ANCOLD Guidelines for Design of Dams and … · ANCOLD Guidelines for Design of Dams and...

ANCOLD Guidelines for Design of Dams and Appurtenant Structures for Earthquake – Draft March 2017 1

ANCOLD Guidelinesfor Design of Dams andAppurtenant Structuresfor EarthquakeDraft Version for publishing on websiteMarch 2017

THIS IS A DRAFT DOCUMENT AND HAS NOT BEEN RELEASED FOR USE.

IMPORTANT DISCLAIMER

To the maximum permitted by law, each of, ANCOLD Incorporated and itsMembers, the Convenor and Members of the Working Group which developedthese Guidelines, and the Independent Reviewers of these Guidelines exclude allliability to any person arising directly or indirectly from that person using thispublication or any information or material contained within it. Any person acting onanything contained in, or omitted from, these Guidelines accepts all risks andresponsibilities for losses, damages, costs and other consequences resultingdirectly or indirectly from such use and should seek appropriate professionaladvice prior to acting on anything contained in the Guidelines.

DRAFT


The committee members were;

Mr Steve O’Brien (Consultant, VIC - Convenor)Mr Ian Landon Jones (Consultant, NSW)Emeritus Professor Robin Fell (Consultant, NSW)Mr Brian Cooper (Consultant, NSW)Mr Jon Williams (Consultant, QLD)Mr Peter Allan (Regulator, QLD)Mr Graeme Bell (Consultant, NSW)Mr David Ryan (Consultant, QLD)Mr Gary Gibson (Consultant, VIC)Dr Gavan Hunter (Consultant, VIC)Mr Andrew Reynolds (Owner, NSW)Mr David Brett (Consultant, NSW)Dr Paul Somerville (Consultant, NSW/USA)Mr Bundala Kendaragama (Consultant, VIC)Mr Dan Forster (Consultant, NZ)Dr Nihal Vitharana (Consultant, NSW)

The following members formed the expert review panel:

xxxxxxx

Additional inputs and review comments were provided by:

Xxxxx

Xxxxxx

Technical Editor

xxxxx

ANCOLD Executive Liaison

xxxxxx DRAFT

Table of Contents


1 INTRODUCTION ......................................................................................................................................... 6

2 ASSESSMENT OF DESIGN SEISMIC GROUND MOTIONS AND ANALYSIS METHOD ....................... 8

2.1 Earthquakes and Their Characteristics .................................................................................................... 8

2.2 Terminology .......................................................................................................................................... 8

2.3 General Principles for Selection of Design Ground Motion and Analysis Method ................................... 8

2.4 Description of a Probabilistic Seismic Hazard Assessment ..................................................................... 9

2.5 Requirements of a Seismic Hazard Assessment ...................................................................................... 9

2.6 Selection of Design Seismic Ground Motion - Deterministic Analysis Approach .................................. 11

2.7 Selection of Design Seismic Ground Motion - Risk Based Analysis Approach ..................................... 12

2.8 Modelling Vertical Ground Motions ..................................................................................................... 12

2.9 Selection of Response Spectra and Time-History Accelerograms .......................................................... 12

2.10 Earthquake Aftershocks ....................................................................................................................... 12

2.11 Seismic Ground Motions from Earthquakes Induced by the Reservoir .................................................. 12

3 ASSESSMENT OF EMBANKMENT DAMS FOR SEISMIC GROUND MOTIONS ................................. 13

3.1 Effect of Earthquakes on Embankment Dams ....................................................................................... 13

3.2 Defensive Design Principles for Embankment Dams ............................................................................ 13

3.3 Seismic Deformation Analysis of Embankment Dams .......................................................................... 15

3.4 Assessment of the Effects of Liquefaction in Embankment Dams and Their Foundations ..................... 16

3.4.1 Overall Approach ......................................................................................................................... 16

3.4.2 Definitions and the Mechanics of Liquefaction ............................................................................. 17

3.4.3 Methods for Identifying Soils which are Susceptible to Liquefaction ............................................ 17

3.4.4 Methods for Assessing Whether Liquefaction may occur .............................................................. 17

3.4.5 Assessing the Strength of Liquefied Soils in the Embankment and Foundation.............................. 18

3.4.6 Methods for Assessing the Post-Earthquake Strength of Non-Liquefied Soils in the Embankmentand Foundation ............................................................................................................................................ 18

3.4.7 Site Investigations Requirements and Development of Geotechnical Model of the Foundation ...... 18

3.4.8 Liquefaction Analysis ................................................................................................................... 18

3.4.9 Assessment of the Likelihood and the Effects of Cracking of Embankment Dams Induced bySeismic Ground Motions ............................................................................................................................. 18

3.5 Methods for Upgrading Embankment Dams for Seismic Ground Motions ............................................ 19

3.5.1 General Approach ........................................................................................................................ 19

3.5.2 Embankment Dams not Subject to Liquefaction ............................................................................ 19

3.5.3 Embankment Dams Subject to Liquefaction.................................................................................. 19

3.6 Flood Retarding Basins ........................................................................................................................ 20

4 ASSESSMENT OF CONCRETE DAMS FOR SEISMIC GROUND MOTIONS ........................................ 21

DRAFT

Table of Contents


4.1 Effect of Earthquakes on Concrete Dams.............................................................................................. 21

4.2 Defensive Design Principles for Concrete Dams ................................................................................... 21

4.3 General Principles of Analysis for Seismic Ground Motions ................................................................. 22

4.4 Seismic Structural Analysis of Concrete Dams ..................................................................................... 22

4.4.1 Material Properties ....................................................................................................................... 22

4.4.2 Loads ........................................................................................................................................... 23

4.4.3 Methods Available and When to Use Them .................................................................................. 24

4.4.4 Analysis in the Frequency Domain ............................................................................................... 26

4.4.5 Analysis in the Time Domain ....................................................................................................... 26

4.5 Approach to Analyses and Acceptance Criteria .................................................................................... 26

4.5.1 During the Earthquake .................................................................................................................. 26

4.5.2 Post-Earthquake ........................................................................................................................... 26

4.5.3 General ........................................................................................................................................ 27

5 ASSESSMENT OF TAILINGS DAMS FOR SEISMIC GROUND MOTIONS ........................................... 28

5.1 Some General Principles ...................................................................................................................... 28

5.2 The Effects of Earthquakes on Tailings Dams ...................................................................................... 28

5.3 Defensive Design Principles for Tailings Dams .................................................................................... 28

5.4 Design Seismic Ground Motions and Analysis Method ........................................................................ 29

5.5 Seismic Deformation Analysis of Tailings Dams not Subject to Liquefaction ....................................... 30

5.6 Assessment of the Effects of Liquefaction in Tailings Dams and Their Foundations ............................. 30

5.7 Methods for Upgrading Tailings Dams for Seismic Loads .................................................................... 30

6 ASSESSMENT OF APPURTENANT STRUCTURES FOR SEISMIC GROUND MOTIONS .................... 31

6.1 The Effects of Earthquakes on Appurtenant Structures ......................................................................... 31

6.2 Intake Towers ...................................................................................................................................... 31

6.2.1 General ........................................................................................................................................ 31

6.2.2 Defensive Design Principles ......................................................................................................... 32

6.2.3 Performance Requirements ........................................................................................................... 32

6.2.4 Analysis Procedures ..................................................................................................................... 32

6.3 Spillways ............................................................................................................................................. 33

6.3.1 General ........................................................................................................................................ 33




6.4 Spillway Gates ..................................................................................................................................... 35

DRAFT

Table of Contents


6.4.1 General ........................................................................................................................................ 35




6.5 Outlet Works - Water Conduits, Gates and Valves ................................................................................ 36

6.5.1 General ........................................................................................................................................ 36




6.6 Retaining Walls ................................................................................................................................... 37

6.6.1 General ........................................................................................................................................ 37




6.7 Parapet Walls ....................................................................................................................................... 38

6.7.1 General ........................................................................................................................................ 38




6.8 Mechanical and Electrical Equipment ................................................................................................... 38

6.8.1 General ........................................................................................................................................ 38




6.9 Access Roads and Bridges .................................................................................................................... 39

6.9.1 General ........................................................................................................................................ 39



6.9.4 Analysis Procedures (Bridges) ...................................................................................................... 40

6.10 Reservoir Rim Instability ..................................................................................................................... 40

DRAFT

1.0 Introduction


1 INTRODUCTIONThe purpose of these Guidelines is to provideOwners, Regulators, Consultants, and othersinvolved in the management of existing andnew dams and appurtenant structures guidanceon the selection of seismic ground motionsresulting from earthquakes, and analysis andassessment procedures. The Guideline coverswater supply dams, tailings dams and retardingbasins (levees are not covered in thisGuideline).

The Guidelines are not:

· A textbook on earthquakes and design forseismic ground motions.

· A design code or Australian Standard.Hence the mandatory “shall” or “ensure”has been avoided in the text.

These Guidelines have been written for theguidance of experienced practitioners. There isa need for the users of the Guideline to applytheir own experience and judgement in theapplication of this Guideline. Therefore, it isstrongly recommended that personnel whoundertake, or at least closely supervise, theassessment procedures outlined within thisGuideline are highly experienced in field of theassessment being undertaken.

There have been significant advances in theunderstanding of earthquakes in Australia sinceANCOLD (1998) “Guidelines for Design ofDams for Earthquake” was published. Theseinclude:

· A recognition that there are active faults insome areas, and other neotectonic faultswhich have been active in the currentregional stress regime and may contributeto the seismic hazard.

· Refinement of earthquake source modelsfor modelling seismic hazard.

· Refinement of ground motion models, anddevelopment of models for Australianconditions.

· Refinement in the modelling of uncertaintyin the seismic hazard.

The Guidelines provide detailed descriptions ofthese and guidance on the requirements ofseismic hazard assessments.

There have also been advances in the methodsavailable for analysis and design including:

Embankment dams:· Assessment of the potential for liquefaction

of dams and their foundations, and post-earthquake liquefied strengths.

· Numerical modelling of deformationsduring seismic ground motions and post-earthquake.

Concrete Dams· Numerical modelling of the dams for

seismic ground motions includingmodelling of deformations anddisplacements.

· Improved understanding of the importanceof identifying and modelling kinematicallyviable mechanisms for sliding in thefoundation.

Appurtenant Structures· Numerical modelling of the structures for

seismic ground motions includingmodelling of deformations anddisplacements.

Importantly, there is a greater understandingand expertise within the Australian damindustry than in 1998.

Given the extensive literature relating toseismic hazard assessment, and seismicanalysis and design, these Guidelines retain alarge body of information to assist users. Theseare in the Commentary. Users should read theCommentary along with the Guideline Sectionsso they understand the details and backgroundto the Guideline.

Since the issue of ANCOLD (1998) a numberof other ANCOLD Guidelines have also beenupdated. Where appropriate, references havebeen made to these other Guidelines rather thanrepeating sections of them. If inconsistenciesoccur then this Guideline supersedes olderdocuments.

DRAFT

1.0 Introduction


The Guideline encourages the use of risk basedmethods for assessing existing and design ofnew dams. This is because of the uncertainty inthe seismic hazard, and the fact that theresponse of dams and appurtenant structures toseismic loads may change significantly withthe seismic ground motion. Therefore, a dam orappurtenant structure that is assessed using asingle design ground motion only may notnecessarily satisfy ANCOLD (2003, 2017(pending)) Risk Management Guidelines.

However, it is recognised that many Ownersprefer to use a deterministic approach toassessment of design seismic ground motionsso both approaches are covered in theGuideline.

For the purposes of defining roles andresponsibilities the following terminology hasbeen adopted throughout this Guideline:

· Owner – the entity responsible for the asset,including operation and maintenance andholds a duty of care to the community thatthe asset risks are as low as reasonablypracticable .

· Regulator – the entity responsible forregulation of dams within each state orterritory.

· Consultant – the entity responsible forprovision of assessment and design servicesreferred within the Guideline. This rolemay be performed by the Owner,educational or government institution, or bya professional consulting entity suitablyqualified to deliver the services.

AS1170 (Structural Design Actions - Part 4:Earthquake actions in Australia) specificallystates that dams are not within the scope of thestandard and therefore AS1170 should not beused for dams and appurtenant structures.

DRAFT

2.0 Assessment Of Design Seismic Ground Motions And Analysis Method


2 ASSESSMENT OF DESIGN SEISMICGROUND MOTIONS AND ANALYSISMETHOD

2.1 Earthquakes and Their CharacteristicsUsers of this Guideline should read theCommentary on earthquakes and theircharacteristics generally and Australianearthquakes in particular. A good understandingof these is important to using the Guideline.

2.2 TerminologyThe following terminology is based onANCOLD (1998), and ICOLD Bulletin 148 -Selecting Seismic Parameters for Large Dams(ICOLD, 2016a).

Operating Basis Earthquake (OBE) – the OBEis that level of ground motion at the dam site forwhich only minor damage is acceptable. Thedam, appurtenant structures and equipmentshould remain functional and damage from theoccurrence of earthquake shaking not exceedingthe OBE should be easily repairable.

Safety Evaluation Earthquake (SEE) – theSEE is the maximum level of ground motion forwhich the dam should be designed or analysed.Damage can be accepted but there should be nouncontrolled release of water from the reservoiror tailings from tailings dams.

The term Safety Evaluation Earthquakereplaces the term Maximum Design Earthquake(MDE) in ANCOLD (1998, 2012a, 2013).

The OBE and SEE are expressed in termsrelevant to the design of the dam or appurtenantstructure. It may be in terms of peak groundacceleration (PGA), peak ground velocity(PGV), or response spectra, which may beaccompanied by ground motion time-historiesor by proxy information such as momentmagnitude Mw representing ground motionduration.

Maximum Credible Earthquake (MCE) – theMCE is the largest reasonably conceivableearthquake magnitude that is consideredpossible along a recognised fault or within ageographically defined tectonic province, under

the presently known or presumed tectonicframework.

Maximum Credible Earthquake GroundMotion – The most severe ground motionaffecting a dam site due to an MCE.

Active Faults – a fault, reasonably identifiedand located, known to have produced historicalearthquakes or showing evidence of movementsin Holocene time (i.e. in the last 11,000 years ),large faults which have moved in LatestPleistocene time (i.e. between 11,000 and35,000 years ago).

Neotectonic fault - a fault that has hosteddisplacement under conditions imposed in thecurrent crustal stress regime, and hence maymove again in the future.

2.3 General Principles for Selection ofDesign Ground Motion and AnalysisMethod

There are two ways of selecting the designseismic ground motion and analysing thestructures:

(a) Deterministic Analysis – This requires theselection of an OBE and SEE, taking intoconsideration the Consequence Category ofthe dam, and the implications of failure ofappurtenant structures. The dam andappurtenant structures are analysed for theseground motions. For the SEE it is usual torequire a factor of safety to be applied toestimated stresses and deformations withinthe dam and its foundations to give a lowlikelihood of the dam failing given the SEEloading.

(b) Risk Based Analysis– which assesses theeffects on the dam and its foundations of arange of seismic loads up to and greaterthan the SEE. The design requirement is tosatisfy ANCOLD (2003, 2016) tolerablerisk criteria. The steps in the process aredetailed in Section C2.3.

Preferred method – Based on consideration ofall the factors ANCOLD prefers the risk basedapproach to the deterministic approach for Highand Extreme Consequence Category dams. For

DRAFT



Significant and Low Consequence Categorydams, deterministic approaches will usually beadequate.

The decision as to whether a risk based ordeterministic approach is adopted is for theOwner in consultation with the Regulator.

The Consequence Category of the dam isassessed in accordance with ANCOLD (2012).

2.4 Description of a Probabilistic SeismicHazard Assessment

A Probabilistic Seismic Hazard Assessment(PSHA) is an evaluation of the groundmotion level that will be exceeded at aspecified frequency or annual probability.PSHA involves the following steps:

· Identify all earthquake sources capableof generating damaging ground motionsat the site, and characterise theirlocation, geometry and sense of slip;

· Characterise the rates at whichearthquakes of various magnitudes areexpected to occur on each source;

· Characterise the distribution of source-to-site distances associated withpotential earthquakes;

· Predict the resulting ground motionlevel for each earthquake scenario usingground motion prediction equations;

· Combine uncertainties in earthquakesize, location and ground motion levelusing the total probability theoremwithin a qualified computer program.

Section C2.4 gives a detailed description of thePSHA process.

2.5 Requirements of a Seismic HazardAssessment

A PSHA is required for both deterministic andrisk based approaches.

Where there are active fault(s) in the vicinity ofthe dam site the deterministic method forExtreme Consequence Category dams alsorequires assessment of the ground motion at thedam site from the MCE on the active fault(s).

The seismic hazard assessment should becarried out by Seismologists experienced inseismic conditions in Australia, which aredifferent in some regards to what occurelsewhere.

When requesting a seismic hazard assessmentthe Owner or Consultant should provide sitespecific details to the Seismologist that mayinclude latitude and longitude for the site,natural frequency of the structures beingassessed, specific return periods to be includedin the seismic hazard assessment, minimummagnitudes, geological information for the site,Vs30 data, etc.

The seismic hazard assessment for High andExtreme Consequence Category dams shouldinclude the following features:

1. Distributed earthquake source models.Analyses should use all viable sourcemodels. The weighting between theseshould be determined by the Seismologist inconsultation with the Owner andConsultant.

2. Fault sources. Active and neotectonic faultswhich could significantly contribute to theground motion for the dam should beidentified, and be accounted for in theseismic hazard assessment.

3. Faults in the dam foundation. Informationon any known active or neotectonic faultswhich have the capability to causedisplacement in the foundation of the dam,appurtenant structures, or reservoir rimshould be provided.

4. Ground motion prediction models.Consideration of multiple alternative groundmotion prediction equations (GMPE’s) toaddress epistemic uncertainty. These mayvary for cratonic and non-cratonic areas.For non-cratonic areas at least one of thecoefficients should be Next GenerationAttenuation (NGA), along with one or moreof the models developed for Australia.Weightings applied to the models should bedetermined by the Seismologist inconsultation with the Owner andConsultant.

DRAFT



5. Shear wave velocity of the dam foundationbedrock. Shear wave velocity of thefoundation rock should be determined andincluded in the ground motion evaluation.The shear wave velocity may vary fromvalley to abutment sections so the seismichazard may be different in the valley andthe abutments.

6. Effects of soil overlying bedrock. If the damis founded on sediments or deep residualsoils and completely weathered rockoverlying the bedrock which are expected tohave significant nonlinear response, thePSHA ground motions should be specifiedat the surface of bedrock beneath thesediments or weathered rock at a depth inthe profile having a suitable shear wavevelocity that is identified jointly with thesoil response Consultant, to provide inputinto a separate study of nonlinear soilresponse that would not be part of theseismic hazard assessment. The bedrockground motions may be amplified or de-amplified by the overlying soil dependingon the soil depth and properties and thelevel of the bedrock seismic ground motion.

7. Response Spectra. The response spectraused for the hazard assessment should bedesigned to allow for the response of thedam and appurtenant structures at the dam.Conditional mean spectra may be used torepresent the design response spectra for thedam site.

8. Minimum magnitude earthquake. Thehazard assessment should be carried out forearthquake magnitudes Mw 5 and more.However, under some circumstances,smaller magnitude earthquakes may formthe lower limit. With masonry dams, slaband buttress dams, older concrete dams andstructural concrete components of dams,Mw 4 earthquake should form the lowerlimit.

9. Quantification and reporting epistemicuncertainty. The epistemic uncertainty inthe true value of the hazard due to epistemicuncertainty in the earthquake source (1) andground motion models (4) should bequantified using the fractiles of the hazard.The report should include median (50th

fractile), 85th fractile and 95th fractilepercentage spectral response spectra plots.

10. De-aggregation plots and selection of time-history motions. The report should includeplots showing de-aggregation of the hazardby earthquake magnitude and distance, andbased on these contributions provideguidance on how to select time-historymotions (accelerograms) suited for the damand the appurtenant works.

11. Reservoir-induced seismicity. For newdams, guidance on whether reservoir-induced seismicity need be considered, andif so, the resultant seismic loading.

For Significant Consequence Category dams,Owners may opt with the advice of theirSeismologist and Consultant to not requireitems (2), (3), and (11); only one GMPE (4),and use estimated shear wave velocities (6); andprovide mean estimates without modellinguncertainty (9). However if that study showsthat the seismic hazard is critical to theassessment of the risks posed by the dam themore complete assessment of the seismichazard will be required.

For Low Consequence Category dams and forpreliminary studies of higher ConsequenceCategory dams it may be appropriate to conductan initial assessment based on existing PSHAfrom nearby dams. Depending on the dam, itscharacteristics and whether its importantpotential failure modes are seismic loadsrelated; e.g. liquefaction, a decision can then bemade as to whether a site specific PSHA isrequired.

In some situations for High and ExtremeConsequence Category dams a staged approachmay be appropriate with a less detailed PSHAas described above for Significant ConsequenceCategory dams used initially, and the moredetailed assessment carried out if it isrecognised that the seismic hazard is critical.Similarly for Significant Consequence Categorydams it may be appropriate to start with aninitial assessment based on an existing PSHAfrom nearby dams, and then conduct to moredetailed site specific assessment if it isrecognised that the seismic hazard is critical.

DRAFT



It should be noted that PSHAs become dated asnew methods are developed and data basesimproved. It is unlikely that a PSHA more thanabout 5 years old will be reliable. The advice ofthe Seismologist who carried out that studyshould be sought to advise if a new study isrequired.

2.6 Selection of Design Seismic GroundMotion - Deterministic AnalysisApproach

Operating Basis Earthquake OBE

The selection of the OBE is a matter for theOwner to consider in consultation with theConsultant and other stakeholders. As shown in

Table 2.1 the OBE is often accepted as aloading which has a 10% chance of beingexceeded in a 50 year period, or an annualprobability of exceedance of 1 in 475. Howeverhigher or lower annual probability loadings maybe adopted depending on the Owners appetitefor risk, and the criticality of the dam andappurtenant structures or tailings dam.

Safety Evaluation Earthquake SEE

The recommended Safety EvaluationEarthquake ground motions are shown in Table2.1.

Table 2.1: Recommended deterministic analysis seismic design ground motions.

Dam ConsequenceCategory

Operating Basis EarthquakeOBE )1(

Safety Evaluation EarthquakeSEE )2(

ExtremeConsequenceCategory Dams

Commonly 1 in 475 AEP upto 1 in 1,000 AEP

The greater of:Ground motion from MCE on knownactive faults )3(

orProbabilistic ground motionExtreme : 1 in 10,000 AEP )4(

High A, B and CConsequenceCategory Dams

Commonly 1 in 475 AEP upto 1 in 1,000 AEP

Probabilistic ground motion )5( )6( :High A: 1 in 10,000 AEPHigh B: 1 in 5,000 AEPHigh C: 1 in 2,000 AEP

SignificantConsequenceCategory Dams

Commonly 1 in 475 AEP Probabilistic ground motion )5( :1 in 1,000 AEP

Low ConsequenceCategory Dams

Commonly 1 in 475 AEP Probabilistic ground motion )5( :1 in 1,000 AEP

Notes(1) Owner and other Stakeholders to determine in consultation with the Consultant.(2) A factor of safety should be applied to estimated stresses and deformations within the dam and itsfoundations to give a low likelihood of the dam failing given the SEE loading.(3) Active faults are as defined in Section 2.2. See C2.6 for discussion on active faults.(4) 85th fractile.(5) Median, 50th fractile.(6) The adoption of these SEE for High B and High C Consequence Category dams may prove insome particular cases not to provide an acceptable level of risk in accordance with ANCOLD RiskManagement Guidelines. It is therefore recommended that some level of risk assessment beundertaken in these cases before adopting the AEP’ stated in the table. If it cannot be demonstratedthat an acceptable level of risk would be achieved then an AEP of 1 in 10,000 should be adopted.

DRAFT


ANCOLD Guidelines for Design of Dams and Appurtenant Structures for Earthquake – Draft March2017 12

2.7 Selection of Design SeismicGround Motion - Risk BasedAnalysis Approach

As discussed in Section 2.3, for risk basedassessments the seismic ground motionsshould cover the complete range offeasible loads up to and above thedeterministic SEE values.

2.8 Modelling Vertical GroundMotions

The vertical component of ground motionsshould be included in the time-historyaccelerograms and used in dynamicanalyses for embankment and concretedams.

For pseudo-static analyses of concretedams the vertical ground motions shouldbe estimated using the method described inthe Commentary.

Vertical ground motions are often notincorporated in simplified deformationanalyses for embankment dams. If requiredthe methods described in the Commentaryshould be used.

2.9 Selection of Response Spectra andTime-History Accelerograms

A response spectrum will be required foreach of the earthquake ground motioncases (annual exceedance probabilities(AEP)) to be used in analysing the dam.The response spectra should be preparedby specialist Seismologists. The responsespectra are to be site specific, and for thedamping factor applicable to the dam. Forexample, response spectra for concretedams with linear elastic response willusually be for a 5% damping factor(viscous damping). This linear elasticanalysis ignores the hysteresis dampingwith inelastic behaviour unless a non-linear inelastic analysis is carried out,which is the most accurate to capture thetrue behaviour of the dam or structures,refer to the Commentary for additionaldetails regarding damping.

For time-history analyses, use at least 3ground motion records. Generally,accelerograms (acceleration vs time) willbe used although some analysis methodsprefer velocity/displacement vs time. Use aspecialist Seismologist to developsynthetic accelerograms (orvelocity/displacement vs time whererequired) if recorded accelerogramsappropriate to the AEP earthquake are notavailable. Use amplitude scaling orspectral matching to produceaccelerograms that will produce a responsespectrum that would approximately matchthe design response spectrum over therange of frequencies equivalent to therange of natural frequencies of interest forthe dam.

2.10 Earthquake AftershocksConsider whether the dam, or appurtenantstructures, which are critical to operationof the dam post-earthquake are likely to bedamaged sufficiently by the mainearthquake ground motion so as to be morevulnerable to further damage byearthquake aftershocks. If so, seek advicefrom the Seismologist on the likelymagnitude, focal location, and seismicground motions at the dam site, andanalyse the effects on the dam andappurtenant structures. It also needs to beconsidered how the safety of the dam willbe managed in the period through theaftershocks until repairs can be completed.

2.11 Seismic Ground Motions fromEarthquakes Induced by theReservoir

The ground motion from ReservoirTriggered Earthquakes (RTE) should beconsidered in the seismic hazardassessment for new dams, either riskbased, or deterministically. This would bea specific request to the Seismologist foradvice on whether RTE should beconsidered and if so what the loadingswould be. Refer to the Commentary fordetails.

DRAFT

3.0 Assessment Of Embankment Dams For Seismic Ground Motions


3 ASSESSMENT OF EMBANKMENTDAMS FOR SEISMIC GROUNDMOTIONS

3.1 Effect of Earthquakes on EmbankmentDams

Earthquakes impose ground motions whichresult in additional loads on embankment damsover those experienced under static conditions.The earthquake ground motion is of shortduration, cyclic and involves motion in thehorizontal and vertical directions. Earthquakescan affect embankment dams by causing any ofthe following:

(a) Settlement and longitudinal and transversecracking of the embankment, particularlynear the crest of the dam.

(b) Liquefaction or loss of shear strength due toincrease in pore pressures induced by theearthquake in the embankment and itsfoundations.

(c) Instability of the upstream and downstreamslopes of the dam if the seismic loadingleads to strength loss e.g. from liquefaction,within the embankment or foundationsufficient to result in post-earthquake factorsof safety less than 1.0.

(d) Reduction of freeboard due to settlement orinstability which may, in the worst case,result in overtopping of the dam.

(e) Differential movement between theembankment, abutments and spillwaystructures leading to transverse cracks.

(f) Transverse cracking in which internalerosion and piping may develop.

(g) Differential movements on active faultspassing through the dam foundation.

(h) Overtopping of the dam by seiches inducedin the reservoir in the event of large tectonicmovement in the reservoir basin.

