,may not appear critic
the dam. In many cases,the ability o
appurtenant structures to operate success
after an earthquake,may be critical to whe
the dam may breach. Eg. if spillway i
become inoperable,delayed failure \
overtopping may result; inability to emp
reservoir behind a badly damaged dam
greatly increase the probability of J
earthquake breaching; a dam may be critic^
water supply,so outlet works must contin
operate,or be able to be repaired rapidly
an earthquake.
4.4Concurrent Load Combinatio
The selection of the water level in the da
the time of the earthquake should be detern
taking into account the combined probabi
of the water level (related to floods and sea
variations in the water level in the dam),an
earthquake. This will preclude,or
32 ANCOLD Guidelines for Design of Dams for Earthquake
correctly render very unlikely,the application
concurrently of low AEP floods with low AEP
earthquakes.
The joint probability of other unusual load
conditions,eg. high pore pressures soon after
construction or raising of a dam needs to be
considered.
It is also essential to consider likely water levels
in the dam after the earthquake. It may,for
example,take some months to draw down the
water level in the reservoir behind a badly
damaged dam,in which time additional
flooding may occur,leading to higher water
levels and greater chances of breaching of the
dam.
4.5 Earthquakes Induced by the
Reservoir
As discussed in Section 2.7and in ICOLD
(1983) and ICOLD (1989),there is documented
evidence to prove that impounding of a
reservoir sometimes results in an increase of
earthquake activity at or near the reservoir.
ICOLD (1983) conclude that:
-earthquakes of magnitude 5 to 6.5 were
induced in 11 of 64recorded events
-the greatest seismic events have been
associated with very large reservoirs (but
there is insufficient data to show any definite
correlation between reservoir size and depth
and seismic activity)
-the load of the reservoir is not the significant
factor,rather it is the increased pore water
pressure in faults,leading to a reduction in
shear strength over already stressed faults.
-in view of the above,a study of possible
induced seismic activity should be made at
least in cases where the reservoir exceeds
109m3in volume,or 100m in depth
ICOLD (1983and 1989) give more details and
references on this issue.
/
Reservoir induced seismicity has been recorded
in Australia.
4.6 Response Spectra and
Accelerograms
In some circumstances,particularly for the
analysis of concrete dams,and dynamic
analysis of earth and rockfill dams,response
spectra,and accelerograms suited to the site and
the design earthquakes will be needed. This is
discussed in section 7.4. Advice from
seismologists should be sought when selecting
suitable response spectra and accelerograms.
5. DESIGN OF
EMBANKMENT DAMS AND
ANALYSIS OF LIQUEFACTION
5.1 Effect of Earthquake on
Embankment Dams
Earthquakes impose additional loads on
embankment dams over those experienced
under static conditions. The earthquake loading
is of short duration,cyclic and involves motion
in the horizontal and vertical directions.
Earthquakes can affect embankment dams by
causing any of the following:
-settlement and cracking of the embankment,
particularly near the crest of the dam
-reduction of freeboard due to settlement,
which may,in the worst case,result in
overtopping of the dam
-instability of the upstream and or
downstream slopes of the dam
-differential movement between the
embankment,abutments and spillway
structures,increasing the likelihood of
leakage and piping failure
-liquefaction or loss of shear strength of
saturated granular soils in the embankment
or its foundations due to increase in pore
pressures induced by the cyclic loading of
the earthquake
-differential movements or instability on
faults or other low strength seams or defects
passing through the dam foundation
-overtopping of the dam in the event of large
tectonic movement in the reservoir basin,or
by seiches induced upstream
ANCOLD Guidelines for Design of Dams for Earthquake 33
-overtopping of the dam by waves due to
earthquake induced landslides into the
reservoir from the valley sides
-damage to outlet works passing through the
embankment leading to leakage and
potential piping erosion of the embankment.
The potential for such problems depend on:
-the seismicity of the area in which the dam
is sited,and the assessed design earthquake
-local foundation and topographic conditions
-the type and detailed construction of the dam
-the water level in the dam at the time of the
earthquake.
The amount of site investigation,design,and
additional construction measures (over those
needed for static conditions) will depend on
these factors,the hazard category and whether
the dam is existing or new.
There are four main issues to consider:
the general (or "defensive") design of the
dam,particularly the provision of filters,to
prevent or control internal erosion of the
dam and the foundation,provision of zones
with good drainage capacity (eg. free
draining rockflll)
the stability of the embankment during and
immediately after the earthquake
deformations induced by the earthquake
(settlement,cracking) and dam freeboard
the potential for liquefaction of saturated
sandy and silty soils in the foundation,and
possibly in the embankment,and how this
affects stability and deformations during and
immediately after the earthquake.
These issues are considered in the following
sections,with references which give more
detailed information.
It should be noted that many low hazard
category dams may not warrant extensive
investigation and design measures. ICOLD
(1986) quote Seed(1979):
"Since there is ample field evidence that
well-built dams can withstand moderate
shaking with peak accelerations up to at
least 0.2g,with no harmful effects,we
should not waste our time and money
analysing this type of problem -rati
we should concentrate our efforts
those dams likely to present problei
either because of strong shaki
involving accelerations well in excess
0.2g,or because they incorporate lar
bodies of cohesionless materii
(usually sands) which,if saturated,m
lose most of their strength duri
earthquake shaking and,thereby,lead
undesirable movements".
In other words,it is only the potent
liquefaction which is critical for such dam
It is important to consider what is me
"well built dams" in applying this j
statement.
Since one of the main effects of earthqual
induce cracking,this will increas
likelihood of piping failure,so earth,am
and rockflll dams which are built withoi
designed filters may not be considei
"well-built". Similarly,care should be ta
considering as "well built" some of Aus
older puddled core,and concrete con
dams,which have poor erosion control,a
usually poorly compacted.
5.2 General ("Defensive") Desig
Principles for Embankment Dams
ICOLD (1986),Sherard (1967),Seed |
and Finn (1993) present lists of dei
design measures. The general philosoph
apply logical,commonsense measures
design of the dam,to take account <
cracking,settlement,and displacements
may occur as the result of an earthquake,
measures are at least as important (pr<
more so) as attempting to calculate acci
the stability during earthquake,or the
deformations. The most important me
which can be taken are:
(a) Provide ample freeboard,above i
operating levels,to allow for sett
or slumping or fault movements;
displace the crest. For exampl
might adopt a narrow spillwa]
large flood rise (and,hence,
34 ANCOLD Guidelines for Design ofDams for Earthquake
freeboard) instead of a wide spillway
with small flood rise and thus usually a
lower freeboard,provided the costs
were similar.
(b)Use well designed and constructed
filters downstream of the earthfill core
(and correctly graded rockfill zones
downstream of a concrete face for
concrete face rockfill dams) to control
erosion in the event the core is cracked
in the earthquake. For larger dams,full
width filters (2.5m to 3m) might be
adopted instead of narrower (1.5m say)
filters placed by spreader boxes. This
would give greater security in the event
of large crest deformations.
(c)Provide ample drainage zones to allow
for discharge of flow through possible
cracks in the core. For example,ensure
that at least part of the rockfill is free
draining,or that extra discharge
capacity is provided in the vertical and
horizontal drains for an earthfill dam
with such drains.
(d)Avoid,densify,drain (to be
non-saturated) or remove potentially
liquefiable materials in the foundation
or in the embankment.
There are a number of other measures which are
listed by Sherard (1967),Seed (1979) and Finn
(1993). These include:
(e)Use a well graded filter zone upstream
of the core to act as a crack stopper,
possibly only to be applied in the upper
part of the dam.. The concept is that in
the event that major cracking of the
core occurs in an earthquake this filter
material will wash into the cracks,
limiting flow,and preventing
enlargement of the crack. If well
designed filters are provided
downstream,this upstream filter is of
/secondary importance.
(f)Provide crest details which will
minimise erosion in the event of
overtopping,eg. by having a wide crest,
and if the core is a highly dispersive
soil,it could be modified with lime to
make it non dispersive. A related issue
is the detailing of the crest to control
internal erosion in the event of
settlement and cracking in an
earthquake. For example,it will be
advantageous to take filters to the crest
level,not just to within say one metre
of the crest. Also,the common practice
of using steeper slopes near the crest
(usually only to provide for camber of
the crest),and the use of wave walls in
concrete face rockfill dams,will make
the crest more susceptible to damage
(Sherard,1967).
(g)Flare the embankment core at abutment
contacts,where cracking can be
expected,in order to provide longer
seepage paths. Just as (or more)
important,is to consider the detailing of
the contact with concrete walls,and the
provision of filters downstream of the
contacts. This detailing is discussed in
Fell etal (1992).
(h)Locate the core to minimize the degree
of saturation of materials (eg. use
sloping upstream core). (Finn (1993)
also suggests positioning chimney
drains near the central section of the
embankment). These measures are
intended to reduce to a minimum the
extent of saturated zones which are
more likely to reduce strength on cyclic
loading. It is particularly relevant
where sand,silty sand and sand-gravel
soils are present as these are most
susceptible to liquefaction.
(i)Stabilize slopes around the reservoir
rim (and appurtenant structures,such as
spillways) to prevent slides into the
reservoir.
(j) Provide special details if there is danger
of movement along faults or seams in
the foundation.
(k) Site the dam on a rock foundation,
rather than soil foundation (particularly
ANCOLD Guidelines for Design ofDams for Earthquake
35
if it is potentially liquefiable) where the
option is available.
(1) Use well graded (densely compacted)
sand/gravel/fmes,or highly plastic clay
for the core,rather than clay of low
plasticity (if the option is available)
(Sherard,1967).
When assessing an existing dam,the use of
these "defensive design" measures is seldom
practical (except in remedial works). However,
it is useful to gauge the degree of security the
existing dam presents by comparing it with this
list. Where the dam fails to meet many or most
of these features,particularly (a) to (d),this may
be a better guide than a lot of analysis,to the
fact that the dam may not be very secure against
earthquake.
Apart from the defensive design measures,there
are some other general points which can be
made:
(m) Some dam zonings are inherently more
earthquake resistant than other
embankment dam types. In general,the
following would be in order of
decreasing resistance:
concrete face rockfill
sloping upstream core earth and
rockfill
central core earth and rockfill
earthfill with chimney and
horizontal drains
zoned earth-earth rockfill (without
filters)
homogeneous earthfill (without
filters).
(n) Consideration should be given to the
effect of earthquake on the strength of
dam foundation,particularly where
there are weak,or strain weakening
seams of rock or clay in the foundation.
(o) Seed et al (1985) suggest that the slopes
of concrete face rockfill dams should
be flattened to limit displacements in
earthquake. They suggest the
following:
Assuming that for areas with
earthquakes magnitude 6.5 or le
maximum ground acceleration!
than 0.3g,the slopes are 1.35
horizontal:! vertical,then
-for design earthquakes magi
6.5 to 7.5,and ground accelera
up to 0.5g,slopes should
flattened to about 1.65H:IV
-for design earthquakes magnii
7.5 to 8.5,with gfj
accelerations up to 0.5g,si
should be flattened to
1.8H:IV.
They suggest these are only general guidel
and that these flatter slopes may only nei
apply to the upper part of the dam
accelerations are the greatest. As pointej
by Finn (1993) this general philosop!
defensive design may be applied to other
of dams. However,the use of such
slopes has not been widely adopted.
5.3Liquefaction of Dam
Embankments and Foundations
One of the most critical issues relating i
effect of earthquakes on dams is whe
liquefaction of the dam or foundation
occur,and if so,what the consequences ma|
Historically,liquefaction has been the m|
cause of dam failures due to earthquake.
The following discussion outlines what is m|
by liquefaction,how it can be assessed,!
measures which can be taken to overcome if
The subject is a broad one,which is subjel
continuing research. Some references w|
give overviews include USNRC (1985),Uj
(1989),Finn (1993),Fell et al (1992),J
Robertson and Fear (1995 and 1996).
5.3.1 Definitions,and the Mechanics i
Liquefaction
The USNRC (1985) gives the folio!
definitions of liquefaction and r<
phenomena:
36 ANCOLD Guidelines for Design of Dams for Earthquake
"The word liquefaction is used to include all
phenomena giving rise to a loss of shearing
resistance and to the development of
excessive strains as a result of transient or
repeated disturbance of saturated
cohesionless soils".
"Initial liquefaction" is the condition when
effective stress momentarily is zero. This
may be considered as the condition when
pore pressure equals total vertical stress.
"Flow failures" describe the condition where
the soil mass deforms continuously under a
shear stress equal to the static shear stress
applied to it,eg. slope instability,total
bearing capacity failure.
"Deformation failures" involve large
permanent displacement or settlement,but
the earth mass remains stable without great
changes of geometry.
Robertson and Fear (1995,1996) point out that
flow liquefaction only applies to strain
weakening soils,and may occur under static,as
well as cyclic loading. They use the term cyclic
softening,split into cyclic liquefaction (in
which zero effective stress is reached) and
cyclic mobility (in which zero effective stress
does not develop) to describe the "deformation
failures" condition.
Extensive laboratory testing has been carried
out by many researchers to explain the
phenomenon of liquefaction. These show that:
(a)Cyclic loading causes densification of
dry granular soils by particle
rearrangement due to the back and forth
straining. If,however,the soil is
saturated and not allowed to drain
during cyclic loading,the decreases in
volume cannot occur and the tendency
to decrease volume is counteracted by
an increase in pore pressure and
decrease in effective stress and
strength.
The pore pressures build up gradually
with the number of cycles of loading,
and only if the pore pressures build up
to equal the total stress does the "initial
liquefaction" (effective stress <7=0)
condition occur.
(b)The number of cycles (of shearing of
the soil) to reach the cf=0condition
depends on the relative density and the
magnitude of the cyclic stress
compared to initial stress xJa'vo where
tc =cyclic shear stress and avo =
vertical stress.
Figure 8 shows the results of laboratory
testing on a saturated sand which shows
these effects. It should be noted that
the number of cycles of loading in an
earthquake depends on the magnitude
of the earthquake. Hence,larger
magnitude earthquakes induce more
cycles of shearing,and are more likely
to induce liquefactions. This is
discussed further in Section 5.3.3.
/
ANCOLD Guidelines for Design of Dams for Earthquake
37
Ntimbar of Cyclts,Ne
Figure 8. Cyclic shear stress ratio iJo\a versus number of cycles and relative density to
liquefaction from laboratory tests (De Alba etal,1976; USNRC,1985),
It will also be apparent that loose sands are
more susceptible to liquefaction than
dense sands.
(c) Laboratory tests have shown also that
the behaviour of saturated cohesionless
soils under cyclic loading is dependent
on stress history,initial stress
conditions (eg. isotropic or anisotropic
loading),and particle grading and
assemblage. It is difficult to simulate
field behaviour in the laboratory,
mainly because it is very difficult to
obtain undisturbed samples of the soils
for testing.
Soils which have been subject to liquefaction do
retain some strength. An understanding of the
stress-strain relationship of these soils in
undrained loading is critical to the
understanding of liquefaction,and is described
in some detail in USER (1989(a)) and USNRC
(1985). Figure 9 illustrates the important
points.
Figure 9(a) illustrates the stress-strain behaviour
of a saturated cohesionless soil loaded in an
undrained state,where the soil is contractive in
behaviour and strain weakening (at large
strains) during shearing. The plots assume that
the loading begins at the static shear stress Td.
Figure 10shows conditions where such a static
shear stress may occur.
As the soil is sheared,positive pore prs
are generated as the soil tries to conti
volume,but cannot do so in the und
condition. In the monotonic loading cai
available strength is greater than the statii
stress,so if this was representative of a
the slope would be stable. On the right:
Figure 9(a) is shown the effect of cyclic li
from an earthquake. If this induces sui
strain,the available undrained strength s
less than the static shear stress,anc
continuing deformation will occur und
static loading,with further reduction in st
until the residual undrained strength is r<
(Sus). (Sus is also known as the stead;
undrained strength or "residual strength")
is a case of flow failure,and would likel)
in large deformations in a dam.
Figure 9(b) illustrates the stress strain beh
of a saturated cohesionless soil loaded
undrained state,where the soil is dilaf
behaviour during shearing. In this case,
monotonic undrained loading,negativf
pressures are generated,so the soil contir
increase in strength until a maximi
reached. In this case,cyclic loading le
larger strains,but a stable,strain har
loading condition remains at the end of cy
38 ANCOLD Guidelines for Design of Dams for Earthquake
(a)
Cyclic
Loading
INSTABILITY and FLOW
i 1
(b)DEFORMATIONS of STABLE SOIL
T =Shear Stressf =Shear Strain
Td =Static (Driving) Shear Stress Sus =Undrained Steady State Strength
Figure 9. (a) Unstable;
USNRC,1985).
and (b) Stable behaviour under static and cyclic loading (Castro,1976;
Whether or not a soil will be contractive or
dilative in behaviour depends on its void ratio
(which is related to the relative density and the
stress conditions). Whether it will reach a flow
failure
condition will depend on the static shear stress,
the undrained strength vs strain behaviour,and
the amount of strain induced by cycling of load.
In practice,some drainage may occur during
the earthquake loading,particularly if the soils
are highly permeable,and/or if the drainage
path is very short. This drainage dissipates pore
pressures,reduces strains,and hence lessens the
likelihood of a flow failure condition.
It should be noted that:
Soils which have liquefied,can liquefy
again,so one cannot rule out the possibility
of liquefaction on this basis.
Liquefaction is an issue in Australia -it has
been observed during earthquakes in the
South Australia-Victoria border region,and
the larger magnitude Australian earthquakes
are quite sufficient to cause liquefaction.
Australian earthquakes may have somewhat
different characteristics to those in other
countries,but it is considered that the
methods for assessing liquefaction potential
described below are sufficiently accurate for
use in Australia.
/
ANCOLD Guidelines for Design of Dams for Earthquake 39
(a)
(b)
FAILURE
SURFACE
SAND
IMPERMEABLE
SOIL
lc)
. . . SAND': . .
Figure 10. Examples of situations involving the existence of static shear stress in the soil: (a):
ground; (b) embankment on level ground; (c) earth dam.
5.3.2 Soils Susceptible to Liquefaction
Saturated sands,silty sands,silts and gravelly
sands are susceptible to liquefaction. Figures
11 and 12,reproduced from USNRC (1985),
show the particle size envelopes for potentially
liquefiable natural soils,and mine tailings.
However,later experience has shown that even
soils with small amounts of clay may liquefy.
It can be seen that the mine tailings ar
susceptible to liquefaction than natura
possibly reflecting their uniform size am
deposition. Troncoso (1990) and Tronco
(1988) present some evidence that tailir
"age" and develop greater resistai
liquefaction with time. The behavi
tailings can be considered to be represe
of dredged fills.
40 ANCOLD Guidelines for Design of Dams for Earthquake
Figure 11. Limits in the gradation curves separating most liquefiable and potentially liquefiable soils.
(Tsuchida,1970; USNRC,1985). Note that soils outside these boundaries may liquefy.
Generally the presence of fines (silt and clay
size particles passing 0.075mm sieve) reduces
the susceptibility to liquefaction. USSR
(1989(a)) indicate that soils with clay fines (as
opposed to silt) are susceptible to liquefaction,
but the soil will not be subject to liquefaction if:
Clay content >20%
or Water content <0.9 (liquid limit)
(Note USSR (1989(a)) define
clay content as % by weight
passing 0.005mm,not the
usual passing 0.002mm
definition)
They indicate that soils are potentially
liquefiable if
and
or
Clay content
Liquid limit
Water content
<15%
<35%
>0.9 (liquid limit)
Figure 12. Ranges of grain sizes for mine tailings slimes with low resistance to liquefaction. (Ishihara,
1985,USNRC,1985).
ANCOLD Guidelines for Design of Dams for Earthquake
41
5.3.3Seed Method for Assessing
Liquefaction
The most widely accepted,simplest and most
practical method of assessing whether there is a
potential for liquefaction for horizontal ground
conditions,has been developed by Seed and his
co-researchers over many years. The method is
semi-empirical,and is based on the maximum
acceleration induced by the earthquake a,,,.,^ the
SPT 'N' value corrected for the SPT hammer
energy and for overburden pressure (Ni)60,
earthquake magnitude (M),and fines content of
the soil (% passing 0.075mm). It is based on
recorded cases of liquefaction during
earthquakes in USA,Japan and China. Details
are given in Seed and De Alba (1986) and in
USNRC (1985). The method is recommended
for the initial assessment of whether a soil will
liquefy.
The steps are:
(a)Estimate a^ for the site,ie. maximum
acceleration at the ground surface
during the design earthquake.
(b)Estimate average cyclic stress ratio
Tav/a'0induced by the earthquake from
=
0-65amaxcTorrf
where tav =average peak shear stress
a^ =maximum acceleration
ground surface
g'0=effective overburden stre;
depth under consideration
ct0=total overburden stress al
same depth
rd =stress reduction factor =
ground surface and 0.9 at 1
g =acceleration due to grs
(9.81m/sec?)
(c) Determine the stress ratio (t,
which will lead to liquefaction,by
following procedure:
(i) Determine (N,)^,from the measi
SPT 'N' value,corrected to 60% em
ratio from Table 12 and to 100
overburden stress using Figure 13.
ie.
and
N,=CnN,
N,
60
:N
ERm
60
where Nm =measured SPT valu
ER,,,=measured rod enerj
Vog
Table 12. Summary of Energy Ratios for SPT Procedures (Seed and De Alba,1986).
Country Hammer Hammer Release Estimated Rod Correction Factor fl
Type Energy (%) 60% Rod Energy f
Japan Donut Free-fall 78 78/60=1.30|
Donut Rope and pulley with 67 67/60=1.12 J
special throw release
United States Safety Rope and pulley 60 60/60=1.00I
Donut Rope and pulley
45 45/60=0.75 J
Argentina Donut Rope and pulley 45 45/60=0.75 1
China Donut Free-fall 60 60/60=1.001
Donut Rope and pulley
50
50/60=0.831
(a) Japanese SPT results have additional
correction for borehole diameter and
frequency effects.
(b) The safety hammer is the preva|
method in the United States.
42 ANCOLD Guidelines for Design ofDams for Earthquake
(c) Pilcon-type hammers develop an energy
ratio of about 60%.
Australian hammers are generally
activated by a trip deyice and are free
fall. Some recent measurements in
USA reported in Drumrightet al (1996)
indicates free fall US safety hammers
gave an energy ratio of 0.9 to 0.95,and
reported tests by Kovacs (1994)
showing energy ratios of 0.8. Tests on
hammers used recently on an
Australian Dam gave Er 59% to 61%.
Robertson and Fear (1996) recommend
that Er be measured directly because it
is influenced by the type of equipment
and its condition.
In the absence of measurements,it will
probably be reasonable to assume that
Australian free fall hammers have an
energy ratio of 60%. It is strongly
recommended that the actual energy ratio
be measured where liquefaction is being
assessed.
(ii) Determinewhich will lead to
liquefaction for a magnitude 7.5
earthquake (from Figures 14or 15.
Figure 14applies to sand with less than 5%
fines passing 0.075mm. Figure 15 applies
where there are more fines. Note that the fines
reduce susceptibility to liquefaction. The
influence of clay in the fines is discussed in
Section 5.3.2.
Cm
Figure 13. Graphs for value of overburden correction factor CN (Seed and De Alba,1986).
/
ANCOLD Guidelines for Design of Dams for Earthquake
43
OjEt
OS
v0"
0.3
.02
10
FINES CONTENTS 5*
OmmC* Id* cMM>0)
jjquiiocScn
Pm+minca\ Mo
Jogonttt daw (
CltMH doll ,A ^4
>Iw^BeNan
2030
*0
40
Figure 14. Relationship between stress ratios causing liquefaction and (N,),#values for clean sa
magnitude 7.5 earthquakes. (Seed and De Alba,1986).