(i) Overtopping of the dam by waves due toearthquake induced landslides into thereservoir from the valley sides.

(j) Damage to outlet works passing through theembankment leading to leakage andpotential piping erosion of the embankment.

The potential for such problems depend on:

· The seismicity of the area in which the damis sited and the assessed design earthquake.

· Foundation materials and topographicconditions at the dam site.

· The type and detailed construction, andnatural period of the dam.

· The water level in the dam at the time of theearthquake.

The amount of site investigation, design, andadditional construction measures (over thoseneeded for static conditions) will depend onthese factors, the consequences of failure, andwhether the dam is existing or new.

There are four main issues to consider:

· Deformations induced by the earthquake(settlement, cracking) and the effects on damfreeboard.

· The potential for liquefaction or strainsoftening of saturated or nearly saturatedsandy and silty soils and gravels with a sandand silt matrix in the foundation, andpossibly in the embankment, and how thisaffects deformations during the earthquakeand stability immediately after theearthquake.

· The zoning and design of the dam,particularly the provision of filters, toprevent or control internal erosion of thedam and the foundation, and provision ofzones with good drainage capacity (e.g. freedraining rockfill).

· For tailings dams using upstream orcentreline construction, the potential forliquefaction of loose to medium dense partlysaturated tailings where perched water tablesor nearly saturated zones may exist abovethe measured phreatic surface.

3.2 Defensive Design Principles forEmbankment Dams

The general philosophy is to apply logical,common-sense measures to the design of thedam, to take account of the cracking, settlementand displacements which may occur as the resultof an earthquake. These measures are at least asimportant (probably more so) as attempting tocalculate accurately the deformations during

DRAFT



earthquake. The most important measures whichcan be taken are:

(a) Provide ample freeboard, above normaloperating levels, to allow for settlement orslumping or fault movements which displacethe crest. For example, one might adopt anarrow spillway with large flood rise andlarge freeboard instead of a wide spillwaywith small flood rise and thus usually alower freeboard, provided the costs weresimilar.

(b) Use well designed and constructed filtersdownstream of the earthfill core (andcorrectly graded rockfill zones downstreamof the face for concrete, asphalt and othermembrane faced rockfill dams) to controlerosion if the core (or face) is cracked in theearthquake. Filters should be taken up to thedam crest level, so they will be effective inthe event of large crest settlements, whichare likely to be associated with transversecracking.

(c) Provide ample drainage zones to allow fordischarge of flow through possible cracks inthe core. For example provide that at leastpart of the downstream zone is free drainingor that extra discharge capacity is providedin the vertical and horizontal drains for anearthfill dam with such drains. In this regardsome embankment dam types are inherentlymore earthquake resistant than others. Ingeneral the following would be in order ofdecreasing resistance:

o Concrete face rockfill.o Central core earth and rockfill.o Sloping upstream core earth and

rockfill.o Earthfill with chimney and

horizontal drains.o Zoned earth-earth rockfill.o Homogeneous earthfill.o Upstream construction tailings and

other hydraulic fill.(d) Avoid, densify, drain (to be non-saturated)

or remove potentially liquefiable materialsin the foundation or in the embankment.Filters, rockfill and other granular materialsin the embankment should be wellcompacted if they are likely to become

saturated, so they will be dilatant and notliquefy.

(e) Avoid founding the dam on the strainweakening clay soils and completelyweathered rock, or rock with the potential tostrain weaken Post- earthquake stability canbe an issue if the earthquake causes evenrelatively minor movements which can takethe foundation strength from peak toresidual effective stress strength. Clay soilswith high clay size fraction, and mudrocksare potentially strain weakening.

There are a number of other less importantmeasures:

(f) Use a well-graded filter zone upstream ofthe core to act as a crack stopper, possiblyonly in the upper part of the dam. Theconcept is that, in the event that majorcracking of the core occurs in an earthquake,this filter material will wash into the cracks,and prevent the core from eroding further bythe crack stopper filtering against thedownstream rockfill or filter, therebylimiting flow and preventing enlargement ofthe crack. If well designed filters areprovided downstream, the upstream filter isof secondary importance.

(g) Flare the embankment core at abutmentcontacts, where cracking can be expected, inorder to provide longer seepage paths. Justas (or more) important is to consider thedetailing of the contact with concrete wallsand the provision of filters downstream ofthe contacts.

(h) Provide special details if there is likelihoodof movement along faults or shears in thefoundation.

(i) Site the dam on a rock foundation ratherthan soil foundation (particularly if it ispotentially liquefiable or subject to strainweakening), where the option is available.

(j) Use well graded (densely compacted)sand/gravel/fines or highly plastic clay forthe core, rather than clay of low plasticity (ifthe option is available) (Sherard, 1967). Theformer is more readily filtered by thedownstream zones, and the latter moreresistant to erosion than clays of lowerplasticity.

DRAFT



When assessing an existing dam, the use ofthese “defensive design” measures is seldompractical (except in remedial works). However,it is useful to gauge the degree of security theexisting dam presents by comparing it with thislist. Where the dam fails to meet many or mostof these features, particularly (a) to (e), this maybe a better guide to the fact that the dam maynot be very secure against earthquake than a lotof analysis.

Dams which have well designed and constructedfilters, have adequate stability against normalloads, and do not have liquefiable or strainweakening zones or foundations, will be likelyto withstand the loading from even very largeearthquakes with only minor deformations.

3.3 Seismic Deformation Analysis ofEmbankment Dams

Embankment dam engineers have recognised formany years that when considering the effects ofseismic ground motions, it is deformations, notstability which should be assessed. Mostembankments will under large seismic groundmotions, yield during part of the loading cycle,resulting in permanent deformations. Howeverthat does not mean the dam has “failed”provided the deformations are tolerable.

There are a number of methods available forestimating the deformations which may occur inembankments and their foundations during andpost seismic loading. These vary considerably intheir degree of sophistication and time to do theanalyses. In view of this a staged approach toestimating deformations is recommended asfollows:

1. Assess whether the embankment or itsfoundation are susceptible to liquefactionusing the methods described in Section 3.4.

2. For embankments and foundations notsusceptible to liquefaction or experiencingsignificant pore pressure build up or strainweakening:2.1.Use one or more of the screening or

database methods to estimate thedeformation. If the estimates ofdeformations are much less than what istolerable; e.g. crest settlements are much

less than the available freeboard beforethe earthquake for the seismic load beingconsidered and the deformation aremuch less than the width of filter,allowing for the uncertainty of themethod, accept that the likelihood orfailure by overtopping and loss of filterfunction is negligible for that seismicground motion.

2.2.If the deformations estimated by thescreening or database methods aregreater than this, use one or more of thesimplified methods to estimatedeformations. If the estimateddeformations are significantly less thanthe available freeboard and filter width,allowing for uncertainty in the methodfor the seismic load being considered,accept that the likelihood of failure byovertopping and loss of filter function isnegligible for that seismic groundmotion.

2.3.If the deformations estimated by thesimplified methods potentially threatenthe dam with excessive settlement,opening of transverse cracks that presenta significant piping risk or shearingacross filters for the maximum seismicground motion being considered, eitheruse risk based methods to assess whetherremedial works or for a new dam, designchanges to reduce deformations arerequired; or use an appropriate advancednumerical method to estimatedeformations and then use risk basedmethods to assess whether remedialworks are required.Repeat this as required for the full rangeof seismic loading being considered.

3. For embankments and foundations subject toliquefaction and / or to significant porepressure build up or strain weakening:3.1 Assess post-earthquake factors of safety

using the limit equilibrium method. Iffactors of safety for all reasonable lowerbound estimates of the liquefied strengthand allowing for pore pressure build upand strain weakening in other zones aregreater than 1.1 accept that thelikelihood of failure will be small and

DRAFT



probably tolerable subject to the qualityand extent of information available. Thismeans that for mean estimates ofliquefied strength the post-earthquakefactor of safety is likely to be > 1.3 or1.5.

3.2 If factors of safety are lower thandescribed above, estimate deformationsusing the simplified deformation analysismethod allowing for the range ofliquefied strengths and other properties.Use these to assess the likelihood offailure taking account of the veryapproximate nature of these estimates.

3.3 For High and Extreme ConsequenceCategory dams, regardless of the resultsof (3.1); and for other dams wherefactors of safety are lower than describedin (3.1), carry out static numericalanalyses of the post-earthquakecondition for a range of liquefiedstrengths and other properties. Use theseto assess the likelihood of failure takingaccount of the approximate nature ofthese estimates.

For cases where the estimated likelihood offailure are intolerable, either design remedialworks or for a new dam modify the design sothe residual likelihoods of failure are tolerable,or use advanced numerical methods to makemore refined estimates of deformations. Allowfor the range of liquefied strengths and otherproperties. Use these to assess the likelihood offailure taking account of the still approximatenature of these estimates.

Also consider interim actions if necessary toreduce the risk until remedial works areundertaken.

The extent to which an embankment is analysedfor deformations should be consistent with theConsequence Category and size of the dam. Inmany cases it may be better to design andconstruct remedial measures or for new damsmodify the design than continuing to do moreand more sophisticated analyses.

There may however be situations where themore sophisticated methods are warranted. Even

for these the deformations are at bestapproximate, and controlled by the quality ofthe data input into the analyses, and thelimitations of the methods of analysis.

The more sophisticated methods require expertinput to the analysis and selection of propertiesand should not be carried out other than byexperienced persons.

3.4 Assessment of the Effects ofLiquefaction in Embankment Damsand Their Foundations

3.4.1 Overall ApproachWhen assessing the likelihood of failure of anembankment dam by loss of freeboard due toliquefaction of the foundation soil, consider:

(i) The likelihood that the soils aresusceptible to liquefaction or cyclicsoftening.

(ii) The probability, given the earthquakeseismic ground motion, that liquefactionoccurs.

(iii)Given liquefaction occurs, whether thecrest settles sufficiently to losefreeboard, taking account of thereservoir level at the time of theearthquake.

(iv)Given freeboard is lost, whether breachoccurs.

(v) Do this for both the upstream anddownstream slopes of the embankmentand for the varying geotechnicalconditions which may apply over thelength of the embankment and itsfoundations.

(vi)If deterministic methods are being usedfor the assessment, it would be commonto assume that the reservoir is at fullsupply level.

This assessment should allow for the modellingof uncertainties in estimating the seismic hazard,the likelihood of liquefaction, the liquefiedstrength, and deformations which may result.These are by their nature very approximate. Theextent to which this is modelled will depend onthe Consequence Category of the dam, andwhether the outcomes of the assessment are

DRAFT



clear cut or marginal. These are more readilymodelled if a risk based approach is beingfollowed.

The following Sections provide guidance onmatters relating to liquefaction. The wording issomewhat prescriptive for the sake of brevity,but they are not a standard and well qualifiedand experienced practitioners may choose toadopt alternative methods if that is appropriatefor the conditions they are assessing.

3.4.2 Definitions and the Mechanics ofLiquefaction

Liquefaction – All phenomena giving rise to areduction in shearing resistance and stiffness,and development of large strains as a result ofincrease in pore pressure under cyclic ormonotonic (static) loading of contractive soils.

Initial liquefaction – is the condition wheneffective stress is momentarily zero duringcyclic loading.

Flow liquefaction – is the condition where thereis a strain weakening response in undrainedloading and the in-situ shear stresses are greaterthan the steady state undrained shear strength.

Temporary liquefaction – is the conditionwhere there is a limited strain weakeningresponse in undrained loading; at larger strainthe behaviour is strain hardening.

Cyclic liquefaction – is a form of temporaryliquefaction, where the cyclic loading causesshear stress reversal and an initial liquefaction(zero effective stress) condition developstemporarily.

Cyclic mobility – is a form of temporaryliquefaction where the shear stresses are alwaysgreater than zero.

Cyclic softening – as a term used to describe thereduction of shear strength and stiffness of claysand plastic silts under cyclic loading.

It is recommended that all those involved inliquefaction assessments read the documentsreferenced in the Commentary to familiarisethemselves with the mechanics of liquefaction.

3.4.3 Methods for Identifying Soils whichare Susceptible to Liquefaction

(a) Geology and age of the deposit.

Soils which are most susceptible to liquefactionare non-plastic or very low plasticity fillsincluding mine tailings and dredged fills, andalluvial, fluvial, marine and deltaic soils.Residual soils are generally likely to have a lowsusceptibility,

There is some indication that Pleistocene andolder soils may be more resistant to liquefactionthan Holocene soils but the evidence for this isnot sufficient to be relied upon for damengineering. This is discussed further in Section3.4.4 and in the Commentary.

The age of the deposit should be allowed fortailings dams and dredged fills as discussed inSection 3.4.4.

(b) Soil gradation, plasticity, moisture content

Use well established methods such as thosedescribed in the Commentary to assess whethera soil is potentially liquefiable. The methodsshould be used with all required data inputsincluding in-situ moisture content, Atterberglimits, and fines content.

Where Cone Penetration Test (CPT) data isavailable use also CPT based methods such asdescribed in the Commentary. CPT or CPTUbased methods should not be relied upon alone.

Use at least two and preferably more of thewell-established methods to assess thelikelihood the soils are potentially liquefiable. Asuggested approach for doing this is given in theCommentary.

3.4.4 Methods for Assessing WhetherLiquefaction may occur

Use well established methods such as thosedescribed in Section C3.4 of the Commentarysubject to the qualifications detailed in theCommentary.

If a new consensus method or updates of themethods described herein are published inrefereed journals, these methods should beadopted.

DRAFT



Take account of the fact that for some of thereferred methods the deterministic methods forestimating Cycle Resistance Ratio (CRR) are for15% probability of liquefaction, that there aresignificant uncertainties in the factors usedwithin these methods and as a result, soils whichplot within the margins of liquefiable and non-liquefiable soils should be assigned somelikelihood of being liquefiable.

These uncertainties are greater for soils belowabout 15 metres from the surface.

Take account also of the uncertainty in thegeotechnical model, and variability of theStandard Penetration Test (SPT) and CPT data,and the uncertainty of the earthquake loading.

Where the Ks and Ka values are critical to theassessment of liquefaction, and / or theliquefiable strata are below 15 metres it may benecessary to seek expert advice and use moreadvanced methods to assess liquefaction.

Do not use a mix of factors from the differentmethods because the authors rely on their ownapproach when developing the data bases uponwhich the methods are developed.

The effects of ageing of the soil deposit may betaken account of as detailed in the Commentary.This should be done for tailings and dredgedfills particularly if they are recently deposited.

3.4.5 Assessing the Strength of LiquefiedSoils in the Embankment andFoundation

Use both “critical state” and “normalisedstrength ratio” methods and apply equalweighting to each method.

Do not allow for any increase in the liquefiedresidual strength beneath a berm which is to beadded to improve post-earthquake stability. If aberm has already been constructed carry outSPT and CPT in the liquefiable soils below theberm and use these data to assess the liquefiedstrength.

Take account of the fact that there aresignificant uncertainties in these methods, and

use strengths between the best estimate andlower bound strengths.

Use methods published in refereed journals suchas those referenced in Section C3.4.5 of theCommentary.

3.4.6 Methods for Assessing the Post-Earthquake Strength of Non-Liquefied Soils in the Embankmentand Foundation

Allow for the effects of pore pressure build upduring cyclic loading in saturated non liquefiednon plastic strata, and for cyclic softening ofsaturated sensitive plastic strata.

Allow for cracking and potential weakening ofcompacted fills.

Use methods such as those described in theCommentary.

3.4.7 Site Investigations Requirements andDevelopment of Geotechnical Modelof the Foundation

The key requirements are:

(a) Review of available data relating to theembankment foundation.

(b) Develop a preliminary geotechnicalmodel and plan any additional siteinvestigations required.

(c) Carry out site investigations using themethods described in the Commentary.

(d) Refine the geotechnical model using allthe available data

Details are given in the Commentary.

3.4.8 Liquefaction AnalysisFollow the procedure detailed in theCommentary.

3.4.9 Assessment of the Likelihood and theEffects of Cracking of EmbankmentDams Induced by Seismic GroundMotions

When assessing the likelihood of failure forcracking under earthquake loading leading tointernal erosion and piping consider:

DRAFT



(i) The probability, given the earthquake,settlement occurs resulting in transversecracking.

(ii) Given it occurs, whether it will persist tobelow the reservoir level at the time ofthe earthquake or before repairs can becarried out.

(iii)Given it does, whether erosion willinitiate along the crack, whether filterswill prevent erosion continuing, whethererosion progression and a breach forms.

This is seldom a dominant failure mode because“normal” and “flood” load conditions tend todominate internal erosion and piping failuremodes. It should be noted that dams which aresusceptible to cross valley differential settlementmay not require large earthquakes to producetransverse cracking.

3.5 Methods for Upgrading EmbankmentDams for Seismic Ground Motions

3.5.1 General ApproachThe method or methods suitable for upgrading adam and its foundation for seismic groundmotions will be site specific, and whereapplicable will require a thorough understandingof the liquefaction mechanics, extent, and theconsequences for the dam. It will also depend onthe objectives, probably measured in residualrisk terms.

The remedial measures may consist of one ormore of the following:

(a) Do nothing and accept the potentialdamage and risks.

(b) Adding filters and possibly raising thecrest level of the embankment so theconsequences of deformations and theresulting risks are reduced to tolerablelevels.

(c) Modifying the liquefiable soil in thefoundation (and the embankment ifapplicable), and / or constructingstabilizing berms to limit deformationsto tolerable levels.

These works may be required only on thedownstream of the embankment, or on bothupstream and downstream. Quite commonly

only part of the length of the embankment mayrequire remedial works; e.g. if only part isfounded upon liquefiable soil. It is notuncommon to use a combination of methods;e.g. to carry out treatment on the downstreambut do nothing upstream and tolerate the risksposed by upstream deformations. The groundimprovement method used will depend on thenature and depth of the liquefiable soils, thetechniques available, and cost.

3.5.2 Embankment Dams not Subject toLiquefaction

The following are the most common remedialmeasures which may be required:

(a) Provision of filters or upgrade ofexisting filters in the upper part of thedam. This is usually done to reduce risksof internal erosion and piping totolerable levels for flood and normalloading, and the upgrade for earthquakeloads is achieved at the same time.

(b) For the very few dams where freeboardis insufficient, the embankment mayhave to be raised, usually along withraising for flood.

(c) Where there are strain weakening soilssuch as high clay content high plasticityover-consolidated clays in which strainsmay localise under earthquakedeformations, it may be necessary to adda stabilizing berm.

3.5.3 Embankment Dams Subject toLiquefaction

The following are the most common remedialmeasures which may be required:

(a) Construction of a stabilising berm, mostcommonly founded upon the potentiallyliquefiable soil after it has been treatedby ground improvement. Alternativelyremove the potentially liquefiable soilfrom the foundation of the berm.

(b) Provision of filters or upgrade ofexisting filters in the upper part of thedam to reduce the risks of internalerosion and piping to tolerable levels.

(c) Where freeboard is insufficient, theembankment may have to be raised.

DRAFT



Section C3.5.3 in the Commentary includes asummary of ground improvement methods, theirlimitations and some design details.

3.6 Flood Retarding BasinsThe assessment of retarding basins underseismic ground motions should be consideredsimilar to the methods described above forembankment dams. It is to be noted that it islikely that a retarding basin will be empty whensubjected to seismic loading conditions andtherefore the resulting risks associated with theearthquake are likely to be low. However it isimportant that the condition and function of theretarding basin post-earthquake be consideredwith regard to the potential filling of theretarding basin following an earthquake.

DRAFT

4.0 Assessment Of Concrete Dams For Seismic Ground Motions


4 ASSESSMENT OF CONCRETEDAMS FOR SEISMIC GROUNDMOTIONS

4.1 Effect of Earthquakes on ConcreteDams

It is important to recognise the differentbehaviours that gravity, arch and buttress damsexhibit during and after earthquake shaking.

With gravity dams, cracking can occur which islikely to initiate on the upstream face or at theupstream heel, and propagate along the base ofthe dam or along lift surfaces. This crackingwould then produce increaseddeformations/movement in the dam, allowincreased uplift pressures to develop post-earthquake which could be exacerbated bypartial or fully blocked drainage due to thedeformations and could damage/rupturepassive or active ground anchors due toexcessive deformations, thereby reducing post-earthquake stability. Gravity dams are mostlikely to fail post-earthquake by sliding alongthe foundation interface, through thefoundations or along a weak lift surface.Failure can also be caused by over-stressing ata sudden change in geometry in the upper partof the dam that leads to cracking severe enoughto cause toppling/sliding of the upper part ofthe dam.

Arch dams generally have more complexmodes of vibration than gravity dams and someover-stressing is more likely to occur above thebase of the dam (as well as along the base ofthe dam). There is likely to be someredistribution of stresses within the dam to thedam’s abutments. Failure is likely to occur dueto sliding along the foundation interface,through the foundation or along any weakperimetric joint. Failure could also be due togreater horizontal thrusts and shears beingtransferred to gravity abutment blocks leadingto instability in those abutment blocks. Failurecould also occur of the abutment rock masswhich supports the arch (hoop) forces from thedam.

Buttress dams may comprise a number ofstructural elements (especially slab and buttressdams). During the course of earthquake

shaking, some of these structural elements mayprogressively fail. In many circumstanceshowever, this will not necessarily lead tofailure of the dam – simply to a redistributionof loads/stresses to other parts of the dam. It isimportant to appreciate this when analysing thedam. Failure may be either structural failure ofthe various elements of the dam or globalsliding.

4.2 Defensive Design Principles forConcrete Dams

The use of defensive design principles for newconcrete dams as well as for the upgrading ofexisting concrete dams is advised.

Defensive design principles to ensure that thesecriteria are met, should address the damstructure, the interface between the dam and itsfoundations, and the dam’s foundations. Ingeneral terms, the dam or the dam’s remedialworks should be designed / assessed such that:

(a) Sliding during seismic excitation isideally prevented, although somedeformation may be inevitable underlarge seismic ground motions. Slidingresistance should be sufficient,potentially with reduced shearresistance along the sliding surface, tomeet the post-earthquake stabilitycriteria.

(b) The dam is stable against overturningpost-earthquake at all levels ofearthquake ground motion up to SEE,allowing for cracking and increaseduplift pressure in the foundationscaused by deformations during theearthquake. It is to be noted that underpost-earthquake conditions that theconcrete dam may sit in this conditionfor some time, this should be taken intoconsideration in the assessment of theoutcomes of the analysis.

(c) Any cracking of the dam or in thefoundations will not lead touncontrolled leakage.

(d) Hydraulic outlet structures are notdamaged to the extent that they eitherallow uncontrolled loss of water whichmight lead to collapse of the dam or

DRAFT



they cannot be used to lower the storageif necessary.

To achieve the above, the following defensivedesign principles are recommended:

· Proportion the cross section of a damwith no sudden changes in shape orsection stiffness.

· Keep any superstructures on the damcrest to a minimum.

· Avoid sudden changes in the abutmentprofile which would give rise to stressconcentrations, or design toaccommodate the differentialdisplacements which are likely to occurat these locations.

· Account for geological features in thefoundations which would control thedam's sliding stability. If there are suchfeatures, then suitable means ofstabilising the dam to account for themshould be employed. This is especiallytrue of arch dams, since the interactionof dam thrusts and foundationcharacteristics is determinative of thewhole dam-foundation stability.

· Provide sufficient internal andfoundation drainage, and see that drainholes are of large enough diameter thatthey will remain operative particularlyif some sliding is expected underseismic ground motions. It is vital thatthe drains keep working or that the damis stable post-earthquake with the drainsnot operative.

· Undertake adequate investigations ofthe dam’s foundations and of the liftjoints to be able to determine thestrength parameters of these features.

· If post-tensioned ground anchors arerequired for stabilising a dam thenensure that they are un-bonded (exceptfor their anchorage length) - this willallow strains due to any overloadingresulting from earthquake loading to betaken over the entire free length of theanchor rather than over a very shortlength either side of a crack in the damor foundations – strains taken over theentire free length results in much lesser

strains and consequently much lesserstresses due to earthquake loading.

· Detail internal features such as galleriesso they do not give rise to stressconcentrations which will could lead toexcessive cracking.

4.3 General Principles of Analysis forSeismic Ground Motions

See Commentary for some general designprinciples and refer ANCOLD (2013).

4.4 Seismic Structural Analysis ofConcrete Dams

4.4.1 Material Properties

4.4.1.1 ConcreteRefer to the ANCOLD (2013) for informationon selecting appropriate strength and stiffnessparameters for concrete, when analysingconcrete dams. Consideration should be giventhough to any increase in strength or stiffnessdue to the short term and high frequency natureof earthquake loading. This is discussed in theCommentary.

In assessing existing concrete dams, it will benecessary to not only investigate the strengthand stiffness of the concrete but also thestrength of lift surfaces.

4.4.1.2 FoundationsThe design of concrete dams requires athorough knowledge of the foundationconditions.

The design process should include a properlyfunded geotechnical investigation of the rockfoundations.

The investigations should provide a detailedgeological model of the dam foundations asdescribed in ANCOLD (2013). The modelshould be used as the basis for estimating thedesign strength, compressibility, design upliftpressures for the foundations and kinematicallyfeasible failure mechanisms.

The design team should include specialists withconsiderable experience in dam foundationinvestigation and design, engineering geologyand rock mechanics. The investigation team

DRAFT



should develop the geological model and workwith the dam engineering design team toestimate foundation design parameters.

The investigations should be done in stageswith the development of the dam designfeeding back into subsequent stages of thegeotechnical investigations.

A critical issue is whether there are continuous,or near continuous, unfavourably orienteddiscontinuities in the foundation, such asbedding surfaces, bedding surface shears, jointsincluding stress relief (sheet) joints, faults andshears. The geological investigations shouldassess the foundation for such features and, ifpresent, the design should make allowance forthem.

The shear strength of the foundation should bedetermined as outlined in Section 5.4 ofANCOLD (2013). The compressibility shouldbe estimated as detailed in the Commentary.

4.4.2 Loads

4.4.2.1 StaticStatic loads on concrete dams shall be inaccordance with ANCOLD (2013).

4.4.2.2 UpliftThe pre-earthquake uplift pressure distributionestimated in accordance with ANCOLD (2013)should be assumed to apply during theearthquake.

The post-earthquake uplift pressure distributionshould be determined considering the crackingdeveloped during earthquake shaking.Consideration may be given to theeffectiveness of internal/foundation drainsfollowing the earthquake. If crackingpropagates past the line of drains, the capacityof the drains should be assessed to checkadequacy for handling potentially largerdischarges due to water travelling along thecrack.

4.4.2.3 SiltConsider any potential liquefaction of silt whenestimating silt loads in a post-earthquakeanalysis.

4.4.2.4 Seismic

4.4.2.4.1 GeneralIn preliminary, simplified studies where only ahorizontal response spectrum is supplied, thevertical response spectrum may be generatedby scaling the horizontal spectral accelerationusing the multipliers given in the Commentary.For more detailed analyses us the methods toapply the ground motions described in SectionC2.9.

4.4.2.4.2 InertiaConsider the mode shape of the dam and theassociated natural frequency, to determine theacceleration of the dam at various levels withinthe height of the dam. Ensure that sufficientmodes of vibration contribute to thedetermination of inertia loads.

4.4.2.4.3 Hydrodynamic

Gravity and Buttress Dams:Pressure distributions which assume the dam isrigid and the water incompressible may be usedfor preliminary studies, especially for lowheight dams. Other methods as described in theCommentary should be applied to morecomplex dam geometry (e.g. fully or partlysloping upstream face) and for high dams.

Arch Dams:

Review the fundamental frequency of the damwith an empty storage compared to thefundamental resonant frequency of the storage.Where the ratio of the two natural frequenciesis near unity, it is necessary to consider theflexibility of the dam and the compressibility ofthe water in the storage. In this case is will benecessary to consider the storage in any finiteelement model of the dam (e.g. model thestorage (all or part) using fluid elements) inorder to determine hydrodynamic pressures onthe upstream face of the dam. See theCommentary for further information oncalculating hydrodynamic pressures for archdams.

4.4.2.5 Dynamic Earth PressuresDynamic earth pressures from silt, fill orabutments should be calculated based ongeotechnical principles as discussed in the

DRAFT



Commentary for Section 6.6.4. Pseudo-static orpseudo-dynamic procedures will be sufficientin most cases.

4.4.2.6 Load combinationsRefer to ANCOLD (2013) for applicable loadcombinations.

Consideration should be given to the behaviourof a concrete dam during an earthquake whenthe dam is empty or near empty. This is likelyto be important in gravity dams which havebeen strengthened using post tensioned anchors(which are likely to be located near the dam’supstream face). Cracking propagating from thedownstream toe of the dam could prejudice thestability of the dam during subsequent floodloading of the dam.

As noted in ANCOLD (2013), considerationalso needs to be given to the possible loss ofstabilising force from damaged post-tensionedcables by shearing due to sliding displacementor tension overload.

For arch dams especially, temperature effectscan have a significant effect on the stressdistribution within the dam. However, in mostarch dams built with vertical contraction joints,the contraction joints tended to be groutedduring winter. Therefore winter represents thetemperature stress neutral condition. Insummer when the dam deflects upstream due toconcrete expansion, the earthquake stressesinduced in the dam are likely to be less than inwinter. However, each dam should beconsidered on its merits and at least aqualitative assessment made of how the dam islikely to perform during earthquake loading incombination with different temperatureconditions.

4.4.3 Methods Available and When to UseThem

Analysis will be either in the frequency domainwhere an earthquake response spectrum isused, or in the time domain where time-histories of velocity, acceleration ordisplacement are used.

For analyses in the frequency domain, obtain aresponse spectrum for the site (in the case ofconcrete dams, attenuation or amplification of

accelerations through overburden is not likelyto be an issue as it may be for embankmentdams founded on soil) and for the relevantdamping factor and design earthquake peakground acceleration.

For analyses in the time domain, obtain aminimum of 3 sets (preferably 5 sets) of time-histories (two orthogonal horizontal time-histories and one vertical time-history in eachset).

For analyses in the time domain either directintegration or modal superposition methodsmay be used to determine stresses in the dam.The former is the more accurate but requiresconsiderably greater computer resources(computer memory and run time). The latterwill be more conservative as it uses a statisticalapproach (e.g. square root, sum of the squares)to combine the maximum stresses from thevarious modes. Consequently, it ignores thesign of the maximum stress for a particularmode.

Undertake structural seismic analysis in ahierarchical manner. That is, analysis shouldstart using simplified linear elastic methods andthen progress as required, to more sophisticatedmethods as given in Table 4-1. TheCommentary gives more details onapplicability of the methods.

The extent to which a concrete dam is analysedshould be consistent with the ConsequenceCategory and size of the dam. In some cases itmay be better to design and construct remedialmeasures than continuing to do more and moresophisticated analyses (e.g. non-linearanalysis). There may however be situationswhere the more sophisticated methods arewarranted. The more sophisticated analysismethods should not be carried out other than byexperienced persons.

DRAFT



Table 4-1 Hierarchy of Seismic Analysis Methods for Concrete Damsi. Simplified, linear elastic methods using a site specific response spectrum on a 2D model of

the dam for pseudo-static cantilever type stability analysis – generally only applicable to aconcrete gravity dam.

ii. Simplified, linear elastic methods using a site specific response spectrum on a 2D finiteelement model (FEM) of the dam and foundations – applicable to a concrete gravity dam orfor the upstream/downstream direction of a concrete buttress dam.

iii. Linear elastic methods using a site specific response spectrum on a 2D or 3D finite elementmodel (FEM) of the dam and foundations – applicable to a concrete gravity dam (2Dsatisfactory for flat sloped abutments or buttress dams considering onlyupstream/downstream direction); 3D required for steep abutments), and for arch or buttressdams (considering full 3D action).

iv. Simplified, linear elastic methods using site specific time-history records of groundacceleration or velocity on a 2D or 3D finite element model (FEM) of the dam andfoundations – applicable to a concrete gravity dam (2D satisfactory for flat sloped abutmentsor buttress dams considering only upstream/downstream direction); 3D required for steepabutments, and for arch or buttress dams (considering full 3D action).

v. Non-linear methods using site specific time-history records of ground acceleration or velocityon a 2D finite element model (FEM) of the dam and foundations so that cracking can besimulated – applicable to a concrete gravity dam or for the upstream/downstream direction ofa concrete buttress dam.

vi. Non-linear methods using site specific time-history records of ground acceleration or velocityon a 3D finite element model (FEM) of the dam and foundations so that cracking can besimulated – applicable to a concrete gravity, an arch dam or for a concrete buttress dam.

The level of complexity of the analysis may be linked to the dam’s Consequence Category as indicatedin Table 4-2.

Table 4-2 Analysis Method Appropriate for Consequence Category

Analysis Methods Consequence CategoryLow Significant High and Extreme

2D linear elastic, simplified response spectrum,cantilever analysis - gravity dam only ((i) in Table4-1)

Yes Yes Yes (1)

2D linear elastic FEA, site specific responsespectrum-gravity dam & buttress dam in u/s-d/sdirection ((ii) in Table 4-1)

Yes Yes

2D or 3D linear elastic FEA, site specific responsespectrum- all concrete dams ((iii) in Table 4-1)

Yes

2D or 3D linear elastic FEA, site specificaccelerograms- all concrete dams ((iv) in Table 4-1)

Yes

2D non-linear FEA, site specific accelerograms -gravity dam & buttress dam in u/s-d/s direction ((v)in Table 4-1)

Yes

3D non-linear FEA, site specific accelerograms – allconcrete dams ((vi) in Table 4-1)

Yes

Notes(1) Consideration needs to be given by the dam Owner in consultation with the Consultant as to whethera 2D analysis is considered sufficient for a High and Extreme Consequence Category dam. In mostcases a higher level of analysis would be required.

DRAFT



4.4.4 Analysis in the Frequency Domain See Commentary for details.

4.4.5 Analysis in the Time DomainSee Commentary for details.

4.5 Approach to Analyses and AcceptanceCriteria

4.5.1 During the Earthquake

4.5.1.1 Gravity Dams:The following are the steps in the analyses andacceptance criteria. The extent of analyses willdepend on the Consequence Category of thedam, and the purpose of the analysis; e.g.concept versus detailed design. For a detaileddesign or assessment of a High or ExtremeConsequence Category dam linear elastic time-history or non-linear -history time analysesshould be carried out. It is likely that a responsespectrum analysis will give overly conservativeresults. This will be important if the responsespectrum analysis gives an extent of crackingduring earthquake shaking that would lead toinadequate post-earthquake stability.

1. Carry out pseudo-static analysis using5% damping factor. If factors of safetyare ≥ minimum required by ANCOLD(2013) then behaviour duringearthquake loading is satisfactory.

2. If factors of safety are < required byANCOLD (2013) carry out linear elasticresponse spectrum analysis or linearelastic time-history analysis. If stressesusing 5% damping factor indicatesstresses ≤ dynamic strengths, thenbehaviour during earthquake loading issatisfactory.

3. If (2) indicates stresses > dynamicstrengths, then carry out analysis withresponse spectrum modified for 10%damping factor. If stresses > dynamicstrengths then undertake linear elastictime-history analysis to estimateDemand Capacity Ratio (DCR) andcumulative time for non-linearbehaviour. Assess according to relevantUSACE manual (e.g. USACE EM1110-2-6053).

4. If (3) indicates that the linear elasticanalyses undertaken indicate

unacceptable DCR, carry out a non-linear time-history analysis to estimatethe extent of sliding/rocking that thedam undergoes. Assess this against theestimated maximum sliding/rocking thatthe dam can undergo. This assessmentshould consider excessive permanentleakage that may occur through the damor its foundations, and the effect oninternal drains and post tensionedanchors.

4.5.1.2 Arch Dams:Carry out a 3D linear elastic finite elementanalysis (FEA) of the dam and its foundationsusing response spectra, in the first instance. Goto more sophisticated 3D linear elastic FEAusing time-histories and 3D non-linear FEAusing time-histories as might be dictated by theresults of the less sophisticated analyses and thedam’s (and foundation’s) geometry/properties.See Commentary for details.

4.5.1.3 Buttress Dams:The requirements are that for bothupstream/downstream and cross valley groundmotions:· The structure has satisfactory factors of

safety against sliding and overturning asrequired for concrete gravity dams or thatpermanent deformations are tolerable.

· The face slabs are not dislodged and satisfynormal structural reinforced concreterequirements;

· The buttresses do not fail in buckling whenassessed using normal reinforced concretedesign principles;

· Sufficient struts remain intact such that thestrength of the buttresses is notcompromised.

4.5.2 Post-Earthquake

4.5.2.1 Gravity Dams:Regardless of the method of analysis used toanalyse the dam during earthquake shaking,carry out a post-earthquake stability analysis.This analysis should consider damage to thedam occurring during the ground motion (e.g.disruption of drains, failure of post tensionedanchors) and changes in uplift pressuredistribution due to cracking caused during

DRAFT



earthquake ground motions. Reduced orresidual shear strength parameters due to strainweakening should be used where appropriatewithin the foundations and at lift joints. Theminimum post-earthquake factor of safety forsliding should satisfy ANCOLD (2013) Table6.1 and 6.2 criteria.

4.5.2.2 Arch DamsConsider extent of cracking at base of damcaused by earthquake shaking in determiningredistribution of stresses in the dam and indetermining post-earthquake sliding resistance.

Consider extent of cracking in both the arch andany gravity abutment sections in consideringthe change of loading on the abutment sections.

Use residual shear strength parameters withinthe dam and the foundation where appropriateaccording to the requirements of ANCOLD(2013).

4.5.2.3 Buttress DamsCarry out post-earthquake analysis followingthe principles for gravity dams as discussed inSection 4.5.2.1.

The minimum factor of safety for sliding andoverturning should satisfy the requirements ofANCOLD (2013) for gravity dams.

4.5.3 GeneralSee Commentary. DRAFT

5.0 Assessment Of Tailings Dams For Seismic Ground Motions


5 ASSESSMENT OF TAILINGS DAMSFOR SEISMIC GROUND MOTIONS

5.1 Some General PrinciplesThis Section of the Guideline supplements theANCOLD Guideline on Tailings Dams(ANCOLD, 2012a). Those involved withtailings dams should refer to that Guideline.

Where there are conflicting data in the twoGuidelines this Guideline representsANCOLD’s position.

Where there are additional State regulations fortailings dams these regulations must becomplied with.

Tailings dams and their potential impacts mustbe effectively managed throughout their life,from design through operation to closure andpost closure. The post closure period may bevery long; e.g. ANCOLD (2012a) suggest 1000years.

The flow diagram below illustrates a cyclicalreturn step during operation indicating possiblemodifications to ensure that the dam achieves itsultimate goal - sustainable closure with minimalpost closure risks and associated costs.

These considerations can impact on the seismicdesign requirements of tailings dams asdescribed below.

D e s ig n C o n s tru c t O p e ra te& M o n ito r

M a in ta in &M o d ify

D e co m m iss io n& C lo se A fte rca re

Tailings dams may be designed and constructedby a number of methods (ANCOLD, 2012a):

· Downstream, usually incorporating astarter dam.

· Centreline, where part of theembankment is built over the tailingsbeach.

· Upstream, where most of theembankment is built over the tailingsbeach.

A combination of these methods may be used;e.g. downstream construction for the majority ofthe dam, with small raises using upstream orcentreline construction.

If downstream construction methods are usedthere are few if any differences between suchdams and conventional water retaining damsfrom the seismic design viewpoint.

However, if centreline or upstream constructionis used there are a number of the methodsdescribed in this Guideline for conventionaldams which are not applicable. These arediscussed below.

There is also a greater susceptibility toliquefaction and strain weakening leading tolarge deformations and potentially slopeinstability which needs to be taken into account.

5.2 The Effects of Earthquakes on TailingsDams

For downstream construction these are as forconventional embankment dams as detailed inSection 3.1. If the water storage is kept remotefrom the dam by good water managementpractices the implications of cracking of the dampotentially leading to internal erosion and pipingmay be less than for a conventional dam.

For tailings dams constructed by centreline andupstream construction methods there will be agreater emphasis on liquefaction and strainweakening, and potential for large deformations,internal erosion and piping in cracks resultingfrom deformations and slope instability.

5.3 Defensive Design Principles for TailingsDams

These are as for conventional embankmentdams as detailed in Section 3.2. For tailingsdams constructed by centreline and upstream

DRAFT



construction methods there will be a greateremphasis on the potential for liquefaction andstrain weakening and slope instability.

Particular care and conservatism is required ifupstream construction methods are proposed forHigh and Extreme Consequence Categorytailings dams if they are constructed of tailingsmaterials which are potentially liquefiable. Thisis due to the difficulty of maintaining goodconstruction practices throughout the life of thedam, and the uncertainty of predicting theliquefied strength of the tailings and post-earthquake deformations. If upstreamconstruction methods are used it should only beby involving Consultants expert in thesematters, using conservative assumptions andrequiring confirmation of assumed parametersduring construction.

Upstream construction may reasonably be usedfor small raises of tailings dams constructed bydownstream methods where it can bedemonstrated with a high degree of confidencethat the potential effects of liquefaction and / orstrain weakening of the tailings can beaccommodated with a large margin of safety.

Where upstream construction is used, this mustrecognise that such dams are not suitable forwater storage other than the minimum amountrequired for achieving decant water quality.Water storage limitations should be set, andstringently adhered to during operation and forthe life of the dam, taking into account the lackof filters for this type of dam and the potentialdeformation, which should ensure no contact ofstored water to the perimeter embankment.

Where centreline undertaken, specific defensivedesign principles include:

(a) Use conservative design methodsincluding estimates of the extent of liquefactionand residual strength parameters for stabilityanalysis;

(b) Include buttresses constructed of highstrength, non-liquefiable materials as part of theouter shell design;

(c) Avoid water storage on the dam surface;

(d) Where feasible use “integrated wastemanagement” principals to constructconservative embankments from mine wasterock;

(e) Provide internal drainage systems tolower phreatic surface and reduce the level ofsaturation of tailings;

(f) Maximise the density and reducesaturation by well-planned and managed tailingsdischarge aimed at minimising thickness offresh tailings discharge layers and maximisingevaporation and drying shrinkage;

(g) Maximise the density and reducesaturation by mechanical working of the tailingsusing purpose built equipment such asAmphirolls (Archimedes Screw Tractors):– notethat this method has been able to successfullycompact tailings to non-liquefiable conditionsequivalent to “dry-stacking”;

(h) Use dry-stacking methodology includingfilter-press technology to dewater tailings andpotentially allow placement and compaction oftailings to non-liquefiable density or saturationlevels.

5.4 Design Seismic Ground Motions andAnalysis Method

Design seismic ground motions for tailingsdams during operation are as for conventionaldams and are given in Table 3.2. TheConsequence Category should be determined asdetailed in ANCOLD (2012), Table 1. It shouldbe noted that for tailings dams the impact on thenatural environment may be a controlling factor.

The design seismic loads for tailings dams postclosure should be the SEE. The SEE should bere-evaluated at the time of closure to allow forany development in understanding of seismichazard and for any changes in the consequencesof failure, and that allowance should be madefor potential increases in the PAR and henceconsequences post closure.

DRAFT



5.5 Seismic Deformation Analysis ofTailings Dams not Subject toLiquefaction

These are as for conventional embankmentdams as detailed in Section 3.3.

For tailings dams constructed by centreline andupstream construction methods there will be agreater emphasis on liquefaction and strainweakening. None of the “screening”, “database”or “simplified” methods are applicable totailings dams where liquefaction or othersignificant strain weakening may occur.Deformations may be estimated by the methodsdescribed for static and advanced numericaldeformation analyses as described in SectionsC3.3.4.3 and C3.3.4.4 of the Commentary.

5.6 Assessment of the Effects ofLiquefaction in Tailings Dams andTheir Foundations


Allow for the effects of age of the tailingsdeposit as described in Section 3.4.3 and SectionC3.4.3 in the Commentary.

5.7 Methods for Upgrading Tailings Damsfor Seismic Loads


DRAFT

6.0 Assessment Of Appurtenant Structures For Seismic Ground Motions


6 ASSESSMENT OF APPURTENANTSTRUCTURES FOR SEISMICGROUND MOTIONS

6.1 The Effects of Earthquakes onAppurtenant Structures

A number of subsidiary structures associatedwith a dam are essential for the dams operation.For these structures it is important that theirfunctional and structural integrity is retained inthe event of a notable earthquake. This isparticularly the case where the appurtenantstructure may be required to release water fromthe reservoir in a controlled manner to lower thestorage following an earthquake. It is thereforenecessary that not only the intake/outletstructures and their gates and valves remainserviceable but also that there is proper accessto these structures. Bridges and roads may needto remain in a sound state after an earthquakedepending upon their importance.

The most important factors in considering theearthquake resistant design of an appurtenantstructure are:

· Whether or not failure of such a structurecould lead to loss of control of the reservoirfollowing an earthquake,

· Where the structure is required for post-earthquake operation to lower or maintainthe reservoir level so that the structuremaintains its functionality or allow repairs.

Generally, appurtenant structures should besuch that:

· They maintain their normal operatingcondition after an OBE

· They are not damaged to an extent wherethey could allow sudden or uncontrolledloss of water from the storage for groundmotions up to the SEE.

· Following ground motions up to the SEE,appurtenant structures should operable toan extent that the dam is able to pass floodswhile repairs are carried out.

The level of assessment and design groundmotions of critical appurtenant structures issubject to the risk assessment and operational

requirements of the component and should bedetermined by the Owner and Consultant. Itmay be required for appurtenant structures thatare critical to the operation and safety of thedam following an earthquake, e.g. low leveloutlet works that have the ability to draw downthe reservoir, that these structures maintainfunction post-earthquake loading conditionswith a high degree of confidence. To achievethis may require them to be designed for theSEE.

When designing and assessing appurtenantstructures it is important to note that these arehydraulic structures and the assessment needs tobe conducted accordingly. Australian Standardssuch as AS1170 (Structural Design Actions) aretypically not appropriate for these structures.

When assessing appurtenant structuresconsideration also needs to be given to otherexternal factors that could influence theoperation of the structure such as rock fallimpacts, restricted access for operation, loss ofpower, etc.

This following sections provide the informationfor the various types of appurtenant structuresincluding:

· Defensive design principles;· Performance criteria for the Operating Basis

Earthquake (OBE) and the SafetyEvaluation Earthquake (SEE); and

· Recommended analysis procedures.

6.2 Intake Towers

6.2.1 GeneralDam intake/outlet facilities typically compriseof the intake/outlet structure, intake/outlettunnel, exit structure, access bridge (at times)and operating equipment including pipework,valves, generators, electrical control panels, etc.

Many intake/outlet towers, termed intaketowers from herein, are either free standing onan enlarged base or foundation mat placed onthe reservoir bottom, or are deeply foundedthrough bedrock or soil, away from the dam.Others are embedded within earthfill dams, or

DRAFT



structurally tied to the upstream face ofconcrete dams. Towers built withinembankment dams may interact dynamicallywith the surrounding materials. Reservoir watersurrounds most towers, sometimes up to asignificant height. Hence they are subjected tohydrodynamic interaction effects. Some containinside water, which also affects seismicresponse.

6.2.2 Defensive Design PrinciplesThe following defensive design principles aresuggested for new structures and upgrades toexisting structures:· Site the intake tower on a foundation where

there are no geological features that couldlead to uneven displacement ordeformations during the earthquake event.

· Where possible, the intake tower should besocketed into rock to assist with preventingthe tower from sliding.

· Where possible, install grouted dowels inthe rock foundation as a redundancy againstuplift and rocking.

· Avoid so far as is practicable suddenchanges in profile that could give rise tostress concentrations.

· Provide horizontal reinforcement designedto prevent vertical reinforcement frombuckling and to confine concrete when it isin compression. The horizontalreinforcement should be placed on theoutside of the vertical reinforcement.

· In the design of towers, practices should beadopted that ensure ductile behaviour whilesuppressing brittle failure modes including:meet minimum reinforcing steelrequirements such that the nominal strengthmoment is equal to 1.2 times that crackingmoment; provide adequate confinement atsplice locations and plastic hinge regions:provide anti-bucking hoops/ties in plasticregions; provide adequate splice and anchorlengths; avoid locating splices in inelasticregions; and provide direct and continuousloads paths.

· For access bridges to intake towers provideappropriate support mechanisms designed toprevent damage to both the bridge and the

intake tower due to deflections occurringduring the seismic event.

6.2.3 Performance RequirementsOBE: Static and dynamic loads to induce

maximum concrete and steelreinforcement stresses within the elasticregion (i.e. limited amount ofreinforcement yielding) for strengthdesign purposes and the tower and itsbase is to remain stable.

SEE: Significant amount of reinforcement canyield and damage to the tower mayoccur. The tower and its base and shouldnot be damaged to an extent that it leadsto an uncontrolled release of water fromthe reservoir. Intake towers required forthe emergency release of waterfollowing an earthquake event shouldremain functional.

6.2.4 Analysis ProceduresThe analysis methods for intake towers aredescribed here in general terms only. It isrecommended that the reader refer to ICOLD(2002), USACE (2003b), USACE (2007) andthe Commentary for additional detailedinformation on the analysis procedures. Themore sophisticated methods require expert inputto the analysis and should not be carried outother than by experienced persons.

The general issues and potential modes offailure that need to be examined in the seismicresponse of intake towers include the following:

· Flexural displacement demands exceedingthe flexural displacement capacity;

· Shear demands exceeding shear (diagonaltension) capacity;

· Shear demands exceeding the sliding shearcapacity; and

· Moment demands exceeding theoverturning capacity (rocking).

There are a number of methods available for theassessment of intake towers, these are providedbelow ranging in increasing level ofcomplexity. It is to be noted that care is to betaken when selecting models with increasing

DRAFT



levels of complexity and is to be based on thejudgement of experienced personnel. This isparticularly the case with non-linear analysiswhere the assessment can often be complex and

time consuming. In this instance it needs to beconsidered whether remedial measures wouldbe more beneficial than further analysis.

Table 6-1: Hierarchy of Seismic Analysis Methods for Intake TowersStrength Assessment

i. Response Spectrum: The response spectrum analysis is adequate for towers whose responsesto earthquakes are within the linear elastic range. If it is determined that the tower remainswithin its elastic range under the OBE and meets the ductility requirements for the SEE, thenthe tower is considered acceptable.

ii. Linear Time-History Analysis: Applicable if tower exceeds requirements of the ResponseSpectrum Analysis. If the demand capacity ratios, cumulative duration of bending momentexcursions and extent of reinforcement yielding meet those described in the Commentary thenthe tower is considered acceptable.

iii. Non Linear Analysis: The nonlinear analysis of towers can be complicated and timeconsuming. In this instance it needs to be considered whether remedial measures would bemore beneficial than further analysis.

Stability Assessment - Slidingi. Response Spectrum: The OBE is considered an unusual condition and a factor of safety of 1.3

is required. The SEE is considered an extreme condition and a factor of safety of 1.1 isrequired.

ii. Linear Time-History Analysis: The tower should meet the requirements described in Section6.2.3 for the OBE and SEE. Within this analysis, the stability is maintained and sliding doesnot occur if the factor of safety is greater than 1.

iii. Non Linear Analysis: In the nonlinear time-history analysis an assessment can be made of thetotal permanent sliding displacement of the tower. It then needs to be assessed whether thepermanent displacements meet the criteria for the SEE described in Section 6.2.3.

Stability Assessment – Rotationali. Tipping Potential Evaluation: The tower may start to tip and start rocking during an

earthquake. The assessment of the tipping potential for the tower, either a rigid or flexibletower, is provided in USACE (2007). If it is determined that no tipping occurs, ie. the structuredoes not break contact with the ground, then the tower is considered stable. If tipping occursthen the rocking block analysis should be conducted.

ii. Rocking Block Analysis: USACE (2007) and the Commentary provide the procedures forassessment of the tower rocking. The tower should meet the requirements described in Section6.2.3 for the OBE and SEE.

Note. For sliding in the foundation the principles of selection of the strength of the foundationand factors of safety should be as detailed in ANCOLD (2013).

6.3 Spillways

6.3.1 GeneralTypically, spillways are constructed of mass orreinforced concrete. Seismic loads often controlthe design of such structures. Spillwaycomponent structures can be grouped into threegeneral classes, including inlet structures (inletand/or crest structures including gates), chutes

(conveyance structures such as floor slab withwalls connecting the inlet structures to theterminal structure), and the terminal structure(hydraulic-jump, stilling basin, flip bucket,impact structure, etc.). Each of these spillwaycomponents is covered in the following section.

DRAFT



6.3.2 Defensive Design PrinciplesFor defensive design principles for mass andstructurally reinforced components of spillways,refer to Section 4.2.

6.3.3 Performance RequirementsOBE: Spillways should maintain their normal

operating condition after an OBE.

SEE: Spillways, including gates that retainpermanent storage at the time of an SEEshould not fail to an extent where wateris released in an uncontrolled manner.Following ground motions up to the SEE

spillways should be operable to an extentthat the dam is able to pass floods whilerepairs are carried out.

6.3.4 Analysis Procedures

General

Any large mass spillway structures that retainthe reservoir should be designed in accordancewith Section 4 – Concrete Dams. For all otherspillway structures, Table 6-2, taken fromICOLD (2002) with some additions, providessome guidance on the recommended analysismethods for the components of spillwaystructure.

Table 6-2: Analysis method of spillway component structures

SpillwayStructures

Components Usual Approaches Recommended models

Inlet and creststructures

Morning glory dropinlet structures

Response spectrumLinear time-history

3D

Overflow structures:Straight ogee crests,Labyrinth, Fuse gates

Pseudo-staticUsing elasticfoundation

2D plane strain of plainstress or3D

Siphon structures Response spectrum 2D or 3DFuse plug structures:zoned embankment

Deformation analysisNewmark method orLiquefactionpotential(refer Section 3)

2D

Chutes Conveyancestructures:Floor slab andconnecting walls


2D plane-strain or plane-stress

Terminal Structures Hydraulic jumpStilling basin

Flip bucketImpact structures


2DDRAFT



Additional inertia loading from operationalequipment, piers, etc. needs to be consideredin the assessment of spillway structures. Forspillway piers, the effect of the bridges andoperating equipment needs to be consideredin the assessment of the piers.

6.4 Spillway Gates

6.4.1 GeneralSpillways often incorporate gate systems thatretain permanent storage and / or fulfil damsafety functions. These gates can, at times, beused to allow the controlled release of water ina potential dam safety emergency, includingpost earthquake.

Spillway gate types typically include crestmounted radial gates, orifice radial gates,vertical lift wheel gates and flap gates.

6.4.2 Defensive Design PrinciplesThe following defensive design measures arerecommended:· Provide sufficient flexibility and details in

the gate system to accommodate expecteddifferential movements during anearthquake event.

· Evaluate the reliability of the operatingsystem, including appropriate access, foroperation post earthquake, refer to Sections6.8 and 6.9 for additional details.

· The gate drive systems should not fail ordistort; such as drive shafts and hydrauliccylinders.

6.4.3 Performance RequirementsOBE: Spillways gates and their operating

gear should maintain their normaloperating capability after an OBE.

SEE: Spillways gates that retain permanentstorage at the time of an SEE shouldnot fail to an extent where water isreleased in an uncontrolled manner.The SEE should not cause gates todistort or gate piers to permanentlydisplace to an extent where the gatesbecome jammed. The hoist system or abackup arrangement should remainfunctional.

6.4.4 Analysis ProceduresIf spillway gates are located on the top of aconcrete dam, the spillway gates will need tobe analysed as an integral part of the dam.Amplification of the ground accelerationscould be significant (in some cases there maybe a two – to fivefold increase in magnitude).