2030
<N*0
Figure 15. Relationship between stress ratios causing liquefaction and (N,)^ values for silty sa
magnitude 7.5 earthquakes (Seed and De Alba,1986).
(iii) For earthquakes of magnitude other
than 7.5,correct the values of Tav/a'0by
the factors in Table 13.
44 ANCOLD Guidelines for Design of Dams for Earthquake
Table 13. Representative Number of Cycles and Corresponding Correction Factors (Seed and De
Alba,1986).
Earthquake
Magnitude (M)
Number of
Representative Cycles
at 0.65 t^
Factor to Correct
Abscissa of Curve in
Figures 14and 15
8.5 26 0.89
7.5 15 1.00
6.75 10 1.13
6.0 5-6 1.32
5.25 2-3 1.5
It should be noted that the magnitude of
the earthquake has a significant
influence on whether liquefaction will
occur. It may be assumed that
liquefaction will not occur for
earthquakes of Magnitude 5 or less
(there are not sufficient cycles of
loading for small earthquakes).
However,where static liquefaction may
occur,even small earthquakes could
trigger failure. Morgenstem (1995)
discusses static liquefaction which is
likely to occur only in very
loose/poorly compacted,saturated soils
including dredged sand,mine
overburden dumps and mine tailings.
Salmon (1995) points out that when
earthquake loading is determined by a
probabilistic analysis of the history of
earthquakes in the seismotectonic zone
around the dam (as is done in
Australia),it is possible to determine
the contribution of different magnitude
earthquakes to the assessed ground
motion. Figure 16 gives an example
which shows that the major
contribution to the estimated ground
motion for a given PGA (in this case
the 1 in 1000PGA) is from small
magnitude earthquakes near the dam.
Given the nonlinear relationship
implied in Table 13,and Figures 14and
15,this has an important influence on
the assessment of the probability of
liquefaction,and should,if practicable,
be included in the assessment.
/
ANCOLD Guidelines for Design of Dams for Earthquake 45
50
GO
OS
I t
o
10
50
Suniat9 0
MAGNITUDE CONTRIBUIIONS
-t-
Sums5 6
lb.
345 6 7
MACNIIUOE (M)
oisiANcr coNfRinunnN^
NnleConUfbolKjftS /ro<nfalonte',<i
Kimptd!othe'a?l mlevnl (?!/0Vnt)
TVvn^.
0204060SO 100120140160ISO 200220240
OISIANCC (km)
Magnitude-distancecontributions (MurrinSubstation)
Fromshallow seismogemc zones loPGA at P-0001
Figure 16. Magnitude-distance contribution (Sail
(d) Compare the estimated ija'0for the
design earthquake with that required to
cause liquefaction.
The "factor of safety" to be applied to
this depends on the degree of
conservatism in selecting a^ and may
be between 1 and 1.3(on tjo'0).
The Seed approach has been extended to use of
the Static Cone Penetration test (CPT) by
correcting CPT q,. values to SPT (N,)60using
Figure 17. Alternatively Figure 18 can be used
in lieu of Figure 15. However,recent
i,1995).
experience on two dam sites in Australia s
the SPT 'N' value vs qc relationships shoi
Figure 17,and implied in Figure 18 d<
apply,so considerable care should be tak
using CPT values. A further problem is th;
unlike the SPT,where a sample is recover!
(and a % fines can be determined direct
the SPT (N,)^ value) for CPT,fines co
must be determined from other data,
brings in a potential for error in the assess
The CPT does have the advantage of pro
continuous data,so a good approach is '
both CPT and SPT on the site.
46
ANCOLD Guidelines for Design of Dams for Earthquake
-4
o
a.
c
o
2 3
-0 O JomfoUowtki tf oL 09091
0MurorrocM andKoboroiM UM2)
A (ihihofoandK090(Oil)
it Robvrlion(1902)
V MHehtll (1903)
O HoriNr >oL (1964)
ne'
bkul/lool
OOI 002 00301 0205 I
MeanGroinSit*,Oj0-mm
Figure 17. Variation of ratio with mean grain size (qc) measured in tsf 100kPa) (Seed and De
Alba,1986).
06
V 0'
Z03
>
u
o^
o.i
-i 1 1i-
M> 7.3orthquaku
% riots >33213slO 13
Ojolmm) 01 0.2 0.230230408
le/Ngo 3.341 4.44,44^33
_1_ X
4080120160200
ModifiadCon*P*n*tralionR*iilanc*. qe|-lf
240
Figure 18. Relationship between stress ratio causing liquefaction and cone tip resistance for sands and
silty sands (1 tsf100kPa (Seed and De Alba,1986).
There are several problems in applying the Seed
et al semi empirical approach:
(a) It has been developed for level or near
level ground conditions. For most dam
applications it can therefore only be
directly applied to assess whether
liquefaction would occur without the
dam.
ANCOLD Guidelines for Design of Dams for Earthquake 47
Seed and Harder (1990) recommend
that to allow for the static driving stress
due to the dam,a correction factor
should be applied to the calculated
(Tav/CT'0). This is calculatedusing
(xayCT'0)a^=(xav/a'0)a^ Ka
where Ka is a correction factor
determined from Figure 19. To
determine Ka,the relative density and
a (the ratio of static driving shear stress
on a horizontal plane to the initial
effective overburden stress,(tav/a'0) has
to be determined,eg. from finite
element analyses.
The correction only applies to soils
where Tav/a'0<300kPa. At higher
initial effective stresses,soils will be
dilative or more contractive,and
values will apply.
It will be noted that for soils y
relative density greater than 45%,
greater than unity,so the effect (
in-situ shear stress is to lessei
likelihood of liquefaction.
(b) Seed and Harder (1990)
recommend that for effi
overburden stresses greater thar
kPa,a further correction is requii
the calculated xja'^ Thisiscalci
using
(tav/o' 0)0*0=tfoo)&o=lOOkPa *KJ
where K^ is a correction factor determinec
a',,and Figure 20.
2.0
Ora. 55 -70%y
X
Op ^ BOOkPa
Thv
oc
Thv
0.1 0.20.30.4
OC
0.5
Figure 19. Correction factor K,,to account for static driving shear stresses (Seed and Harder,1990
48 ANCOLD Guidelines for Design of Dams for Earthquake
i
(c) In most natural soil deposits the SPT
values are variable (over that increase
which occurs with the increasing
overburden stresses). There are no
clear guidelines given by Seed et al as
to how this should be accounted for.
USSR (1989(a)) discuss this issue and
present some useful practical points
which are recommended for use. These
include:
The results of each interval of each
drill hole,with regard to liquefaction
potential,should be prepared in a
table and should be presented on
geologic cross sections and profiles
to allow examination of the
frequency and continuity of those
intervals indicating liquefaction
susceptibility. From such a
presentation a judgement is drawn
as to whether or not the continuity
of potential liquefaction intervals
indicated is great enough to be of
concern.
If the deposit being sampled is
known to contain,or may contain
gravel,these coarse particles may
increase the blow count,implying a
more dense,less potentially
liquefiable soil. To check for this,
USBR (1989(a)) recommend
recording SPT blow counts for each
300mm of penetration,and
correcting for the erratic effects of
gravel (as a minimum,the three sets
of 150mm blow counts should be
checked to see whether this potential
irregularity is present). Where
gravel is extensive,shear wave
velocity methods should be used to
assess liquefaction potential.
X
}*
*x
V
X
X
X
X
5s .* *
X
t
X
<
; X
X
:
< X
:
X
[
X
0100200300400S00600700800
EFFECTIVE CONFINING PRESSURE (To' kPa
Figure 20. Correction factor K,to account for in-situ effective stress ct'0greater than lOOkPa (Seed and
Harder,1990).
Fear and McRoberts (1995) reconsidered thethe Seed method used average SPT 'N' values in
Seed database,and determined that in general,the critical zones. The USBR procedure
ANCOLD Guidelines for Design of Dams for Earthquake
49
discussed above effectively does the same
thing,but this emphasises that it would be over
conservative to use the minimum 'N' value in
the Seed method. Fear and McRoberts (1995)
present revised graphs of liquefaction potential
based on the use of minimum 'N' values. These
would need to be used with caution since it is
not uncommon for individual SPT tests to be
affected by loosening in the drilling process
giving unrealistically low values.
Robertson and Fear (1996) indicate that at the
NCEER Workshop it was agreed that the
curves shown in Figure 21 should be adopted in
preference to those in Figures 14and 15. In
figure 21,t,. represents the limiting shear strain
at liquefaction. The 3% line is equivalent to the
design curve in Figure 14. It will be noted
there are changes in the graph at the low SPT
values,Robertson and Fear (1996) suggest that
the fines content can be allowed for by adding
the following corrections to (N,)^.
ACNOeo =7ifFC > 35%
=0ifFC <5%
ACNOao =(FC-5) if5%<FC<35%
Having adjusted for fines content,the clean
sand curve can be used.
ctOr
:%>-
in I
=
for
Ejlimettdrqn^t (S64>
_Ol
0.6
PropoiH
by Uti 119791
Bandenttit doteBy
ToiumcViuendYoihin
(13041
i"iqnesire^i coKnnei j-iiim3clT"Noti^mlieamdome?!--
Licui'eetienwitn
7^-20% -10% -a*
05-
C-f-
3k
<=1
o.-
02-
01
Ne
10 20 30 40 50
<N,)
to
Figure 21. Recommended cyclic resistance ratio (CRR) for clean sands under level ground conditions
based on SPT (Robertson and Fear,1996).
50
ANCOLD Guidelines for Design of Dams for Earthquake
It is important to note that a positive result from
the Seed method only indicates that a soil may
be susceptible to liquefaction. It does NOT
mean that just because (part of) the dam
foundation will liquefy,the dam will fail. What
should follow such an outcome,is an
assessment of the extent and continuity of
potentially liquefiable soil,and the residual
undrained strength of that soil,for use in post
liquefaction analysis.
Liao,Venziano and Whitman (1998) provide
some useful information which will help in the
application of liquefaction in a probabilistic
framework.
5.3.4Shear Wave Velocity Methods for
Assessing Liquefaction Potential
Shear wave velocity is affected by many of the
variables which influence liquefaction,eg.
(relative) density,confining pressures,stress
history,geologic age. Hence,it has some use as
an indicator of potential for liquefaction.
The shear wave velocity may be obtained by
downhole,crosshole or surface to downhole
seismic methods,or by a seismic cone
penetration test (a modification of the
piezocone test) developed by Campanella et al
(1986).
As discussed above,some coarse grained
cohesionless materials (gravels,cobbles)
suspected of being potentially liquefiable
cannot be successfully sampled using SPT. If
crosshole shear wave velocity data have been
obtained on the materials,and these data
accurately represent the deposits in plan and
section,then they provide a viable means for
making a judgement on liquefaction potential.
USBR (1989) recommend that:
if shear wave velocities are >365m/s,the
deposits may be judged non liquefiable
if shear wave velocities are between 245 and
365m/s,the deposit may be considered
likely to be non liquefiable,but supporting
evidence should be obtained
if shear wave velocities are < 245m/s,the
deposit may be judged liquefiable.
In practice,this will leave many sites in the
"grey" zone.
Shear wave velocity may also be used to assess
liquefaction by relating peak ground
acceleration,and a history of performance of
sites in earthquakes. Bierschwale and Stokoe
(1984) and USNRC (1985) give details.
USNRC (1985) also gives details of a method
which assesses peak strains due to the
earthquake from the shear wave velocity,and
compares this to threshold strain. Brief details
of these methods are given in Fell et al (1992).
5.3.5 Determination of Residual Undrained
Strength
In recent years there has been quite a lot of
discussion of the post liquefaction condition.
This is usually discussed in terms of "residual
(undrained) strength","field residual strength",
or "steady state undrained strength". Some of
these are expressed as plots of residual
undrained strength versus SPT 'N' value (eg.
Seed (1987),Lo and Klohn (1990),Seed and
Harder (1990). These plots are developed from
backanalyses of liquefaction failures. Figure 22
is the plot presented by Finn (1993).
Finn (1993) indicates that lower bound
strengths are often used for these analyses,
although 33rd percentile values have sometimes
been used. In either case,the lower bound or 33
percentile gives very low (-zero) residual
undrained strengths at SPT less than about N =
6.
An alternative approach adopted by a number of
authors including Ishihara (1994) and Finn
(1993),is to relate the normalised residual
undrained strength Su/ct'vo (where SU5 =residual
undrained strength and ct'Vo effective vertical
stress) to either SPT (N,)^ values (N values
corrected to 100kPa effective stress,60%
energy ratio hammer) or to cone penetration
resistanceqc.
ANCOLD Guidelines for Design ofDams for Earthquake 51
2000
I60C
H
O
Z
tu
cr
e
<
u
a
UJ
z
<
s
Q
Z
3
-I
<
O
a
35
1200
800
400
O
<
tTM0Ut " INDUCCO ^PJ"!V"0tVt0|>?<l-t 'fjHtO.'
JT 0*r M0DCHOUll. lTeSTH .
CANTHOUARC -IMOUCO LKiucrwriM ultT'
JPT 3ATJL AMD KtJIOUAI. MKA-tTim "*WTII.
CO NTT RUCTION -INOUCCO UIOUtfieTION **8 tUUMt MJt KUTOmtl.
iomunr*>NooOA"
48 12 16 zo
EQUIVALENT CLEAN SANO SPT BLOWCOUNT,
Figure 22. Correlation of residual undrained shear strength with SPT (NOso (Finn,1993).
Finn (1993) indicates that values of Sus/a'vo
around 0.06 to 0.1 have been used for the design
of a number of dams. Ishihara (1994) presents
data from backanalysisof field failures showing
a lower limit of Su/a'vo of 0.05. He indicates
that laboratory tests,particularly on
reconstructed samples,gives lower bound
results.
Fear (1996) reviewed the field data used by
Seed and Stark and Mesri (1993). This shows
that the data used to produce Figure 22 is very
approximate. Many SPT 'N' values were only
estimated,and a variety of analysis methods
with often poor modelling of geotechnical
conditions were used to determine the residual
undrained strength. The range shown in Figure
22 is a true reflection of the data.
Stark and Mesri (1993) argue that these
strengths obtained from case histories are
sometimes affected by drainage,and hence
should not be used where drainage may not
occur. They recommend use of a formula
relating SJ<j'Y0to SPT (N,^ values which they
determined from laboratory tests done by a
number of researchers. Their recommended
equation is:
SuMo =0.0055 (NJXsoc
where (N,)'eocs
=SPT 'N' value corrected to
100kPa overburden stress,
and 60% energy ratio,for
clean sand (cs).
For soils with fines (passing 0.075mm),they
recommend correcting the measured (NOeo
values by adding A^,)^ taken from Table 14.
This correctipn is similar to that detailed above.
'-vYtV/'V "
52
ANCOLD Guidelines for Design of
uake
Table 14. Correction to SPT (Ni)60Values to Allow for Fines (Stark and Mesri,1993).
Fines Content
10 2.5
15 4
20 5
25 6
30 6.5
35 7
50 7
75 7
There is a diversity of views on what strength to
use,particularly at low SPT (N,)^ values. All
that can be recommended is that those involved
read the relevant papers and make some
judgement for the site under consideration. Of
the available methods,that proposed by Stark
and Mesri (1993) is the most complete,and is
recommended. For silty sands and sand with
some silt,or interlayered sand with sandy clay,
silt or clay,drainage during earthquake is
unlikely to occur,so the Stark and Mesri
method is likely to give reasonable estimates.
For uniform clean sands which may drain it is
possibly on the conservative side. It is
reasonable to expect that there will be more
papers published on this topic,as it is an
important one receiving the attention of a
number of prominent researchers. Some
information in Fear (1996) seems to indicate
that at low SPT 'N' values,Stark and Mesri is
generally acceptable,but might not be
conservative in particular situations,which
cannot be defined other than by laboratory
testing. At high SPT 'N' values,Stark and
Mesri's (1993) method is likely to be
conservative,and laboratory testing is needed to
determine the actual strength.
The use of analysis for residual undrained
strength is discussed further in Section 6.4.
5.3.6 Measures to Improve Liquefaction
Resistance
USNRC (1985) reproduced in Fell et al (1992)
discuss methods for altering the hazard,
reducing the risk of liquefaction or
improvement of stability during and after
earthquake. These may apply to an existing
dam or to new dam construction. USNRC
(1985) suggest there are four general classes of
mitigation measures:
(a)Changing operational procedures for the
project,eg:
(i)lowering the maximum water level
in the reservoir
(ii)limiting public access to the area
downstream of the dam
(iii)institute early warning systems
downstream.
(b)Improve in-situ foundation conditions to
reduce liquefaction susceptibility including:
(i)excavation of existing materials of
potentially liquefiable soils of
inadequate relative density sand,and
replacement with compacted non
liquefiable soils,eg. replacement of silt
with well compacted sandy gravel,
sand,or gravel
(ii)densification and increase in in-situ
lateral stress by in-situ methods:
blasting,compaction piles;
vibroflotation or vibratory probes;
compaction grouting; dynamic
compaction (consolidation). These
methods are more likely to be applied
to new construction than to remedial
works for existing dams.
Vibroflotation and dynamic
consolidation are relatively common
procedures,and particularly in the case
of vibroflotation,high relative densities
(70% to 85%) can be achieved at
moderate cost to depths of up to 30m in
clean sands,sandy gravels and sands
with minor silt. Brown (1977) discusses
the applicability of the methods. Figure
23shows the soils which can be treated
by vibroflotation.
ANCOLD Guidelines for Design ofDams for Earthquake
53
particlesue-millimetres
Uay
tilt sand qravtl
r otin
Mn |mdium|coart |mdnjH ceont fin*|tndiuin|coars*
Gradingof soilswhich canbednsifidby vibroflotalion
Zon B Most suited
Zon C Canb treatedbut finascaneauM probims
Zon A Canb (rat*dbul grav*4eausswar inquipmtnt andlow progress
Figure 23. Soils suitable for vibroflotation(adapted from Brown,1977,and industry data).
(iii)in-situ improvement by alteration of
material,eg. mixing in-place material
with additives,eg. lime,cement or
asphalt introduced in augered piles;
removing in-place material by jetting
and replacing with suitable materials.
These methods are costly and
unlikely to find wide application
(iv)grouting or chemical stabilization.
Chemical or cement grouts may be
used to improve the strength and
stiffness of soils. Generally,grouting
of soils is difficult and costly,and not
likely to be the economic solution. It
also has the disadvantage that
permeability is reduced,lessening the
ability of the soil to dissipate pore
pressures built up by the cyclic
loading of an earthquake.
Figure 24shows schematically some of the
techniques. USNRC (1985) gives
additional details of the techniques and the
applicability.
(c) Structural solutions:
(i) berms -these may be added to an
existing dam or be incorporated
into a new dam design. The'
several effects
the effective vertical stre
the foundation (of poter
liquifiable soil) is increa
This increases the cyclic
strength and shear moduli
the static shear stress
reduced,improving
earthquake stability
the freeboard may
increased by raising the
level of an existing dam \
berm
(ii)piles,caisons,freezing
more likely to be usee
buildings than for dams bec|
the large costs involved
(iii)rebuilding the structure
extreme cases a new dam if
needed as shown in Figured
(d) Drainage solutions -drainage so
set out to relieve pore pressuresb
during the cyclic loading
earthquake by providing
permeable drainage paths,
dewatering the soil to a 0
saturated condition
54 ANCOLD Guidelines for Design of Dams for Earthquake
DENSIFICATION AND INCREASE OF
LATERAL STRESS
-COMPACTION PILES
-VIBRATORY PROBE
-VIBROFLOTAtlON
-COMPACTION GROUTING
MATERIAL IMPROVEMENT
-INPLACE MIXING WITH ADDITIVES
-INPLACE JETTING AND REPLACEMENT
GROUTING (CEMENTATION)
-CHEMICAL GROUTING
-PARTICULATE GROUTING
Figure 24. Schematic illustration of the use of some in-situ improvement techniques to reduce the
liquefaction hazard for earth dams (USNRC,1985; Silver,1985).
(i) Pressure relief wells. These may be
conventional water wells with stainless
steel well screens,spaced at
sufficiently close centres to allow
dissipation of pore pressures from the
cyclic loading. Since many alluvial
soils are stratified,the horizontal
permeability is much greater than the
vertical,so the drainage wells will
significantly improve the drainage.
Stone columns may be used in lieu of
conventional wells,and may be built
by backfilling vibroflot probe holes
with gravel. However,in this case
control of erosion of fines into the
stone columns may not be possible and
they may not be as effective as
screened wells.
In addition to helping dissipate pore
pressures built up by the cyclic
loading,pressure relief wells may be
used to reduce uplift pore pressures
under static conditions. Since
"liquefaction" is a process of reduction
of effective stresses to zero by buildup
of pore pressure,to start at lower static
pore pressures may be critically
/important to prevention of
liquefaction.
(ii) Drainage layers. Drainage layers,ie.
horizontal drains,vertical drains,
would be incorporated into the zones
of the dam embankment to ensure
that so far as practicable the
embankment is partially saturated. If
sands,eg. tailings sand are used to
construct zones which will become
saturated,layers of more permeable
sand or gravel would be included,to
dissipate pore pressures.
(iii) Dewatering and air injection. These
are shown in principle in Figure 26
and are likely to be used as remedial
measures in existing structures,and
then only when other methods are
not adequate. The major problem is
that pumping or injection of air must
be continuous which is undesirable
from dam safety and cost viewpoints.
As discussed above,it may well be good
engineering practice to incorporate into the
design one or more of these measures to
improve the resistance of a dam foundation to
liquefaction,rather than to rely on the ability to
predict what will happen in a design earthquake.
Positive measures to density loose-medium
dense sands,and/or to add a berm to reduce
static shear stresses can be carried out at a
relatively small incremental cost to a project,
and improve the degree of confidence of the
safety of the dam very significantly. In
particular the risk of flow failure can be
considerably reduced.
ANCOLD Guidelines for Design of Dams for Earthquake 55
/.////////////m/MW/M/SMmmrm
ADtOUATf FOUNDATION MIMIC STAilLfTY
CONSTRUCT DOWNSTREAM EARTH BERM
AOCOUAT*FOUNDATION ttltMIC JTA1IUTY
INCREASE FREEBOARD WITH BERM
DOWNSTREAM BOLSTER SECTION
-ROLLCRETE STRUCTURE THROUGH THE
DOWNSTREAM TOE
-REINFORCED EARTH STRUCTURE THROUGH
THE DOWNSTREAM TOE
NEW DOWNSTREAM CROSS SECTION
-PHYSICAL SEPARATION FROM EXISTING CROSS SECTION
Figure 25. Schematic illustration of of the use of structural measures to reduce the liquefaction hazard
for earth dams (USNRC,1985; Silver,1985).
Ledbetter(1985) also gives some details on methods of reducing liquefaction potential of foundations.
AIR HfAOCR
HOKIZONTAl ORAMACI
T&77777?'/,,>////////////////7////////////,
STONE COLUMNS
!LiL
V tTOtd
'////////W,/////,
STONE COLUMNS
INJECTION OF AIR
PARTIALLY SATURATE LIQUEFIABLE MATERIALS
KAVINOHCAOCflS KINC tint
f ARTIALLY tATURATfD (OILS
DEWATERING
-PARTIALLY SATURATE LIQUEFIABLE SOIL
FREEZING
-FREEZE SOILS THAT CAN LOOSE STRENGTH
Figure 26. Schematic illustration of the use of groundwater control measures to reduce the liquefaction
hazard for earth dams (USNRC,1985; Silver,1985).