Seismic loads transmitted from spillway gatesto trunnion pins and trunnion blocks shouldalso be accounted for when designing orassessing a gated crest structure. These loadscan be significant due to their concentratednature.

Hydrodynamic loads are typically assessed asdescribed in Section 4.4.2.4.3, and applied asadded masses to a model. When using thisapproach, the following items need to beconsidered as they can impact on thehydrodynamic loads applied to the gates:

· The depth of the water against the gate;· The depth of water against the full spillway

structure;· The fundamental frequency of the gate

compared to the fundamental resonantfrequency of the storage; and

· The position of the gate relative to theupstream face of the spillway.

With the development of numerical modelling,the use of fluid or acoustic elements within athree dimensional model is now possible andcan be adopted, particularly for structures witha complex geometry. For further details onseismic induced loads on spillway gates referto USBR (2011).

The complete gate system should beconsidered, from incoming power supply,electrical components, backup supplies, hoistdesign and gate design, emergency bulkheadsand cranes, operator access.

The effect of dynamic amplification on criticalcomponents (eg hydraulic cylinders) may needto be considered.

DRAFT



6.5 Outlet Works - Water Conduits, Gatesand Valves

6.5.1 GeneralWater conduits such as pipelines, penstocks,tunnels and low-level outlets may be requiredfor reliable, controlled, rapid emptying of thereservoir. Water conduit design should be suchthat it does not lead to failure or compromisethe functioning of the dam and its foundation.In addition, in the case of water supplyreservoirs in populated areas, the safety andoperability of the outlet pipelines, gates andvalves become significant factors affecting themaintenance of drinking water supplies as wellas water to fight-fires and assisting with post-earthquake recovery functions.

6.5.2 Defensive Design PrinciplesThe following defensive design measures arerecommended:· Provide sufficient flexibility in the outlet

conduit system to accommodate expecteddifferential movements during anearthquake event.

· Weak zones, faults and active faultcrossings should be avoided wherepossible for outlet conduits. If such areascannot be avoided, then design detailsshould be utilised which can accommodatedisplacement and differential movements.

· Tunnel plugs can be incorporated as part ofthe conduit design to prevent uncontrolledreleases of water should damage or conduitfailure occur.

· Where possible, pipelines or penstocks,including supports and anchor blocks,should be founded on suitable rock, soil, orstabilised soil which is capable ofminimising differential movement andsettlement due to seismic ground motion.

· Pipelines or penstocks should not befounded on low-density, non-plastic soilsthat are subject to liquefaction or highlevels of strain that could cause damageeven during a moderate earthquake.

· Consider the differential loadings that canoccur on conduits which pass throughdifferent zones in an embankment.

· Consider rock falls that could occur as aresult of an earthquake and potentiallydamage the pipeline or penstock.

· For pressurised systems considerationneeds to be given to the hydrodynamicforces that can be generated during anearthquake, refer to the Commentary.

· Ensure the gate actuation systems canwithstand vibration. Hydraulic cylindersare vulnerable to vibration. Rope supportedgates may be susceptible to vibration undercertain conditions and position.

6.5.3 Performance Requirements

Outlet Conduits

OBE: Static and dynamic loads to induceconcrete and steel reinforcementstresses which satisfy AS3600 ConcreteStructures.

SEE: Conduit not to collapse or rupture.Collapse could lead to an underminingand subsequent failure of theembankment. Rupture could causepiping or destabilise an embankment bya marked increase in pore pressure.

Pipelines and Penstocks

OBE: Static and dynamic loads to inducesteel stresses which satisfy AS4100Steel Structures.

SEE: Pipelines and penstocks required foremergency releases or post-earthquakerecovery not to collapse or rupture.

Gates and Valves

OBE: All gates and valves to maintain theirnormal operating capabilities.

SEE: Emergency closure and regulating gatesand valves (especially low level releasevalves) to maintain operating capability– the storage may need to be quicklylowered if parts of the dam aredamaged and need remedial works orrelief of hydrostatic loads.

DRAFT



6.5.4 Analysis ProceduresThe complete outlet system should beconsidered, including power supply, electricalcomponents, backup supplies, valve and gatedesign and operation, emergency bulkheadsand cranes, operator access.

Outlet Conduits

Use methods detailed in the Commentary. A2D analysis is usually sufficient for the seismicassessment of an outlet conduit.

Pipelines and Penstocks

Consideration needs to be given to thepotential of rocking of pipelines and penstockswithin outlet conduits and the potentialamplification effects that can occur.

6.6 Retaining Walls

6.6.1 GeneralRetaining walls are often critical componentsto a spillway or dam structure. They includegravity, semi gravity and non-gravity wallswith seismic loads often controlling the designof such structures. During an earthquake, aretaining wall can be subjected to dynamic soilpressures caused by motions of the ground andthe wall that need to be accounted for in thedesign and assessment of the wall. Forretaining walls that retain embankments,consideration needs to be given to theperformance of the backfill material as this hasthe potential to lead to piping or largedeformations, particular consideration needs tobe given to backfill materials that arepotentially liquefiable.

6.6.2 Defensive Design PrinciplesThe following defensive design measures arerecommended:· Avoid sudden changes in the retaining wall

profile which would give rise to stressconcentrations;

· Site where there are no geological featuresin the foundations which would decreasethe retaining walls sliding stability - ifthere are, then suitable means of stabilisingthese features or accounting for them in thedesign should be employed.

· Provide sufficient drainage providedbehind the retaining wall. It is vital that thedrains keep working or that the retainingwall is stable post-earthquake with thedrains not operative;

· If post-tensioned ground anchors arerequired for stabilising a retaining walldesign so that they are un-bonded (exceptfor their anchorage length), refer to Section4 for additional details on anchors;

· For new structures and upgrades designand specify the backfill to the retainingwall so that the backfill materials are notliquefiable as this can lead to significantdeformations of the backfill material,increasing the potential for opening a gapbetween the wall and backfill resulting inpiping and increasing the loads on theretaining wall.

· For retaining walls that retainembankments, the wall be sloped on theembankment side to maintain positivecontact with the embankment followingdeformations that may occur due to theearthquake.

6.6.3 Performance RequirementsOBE: Retaining walls should maintain their

normal operating condition after anOBE. Static and dynamic loads toinduce maximum concrete and steelreinforcement stresses which arewithin the elastic region.

SEE: Walls which retain part of anembankment dam shall not collapse ordeform to an extent that could lead toembankment failure due to piping orbreach due to significant deformations.

6.6.4 Analysis ProceduresGravity retaining wall structures which if theyfail could result in breach of the dam should bedesigned in accordance with Section 4 –Concrete Dams. Cantilever retaining wallswhich if they fail could result in breach of thedam should be designed to the same principlesbut using design methods appropriate for suchwalls as described in the Commentary.

DRAFT



The dynamic loads applied from the backfillmaterial need to be included. Depending on themagnitude of wall movements the backfillmaterial is said to be in yielding, non-yielding,or intermediate state. Further information onthe assessment of the backfill loads is providedin USACE (2007) and in the Commentary.

6.7 Parapet Walls

6.7.1 GeneralParapet walls are often used to increase theheight of embankment dams and are generallyconstructed using either cast-in-situ or precastconcrete components. These walls are abovethe full supply level of the dam and do notretain permanent storage.

6.7.2 Defensive Design Principles

· Provide sufficient flexibility in parapetwall joint details to allow for deformationsdue to the earthquake event.

· Consider the potential for internal erosionand piping underneath the walls and ifrequired install suitable filter zones.

6.7.3 Performance RequirementsOBE: Parapet walls should maintain their

normal operating condition after anOBE.

SEE: Parapet walls can be expected todeform and be damaged in a SEE. TheOwner and Consultant should considerthe potential for damage, how this maybe minimised and details provided sothat the dam will satisfy tolerable riskcriteria resulting from floods followingthe earthquake and before the parapetwall can be repaired.

6.7.4 Analysis ProceduresRefer to the Section 6.6: Retaining Walls foranalysis procedures. Consideration of theperformance and deformation of the parapetwall needs to be given as part of theearthquake assessment of the underlying damstructure. Consider the amplification effectsthat can occur at the top of the dam.

6.8 Mechanical and Electrical Equipment

6.8.1 GeneralFor gates and outlet works to remainfunctional following an earthquake event anymechanical and electrical equipment requiredfor the operation of the gates and valves alsoneed to remain operational. These items inexisting dams are often not designed forseismic ground motions and can sometimes befragile and may be subjected to accelerationsmany times greater than the peak groundacceleration due to amplification of the groundmotion. Depending on its resonant frequency,equipment may be susceptible to dynamicamplification resulting in high stresses. Some‘off the shelf’ equipment may not be suitablefor seismic loading and vibration.

6.8.2 Defensive Design PrinciplesThe following defensive design measures arerecommended:· Anchor mechanical and electrical

components to maintain their stabilityduring earthquake events and theiroperability following the event. This istypically relatively inexpensive toincorporate into new and existingstructures.

· The reliability of the power supply shouldbe evaluated. Consideration should begiven to alternative power supplies forcritical components. Consider thepossibility of landslides and structuralmovement on cable runs and hydraulicpipework. Control buildings should bedesigned to withstand earthquake loading.Consider fuel tanks, switchboards,transformers, cable trays and support ofstarter batteries.

· Critical crane equipment should beappropriately anchored to avoidmisalignment. Especially applicable tocranes supporting emergency bulkheads orrequired for emergency remedial works.

· Hoist systems (shafts, hydraulic cylindersand pipework etc.) should be appropriatelydesigned.

DRAFT



6.8.3 Performance RequirementsAll mechanical and electrical equipment andcomponents should meet the designrequirements of the equipment that theyoperate. All equipment should be appropriatelyanchored to prevent sliding, overturning orimpact loading. All power facilities, includingelectrical conduits and buildings that mighthouse engine generator sets, should bedesigned to functionally survive the designseismic ground motion.

6.8.4 Analysis ProceduresA pseudo-static method of analysis is usuallysufficient for the assessment of mechanical andelectrical components. This needs to take intoaccount that the accelerations may be greaterthan the bedrock peak ground acceleration, andthe potential for dynamic amplification of thevibration in some equipment.

6.9 Access Roads and Bridges

6.9.1 GeneralIn order to allow access to the dam forinspection or operation of outlet worksfollowing an earthquake event, the bridges androads may need to remain in a sound statedepending upon the importance of the dam andits appurtenant structures.

There is a limit to what a Dam Owner hascontrol over in this instance as typically roadsand bridges to access dams are not owned ormaintained by the Dam Owner, this needs tobe considered in assessing the requirements foroperation of appurtenant structures followingan earthquake.

6.9.2 Defensive Design PrinciplesThe following defensive design measures arerecommended:· For bridges over spillways that are

combined with or attached to thesupporting structure for operatingmechanisms that open and close gates anumber of defensive design principles canbe incorporate to prevent the followingfrom occurring:o Distortion of the piers that could

cause binding of gates or removal of

support for the bridge from itsbearings.

o Failure of the girders, supports,handrails, etc. that could interferewith the gate operating machinery orinterfere with opening of the gates.

o Failure of the beam seats that couldinterfere with the gate operation

o Differential movement that couldcause hydraulic lines or other powersupply functions to fail.

o Distortion that could interfere withcommunication/control functions.

· For access bridges to intake towers theappropriate support mechanisms should bechosen to prevent or limit damage to boththe bridge and the intake tower due todeflections occurring during the seismicevent. This includes provision thatsufficient longitudinal and lateralmovement is available in the bridgesupports.

· For critical structures an assessment mayneed to be conducted on the potential forrock falls or landslides and the impact theymay have on either the structure itself or onaccess to the structure. It may beimpractical to prevent rock falls, even inevents more frequent than the OBE, androck fall protection structures may berequired.

6.9.3 Performance RequirementsOBE: Access roads and bridges to the dam

and its appurtenant works that areowned and maintained by the DamOwner should so far as is practicableremain passable after an OBE event.Therefore, likely rock fall or landslipareas along the access roads may needto be checked for potential instability.The impact of this on access roads andbridges that are not owned ormaintained by the dam ownerbecoming impassable needs to beconsidered in assessing the requiredoperation of appurtenant structuresfollowing an earthquake.

SEE: Access roads to the dam and itsappurtenant works may become

DRAFT



impassable after an SEE event.However, for access roads that areowned and maintained by the DamOwner they should so far as ispracticable be easily cleared or madeaccessible if there are no alternativeaccess routes available. Access bridgesthat are owned and maintained by theDam Owner should remain capable ofcarrying the design loads if there are noalternative access routes available. Theimpact of this on access roads andbridges that are not owned ormaintained by the Dam Ownerbecoming impassable needs to beconsidered in assessing the requiredoperation of appurtenant structuresfollowing an earthquake.

For bridges over spillways these maybe combined with or attached to thesupporting structure for operatingmechanisms that open and close gates.For purposes of dam safety, suchbridges should not impede the ability toopen and close spillway gatesfollowing the occurrence of the SEE.

6.9.4 Analysis Procedures (Bridges)Bridges should be assessed using anappropriate dynamic analysis of the pier andbridge system and should be in accordance

with AS5100 – Bridge Structures. For intaketower access bridges, the analysis may be partof an overall dynamic analysis for an intaketower. The pier and bridge system should beexamined for seismic ground motions in thedirection of the bridge access andperpendicular to the bridge access. Loadsresulting from seismic ground motions may becombined on a square root of the sum of thesquares (SRSS) basis.

6.10 Reservoir Rim InstabilityIf there are existing or potential landslides onthe reservoir rim, assess the likelihood that theslides will activate under seismic loads, and ifso, whether the slide debris will reach thereservoir and at what velocity and volume.Then assess the potential for waves to begenerated by the landslide debris, and if so, theeffects on the dam and appurtenant structures.

DRAFT

Commentary - Table of Contents


C1 INTRODUCTION………...………………………………………… ........ ……………………44

C2 ASSESSMENT OF DESIGN SEISMIC GROUND MOTIONS AND ANALYSIS METHOD 45

C2.1 Earthquakes and Their Characteristics ............................................................................... 45

C2.1.1 Seismology for Earthquake Hazard Studies ................................................................. 45

C2.1.2. Earthquake Recurrence .............................................................................................. 51

C2.1.3. Seismic Ground Motion ............................................................................................. 55

C2.1.4. Earthquake Hazard in Australia .................................................................................. 60

C2.1.5. Reservoir Triggered Earthquake ................................................................................. 64

C2.2Terminology....................................................................................................................... 66

C2.3 General Principles for Selection of Design Ground Motion and Analysis Method.............. 66

C2.4 Description of a Probabilistic Seismic Hazard Assessment ................................................ 67

C2.5 Requirements of a Seismic Hazard Assessment ................................................................. 76

C2.6 Selection of Design Seismic Ground Motion - Deterministic Approach ............................. 90

C2.7 Selection of Design Seismic Loading- Risk Based Approach ............................................. 92

C2.8 Modelling vertical ground motions .................................................................................... 93

C2.9 Selection of Response Spectra and Time History Accelerograms ....................................... 93

C2.10 Earthquake Aftershocks ................................................................................................... 94

C2.11 Seismic Ground Motions from Earthquakes Induced by the Reservoir ............................. 95

C3 ASSESSMENT OF EMBANKMENT DAMS FOR SEISMIC GROUND MOTIONS 96

C3.1 Effect of Earthquakes on Embankment Dams. ................................................................... 96

C3.2 Defensive Design Principles for Embankment Dams. ........................................................ 96

C3.3 Seismic Deformation Analysis of Embankment Dams. ...................................................... 96

C3.3.1 The Methods Available and When to use them ............................................................ 96

C3.3.2 Screening and Empirical Database Methods ................................................................ 97

C3.3.3 Simplified Methods for Estimating Deformations during Earthquakes ........................ 99

C3.3.4 Post-earthquake Deformations for Liquefied Conditions ........................................... 100

C3.3.5 Advanced Numerical Methods for Estimating Deformations During and Post-Earthquake for Non-Liquefied and Liquefied Conditions ..................................................... 101

C3.4 Assessment of the Effects of Liquefaction in Embankment Dams and Their Foundations 103

C3.4.1 Overall Approach ...................................................................................................... 103

C3.4.2 Definitions and the Mechanics of Liquefaction ......................................................... 103

C3.4.3 Methods for Identifying Soils which are Susceptible to Liquefaction ........................ 104

C3.4.4 Methods for Assessing Whether Liquefaction may occur .......................................... 106

DRAFT



C3.4.5 Methods for Assessing the Strength of Liquefied Soils in the Embankment andFoundation ........................................................................................................................... 112

C3.4.6 Methods for Assessing the Post-Earthquake Strength of Non-Liquefied Soils in theEmbankment and Foundation ............................................................................................... 114

C3.4.7 Site Investigations Requirements and Development of Geotechnical Model of theFoundation ........................................................................................................................... 115

C3.4.8 Liquefaction Analysis ............................................................................................... 117

C3.4.9 Assessment of the Likelihood and the Effects of Cracking of Embankment DamsInduced by Seismic Ground Motions .................................................................................... 117

C3.5 Methods for Upgrading Embankment Dams for Seismic Ground Motions ....................... 117

C3.5.1 General approach ...................................................................................................... 117

C3.5.2 Embankment Dams not Subject to Liquefaction ........................................................ 118

C3.5.3 Embankment Dams Subject to Liquefaction .............................................................. 118

C3.6 Flood Retarding Basins ................................................................................................... 119

C4 ASSESSMENT OF CONCRETE DAMS FOR SEISMIC GROUND MOTIONS……………120

C4.1 Effect of Earthquakes on Concrete Dams......................................................................... 120

C4.1.1 General ..................................................................................................................... 120

C4.1.2 ICOLD ..................................................................................................................... 121

C4.2 Defensive Design Principles for Concrete Dams ............................................................. 121

C4.3 General Principles of Analysis for Seismic Ground Motions ........................................... 121

C4.3.1 Gravity Dams............................................................................................................ 121

C4.3.2 Arch Dams ................................................................................................................ 121

C4.3.3 Buttress Dams ........................................................................................................... 122

C4.3.4 Other Types .............................................................................................................. 122

C4.4 Seismic Structural Analysis of Concrete Dams ................................................................ 122

C4.4.1 Material Properties .................................................................................................... 122

C4.4.2 Loads ........................................................................................................................ 123

C4.4.3 Methods Available and When to Use Them .............................................................. 124

C4.4.4 Analysis in the Frequency Domain ............................................................................ 126

C4.4.5 Analysis in the Time Domain .................................................................................... 128

C 4.5 Approach to Analyses and Acceptance Criteria .............................................................. 128

C4.5.1 During the Earthquake .............................................................................................. 128

C4.5.2 Post-Earthquake ........................................................................................................ 129

C4.5.3 General ..................................................................................................................... 130

DRAFT



C5 ASSESSMENT OF TAILINGS DAMS FOR SEISMIC GROUND MOTIONS ................. 131

C5.1 Some General Principles ................................................................................................. 131

C5.2 The Effects of Earthquakes on Tailings Dams ................................................................. 131

C5.3 Defensive Design Principles for Tailings Dams ............................................................... 131

C5.4 Design Seismic Loads and Analysis Method ................................................................... 131

C5.5 Seismic Deformation Analysis of Tailings Dams not Subject to Liquefaction .................. 132

C5.6 Assessment of the Effects of Liquefaction in Tailings Dams and Their Foundations ........ 132

C5.7 Methods for Upgrading Tailings Dams for Seismic Loads ............................................... 132

C6 ASSESSMENT OF APPURTENANT STRUCTURES FOR SEISMIC GROUND MOTIONS ................................................................................................................................................ 133

C6.1 The Effects of Earthquakes on Appurtenant Structures .................................................... 133

C6.2 Intake/Outlet Towers ....................................................................................................... 133

C6.2.1 General ..................................................................................................................... 133

C6.2.2 Defensive Design Principles ..................................................................................... 133

C6.2.3 Performance Requirements ....................................................................................... 133

C6.2.4 Analysis Procedures .................................................................................................. 133

C6.3 Spillways ........................................................................................................................ 135

C6.4 Spillway Gates ................................................................................................................ 136

C6.5 Outlet Works - Water Conduits, Gates and Valves........................................................... 136

C6.5.1 General ..................................................................................................................... 136




C6.6 Retaining Walls ............................................................................................................... 136

C6.6.1 General ..................................................................................................................... 136




C6.7 Parapet Walls .................................................................................................................. 137

C6.8 Mechanical and Electrical Equipment .............................................................................. 137

C6.9 Access Roads and Bridges ............................................................................................... 137

DRAFT

Commentary – 1.0 Introduction


C1 INTRODUCTIONNo commentary.

DRAFT

Commentary – 2.0 Assessment of Design Seismic Ground Motions AnalysisMethod

ANCOLD Guidelines for Design of Dams and Appurtenant Structures for Earthquake – Draft March 201745

C2 ASSESSMENT OF DESIGN SEISMICGROUND MOTIONS ANDANALYSIS METHOD

C2.1 Earthquakes and TheirCharacteristics

C2.1.1 Seismology for Earthquake HazardStudies

C2.1.1.1 Earthquake Mechanisms andTerminologyAn earthquake is the sudden slip on a faultthat is produced when stress within the earthbuilds up over a long period of time until iteventually exceeds the strength of the rock,which then fails and a break along a fault isproduced. It may take tens, hundreds orthousands of years for the stress to build upin a particular area, and it is then released inseconds. Part of the energy is transmittedaway as seismic waves and part as heat.

The fault displacement in a particularearthquake may vary from a few millimetres

up to a few metres. Once ruptured, the fault isa weakness, which is more likely to furtherrupture in future earthquakes, so a large totaldisplacement may build up from manyearthquakes over a long period of time. Thismay eventually measure kilometres for thrustfaults produced by compression, or hundredsof kilometres for horizontal strike-slip faultssuch as the San Andreas fault in California.

As shown in Figure C2.1 the point on thefault surface where a displacementcommences is called the hypocentre or focus,and the earthquake epicentre is the point onthe ground surface vertically above thehypocentre.

Figure C2.1 Definition of earthquake terms.

The fault slip initially propagates away fromthe hypocentre in a circular manner, but itseventual rupture extent is variable, sometimes

propagating in both directions along strikeand sometimes only in one direction.Maximum energy release is not necessarily

DRAFT



near the hypocentre, and is often not evenclose to the hypocentre.

The hypocentral distance from an earthquaketo a point is the three-dimensional slantdistance from the hypocentre to the point,while the epicentral distance is the horizontaldistance from the epicentre to the point.

Tectonic stress within the earth is caused bydeformation that results from platemovement. The stress can be resolved intothree orthogonal principal stresses, themaximum, intermediate and minimumprincipal stresses, usually denoted by σ1, σ2,and σ3.

For crustal earthquakes, one of theseprincipal stresses is usually near vertical, sothe other two will be near horizontal. Thestress directions determine the type andorientation of the faulting.

The vertical principal stress at any depthusually has a value comparable with the‘lithostatic pressure’, or ρgh, where ρ is thedensity of the rocks above the point. The twonear horizontal principal stresses can behigher or lower than this value, or one can behigher and the other lower, depending onwhich of σ1, σ2, or σ3 is vertical.

If maximum principal stress σ1 is vertical, or‘normal’ to the Earth’s surface we get anormal fault, resulting in a horizontalextension. If the minimum principal stress σ2is vertical we get a reverse fault, resulting inhorizontal compression. If the intermediateprincipal stress σ3 is vertical a strike-slip faultis produced, in which two blocks movehorizontally relative to each other.

Figure C2.2: Principal stress orientation and fault types (P: Maximum, B: Intermediate, T:Minimum)

.

Extension Shortening

Fault Plane

NormalFault

ReversedFault

Strike-SlipFault

DRAFT



A fault will only be active if its type andorientation is consistent with the principalstress orientation, to within about 20degrees. Ancient faults that are not orientedappropriately are very unlikely to fail, butthose that are so oriented may bereactivated. Stress orientations may beinferred from earthquake mechanisms, orobtained from the World Stress Map.

In Australia, all measurements to date showthe maximum principal stress is nearhorizontal, with a high level of stresscompared with the lithostatic pressure. Theminimum principal stress is usually verticalso most faults are reverse. In some locationsthe intermediate principal stress is nearvertical, so active faults at these places arelikely to be strike-slip.

Using traditional mining terms for normaland reverse faults, the block that lies overthe dipping fault is called the hanging-wallblock, and the one under the fault is thefoot-wall block.

For strike-slip faulting, standing on oneblock and facing the other, then observingwhich way the other is moving (right orleft), defines right- or left-lateral strike-slipfaulting.

C2.1.1.2 The Fault Rupture ProcessThe earthquake process is more complexthan the simple displacement between twoblocks shown in Figure C2.2.

A fault is a weakness that has experiencedmany ruptures in the past, each oneproviding an incremental shift between theblocks on either side. As tectonicdeformation takes place at a very slow butfairly constant rate, the relative motionbetween blocks is concentrated along thefault, with bending that results in shearstrain, and the accumulation of shear strainenergy in the blocks on both sides of thefault as shown in Figure C2.3.

Since the Earth’s surface is a free surface,this will result in folding and a verticaloffset. This deformation happens veryslowly over a very long period of time. Thefolding has low amplitude relative to thehorizontal extent, metres vertical overkilometres or tens of kilometres horizontaldistance. Reverse faults give folding in thehanging-wall block, similar to the foldsproduced by subduction in the over-ridingplate, but without outer-rise folding on thesubducting plate. Small earthquakes occurthroughout the folded region duringdeformation.

Figure C2.3: Deformation associated with reverse faulting.

The main active reverse fault dips under theuplifted block, so epicentres of foreshocksand aftershocks that are on this fault extendfor tens of kilometres on the up-thrownblock side of the fault surface rupture, with

few earthquakes in the footwall block. Someforeshocks and aftershocks may also occurabove the main fault in the up-thrown block.

DRAFT



Fault Rupture and generated seismic waves

When the stress level exceeds the strengthof the fault at a point on the fault, itruptures. The stored strain energy allows therupture to propagate along the fault, despiteconsiderable loss of energy due to frictionand some due to radiation of seismic waves.Propagation continues until there isinsufficient strain energy to continue therupture, kinetic energy drops to zero, andthe rupture displacement stops.

When the rupture occurs the release of shearstrain energy leads to the production of ashear wave (S wave) that travels throughrock at speeds of about two-thirds the speedof the P wave (compressional wave), whosemotion is in the longitudinal and thusstrongest on the vertical component on theground surface. The S wave is a transversewave with motion perpendicular to thedirection of propagation, and is strongest inthe horizontal plane on the ground surface.The amplitudes of the P and S wavesdecrease with distance due to geometricspreading and absorption of wave energy.

The stored compressional and shear strainenergy are the tectonic energy that will bereleased in the earthquake.

The strain energy accumulates slowly over along period, until it exceeds the strength at apoint on the fault plane, which thenruptures. The rupture front spreads over thefault plane at about 3 km/s, with one blockmoving at about 1 m/s relative to the other,decreasing with time due to friction and lossof available strain energy.

Seismic waves leave the fault rupture at awave velocity depending on local rocks, buttypically 4 to 6 km/s for P waves, withground motion velocities starting at aboutthe relative block motion velocity (1 m/s).

Surface Rupture

Almost all earthquakes, especially largerearthquakes, occur on existing faults. This is

because faults are weaker than surroundingunbroken rock, and are much more likely tofail again when stress rebuilds. Earthquakesnucleate within the brittle zone of theshallow crust, which is typically about 10 to20 km thick. There is a brittle-to-ductiletransition at the base of this seismogeniczone, typically at a depth of 10 to 20 km,caused by the increase in temperature withdepth in the earth. Because rocks near thesurface are relatively weak in non-cratonicregions of Australia (as defined in Clark etal., 2011; 2012; Figure C2.20), there isanother brittle-to-ductile transition atshallow depths, typically a few km, so thatfew earthquakes originate at shallow depths(in the top one or two kilometres). It iscommon for surface rocks to be folded inresponse to faulting at depth, giving amonocline and scarp at the surface, butwithout a surface fault. In cratonic regionsof Australia, which include most of thewestern part of Australia, rock near thesurface is quite strong and shallowearthquakes are more common that in non-cratonic regions.