56
ANCOLD Guidelines for Design of Dams for Earthquake
6. SEISMIC STABILITY
ANALYSIS OF
EMBANKMENTS
6.1 Preamble
The methods of analysis currently used in
practice to evaluate seismic stability of
embankment dams vary widely,ranging from
simple limit equilibrium type analyses to highly
sophisticated numerical modelling techniques.
For the purposes of these guidelines they are
classified as:
Pseudo-Static Analysis
Simplified Methods of Deformation
Analysis
Post Liquefaction Analysis
Numerical Modelling Techniques
-Total Stress
-Effective Stress
In this section,firstly,a brief description of
each method of analysis is provided,and then a
set of guidelines is outlined for seismic stability
assessment of embankment dams in Australia.
The simplified methods of analyses including
pseudo static,and post liquefaction,rely heavily
on the lessons learnt from the performance of
dams during past earthquakes. Major reviews
of past performance have been conduced by
Sherard (1967),Sherard et al (1974),Seed
(1979,1983) and Seed et al (1978,1985).
These studies show that:
Even at short distances from the epicentres,
there have been no complete failures of
embankments built of clay soils,but several
dams have come close to failure.
Well constructed dams of clay soils on clay
or rock foundation not susceptible to strain
weakening,can withstand extremely strong
shaking resulting from earthquakes of up to
magnitude 8.25 with peak ground
accelerations ranging from 0.35g to 0.8g.
(They suffer cracking,but few,if any,have
failed catastrophically).
Dams which have suffered complete failure
as a result of earthquake shaking have been
/constructed primarily with saturated sandy
materials or on saturated sand foundations.
Liquefaction is a major contributory factor
in these failures.
The most susceptible to failure under
earthquake loading are hydraulic fill dams,
because these are susceptible to liquefaction
if the fill is granular and saturated.
In dams constructed of saturated
cohesionless soils,the primary cause of
damage or failure is the buildup of pore
water pressure (liquefaction) under the
earthquake loading.
There are very few cases of dam failures
during the earthquake shaking. Most of the
failures occur from a few minutes up to
twenty-four hours after the earthquake.
However,cracking and displacements do
occur during the earthquake.
Based on the above and similar observations,
the US Bureau of Reclamation (1989) classifies
embankment dams into two main categories: (1)
not susceptible to liquefaction and (2)
susceptible to liquefaction For the purposes of
seismic stability assessment,they also identify
two types of analyses: deformation analysis
and post liquefaction (earthquake) analysis.
They recommend that deformation analysis be
performed on dams not susceptible to
liquefaction and post earthquake stability
analysis be performed on dams susceptible to
liquefaction. This guideline recommends that a
similar approach be adopted but adds the
proviso,that where significant strain weakening
may occur due to displacements induced by the
earthquake,post earthquake stability should be
done taking account of the strain weakening.
6.2 Pseudo-Static Analysis
Up until the 1970s,the pseudo-static analysis
was the standard method of stability assessment
for embankment dams under earthquake
loading. The approach involved a conventional
limit equilibrium stability analysis,
incorporating a horizontal inertia force to
represent the effects of earthquake loading. The
inertia force was often expressed as a product of
a seismic coefficient k and the weight of the
sliding mass W as shown in Figure 27. The
larger the inertia force,the smaller the safety
factor under the seismic conditions. In this
approach,a factor of safety (FOS) of less than
one implied failure,whereas F0S>1 represents
seismically safe conditions,although as shown
in Table 15,a higher factor of safety was often
utilised to take into account material
degradation under cyclic loading.
ANCOLD Guidelines for Design of Dams for Earthquake
57
Figure 27. Pseudo-static method of assessing seismic stability of embankments.
The seismic coefficients used in this approach
were typically less than 0.2 and were related to
the relative seismic activity of the areas to
which they apply. In the United States,for
example,they ranged from 0.05 to 0.
Japan they have characteristicallybeen 1<
about 0.2,and similar values have been
other highly seismic regions through
world,as shown in Table 15.
Table 15. Seismic coefficients used in selected embankment dams (Seed,1979).
Horizontal seismic
coefficient
0.1
0.1
0.1
0.1
0.1
0.12
0.1
0.12
0.15
0.1
0.12 to 0.2
0.12
0.15
0.12
Dam
Aviemore
Bersemisnoi
Digma
Globocica
Karamauri
Kisenyama
Mica
Misakubo
Netzahualcoyote
Oroville
Paloma
Ramganga
Tercan
Yeso
Country
New Zealand
Canada
Chile
Yugoslavia
Turkey
Japan
Canada
Japan
Mexico
USA
Chile
India
Turkey
Chile
For dams not susceptible to liquefaction,Seed
(1979) suggested a seismic coefficient of 0.1 g
for magnitude 6.5 earthquakes and 0.15g for
magnitude 8.25 earthquakes to obtain a safety
factor of 1.15. However,he qualified this
recommendation to point out that it applied to
"most" earthquakes,and was "often adequate".
The cases shown in Table 15 are all relatively
old,and of interest only in giving the history of
developmentof the methods of analysis.
The US Army Corps of Engineers (198'
extended the basic pseudo-static method^
as a screening method for dams not susc
to liquefaction. They recommend us
seismic coefficient equal to one-half c
ground acceleration and the use of uni
conditions for cohesive soils and <
conditions for free draining granular mi
with a 20percent strength reduction to al
strain weakening during the earthquake 1<
They require a factor of safety greater th
If a dam fails to satisfy this,more accur
58
ANCOLD Guidelines for Design of Dams for Earthquake
detailed analyses are required. Their approach
has been calibrated against a large number of
deformation analyses,and they state that up to
1m of deformation may occur.
The pseudo-static method of analysis,despite
its earlier popularity,was based on a number of
restrictive assumptions. For instance,it
assumed that the seismic coefficient acting on
the potential unstable mass is permanent and in
one direction only. In reality,earthquake
accelerations are cyclic,with direction
reversals. Therefore,the concept it conveyed of
earthquake effects on embankments was very
inaccurate. Also,the concept of failure used in
the approach was influenced by that used in
static problems. It is clear that a factor of safety
of less than one cannot be permitted under static
conditions as the stresses producing this stage
will exist until large deformations change the
geometry of the structure. However,under
seismic conditions,it may be possible to allow
the FOS to drop below one,as this state exists
only for a short time. During this time,
earthquake induced inertia forces cause the
potential unstable masses to move down the
slope. However,before any significant
movement takes place,the direction of the
inertia forces is reversed and the movement of
the soil masses stop and once again,the FOS
rises above one. In fact,experience shows that
a slope may remain stable despite having a
calculated FOS less than one and it may fail at
F0S>1,depending on the dynamic
characteristics of the slope-forming material.
The deficiencies associated with the
pseudo-static analysis were clearly
demonstrated during the San Fernando
Earthquake (M=6.6) in 1971. In this
earthquake,the Lower San Fernando Dam
experienced a massive slide in its upstream
shell. This slide was very significant as the
seismic stability of the Lower San Fernando
Dam had been evaluated only five years before
the earthquake and a number of reputable
design agencies had concluded that the dam was
safe against any earthquake that it might be
subjected to. Clearly this was not the case and
the dam failed despite having a pseudo-static
FOS of around 1.3because liquefaction
occurred,with resultant large loss of shear
strength. This near-disastrous event,more than
any other single event,resulted in an extensive
re-appraisal and gradual demise of the
pseudo-static analysis.
Today,it is generally accepted that the
pseudo-static analysis is not an accurate tool of
seismic stability assessment of embankment
dams,and that it may only be used as a
screening tool (for dams not susceptible to
liquefaction). It is recommended that the US
Corps of Engineers (1984) method may be used
as a screening method for well constructed earth
and earth and rockfill dams,which are not
susceptible to liquefaction or significant strain
weakening in the dam or its foundation.
6.3Simplified Methods of
Deformation Analysis
6.3.1 Initial Screening
The US Bureau of Reclamation (1989)
recommend that for dams not susceptible to
liquefaction,dynamic deformations should not
be a problem,and need not be analysed for such
if the following conditions are satisfied.
1. The dam is a well built dam (densely
compacted),and peak accelerations at
the base of the dam are 0.2g (gravity) or
less; or the dam is constructed of clay
soils,is on clay or rock foundations,
and peak accelerations are 0.35g or less.
2. The slopes of the dam are 3:1 (H:V) or
flatter.
3. The static factors of safety of the
critical failure surfaces involving loss
of crest elevation,(ie. other than the
infinite slope case) are greater than 1.5
under loading conditions expected prior
to an earthquake.
4. The freeboard at the time of the
earthquake is a minimum of 2 to 3
percent of the embankment height (not
less than three feet (0.9m)). Fault
displacement and reservoir seiches with
regard to freeboard should be
considered as separate problems.
These criteria are quite conservative,
particularly for dams with rock fill zones,and
ANCOLD Guidelines for Design of Dams for Earthquake 59
are seldom met. However they,along with the
US Corps of Engineers (1984) pseudo-static
method,may be used to determine whether
more detailed analyses are required. If they are,
the normal second step for dams not susceptible
to liquefaction will be to use the Newmark
(1965) and Makdisi and Seed (1978)
approaches. Care needs to be taken with
screening methods if there are strain weakening
soils in the dam or its foundation.
6,3.2 The Newmark Approach
In 1965,Newmark introduced the basic
elements of a procedure for evaluatingthe
potential deformations of an embankme
earthquake loading (Newmark,1965).
contribution,sliding of a soil mass along
failure surface was likened to slippitf
block on an inclined plane. He envisa
failure would initiate and movements^
begin when the inertia forces exceed V
resistance,and that movements woul
when the inertia forces were reversed. T
proposed that once the yield accelerate
the acceleration time history of a slippirf
are determined the permanent displac
can be calculated by double integratf
acceleration history above the
acceleration,Figure 28.
Disp. of
Soil Block'
Toward
Bluff
Figure 28. Double integration method for determination of the permanent deformation period
embankment,(Newmark 1965).
According to this approach,the permanent
deformation in a sliding mass is a function of:
The amplitude of the average acceleration
time history of the sliding mass,which in
turn is a function of the base motion,the
amplifying factor of the embankment and
the location of the sliding mass within the
embankment.
The duration of the earthquake,whic
function of the magnitude of the earthc
Yield acceleration of the potential s
mass.
The validity of the basic principl
Newmarks's approach has been demon
by many investigators [eg. Goodman anc
(1966),Ambraseys (1973),Sarma (197
Makdisi and Seed (1978)]. It is generally
60
ANCOLD Guidelines for Design of Dams for Earthquake
that,provided the yield acceleration is
accurately evaluated,the approach can estimate
the permanent displacement of a soil mass in
reasonably good agreement with those observed
during past earthquakes.
However,it should be noted that Newmark's
approach was developed at a time when there
were no direct methods of computing
permanent deformations. Today,deformation
calculations can be made directly using
advanced,readily available,numerical codes,
giving a global picture of dam behaviour. It
must also be stressed that Newmark's approach
is limited in application to compacted clayey
embankments and dry or dense cohesionless
soils that experience very little reduction in
strength due to cyclic loading. The approach
should not be applied where embankments or
their foundations are susceptible to liquefaction
or strain weakening because it will significantly
underestimatedisplacements.
6.3.3Makdisi and Seed Analysis
The Makdisi and Seed (1978) approach is based
on Newmark's method,but modified to allow
for the dynamic response of the embankment as
proposed by Seed and Martin (1966). The
approach was developed based on a series of
deformation analyses performed on a large
number of embankments subjected to
earthquake loading. The approach involves the
following main steps:
(a)Determine y/h ratio for the potential
sliding mass,where y is the depth to the
base of the sliding mass and h is the
embankment height.
(b)Calculate the yield acceleration ky for
the potential sliding mass.
(c)Determine the maximum crest
acceleration U*and the predominant
period of the embankment T0(in
seconds).
(d)/Determine the maximum value of the
acceleration history kmax using the
normalised relationship given in Figure
29 and the values calculated in steps (a)
and (c),
(e) Enter Figure 30with the calculated
values of kmax and T0to determine the
horizontal component of earthquake
induced permanent displacement,U,in
the potential sliding mass.
It should be noted that an important step in
determination of kn,ax in step (d) is to establish
the dynamic properties of the material forming
the embankment and the foundation. This can
be achieved by:
triaxial compression tests,simple shear tests
or torsional shear tests conducted under
cyclic loading conditions
resonant column testing
field measurement of shear wave velocities,
either by downhole or crosshole techniques,
or
back calculation using finite element
techniques,modelling measured responses
to earthquake events
empirical relationships,such as those given
by Hardin and Drenerich (1972) and Seed et
al(1986).
However,given that the laboratory and field
based methods are expensive,it is
recommended that the values of kmax be initially
calculated using a range of Gmax values obtained
from the empirical relationships. Should the
results of the analysis be unacceptable then a
more elaborate program of laboratory and/or
field testing may be warranted.
It should be noted that relating satisfactory dam
performance to earthquake induced deformation
is very subjective,and generally depends on
dam specific criteria about the allowable loss of
freeboard,or the tolerable extent of horizontal
displacements.
The Makdisi and Seed approach,is widely used
and accepted among practising engineers.
However,like Newmark's approach,it is
limited in application to dams not susceptible to
liquefaction or strain weakening in the
embankment or its foundations.
ANCOLD Guidelines for Design of Dams for Earthquake 61
v/f
X
"N.
Figure 29. Variation of seismic coefficient k,,,,,,with depth of the base of the potential sliding
62
ANCOLD Guidelines for Design of Dams for Earthquake
VI
-o
c
o
LJ
<u
i/i
3
0.001
0.01
00001
00.2OA 0.6
mo*
Figure 30. Variation of yield acceleration with normalised permanent displacement (Makdisi and Seed,
1978).
6.4Post Liquefaction Stability and
Deformation Analysis
As stated previously,studies of past
performance of embankment dams during
earthquakes indicate that most cases of dam
failures have been associated with dams
susceptibleto liquefaction.
In order to assess post liquefaction stability of
embankments by methods such as those
described in Section 5.3,two types of analysis
are often performed: (1) limit equilibrium
analysis,and (2) deformation analysis. The
limit equilibrium analysis is performed to
calculate post liquefaction factor of safety of the
dam,and the deformation analysis is performed
to obtain a global picture of the liquefaction
induced deformation within the dam.
The deformation analysis,is increasingly
becoming an essential part of a post liquefaction
analysis. This is because;,the picture of dam
behaviour provided by such an analysis allows
the design engineer to make a better judgement
as to the extent of the remedial measures. A
factor of safety alone,in general,is not a very
good discriminating tool for deciding on the
extent of remedial work. A factor of safety of
0.9,for instance,can have very different
connotations depending on the geometry and
the extent and location of the liquefied zones. It
may imply a massive failure or it may simply
ANCOLD Guidelines for Design of Dams for Earthquake 63
n
result in a limited movement along the slip
surface before the embankment attains a new
stable geometry. But,displacements can be
interpreted in terms of performance criteria
such as the allowable loss of freeboard and/or
the tolerable deformation,provided one keeps
in mind the limitations of accuracy of the
methods being used.
6.4,1 Post Liquefaction Limit Equilibrium
Analysis
The basic steps adopted in a post-liquefaction
limit equilibrium analysis are as follows:
(1)Identify soils susceptibleto liquefaction
within the embankment and the
foundation by methods such as those
described in Section 5.3.
(2)Evaluate earthquake induced excess
pore pressures within the liquefiable
materials (note ru =1 implies
liquefaction).
(3)Conduct a conventional limit
equilibrium stability analysis using the
pore pressures calculated in step (2).
For liquefied materials use the residual
undrained strength as defined and
determined in Section 5.3.5.
Estimates of earthquake induced pore pressi
can be obtained through a program
laboratory testing and/or the empif
procedure proposed by Seed and De
(1986). In practice the procedure propos|
Seed and De Alba (1986) is often preferrej
to its simplicity and ease of use. Cyclic tj
of liquefiable soils is very expensive and h|
specialised. Furthermore,such tests rdj
high quality undisturbed samples,whicj
most cases) would be very difficult to obta
For many practical problems,the proc|
proposed by See4and De Alba (1986)
degree of accuracy comparable with thjj
input data. Seed's procedure involves!
following steps:
(1)Determine stress ratios t^/ct1,,)
the soil layers susceptible!
liquefaction (see Section 5.3.3).
(2)Establish N the equivalent numbi
effective stress cycles in the df
earthquake,(Table 13).
(3)Determine the number of cycles!
required to cause liquefaction fo|
stress ratio calculated in (1).
(4)Using the ratio (N/N,) read!
earthquake induced pore pressurej
from Figure 31 below.
0.40
CjcU *0l,N/N|
Figure 31. Rate of pore water pressure built up in cyclic simple shear tests.
It should be noted that in many cases,it will be
sufficient to assume that all potentially
liquefiable zones (as determined by the methods
given in Section 5.3.3) reach the residual
undrained strength,with clayey zones retaining
full (undrained) strength and rockfill zones
retaining full drained strength. In this caS
not necessary to evaluate the induced
pressures,and it is sufficient to use the
pore pressures at the beginning
earthquake.
64 ANCOLD Guidelines for Design of Dams for Earthquake
6.4.2 Post Liquefaction Deformation
Analysis
Indicative estimates of potential liquefaction
induced deformations can be obtained by
performing a static deformation analysis which
incorporates the earthquake induced pore
pressures and the residual strength of the
liquefied soils (Finn,1993and Khalili,1994).
The analysis is often performed in two stages.
In the first stage,the numerical model is
initialised to the pre-earthquake conditions of
the dam by simulating the current in-situ
stresses. Then,in the second stage,the
earthquake induced pore pressures and residual
strengths of the liquefied soils are incorporated
into the model to simulate post-liquefaction
conditions.
This type of analysis is also referred to as
uncoupled deformation analysis and generally
leads to conservative estimates of post
liquefaction deformations,as it does not allow
for dissipation of earthquake induced pore
pressures with time. More accurate estimates of
post liquefaction deformations can be obtained
using fully and semi-coupled methods of
analysis,as discussed in the following sections.
6.5 Numerical Methods
Numerical modelling techniques such as the
finite element method were first applied to the
dynamic analysis of embankment dams by
Clough and Chopra (1966). This was followed
by major improvements by Gaboussi (1967),
Schnabel et al (1972),Gaboussi and Wilson
(1973),Idriss et al (1974),Martin et al (1975),
Finn et al (1977),Lee and Finn (1978),White
et al (1979),Zienkiewiczand Shiomi (1984),
Finn el al (1986),Medina et al (1990) and more
recently Li et al (1992). Today,numerical
methods are routinely used as both investigative
and design tools in many geotechnical
earthquake engineering problems.
The ^dynamic numerical codes used in practice
may be divided into two main categories: total
stress codes,and effective stress codes
(Zienkiewiczet al,1986 and Finn 1993). A
brief discussion of some of the more frequently
used codes within each category is provided in
the following sub-sections.
6.5.1 Total Stress Codes
The total stress codes,as can be inferred from
the classification,are based on the total stress
concept and do not take account of pore
pressures in the analysis. Therefore,they are
used in situations where the seismically induced
pore pressures are negligible. The total stress
codes may be divided into two main categories:
(1) codes based on the equivalent linear (EQL)
method of analysis,and (2) fully non-linear
codes.
(a)Equivalent Linear Analysis (EQL)
The earlier total stress codes were based on the
EQL method of analysis developed by Seed and
his colleagues in 1972. EQL is essentially an
elastic analysis,and was developed for
approximating non-linear behaviour of soils
under cyclic loading. Typical of the EQL codes
currently used in practice are: SHAKE
(Schnabel et al,1972),QUAD-4(Idriss et al,
1973) and FLUSH (Lysmer et al,1975).
SHAKE is a one dimensional wave propagation
program,and is used primarily for site response
analysis. QUAD-4and FLUSH are two
dimensional versions of SHAKE,and are used
for seismic response analysis of dams and
embankments. Given the elastic nature of the
EQL analysis,however,these codes cannot take
account of material yielding and material
degradation under cyclic loading.
Therefore,they tend to predict a stronger
response than actually occurs. Also,they
cannot predict the permanent deformations
directly. Indirect estimates of permanent
deformations can however be obtained using
the acceleration or stress data obtained from an
EQL analysis and the semi-empirical methods
proposed by Newmark (1965) and/or Seed et al
(1973).
(b)Fully Non Linear Analysis
More accurate and reliable predictions of
permanent deformations can be obtained using
the elasto-plastic nonlinear codes. Typical of
AN COLD Guidelines for Design of Daws for Earthquake
65
the elasto-plastic non-linear codes used in the
analysis of embankments are DIANA (Kawai,
1985),ANSYS (Swanson,1992),FLAG
(Cundull,1993),etc. The constitutive models
used in these codes vary from simple hysteric
non-linear models to more complex
elastic-kinematic hardening plasticity models.
Compared to the EQL codes,the elasto-plastic
nonlinear codes are more complex and put
heavy demand on computing time. However,
they provide more realistic analyses of
embankments under earthquake loading,
especially under strong shakings. Critical
assessments of non-linear elasto-plastic codes
can be found in Marcuson et al (1992) and Finn
(1993).
6.5.2 Effective Stress Codes
A major stimulus for the development of the
effective stress codes has been the need for
modelling pore pressure generation and
dissipation in materials susceptible to
liquefaction and thus to obtain better estimates
of permanent deformations under seismic
loading. The effective stress codes may be
divided into three main categories: fully
coupled,semi-coupled and uncoupled.
(a) Fully Coupled Codes
In the fully coupled codes,the soil is treated as
a two-phase medium,consisting of soil and
water phases. Two types of pore pressures are
considered,transient and residual. The transient
pore pressures are related to recoverable
(elastic) deformations,and the residual pore
pressures are related to non-recoverable
(plastic) deformations. A major challenge in
fully coupled codes is to predict residual pore
pressures. The residual pore pressures,unlike
the transient pore pressures,are persistent and
cumulative,and thus exert a major influence on
the strength and stiffness of the soil skeleton.
The transient pore pressures are cyclic in nature
and their net effect within one loading cycle is
often equal to zero. An accurate prediction of
residual pore pressures requires an accurate
prediction of plastic volumetric deformations.
In the fully coupled codes,this is often achieved
by utilising elastic-plastic models based on
kinematic hardening theory of plasticity
(utilising multi-yield surfaces) or JM
surface theory with a hardening la\3M
models are very complex and put
demand on computing time. JH
Generally speaking,fully coupled prefM
pore pressures under cyclic loadingB
complex and difficult. To date,main
numerical codes developed in this are
fully validated and still are 'M
developmental stages. The validationl
performed on a number of these codes I
that the quality of response predictil
strongly path dependent (Saada and Bil
1987). When the loading paths are sitl
the stress paths used in calibrating the tl
the predictions are good. As the loadil
deviates from the calibration pat!
predictions become less reliable. Apal
the numerical difficulties,part M
unreliability is also due to the poor or la
satisfactory characterisation of thi
properties required in the models. For iti
because of sampling problems,it is ofti
difficult to (accurately) determine 1
change characteristics of loose sands as n
by these models. In general,the accui
pore pressure predictions in fully c
models is highly dependent upon the qui
the input data. Typical of the fully c
codes are DNAFLOW (Prevost,
DYNARD (Moriwaki et al,
SWANDYNE (Zienkiewicz,1991)
SUMDES(Lietal,1992).
(b) Semi-Coupled Codes
Compared to the fully coupled code
semi-coupled codes are more robust art
susceptible to numerical difficulties. Ho
they are theoretically less rigorous. In
codes empirical relationships such as
proposed by Martin et al (1975) and See
(1983) are used to relate cyclic
strains/stressesto pore pressures. The em
nature of the pore pressure generation in
codes generally puts less restrictions on th
of plasticity models used in the codes,
semi-coupled codes are in general less coi
and computationally demanding. Als(
parameters they require are often rou
obtained in the laboratory or in the field. '
66
ANCOLD Guidelines for Design of Dams for Earthquake
is extensive experience in using semi-coupled
codes in practice
Typical of the semi coupled codes are
DESRA-2 (Lee and Finn,1978),DSAGE
(Roth,1985),TARA-3(Finn et al,1986),and
FLAG (Cundall,1993).
(c) Uncoupled Codes
In the uncoupled analysis,the pore pressures
are estimated separately using either a program
of laboratory testing or an empirical
relationship such as the one proposed by Seed et
al (1983). Then,they are incorporated into an3
elasto-plastic nonlinear code to obtain
permanent deformations. The uncoupled
analysis is widely used in practice and is
generally believed to provide indicative
estimates of the post liquefaction behaviour of
earth dams. The permanent deformations
obtained using this approach are often on the
conservative side,as the analysis does not allow
for dissipation with time of the estimated pore
pressures.
6.6 Proposed Guidelines
The guidelines proposed for seismic stability
assessment of embankment dams in Australia
may be best illustrated by the flow chart shown
in Figure 32. The procedure involves the
following main steps:
1. Establish if the embankment or its
foundation is susceptible to liquefaction
or not susceptible to liquefaction using
the methods detailed in Section 5.3.
2. For dams not susceptible to
liquefaction:
-perform an initial screening
assessment detailed in 6.3.1
-if the initial screening criteria are
not satisfied,conduct a
pseudo-static analysis using the US
/Army Corps of Engineers
Recommendations
-if the safety factor is greater than
one,the earthquake performance of
the dam is acceptable and no
deformation analysis will be
required
-if the safety factor is less than one,
perform a Newmark type
deformation analysis using the
Makdisi and Seed (1978) approach
-should the results of the Makdisi
and Seed analysis be unacceptable
and further analysis is considered
warranted,perform a seismic
response analysis using a fully
validated elasto-plastic nonlinear
dynamic code.
For dams susceptible to liquefaction
perform a post liquefaction analysis
consisting of the following main steps:
-evaluate earthquake induced pore
pressures in the materials
susceptible to liquefaction (note ru
=1 implies liquefaction)
-conduct a conventional limiting
equilibrium stability analysis using
the above-estimated pore pressures.
For liquefied materials use the
residual undrained strength. For
clayey materials consider a strength
reduction factor of 15%
ANCOLD Guidelines for Design of Dams for Earthquake 67
Figure 32. Embankment seismic stability assessment chart.
68
ANCOLD Guidelines for Design of Dams for Earthquake
-if the factor of safety using this
analysis is less than 1.0using
reasonable estimates of residual
undrained strength,perform a post
liquefaction deformation analysis as
described in 6.4.2.
4. Should the results of the post
liquefaction analyses be unsatisfactory,
perform a real time coupled effective
stress dynamic analysis. Both
semi-coupled and fully coupled codes
may be used for this purpose.
Preferably,both pore pressure
generation and dissipation should be
considered in the analysis. However,
these analyses are expensive in
computation and data required,and
they will only be needed for large
dams,where the other analyses are
giving marginal answers.
In all cases the extent to which one
analyses stability and deformation
should be consistent with the hazard
rating and size of the dam. In many
cases,it will be better to design
remedial works,rather than continuing
to do more and more sophisticated
analysis.
Some soils (and rocks) in embankments
and foundations,eg. overconsolidated
clays,may be strain weakening. For these
cases,post earthquake analysis should be
carried out to consider the effects of
earthquake induced strains on the
available strength.
There should not be an overeliance on stability
and deformation analysis,at the expense of
good engineering judgement,and the
consideration of the general design principles
given in Section 5.2. At best,the analysis
methods are approximate,and controlled by the
quality of data put into the analyses,and the
limitations of the methods of analysis.
7. ANALYSIS AND DESIGN
OF CONCRETE DAMS
7.1 Past Performance of Concrete
Dams in Earthquakes
To date,concrete dams have performed well
under earthquake conditions. No concrete
dam has failed due to earthquake with loss of
part or whole of the reservoir. However,a
number of dams have suffered serious damage,
including cracking right through the concrete
eg. Koyna and Sefid Rud Dams. A summary
of the earthquake effects on some concrete
dams is presented in Table 16. A number of
these have been discussed in the literature (e.g.
Clough and Ghanaat,1994; Hinks and
Table 16. Earthquake Effects on some Concrete Dams
Dam Height
m
Type Country Date Damage Magnitude
M
Koyna 103Concrete gravity India 1967Major cracking 6.5
Sefid Rud 106 Buttress Iran 1990Major cracking 7.3to 7.7
Pacoima 113Arch California,
USA
1971 Cracking at left
abutment
6.6
Lower Crystal
Springs
47 Curved gravity California,
USA
1906 No damage 8.3
Blackbrook 29 Concrete and
masonry gravity
dam
UK 1957Copings displaced.
Cracking
5.3
Hsingfengkiang 105 Buttress China 1962 Major crack 6.1
Honen-Ike 30 Multiple arch Japan 1946 Crack in arch near
buttress
N/A
Ambiesta 59 Arch Italy 1976 No damage 6.5
Maina di Sauris 136 Arch Italy 1976 No damage 6.5
Shenwao 53Concrete gravity China 1975 Cracking N/A
Redflag 35 Masonry gravity China 1970Cracking,leakage N/A
Rapel 110 Arch Chile 1985 Damage to spillway
and intake tower
7.8
(from Hinks and Gosschalk,1993)
ANCOLD Guidelines for Design of Dams for Earthquake 69
Gosschalk,1993;Serafim,1981,Chopra and
Chakrabarti,1972,USCOLD,1992). As an
indication of the severity of shaking at some of
the above dams,Pacoima Dam was subject to
a peak ground acceleration of 1.2g in the
horizontal direction in the 1971 earthquake
(Serafkn,1981) and sustained minor damage.
It suffered relatively minor damage during the
Northridge Earthquake of 1994despite peak
ground accelerations of 2.0g horizontal and
1.4g vertical on the left abutment and 0.5g
horizontal and vertical accelerations at the
dam base. The water level in the dam was
41m below FSL at the time (Dames and
Moore,1994).
Besides the dams in the above table,there
have been many other dams subjected to
severe earthquake loadings which have
suffered no or very little damage at all
(Serafim,1981; Hinks and Gosschalk,1993,
USCOLD,1992).
Of concrete dams generally,arch dams have
been the best performers during earthquakes
These are followed by gravity dams. Buttress
dams have not performed as well as the former
two types,probably due to the lower stiffness
in the cross-valley direction and the structural
discontinuity of the dam.
While concrete dams have not generally
suffered major damage from earthquakes,
there is no cause for complacency. Many
older dams have had earthquake loads
imposed on them which have been much
greater than their design loadings (Hanson and
Roehm,1979). However,these dams have
generally been designed using a simplistic
approach coupled with a low demand on the
concrete tensile strength. Factors such as
dynamic elastic modulus,interaction between
the foundations,the dam and the reservoir,
material strengths and stiffnesses under
dynamic conditions were not normally
considered. Now however,with dams being
designed using more sophisticated methods
and with some dams relying on post-tensioned
ground anchors for stability,there may not be
the same degree of built-in security. It is
therefore necessary that the earthquake model
adopted and the properties on which it is based
for a particular dam and its foundations be
well considered.
It is unrealistic to carry out a refined analysis
if the structure and properties of the dam and
its foundation is not well known,
particularly important for foundatf
gently dipping or near horizontally !
sedimentary rocks,or any rocks co
continuous gently dipping joints or sea
7.2 Defensive Design Measure
As for embankment dams,the use of d
measures for new dams,or for remedi
on old dams is advised.
The two main criteria for a concr
subjected to earthquake loading are
remains serviceable after high pro4
earthquakes (e.g. the OBE) and t
security of the storage is not prejudice
to cause dangerous flooding after f
probability earthquake (e.g. the MD
ensure that these criteria are both sai
there are a number of defensive
measures which can be incorporated
design of new dams or remedial work
defensive design measures address
dam structure and the interface of
with its foundations,and the foundation
In general terms,a dam should be d
such that:-
sliding is prevented -or at least
resistance should be suffici
prevent failure in the post ea
condition after the MDE
the dam is stable against ove
at all levels even after crac'
increased uplift pressure ind
earthquake
any cracking will not I
uncontrolled leakage
hydraulic outlet structures
damaged to the extent that th
allow uncontrolled loss of wat
might lead to collapse of the
they cannot be used to lo
storage if necessary.
To achieve the above,the list belo
some defensive measures that
implemented:
ensure that the cross section of a
well proportioned with no sudden
in shape or section stiffness.
ensure that any superstructure on t'
crest is minimal.
70 ANCOLD Guidelines for Design of Dams for Earthquake
ensure that there are no sudden changes in
the abutment profile which would give rise
to stress concentrations.
ensure that there are no geological features
in the foundations which would decrease
the dam's sliding stability -if there are,
then suitable means of stabilising these
features or accounting for them in the
design should be employed. This is
especially true of arch dams,since the
interaction of dam thrusts and foundation
characteristics is determinative of the
whole dam-foundation stability.
ensure that there is sufficient internal and
foundation drainage,and that drain holes
are of large enough diameter -the diameter
of drain holes is especially important if
some controlled sliding is to be permitted.
It is vital that the drains keep working or
that the dam is stable post earthquake with
the drains not operative.
if post-tensioned ground anchors are
required for stabilising a dam then ensure
that they are unbonded (except for their
anchorage length) -this will allow strains
due to any overload resulting from
earthquake loading to be taken over the
entire free length of the cable rather than
over a very short length either side of a
crack in the dam or foundations -strains
taken over the entire free length results in
much lesser strains and consequently much
lesser stresses due to earthquake loading.
ensure that internal features such as
galleries do not give rise to stress
concentrations which will lead to excessive
cracking.
provide so far as is practicable,sufficient
outlet works discharge capacity to rapidly
lower the storage if necessary,after an
earthquake.
7.3Analysis Methods
7.3.1 General
Until approximately forty years ago,the only
concession made to earthquake effects in the
design of concrete dams was to consider an
additional inertial load acting in the
downstream direction. This load was
detefmined by multiplying the mass of the
dam by an assumed earthquake coefficient
equivalent to an earthquake acceleration of
0.05g to O.lg. Later on,"added mass" was
applied to the upstream face of the dam to
represent the dynamic effect of the water in
the vicinity of the dam. The distribution of
this "added mass" was commonly based on
Westergaard hydrodynamic pressure
distribution.
At this time,the flexibility of the dam
(especially in the case of gravity dams) and its
dynamic response were not recognised. The
next stage in designing concrete dams for
earthquake was to recognise the flexibility of
the dam and linear elastic methods were
established to determine a dam's response in
the frequency domain (i.e. using an earthquake
response spectrum). It is now becoming
increasingly popular to determine a dam's
response in the time domain (i.e. using
suitable accelerograms) when analysing the
dam with non-linear elastic methods.
The three main types of concrete dams are :
gravity dams
arch dams
buttress dams
Each type of dam puts particular demands on
the numerical methods and the various
parameters required for earthquake design and
analysis. In these guidelines however,
emphasis will be given to concrete gravity
dams because most concrete dams in Australia
are concrete gravity structures,and arch and
buttress dam analysis cannot be simply
generalised. A review paper by Hall (1988)
contains a good summary of the state of
concrete dam seismic studies throughout the
world.
Over recent years,international dam
authorities have presented guidelines for
earthquake analysis of concrete dams e.g.
USER (1977) and ICOLD (1986). There has
been considerable published work by Chopra
(1979,1987,1991). A recent paper by Clough
and Ghanaat (1993) presents an up to date
summary of the development of concrete dam
earthquake analysis and describes recent
advances.
Methods of analysis range from simplified 2D
methods such as presented by Fenves and
Chopra (1987) to more complex,non-linear,
3D finite element,dynamic analysis.
The type and degree of sophistication of the
analysis method used for a particular dam will
depend on a number of issues. These include
the size,type and hazard classification of the
ANCOLD Guidelines for Design of Dams for Earthquake 71
dam,the valley configuration,foundation
characteristics,the presence or otherwise of
storage siltation adjacent to the dam,and the
characteristics of the earthquake.
In addition to the above issues there are a
number of factors which need to be considered
in assessing a concrete dam's ability to
withstand an earthquake. These factors are
concrete crushing strength,dynamic tensile
strength of lift surfaces,foundation strength
and the amount of allowable sliding of the
dam on its foundation for extreme
earthquakes. For all dams,foundation stability
is paramount and in many circumstances it
will be more critical than maximum tensile
stresses induced within the dam body.
Relevant factors include the presence of weak
seams,bedding surfaces and sheared joints and
zones,joint orientations in relation to the
thrust vectors imposed by the dam,continuity,
roughness and shape,and the presence of low
strength and/or highly jointed rock.
7.3.2 General Description of Analysis
Methods
(a) Type of Analysis
The increased power of personal computers
and of finite element analysis programs means
that much more sophisticated numerical
models of concrete dams can be used to
determine the dams' behaviour during
earthquakes. However,not all concrete dams
will need dynamic,numerical analysis having
a high level of complexity. There is still a
need to carry out simple analyses without
having to resort to finite element analysis. For
example,a small (less than 15m high) concrete
gravity dam with a length more than twice the
height can be quite adequately examined using
simple,pseudo-static methods. In this case,
the dam is considered to be a rigid body on
rigid foundations with an added "virtual" mass
of water attached to the upstream face. An
inertial load due to the earthquake is included
in the traditional,cantilever stability analysis
(see the ANCOLD "Guidelines on Design
Criteria for Concrete Gravity Dams",
ANCOLD,1991).
For higher concrete gravity dams which are
still essentially two dimensional structures,the
simplified dynamic analysis methods of
Chopra and others are appropriate,especially
in preliminary analysis and design.
The Fenves and Chopra (1986,1987)1
for gravity dams uses a 2D analysis 1
based on modal superposition. It (ielcij
the fundamental natural frequency and
shape for the dam on a rigid foundalidn
an empty storage. Empirically (U
formulae based on a "standard" gravity
cross section are used for this. 1
fundamental natural frequency is then ad
to account for the interaction of the dan
the storage,the flexibility of the
foundations,the compressibility of the^
in the storage and the effect of sedinic
the bottom of the storage. 1 Kinj
fundamental natural frequency correcte
these factors and an appropriate euuh
response spectrum,the spectral accele
(maximum acceleration of a single dcgi
freedom (SDOF) oscillator) is delerini
This is adjusted to account for the distril
of mass and stiffness within the da
compared to the equivalent SDOF oscil
Knowing the distribution of accele
through the dam,inertia and hydrodf
forces for the fundamental mode a
calculated. Formulae are provided to al
higher modes of vibration. The eart
forces thus calculated can be include"
stability analysis or finite element ana'
determine instantaneous stresses in the
The above methods would normal
appropriate when cracking of the dam
expected from an earthquake.
With the Fenves and Chopra (1986,
method,appropriate parameters (repr
the influences of dam/foundation inte
dam/storage interaction,mitigating effl
silt,higher vibration modes) need/
chosen. If there is sediment at the bo
the storage then there will be some abs
of the hydrodynamic wave. This will t'
a reduction of the lateral forces on the
the dam. The paper gives some typical ,
The Fenves and Chopra method uses
frequencies and mode shapes basedj
triangular cross section dam. For i
accurate estimation of natural frequenci
mode shapes,especially if the dam's?
section is more trapezoidal or has a Vj
slope on the downstream face,a 2
element model can be used. ^
When cracking of the dam due to co
static and dynamic loads is expecte
72 ANCOLD Guidelines for Design of Dams for Earthquake
linear methods can be used. While linear
elastic models can be analysed in either the
frequency (using a response spectrum),or time
domain (using an accelerogram),a non-linear
analysis would normally be analysed in the
time domain. This type of analysis would
normally require a quite sophisticated finite
element program. However,an approximate
analysis can be made using the approach of the
US. Corps of Engineers as described by
Guthrie (1986) which is based on a linear
elastic analysis. Larger damping factors are
used to simulate the energy absorbing effects
of cracking.
For arch dams,buttress dams and gravity dams
where length is less than twice the height,a
three dimensional finite element model can
provide natural frequencies and mode shapes
for determining earthquake loads. As the
mode shapes are more complex than those for
a gravity dam analysed in two dimensions,it is
more appropriate to analyse these dams in the
time domain.
Where cracking is expected in arch dams,
buttress dams and gravity dams with lengths
less than twice their heights,non-linear
analysis is complex. A non-linear analysis
Table 17. Analysis Methods
(c) Foundation Characteristics
Unless the foundation rock is considerably
stiffer than the concrete in the dam which
would oe unusual,it will be necessary to
include the effects of the dam-foundation
interaction. As the foundation rock becomes
less stiff in comparison with the concrete in
the dam,the natural frequency of the dam is
would normally be considered,only for final
design work where the high cost of the
analysis is an economic proposition when
compared to the total project cost.
(b) Dimension of Analysis
Whether to do a 2D or a 3D analysis will
largely depend on the type of dam and the
valley geometry:
Arch dams will certainly require a 3D
analysis.
Buttress dams,especially those with
buttresses having low stiffness in the cross-
valley direction,will also require a 3D
analysis.
Gravity dams in wide valleys may be
examined using a 2D analysis. This is
especially so if the ends of the dam are
relatively unrestrained (e.g. earth
embankments at the ends of the concrete
dam).
Gravity dams in narrow valleys where the
length of the dam is less than twice the
height of the dam should be examined
using a 3D analysis. However,2D analysis
may be used for preliminary design.
Table 17summarises this advice.
reduced and the effective damping is
increased. If the foundations are included in a
finite element model,they are given zero
mass,but the modulus and damping
coefficient of the rockmass is included.
(d) Hydrodynamic Pressure
Method Dimension Applicability
Pseudo-static 2-D Gravity Dams < 15m high -
Length > 2*Height
Simplified Dynamic Analysis or
linear elastic finite element method
2-D Gravity Dams -No cracking
Length > 2*Height
US Corps of Engineers
(as described by Guthrie (1986))
2-D Gravity Dams -with cracking
Length > 2*Height
Linear elastic finite element method 3-D Gravity Dams -Length < 2*Height
Arch Dams
Buttress Dams
Non-linear Dynamic Analysis
3-D Gravity Dams -Length < 2*Height
Arch Dams
Buttress Dams
ANCOLD Guidelines for Design of Dams for Earthquake 73
The water pressure on a dam is increased
during an earthquake by an added
hydrodynamic pressure. The hydrodynamic
pressure results from the dynamic interaction
of the dam and the stored water. The lateral
forces on the dam resulting from
hydrodynamic pressure are mostly determined
by the slope of the dam's upstream face,the
flexibility of the dam and the amount of
sediment on the bottom of the storage. The
storage water affects the dynamic behaviour of
the dam which in turn affects the
hydrodynamic pressure on the dam.
Hydrodynamic pressure is discussed in more
detail in Sub-section 7.4.2.
(e) Non-linear Behaviour and Cracking
When the combined maximum static and
dynamic tensile stress in a concrete dam
exceeds the dynamic tensile strength of the
concrete (especially at lift surfaces) or of the
foundation interface,cracking and non-linear
behaviour can be expected. The cracking and
non-linear behaviour will reduce the stiffness
of the dam and provide a mechanism for
energy absorption. The peak values of tensile
stress and extent of tensile zones will tend to
reduce.
If extreme earthquakes are permitted to cause
significant cracking of the dam then
consideration has to be given to additional
factors:
(i)what uplift pressures are induced in
the cracked zone?
(ii)how is crack propagation modelled?
(iii)what are the limit states for the dam?
Point (i) is discussed in more detail in Sub
section 7.4.3.
On point (ii),crack propagation may be
modelled using zero tensile strength finite
elements (i.e. cracking is distributed
throughout elements,modelling discrete
cracks where there are tensile stresses,or
using a fracture mechanics approach).
Point (iii) is discussed in Sub-section 7.3.4.
(1) Dynamic Strengths
Dynamic strengths and modulus of concrete
and the rock foundation in tension.
compression are discussed in S
7.6.1.
(g) Analysis Program
A number of the papers referred to i
may refer specifically to a progra
EAGD-84for 2-D analyses and BAG
3-D analyses.
These programs were written by C
association with others,and are availa
the Earthquake Engineering Research
University of California,Berkeley.
Program EAGD-84is user frien
includes the effects of: 1
the reservoir;
sedimentation at the reservoir botto
water compressibility; and
dam-foundation rock interaction.
Input into the program include
horizontal and vertical accelerati
histories. The difficulty here is in
locally recorded accelerograms upw|
represent an accelerogram for say a
Design Earthquake. This is discus
Section 7.4. A suitable plot of resul
different accelerograms will generally it
whether the sensitivity range is too grea
7.3.3Details of Analysis Methods
(a) Pseudo-static Method
This method assumes that the dam
foundations are rigid and that the wate
storage is incompressible. It assumes f,
entire dam has an acceleration the sam
peak ground acceleration.
The hydrodynamic pressure distribute
calculated according to Westergaard (1
Zangar (1952) e.g.
Ph =Cwh
g
74 ANCOLD Guidelines for Design ofDams for Earthquake
Ph =Hydrodynamic pressure
ag=Peak groundacceleration
g=AccelerationduetoGravity
w =Unit weight of water
C =Dimensionless coefficient:
C =^(2-
2 h
y =Depth fromstoragesurfacetoparticular level
h =Total depth of reservoir at Section
Cm=Maximumvalueof C (073for Vertical Face)
The force derived from this pressure
distribution,and the inertia force equal to the
mass of the dam multiplied by the peak ground
acceleration,are then included in the
traditional cantilever stability analysis with the
appropriate static forces.
The stability analysis gives peak stresses and
sliding stability for the dam. However,it
should be remembered that the resulting peak
stresses only act for a very short time during a
cycle of the earthquake. Consequently,peak
stresses greater than dynamic strength may be
indicative of zones of potential cracking.
Analysis of the dam in a post earthquake
cracked condition will be required to assess
the overall stability.
Generally,this method would only be used as
a screening method to determine if a gravity
dam has a potential problem with earthquakes
or for small dams less than 15m high,in a
wide vallty.
(b) Linear Elastic Dynamic Methods
(i) Simplified Dynamic Analysis Method
earthquake
-peak ground acceleration
-response spectrum for the peak ground
acceleration and a range of damping
ratios
dam/foundation/storage interface
-period lengthening ratio and added
damping ratio due to hydrodynamic
effects
-period lengthening ratio and added
damping ratio due to dam -foundation
rock interaction
-hydrodynamic pressure distribution
(depending on the degree of
hydrodynamic wave absorption in the
alluvium and sediments at the bottom of
the storage).
The computational steps for the method are:
1. Compute the fundamental period of
the dam for an empty reservoir and
rigid foundations.
For dams of triangular cross section,
use the formula:
Ti =0.38
Hs
VEs"
where:
Ti =
Hs =
Eo =
Fundamental period for dam
of essentially triangular cross
section with an empty
reservoir and a rigid
foundation
Dam Height in metres
Dynamic Elastic Modulus of
Concrete in MPa.
This method is presented by Fenves and
Chopra (1986,1987). This sub-section will
describe the method in general terms but
Fenves and Chopra (1986) should be read for
more details.
The input parameters required fall into three
groups -
dam geometry and properties
-height of dam
-modulus of elasticity of concrete
-the generalised mass of the dam
-the generalised earthquake force
coefficient for the dam
2. Adjust the fundamental period using
the period lengthening ratios to
account for the influence of the
impounded water ,the flexibility of
the foundations and for the presence of
sediments at the bottom of the
reservoir.
3. Compute the ratio of the fundamental
vibration period of the impounded
water to the fundamental period of the
dam.