Surface rupture occurs when a fault breakreaches the ground surface. It may producea vertical or horizontal offset, or both, witha displacement of millimetres to a fewmetres, and a length from metres to tens ofkilometres. Surface rupture has occurredfrequently in the past century in cratonicregions of Australia, but is expected to beless common in non-cratonic regions ofAustralia.

A site will have surface rupture potential ifan existing fault is found which has beenactive during the current stress regime.Clark et al. (2011) describe neotectonicfeatures have undergone displacement underthe current stress regime in Australia, andhence may have the potential fordisplacement in the future. The age of thecurrent stress regime in Australia isestimated to lie in the range of 5 to 10million years (Sandiford et al., 2004).

DRAFT



C2.1.1.3. Earthquake MagnitudeEarthquakes vary enormously in size.Richter (1935) defined a magnitude scale toindicate the size of an earthquake as

L 10M log A=

where

ML is the Richter local magnitudescale

A = maximum seismic waveamplitude (in thousandths of amillimetre) recorded from astandard seismograph at adistance 100 km from theearthquake epicentre

or

L 10M log A F( ) k= - D +

where

F(D) = distance correction; k =scaling constant.This allows calculation of ML fromseismographs which have recordedthe earthquake at different distancesfrom the earthquake.

Other magnitude scales have been defined,including the Moment Magnitude (MW)which assigns a magnitude to the earthquakein accordance with its seismic moment MO,which is directly related to the energyreleased by the earthquake:

W 10 OM (log M /1.5) 10.7= -

where

MO is the seismic moment in dyn-cm.

Moment magnitude is the scale mostcommonly used for engineeringapplications.

These magnitude scales give similar valuesthat can range from 0.0 to over 9.0. Foreach unit of magnitude there is a tenfoldincrease in ground displacement, and athirtyfold increase in seismic energy release.

Another measure of earthquake size is thefault area, or the area of the fault surface,which is ruptured. The fault area ruptured inan earthquake depends on the magnitudeand stress drop in the earthquake. Typically,a magnitude 4.0 earthquake ruptures a faultarea of about 1 square kilometre, magnitude5.0 about 10 square kilometres, andmagnitude 6.0 about 100 square kilometres(perhaps 10 x 10 kilometres).

There are approximate relationshipsbetween the magnitude of an earthquake andthe rupture area of a fault and otherparameters, such as Gibson (1994) as shownin Table C2.1 and Leonard (2010).

DRAFT



Table C2.1: Approximate measures of shallow earthquake ruptures. (Modified from Gibson,1994).

Magnitude

RuptureArea

(km2)

TypicalLength x

Width(km x km)

Fault Slip~Length/20,00

0(metres)

RuptureDuration

~Length/3(seconds)

EnergyReleased

(MJ)

Global AverageNumberper year

2.0 0.01 0.1 x 0.1 0.005 0.03 60 2,000,000

3.0 0.1 0.3 x 0.3 0.015 0.1 2,000 200,000

4.0 1 1 x 1 0.05 0.3 60,000 20,000

5.0 10 3 x 3 0.15 1 2 million 2,000

6.0 100 10 x 10 0.50 3 60 million 200

7.0 1000 30 x 30 1.5 10 2,000million

20

8.0 10,000 200 x 50 5.0 60 60,000million

1.0

C2.1.1.4 Maximum Credible EarthquakeMagnitudeIn view of the impracticability of identifyingall the faults on which major earthquakesmay occur, it is necessary to useprobabilistic methods to estimate expectedground motion versus Annual ExceedanceProbability (AEP). For this, it is necessaryfirst to assess the magnitude of themaximum credible earthquake, Mmax.

If the earthquake catalogue only covers ashort period compared with the requiredreturn period, then the activity from a largesurrounding area may be considered whenestimating Mmax, perhaps as large as thewhole of Australia or even including otherintraplate areas over the earth. To give someappreciation of the likely maximum crediblemagnitude, Seismologists have consideredthe credible maximum lengths and widths offaults that may rupture to cause theearthquake. Because of the limitedseismogenic width of the shallow crust, largeearthquakes have long rupture lengths, asindicated in Table C2.1.

Very few intraplate earthquakes are largerthan magnitude 7.5. A series of very largeintraplate earthquakes in the New Madridarea of Missouri, USA, in 1811 to 1812, hadoriginal published magnitude valuesexceeding 8.0 (Johnston and Shedlock,1992). Recent authors believe that the NewMadrid earthquakes were considerablysmaller than this (Evernden, 1975, M6.9;Gomberg, 1992 M7.3; Hough et al, 2000).

There is general consensus amongAustralian Seismologists that the maximumcredible magnitude of an earthquake inAustralia is about magnitude 7.5. This isbased on consideration of the limited depthrange at which crustal earthquakes canoccur, and the sensible maximum length offault that will rupture. It was originallysuggested that a magnitude 7.5 earthquakecould occur anywhere in the country, but itis possible that this value may be reduced ifthere is evidence that active faulting,especially surface faulting, has not occurredin particular locations.

DRAFT



Given the shallow seismogenic depths inEastern Australia, the value of Mmax 7.5 ispossibly conservatively high. However forreturn periods to 1000 years, (annualprobability of 0.001) decreasing Mmax to7.2 or 7.3 would give negligible effect onprobabilistic ground motion estimates. Forlonger return period (lower annualprobability) motion, the lower value couldstill be justified, especially if there is littleseismological and especially geologicalevidence for the existence of large nearbyfaults.

C2.1.2. Earthquake Recurrence

C2.1.2.1. IntroductionEarthquake source regions are either knownfaults, or source zones where faults areunknown or earthquakes are distributed overmany small faults. A zone can be anyvolume in the earth, but is usually defined asa polygonal prism with vertical sides andhorizontal top and bottom. For determinationof earthquake recurrence, the maincharacteristic of each zone is that it isreasonable to expect that earthquake activityis uniform throughout the zone.

For known faults we need to know the typeof fault, location and dip direction. The bestmeasure of the activity on the fault is theaverage slip rate, usually given inmillimetres/year for active regions, ormetres/million years in stable regions.

If the earthquake catalogue included allevents within the entire zone, at all timesduring the observation period, then arecurrence plot of rate of activity againstmagnitude could be determined by simplycounting the number of events with eachmagnitude and dividing by the number ofyears of observation.

Unfortunately earthquakes and seismographcoverage are not so simple. Earthquakescluster in time and space with long periodsof quiescence. Seismograph coverage varieswith time, usually giving complete coverageto lower magnitudes as seismograph density

increases, but sometimes deteriorating asindividual seismographs are removed, orseriously deteriorating as seismographnetworks are removed.

The coverage of a zone depends on thelocation of seismographs relative to thezone. If seismographs are aligned in onedirection relative to the zone, either inside oroutside the zone, then coverage may varywidely across the zone, with completecoverage being limited to larger events.

C2.1.2.2. The Gutenberg-RichterRelationship

In seismology, the Gutenberg–Richter lawexpresses the relationship between themagnitude and the total number ofearthquakes in any given region and timeperiod of at least that magnitude.

log = −

Or

= 10

where:

· N is the number of events having amagnitude ≥

· a and b are constants

The parameter b (commonly referred to asthe "b-value") is commonly close to 1.0 inseismically active regions. This means thatfor given a frequency of magnitude 4.0events there will be 10 times as manymagnitude 3.0 earthquakes and 100 times asmany magnitude 2.0 earthquakes.

The a-value indicates the total seismicityrate or activity of the region.

C2.1.2.3. Random and Non-RandomProcessesThe earthquake recurrence rate is usuallyestimated by assuming that the earthquake

DRAFT



activity rate observed over the recent pastwill continue unchanged.

Earthquake hazard studies usually assumethat earthquake recurrence is a randomprocess. That is, the probability of anearthquake does not change with space ortime, and all earthquakes are independent ofeach other.

One of the most obvious examples of non-random activity in time is the foreshock-main shock-aftershock sequence. In hazardstudies, if it is assumed that the main shockwill be the most damaging event in thesequence, foreshocks and aftershocks can beremoved (de-clustered) and the recurrence ofmain shocks is determined, normally leadingto an increase in estimated hazard becausethe “b-value” is then lower.

Non-random activity in time is clear in anearthquake cycle represented by a sequenceincluding a period of quiescence, possibleprecursory events for large earthquakes,foreshocks, main shock, aftershocks,possible adjustment events and anotherquiescence period.

Non-random activity in space is largelyrelated to faults. Large earthquakes onlyoccur on large faults and small earthquakesmost often occur on small faults. This meansthat the relative number of small to largeearthquakes, the Gutenberg-Richter b-value,varies depending on local geology, not onlyin its value but also in the magnitude rangeof applicability. Variations in b-value areparticularly obvious with earthquake depthwith high values at shallow depths and lowvalues for deep earthquakes. There isincreasing evidence of earthquake activitybeing affected by recent past activity inneighbouring regions, both on a single largefault (or plate boundary) and on other nearbyfaults.

One of the most common ways to force theseismicity of an area to fit a singleGutenberg-Richter distribution is to uselarge zones that include many earthquakes

from a wide range of sources. This leads to“large-zone hazard dilution”, where theestimated hazard in more active regions issignificantly reduced, while that in largerareas of low activity is increased slightly.

Future earthquake hazard studies willcontinue to rely on past earthquakes, but willgive greater consideration to geological andgeodetic data and processes, including stress,strain energy, and the dynamics ofdeformation in a complex geologicalenvironment.

C2.1.2.4. Time-Magnitude PlotsThe earthquake recurrence or seismicity(seismic activity) of an area must take therange of earthquake sizes into account.There are many more small earthquakes thanlarge. As discussed above, in most placesaround the earth the b-value is 1.0. The bvalue may be 1.3 or higher if there are manysmall earthquakes, or 0.7 or lower if thereare relatively few small earthquakes.

The recurrence rate is complicated by thewide range of earthquake magnitudes thatare reported, or recorded and located. Mosthistorical earthquakes larger than aboutmagnitude 4.0 or 5.0 are reported innewspapers because of the intensity ofground motion that was felt, or the damagethat was caused. National seismographnetworks can currently record allearthquakes exceeding about magnitude 2.0to 3.0 depending on the seismograph densityand distribution. A dense local network ofseismographs about a site can recordearthquakes exceeding about magnitude 0.0to 1.0.

The Gutenberg-Richter relationship is usedto quantify the earthquake recurrence for aregion. The activity (a) is given by theintercept on the vertical axis, the relativenumber of small to large magnitudes by thegradient (or b-value) and a maximummagnitude is applied.

The b-value is not a constant, but varies withspace and time. If a number of different

DRAFT



processes are included, each with their ownb-value and other parameters, the apparentseismicity of the region may still have awell-defined b-value, but at someintermediate value depending on the regionand time interval considered (the average ofa set of straight lines gives another straightline, provided that linearity occurs for allearthquakes over the magnitude rangeconsidered).

If the seismicity is dominated by a swarm ofshallow earthquakes then the observed bvalue will be too high to allow extrapolationto higher magnitudes, resulting in hazardestimates too low, or if it is dominated by afew moderate to larger earthquakes (or thecatalogue is incomplete), the observed b-value will be too low to allow reliableextrapolation, resulting in hazard estimatesthat are too high.

This assumption of temporally randomearthquake occurrence clearly does not applyto earthquake clusters consisting of possibleforeshocks, a main shock, and aftershocks.For many years, standard practice has beento consider the recurrence of clusters ratherthan individual events, with the clusterrepresented by the main shock, which is theevent in the cluster that will give strongestmotion, and usually (but not always) causethe most damage. A de-clustered earthquakehistory is used, and foreshocks andaftershocks are simply not considered in theearthquake recurrence calculations.

Figure C2.4: Earthquake history for the West Sydney Basin zone.

As an example the seismological history ofthe West Sydney Basin zone is shown inC2.4, with the dashed line showing theestimated variation in magnitude threshold

for complete coverage of larger magnitudes,generally decreasing as the seismographnetwork density improves over time.

DRAFT



The difference in rates of activity formagnitudes 1.5 to 2.2 between 1970 and1988, and between 1992 to the present isprobably because the original catalogue wasnot fully “de-blasted”, so contains manyquarry and mine blasts as well asearthquakes. The vertical lines of events

show clusters, usually a main shock withforeshocks and aftershocks.

The next stage in quantifying the activitywithin the zone is to convert the historicaldata into cumulative recurrence data. This isshown in Figures C2.5 and C2.6.

Figure C2.5: Earthquake magnitude recurrence, not de-clustered.

Figure C2.6: Earthquake magnitude recurrence, de-clustered.

DRAFT



Note that in this example that the earthquakewith a 1/1000-year average recurrenceinterval is magnitude 6.0 for the originaldata and 6.4 for the de-clustered data. Thecorresponding values for 10,000-yearaverage recurrence interval are magnitude6.9 and magnitude 7.2 respectively.

C2.1.2.5. Geologically Observed SlipRates vs. Extrapolated ObservedSeismicityLarger active faults appear to havegeologically observed slip rates that require

greater earthquake activity rates than theextrapolated seismicity estimates (FiguresC2.6 and C2.7,). Earthquake recurrenceestimates made by extrapolating smallearthquake recurrence will under-estimatethe hazard in these regions.

Figure C2.7: Earthquake magnitude recurrence for the Newcastle zone.

In areas with no obvious large active faults,such as large areas of undeformed flat-lyingsediments, the observed recurrence rates arelower than the extrapolated estimates (orzero), with fewer (or not so large)earthquakes. This suggests that these smallerearthquakes occur in regions with smallfaults, and lower maximum magnitude.Earthquake recurrence estimates made byextrapolating small earthquake recurrence

will over-estimate the hazard in theseregions.

C2.1.3. Seismic Ground Motion

C2.1.3.1. Ground Motion MeasuresEarthquake ground vibration is recorded bya seismograph or a seismogram. Mostmodern seismographs record three

DRAFT



components of motion: east-west, north-south and vertical.

The rupture time for a small earthquake is afraction of a second, for earthquakes ofmagnitude 5.0 it is about a second, and forlarge earthquakes may be up to tens ofseconds. However the radiated seismicwaves travel at different velocities, and arereflected and refracted over many travelpaths, so the total duration of vibrations at asite persist longer than the rupture time, andshow an exponential decay.

As discussed above, several types of seismicwave are radiated from an earthquake. Bodywaves travel in three dimensions through theearth, while surface waves travel over thetwo-dimensional surface like ripples on apond. There are two types of body wave (Pwaves and S waves), and two types ofsurface waves (Rayleigh and Love waves).

Primary or P waves are ordinary soundwaves travelling through the earth. They arecompressional waves, with particle motionparallel to the direction of propagation.

Secondary or S waves are shear waves, withparticle motion perpendicular to thedirection of propagation. The amplitude of Swaves from an earthquake is usually largerthan that of the P waves.

P waves travel through rock faster than Swaves, so they always arrive at aseismograph before the S wave.

The frequency content of seismic groundmotion covers a wide range of frequenciesup to a few tens of hertz (cycles per second).Most engineering studies consider motionbetween about 0.2 and 25 Hz.

The amplitude, duration and frequencycontent of seismic ground motion at a sitedepend on many factors, including themagnitude of the earthquake, the distancefrom the earthquake to the site, and local siteconditions.

The larger the earthquake magnitude, thegreater the amplitude, the longer the durationof motion, and the greater the proportion ofseismic energy at lower frequencies. A smallearthquake has low amplitude (unless it isvery close), short duration, and has mainlyhigh frequencies.

The smaller the distance from an earthquaketo the site, the higher the amplitude. Theduration is not strongly affected by distance.High frequencies are attenuated byabsorption within the ground more quicklythan low frequencies, so at greater distancesthe proportion of seismic vibration energy athigh frequencies will decrease.

C2.1.3.2. Earthquake IntensityEarthquake Intensity is a measure of theeffect of the seismic waves at the surface,and is normally given on the ModifiedMercalli Intensity scale; a copy of which isattached in Appendix A. This is an arbitraryscale defined by the effects observed(whether sleeping people were woken, treesshaken, etc.) and on the damage caused.Normally the maximum intensity occursnear the epicentre of the earthquake, andintensity then decreases with distance.However this may be affected by theorientation of the rupture, or local groundconditions such as topography or surfacesediments.

C2.1.3.3. Attenuation and Amplificationof Ground MotionSeismic ground motion generally attenuateswith increasing distance from the source dueto radiation and hysteretic damping. Highfrequency motion is attenuated more quicklywith distance than lower frequency motion.

In the old, hard rocks in cratonic regionssuch as the shield areas in Western Australiaand in eastern Canada, earthquakes ofmagnitude 5.0 can be felt at much greaterdistances than in Eastern Australia, over 400km compared with less than 150 km, andcratonic ground motion models must reflectthis. Unfortunately, there is very little strongmotion data available from such areas.

DRAFT



In cratonic areas there are often very hardcrystalline rocks to near the surface andearthquake depths tend to be quite shallow.This can lead to generation of strong surfacewaves by relatively small earthquakes withlong period motion (about 2.0 seconds in theYilgarn of Western Australia). Groundmotion models for this region (Liang et al.,2008; Somerville et al., 2009) incorporatesuch unusual characteristics.

In selecting attenuation relationships orground motion prediction equations care isneeded, and attention paid to the mechanismof the source earthquake, e.g. whethershallow intraplate or deep plate boundaryearthquakes.

At low amplitude levels, the ground motionis amplified by the near-surface geologybecause it is generally more flexible thanunweathered bedrock. The amplificationincreases as the shear wave velocity of thesurface geology decreases due to impedanceamplification. The shallow shear wavevelocity is quantified, for example, by Vs30,the time-averaged shear wave velocity in theupper 30 metres of the ground. However, athigh amplitude levels in soft soils, weaksurface materials absorb seismic energyrather than transmit it unchanged, thustending to reduce amplitudes at the surface.The amount of attenuation depends on theproperties of the materials, and especiallytheir thickness.

In horizontally stratified sediments, the nearsurface layers will vibrate preferentially attheir own natural frequencies, depending ontheir thickness and elastic properties. Theearthquake motion at the natural frequenciesof the near-surface layers is amplified, whilemotion at other frequencies may be littleaffected or even attenuated. Theamplification effect can be especiallypronounced for deep soft sediments thathave very low damping, such as thoseunderlying Mexico City.

Dams (like all other structures) each havetheir own natural frequency depending on

their mass and stiffness, usually in the rangefrom about 0.5 to 5 Hz for embankmentdams and 2 hertz to 20 hertz for concretegravity dams.

C2.1.3.4. Site Response – Near-SurfaceAmplification and AttenuationGround motion at a site on the Earth’ssurface may be significantly affected bynear-surface geology and topography. Thereare many phenomena that affect near-surfacemotion, including:

· Impedance amplification· Resonance of surface sediments· Additional sedimentary basin effects· Variation of frequency-dependent

attenuation, Q(f), in surface sediments· Scattering in complex geology· Reflections in stratified surface geology· Groundwater· Wave conversions at interfaces· Focusing by complex structures· Topographic effects

Combinations of impedance amplificationand resonance can give amplifications ofbedrock ground motion that can exceed x 4to x 8 at resonant frequencies. Resonantmotion gives greater amplification tohorizontal motion than to vertical motion.Site response can vary over very shortdistances, such as tens of metres.

Site response can be measured by comparingsurface motion with bedrock motion.

Measurement of ambient vibrations at apoint on the surface can clearly indicate thenatural frequencies of the sedimentarycolumn simply because the spectrum showsa high peak. In other cases using single pointmeasurements, the ratio of horizontalspectral motion over vertical spectral motionwill give some idea of both naturalfrequencies and the degree of amplification,because horizontal motion is amplified muchmore than vertical motion.

Measuring ambient vibrations with a smallarray can reveal more information about the

DRAFT



sedimentary column, such as the SpectralAutocorrelation (SPAC) method which givesa velocity dispersion curve from which thevariation of shear wave velocity with depthcan be estimated. These methods usuallyconsider depth ranges from metres tohundreds of metres.

Site amplification has little effect on hardrock sites, but can dominate the hazard onsites with deep soft sediments.

Free-Surface Amplification

The amplitude of seismic waves increases asthey approach the Earth’s (free) surface, atwhich they are reflected. This amplificationcan be considered in terms of constructiveinterference between the up going and downgoing waves, and can give surface motion upto about double that at depth for a uniformhalf space. However, as noted above, thenear surface material usually has lower shearwave velocities than bedrock, so the surfacemotion is usually more than twice that atdepth. The nature and degree of free surfaceamplification varies with topography, evenin fresh, strong rock. Changes in soilthickness above an irregular bedrock surfacecan give complex surface amplification thatvaries with wave duration.

Resonance

Resonance in the surface sediments causesamplification at particular frequencies,especially at the natural frequency of thesediments. This depends on the thicknessand elastic properties of the sediments.Seismic motion recorded on hard rock tendsto be broadband, while that recorded on softsediments is usually dominated by theresonant frequency.

Attenuation

In surface sediments, high frequencyvibrations are attenuated much more withdistance than low frequencies. If sedimentsare very thick, much of the high frequencymotion will be lost and peak surfaceaccelerations will be relatively low, even if

resonance has amplified motion at the lowresonant frequency.

Local site conditions on a relatively smallscale, particularly soft surface sediments butalso topography, can significantly affectsurface motion from an earthquake. Theseeffects can vary rapidly with location,showing significant changes over distanceswithin tens or hundreds of metres.

Estimating Site Response in Practice

Site response, or site amplification, dependson many phenomena, including variation inimpedance depending on the rock properties,variations in attenuation especially throughsoft rocks, resonance in surface layersespecially in soft rock, resonance ofsedimentary basins, conversion of wavetypes between P, S and surface waves at thesurface or at boundaries beneath the surface,and focusing of waves by irregular surfaces.

The treatment of each of these can be fromsimplistic to complex. The criticalphenomena will vary from site to site. Someof the phenomena are period (or frequency)dependent and require spectral variations(e.g. resonance, attenuation), and others varylittle with frequency (e.g. impedanceamplification).

For many earthquake hazard studies it isnormal practice to calculate bedrock motionfor surface outcrop or at the surface ofbedrock below sediments. Unless the site isnear to an active fault, there is little variationin estimated bedrock motion over distancesof kilometres, and the same bedrock motioncan be applied at various locations about thesites. That does not hold if the rockfoundation conditions are greatly differentbetween the river bed and the abutments.The bedrock motion can then be used toestimate surface motion or motion of thedam, listed in order of increasingcomplexity, accuracy and cost:

DRAFT



· When using Australian Loading codeAS1170.4 (2007), the bedrock groundmotion values given would be multipliedusing the AS1170.4 site sub-soil classfactors, which vary from 0.8 for hardrock to 3.5 for very soft soils. This doesnot consider frequency dependentresonance. AS1170.4 has limitedapplication for dams because it onlyprovides loads for the 10% chance ofexceedance in 50 years, or 1 in 475annual probability.

· For the next level of design, forhorizontally stratified surface layers, alocal value of Vs30 allowing for the Vsof the sediments overlying bedrock canbe used to recalculate the results to givean average estimate of site response thatconsiders impedance amplification, butdoes not consider frequency dependentphenomena such as resonance. Vs30 isthe time averaged shear wave velocity inthe top 30 metres of rock and sediments,calculated from the inverse of the meanslowness in the top 30 metres (usingseconds per metre rather than metres persecond), or 30 divided by the totalvertical shear wave velocity (Vs) traveltime through these 30 metres. This givesa greater weighting to low velocitiesrather than to high velocities. Use ofVs30 to estimate soil amplification at aspecific site assumes that the averageshear wave velocity, shear modulusreduction, and damping increase profilesrepresented in the strong motiondatabase used to develop the groundmotion prediction equations are areasonably accurate representation of thespecific profiles at the site. If this is notthe case, then a site-specific analysis ofthe soil amplification (described next)should be done.

· If site specific response must beconsidered, and if the site has nearhorizontal stratification, the resultsprovided can be used to select designearthquakes for use with a SHAKE typeprogram to estimate surface motion andcyclic stresses in the foundationoverburden and embankment. The siteis represented by a one-dimensionalmodel giving the shear wave velocitiesof each layer down to bedrock (not justthe surface 30 metres), plus densities,and the depths of the layer interfaces. Ifconditions are appropriate and the shearwave velocity variation is known or canbe estimated, this is relatively easy todo.

· If frequency dependent site response isconsidered, and if the site hastopographic variations, or does not havenear horizontal sub-surfacestratification, the results provided canbe used to select design earthquakes foruse with a finite-element or finite-difference computer program toestimate surface motion. The site isrepresented by a two- or three-dimensional model giving thedistribution of sub-surface materials. Insome cases this may be a combined site-structure model, incorporating site-structure interaction. Two-dimensionalmodels require much data and complexcomputation, and three-dimensionalmodels require a great deal more. Thisis seldom done in practice as onedimensional models are usuallyadequate. More than one 1D model maybe used if conditions vary across a damsite.

· A time-consuming and generallyimpractical method in anything otherthan highly seismically active areasis to determine a site transferfunction empirically by installingseismographs at the site and nearbyon bedrock (either in a borehole oron a rock outcrop within a distanceof some kilometres). Comparison ofspectra from regional and distant

DRAFT



earthquakes will give frequencydependent site response for short andlong period motion respectively.

In the case of foundations on rock with afairly typical Vs30 of 760 m/s, there will berelatively little resonance or other frequencydependent site response. Inclusion of theVs30 term will account for some impedanceamplification.

C2.1.4. Earthquake Hazard in Australia

C2.1.4.1. Australian EarthquakesThe Australian continent is within a tectonicplate shared with most of India and some ofNew Zealand, and all of its earthquakes areintraplate. The plate boundaries to the northand east are among the most active on theEarth. Possibly as a result of this, Australiais one of the most active intraplate areas onthe Earth. Despite this, the hazard is quitelow when compared with active plateboundary areas.

Most people in Australia can expect to feelan earthquake about every five to ten years,although many of these will not berecognised as an earthquake. MostAustralian earthquakes that are reported aresmaller than magnitude 3.0 and are heardwith a noise like a distant quarry blast orthunder (due to coupling of P-waves in theground into sound waves in the air), oftenfollowed by a slight vibration.

Only a small proportion of the earthquakesthat are felt, perhaps about 5%, will causeany damage in their epicentral area. If theyoccur in an inhabited area, most earthquakesof magnitude 4.0 to 5.0 will cause someminor damage, while magnitudes greaterthan 5.0 may cause significant damage.

By contrast, in an active plate boundary arealike New Britain or Bougainville in PapuaNew Guinea, earthquakes are felt very often,on average every week or two. These are

normally felt rather than heard, with anysounds being the reaction of a building to thevibrations rather than the earthquake itself.A very small proportion of these felt PapuaNew Guinea earthquakes, perhaps about0.1%, will cause damage in their epicentralarea, and earthquakes smaller thanmagnitude 6.0 rarely cause damage.

There are a number of factors that influencethe estimation of earthquake hazard inAustralia. One is the short duration ofdocumented history, with a little over 200years of data about Sydney, andconsiderably less for most of the continent.

Another factor is the large area of thecountry relative to the size of the population.Seismographs are distributed relativelysparsely, limiting the accuracy of earthquakelocations and magnitudes.

As was the case over most of the Earth,seismograph coverage of local earthquakesin Australia was only established in about1960, following the InternationalGeophysical Year. While there should besome link between population andseismograph density with moreinstrumentation in the populated southeast,coverage is highly variable and still far fromcomplete. The Australian NationalSeismograph Network, operated byGeoscience Australia, aims to locate allearthquakes in Australia larger thanmagnitude ML 3.0. Local seismographnetworks have non-uniform coverage, oftenwith good coverage of large dams and poorcoverage of major cities.