4. Compute the damping ratio for the
dam using the basic viscous damping
ratio for the dam on rigid foundations
ANCOLD Guidelines for Design of Dams for Earthquake 75
and empty reservoir (say 5%) and the
added damping ratios accounting for
the extra damping due to the dam
foundation rock interaction and
hydrodynamic effects.
5. Determine the hydrodynamic pressure
distribution using the period ratio
calculated in step 3.
6. Determine the fundamental mode
shape for the dam and compute the
generalised mass (Mj) and
earthquake coefficient (Lj).
Conservatively,Lj/Mj =4for dams
with full storages and Lj/Mj =3for
dams with empty storages.
7. Compute the equivalent lateral
earthquake forces distributed through
the dam for the fundamental mode,
using:
^ (y) =Sa^"^ [(ds (y)Ky) + gPi (y.T r)]
Mi g
where :
f,(jy) =Lateral earthquake force at
elevation y
Li =Generalised earthquake
coefficient
Mi ::=Generalised Mass
SaCf,,^,) " Spectral acceleration for period
7,and damping^
g =9.806 m/s^
fi)s(y) =Weight of dam per unit height
<Ky) =Normalised horizontal
displacement for the
fundamental mode
Pi(y>Tr) ~ Hydrodynamic pressure for the
adjusted fundamental period of
the dam
8 Compute earthquake forces due to
higher vibration modes.
9. Compute the square root of the sum of
the squares of the forces calculated in
steps 7and 8. Add the resulting forces
to the static forces present pri(
earthquake in a stability analys
The above calculations can convene
done on a one page spreadsheet
Figures in Fenves and Chopr;
1987) show the hydrot
pressure profiles required for ai
A parametric study of a large concrete
dam was carried out by Chavezand
(1993) using the above method. The r<
the study indicates the sensitivity of t
sliding to the various parameters.
(ii) US Corps of Engineers Method
This method is as presented by Guthrie
It can be used for both 2D and 3D prob
For 2D problems,the Simplified E
Analysis above can be used to de
earthquake forces using a viscous d
ratio for the dam of 5%. The forces i
applied as equivalent static loads togetl
other static loads to a 2D finite elemen
of the dam and foundations. Peak stres
computed using the finite element modi
If the dam has to be considered a?
problem then a full dynamic analysis i
3D finite element model has to be cari
to determine peak stresses for the 5% 1
damping. For a more detailed descrip
the full dynamic analysis see below.
The analysis is carried out for be
operating basis earthquake and the ma
design or credible earthquake. It cQ
that the dam behaves elastically for
stresses less than 10% of the c
compressive strength. For tensile i
beyond that value the concrete will craq
method assumes that even though the d?
now behave non-linearly,it can s
modelled elastically provided the di
ratio is increased. The increase in di
ratio is assumed to represent the in<
energy dissipation caused by tensile en
The steps in the method are best repn
by flow charts which are taken from <
(1986). However in general terms,the I
follows two streams -one for the ma
credible earthquake (or maximum
earthquake) and one for the operatinj
earthquake:
76 ANCOLD Guidelines for Design ofDams for Earthquake
-4
Figure 33Sequence of Analysis for Maximum Design Earthquake
(Guthrie 1986)
ANCOLD Guidelines for Design of Dams for Earthquake
CONTINUED
Figure 33(continued) Sequence of Analysis for Maximum Design Earthquake
(Guthrie 1986)
ANCOLD Guidelines for Design of Dams for Earthquake
Maximum Design Earthquake (refer Figure
33)-
(i)Initial 2D analysis,e.g. as
described by Fenves and
Chopra (1986,1987) with 5%
viscous damping.
(ii)If peak tensile stress (static +
dynamic) >15% UCS (fc) of
concrete then re-model using
FEM (good quality lift
surfaces assumed). If peak
tensile stress (static +
dynamic) <15% UCS of
concrete then the dam is
adequate for MDE.
(iii)If peak tensile stress (static +
dynamic) is >15% but <20%
UCS of concrete then repeat
FEA with 7% Damping Ratio.
Assume cracking wherever
combined tensile stress (static
+ dynamic) is >10% UCS of
concrete.
(iv)If peak tensile stress is >20%
UCS of concrete carry out (iii)
but use a 10% Damping Ratio.
(v)On horizontal planes with
cracking,carry out sliding
analysis with cohesion only
applied to the uncracked
portion.
(vi)If sliding safety factor
(factored sliding strength
divided by horizontal forces
causing sliding) >1.0then
check for post earthquake
stability. If sliding safety
factor >1.3and maximum
compressive stress <0.5 UCS,
concrete dam is satisfactory.
If either these two criteria are
not satisfied then the dam is
unsatisfactory and
strengthening (or some other
/remedial option) for
earthquake loading is
required.
(vii)If sliding safety factor <1.0
then compute the
displacement along a
horizontal plane. Refer to the
sub-section below on
Permanent Deformations for a
suitable method. If the
permanent displacement is
considered acceptable then
check post earthquake stability
for maximum static loading,
considering the cracking
caused by the earthquake and
the extra uplift force that
might be caused by that
cracking. If the permanent
displacement is considered
unacceptable then the dam is
unsatisfactory and
strengthening (or some other
remedial option) for
earthquake loading is
required.
Operating Basis Earthquake (refer Figure 34) -
(i)If the above analysis indicates
no cracking then further
analysis for the operating
basis earthquake is not
required.
(ii)If analysis for the operating
basis earthquake is required
then carry out an analysis
similar to (i) for the maximum
credible earthquake using 5%
damping ratio. The peak
tensile stress (static +
dynamic) is not to be greater
than 10% UCS concrete
provided the lift joints are
sound.
Fenves (1989) compared the analysis of a
gravity dam using the above method with
more rigorous non-linear methods. His
conclusions were that assessing the extent of
cracking from the linear analysis appeared to
give reasonable accuracy for the five cases of
the one dam analysed. He warned however,of
extrapolating the generality of the method.
This is due in part to the arbitrary levels of
maximum tensile stress and viscous damping
ratio. He finishes by saying that "there is no
theoretical basis for assuming that tensile
cracking results in the increased energy
dissipation implied by greater damping ratios."
ANCOLD Guidelines for Design of Dams for Earthquake 79
Figure 34Sequence of Analysis for Operating Basis Earthquake
(Guthrie 1986)
The preceding method could probably be used
for arch and buttress dams as well. However,
consideration would have to be given to the
appropriateness of the damping ratios and
corresponding limiting tensile strengths used.
(c) Non-linear Dynamic Analysis
A non-linear dynamic analysis of a concrete
dam is a complex analysis and would normally
be undertaken by specialist numerical analysts
experienced in such work. It would normally
be done only for major dams where the cost of
the new dam or the cost of remedial works for
80
ANCOLD Guidelines for Design of Dams for Earthquake
an existing dam is sufficient to justify the
greater expense of this type of analysis.
Provided peak tensile stresses (static +
dynamic) do not cause cracking in the dam,
then a linear analysis will suffice. When
cracking occurs,stresses will re-distribute.
Therefore,when cracking is expected or there
are pre-existing cracks (e.g. open vertical
contraction joints) and the dam is essentially
3D m nature then a non-linear analysis should
be done.
There are three main ways in which cracks can
be modelled:
as distributed (smeared) cracks (zero elastic
modulus in direction perpendicular to
cracks)
as discrete cracks at finite element
interfaces
using fracture mechanics.
The first of the alternatives is computationally
the simplest. Further details on cracking are
given by Zienkiewicz,Valliappan and King
(1968),Rashid (1968),Mohraz,Schnobrich
and Gomez(1970),Darwin and Pecknold
(1978),Phillips and Zienkiewicz(1976),
Bazant and Cedolin (1979),Bazant and Ob
(1979),Argyris,Krempl and William (1977),
Gerstle (1981),Kotsovos and Newman (1978),
William and Warnke (1975),Cedolin,Crutzon
and Dei Poli (1977),Bicanic and Zienkiewicz
(1983),Zienkiewicz,Fejzo and Bicanic
(1983),Zienkiewicz,Hinton,Bicanic and
Fejze (1980),Pande and Shen (1982),Pal
(1974) and Chapuis,Rebora and Zimmermann
(1985).
A non-linear analysis will by its nature,
require a time-history analysis. Consideration
will have to be given to:
the way the compressibility of the storage
water is modelled
the way seepage pressures particularly
those due to water penetrating cracked
zones,are modelled.
The non-linear analysis of concrete dams is
still a developing and specialised field. Other
relevant papers include Waggoner,Plizzari
and Saouma (1993),Gao Lin,Jing Zhou and
Chuiyi Fan (1993),Greeves and Taylor
(1992),Cervera,Oliver and Galindo (1992),
Jing Zhou and Gao Lin (1992),Clough and
Ghanaat (1993),Fenves and Mojtahedi (1993).
7.3.4Analysis of Permanent Deformations
In some concrete gravity dams subjected to
say the maximum credible earthquake,it may
be permissible for the dam to slide on its base
or within the foundations and be permanently
deformed after the earthquake. This of course
assumes that during or after the deforming
process,the security of the storage is not
prejudiced allowing for the potentially
increased uplift pressures and lower strength
which may apply. Researchers such as Chopra
and Zhang (1991),Leger and Katsouli (1989)
and Danay and Adeghe (1993) discuss
calculations which indicate that typical
permanent displacements for large concrete
gravity dams subjected to earthquakes having
peak ground accelerations the order of 0.5g
can range from tens of centimetres to more
than half a metre. However,some dams with
suitable foundations would be able to
withstand small displacements. Dams relying
on post-tensioned ground anchors or drain
holes for normal load static stability,might not
be able to withstand these sort of movements
if the displacement was sufficient to shear the
anchors. However it may be acceptable to
have the anchors sheared for a low probability
earthquake provided the dam was stable under
the post earthquake load case.
The type of analysis required to compute
permanent deformations is similar to that
carried out for embankment dams i.e. Makdisi
and Seed (1978). The dam is considered as a
block with a limiting sliding strength along its
foundations. The dam is subjected to a time
varying input of acceleration. When the
acceleration is greater than the limiting
acceleration (the acceleration causing inertia
forces which are greater than the sliding
strength of the foundations) the dam will move
on its foundations. The parts of the
accelerogram greater than the limiting
acceleration are double integrated to obtain
cumulative displacements.
The type of analysis just described is a
requirement of the US Corps of Engineers
method for dynamic analysis of concrete
gravity dams when the sliding safety factor is
less than one. The sliding analysis is carried
out for horizontal planes through the dam
where there is cracking. A crack is assumed
through the dam with suitable slip elements
along the crack interface.
ANCOLD Guidelines for Design of Dams for Earthquake 81
T!
In their study of base sliding,Chavezand
Fenves (1993) found that sliding was not
likely to occur if the storage is less than half
full. Other conclusions were :
vertical ground motion has almost no effect
on the sliding displacement (it slightly
increased the maximum stresses). This is
partly because the vertical and horizontal
ground motion do not necessarily coincide.
the assumption of rigid foundation rock can
significantly overestimate the amount of
base sliding
At the present stage of development of the
computation of permanent deformations and
the determination of suitable maximum
displacements there is still much work to be
done especially regarding local conditions.
These guidelines therefore counsel that
considerable care be used if a dam is to be
allowed a permanent deformation following an
earthquake. Adequate sliding and overturning
stability must exist after the earthquake using
foundation strengths appropriate to the
displacement (usually the residual strength)
and uplift appropriate to the displaced
condition,allowing for opening of joints and
bedding,and for reduced (or no) drainage
capacity.
7.4Design Earthquake and
Hydrodynamic Loads
7.4.1 Earthquake Parameters
As discussed elsewhere in these guidelines,
dynamic analysis can be carried out in the
frequency domain or the time domain. For the
former,response spectra are required while for
the latter,accelerograms are required.
(a) Response Spectra
A response spectrum shows the extent to
which any single degree of freedom structure
with an assumed level of damping would
respond to particular earthquakes.
Knowing the natural frequencies of vibration
and the corresponding mode shape for a
structure,the spectral accelerations
corresponding to particular natural frequencies
and damping ratios can be converted to inertial
loads.
The response spectrum used in the analysis of
a dam should be site specific and relate to the
peak ground accelerations examined. The
response spectrum should also reflect the
frequency mix and duration of the design
earthquakes. It will therefore be derived from
a number of earthquakes having various
epicentral distances from the site and
consequently,different acceleration
attenuation functions. The response spectra
should be obtained from a seismologist as part
of the assessment of seismicity of the dam site.
(b) Accelerograms
Where a time-history analysis is to be done,at
least three different accelerograms appropriate
to the dam site and for a particular peak
ground acceleration,should be used. These
accelerograms may be recorded accelerograms
which are suitably scaled (accounting for
change in frequency mix and phase with
change in peak ground acceleration) or
synthetic accelerograms which fit the response
spectra for the site. Care should be taken in
selecting accelerograms which are similar to
Australian earthquake conditions,and advice
should be obtained from a seismologist.
7.4.2 Hydrodynamic Pressures
Any movement of the dam and foundation will
cause movement in the water of the storage
and in turn,the pressures generated by the
water will impose forces on the dam.
Engineers have traditionally used
hydrodynamic pressures derived by
Westergaard (1933). These pressures are
commonly converted into equivalent lumped
'virtual' masses which are attached to the dam.
Westergaard's pressure distribution assumes
that the water in the storage is incompressible
and that the dam and its foundations are rigid.
However,this is not always so. In high
gravity dams and slender arch dams especially,
where the dam is flexible,there can be
considerable interaction or coupling between
the dam and the storage.
Considerable work on the interaction of
gravity dams and their storages has been done
by Chopra and his fellow researchers. Details
of the work are given in Chopra (1967),
Chakrabarti and Chopra (1974),Chopra,
Chakrabarti and Gupta (1980),Chopra and
Gupta (1981). Other relevant papers include
82
ANCOLD Guidelines for Design ofDams for Earthquake
Clough and Chang (1980),Dungar (1978),
Hall (1988),Zienkiewicz,Paul and Hinton
(1983) and Tsai and Lee (1989).
Besides the lumped,'virtual' mass approach,
the interaction of a dam with the water in its
storage can be determined by treating the
water as a 'solid' having zero shear modulus
but retaining compressibility. This approach,
although simple in principle,has a number of
numerical problems. An alternative approach
is therefore preferable where from the
beginning,the shear components of stress in
the fluid are neglected. This latter approach
includes the effect of water compressibility. It
also will allow the limiting case of
incompressible water to be considered.
Accounting for water compressibility and
coupling of a dam with the water in its storage
adds considerable computational effort to
determining the effects of earthquake loading
on a dam. Neglecting coupling and water
compressibility in the simpler "added mass"
approach is not considered significant for
excitations at frequencies below the natural
frequency of the reservoir.
An estimate for the fundamental frequency of
a reservoir can be obtained from:
f =__
w 4H
eff
where: fw =the fundamental frequency
C =the compression wave speed
in water (l,439m/s)
Heff =an effective depth of the
reservoir.
The above relationship was obtained from
Duron,Ostrom and Aagaard (1994) and
applies strictly to an infinite reservoir of
constant cross section. Duron and Hall (1988)
indicate that if the ratio of to the
fundamental frequency of the dam-foundation
alone (fj) is near unity,water compressibility
will have a significant effect. For ratios of fw
to fj much greater than one (e.g. >1.5)
incompressible fluid behaviour can be
assumed.
7.4. i Uplift/Seepage Pressures
The USBR (1977) considers that the uplift
pressure within the crack is zero while ICOLD
(1986) assumes full headwater pressure but
recognises the need for further research.
Guthrie (1986) uses the pre-crack uplift
pressure diagram.
These guidelines recommend that for the
duration of the earthquake,the pre-earthquake
uplift pressure distribution is used for the
stability analysis. However for the post
earthquake analysis,consideration should be
given to the amount of cracking and the post-
earthquake efficiency of the dam's drainage
system. In the post-earthquake situation,full
headwater pressure is assumed to exist in a
crack at least as far as the line of drains. If the
drains have sufficient capacity and they have
not been disrupted by sliding of the dam,then,
if the crack extends,past the drains,a
significant reduction in uplift pressure should
be considered. Typically,the pressure at the
line of drains in this case might be the
tailwater pressure plus 50% of the difference
between headwater and tailwater pressures.
The pre-earthquake uplift pressure distribution
might have had a 67% reduction. The lesser
reduction for the post-earthquake case allows
for the greater amount of drainage with which
the drains would have to cope.
7.5 Design Criteria
7.5.1 General Approach
The working group has had two major
difficulties in preparing suitable design criteria
for concrete gravity dams.
(1)The current ANCOLD (1991)
guidelines for design criteria for concrete
gravity dams are based on a limit state
approach with partial factors of safety. This
method has proven to be difficult to use on
dams subject to significant earthquake loads.
The ANCOLD (1991) guideline is under
review to address these problems. In the
interim,it is recommended that the design
loadings and acceptance criteria described
herein are used. These are based largely on
the BC Hydro (1995) guidelines.
(2)Existing guidelines for design of
concrete gravity dams are not simply applied
to a risk based approach. As a result it has not
been practicable to develop these guidelines to
directly apply to a risk based approach,and
they are given in terms of a deterministic
method using OBE and MDE. For
completeness,flood and static load case are
also listed. Those wishing to use a risk based
ANCOLD Guidelines for Design of Dams for Earthquake 83
approach may do so,taking account of the
general intention of the load cases detailed
herein.
7.5.2 Loads and Load Cases
The loads considered in the assessment of
concrete gravity dams and their foundations
should include the following:
Dead loads of permanent structures,
equipment and foundation rock (D).
Water load due to maximum normal
headwater level combined with the most
critical concurrent tailwater level (H).
Water load due to maximum flood
headwater level based on the Inflow Design
Flood (IDF) with corresponding tailwater
levels (H).
Foundation uplift (U),both at the concrete-
rock contact and at critical discontinuities
within the foundation.
Static and dynamic thrust created by an ice
sheet,for reservoirs subject to freezing (I).
Vertical and horizontal loading due to rock
or soil backfill (both natural or engineered),
including potential effects of liquefaction
and loads from silt deposited against the
dam (S).
Load due to Operating Basis Earthquake
(OBE) Q'
Load due to Maximum Design Earthquake
(MDE) (Q).
Determination of the loads should take into
account the actual field conditions and
instrumentation records.
Foundation uplift assumptions should reflect
the stress state and condition of the dam and
foundation. Disruption of the dam and/or
foundation condition due to an earthquake
should be recognised in assessing the uplift
assumptions for the post-earthquake case.
The dam and foundation should be assessed
for the following load cases:
(a) Usual Load Case
permanent and operating loads should be
considered for both summer and winter
conditions including self-weight,ice (where
applicable),silt,earth pressure,and the
maximum normal reservoir level with
appropriate uplift pressures and tailwater level.
(D + H +1 + S + U)
(b)Unusual (Flood) Load Case
Permanent and operating loads of the Usual
Load Case,except for ice loading,should be
considered in conjunction with reservoir and
tailwater levels and uplift resulting from the
passage of the IDF.
(D + H' + S + UF)
where subscript "F" refers to the flood case.
The potential should also be assessed for
reservoir levels higher than would result from
passage of the IDF,such as those due to
operating failures or other unusual conditions.
The effects of ice loads should not be
considered simultaneously with flood
conditions.
(c)Unusual (Earthquake) Load Case
Permanent and operating loads of the usual
load case except for ice loading,should be
considered in conjunction with earthquake
loading associated with the Operating Basis
Earthquake (OBE). The effect of ice loads
should not be considered simultaneously with
OBE earthquake conditions.
The analysis should be carried out for the dam
empty case.
(d)Extreme Load Case
Permanent and operating loads of the Usual
Load Case should be considered in
conjunction with seismic loads of the
Maximum Design Earthquake (MDE).
(D + H + S + Q + U)
The effects of ice should be given special
consideration,recognising the high uncertainty
associated with ice loading on earthquake
loading,and its effects on the dam.
(e)Other Load Cases
Where earthquake-induced cracking at the
concrete-rock interface or any weak section is
identified,a stability analysis should be
carried out to assess whether the dam in its
post-earthquake condition is capable of
resisting loads of the Usual Load Case.
84 ANCOLD Guidelines for Design ofDams for Earthquake
(D + H + S + UpQ)
where subscript "pg" refers to the post-
earthquake case.
Concurrent ice loading (with the post-
earthquake condition) may be considered in
areas where appropriate.
A landslide generated wave case should be
considered where existing or potential
landslides,which may affect the reservoir,
have been identified. Combinations with other
loads would be site specific.
An inoperative drain case assuming plugged
drains may be assessed and could be
considered as an Unusual Load Case.
7.5.3Acceptance Criteria
All kinematically feasible failure modes,
analysed by the single-slice,rigid-body,force
equilibrium method,should satisfy the
acceptance criteria shown in Table 18.
Table 18
Stability Index Acceptance Criteria
Load Sliding Factor
(Note 1)
Position of Resultant Force
(Note 2)
Minimum Compressive
Stress Factor
(Note 3)
Usual 1.5-2.0 Mid-third of surface
No tension
4.0
Unusual (Flood)
and
Unusual (OBE)
1.3-1.5 Mid-half of surface
One-quarter tension
2.7
Extreme (MDE) 1.1 -1.3 Within surface 1.3
Post-earthquake 1.2-1.4 Mid-half of surface
One-quarter tension
2.7
Notes: 1. Sliding Factor (Frictional Analysis) =Resisting Forces
Applied Force
Lower values of the range apply where the geology and the strength parameters are
reasonably well known.
2. Vector summation of all forces,including uplift,acting on the analysis surface
3. Compressive Stress Factor =Unconfined Compressive Strength
Compressive Stress Normal to Surface
To be considered primarily for massive but low strength rock and weak deteriorated
concrete.
7.5.4Post Earthquake Stability
If a dam is likely to be severely damaged after
being subjected to the MDE,considerable time
may elapse before the dam can be repaired or
the storage lowered. Consequently,all parts of
the dam will need to remain stable after an
extreme earthquake event. The stability of the
dam should therefore be checked for static
loading conditions. The assumed uplift
pressure distribution should be as discussed in
sub-section 7.4.3,i.e. full headwater pressure
within cracks emanating from the upstream
face.
7.5.5 Foundation Stability
Where there is the possibility of a sliding
failure along faults,shears and/or joints,the
stability of the foundations should be
ANCOLD Guidelines for Design of Dams for Earthquake 85
examined. This examination should be made
for both the earthquake and post-earthquake
cases. The dam itself should be examined for
local overstressing due to foundation
deficiencies.
Sliding failure is especially likely when
discontinuities and/or horizontal or sub-
horizontal seams close to the foundation
surface contain clay,or have been previously
sheared eg. bedding surface shears due to
stress relief,folding,or associated with faults.
7.6 Dynamic Material Properties
7.6.1 Concrete
The compressive and tensile strengths of
concrete increase with increased rate of
loading. The dynamic compressive and tensile
strengths of concrete can therefore be expected
to be greater than the static strengths. As
dynamic compressive stresses are rarely of
concern,the allowable compressive stress for
static loading can be used also for dynamic
loading.
Raphael (1984) states that the apparent tensile
strength of concrete under seismic loading
which should be used with linear finite
element analyses is given by:
fr =0.65 fc 2P
where fc is the concrete compressive strength
in MPa and fr is the apparent seismic tensile
strength in MPa. Values given by this formula
are some 50% greater than the apparent tensile
strength for static loading. Raphael suggests
that fr is twice the splitting strength of the
concrete under static loading.
Clough and Ghanaat (1993) suggest that the
apparent dynamic tensile strength is about
25% greater than the measured static value
which gives apparent tensile strength about
20% of the standard compressive strength.