Figure C2.8 shows the location of Australianearthquakes with magnitude exceeding ML4.0 in the period 1850 to 2014. Note thatmany earthquakes prior to 1960 were notlocated so are not shown.

DRAFT



Figure C2.8: Australian earthquakes exceeding magnitude 4.0 from 1850 to 2016

Table C2.2 lists some of the largest earthquakes which have occurred in Australia. It can be seenthat there are several in the range M6.5 to 7.0.

DRAFT



Table C2.2: Significant Australian earthquakes

Date Place Mag Imax Damage, Approximate A$(2016).

1892-01-26

Flinders Island,Tas

MP6.9

7+ Offshore epicentre, little damage. Felt throughTasmania and East Victoria.

1897-05-10

Beachport, SA MP6.7

8 Damage in Kingston, Robe, Beachport, majorliquefaction; felt from Ceduna to Melbourne.

1902-09-19

Warooka, SA MP6.0

8 At Warooka chimneys fell, walls partiallydemolished, few buildings without damage.

1903-07-14

Warrnambool,Vic

MP5.3

7 Extensive minor damage and liquefaction atWarrnambool. Followed similar event in April.

1918-06-06

Bundaberg, Qld MP6.0

6 Epicentre offshore, minor damage inRockhampton.

1941-04-29

Meeberrie, WA MS7.2

8 Isolated area. Damage to remote farm housesincluding cracked walls, burst tanks.

1954-02-28

Adelaide, SA ML5.4

8 Widespread minor damage, A$200m. Epicentralwas rural but is now urban.

1961-05-21

Robertson,NSW

ML5.6

7 Damage in the Moss Vale, Robertson and Bowralarea, A$10m.

1968-10-14

Meckering, WA MS6.8

9 Most buildings in Meckering destroyed, $90m.32 km surface rupture.

1969-10-14

Boolarra, Vic ML5.6

6 Cracked walls and fallen chimneys in theepicentral area.

1973-03-09

Burragorang,NSW

ML5.0

6 Minor damage in Picton, Bowral andWollongong, A$7m.

1979-06-02

Cadoux, WA MS6.2

9 Many buildings at Cadoux destroyed, $9m. Onlyone injury. 15 km surface rupture.

1986-03-30

Marryat Creek,SA

MS5.9

8 Epicentre in remote area. Cracked walls innearest homestead, 13 km surface rupture.

1988-01-22

Tennant Creek,NT

MS6.8

9 Epicentre in remote area. Damage A$4m, mainlyto damaged gas pipeline, 35 km surface rupture.

1989-05-28

Uluru, NT mb5.7

6 Minor damage at Uluru National Park (AyersRock). Epicentre west of Mt Olga.

1989-12-27

Newcastle,NSW

ML5.6

8 Thirteen fatalities, $4500m plus damage.Widespread minor to moderate damage.

C2.1.4.2. Mechanism of AustralianEarthquakesAlmost all Australian earthquakes havemechanisms with the maximum principalstress near horizontal. Most have theminimum principal stress near vertical,producing reverse or thrust faults. There maybe some strike-slip movement, when the

failure is on a fault that is oriented at anangle other than 90° to the principal stressdirection.

Compression giving reverse and thrust faultsproduces surface uplift, so these earthquakesare most likely to occur in areas wheremountains are developing (such as the

DRAFT



eastern Highlands or Flinders Ranges), andless likely under sedimentary basins (such asthe Murray-Darling Basin or the GreatArtesian Basin).

There are a number of active faults uponwhich several episodes of movement havebeen proven in the last 100,000 years. Theseinclude the Edgar fault; Tasmania, Hydenfault, Western Australia; Wilkatana andRoopena, South Australia; and Cadell,Victoria/New South Wales (Clark et al,2012).

Large Australian earthquakes (e.g. TennantCreek, 1988), have occurred on faults hadnot been previously been recognised andmapped. Crone et al. (1997) describeevidence that there had been no movementon the fault at Tennant Creek for 200,000years.

Based on the available information, it wouldappear that differences between earthquakeground motion s in non-cratonic regions ofAustralia and those from reverse faultearthquakes in USA, China, etc., data fromwhich are included in some empiricalground motion prediction models are based,are not sufficiently great as to invalidate theuse of those models in Australia. This is anaspect that will need to be further assessedas more ground motion data from Australianearthquakes is gathered.

More detailed design methods are basedupon seismic response analyses to actualground motions rather than empiricalmethods so this is not an issue providedthere is a careful selection of ground motionhistories to represent the earthquakeconditions in Australia.

C2.1.4.3. Earthquake DepthsUsually, earthquake depths can be preciselydetermined from local networks only if thedistance to the nearest seismograph is notgreater than about twice the earthquakedepth. The depths of most Australianearthquakes are poorly constrained, because

of their relatively shallow depths and the lowdensity of seismographs.

If it is not possible to constrain anearthquake depth using seismograph data orother observations (e.g. a magnitude 1.0earthquake that is felt must be within acouple of kilometres of the surface), anarbitrary “normal” depth may be used.Typical values used in Australia include 0, 5,10 or 33 kilometres. These may or may notbe realistic. Some observatories may select adepth from a range of standard values,depending on the character of the recordedwaveforms (e.g. a small magnitudeearthquake with large surface waves must beshallow).

Eastern Australian earthquakes are usually atdepths between 1 and 20 km, while those inCentral Australia may be a little deeper, andthose in Western Australia may be a littleshallower. Australia’s deepest knownearthquake occurred at 39 km depth offshorefrom Arnhem Land in 1992 (McCue andMichael-Leiba, 1993).

In Eastern Australia, earthquakes at depthsof less than 5 km are regarded as shallow,and greater than 15 km as deep. TheNewcastle earthquake of December 1989was at a depth of about 12 km.

All of these Australian earthquakes are veryshallow compared with those in plateboundary areas, where subductionearthquakes can occur as deep as 700 km,and depths of less than 70 km are regardedas shallow, and greater than 300 km as deep.

It seems that small earthquakes tend to occurmore often at shallow depths, whilemoderate to large earthquakes rupture atgreater depths (Allen et.al, 2004). Many ofthe larger Australian earthquakes haveoccurred in the cratonic regions and haverupture through to the surface due to thehigh strength of the surface rocks, giving asurface fault rupture.

Shallow earthquakes often have manyaftershocks, some with magnitudes

DRAFT



approaching that of the main shock. Deepearthquakes usually have few aftershocks,and these are usually no larger than onemagnitude unit below the main shockmagnitude.

Because of their shallow depth, smallAustralian earthquakes are often heard orfelt. Magnitude 1.0 events can be felt to adistance of about 1 km, and magnitude 2.0 toabout four kilometres. These are slantdistances, so only very shallow events ofthese magnitudes will be felt.

Because the short travel distances fromshallow earthquakes do not give muchattenuation of high frequency vibrations, andif a relatively high stress drop gives a highproportion of high frequency vibration fromthe earthquake source, these events haveenough energy in the audio range to allowthem to be heard. For many such smallearthquakes the sound heard is moresignificant to an observer than the vibrationsfelt.

For similar reasons, moderate magnitudeearthquakes in Australia produce motionwith strong peak ground accelerations, andcan cause significant damage. TheNewcastle earthquake was only ofmagnitude ML 5.6, but caused extensivedamage. However much of this occurred inareas where the significant depth of alluviumoverlying bedrock amplified the groundmotions.

In summary, Australian earthquakes are:

Intraplate and continental, so they areinfrequent

· Most people feel earthquakes just a fewtimes in their lifetime

· However, Australia is one of the mostactive intraplate areas

Distributed over many small to moderatefaults

· Fault lengths rarely exceed 100 km,

· Low maximum magnitude, perhaps Mw7.5

· Hazard is quite widely distributed

Shallow, from surface to 20 km

· Small events often felt and heard withina very limited area

· Moderate magnitudes can cause damage· Above Mw 6 usually gives surface

rupture, especially in cratonic regions

Dominant horizontal compression

· Reverse faults predominate, strongand relatively inactive, so stresslevels are high

· Earthquakes may have high stressdrops, giving high frequency, highacceleration, short duration motion

C2.1.4.4. Australian Earthquake HazardMapsAs presented in ANCOLD (1998)earthquake hazard maps for Australia havechanged considerably over the years as moreearthquakes are recorded in an increasingseismograph recording network, and asearthquakes such as Tennant Creek haveoccurred in what was thought to be a lowseismic hazard area.

In practice these hazard maps are of limiteduse for dams because they only consider the10% chance of exceedance in 50 years, or 1in 475 / annum hazard which is too frequentfor design ground motions for most dams.

As a result site specific hazard studies aregenerally required for dams.

C2.1.5. Reservoir Triggered EarthquakeReservoirs may induce seismicity by twomechanisms. Either the weight of the watermay change the stress field under thereservoir, or increased ground water porepressure may decrease the stress required tocause an earthquake. In either case, reservoirtriggered earthquakes (RTE) will only occurif relatively high stresses already exist in thearea. If a recent large earthquake hasrelieved the stress, perhaps in the last few

DRAFT



hundred years for low seismicity areas likeAustralia, then RTEs are unlikely to occur.

RTE events initially occur at shallow depthunder or immediately alongside a reservoir.As years pass after first filling, andgroundwater pore pressure increasespermeate to greater depths and distances, theevents may occur further from the reservoir.This occurs at a rate of something like onekilometre per year. RTE’s are experiencedunder new reservoirs, usually starting withina few months or years of commencement offilling, and usually not lasting for more thanabout twenty years. Once the stress field andthe pore pressure field under a reservoir havestabilised, then the probability of futureearthquakes reverts to values similar to thoseestimated if the reservoir had not beenproduced. Most of the earthquake energydoes not come from the reservoir, but fromthe normal tectonic processes. The reservoirsimply acts as a trigger.

If there is a major fault near the reservoir,RTE events can exceed magnitude 6.0 (e.g.Xinfengjiang, China, 1962, M 6.1; Koyna,India, 1967 M 6.3). Such events will onlyoccur if the fault is already under high stress.Several Australian reservoirs may havetriggered earthquakes exceeding magnitude5.0 (e.g. Eucumbene, 1959, M 5.0;Warragamba, 1973, M 5.0; Thomson, 1996,M 5.2), and others may have triggeredsmaller earthquakes.

It is more common for a reservoir to triggera large number of small shallowearthquakes, especially if the underlyingrock consists of jointed crystalline rock likegranite (e.g. Talbingo 1973 to 1975;Thomson 1986 to 1995). These eventspossibly occur on joints rather thanestablished faults, so are limited in size, andonly give magnitudes up to 3 or 4. Theirshallow depth means that even eventssmaller than magnitude 1.0 may be felt orheard.

RTEs have been observed for over onehundred reservoirs throughout the world,

and small shallow induced events haveprobably occurred under many others. Arelatively high proportion of reservoirs withRTE seismograph networks do record suchactivity. A high proportion of RTE examplesoccur in intraplate areas, with above averagerates in China, Australia, Africa and India.

It is not easy to predict whether a particularnew reservoir will experience RTE becausethe stress and strength at earthquake depthscannot normally be measured. For the samereason, prediction of normal tectonicearthquakes has been unsuccessful in mostparts of the world.

It seems that RTE with many small events ismore likely to occur in intraplate areas withnear-surface crystalline rocks like granite,rather than sedimentary rocks. A largermagnitude RTE event can only occur if thereis an existing fault of sufficient dimensionthat is late in its earthquake cycle, with thestress already approaching the strength ofthe fault.

ICOLD (1983, 1989, 2012), conclude thatthere is documented evidence to prove thatimpounding of a reservoir sometimes resultsin an increase of earthquake activity at ornear the reservoir. ICOLD (1983) concludethat:

– Earthquakes of magnitude 5 to 6.5were induced in 11 of 64 recordedevents.

– The greatest seismic events have beenassociated with very large reservoirs(but there is insufficient data to showany definite correlation betweenreservoir size and depth and seismicactivity).

– In view of the above, a study ofpossible induced seismic activityshould be made at least in caseswhere the reservoir exceeds 5x108m3

in volume, or 100 m in depth.– The load of the reservoir is not the

significant factor; rather it is theincreased pore water pressure in

DRAFT



faults, leading to a reduction in shearstrength over already stressed faults.

ICOLD (2012) indicate that there are so faronly six generally accepted cases ofreservoir triggered seismicity where themagnitude of the event exceeded 5.7. Thelargest recorded magnitude event that isbelieved to be due to a reservoir-triggeredevent is 6.3.

C2.2TerminologyActive fault. The definition adopted for anactive fault is adapted from ICOLD (2016a).

ICOLD (2016a) include “or very long faultswhich have moved repeatedly in Quaternarytime (1.8 million years). For this guidelinethis is not included as such very infrequentactivity would be too conservative for use inthe deterministic analysis approach. Thesefaults are included in PHSA as neotectonicfaults.

The United States Nuclear RegulatoryCommission (USNRC) define a capablefault as a fault which has exhibited one ormore of the following characteristics:

(a) Movement at or near the ground surfaceat least once within the past 35,000 yearsor movement of a recurring nature withinthe past 500,000 years.

(b) Macro-seismicity instrumentallydetermined with records of sufficientprecision to demonstrate a directrelationship with the fault.

(c) A structural relationship to an activefault according to characteristics (a) or(b) that movement on one could bereasonably expected to be accompaniedby movement on the other.

This is broadly consistent with the ICOLD(2016) definition. Active faults are includedin the Probabilistic Seismic HazardAssessment (PSHA) and in assessing theMCE for Standards based design.

Neotectonic fault. The definition is takenfrom Clark et al (2012). Such faults are

assumed to be potentially active in assessingthe PSHA. They represent faults which haveruptured in the last 5 to 10 million years.

C2.3 General Principles for Selection ofDesign Ground Motion and AnalysisMethodThere are two ways to select the designground motion. These are related to theanalysis method. They are:

(a) Deterministic Analysis. Design groundmotions at the dam site are defined inprobabilistic terms; and where there areactive faults in the vicinity of the dam,also by the ground motion resulting fromthe MCE on the faults.

(b) Risk based Analysis. A risk basedapproach requires assessment of theseismic ground motions at the dam siteup to and beyond the SEE for use in riskanalyses.

The steps involved in the risk analysisprocess are:

(i) Determine the AEP of earthquakeground motion (PE) over the range ofearthquake events which may affectthe dam. Table C2.3 gives anexample for AEP vs groundacceleration.

(ii) Determine the conditional probability(PBC) that for each of the groundmotion ranges (e.g. 0.125 g to 0.175g in Table C2.3) the dam will failresulting in uncontrolled release ofthe reservoir. In assessing thisconditional probability all modes offailure should be considered and theprobabilities combined, makingallowance for interdependence andmutual exclusivity or otherwise (e.g.for embankment dams, slopeinstability, piping,liquefaction/instability, and forconcrete gravity dams, overturning,and sliding).

(iii)Assess the probability of failure foreach range of ground motion by

DRAFT



multiplying the AEP with PBC i.e. PB= PE x PBC – see Table C2.3.

(iv)Sum the probabilities to give theoverall annual probability of failuredue to earthquake.

(v) Assess whether the resulting risks aretolerable.

Table C2.3. Example of assessing the probability of failure by earthquake.Acceleration Annual

Probability (PE)Conditional(1)

Probability(PBC)

PB(2)

<0.075g 0.874 0.0005 0.00040.075g to 0.125g 0.100 0.005 0.00050.125g to 0.175g 0.015 0.05 0.00070.175g to 0.225g 0.007 0.1 0.00070.225g to 0.3g 0.003 0.3 0.0009>0.3g 0.001 0.5 0.0005Total 1.000 0.0037(1) Given the earthquake occurs(2) PB = PE x PBC

Preferred method.

The risk based approach is preferred forassessment of existing dams for thefollowing reasons:

1. In the deterministic approach it isdifficult to account for the fact that theconditional probability of failure of thedam given the SEE is not 1, and may bevery different for different types of damsand foundation conditions. Hence eitherthese conditional probabilities have to beestimated, or it has to be understood thatthe different dams have quite differentannual probabilities of failure, eventhough they are assessed for the sameSEE.

2. Deterministic methods potentially maskthe fact that many dams have somelikelihood of failure at ground motionsless than the SEE.

3. Some design methods are by their natureprobabilistic; e.g. liquefactionassessments, and hence inherently riskbased.

4. Risk based methods are commonly usedfor Portfolio Risk Assessments (PRA) inAustralia, and if deterministic methodsare used they may result in a dam

passing the deterministic method but notsatisfying the PRA tolerable risk criteria.

C2.4 Description of a ProbabilisticSeismic Hazard Assessment

C2.4.1 Inputs into PSHA.

(a) Earthquake Forecast (EQF)

The PSHA uses an earthquake forecast,which predicts the locations,magnitudes, and frequencies ofoccurrence of all earthquakes thatcontribute seismic hazard at the site.The earthquake forecast is based onearthquake source models, which are oftwo kinds: distributed earthquakesources, and fault sources. To accountfor epistemic uncertainty in earthquakesource models, alternative viableearthquake source models should beused in a logic tree framework, withweights given to each alternative model.

(b) Ground Motion Prediction Equations(GMPE)

Ground motion prediction equationsprovide estimates of the ground motionsat a site having specified site conditions

DRAFT



located at a specified distance from anearthquake of specified magnitude andother source properties (such as style offaulting). They specify the groundmotion by peak ground acceleration andresponse spectral acceleration (5%damping).

In the past, site conditions wereparameterized in GMPE’s as a genericgeological category such as rock or soil,but most current GMPE’s typically nowuse the time-averaged shear-wavevelocity in the upper 30 metres of thebedrock profile (Vs30) that underlies thesite.

In cases where soil overlies bedrock,especially where the soils are expectedto have significantly non-linearbehavior, non-linear site responseanalysis should be carried out usingground motions estimated at an“engineering bedrock” level below thesoil profile by the PSHA. These analysesare carried out by the dam Consultant.

To account for epistemic uncertainty inearthquake ground motion models,alternative viable earthquake sourcemodels should be used in a logic treeframework, with weights given to eachalternative model.

C2.4.2 The PSHA Process

(a) Method

PSHA is based on methodologyoriginally proposed by Cornell (1968),outlined schematically in Figure C2.9.If seismicity is considered to follow arandom Poisson process, then theprobability that a ground motion, such asSpectral Acceleration (SA) exceeds acertain value (s) in a time period t isgiven by:

Equation 1

where is the annual mean numberof events (also known as “annualprobability of exceedance”) in which the

ground motion parameter of interestexceeds the value ‘s’.Some prefer to report this in terms ofcomputing s for a certain probability ofoccurrence, P, in a time period t. Forexample a ground motion level having a10 4- / annum probability of exceedancemay be expressed as equivalent to areturn period of 10,000 years.

The annual probability of exceedance iscalculated as follows:

Equation 2where:

f(mi) = probabilitydensity function for events of magnitudemi

(from the EQF).

P(SA>s|m,r) = probability thatSA exceeds s for a given magnitude mand

distance r (from the GMPE).

P(r|m) = probability thatthe source to sitedistance is r,given a

source ofmagnitude m(from the EQF).

The probabilistic analysis is performedusing a computer program of the typedescribed further below. The computerprogram should allow the treatment oftwo types of uncertainty: epistemicuncertainty and aleatory variability.(b) Treatment of Random Variability within

the PSHA Hazard Integral

Aleatory variability (randomness) resultsfrom randomness in natural physicalprocesses that coexist in nature. Thesize, location, and occurrence time of thenext earthquake on a fault, and the site-

P(SA > s) = 1 - e-f (s )t

f (s)

f(s) = ( f (m)(P(SA > s | m,r)P(r | m)dmdr)im,ròò

i =1

Faults

å

DRAFT



to-site random variability in the groundmotion level at a given distance from agiven earthquake (shown in Box 4 ofFigure C2.9) are examples of quantitiesconsidered aleatory. In current practice,these quantities cannot be predicted,

even with the collection of additionaldata. Integration over aleatoryvariabilities is carried out within thehazard curve (Equation 2) to yield asingle hazard curve.

Figure C2.9 Schematic diagram of probabilistic seismic hazard analysis. The attenuationequations shown in Box 4 are also called ground motion prediction equations.(Source:Paul Somerville)

(c) Treatment of Epistemic Uncertaintyoutside the PSHA Hazard Integral

Epistemic uncertainty results fromimperfect knowledge about the processof earthquake generation (e.g. in

DRAFT



alternative viable earthquake forecastmodels) and the assessment of theireffects (e.g. in alternative viable groundmotion prediction models). Viablealternatives are mutually exclusive, andare treated using logic trees outside thePSHA hazard integral. Epistemicuncertainties are thus expressed byincorporating multiple hypotheses,models, or parameter values. These

multiple interpretations are eachassigned a weight within the logic treeframework, resulting in a suite of hazardcurves and their associated weights.Examples of logic trees for distributedearthquake sources and fault sources areshown in the upper and lower parts ofFigure C2.10.

Figure C2.10. Example logic trees for the treatment of epistemic uncertainty in distributedearthquake sources (top) and fault sources (bottom) in PSHA. (Source: Paul Somerville)

DRAFT



C2.4.3 Products of PSHA and theiranalysis.

(a) Hazard Curves

The PSHA estimates the ground motionlevel as a function of annual probability ofexceedance, or return period. Figure

C2.11 shows peak ground accelerationhazard curves for a group of distributedearthquake sources and fault sources,shown by the colored lines, with the blackcurve showing the combined total hazard.

Figure C2.11 Example hazard curve for a group of distributed earthquake sources and faultsources, with the uppermost black curve showing the combined total hazard. (Source: PaulSomerville)

(b) Uniform Hazard Response Spectra(UHS)

Uniform hazard curves for a set of returnperiods are obtained from the hazard

curves for each of a set of groundmotion periods, including PGA, which isequivalent to the zero period responsespectral acceleration (Figure C2.12).

DRAFT



Figure C2.12. Equal hazard response spectra for a range of return periods.

(c) Fractiles of the Uniform HazardResponse Spectrum (UHS)

There is epistemic uncertainty in the truevalue of the seismic hazard results dueto uncertainty in the true state of nature.In the example shown in Figure C2.13,three distributed earthquake sourcemodels (AUS5, RF and GA) were usedbecause of uncertainty about the degreeto which each is correct. Similarly, sixground motion models were usedbecause of uncertainty about the degreeto which each is correct. This kind ofuncertainty is modeled using logic trees,with the alternative branches(corresponding to the alternativechoices) being given weights. Theseismic hazard is calculated for each ofthese branches, in this case, for each ofthe eighteen combinations of distributedearthquake source model and groundmotion model. As a result, eighteen

seismic hazard curves are obtained. Theusual practice in seismic hazard analysisis to use the mean of these hazardcurves, to ensure that large hazardvalues are given appropriate weight.

The fractiles of these multiple hazardcurves are used to quantify the epistemicuncertainty in the true value of thehazard. Figure C2.13 shows the 95th,85th, 50th (i.e. median), 15th, and 5th

percentiles of the hazard for a givenreturn period, as well as the meanhazard. The mean hazard is generallyslightly higher than the median. Thefractiles represent the degree of certaintythat the true value of the hazard does notexceed the value given by the fractile.For example, there is a 50% chance thatthe true value of the hazard exceeds themedian (50th fractile) value, and a 15%chance that it exceeds the 85th fractile.

DRAFT



Figure C2.13. Fractiles of the hazard for a return period of 10,000 years (10 4- /annum)

(d) De-aggregation of the UHS

The probabilistic UHS obtained throughPSHA takes account of all of theearthquake scenarios that could affectthe site, and describes the ground motionlevel that corresponds to a specifiedreturn period or annual probability ofexceedance. The UHS represents thecontributions from many differentearthquakes, perhaps including smallnearby earthquakes and more distantlarger earthquakes, and may not be arealistic representation of the responsespectrum of any individual scenarioearthquake. For this reason, if it isdesired to use ground motion time-histories to represent the responsespectrum, it is then necessary to identifyone or more scenario earthquakes thatdominate the hazard for that returnperiod in the ground motion periodrange of importance for the structure.This process, termed de-aggregation ofthe UHS, results in one or more

earthquake scenarios, each having aspecified magnitude, distance, andseverity (described by the parameterepsilon).

The de-aggregation of the hazard varieswith the return period and the groundmotion period of interest, as shown inFigure C2.14. For short return periods,the hazard tends to be dominated bysmaller, more distant earthquakes (toprow of Figure C2.14) while for longreturn periods the hazard tends to bedominated by larger, more nearbyearthquakes (bottom row). For shortground motion periods, the hazard tendsto be dominated by smaller earthquakes(left side of Figure C2.14) while for longground motion periods the hazard tendsto be dominated by larger earthquakes(right side). Selection of time-historiestherefore needs to take account of thereturn period of the UHS and the groundmotion period of the structure that is tobe analysed.

DRAFT



Figure C2.14. De-aggregation of the hazard for peak acceleration (left) and 1 s responsespectral acceleration (right) for return periods of 500 years (top) and 10,000 years(bottom).

The de-aggregation of the UHS is usedwhenever it is necessary to represent thedesign response spectrum by a singleearthquake. This is done wheneverdynamic analyses of a structure areperformed using a ground motion time-history.

We next consider the suitability of theUHS as a representation of the responsespectrum of a single earthquake time-history. We show that there are twosources of over conservatism in the useof the UHS as a representation of asingle earthquake time-history. In bothcases, the UHS is too “broadband” (it islarge over a range of periods that isunrealistically broad), rendering it sub-optimal for use as a target for scaling or

spectral matching of time-histories torepresent the design ground motions.

(e) Scenario Spectrum in Place of the UHS

As discussed above the UHS representsthe contributions from many differentearthquakes, perhaps including smallnearby earthquakes and more distantlarger earthquakes, and may not be arealistic representation of the responsespectrum of any individual scenarioearthquake. For example, small nearbyearthquakes are expected to haverelatively high ground motion levels atshort periods and relatively low groundmotion levels at long periods.Conversely, large distant earthquakes areexpected to have relatively low ground

DRAFT



motion levels at short periods andrelatively high ground motion levels atlong periods.

Instead, it is preferable to use groundmotion time-histories that are scaled orspectrally matched to a set of scenariospectra representing a small set ofmagnitude and distance combinationsthat dominate de-aggregation of theUHS. These scenario spectra are in turnscaled so that they are enveloped by theUHS and collectively represent thatbroadband spectrum. One scenario froma small nearby earthquake may representthe short period part of the UHS, whileanother scenario from a larger moredistant earthquake may represent thelong period part of the UHS.

(f) Conditional Mean Spectrum in Place ofthe UHS

There is another source of overconservatism in the UHS that maybecome important when the scaledscenario spectrum lies considerablyabove the median level for thatearthquake scenario. Baker (2011)showed that the UHS conservatively

implies that large-amplitude spectralvalues will occur at all periods within asingle ground motion time-history. Analternative, termed a Conditional MeanSpectrum (CMS), provides the expected(i.e., mean) response spectrum,conditioned on the occurrence of a targetspectral acceleration value at the periodof interest (Baker and Cornell, 2006;Baker, 2011). Baker (2011) shows thisspectrum to be the appropriate targetresponse spectrum for the goal describedabove, and it is thus a more appropriatetarget for scaling and spectrally matchingground motion time-histories as input todynamic analyses. Baker (2011)demonstrates that the CMS spectrummaintains the probabilistic rigour ofPSHA, so that consistency is achievedbetween the PSHA and the groundmotion selection. This enablesquantitative statements to be made aboutthe probability of observing the structuralresponse levels obtained from dynamicanalyses that use this spectrum; incontrast, the UHS does not allow forsuch statements (Baker, 2011). FigureC2.15 shows an example.

DRAFT



Figure C2.15 Left: Median (solid line) and median plus two standard deviations(dashed line) design spectrum for a magnitude 7 earthquake at a closest distance of 12km, shown with recorded response spectra (green), one of which (blue) is close to thedesign spectrum at a period of 1 second. Right: Derivation of the conditional meanspectrum for a period of 1 second from a scenario spectrum. The conditional mean isderived from the correlation between response spectral values at adjacent periods.(Baker, 2010).