They further suggest that there may be a 15 to
20% loss of strength across lift joints. These
figures may be even lower for poorly
constructed or defective lift surfaces,
fiowever,the peak dynamic tensile stresses
only exist during a fraction of a response
cycle. Even though these peak stresses may
greatly exceed the tensile strength of the
concrete,any cracking that might be initiated
will not have time to fully develop. It is well
recognised that a single spike of excessive
localised tension should not be taken to
represent dam failure.
In consideration of the above however,these
guidelines recommend that for sound lift
surfaces,the apparent tensile strength to be
used is 16% of the standard compressive
strength.
For dynamic modulus of elasticity,Clough and
Ghanaat (1993) suggest a value 25% greater
than the static value and these guidelines
recommend this be adopted. For existing
dams,the elastic modulus of the concrete mass
may be determined using geophysical means
(e.g. derived from measured shear wave
velocity). Values obtained should be
compared with static and dynamic small
sample laboratory test values for credibility.
7.6.2 Rock
In most cases the stability of the dam will be
controlled by sliding in or on the dam rock
foundation. To carry out static and dynamic
analyses,it will be necessary to:
assess and map surface exposure,
excavations and drill core to define the
geology of the site. In particular the 3-
dimensional orientation,continuity and
detailed nature of bedding,joints,and other
features such as shears are required.
the shear strength of the foundation rock
should be determined using appropriate
rock mechanics techniques,such as those
described in Hoek (1983,1990,1994),
Hoek and Brown (1980),Patton (1966),
Barton and Choubey (1977),Barton and
Bandis (1991). These require an
assessment of the orientation,spacing,
continuity,shape and roughness of
discontinuities in the rock (e.g. joints),and
the strength of the rock substance as this
varies with confining stress. Care should
be taken in applying these techniques,to
account for the presence of continuous,
adversely oriented low strength surfaces
such as bedding surface shears,faults or
shears,and to take account of the
mechanisms of failure.
The Hoek,and Hoek and Brown methods give
strengths for "undisturbed rock" and
"disturbed rock". The latter should generally
be adopted unless advice from an experienced
86 ANCOLD Guidelines for Design of Dams for Earthquake
rock mechanics specialist indicates otherwise,
because of the uncertainty as to the accuracy
of the Hoek methods,and since the
undisturbed rock strengths apply to confined
conditions such as underground openings,
where dilation or shearing causes increases in
normal stress.
It will be noted these stress methods give non
linear failure envelopes,with high friction
angle and low cohesion at the normal stresses
usually applicable to dams. They also allow
estimation of the modulus of the rock mass
(Hoek,1994). The dynamic modulus may be
higher than the static modulus as discussed in
Clough and Ghanaat (1993) and Scott and Von
Thun (1993) and may,as for concrete,be
obtained by geophysical means,or by relation
to the static modulus.
A dam's foundations will normally contain
joints,shears,and bedding. Consequently,it
will not be possible to transmit tensile stress
within the foundations and the allowable
tensile strength for the foundations will
therefore normally be assumed to be zero.
However,if extensive site investigation and
strength testing is able to prove that the
foundations for a particular dam site are
capable of transferring tensile stresses,then
the tensile strength of the rock may be
included.
Where foundation rock strength becomes
critical,as they often will,advice should be
obtained from a person expert in rock
mechanics.
8. APPURTENANT
STRUCTURES
8.1 Introduction
A number of subsidiary structures associated
with a dam are essential for the dam's
operation. Consequently damage to or
destruction of these appurtenant structures
would be prejudicial to the dam's safety. An
important facility at a dam is one that allows
water to be released in a controlled manner. If
therie has been an earthquake and the dam is
damaged to the extent that the dam is not
serviceable then it may be necessary to lower
the storage so that remedial works can be
undertaken. It will therefore be necessary that
not only the outlet structures and their gates
and valves remain serviceable but also that
there is proper access to these structures.
Bridges and roads may need to remain in a
sound state after an earthquake depending on
their importance.
Generally,appurtenant structures should be
such that:
they maintain their normal operating
condition after an operating basis
earthquake
they are not damaged to an extent where
they could allow sudden or uncontrolled
loss of water from the storage for a more
extreme earthquake up to the maximum
design earthquake.
8.2 Performance Requirements
This sub-section gives the performance criteria
for the operating basis earthquake (OBE) and
the maximum design earthquake (MDE) which
could be the maximum credible earthquake
(MCE). Performance requirements are given
for a number of appurtenant structures
including intake towers,outlet conduits,outlet
works,spillway gates,spillway piers,spillway
gate hoist piers,access bridges and piers,
access roads.
Intake Tower
OBE: Static and dynamic loads to
induce maximum concrete and
steel reinforcement stresses
which satisfy AS3600
(Concrete Structures Code)
(i.e. limited amount of
reinforcement yielding) and
the tower and its base remain
stable.
MDE: Significant amount of
reinforcement can yield
horizontal reinforcement
designed to prevent vertical
reinforcement from buckling
and to contain concrete when
it is in compression (i.e.
concrete contained between
inner and outer layers of
vertical reinforcement will not
spall away).
Outlet Conduit
ANCOLD Guidelines for Design ofDams for Earthquake 87
OBE:
MDE:
Outlet Works
OBE:
Static and dynamic loads to
induce maximum concrete and
steel reinforcement stresses
which satisfy AS3600
(Concrete Structures Code).
Dynamic loads to include
those induced from the
earthquake effect on the
overlying dam.
Conduit not to collapse or
rupture -Collapse could lead
to undermining and
subsequent failure of an
overlying embankment.
Rupture could cause piping or
destabilise an embankment by
a marked increase in pore
pressure.
All valves to maintain their
normal operating capabilities.
MDE: Emergency closure and
regulating valves (especially
low level release valves) to
maintain operating capability -
storage may have to be
quickly lowered if parts of the
dam are damaged and need
remedial works or relief of
hydrostatic loads.
Spillway Gates
OBE: Gates to maintain normal
operating capability.
MDE: Gates retaining permanent
storage at the time of an MDE
should not fail to the extent
where water from the storage
is released in an uncontrolled
manner. MDE should not
cause the gates to distort to an
extent that they cannot be
opened or closed.
Spillway Piers
OBE: Carry out appropriate dynamic
analysis for the spillway piers
for earthquake loading in the
upstream/downstream and
transverse directions.
Combined static and dynamic
loads should satisfy ASS 600
(Concrete Structures Code).
Earthquake loads from the
orthogonal directions may be
combined on a square root of
the sum of the squares basis.
MDE: Carry out dynamic analysis as
for the OBE. Combined static
and dynamic loads may cause
cracking but piers must
remain stable for overturning
and sliding. Piers should not
be permanently displaced to
the extent where the spillway
gates become jammed.
Spillway Gate Hoist Piers
OBE: Carry out appropriate dynamic
analysis similar to that for
spillway piers. Combined
static and dynamic loads
should satisfy AS3600
(Concrete Structures Code).
MDE: If the gates can be operated
from an alternative position
(possibly,in a less efficient
manner),then the spillway
gate hoist piers (and hoist
bridge) can be allowed to fail.
If the continuing operation of
the gates depends on the
continued viability of the hoist
bridge piers (and hoist bridge)
then carry out an appropriate
dynamic analysis similar to
that for the spillway piers.
Combined static and dynamic
loads may cause cracking but
the piers must remain stable
for overturning and sliding.
Note: theappropriate
dynamic analysis
required for spillway
piers and spillway
hoist piers may be part
of an overall dynamic
analysis for a concrete
dam.
Access Bridge arid Piers
OBE: Carry out appropriate dynamic
/of the pier and bridge
' The analysis may be
88 ANCOLD Guidelines for Design
Wke'
*4
part of an overall dynamic
analysis for an intake tower.
The pier and bridge system
should be examined for an
earthquake in the direction of
the bridge access and
perpendicular to the bridge
access. Earthquake loads may
be combined on a square root of
the sum of the squares basis.
Combined static and dynamic
loads should satisfy the
appropriate Structures Code).
MDE: If there are alternative means
of access then the access bridge
can be allowed to fail. If there
are no alternative means of
access then the bridge and piers
should remain capable of
carrying their design loads.
Access Roads
OBE: Access roads to the dam and
its appurtenant works should
remain passable immediately
after the OBE. Therefore any
likely land slip areas along the
access roads should be checked
for stability. There should be no
land slips either during or after
an OBE.
MDE: Access roads to the dam and
its appurtenant works may
become impassable during or
immediately after an MDE.
However,they should be easily
cleared. Therefore,while there
may be land slips onto the
roads,the roads themselves,
should not be allowed to
collapse where there is no easy,
alternative access route.
8.3Intake Towers
8.3.1 Analysis
The analysis method for intake towers is
described here in general terms only. A more
complete description of the method on which
the following general description is based,is
given by Chopra and Goyal (1991) and Goyal
and Chopra (1989).
The method presented here is based on a
simplified dynamic analysis in the frequency
domain using a suitable response spectrum. It
uses an added mass representation of
hydrodynamic effects due to surrounding
(outside) water and contained (inside) water
(in the case of wet towers). In addition,it
includes the effects of tower/foundation
interaction.
The steps of the method are:
(i)Select suitable response spectrum.
(ii)Compute the added hydrodynamic
mass of water using Goyal and Chopra
(1989).
(in) Determine the structural properties of
the tower
mass per unit height
flexural stiffness
modal damping ratios.
(iv)Compute natural periods and mode
shapes for the first two modes of
vibration.
(v)Determine the spectral accelerations
for the first two modes of vibration
from the response spectrum.
(vi)Compute the generalised mass and
generalised excitation terms using the
mass distribution and the mode
shapes.
(vii)Compute the inertia forces using the
spectral accelerations,generalised
mass and excitation terms,mass
distribution and mode shapes.
(viii)Add the inertia loads from the first
two vibration modes on a square root
of the sum of the squares basis.
(ix)Compute bending moments and shear
forces from (viii).
(x)Design tower according to AS3600
(Concrete Structures Code).
8.3.2 Design Criteria
As discussed in Sub-section 8.2,an intake
tower is designed elastically for the OBE.
Generally,it will not be necessary for the
ANCOLD Guidelines for Design of Dams for Earthquake 89
tower to behave elastically during the MDE.
In order to ensure that the tower can survive
intense ground shaking due to the MDE with
limited damage,it should possess a ductility
capacity greater than the ductility
requirements imposed by the ground motion.
A suitable method for this is described in
Chopra and Liaw (1975) where a ductility
factor of two is recommended (ratio of
maximum permissible displacement to the
yield displacement).
/
90 ANCOLD Guidelines for Design of Dams for Earthquake
REFERENCES
/
ANCOLD Guidelines for Design of Dams for Earthquake
REFERENCES
Ambraseys,N.N. (1973). Dynamics and response of foundation materials in epicentral regions of strong
earthquake&Proc. 5th World Conference in Earthquake Engineering,Rome,Italy.
ANCOLD (1991). Guidel ines on Design Criteria for Concrete Gravity Dams.
ANCOLD(1994). Guidelines on Risk Assessment.
ANCOLD (1996). Commentary on ANCOLD Guidelines on Risk Assessment (in preparation).
Argyris,J.A.,Krempl,E. and William,K.J. (1977). Constitutive Models and Finite Element Solution of
Inelastic Behaviour,in Formation and Algorithm in Finite Element Analysis (eds. K.J. Bathe et al).
Barton,N. and Bandis,S. (1991). Review of predictive capabilities of JRC-JCS model in engineering
practice. Pub. No. 182,Norwegian GeotechnicalInstitute,Oslo,Norway,8p.
Barton,N. and Choubey,V (1977). The shear strength of rock joints in theory and practice. Rock
Mechanics 10,1-54.
Bazant,Z.P. and Ob,B.M. (1979). Blunt crack band propagation in finite element analysis.
J.Eng.Mech. Div.,ASCE,105,EM2,297-315.
Bazanl,Z.P. and Cedolin,L. (1981). Concrete fracture via stress strain relations. P.I. Theory,P.2
Verification. Report 81-10/655c,Centre of Concrete and Geomechanics,Northwestern University,
Evanston,Illinois.
BC Hydro (1993). Guidelines for Consequence Based Dam Safety Evaluations and Improvements. BC
Hydro,Report No. H252B,August 1993.
BC Hydro (1995). Guidelines for the Assessment of Rock Foundations of Existing Concrete Gravity
Dams. BC Hydro,ReportNo. MEP67,Vancouver,Canada.
Bicanic,N. and Zienkiewicz; O.C. (1983). Constitutive model for concrete under dynamic loading.
J.EarthquakeEng. and Struct Dyn,Vol 11,689-710.
Bierschwale,J.G. and Stokoe,K.H. (1984). Analytical evaluation of sands subjected to the 1981
Westmorland earthquake. Geotechnical Engineering Report GR-84-15,Civil Engineering
Department,University of Texas.
Brown,R.E. (1977). Vibroflotationcompaction of cohesionless soils. J.Geotech. Eng. Div.,ASCE,Vol
103,NoGT12,1437-1451.
Campanella,R.,Robertson,P.K. and Gillespie^ D. (1986). Seismic cone penetration test,in Use of
In-Situ Tests in Geotechnical Engineering. ASCE Geotechnical Special PublicationNo. 6.
Canadian Dam Safety Association (1994). Draft Dam Safety Guidelines for Existing Dams. Canadian
Dam Safety Association,April 1994.
Castro,G. (1976). Comments on seismic stability evaluation of embankment dams. Proc. Conference
on Evaluation of Dam Safety.
Cedolin,L.,Crutzen,Y.R.J,and Dei Poli,S. (1977). Stress-strain relationship and ultimate strength of
concrete under triaxial loading conditions. Jnl. ofEng.Mechs. Div.,Vol 103,No EM3,423-440.
Cervera,M.,Oliver,J. and Calindo,M. (1992). Numerical analysis of dams with extensive cracking
resulting from concrete hydration; Simulation of a real case. Dam Engineering,Vol III,Issue 3,
February.
Chakrabarti,P. and Chopra,A.K. (1974). Hydrodynamic effects in earthquake response of gravity
dams. J.StructuralEng.,ASCE,100,ST6,1211-1224.
Chapuis,T.,Rebora,B. and Zimmerman,T. (1985). Numerical approach to crack propagation analysis
in gravity dams during earthquakes. Proc. 15th Int. Congress on Large Dams,Lausanne.
Chavez,J.W. and Fenves,G.L. (1993). Earthquake analysis and response of concrete gravity dams
including base sliding. ReportNo UCB/EERC-93/07,University of California,Berkeley.
Chong-Shien Tsai and Lee,G.C. (1989). Hydrodynamic pressure on gravity dams subjected to ground
motions. Jour,of Engineering Mechanics,ASCE,Vol 115,No. 3,March.
Chopra,A.K. (1967). Hydrodynamic pressure on dams during earthquakes. J.Eng. Mech. Div.,ASCE,
93,EM6,205-223.
Chopra,A.K. and Chakrabarti,P. (1972). The earthquake experience at Koyna Dam and stresses in
concrete gravity dams. Earthquake Engineering and Structural Dynamics.
ANCOLD Guidelines for Design ofDams for Earthquake
Chopra,A.K. Chakrabarti P. and Gupta,S. (1980). Earthquake response of concrete dams including
hydrodynamic and foundation interaction effects. EERC Report No UCB/EERC-80/01,University
of California,Berkeley.
Chopra,A.K. and Goyal,A. (1991). Simplified earthquake analysis of intake-outlet towers. Jour,of
Struct Eng.,Vol 117,No 3,March.
Chopra,A.K. and Gupta,S. (1981). Hydrodynamic and foundation interaction effects in earthquake
response of a concrete gravity dam. J.Structural Div.,ASCE,107,ST8,1399-1412.
Chopra,A.K. and Liaw,C-Y (1975). Earthquake resistant design of intake-outlet towers,Jour. Struct
Div.,ASCE,No 577,July.
Chopra,A.K.,Liping Zhang (1991). Earthquake-induced base sliding of concrete gravity dams. Jour,of
Struct Eng.,ASCE,Vol 117,No 12,December.
Clough,R.W. and Chang,C.H. (1980). Seismic cavitation of gravity dam reservoirs. Proc. Int. Conf. on
Num. Meth. for Coupled Problems,Swansea,1985-1996. Pineridge Press,Swansea.
Clough,R.W. and Chopra,A.K. (1966). Earthquake stress analysis in earth dams. Proc. ASCE,Vol 92,
No EM2.
Clough,R.W. and Ghanaat Y. (1993). Concrete Dams: Evaluation for Seismic Loading. Dam
EngineeringACOLDInternational Workshop on Dam Safety Evaluation -Vol IV.
Committee on Safety Criteria for Dams (1985). Safety of Dams. Flood and Earthquake Criteria.
National Research Council,National Academy Press,Washington,DC.
Crone,A. and Machette,M. (1992). Paleoseismicityand the recurrence of surface faulting in the stable
interior of continents -Paleoseismic Investigations in Australia,US Geological Survey Geologic
Division. The Cross Section,Vol 23,No. 7,5-8.
Cundall,P. (1993). FLAC -Fast Langrangian analysis of continua,ITASCA.
Danay,A. and Adeghe,L.N. (1993). Seismic induced slip of concrete gravity dams. Jour,of Struc.
Eng.,ASCE,Vol 119,No 1,January.
Darwin,D. and Pecknold,D.A. (1978). Analysis of R.C. shear panels under cyclic loading. Jour,of
Struc. Eng.,ASCE,102,ST2,355-369.
De Alba,P.,Seed,H.B. and Chan C.K. (1976). Sand liquefaction in large scale simple shear tests. J.
GeotechEng,ASCE,Vol 102,GT9,909-927.
DeKay,M.L. and McClelland,G.H. (1993). Predicting loss of life in cases of dam failure and flash
flood. Risk Analysis,Vol 13,No 2,193-205.
Drumright,E.E.,Pfingsten,C.W. and Lukas,R.G. (1996). Influence of hammer type on SPT results.
J.Geotech.Eng,ASCE,Vol 122,No. 7,598-599.
Dungar,R. (1978). An efficient method of fluid structure complying in the dynamic analysis of
structures. Int. Jour,of Num. Methods in Engineering,Vol 13,No 3.
Dungar,R. (1994). Complexity,uncertainty and realism in the seismic safety evaluation of concrete
dams: linear analysis,Hydropower and Dams,May.
Duron,Z.H. and Hall,J.F. (1988). Experimental and finite element studies of the forced vibration
response of Morrow Point Dam. Jour,of Earthquake Eng. and Struct Dynamics,Vol 16,
November.
Duron,Z.H.,Ostrom,D.K. and Aaagard,B. (1994). Measured steady state,ambient and transient
responses of a small arch dam. Dam Engineering,Vol V,Issue 1,May.
Esteva,L. and Rosenblueth,E. (1969). Espectos de temblores a distancias moderadas y grandes. Bo.
Soc. Mexicano Ing. Simica,2,1-18.
Evernden,J.F. (1975). Seismic intensities,"size" of earthquakes and related parameters. Bull. Seis.
Soc. America,86,1287-1313.
Fear,C.E. and McRoberts,E.C. (1995). Reconsideration of initiation of liquefaction in sandy soils.
/J.Geotech.Eng,ASCE,Vol 121,No. 3,249-261.
Fell,R.,MacGregor,J.P. and Stapledon,D.H. (1992). Geotechnical Engineering of Embankment Dams,
Balkema.
ANCOLD Guidelines for Design ofDams for Earthquake
Fell,R. (1995). Estimating the probability of failure of embankment dams under normal operating
conditions and earthquake loading. Proc. NZSOLD/ANCOLD Symposium on Dams,The
implications of ownership. IPENZ,Vol 21,Issue 1,68-82.
FEMA (1985). Federal Guidelines for Earthquake Analysis and Design of Dams. Federal Emergency
Management Agency,USER (1989). Design Standards No. 13-Embankment Dams,Chapter 13-
Seismic Design and Analysis. US Bureau of Reclamation,Denver.
Fenves,G.L. (1989). Earthquake induced cracking in concrete gravity dams. Seismic Engineering
Structures Congress.
Fenves,G. and Chopra,A.K. (1986). Simplified analysis for earthquake resistant design of concrete
gravity dams. Report No UCB/EERC-85/10,Earthquake Engineering Research Centre,University
of California,Berkeley,June.
Fenves,G. and Chopra,A.K. (1987). Simplified earthquake analysis of concrete gravity dams. Jour,of
Struc. Eng.,ASCE,^Vol 113,No 8,August.
Fenves,G.L. and Mojtahedi,S. (1993). Earthquake response of an arch dam with contraction joint
opening. Dam Engineering,Vol IV,Issue 2,May.
Finn,W.D.L. (1993). Seismic safety evaluation of embankment dams,in International Workshop on
Dam Safety Evaluation,Grundewald,Switzerland,26-27April,Vol 4,91-135.
Finn,W.D.L.,Lee,K.W. and Martin,G.R. (1977). An effective stress model for liquefaction.
J.Geotech. Engng. Div.,ASCE,103,No GT6,517-533.
Finn,W.D.L.,Yogendrakumar,M.,Yoshida,N. and Yoshida,H. (1986). TARA-3: A program to
compute the response of 2-D embankments and soil-structure interaction systems to seismic
loadings. Department of Civil Engineering,University of British Columbia,Canada.
Gao Lin,Jing,Zhou and Chuiyi Fan (1993). Dynamic model rupture test and safety evaluation of
concrete gravity dams. Dam Engineering,Vol IV,Issue 3,August.
Gaull,B.A.,Michael-Leiba,M.O. and Rynn,J.M.W. (1990). Probabilistic earthquake risk maps of
Australia. Australian Journal of Earth Sciences,Vol 37,169-187.
Gerstle,K.H. (1981). Simple formulation of triaxial concrete behaviour. Journal of American Concrete
Inst.,Vol 78,No 5,382-387.
Ghaboussi,J. (1967). Dynamic stress analysis of porous elastic solids saturated with compressible
fluids. PhD Thesis,University of California,Berkeley.
Ghaboussi,J. and Wilson,E.L. (1973). Liquefaction analysis of saturated granular soils. Proc. 5th
World Conference in Earthquake Engineering,Rome,Italy,380-389.
Gibson,G. (1994). Earthquake hazard in Australia. Seminar,Acceptable Risks for Extreme Events in
the Planning and Design of Major Infrastructure. ANCOLD-Munro Centre,26-27April 1994,
Sydney.
Gomberg,J.S. (1992). Tectonic deformation in the New Madrid seismic zone: inferences from
boundary element modeling. SeismologicalResearch Letters,63,3,107-425.
Goodman,R.E. and Seed,H.B. (1966). Earthquake induced displacements in sand embankments.
Journal of the Soil Mechanics and Foundation Division,ASCE,Vol 92,No SM2,125-146.
Goyal,A. and Chopra,A.K. (1989). Simplified evaluation of added hydrodynamic mass for intake
towers. Jour,of Eng. Mech,ASCE,Vol 115,No 7,July.
Greeves,EJ. and Taylor,C.A. (1992). The use of displacement type fluid finite elements for the
analysis of dam-reservoir interaction. Dam Engineering,Vol III,Issue 3,August.
Guthrie,L.G. (1986). Earthquake analysis and design of concrete gravity dams. 3rd US National Conf.
on Earthquake Engineering,Vol 1.
Hackney,G. (1994). Embankment dams,and their probability of failure. BE(Civil) Thesis,School of
Civil Engineering,Universityof New South Wales.
Hall,J.F. (1988). The dynamic and earthquake behaviour of concrete dams: Review of experimental
behaviour and observational evidence. Soil Dynamics and StructuralEngineering,Vol7,No. 2.
Hansen,K.D. and Roehm,L.H. (1979). The response of concrete dams to earthquakes. Water Power
and Dam Construction,April.
ANCOLD Guidelines for Design ofDams for Earthquake
Hardin,B.O. and Drenerich,V.P. (1972). Shear modulus and damping in soils. Design equations and
curves. J.Soil Mech. and Found. Div.,ASCE,98(7).
Hinks,J.L. and Gosschalk,E.M. (1993). Dams and Earthquakes -A review. Dam Engineering,Vol IV,
Issue 1,9-26.
Hoek,E. (1983). Strength of jointed rock masses. Geotechnique33,No 3,187-223.
Hoek,E. (1990). Estimating Mohr-Coulomb friction and cohesion values from the Hoek-Brown failure
criterion. Int. J.Rock Mech. Si. and Geomech. Abstr.,Vol 27,No. 3,227-229.
Hoek,E. (1994). Strength of rock and rock masses. ISRM News Journal,Vol 2,No 2,4-16.
Hoek,E. and Brown,E.T. (1988). The Hoek-Brown failure criterion -A 1988 update. Proc. 15th
Canadian Rock Mechanics Symposium,31-38.
Hoek,E.,Wood,D. and Shah,S. (1992). A modified Hoek-Brown failure criterion for jointed rock
masses. Eurock92. Thomas Telford,London,209-213.
ICOLD (1983). Deterioration of dams and reservoirs,examples and their analysis. International
Commission on Large Dams.
ICOLD(1986). Earthquake Analysis Procedures for Dams-State of the Art. International Commission
on Large Dams,Bulletin 52.
ICOLD (1989). Selecting seismic parameters. InternationalCommission on Large Dams,Bulletin 72.
Idriss,I.M.,Lysmer,J.,Hwang,R. and Seed,H.B. (1973). QUAD-4: A computer program for
evaluating the seismic response of soil structures by variable damping finite element procedures.
Report No EERC73-16,University of California,Berkeley.
Ishihara,K. (1985). Stability of natural deposits during earthquakes. Proc. 11th ICSMFE. Balkema.
Ishihara,K. (1994). Evaluation of residual strength of sandy soils. XIII ICSMFE,Vol 5,175-181,New
Delhi.
Jing Zhow and Gao Lin (1992). Seismic fracture analysis and model testing of concrete gravity dams.
Dam Engineering,Vol III,Issue 1,February.
Johnston,A.C. and Shedlock,K.M. (1992). Overview of research in the New Madrid seismic zone:
inferences from boundary element modeling. Seismological Research Letters,63,3,193-208.
Kawai,T. (1985). Summary report on the development of the computer program DIANA -Dynamic
interaction approach and non-linear analysis. Science University of Tokyo.
Khalili,N. (1994). Post liquefaction analysis,Hume Dam -Section Ch3730. Report for the Public
Works Department,NSW. University of New South Wales,Sydney,Australia.
Kotsovos,M.D. and Newman,J.B. (1978). Generalised stress-strain relations for concrete. Journal of
Eng. Mechs. Div.,ASCE,Vol 104,No EM4,845-856.
Kovacs,W.D. (1994). Effects of SPT equipment and procedures on the design of shallow foundations
on sand. Proc. Settlement 94,ASCE Geotechnical Special Publication,No. 40,Vol 1,Issue 1,
ASCE,55-67.
Landon-Jones,I.,Wellington,N. and Bell,G. (1995). Risk assessment for Prospect Dam. Proc.
NZSOLD/ANCOLD Symposium on Dams,The implications of ownership. IPENZ,Vol 21,Issue 1,
68-82.
Ledbetter,R.H. (1985). Improvement of liquefiable foundation conditions beneath existing structures.
US Corps of Engineers Technical Report REMR-GT-2,August 1985.
Lee,M.K. and Finn,W.D.L. (1978). DESRA-2,Dynamic effective stress response: Analysis of soil
deposits with energy transmitting boundary including assessment of liquefaction potential. Soil
Mechanics Series No 38,Department of Civil Engineering,University of British Columbia,
Vancouver,Canada.
Leger,P. and Katsouli,M. (1989). Seismic stability of concrete gravity dams. Earthquake Engineering
and Structural Dynamics,John Wiley and Sons.
Li,6c.S.,Wang,Z.L. and Shen,C.K. (1992). SUMDES,a nonlinear procedure for response analysis of
horizontally-layered sites subjected to multi-directional earthquake loading. Department of Civil
Engineering,University of California,Davis.
Liao,S.S.C.,Veneziano,D. and Whitman,R.V. (1988). Regression models for evaluating liquefaction
probability. JASCEGeotech.Eng.,Vol 114,No 4.
ANCOLD Guidelines for Design of Dams for Earthquake
Lo,R.L. and Klohn,E.J. (1990). Seismic stability of tailings dams. ICOLD/ANCOLD International
symposium on Safety and Rehabilitation of Tailings Dams,Sydney.
Lysmer,J.,Udaka,T.,Tsai,C.F. and Seed,H.B. (1975). FLUSH -A computer program for approximate
3-D analysis of soil-structure interaction problems. Report No EERC75-30,University of
California,Berkeley.
McCann,M.W.,Franzinic,J.B.,Kavazanyan,E. and Shah,H.C. (1985). Preliminary safety evaluation
of existing dams. Report to Federal Emergency Management Agency,Stanford University,
California.
McCue,K. and Michael-Leiba,M. (1993). Australia's deepest known earthquake. Seismological
Research Letters,64,3-4,201-206.
McEwin,A.,Underwood,R. and Denham,D. (1976). Earthquake risk in Australia,BMR Journal,Vol
l,No 1,15-21.
Makdisi,F.I. and Seed,H.B. (1978). Simplified procedure for estimating dam and embankment
earthquake-induced deformations. Jour,of Geotech. Eng. Div.,ASCE,Vol 104,No GT7,569-867,
July.
Marcuson,W.F.,Hynes,M.E. and Franklin,A.G. (1992). Seismic stability and permanent deformation
analyses: the last twenty five years. Proc. ASCE Specialty Conference on Stability and
Performance of Slopes and Embankments -II,Geotech. Special Publ. No 31 (eds. R.B. Seed and
R.W.Boulanger,ASCE,New York,Vol 1,552-592.
Martin,G.R.,Finn,W.D.L. and Seed,H.B. (1975). Fundamentals of liquefaction under cyclic loading.
Journal of the Geotechnical Engineering Division,ASCE,Vol 101,423-438.
Medina,F.,Domingues,J. and Tassoulas,J.L. (1990). Response of dams to earthquakes including
effects of sediments. Jour,of Struc. Engng Div.,ASCE,101,423-438.
Moriwaki,Y.,Beikae,M. and Idriss,I.M. (1988). Non-linear seismic analysis of the Upper San
Fernando Dam under the 1971 San Fernando earthquake. Proc. 9th World Conference on
Earthquake Engng.,Tokyo and Kyoto,Japan,Vol III,237-241.
Mohraz,G.,Schnobrich,C. and Gomez,A.E. (1970). Crack development in a prestressed concrete
reactor vessel as determined by a lumped parameters methods. Nuclear Eng. and Design,11,
286-294.
National Research Council (1983). Safety of Existing Dams,Evaluation and Improvement. National
Academy Press.
Newmark,N.M. (1965). Effects of earthquakes on dams and embankments. Geotechnique 15,No 2,
139-160.
NSWDSC (1993). Interim Requirements for Seismic Assessment of Dams. New South Wales Dams
Safety Committee Information Sheet DSC 16.
Pal,N. (1974). Non-linear earthquake response of concrete gravity dams. EERC Report No UCB
/EERC 74-14,University of California,Berkeley.
Pande,G.N. and Shen,S. (1982). A two surface multi-laminate model for dynamic analysis of rock
structures. Proc. 4th Int. Conf. of Numerical Methods in Geomechanics,Vol 1,421-426.
Patton,F.D. (1966). Multiple modes of shear failure in rock. Proc. Int. Cong. Rock Mech.,Lisbon,Vol
1,509-513.
Phillips,D.V. and Zienkiewicz,O.C. (1976). Finite element non-linear analysis of concrete structure.
Proc. Int. Civ. Eng.,61,P.2,59-88.
Prevost,J.H. (1981). DYNAFLOW: A nonlinear transient finite element analysis program. Princeton
University,Department of Civil Engineering,Princeton,NJ.
Raphael,J.M. (1984). Tensile strength of concrete. ACI Journal,March-April.
Rashid,Y.R. (1968). Ultimate strength analysis of prestressed concrete pressure vessels. Nuclear Eng.
/and Design,7,334-344.
Robertson,P.K. and Fear,C.E. (1995). Liquefaction of sand and its evaluation. IS Tokyo 95. First Int.
Conf. on Earthquake Geotech. Eng.
Robertson,P.K. and Fear,C.E. (1996). Soil liquefaction and its evaluation. Proc. NCEER Liquefaction
Workshop.
ANCOLD Guidelines for Design of Dams for Earthquake
Roth,W.H. (1985). Evaluation of earthquake induced deformations of Pleasant Valley Dam. Report for
the City of Los Angeles. Dames and Moore,Los Angeles.
Salmon,G.M. (1995). Canadian Dam Safety Guidelines. Their adoption by regulatory agencies in the
light of trends towards utility deregulation. Proc. NZSOLD/ANCOLD Symposium on Dams,The
implications of ownership. IPENZ,Vol 21,Issue 1.
Salmon,G.M. and Hartford,D.N.D. (1995(a)). Risk analysis for dam safety. Int. Water Power and Dam
Construction,March,42-47.
Salmon,G.M. and Hartford,D.N.D. (1995(b)). Risk analysis for dam safety -Part II. Int. Water Power
and Dam Construction,April,38-39.
Salmon,G.M. and von Hehn,G.R. (1993). Consequence based dam safety criteria for floods and
earthquakes,in International Workshop on Dam Safety Evaluation,Grundewald,Switzerland,26-27
April,Vol 3,55-62.
Sarma,S.K. (1975). Seismic stability of earth dams and embankments. Geotechnique,Vol 25,No 4,
743-761.
Schnabel,P.L.,Lysmer,J. and Seed,H.B. (1972). SHAKE: A computer program for earthquake
response analysis of horizontally layered sites. Report No EERC72-12,University of California,
Berkeley.
Scott,G.A. and Von Thun,J.L. (1993). Interim Guidelines -Geotechnical Studies for Concrete Dams,
USBR,January.
Seed,H.B. (1979). Consideration in the earthquake resistant design of earth and rockfill dams.
Geotechnique 29,215-263.
Seed,H.B. (1983). Earthquake resistant design of earth dams. Seismic design of embankments and
caverns. Proc. of ASCE Symposium,Philadelphia,41-64.
Seed,H.B.,Seed,R.B.,Lai,S.S. and Khemenchpour,B. (1985). Seismic design of concrete faced
rockfill dams,in Concrete Faced Rockfill Dams (eds. J.B. Cook and J.L. Sherard),ASCE,Detroit,
MI,459-478.
Seed,H.B. and De Alba,P. (1986). Use of the SPT and CPT tests for evaluating the liquefaction
resistance of sands. In-Situ 86,Conference on Use of In-Situ Tests in Geotechnical Engineering.
S.P. Clemence (ed). ASCE Geotechnical Special PublicationNo. 6.
Seed,H.B. and Harder,L.F. (1990). SPT-based analysis of cyclic pore pressure generation and
undrained residual strength. In H. Bolton Seed Memorial Symposium Proceedings,BiTech
Publishers,Vancouver,Canada.
Seed,H.B. and Idriss,I.M. (1982). Ground motions and soil liquefaction during earthquakes.
Monograph Series,Earthquake Engineering Research Institute,Berkeley,California.
Seed,H.B.,Idriss,I.M. and Arango,I. (1983). Evaluation of liquefaction potential using field
performance data. ASCE,J.Geotech. Engng,109(3),458-482.
Seed,H.B.,Lee,K.L.,Idriss,I.M. and Makdisi,F.I. (1973). Analysis of slides in the San Fernando
Dams during the earthquake of February 9 1971. Report No EERC/73-2,University of California,
Berkeley.
Seed,H.B.,Makdisi,F.I. and De Alba,P. (1978). Performance of earth dams during earthquakes.
J.Geotech. Engng. Div.,104,No GT7.
Seed,H.B. and Martin,G.R. (1966). The seismic coefficient of earth dam design. Journal of the Soil
Mechanics and Foundation Division,ASCE,Vol 92,No SM3,25-48.
Seed,H.B.,Wong,R.T.,Idriss,I.M. and Tokimatsu (1986). Moduli and damping factors for dynamic
analysis of cohesionlesssoils. J.Geotech. Engng,ASCE,112(11).
Serafim,J.L. (1981). Criteria for the earthquake resistant design of concrete dams. Dams and
Earthquake,TTL,London.
Sherard,J.L. (1967). Earthquake considerations in earth dam design. Journal of the Soil Mechanics and
Foundations Division,American Society of Civil Engineers,Vol 93,No. SM4,July,377-401.
Sherard,J.L.,Cluff,L.S. and Allen,C.R. (1974). Potentially active faults in dam foundations.
Geotechnique^ Vol 24,No 3,367-428.
ANCOLD Guidelines for Design of Dams for Earthquake
Silver,M.L. (1985). Remedial measures to improve the seismic strength of embankment dams. Report
No 85-10,Department of Civil Engineering,University of Illinois,Chicago,Illinois.
Standards Australia (1993). Minimum Design Loads on Structures,Part 4: Earthquake Loads.
Australian Standard 1170.4-1993,Sydney.
Stark,T. and Mesri,G. (1993). Undrained shear strength of liquefied sands for stability analysis.
JASCE Geotechnical Engineering,Vol 118,No 11,1727-1747.
Swanson (1992). A computer program designed by Swanson Analysis Systems,Inc.
Troncoso,J.H. (1990). Failure risk of abandoned tailings dams. International Symposium on Safety and
Rehabilitation of Tailings Dams,ICOLD/ANCOLD,Sydney.
Troncoso,J.H.,Ishihara,K. and Verdugo,R. (1988). Aging effects on cyclic shear strength of tailings
materials. Proc. IX World Conference on Earthquake Engineering,Kyoto.
Tsai,Chong-Shien and Lee,C. (1989). Hydrodynamic pressure on gravity dam subjected to ground
motions. Jour. Eng. Mech.,ASCE,Vol 115,No 3,March.
Tsuchida (1970). Prediction and countermeasure against the liquefaction in sand deposits. In Abstract
of the Seminar in the Port and Harbour Research Institute (in Japanese),31-333.
USBR(1977). Design criteria for concrete arch and gravity dams. EngineeringMonographNo 19.
USER (1989(a)). Policy and Procedures for Dam Safety Modification and Decision Making. US
Bureau of Reclamation,Denver.
USBR (1989(a)). Design Standard % Embankment Dams,No 13,Seismic Design and Analysis.
USCOLD (1992). Observed performance of dams during earthquakes.
US Corps of Engineers (1984). Rationalising the seismic coefficient method. Miscellaneous Paper
GL84-13.
US National Research Council (1985). Liquefaction of soils during earthquakes. National Academy
Press,Washington DC.
Waggoner,F.,Plizzari,G. and Saouma,V.E. (1993). Centrifuge tests of concrete gravity dams. Dams
Engineering,Vol IV,Issue 3,August.
Westergaard,H.M. (1933). Water pressure on dams during earthquake. Transaction,ASCE,98,
418-433(see also Proc. ASCE,November 1931,1303-1318).
White,W.,Valliappan,S. and Lee,I.K. (1979). Finite element mesh constraints for wave propagation
problems. Proc. of the Third International Conference in Australia on Finite Element Methods. The
University of New South Wales,Sydney,531-539.
William,K.J. and Wamke,E.P. (1975). Constitutive model for the triaxial behaviour of concrete. Int.
Assoc. for Bridge and Struct. Eng. Procs.,Vol 19.
Zangar,C.N. (1952). Hydrodynamic pressure on dams due to horizontal earthquake effects. US Bureau
of Reclamation.
Zienkiewicz,O.C.,Clough,R.W. and Seed,H.B. (1986). Earthquake analysis procedures for dams -
state of the art. ICOLD/CIGB,Bulletin No 52. International Commission on Large Dams,Paris,
148pp.
Zienkiewicz,O.C.,Fejzo,R. and Bicanic,N. (1983). Experience in analysis plane concrete structures
using a rate sensitive model with crack monitoring capabilities. Proc. of the Int. Conf. on
Constitutive Laws for Engineering Materials,University of Arizona,Tucson,January.
Zienkiewicz,O.C.,Hinton,E.,Bicanic,N. and Fejzo,R. (1980). Computational models for the transient
dynamic analysis of concrete dams. Proc. Conf. on Design of Dams to Resist Earthquakes,Int. Civ.
Eng.,171-178,London,October.
Zienkiewicz,O.C.,Paul,D.K. and Hinton,E. (1983). Cavitation in fluid-structure response (with
particular reference to dams under earthquake loading). J.EarthquakeEng. and Struct. Dyn,11.
Zienkiewicz,O.C. and Shiomi,T. (1984). Dynamic behaviour of saturated porous media: the
generalised biot formulation and its numerical solution. Int. J.Num. and Anal. Meth. in Geomech,8,
71-96.
Zienkiewicz,O.C.,Valliappan,S. and King,LP. (1986). Stress analysis of rock as a non-tension
material. Geotechnique 18,55-56.
ZienkiewiczO.C. and Xie,Y.M. (1991). Analysis of the Lower San Fernando Dam failure under
earthquake. Dam Engineering,2,302-322.
ANCOLD Guidelines for Design of Dams for Earthquake
APPENDICES
/
ANCOLD Guidelines for Design of Dams for Earthquake
APPENDIX A
AUSTRALIAN NATIONAL COMMITTEE ON LARGE DAMS
GUIDELINES FOR THE DESIGN OF DAMS
FOR EARTHQUAKE
TERMS OF REFERENCE
OBJECTIVE
To prepare guidelines for the design of dams in Australia for earthquake. The guidelines are to be
sufficiently detailed to allow users to carry out preliminary assessments without reference to other
material,and to give guidance on the methods available,their advantages and shortcomings for more
detailed assessment where this is shown to be necessary by the preliminary assessment.
SCOPE OF GUIDELINES
1. Type of Dam -the guideline is to cover all dam types
2. Earthquake Loading in Australia
An overview of earthquake activity in Australia
Description of design earthquakes and acceptable damage
design basis earthquake
low probability events (eg. Maxm Credible Earthquake)
concurrent load combinations
relation to risk assessment
dam and ancillary structures
Calculation of design loadings from design earthquakes
attenuation functions applicable in Australia
effect of local geology,ground conditions,topography
peak accelerations,spectra,components
Methods for calculating design earthquakes
probabilistic methods
deterministic methods (relating to known faults etc)
who should make this assessment
/
ANCOLD Guidelines for Design ofDams for Earthquake
3. Analysis of Effect of Earthquake Loading
For embankment dams
simplified "screening" methods to assess liquefaction potential and stability
"second level" methods for analysing deformation
"advanced level" analysis methods,eg. dynamic finite element
data requirements for the above
acceptability criteria for the methods as related to stage of design,dam hazard
rating and type
For concrete dams (gravity,arch,buttress etc)
simplified "screening" methods to assess stability
"second level" methods for analysing stability,eg. 2D finite element
"advanced level" analysis methods,eg. including crack propagation,3D effects
data requirements for the above
acceptability criteria for the methods as related to the stage of design,dam
hazard rating and type
For tailings dams
specific issues compared to embankment dams
For ancillary structures
simplified screening methods
second level methods
advanced level methods
data requirements for the above
acceptability criteria as related to the stage of design,dam hazard rating and
type of ancillary structure
4. Defensive Design for Earthquake
Details of features which should be incorporated into dam designs which will reduce the
probability of failure under earthquake. Issues include,for example,freeboard,filters,materials
selection and zoning,drains.
5. Other Features of Earthquake Design
Seiche
-Landslides in reservoir
-Earthquakes induced by dam
(not to be covered in detail)
ANCOLD Guidelines for Design ofDams for Earthquake
APPENDIX B
-TYPICAL EASTERN AUSTRALIAN PEAK GROUND
ACCELERATION VS AEP
-RESPONSE SPECTRUM FOR 1 IN 1000AEP
-MODIFIED MERCALLI SCALE
ANCOLD Guidelines for Design of Dams for Earthquake
Earthquake Ground Motion Recurrence
PEAK GROUND ACCELERATION RECURRENCE
Tectonic Model using Esteva S Rosenblueth 1964Attenuation
25 10205010020050010002000500010000
Return Period (Years)
Peak Ground Acceleration Recurrence
Earthquake Ground Motion Recurrence
RESPONSE SPECTRUM -Trifunac 1980Attenuation
Tectonic Model Return Period 1000years
Probability of Exceedance =0.50,Horizontal motion,No sediments
0.5i 25
Frequency (hertz)
Response Recurrence for 1000Year Return Period
TABLE H.4
MODIFIED MERCAULI SCALE,1956 VERSION*
Intensity Effocu v.t cm/t r*
M I. Not feit. Marginal and long-pertod effects of large earthquakes (for
details seetext).
3II. Felt by persons at rest,on upper floors,or favorably placed.
III. Felt indoors Hangingobjects swing. Vibrationlikepassingof light trucks.
Durationestimated. May not berecognizedas anearthquake.
IV. Hangingobjects swing. Vibrationlikepassingof heavy trucks; or
4sensation of a jolt like a heavy ball striking the walls. Standing motor cars
rock. Windows,dishes,doors rattle. Glasses clink. Crockery clashes In
theupper rangeof IV woodenwalls andframecreak.
V Felt outdoors; directionestimated. Sleepers wakenedLiquids disturbed,
somespilled. Small unstableobjects displacedor upset. Doors swing,
close,open. Shutters,pictures move. Pendulumclocks stop,start,change
rate.
VI. Felt by all. Many frightenedandrunoutdoors. Persons walk unsteadily.
5Windows,dishes,glassware broken. Knickknacks,books,etc.,off shelves.
Pictures off walls. Furnituremovedor overturned. Weak plaster and
masonry D cracked. Small bells ring(church,school). Trees,bushes
shaken(visibly,or heardtorustleCFR).
VII. Difficult tostand. Noticedby drivers of motor cars. Hangingobjects
quiver. Furniturebroken. Damagetomasonry D,includingcracks. Weak
chimneys brokenat roof line. Fall of plaster,loosebricks,stones,tiles,
cornices (alsounbracedparapets andarchitectural ornamentsCFR).
6Some cracks in masonry C. Waves on ponds; water turbid with mud.
Small slides andcavinginalongsandor gravel banks Largebells ring.
Concreteirrigationditches damaged.
VIII. Steeringof motor cars affected. Damagetomasonry C; partial collapse.
Somedamagetomasonry B; nonetomasonry A. Fall of stuccoandsome
masonry walls. Twisting,fall of chimneys,factory stacks,monuments,
towers,elevatedtanks. Framehouses movedonfoundations if not bolted
down; loosepanel walls thrownout. Decayedpilingbrokenoff. Branches
brokenfromtrees. Changes inflow or temperatureof springs andwells.
Cracks inwet groundandonsteep slopes.
IX. General panic. Masonry D destroyed; masonry C heavily damaged,
sometimes with completecollapse; masonry B seriously damaged.
7(General damage to foundationsCFR.) Frame structures,if not bolted,
shiftedoff foundations Frames racked. Serious damagetoreservoirs.
Undergroundpipes broken. Conspicuous cracks inground. Inalluviated
areas sandandmudejected,earthquakefountains,sandcraters.
X. Most masonry andframestructures destroyedwith their foundations.
Somewell-built woodenstructures andbridges destroyed. Serious
8 damage to dams,dikes,embankments. Large landslides. Water thrown on
banks of canals,rivers,lakes,etc. Sandandmudshiftedhorizontally on
beaches andflat land. Rails bent slightly.