(g) Vertical Response Spectrum.

In dam engineering, the horizontalcomponents of ground motions usuallydominate the design. Accordingly, thevertical component of ground motion thatis used in design should be compatiblewith the horizontal components; it shouldrepresent the characteristics of thevertical component that are expected tooccur in conjunction with the horizontalcomponents. Therefore, the verticalcomponent is not derived from the PSHA– that would represent earthquakemagnitudes and distances that aregenerally quite different from those of thehorizontal component. Instead, thevertical component is derived from thede-aggregation of the horizontalcomponent so that the vertical responsespectrum represents the samecombination of earthquake magnitudeand distance as does the dominantcontribution to the horizontal spectrum.This is accomplished by calculating thehorizontal response spectrum using thede-aggregated magnitude and distance

from the PSHA, and then scaling thatspectrum by the vertical/horizontal ratioof a model such as Gulerce andAbrahamson (2011).

C2.5 Requirements of a Seismic HazardAssessment

1. Distributed Earthquake source models.

In current earthquake source models forAustralia, distributed seismicity ismodelled in quite different ways that giverise to significant epistemic uncertainty.Distributed seismicity is represented bydiscrete seismic source zones (Brown andGibson, 2004); by spatially smoothedseismicity (Somerville et al., 2009), andby a layered seismicity approach(Burbidge and Leonard, 2011), asdescribed in more detail below. It isnecessary to include these alternativesource models in the seismic hazardanalysis in order to account for thisepistemic uncertainty. None of thesethree source models contains any faultsources; these need to be treated

DRAFT



separately.

ES&S AUS5 Seismic Source Zone Model

The Environmental Systems and Services(ES&S) AUS5 source zone modeloriginally developed by Brown andGibson (2004) uses geological andgeophysical criteria in combination withhistorical seismicity to identify zones ofuniform seismic potential, and then useshistorical seismicity to characterise theseismic potential of each zone by meansof the a-values and b-values of theGutenberg-Richter earthquake recurrencemodel, together with an estimate of themaximum magnitude of earthquakes ineach zone. This approach has theadvantage of allowing for theincorporation of geological andgeophysical information as well asseismicity data in the identification ofseismic source zones.

Risk Frontiers Spatially SmoothedSeismicity Model

Judgment is required in defining thesource zone boundaries of models such asBrown and Gibson (2004), and it isunclear what would cause abrupt changesin seismicity levels across source zoneboundaries. These considerationsmotivate the use of spatially smoothedhistorical seismicity to define theearthquake forecast, developed by RiskFrontiers (Hall et al., 2007). Thisapproach gives a spatially continuoussource model without boundaries exceptin b-value. The spatial smoothingapproach has the advantages of simplicityand of avoiding uncertainty in thegeological definitions of zones, but hasthe disadvantage of not making use ofpotentially informative geological data.

GA Layered Seismicity Model

Leonard et al. (2012) developed anearthquake source model for Australiathat is based entirely on historicalseismicity. They identify a set of back

ground zones, a set of regional zoneshaving higher seismic activity rates thanthe background zones, and hotspots thatcontain concentrations of earthquakeactivity within regional source zones. The2012 Australian Seismic Hazard Map(Burbidge and Leonard, 2011) iscomposed of three layers of seismicity.

• Background Source Zones. There aretwo of these, one for the cratonicregion of the western part ofAustralia, and another for the non-cratonic region of the eastern part ofAustralia. For each region there aretwo models: one that reflects theactual data and another that uses aminimum value (floor).

• Regional Source Zones. Theserepresent the long-term earthquakeactivity above the background level inspecified broad regions

• Hotspot Source Zones. Theserepresent the short-term earthquakeactivity, including earthquake swarmsand clusters, above the backgroundand regional source zones, that havebeen occurring recently in local zones

Burbidge and Leonard (2011) propose anumber of different ways in which thehazard maps from each layer can becombined. One way is to use the highesthazard value from among the threedifferent layers at each site. In thisapproach, the total integrated seismicmoment across Australia is notconserved, but the hazard in the areas oflow seismic activity is allowed toincrease without decreasing the hazard inareas of high seismic activity. Thisapproach may be justifiable forapplication to the new building codeprovisions (AS1170.4), but it is notsuitable for site-specific studies such asthose for dams, where rigorousprobabilistic analysis of hazard and riskis required. In applications to dams, it ispreferable to use the version of theBackground Source Zones that does not

DRAFT



contain a floor (minimum value),together with the Regional Source Zones.If any of the Hotspot Source Zonescontributes significantly to the seismichazard at the site, they should also beincluded as source zones.

The three distributed earthquake sourcemodels all specify the maximum earthquakemagnitudes in their recurrence models. TheES&S and RF models both assume amaximum earthquake magnitude of 7.5. The

GA model has maximum magnitudes rangingfrom 7.3 to 7.7.

As shown in Figures C2.16 and C2.17 themodels can give significantly differentseismic hazard in some areas in Australia,and there is no systematic outcome, with oneor the other model giving higher hazard. Inview of this any seismic hazard assessmentwhich uses only one source model ispotentially unreliable.

Figure C2.16. Example of differences in event rates for seismic hazard assessment from AUS5discrete seismic source zones model and Risk Frontiers spatially smoothed seismic source model.DRAFT



Figure C2.17 Seismic Response Spectra using the AUS5 and GA discrete seismic source zonemodels and the Risk Frontiers spatially smoothed seismic source model, and their average.

2. Fault sources.

(a) Active and Neotectonic Faulting inAustralia

Fault sources here means active faultsand neotectonic faults as defined inSection 2.2.

The long term uplift and subsidencecaused by active faulting in geologicaltime has a significant effect on drainagepatterns in many parts of Australia, withthe result that dams are frequently locatedin close proximity to active orneotectonic faults.

An Australia-wide assessment of faultingwas made by Clark et al. (2011; 2012).They analysed a catalogue of over 200neotectonic features, 47 of which areassociated with named fault scarps. Thedata were derived from analysis ofDEMs, aerial photos, satellite imagery,geological maps and consultation withstate survey geologists and a range ofother geoscientists. Verifying the featuresas active as defined for neotectonic faultsis an ongoing process. The catalogue

varies in completeness because samplingis biased by the available data bases, theextent of unconsolidated sedimentarycover, and the relative rates of landscapeand tectonic processes.

Following Clark et al. (2011) andSandiford et al. (2004), Seismologistsconsider neotectonic faults as potentiallyactive if they have undergonedisplacement under the current stressregime in Australia, and hence may havethe potential for displacement in thefuture. The age of the current stressregime in Australia is estimated to lie inthe range of 5 to 10 million years(Sandiford et al., 2004).

In Australia, geological maps typicallyshow numerous faults but do not indicatewhether they are active in the currentstress regime; most of them are probablynot. For example, if these faults werepreviously active under a different stressorientation, it is possible that they areunfavorably oriented to undergo slipunder the current stress orientation.However, if they are favourably oriented,then consideration should be given to the

Example Dam – Equal Hazard Spectra – 5% Damping

DRAFT



possibility that they have been reactivatedunder the current stress regime.

Clark et al (2011) list a number of faultswhich have been investigated and foundto have had multiple surface ruptureevents over intervals of hundreds ofthousands of years, with intervals of afew tens of thousands of year betweenevent. In some cases these faults undergoa long period of quiescence, and maycurrently be in a quiescent phase.However, others may have had a large

event within the past 35,000 years or mayhave the potential to generate a largeearthquake within a 35,000 year timeframe, and so may fit the definition ofactive fault that these guidelines hasadopted. Figure C2.18 summarizes datafor some of these faults which haveresulted in surface rupture. Others whichdid not result in surface rupture are listedwithin the text of Clark et al (2012).

Figure C2.18 Compilation of surface breaking earthquake recurrence data. (Clark et al. 2012)

DRAFT



(b) Alternative Approaches toIncorporating Active and NeotectonicFaults

In view of the short time span of thehistorical earthquake catalogue inAustralia, it is unclear how best toincorporate active and neotectonic faultsinto the earthquake forecast.

Brown and Gibson (2000) proposed amethod which subtracts fault-relatedseismicity from the area source zone inwhich the fault occurs, and insert a faultsource having that seismicity, using aGutenberg-Richter recurrence model.This approach assumes that the faultseismicity is represented in thebackground seismicity of the area source.

An alternative approach (Somerville et al,2008) is to add a fault source whoseseismicity is based on slip rate, withoutmodifying the background seismicity ofthe area source, using a characteristicearthquake recurrence model. Thisapproach assumes that the fault

seismicity is not represented in thedistributed source zones, consistent withthe Characteristic earthquake recurrencemodel, which is described next. Thisapproach requires an estimate of the sliprate of the fault. The episodic nature oflarge earthquake ocurrence, describedbelow, necessitates the carefulconsideration of the time period overwhich the slip rate of the fault should beestimated.

(c) Alternative Earthquake Recurrencemodels

The distribution of earthquakemagnitudes in distributed earthquakesource zones is usually assumed to followthe Gutenberg-Richter model (FigureC2.19, top left). However, thedistribution of earthquake magnitudes ondiscrete active and neotectonic faults maybe better represented by the characteristicrecurrence model (Schwartz andCoppersmith, 1984), in which most of thefault slip is taken up in large earthquakes(Figure C2.19, bottom left and right).

Figure C2.19. Left: Interval recurrence for the Gutenberg-Richter (top) andCharacteristic earthquake recurrence models (bottom). (Wesnousky et al., 1983). Right:Cumulative recurrence for the Characteristic earthquake recurrence model. Source:Schwartz and Coppersmith, 1984.

DRAFT



If the characteristic recurrence modelapplies, then the recurrence rate of largeearthquakes may be underestimated bythe Gutenberg-Richter model that isderived from historical seismicity if itonly contains small earthquakes, asindicated on the right side of FigureC2.19. For active faults, it is preferableto estimate the recurrence rate of largeearthquakes from geological data, such asfault slip rates, rather than historicalseismicity, because it is unclear whetherthe earthquake recurrence of fault sourcesfollows the Gutenberg-Richter model.

(d) Alternative Recurrence Behaviour ofActive Faults

Clark and Van Dissen (2006) and Clarket al. (2012) found that earthquakeactivity on faults in Australia is episodic,with clusters of earthquakes on a givenfault occurring close together in time(several tens of thousands of years),separated by longer periods (severalhundreds of thousands of years) of nolarge earthquake activity. This isinconsistent with the random temporal(Poisson) distribution of earthquakes thatis usually assumed in seismic hazardanalysis. Using the results of Clark et al.(2006), it may be possible to identifywhich faults are currently in an activephase and which are currently in aninactive phase. This could then be appliedto the evaluation of the seismic potentialof active faults in seismic hazardevaluations.

Clark et al. (2011) reviewed knowledgepertaining to the seismogenic deformation ofthe Australian continent over the last 5-10 Ma(the Neotectonic Era). Based upon perceiveddifferences in character of the seismogenicfaults across the continent, and guided byvariations in the geologic and geophysicalmakeup of the crust, they propose six onshoreneotectonic domains. A seventh offshoredomain was defined based upon analogy withthe eastern United States. These domains are

characterised by different earthquakerecurrence behaviour.

In practice unless there is definitiveinformation to the contrary, faults identifiedas active as defined in Section 2.2 should beconsidered still active. If there are differencesof view from Seismologists as to whether afault is still active this can be managed byassigning a conditional probability to theoccurrence of the earthquake motion toreflect these opinions.

(e) Modelling the Maximum CredibleGround Motion at the Dam site from anActive Fault

Where an active fault has been identifiedwhich could result in significant groundmotions at the dam site its contribution toground motion will have been considered inthe PSHA as described above.

As detailed in Table 2.1 of Section 2.6, fordeterministic seismic hazard for ExtremeConsequence Category dams the maximumcredible ground motions which might occurat the dam site from rupture of active faultsare required

There are however uncertainties in doing thisincluding whether the whole fault or only partruptures: the magnitudes of the resultingearthquakes which might occur; the locationof the focus of the earthquake; the groundmotion models to be adopted. These shouldbe discussed with the Seismologist so theOwner and Consultant understand thepotential degree of conservatism inherent inthe ground motion assessment.

It should be noted that in cases where theactive fault is close to the dam site theseground motion estimates are likely to begreater than obtained from PSHA. This isbecause the PSHA will have assigned anannual probability to the occurrence ofearthquakes on the fault while the processabove takes no account of this. The processcan be considered as equivalent to ProbableMaximum Flood estimate which also has noannual probability assigned to it.

DRAFT



3. Faults in the dam foundation

In practice only active faults are likely tobe of significance to the dam. Howeverthere may be neotectonic faults whichwarrant further investigation to determineif they are active.

4. Ground motion prediction models.

There are insufficient strong motionrecordings from earthquakes in stablecontinental regions anywhere in theworld, including Australia, to form abasis for the development of groundmotion prediction models usingregression analysis of recorded strongground motions. Such analyses haveonly been feasible for crustal earthquakesin tectonically active regions, and whilethe ground motion prediction equationsderived this way have been usedextensively in Australia, the applicabilityof these models in Australia is still notwell established. Accordingly, recentinvestigators (Liang et al., 2008;Somerville et al., 2009; and Allen, 2011),described further below, have usedseismologically based methods to

develop ground motion prediction modelsfor Australia. In view of the significantdifferences in the ground motions that arepredicted by the different ground models,it is necessary to include alternativeground motion prediction models in theseismic hazard analysis in order toaccount for epistemic uncertainty inwhich of these models is more applicablein Australia. The following brieflysummarizes the available models (in2016).

Liang et al. (2008).

Liang et al. (2008) estimated strongground motions in southwest WesternAustralia using a combined Green'sfunction and stochastic approach. Thismodel is applicable to the Yilgarn craton,and may also be applicable to othercratonic regions of Australia, but is notapplicable to non-cratonic regions,including Perth. This model usesepicentral distance as the distancemeasure. Figure C2.20 shows the crustaldomains in Australia as defined by Clarket al (2012).

DRAFT



Figure C2.20 Neotectonic domains map of the Australian continent. Neotectonic featuresrepresent known or suspected features, primarily topographic fault scarps. (Clark et al,2012)

Somerville et al. (2009).

Somerville et al. (2009) demonstratedtheir ability to simulate the recordedground motions of small earthquakes thatoccurred in Eastern and WesternAustralia, and developed earthquakesource scaling models for Australianearthquakes based on earthquake sourcemodelling of the Mw 6.8 1968 Meckeringand the Mw 6.25, 6.4 and 6.6 1968Tennant Creek earthquakes. They thenused a broadband strong ground motionsimulation procedure based on the elasto-dynamic representation theorem andGreen’s functions calculated from crustalstructure models for various regions ofAustralia to calculate ground motions forearthquakes in the magnitude range of 5.0to 7.5. These ground motions were thenused to develop ground motion predictionequations, which were checked forconsistency with available data fromAustralian earthquakes at each step. Theseground motion models predict responsespectra for two crustal domain categories:

Cratonic Australia and Non-CratonicAustralia. The cratonic regions ofAustralia include much of WesternAustralia (but not the coastal strip west ofthe Darling Fault, including Perth); south-central South Australia (including thesite); the northern part of the NorthernTerritory; and north western Queensland(Clark et al, 2011). Non-CratonicAustralia consists of the remainder ofAustralia, including Eastern Australia andpart of the coastal margin of WesternAustralia, which includes all of the statecapital cities, is on Non-CratonicAustralia.

Allen (2012).

Allen (2012) developed a ground motionmodel for south eastern Australia basedon the stochastic model, having calibratedthe parameters of the stochastic modelusing recordings of small earthquakes insouth eastern Australia (SEA). The sourceand attenuation parameters provided inAllen et al. (2007) were first reviewed and

DRAFT

Commentary – 2.0 Assessment of Design Seismic Ground Motions Analysis Method


modified in light of additional availabledata and more rigorous statistical analysis.The dependence of stress drop onearthquake depth was examined, andoptions were provided for variable stressparameter values in the ground motionprediction equation. The near-surface,path-independent diminution parameterimmediately beneath the station, k0,(Anderson and Hough, 1984; Campbell,2009; Van Houtte et al., 2011) was alsoexamined for average station conditionsin SEA, in addition to the parameter’scorrespondence with a limited dataset ofaverage shear-wave velocitymeasurements in the upper 30 m (Vs30)at seismic recording stations acrossAustralia. These updated source andattenuation parameters were used asinputs to the stochastic finite-faultsoftware package, EXSIM (Motazedianand Atkinson, 2005; Atkinson and Boore,2006). Five percent damped responsespectral accelerations were simulated forearthquakes of moment magnitude Mw4.0 to 7.5. These stochastic data were thenregressed to obtain model coefficients andthe resulting ground motion predictionmodel was evaluated against recordedresponse spectral data for moderate-magnitude earthquakes recorded in southeastern Australia.

PEER-NGA West 2.

These are the most recently developedground motion models for shallow crustalearthquakes in tectonically active regions.They were developed by five groups:Abrahamson et al. (2014), Boore et al.(2014); Campbell and Bozorgnia (2014),Chiou and Youngs (2014), and Idriss(2014), and are compared by Gregor et al.(2014). In view of the great care that wasput into documenting the NGA West 2metadata that describe the strong motionrecordings, the vastly larger size of thedata set that has been used, and thediligence that has been applied by themodellers, the NGA Program has resultedin a set of ground motion models thathave a much more substantial basis thanthe earlier 1997 generation of models

(Abrahamson and Shedlock, 1997).

Selection of Ground Motion Models foruse in Australia

The NGA West 2 models are basedmostly on a large global set of recordedground motion data from tectonicallyactive regions, but none of those data arefrom Australia. The Allen (2012) andSomerville et al. (2009) ground motionprediction models were both developedfor Australia, but are not based on a largeset of recorded strong motion data fromAustralia.

In view of this the NGA West 2 modelshould be included.

Judgments need to be made as to therelative weights that should be given tothe Australian and global models. This isa matter for the Seismologist todetermine, in consultation with the Ownerand the Consultant, taking account ofwhether the site is in cratonic or non-cratonic conditions.

It can be expected that new models willbe developed for Australian and relevantInternational conditions and given theyare rigorously peer reviewed, they may beincluded in the PSHA with appropriateweighting.

Impact of Site Conditions on GroundMotion Level

Ground motion prediction models used inearthquake engineering are based on threemain parameters: the magnitude of theearthquake, the distance of the earthquakefrom the site, and the site characteristics.It has long been known that sitecharacteristics have a strong influence onground motion level. Until recently, sitecharacteristics have been represented bybroad geological categories such as“rock” or “soil.” In eastern Australia, ithas been common to assume that the sitecharacteristics of dam abutments can berepresented by the “rock” site category inground motion models such as Sadigh etal. (1997). However, this ground motion

DRAFT



model was found in the course of theNGA Project to be representative of softrock sites in California having an averageshear wave velocity (Vs30) of only 520m/sec, while many dams in Australia maybe founded on hard rock having Vs30 of1,000 m/sec or more. In view of this, re-evaluation of the seismic hazard whichwas carried out using Sadigh et al (1997)may identify a significant level ofconservatism in those hazard assessmentsfor those dams located on hard rockfoundations.

Ground motion models, such as the NGAWest 2 models (Gregor et al., 2014) havebeen developed that quantify sitecharacteristics in a much more rigorousway. Specifically, these models specifythe site characteristics using Vs30, whichis the average shear wave velocity in theuppermost 30 meters below the groundsurface. Amplification of ground motionis inversely proportional to Vs30. Theamplification is roughly equal to thesquare root of the ratio of subsurface tosurface shear wave velocity, andillustrated in Figure C2.21. AlthoughVs30 is not yet routinely measured in thefoundation investigations for dams, it canusually be inferred from the P-wavevelocities obtained from seismic

refraction surveys which were oftencarried out as part of the original siteinvestigation for the dam site.

The ground motion models for Australiathat were developed by Somerville et al.(2009) and Allen (2011) do not haveVs30 as a variable, and instead are forrock site conditions. The Vs30 used bySomerville et al. (2009) is 865 m/sec,which is consistent with average rock siteconditions in Australia. Similarly, theVs30 of the Allen (2011) model isassumed to represent a Vs30 of 820m/sec. The site amplification model ofCampbell and Bozorgnia (2008) can andshould be used to adjust the shear wavevelocities represented in these models tothat of a specific site where the Vs30 ofthe site is significantly different to that inthe models.

DRAFT



Figure C2.21 Dependence of response spectral acceleration on Vs30 for a magnitude 7earthquake at a distance of 30 km. (Abrahamson and Silva, 2008).

5. Shear wave velocity of the damfoundation bedrock

The shear wave velocity can be measureddirectly or may be assessed indirectlyfrom seismic refraction “P” wavevelocities. Many dams have had seismicrefraction (“P” wave) surveys carried outas part of the site investigations for thedam and these may be used to estimateVs30.

If there are no data for the dam, a guide to

what might be expected can be obtainedfrom Table C2.4 and / or by consulting anexperienced engineering Geologist.However in most cases if there are nodata, or what data there are only old “P”wave data, site specific “S” wave testingshould be carried out because the seismichazard is significantly dependent on Vs30and as can be seen from Table C2.4unless measurements are made at the damsite the potential range of Vs30 is large.DRAFT



Table C2.4 Shear wave velocity Vs30 (m/s) site classification applicable to Australian regolithconditions (Mcpherson and Hall, 2007)

Site Class Vs30 (m/s) Geological Materials

B >760 Fresh to moderately weathered hard rock units (Plutonic &metamorphic rocks, most volcanic rocks, coarse-grainedsedimentary rocks Cretaceous & older)

BC 555 - 1000 Highly weathered hard rock; some Tertiary volcanics

C 360 - 760 Sedimentary rocks of Oligocene to Cretaceous age; coarse-grained; sedimentary rocks of younger age; extremelyweathered hard rock units

CD 270 - 555 Sedimentary rocks Miocene and younger age, unlessformation is notably coarse grained; Plio-Pleistocene alluvialunits; older (Pleistocene) alluvium, some areas of coarseyounger alluvium

D 180 - 360 Younger (Holocene to Late Pleistocene) alluvium

DE 90 - 270 Fine grained alluvial, deltaic, lacustrine and estuarinedeposits

E <180 Intertidal and back-barrier swamp deposits

6. Effects of soil overlying bedrockCurrent ground motion prediction modelsestimate the amplification of ground motionsby shallow geology using Vs30, the time-averaged shear wave velocity over the upper30 metres of the ground. Vs30 is a continuousvariable that spans the range from soft soils,stiff soils, weathered rock, soft rock, to hardrock. Vs30 is specified in the range of 180m/sec to 1100 m/sec (ranging from soft soil tohard rock). The data are insufficient toconstrain amplification in rock above 1100m/sec; it is expected to decrease slowly.Below 180 m/sec, the use of Vs30 is notviable and a site-specific soil response studyis necessary. Such soils are described inUnited States codes as NEHRP sites E and F.

Use of Vs30 to estimate soil amplification at aspecific site assumes that the average shearwave velocity, shear modulus reduction, anddamping increase profiles represented in thestrong motion data base used to develop theground motion prediction models are areasonably accurate representation of thespecific profiles at the site. If this is not thecase, then a site-specific analysis of the soilamplification should be done as describedbelow. For most dam sites where the dam is

founded on soft or deep valley alluvium, asite-specific analysis may be required.

Generally speaking, the larger the value ofVs30, the more likely it is that it provides anadequate representation of the amplificationof the site while avoiding the uncertaintiesassociated with nonlinear soil responseanalysis. Also, generally speaking, the lowerthe level of the ground motion that is inputinto the soil, the less nonlinear will be theresponse of the soil, and the more reliable willbe the estimate of the soil amplification usingVs30.

However if the dam is founded on sedimentsor deep residual soils and completelyweathered rock overlying the bedrock whichare expected to have significant nonlinearresponse, then the response of these materialsshould be analysed in a separate study ofnonlinear soil response using SHAKE,QUAD4M or similar programs that would notbe part of the seismic hazard assessment. TheConsultant who performs this analysis shouldspecify the subsurface level (depth in theprofile) and the associated (subsurface) Vs30,Z1.0 and Z2.5 at which the input groundmotions are to be provided. In this case, Vs30

DRAFT



is the time-averaged shear wave velocity overa depth of 30 meters (m/s) below the selectedsubsurface level, and the Z1.0 and Z2.5(depths to shear wave velocities of 1.0 and 2.5km/sec) are also adjusted accordingly. Theground motions developed to represent free-field ground motions at this subsurface levelare then used as input into the nonlinearanalysis.

The analyses should use the non-liquefiedproperties of the soil.

One dimensional analyses such as SHAKEare potentially conservative for 2D and 3Dstructures such as embankment dams, sowhere the results are important to dam safetydecision making it may be necessary to carryout 2D modelling such as QUAD4M orQUAKEW.

7. Response spectra

No commentary.

8. Minimum Magnitude Earthquake

Even small earthquakes can producerelatively large peak accelerations at closedistances, but such earthquakes generallyhave low damage potential, and it isdesirable to exclude earthquakes below aspecified minimum magnitude to avoidsuch events contributing to anunrealistically high hazard level,especially for PGA and short period

ground motions. The minimummagnitude considered by Burbidge andLeonard (2011) in the new draft seismichazard maps of Australia for generalbuilding code applications is magnitude4.5. For dams, it will be appropriate touse a higher minimum magnitude.

9. Quantification and reporting epistemicuncertainty

Uncertainty in the inputs and models shouldbe modelled and reported. There existsepistemic uncertainty (about the true state ofnature) in earthquake source models (and thehistorical earthquake catalogues on whichthey are based); about how to incorporateactive faults and in the source parameters ofthese faults; in ground motion predictionmodels; and in the shear wave velocityassigned to the dam foundation and thedamping coefficients adopted. Theseuncertainties lead to uncertainty in the truevalue of the hazard at a specified returnperiod. These uncertainties can be significant(Somerville and Thio, 2011). Figure C2.22shows the results of a seismic hazardassessment where the mean (best estimate) ofthe hazard would result in the foundationlargely being non-liquefiable, but the 85%fractile and 95th fractile would likely result inliquefaction, giving a totally differentoutcome for only marginally less likely loads.DRAFT



Example Dam: Equal Hazard Spectra – 10,000 yr – 5%

Figure C2.22 Example of horizontal spectral response acceleration showing the epistemicuncertainty in the true value of the hazard. The mean hazard is shown by the red line and the 5%and 95% fractiles are shown by the blue lines.

10. De-aggregation plots and selection oftime-history motions.Earthquake de-aggregation plots showingmagnitude – distance contributions should beprovided because these are a significant inputinto analyses for assessing embankment damdeformations and liquefaction. The de-aggregation of the hazard varies with thereturn period and with the period of theground motion, so the magnitude and distanceshould be selected based on knowledge of thestructure being analysed and the return periodbeing considered.

11. Reservoir-induced seismicity.

See Sections C2.1.5 and C2.11.

In some situations for high and extremeconsequence category dams a staged approach

may be appropriate with a less detailed PSHAas described below for significantconsequence category dams used initially, andthe more detailed assessment carried out if itis recognised that the seismic hazard iscritical.

For significant and low consequence categorydams situations where a site specific anddetailed PSHA may be required include damswhich have potentially liquefiablefoundations, or other features such as wallsretaining the embankment which are sensitiveto the seismic loading.

C2.6 Selection of Design Seismic GroundMotion - Deterministic Analysis ApproachOperating Basis Earthquake

DRAFT



Other stakeholders may for example include alocal government council or other watersupply authority which obtains their waterfrom the dam Owner.

Safety Evaluation Earthquake

In assessing the required factor of safety toachieve a low likelihood of failure the effectof the 85th fractile (and possibly 95th fractilefor Extreme Consequence Category dams)ground motions on the dam should beassessed. These are about 3 times and 10times less likely than the median or 50th

fractile ground motion. Alternatively the 1 in30,000 AEP or 1 in 100,000 AEP groundmotions may be used.