XI. Rails bent greatly Undergroundpipelines completely out of service.
XII. Damagenearly total. Largerock masses displaced. Lines of sight and
level distortedObjects thrownintotheair.
00035-0007
0007-0.015
1-3 0.015-0035
3-7 0.035-0.07
7-20 0.07-0.15
20-60 0.15-0.35
60-200 035-0.7
200-500 0.7-1.2
>1 2
FromFig. 11.14
NOTE. Masonry A,B,C,D. Toavoidambiguity of language,thequality of masonry,brick or otherwise,isspecifiedby thefollowinglettering(which hasnoconnection
with theconventional ClassA. 8,C construction)
Masonry A. Goodworkmanship,mortar,anddesign,reinforced,especially laterally,andboundtogether by usingsteel,concrete,etc.; designedtoresist lateral
forces.
Masonry 8*Goodworkmanship andmortar; reinforced,but not designedtoresist lateral forces
Masonry C. Ordinary workmanship andmortar,noextremeweaknessessuch asnon-lied-mcorners,but masonry isneither reinforcednor designedagainst
horizontal forces.
Masonry 0*Weak materials,such asadobe,poor mortar,low standardsof workmanship,weak horizontally
FromRlchter|l958|1 Adaptedwith permissionof W H FreemanandCompany
^Averagept/ak groundvelocify,cm/s
f Averagepeak acceleration(away fromsource)
^Magnitudecorrelation
APPENDIX C
EXTRACTS FROM
CANADIAN DAM SAFETY GUIDELINES
ANCOLD Guidelines for Design of Dams for Earthquake
CDSA
Dam Safety Guidelines
1.4CLASSIFICATION OF DAMS
Requirement: Each dam shall be classified In terms of the reasonably
foreseeable consequences of failure. Each water retaining
structure,including water passages,shall be classified
separately.
Each dam should be classified in accordance with the consequences of failure. The
classification constitutes the basis for analysing the dam's safety and setting appropriate
levels of surveillance activities. Table 1-1 represents a commonly-accepted classification
system which is based on the potential loss of life and economic damages associated
vrfth dam failure. This classification system is used to link the consequences of failure
to the requirements contained in Sections 2 through 10.
Aftemative classification systems may be adopted for interpreting and addressing the
requirements for dam surveillance and dam safety reviews,as set out in Sections 2 and
3of these Guidelines. Such classification systems may incorporate the physical
characteristics of the dam,its condition and the perceived risk of its failure,as well as
the consequences of failure.
Appurtenances may be classified and evaluated separately. Thus the water passages
could be in a different category from the dam,depending on the consequences of failure.
If warning systems are considered to reduce the potential loss of life,the reliability of
such warning systems must be incorporated into all analyses and evaluations.
The consequence categories listed in Table 1-1 are based on the incremental losses
which a failure might inflict on downstream or upstream areas or at the dam.
"Incremental losses" are those over and above losses which might have occurred for the
same natural event or conditions,had the dam not failed.
The distinction between consequence categories,and the link with safety requirements,
is intended to reflect society's values and priorities in allocating and distributing resources
and funds to be used for protecting and saving lives,and for safeguarding property.
The incremental consequences of dam failure should be evaluated in terms of:
Loss of life
Economic value of other losses and/or damage to property,fadi'rties,other utilities
and dam,as well as loss of power generation or water supply. Where appropriate,
costs are assigned to social,cultural and environmental impacts.
Other less quantifiable consequences related to social,cultural and environmental
damages.
The most severe consequences should prevail -if economic losses are Very High and
loss of life is High,the dam would be classified as a Very High Consequence dam.
/Evaluation of potential losses,both with and without dam failure,should be based on
inundation studies and should consider existing and anticipated future downstream
development and land uses. At the same time,the appropriate study level of inundation
would depend on the potential consequences of failure. For dams where there is
uncertainty about the consequences of a dam break,a simplified and conservative
January 1,1995
Page 1-3
CDSA
Dam Safety Guidelines
mmmmmmmmvinimw iMMVk**
analysis should be used to make a preliminary assessment. If this analysis demonstrates
a potential hazard,a more sophisticated analysis should then be undertaken. In the case
of dams where the consequences of failure clearly fall within the "Very Low" category,
a formal inundation study is not required.
A dam may be in one category for flood hazard and a different category for earthquakes,
depending on the incremental damage attributable to dam failure from each cause.
A screening level estimation of the incremental consequences of failure may be
appropriate for a dam to be classified in the Low Consequence category. However,if
a dam is likely to be classified in the High or Very High Consequence categories,the
evaluation of incremental consequences of failure should be based on she-specific
analysis,and may require detailed site investigation.
Consequences of dam failure due to earthquakes should be based on average discharge
conditions and maximum normal operating levels. Consequences attributable to reservoir
slope failure or slope-failure-induced waves should be based on average discharge and
maximum normal operating levels,unless the slide would have been induced by extreme
rainfall associated with an extreme flood.
January 1,1995
Page 1-4
CDSA Dam Safety Guidelines
TABLE 1-1
CONSEQUENCE CLASSIFICATION OF DAMS
CONSEQUENCE
CATEGORY
POTENTIAL INCREMENTAL
CONSEQUENCES OF FAILUREl*1
LOSS OF LIFE ECONOMIC,SOCIAL,ENVIRONMENTAL
VERY HIGH
Large increase expected ^ Excessiveincreaseinsodal,economic
and/or environmental losses.
HIGH
Some increase expected ^ Substantial increaseinsodal,economic
and/or environmental losses.
LOW
Noincreaseexpected Low sodal,economic and/or environmental
losses.
VERY LOW No increase Small dams with mtnimal sodal,economic
and/br environmental losses. Losses
generally limitedtotheowner's property;
damages toother property areacceptable
tosodety.
[a] Incremental to the impacts which would occur under the same natural conditions
(flood,earthquake or other event) but without failure of the dam. The type of
consequence(e.g. loss of life,or economic losses) with thehighest ratingdetermines
which category is assignedtothestructure.
[b] The loss-of-life criteria which separate the High and Very High categories may be
basedonrisks which areacceptableor tolerabletosociety,takentobe0.001 lives
per year for each dam. Consistent with this tolerable societal risk,the minimum
criteria for a Very High Consequence dam (PMF and MCE) should result in an
annual probability of failureless than1/100,000.
January 1,1995 Page 1-5
CDSA
Dam Safety Guidelines
MMNMMMMMMMimMliilMlttWrtMiliMMff*-
1.5 SELECTION OF SAFETY CRtTERIA
Requirements: The dam,alongj with Its foundation and abutments,shall have
adequate stability to safely withstand extreme loads as well
as the normal design loads.
The selection of loading criteria for extreme loads shall be
based on the consequences of failure of the dam.
Methods to determine appropriate normal design loadings and factors of safety are
covered in Sections 5 through 9 of this document. Sections 5 and 6 address earthquake
loadings and floods,respectively.
To select criteria for extreme events,a risk-based approach may be used. The principle
is that a dam whose failure would cause excessive damage or the loss of many lives
should be designed to a proportionately higher standard than a dam whose failure would
result in less damage or fewer lives lost. In assessing the safety of an existing dam,
probabilistic risk analysis (PRA) methods can help verify that qualitative factors such as
internal erosion and spillway debris blockage are not overlooked and that they receive
attention commensurate with their contribution to the failure probability. The level of
safety of a dam can sometimes be improved by addressing conditions less severe but
more likely than those associated with such extreme events as the Maximum Credible
Earthquake (MCE) and the Probable Maximum Flood (PMF).
Criteria for extreme events other than floods and earthquakes should be consistent with
the levels required for flood and earthquakes.
January 1,1B95 Page 1-6
CDSA Dam Safety Guidelines
5.0EARTHQUAKES M
The use of criteria other than those indicated in this document may be appropriate
or necessary,taking into account specific conditions arising at some dam projects,
and to permit development in the application and use of new knowledge and
improved techniques.
Requirement: Dams shall be designed and evaluated to withstand ground
motions associated with a Maximum Design Earthquake
(MDE),without release of the reservoir.
Selection of the MDE for a dam shall be based on the
consequences of dam failure.
The MDE is usually represented by the most severe ground motion which has been
selected for design or safety evaluation of the dam. Site-specific ground motion
parameters required for design or evaluation are determined from the MDE.
For a given site,the MDE should increase with increasing consequences of dam failure,
as illustrated in Table 5-1. For a given Annual Exceedance Probability (AEP),the MDE
from site to site may also vary with the tectonic setting of the site and the distance from
the earthquake source. In some cases,the MDE selection may be based on seismic
loading that could be triggered by human activity,some examples being extraction or
injection for oil fields,and reservoir-induced seismicity.
The development of site-specific seismic parameters such as ground velocities,
accelerations and response spectra,shall be derived from the criteria for design
earthquakes in Table 5-1. Methods to achieve this should conform to currently accepted
practice in different regions of Canada. Derivation of seismic parameters should be
undertaken or supervised by persons with appropriate specialties in earthquake
engineering.
Well built embankment dams that are sited on firm non-fiquefiable foundations and that
do not incorporate large bodies of materials which,if saturated,might lose most of their
strength during earthquakes,can be designed and evaluated using seismic coefficient
methods (pseudostatic analysis) under the conditions outlined in Section 8.1. The
seismic coefficient should reflect the seismicity of the dam site and it can be obtained
from zoning maps created for that purpose.
/
[a] This section addresses criteria for design earthquakes only. The requirements for
structural resistance to the earthquake are presented in Sections 8 and 9.
January 1. 1995Page 5-1
CDSA
Dam Safety Guidelines
mmmommmmmmmimiitmimmsm
TABLE 5-1
USUAL MINIMUM CRITERIA FOR DESIGN EARTHQUAKES
CONSEQUENCE
CATEGORY
MAXIMUM DESIGN EARTHQUAKE (MDE)
DETERMINISTICALLY DERIVED
PROBABILISTICALLY DERIVED
(Annual exceedance probability)
Very High
MCE MM [d] 1/10,000M M
High
50% to 100% MCE[e] 1/1000to 1/10,000W
Low
. [g] 1/100to 1/1000k!
[a] See Section 1.4for consequence classification.
[b] For a recognized fault or geographically defined tectonic province,the Maximum Credible
Earthquake(MCE) is thelargest reasonably conceivableearthquakethat appears possible.
For a damsite,MCE groundmotions arethemost severegroundmotions capableof being
producedat thesiteunder thepresently knownor interpretedtectonic framework.
[c] In Hydro-Quebec's practice,the MDE for Very High Consequence structures involves a
combinationof deterministic andprobabilistic approaches that reflect current knowledgeof
seismo-tectonic conditions inEasternCanada. Hydro-Qu6bec's deterministically derivedMDE
magnitudeis themaximumhistorically recordedearthquake,increasedby one-half magnitude,
whiletheir probabilistically derivedearthquakehas anestimatedprobability of exceedanceof
1/2000.
[d] An appropriate level of conservatism shall be applied to the factor of safety calculated from
these loads,to reduce the risks of dam failure to tolerable values. Thus,the probability of
dam failurecouldbemuch lower thantheprobability of extremeevent loading.
[e] MDE firmgroundaccelerations andvelocities canbetakenas 50% to100% of MCE values.
For design purposes the magnitude should remain the same as the MCE
[f] In the High Consequence category,the MDE is based on the consequences of failure. For
example,if one incremental fatality would result from failure,an AEP of 1/1000could be
acceptable,but for consequences approaching those of a Very High Consequence dam,
designearthquakes approachingtheMCE wouldberequired.
[g] If a Low Consequencestructurecannot withstandtheminimumcriteria,thelevel of upgrading
may bedeterminedby economic risk analysis,with considerattenof environmental andsocial
impacts.
January 1,1995
mmMXfwtttMttrfi'itftrtmrfmpti nlyrrfr
Page 5-2
APPENDIX D
ADDITIONAL INFORMATION ON
ACCEPTABLE RISKS
ANCOLD Guidelines for Design of Dams for Earthquake
REFERENCES
1. ANCOLD (Australian National Committee on Large Dams),"Guidelines on Risk
Assessment",January 1994.
2. Appleyard,L.D. and Narwar,G.,"Risk Assessment in the Development of
Serviceability Based Design Criteria for Small Buildings",Proceedings of the
Conference on Probabilistic Risk and Hazard Assessment,Newcastle NSW,22-23
September 1993.
3. BC Hydro,"Guidelines for Consequence Based Dam Safety Evaluations and
Improvements",Report No. H2528,Hydroelectric Engineering Division,Bumaby,
BC,Canada,August 1993.
4. CERIA (Construction Industry Research and Information Association),
"Rationalisation of Safety and Serviceability Factors in Structural Codes",Report 63,
London,July 1977.
5. DOP (Department of Planning),New South Wales,"Risk Criteria for Land Use Safety
Planning",Hazardous Industry Planning Advisory Paper No. 4,1992.
6. Higson,D.J.,"Nuclear Safety Assessment Criteria" Nuclear Safety,Vol. 31,No. 2,
April-June 1990.
7. HSE (Health and Safety Executive,United Kingdom),"Risk Criteria for Land Use
Planning in the Vicinity of Major Industrial Hazards",London,Her Majesty's
Stationery Office,1989.
8. HSE (Health and Safety Executive,United Kingdom),"Quantified Risk Assessment:
Its Input to Decision Making",London,Her Majesty's Stationery Office,1989.
9. Kletz,T.A.,"The Application of Hazard Analysis to Risks to the Public at Large".
Paper presented to World Congress of Chemical Engineering,Amsterdam,1976.
10. Reid,S.G.,"Practical Procedures for Setting Standards",Lecture 8,One Day Post
graduate Course on Engineering Risk Assessment,University of Sydney,8th March,
1991.
11. Technical Advisory Committee on Water Defences,Centre for Civil Engineering
Research and Codes,"Probabilistic Design of Flood Defences",Gouda,The
Netherlands,June 1990.
12. The Royal Society of London,"Risk Assessment: A Study Group Report",The Royal
Society,London,1983.
13. Whitman,R.V.,"Evaluating CaJculated Risk in Geotechnical Engineering",Journal of
Geotechnical Engineering,ASCE,Vol. 110,No. 2,February 1984,pp. 145-188.
14. Wilson,R. and Crouch,E.A.C.,"Risk Assessment and Comparisons: An
Introduction",Science,Vol. 236,April 1987.
/
10
ig
10
10
\
-X
|
i
i
i
i
i
i

i
i
i
i
1
S,\
\
\
\
\\
\\
XK
% \
\ -r-s
//
//
//
^
!
I
1
I
I
1
ccnerall y
accepted
range of
maxi mum
tolerable
risk to an /
individual^
(j
\\
\\ \
f \Xx^
\\T-
\% X\
\ < ^
1 \ \
i
V<r ! \\
V,W.
\++-\ \
\ \ \-p
Vo. '
V
\ X
\
\
\
s
\
\
\
\
^ proposec
intolera
an i no i
b. c. hyorb ^
ble risk toxoa
vidual
v
c \
s-<\
^ \
\\V'<n4
%
v\-
NaVi>
\ \
\
\
\
\
\
\
\
\ i
\
\ 1
\
\
\
N j
\l
w
&
vrt,
\.
<? vt
\Y? X
^ \
0-1 I 10100100010000
potential number of premature fatalities in the event of a dam incident
/
COMPARISON OF PROPOSED INDIVIDUAL AND SOCIETAL RISK CRITERIA
AND RISK CRITERIA USED IN THE NETHERLANDS AND UNITED
KINGDOM (BC HYDRO,1993,AND ANCOLD,1994).
COMPARATIVE RISK LEVELS
The risk levels quoted in the following Individual Risk table are the additional
increment of risk arising from the activity in question; except for the first two risks.
The risk levels are as quoted in the references given in brackets (see Attachment C for
references).
Individual Risk is the total additional increment of risk imposed by a facility (such as
a dam) on a particular person. All such risks may be averaged over a particular
group.
Societal Risk is a measure of society's aversion to loss of life,especially to
catastrophes involving large loss of life. It has no regard to the identities of
persons and is related to the occurrence of an event at a facility.
COMPARATIVE INDIVIDUAL RISK LEVELS
INDIVIDUAL RISK
SITUATION
chanceper personper year
Male,age 50,UK (7)
1 in 55
Female,age 20,UK (7)-
1 in 3000
Rock climbing,UK (7)
1 in 125
Smoking 20cigarettes per day (6)
1 in 200
Smoking 20cigarettes per day,persons at risk (5)
1 in 200
Commercial Diving (3)
1 in 350
Deep Sea Fishing,US (10)
1 in 360
Parachuting,US (10)
1 in 530
Offshore Oil/Gas Workers (12) I in600
Hang GUding,UK (7)
1 in 670
Air Crew,persons at risk (4)
1 in 1,000
Mountaineering,persons at risk (14)
1 in 1,660
Drinking,alcohol,persons at risk (5)
1 in 2,600
Car Travel,persons at risk (2)
1 in 2,800
Road Accidents,US (10)
1 in 3,300
Quarry Workers (10)
1 in 3,400
Car Travel,10,000km/year,British Columbia (3) 1 in 3,500
Coal Mining,US (10)
1 in 4,800
Road Accidents,whole population (6)
1 in 5,000
Air Pollution,Eastern USA (14)
1 in 5,000
Coal Mining,UK (7)
1 in 9.500
/
Commercial Air Travel,whole population (2)
1 in 10,000
Road Accidents,UK (7)
I in 10,000
Construction Worker,UK (7)
I in 10,800
Drinking I bottle wine/day (9)
1 in 13,300
Swimrrung,persons at risk (6)
1 in 20,000
Pedestrians struck by Car,whole population (5)
1 in 28,600
Air Travel (3)
1 in 30,000
Train Travel,persons at risk (6)
1 in 33,000
Drowning,US (10)
1 in 33,300
Homicide,whole population (5)
I in 50,000
Domestic electrocution,Canada (3)
I in65,000
Air Travel,persons at risk (6)
1 in 100,000
Petrochemical,criterion,maximum to public (9)
1 in 100,000
Dam Failure,limit criterion,exposed persons (I)
1 in 100,000
Industrial plants,HSE limit,most exposed group (7)
1 in 100,000
Nuclear plants,ICRP limit,to public (6)
1 in 100,000
Air Travel,US (10)
I in 110,000
Drowning,UK (7)
I in 170,000
Electrocution (14)
1 in 190,000
Building fires,Australia (10)
I in250,000
Railway travel,US (10)
1 in 250,000
Electrocution,whole population (6)
1 in 330,000
Earthquake,California (9)
1 in 590,000
Dam Failure,objective criterion,exposed persons (1)
I in 1,000,000
Industrial plants,DOP objective criterion (5)
1 in 1,000,000
Industrial plants,HSE objective,most exposed roup (7)
1 in 1,000,000
Argentine nuclear standards,limit criterion(6)
1 in 1,000,000
Nuclear plants,USNRC objective,prompt fatality (6)
1 in 2,000,000
Nuclear reactor accidents (3)
1 in 2,500,000to
1 in 5,000,000
Building collapse,persons at risk (2)
1 in 5,000,000
Structural collapse,UK (10)
1 in 7,000,000
Lightning strike,UK (7)
1 in 10,000,000
Lightning strike,whole population (6)
1 in 10,000,000
Lightning,whole population (5)
1 in 10,000,000
Petrochemical,criterion,average to the public (9)
1 in 10,000,000
Nuclear power station at 1km,UK (9)
1 in 10,000,000
Nuclear power station at boundary,US (9)
1 in 10,000,000
Flooding,North Sea Dikes (9)
1 in 10,000,000
Building Collapse,whole population (2)
1 in 10,000,000
Structural Failure (4)
1 in 10,000,000
J Meteorite strike,whole population (6)
I in 1,000,000,000
f Meteorite strike,whole population (5)
1 in 1,000,000,000
/
Abbreviations:
DOP
HSE
ICRP
USNRC
Department of Planning NSW
Health and Safety Executive,UK
International Commission for Radiological Protection
US Nuclear Regulatory Commission.
COMPARATIVE SOCIETAL RISK LEVELS
L -Limit O -Objective NA -Not applicable
SOURCE
SOCIETAL RISK CRITERIA -chance per facility per year
10Lives 100Lives 1,000Lives
ANCOLD for
dams (1)
BC Hydro
for dams (3)
UK Nuclear
Industry (5)
Netherlands Planning
Rules (5)
Eastern Scheldt Storm
Surge Barrier (11)
Environmental Norms
Province of Groningen (11)
Netherlands Liquefied
Petroleum Gas Integral
Note (11)
Sizewell B Nuclear Station,
UK (8)
Thames Barrier (8)
Rail Tunnel,England to
France (3)
Canvey Oil Refinery,
Chemical Plant,UK
Second Report (8)
Sequoyah Nuclear Plant,
US (6)
Grand Gulf Nuclear Plant,
US (6)
Dajn Failure US (13)
1 . 10,000(L)
1 250,000(0)
1 ' 10,000(L)
No objective set
No Limit given
1 : 200,000(0)
1 : 100,000(L)
I : 10,000,000(0)
NA
NA
1 : 10,000(L)
1 : 100,000,000(0)
1 : 100,000(L)
1 : 10,000,000(0)
1 : 3,000,000(0)
1 :100(0)
1 : 1,250(L)
NA
1 : 100,000
(actual)
1 :70,000,000
(actual)
1 : 10,000
(estimated)
1 140,000(L)
1 : 12,500,000(0)
1 . 100,000(L)
No objective set
No Limit given
1 : 10,000,000(0)
1 : 10,Q00,000(L)
1 : 100,000,000(0)
NA
NA
1 : 1,000,000(L)
1 : 1,000,000,000(0)
1 : 10,000,000(L)
1 : 1,000,000,000(0)
1 :7,000,000(0)
1 : 1,000(0)
1 : 1,100(0)
1 :2,000(L)
NA
1 : 10,000,000
(actual)
1 : 1,000,000,000
(actual)
1 :25,000
(estimated)
1 : 10,000,000(L)
1 : 1,000,000,000(0)
I : 1,000,000(L)
No objective set
No Limit given
1 : 100,000,000(0)
1 100,000,000(L)
1 : 1,000,000,000(0)
NA
I : 10,000,000(O)
I : 100,000,000(L)
Too small
1 : 1,000,000,000(L)
Too small
1 : 12,500,000(0)
1 : 5,000(L)
NA
1 : 1,000,000,000
(actual)
Too small
1 : 100,000
(estimated)
Notes:
Risks lower than 1 : 100,000,000have been rounded up to the nearest whole order.
Canvey Complex,UK is an existing facility,important to the national economy. The
accepted risks are regarded as high.
-->'
GUIDELINES FOR
DESIGN OF DAMS
FOR EARTHQUAKE
ij
Dam engineers in Australia have been conscious of earthquakes for many
years,but it was the earthquakes at Tennant Creek in 1988,which were
M6.3,' M6.4and M6.7,with a total fault scarp length of 32 km which raised
the question most acutely as to whether dams in Australia could be subject
to large earthquakes,and if so,could they withstand them without resultant
loss of the facilities and lives,property,and environmental values downstream.
In recognition of the need to provide some guidance to dam engineers and
owners in Australia,the Australian National Committee on Large Dams
(ANCOLD) established a Working Group to prepare Guidelines for Design
of Dams for Earthquake.
These guidelines cover all types of dams,including tailings dams,and apply
to existing and new dams. They cover the selection of the design earthquake,
analysis and design of embankment and concrete dams,and appurtenant
structures. Whilst specific to Australian considerations,the majority of this
guideline could be applied to dam structures throughout the world.
AUSTRALIAN NATIONAL COMMITTEE ON LARGE DAMS

S-ar putea să vă placă și