If the dam would fail leading to breach forthese ground motions the likelihood of failurewould probably be too high to satisfyANCOLD (2003, 2016) tolerable risk criteria.Some additional information to assist in thisassessment is given below.

For High B, High C , Significant and LowConsequence Category dams, if the structureis susceptible to liquefaction or hascomponents which will fail at loads only alittle greater than the loads in Table 2.1, checkthe design for the critical load and assess theadequacy of the design using risk assessmentmethods.

It should be noted that in Australia the MCEloading on known active faults may besignificantly greater than the probabilisticloadings in Table 2.1 even though theprobabilistic loadings include the effects ofthe active faults. This is because the MCEapproach effectively ignores the AEP of theearthquake on the fault and this may be verylow, so it does not have as much effect in theprobabilistic assessment as it does in the MCEapproach. In seismically active areas thereverse may apply.

If the MCE loading is significantly larger thanthe probabilistic loading detailed discussionsshould be held with the Seismologist to better

understand the characteristics of the activefault. This may result in additionalinvestigations being required such astrenching across the fault and dating theepisodes of displacement. Once the MCEloading is finalised either adopt it and / oradopt a risk based approach.

If there is still some doubt about whether afault is active or not, consult the Seismologist(s) and Geologist involved and assign aprobability that the fault is active and applythis in the event tree logic in a risk analysis.

The use of a deterministic analysis approachrequires that a factor of safety should beapplied to estimated stresses and deformationswithin the dam and its foundations to give alow likelihood of the dam failing given theSEE loading. Table C2.5 shows theconditional probabilities of failure required tojust achieve ANCOLD (2003) tolerable riskguidelines. To allow for the fact the tolerablerisk criteria are for the sum of all potentialfailure modes, and the seismic loadingcomponent may only be a small part of theoverall probability of failure, and to givesome margin of safety that the risks are belowthe limit of tolerability, the conditionalprobability of failure for seismic loading forexisting dams, given the SEE should be about1 in 100, and for new dams or majoraugmentation, about 1 in 100 to 1 in 1000.This requires a significant factor of safety onstresses or deformations for the SEE load. Itshould be noted that individual risk is thecontrolling factor for High C, Significant andLow Consequence Category dams.

The requirement for a factor of safety onstresses or deformations applies to allconsequence categories and is a criticalrequirement if a deterministic analysisapproach is to be followed.As there are no unique relationships betweenfactor of safety and likelihood of failure theConsultant will have to apply engineeringjudgement to achieve the required lowlikelihood of failure.

DRAFT



Table C2.5 Indicative required conditional probabilities of failure given the ANCOLD (2013)Design Earthquake

DamConsequenceCategory, or

“class”

ANCOLD(2013)

Design ofConcrete

Gravity DamsRecommendedDesign Loads

ANCOLD(2012)

ConsequenceCategoriesPotentialLife Loss

(PLL)

ANCOLD(2003)

SocietalRisk

AnnualProbabilityof FailureLimit (a)

ConditionalProbability of

FailureTo achieve

ANCOLD(2003)Limit Criteria

(a) (b)

ConditionalProbability of

FailureTo achieve

ANCOLD(2003)Limit Criteria

(c)

Extreme 1:10,000 AEP >50100

1000

2E-051E-05

1E-06(d)

<0.2<0.1<0.01

<0.10.050.005

High A 1:10,000 AEP 1 to 50 1.0E-03 to2E-05

< 0.2 < 0.1

High B 1:5,000 AEP 0.1 to 5 1.0E-02 to2.0E-04

<1(0.5,0.05)

<0.5(< 0.25 to<0.025)

High C 1:2,000 AEP <0.1 to 5 1.0E-02 to2.0E-04

<0.4

(<0.2,0.02)

<0.2

(<0.1, 0.01)

Significant 1:1,000 AEP <0.1 to 1 1.0E-02 to1.0E-03

<1

(<0.1,0.01)

(<0.5)

(<0.05, 0.0005)

Low 1:1,000 AEP <0.1 1.0E-02 <1

(<0.1,0.01)

(<0.1)

(<0.005, 0.0005)

Note(a) The figures include the contribution to the annual probability of failure from other failure modes.(b) Figures in brackets controlled by individual risk criteria, 1.0E-04 / annum for existing dams;1.0E-05 for new dams or major augmentations. Figures assume vulnerability of the person most atrisk = 1.(c) These figures assume that the contribution to the risk from seismic loads is half the total.(d) This assumes that the horizontal truncation in the societal risk plot is removed in ANCOLD(2016).

C2.7 Selection of Design Seismic Loading-Risk Based Analysis ApproachFor the risk analysis the seismic loads shouldbe partitioned taking account of the fragilityof the structure. For example:

(a) If liquefaction is an issue, theearthquake loading with an AEP atwhich liquefaction is widespreadwould be used as a partition boundary.

(b) The AEP of the earthquake loadingwhich results in critical retaining

walls; e.g. between spillway andembankment; failing.

(c) The AEP of the earthquake loadingwhich is likely to result indisplacement of a concrete gravitydam.

The uncertainty in the earthquake loadingshould be considered and modelled in the riskanalysis.

DRAFT



C2.8 Modelling vertical ground motionsGulerce and Abrahamson (2011) provideground-motion prediction equations (GMPEs)for the vertical-to-horizontal spectralacceleration (V/H) ratio, and the methods forconstructing vertical design spectra that areconsistent with the probabilistic seismichazard assessment results for the horizontalground motion component.

The proposed V/H ratio GMPE is dependenton the earthquake magnitude and distance,and accounts for the differences in the non-linear site-response effects on the horizontaland vertical components. This results in largeV/H ratios at short spectral periods for soilsites located close to large earthquakes.

It is suggested that this method be used ratherthan using a constant ratio of vertical tohorizontal ground motion as has beencommonly done.

C2.9 Selection of Response Spectra andTime-History AccelerogramsThe amount of damping assumed for theresponse spectrum and analyses will dependon the type of structure. For embankmentdams the amount of damping will depend onthe extent of straining within the dam. Forexample, the Makdisi-Seed method forestimating earthquake induced deformationsin an embankment varies both the bulkmodulus and the damping factor according tothe shear strain in the embankment. Forconcrete dams, the USACE (1999) states:“Energy dissipation in the form of a dampingratio is included as part of the responsespectrum curves. For the linear elastic ornearly elastic response during an OBE event,the damping value should be limited to 5percent. For the SEE excitation, a dampingconstant of 7 or 10 percent may be useddepending on the level of strains and theamount of inelastic response developed in thestructure”.

Suitable methods can be used to convert aresponse spectrum to account for higherdamping factors than the one for which theresponse spectrum was prepared (e.g. Rezaeinet al., 2014a, b).

At least four or five ground motion recordsare generally needed for advanceddeformation analyses. A relatively large suiteof records is required due to the range ofdeformation predictions that may occur evenfor a carefully selected set of motions. Sinceonly 4 or 5 records are being used, the intentof the study is not to define the full range ofpotential displacements but to determine theaverage, expected response for the specifiedlevel of earthquake loading.

Three ground motion components should beprovided for each record: two horizontal andone vertical. While vertical motions are oftenconsidered to have a modest effect ondeformation predictions, advances indeveloping appropriate and consistentmotions and the ease with which they can beincluded in many sophisticated analyseswarrants their routine use in deformationanalyses.

The suite of records should be obtained fromdifferent source earthquakes to reduceunintended bias in the record selection. Thefollowing criteria may be considered in theselection of earthquake record:

a) Records to be used for preparation ofsite specific time-histories shouldoriginate from a seismic event similarto the target design earthquake (e.g.,magnitude, fault distance, and focaldepth). The site condition for eachrecord should reasonably correspondto the site condition for the targetresponse spectrum. For example, itmay be appropriate to use a recordfrom a shallow, stiff soil site torepresent soft rock conditions, but nota deep soil record.

b) The shape of the response spectrumfor each record should reasonablymatch the target response spectrumover the frequency range of interest.This frequency range may be ratherlarge and will typically include lowfrequencies (long periods).

c) Scaling factors may be applied to therecord to provide a best fit to theresponse spectrum over the periodrange of interest (see Figure C2.23).

DRAFT



Alternatively, spectral matchingprograms such as RSPMatch can beused to more closely match the targetresponse spectrum over a wide rangeof frequencies.

d) Spectral matching techniques shouldbe carefully applied to preserve asmuch of the original character of theearthquake record as possible (e.g.,relative magnitude and duration ofvelocity peaks). Although scalingfactors are traditionally limited tovalues between 0.5 and 2.0, valuesoutside of this range may be permittedin some cases (Watson-Lamprey andAbrahamson, 2006).

e) Additional criteria can be useful indefining an appropriate suite of ground

motions, such as Arias Intensity orsignificant duration. Attenuationrelationships are available for theseparameters allowing their inclusion indeterministic or probabilistic hazardestimates (e.g., Watson-Lamprey andAbrahamson, 2006; Travasarou, et al.,2003; Kempton and Stewart, 2006].

f) Original earthquake records can beobtained from a number of onlinesources, including the COSMOS andPEER websites. Syntheticaccelerograms should be consideredwhen the design earthquake is notwell-represented by the database ofavailable records.

Figure C2.23 Scaling of the input time-history for best fit with target response spectrum overperiod range of interest (Perlea and Beaty, 2010).

C2.10 Earthquake AftershocksThis is more likely to be an issue where anactive fault or individual neotectonic fault is amajor contributor to the seismic hazard.

An example of a dam which might besusceptible to aftershock seismic groundmotions is a concrete face, asphalt face orother membrane face rockfill dam which haspoor drainage capacity, so the rockfill may

saturate from leakage through the damagedface slab or membrane after the mainearthquake, leaving it with a low static factorof safety and susceptible to much largerdeformations and damage during theaftershock.

Concrete dams and appurtenant structuressuch as walls retaining the embankmentwhich may experience significant

DRAFT



displacement during the primary groundmotion, sufficient to compromise thefoundation drainage or passive or posttensioned anchors should be considered foraftershocks.

C2.11 Seismic Ground Motions fromEarthquakes Induced by the ReservoirThe following is adapted from ICOLD(2016a).

The Reservoir-Triggered Earthquake (RTE)represents the maximum level of groundmotion capable of being triggered at the damsite by the filling, drawdown, or the presenceof the reservoir. ICOLD Bulletin 137 onReservoirs and Seismicity provides the state-of-knowledge on reservoir-triggeredseismicity. Section C2.1.5 gives some details.

While there exist differences of technicalopinion regarding the conditions which causereservoir-triggered seismicity, it should beconsidered as a credible event if the proposedreservoir contains active or neotectonic faultswithin its hydraulic regime and if the regionaland local geology and seismic record withinthat area are judged to indicate potential forreservoir-triggered seismicity. Even if all thefaults within a reservoir are consideredtectonically inactive, the possibility ofreservoir-triggered seismicity should not betotally ruled out, if the local and regional

geology and seismicity suggest that the areacould be subject to reservoir-triggeredseismicity.

Depending on the dam location and prevailingseismotectonic conditions, the RTE mayrepresent ground motion less than, equal to, orgreater than the OBE ground motion. RTEground motion should in no case be greaterthan the Safety Evaluation Earthquake groundmotion because the faults considered capableof triggering seismicity should be taken intoconsideration during the seismic hazardevaluation. However the result might be thepremature triggering of seismic events due tothe impounding of the reservoir that wouldhave occurred naturally at some longer time inthe future. It is therefore justified in the caseof larger dams and storages located inseismically active regions and regions withhigh tectonic stresses to install a micro-seismic network and to monitor the seismicityprior to, during and after impounding.

This has been done on some Australian dams,e.g. Thomson and Dartmouth, and micro-seismicity was detected associated with faultsin the reservoir as discussed in SectionC2.1.5.

For existing dams there is unlikely to be anyRTE from reservoirs which have been filledfor several decades.

DRAFT

Commentary – 3.0 Assessment of Embankment Dams for Seismic Ground Motion


C3 ASSESSMENT OF EMBANKMENTDAMS FOR SEISMIC GROUNDMOTIONS

C3.1 Effect of Earthquakes onEmbankment Dams.

No commentary.

C3.2 Defensive Design Principles forEmbankment Dams.The concept of ‘defensive design’ ofembankment dams for earthquake wasdeveloped by Sherard (1967) and Seed (1979)and endorsed by Finn (1993), ICOLD (1986,1999a) and ANCOLD (1998).

C3.3 Seismic Deformation Analysis ofEmbankment Dams.

C3.3.1 The Methods Available and Whento use themThere are a number of methods available forestimating the deformations which may occurin embankments and their foundations duringand post seismic loading. These can besummarized as:

(a) Screening and empirical databasemethods. These are applicable toembankments and their foundationswhich do not liquefy, or experiencesignificant loss of strength due eitherto build up of pore pressure or strainweakening. These should only beapplied if the criteria listed in SectionC3.3.2 are satisfied, and should onlybe relied upon if the estimateddeformations are much less than whatis tolerable; e.g. crest settlements aremuch less than the available freeboard.

(b) Simplified methods for estimatingdeformations during earthquakes.These methods are also onlyapplicable to embankments and theirfoundations which do not liquefy.They assume the post-earthquakedeformations are negligible and thedeformations during the earthquakeare due to the action of the horizontalinertia forces induced by theearthquake. They are commonly basedon the Newmark (1965) principle.

(c) Post-earthquake deformations forliquefied conditions. These methodsassume that the deformations areprimarily caused by gravitationalforces acting on an embankmentfollowing the earthquake, and allowfor the reduction in strength caused byliquefaction or strain weakening ofother soils. They may be carried outusing limit equilibrium and or staticnumerical methods.

(d) Advanced numerical methods forestimating deformations during andpost-earthquake for non-liquefied andliquefied conditions. These cover awide range of sophistication andcomplexity and include dynamicanalyses using total and effectivestress methods, linear and non-linearmodels, and varying degrees ofrefinement of how pore pressures aredeveloped and coupled todeformations.

Perlea and Beaty (2010) give an overview ofthe methods and their application.

The Guideline and the information presentedbelow give guidance on which methodsshould be used.

It should not be assumed that the screeningand simplified methods are conservative andwhere estimated deformations are near to thetolerable deformations more advancedmethods should be used or measures taken toreduce deformations.

For risk based methods these calculations willin principle need to be done for a range ofseismic loads. In practice for many damsdeformations will be very small even for thelargest loads so it will be unnecessary to dothe calculations for the smaller loads.

For deterministic methods only the OBE andSEE load will need to be analysed subject tothe qualifications in Sections 2.6 and C2.6.

The screening, database and simplifiedmethods are not applicable to tailings damsconstructed from hydraulically placed tailingsbecause the tailings are not compacted to thedegree required for conventional earthfill and

DRAFT



rockfill, and will as a result be more subject todensification during seismic loading, and maybe subject to pore pressure generation andliquefaction.

C3.3.2 Screening and Empirical DatabaseMethods

Screening Method

Perlea and Beaty (2010) report that theUSACE do not require deformation analysesfor low to moderate height dams < 60m highif all the following criteria are met:

1. Dam and foundation materials aredense, not subject to liquefaction, anddo not include sensitive clays.

2. The dam is well built and denselycompacted to at least 95% of thelaboratory maximum dry density, or toa relative density greater than 80%.

3. The slopes of the dam are 3:1 (H:V) orflatter, and/or (the slopes are steeperbut) the phreatic line is well below thedownstream face of the embankment.

4. The predicted peak horizontal groundacceleration (PGA) at the base of theembankment is no more than 0.20g.Compacted clay embankments on rockor stiff clay foundations may offeradditional resistance to deformations.Somewhat higher allowable PGAvalues may be justifiable for thesedams on a case-by-case basis,although the PGA criterion should notexceed 0.35g (USBR, 1989).

5. The static factors of safety for allpotential failure surfaces involvingloss of crest elevation (other thanshallow surficial slides) are greaterthan 1.5 under the loading and pore-pressure conditions expectedimmediately prior to the earthquake.

6. The freeboard at the time of theearthquake is at least 3 to 5 percent ofthe dam height plus alluvialfoundation, and not less than 0.9 m.Special attention should be given tothe presence and suitability of filtersfor dams with modest freeboard.

7. There are no appurtenant featuresrelated to the safety of the dam that

would be harmed by small movementsof the embankment.

8. There have been no historic incidentsat the dam that may indicate alimitation in its ability to survive anearthquake.

The words in italics have been added forclarity.

This approach may be used as a screeningmethod. However for the SEE most dams inAustralia will have a PGA greater than 0.2g(0.35g).

Empirical Database Methods

There are a number of these methods whichare developed by gathering deformation andother data on embankment dams which havebeen subject to earthquake, and relating theamount of deformation and cracking to theearthquake loading experienced by the dam.Of these the Swaisgood (1998) method isbased on a large database and is referred to inPerlea and Beaty (2010) as being used byUSACE, and is in use in Australia.

The Pells and Fell (2002, 2003) method uses alarger database and also records the amount ofdamage as evidenced by longitudinal andtransverse cracking observed. It also givesguidance on for what seismic loading(measured by Mw and PGA) transversecracking can be expected. This is importantfor considering concentrated leak erosionfollowing the earthquake.

Both methods are described in Fell et al(2005, 2015). They only apply to dams wherethere is no potential for liquefaction orsignificant strain weakening. They should beapplied conservatively, e.g. use upper boundestimates, allowing for the scatter in the dataused to form the plots. They are for manydams quite sufficient provided the dams arewell constructed and there is a large marginbetween estimated settlement and thefreeboard.

DRAFT



Hynes-Griffin and Franklin (1984) Pseudo-Static Seismic Coefficient Method.

This method is based on Newmark slidingblock analyses of 349 horizontal componentsof natural earthquakes and 6 synthetic recordsfor a range of yield accelerations. From thispermanent displacements u in cm were plottedversus N/A where N is the yield accelerationand A is the peak value of the earthquakeacceleration at the base of the dam.

Hynes-Griffin and Franklin (1984) then usedan arbitrary limit of 100cm (one metre)displacement as representing tolerabledisplacements, and from Figure C3.1, used theupper bound plot of displacements, whichgave N/A = 0.17. After allowing foramplification factor of 3 between the baseacceleration and that at the crest of the dam,they concluded that a factor of safety of 1.0

with a seismic coefficient of one half of thepeak ground acceleration at the base of thedam, would assure that deformations wouldnot exceed 1 metre. Their suggested methodis:

(a) Carry out a conventional pseudo-staticstability analysis using a seismiccoefficient equal to one-half thepredicted peak ground acceleration.

(b) Use a composite S-R strengthenvelope (Effective stress strength atlow stresses, undrained strength athigh stresses) for pervious soils and Rundrained strength for clays,multiplying the strength in either caseby 0.8.

(c) Use a minimum factor of safety of 1.0.

Figure C3.1. Permanent displacement u versus N/A based on 354 accelerograms (Hynes-Griffin and Franklin, 1984)

It should be noted that this method cannot beused in reverse. That is, just because thefactor of safety is less than 1.0, it should notbe assumed that the deformations will begreater than 1 metre. This is because theupper bound deformation curve in FigureC3.1 has been used in the method.

In practice for most dams with static factorsof safety about 1.5, the SEE PGA will resultin factors of safety using this method lessthan 1.0 so the method is not particularlyuseful. It is included here because thebackground to the method is often notappreciated by practitioners, and to avoid itsmisuse.

DRAFT



C3.3.3 Simplified Methods for EstimatingDeformations during Earthquakes

These methods are all based on theNewmark (1965) approach. He introducedthe basic elements of a procedure forevaluating the potential deformations of anembankment under earthquake loading. Inthis procedure, sliding of a soil mass along afailure surface was likened to slipping of ablock on an inclined plane. Newmark (1965)envisaged that failure would initiate andmovements would begin when the inertiaforces exceed the yield resistance, and thatmovements would stop when the inertiaforces were reversed. Thus, he proposed thatonce the yield acceleration and theacceleration time-history of a slipping mass

are determined the permanent displacementscan be calculated by double integrating theacceleration history above the yieldacceleration as shown in Figure C3.2.

Newmark’s approach is limited inapplication to compacted clayeyembankments and dry or dense cohesionlesssoils that experience very little reduction instrength due to cyclic loading. Theapproaches using this principle should not beapplied where embankments or theirfoundations are susceptible to liquefaction orstrain weakening because they willsignificantly underestimate displacements.

There are a number of methods based on theNewmark (1965) principle:

Figure C3.2. Double integration method for determination of the permanent deformation periodof an embankment (Newmark, 1965).

Makdisi and Seed (1978) Method

The Makdisi and Seed (1978) approach isbased on Newmark’s method, but modifiedto allow for the dynamic response of theembankment as proposed by Seed andMartin (1966). The approach was developedfrom a series of deformation analysesperformed on a large number of

embankments subjected to earthquakeloading.

The approach has been widely used andaccepted among practicing engineers.However it is limited in application to damsnot susceptible to liquefaction or strainweakening in the embankment or itsfoundations. Perlea and Beaty (2010) alsowarn the method may not be appropriate for

DRAFT



severe ground shaking because it does notmodel strength losses. They also point out itwas based on few earthquake records.

Bray and Travasarou (2007) used thismethod as well as their own (describedbelow) and for the dams they analysed theMakdisi and Seed (1978) method generallyunderestimated the actual displacements.

Bray and Travasarou (2007) Method.

A simplified procedure was proposed byBray and Travasarou (2007) for estimatingearthquake induced permanentdisplacements in earth dams using aNewmark-type model. This procedure isbased on the results from a set of simplifiednon-linear analysis using 688 recordedground motions from 41 earthquakes. Theflexibility of the dam system, and theinteraction between yielding and seismicloading, were considered by using a non-linear coupled stick-slip deformable slidingmodel. The flexibility of the dam structure iscaptured through an estimate of the initialfundamental period Ts.

Key parameters of this procedure are theyield acceleration ky (in g), the initialfundamental period of the embankment, Ts,and the value of spectral acceleration for adamping of 5% and a degraded responseperiod equal to 1.5Ts.

Bray and Travasarou (2007) claim themethod provides improved characterisationof the uncertainty involved in the estimate ofseismic displacement. It can be also beincorporated into a probabilistic framework.

C3.3.4 Post-earthquake Deformations forLiquefied Conditions

C3.3.4.1 Limit Equilibrium Analysis.The post liquefaction stability is assessed asfollows (Fell et al, 2005, 2015):

(a) Determine the zones which haveliquefied (i.e. FS < 1.0) under theearthquake loading using themethods in Section C3.4.4.

(b) Determine the liquefied shearstrength (Su(LIQ)) for these zones

using the methods described inSection C3.4.5.

(c) For potentially liquefiable soils witha factor of safety against liquefactiongreater than 1.0, determine theresidual excess pore pressure asdetailed in Section C3.4.6.1.

(d) For clay soil zones in theembankment and foundation assignstrength and pore pressure consistentwith the soil’s behaviour in staticloading after being cracked anddisturbed by the earthquake. If theclay is contractive in nature, useundrained strengths. If it is dilativeon shearing, use effective stressstrengths c′, f′. Usually there will besome cracking and loosening, and ifso fully softened strengths, wouldapply.. Some apply an arbitrary 10%or 15% loss of strength.

(e) For well compacted free drainingrockfill filters and dense sands andgravels adopt the effective stressstrengths c′, f′, with no change in thepore pressures. If large deformationsare expected in the earthquake thedense granular materials may haveloosened and will have a strengthapproaching the critical statestrength, rather than the peakstrength. In practice this can beaccommodated by a small reductionfrom the expected peak strengths.

The analysis is done with conventional limitequilibrium analysis methods. No loadingfrom the earthquake is applied, since this is apost-earthquake analysis.

If the liquefied zone is subject to flowliquefaction and the post-earthquake factorof safety is significantly less than 1.0, large,rapid deformations and flow sliding can beexpected. If the factor of safety is onlymarginally less than 1.0, or marginallyabove 1.0, deformations may not be so largeas to lose freeboard between the dam crestand the reservoir level.

In considering these factors of safetyaccount should be taken of the uncertainty in

DRAFT



the assessment of the extent of liquefiedzones and the liquefied strength. Perlea andBeaty (2010) indicate that a factor of safetyin excess of 1.2 to 1.5 may be required toachieve tolerable displacements.

Because of the uncertainty in what postliquefaction factor of safety will givetolerable displacements, this approach andthe simplified method described in SectionC3.3.4.2 should only be used as described inthe Guideline.

C3.3.4.2 Simplified Deformation Analysis.An approximate estimate of thedeformations can be obtained using theKhalili et al. (1996) method which isdetailed Fell et al (2005, 2015).

C4.3.4.3 Static Numerical DeformationAnalyses.

Indicative estimates of deformations can beobtained by performing a static deformationnumerical analysis which incorporates theearthquake induced pore pressures and theliquefied strength of the liquefied soils(Finn, 1993). The analysis is oftenperformed in two stages. In the first stage,the numerical model is initialised to the pre-earthquake conditions of the dam bysimulating the current in-situ stresses. Then,in the second stage, the earthquake inducedpore pressures and residual strengths of theliquefied soils are incorporated into themodel to simulate post-liquefactionconditions.

This type of analysis is also referred to asuncoupled deformation analysis andgenerally leads to conservative estimates ofpost liquefaction deformations, as it does notallow for dissipation of earthquake inducedpore pressures with time. More accurateestimates of post liquefaction deformationscan be obtained using fully and semi-coupled methods of analysis, as discussed inthe following sections.

However for many projects the uncoupleddeformation analyses are sufficient. As forthe more simplified methods, account shouldbe taken of the uncertainty in the assessmentof the extent of liquefied zones and the

liquefied strength. These are best accountedfor in a risk framework by assigning alikelihood to a range of liquefied strength,and analysing for each liquefied strength.

C3.3.5 Advanced Numerical Methods forEstimating Deformations Duringand Post-Earthquake for Non-Liquefied and LiquefiedConditions

The dynamic numerical codes used inpractice may be divided into two maincategories: total stress codes, and effectivestress codes (Zienkiewicz et al., 1986 andFinn, 1993). They are reviewed inZienkiewicz et al., (1986), Finn (1988,1993,2000), Marcuson et al. (1992), and Perleaand Beaty (2010). A brief discussion ofsome of the more frequently used codeswithin each category is provided in thefollowing sub-sections.

C3.3.5.1 Total Stress CodesThe total stress codes, as can be inferredfrom the classification, are based on the totalstress concept and do not take account ofpore pressures in the analysis. Thereforethey should only be used in situations wherethe seismically induced pore pressures arenegligible. The total stress codes may bedivided into two main categories: (1) codesbased on the equivalent linear (EQL) methodof analysis, and (2) fully non-linear codes.

(a) Equivalent linear analysis (EQL)

The earlier total stress codes were based onthe EQL method of analysis developed byH.B. Seed and his colleagues in 1972. EQLis essentially an elastic analysis and wasdeveloped for approximating non-linearbehaviour of soils under cyclic loading.Typical of the EQL codes used in practiceare: SHAKE (Schnabel et al., 1972), QUAD-4 (Idriss et al., 1973), QUAD4M (Hudson etal 1994) and FLUSH (Lysmer et al., 1975).SHAKE is a one dimensional wavepropagation program and is used primarilyfor site response analysis. QUAD-4,QUAD4M and FLUSH are two-dimensionalversions of SHAKE and are used for seismicresponse analysis of dams andembankments.

DRAFT



Given the elastic nature of the EQL analysis,however, these codes cannot take account ofmaterial yielding and material degradationunder cyclic loading. Therefore, they tend topredict a stronger response than actuallyoccurs. Also, they cannot predict thepermanent deformations directly. Indirectestimates of permanent deformations canhowever be obtained using the accelerationor stress data obtained from an EQL analysisand the semi-empirical methods proposed byNewmark (1965) and / or Seed et al. (1973).

(b) Fully non-linear analysis

More accurate and reliable predictions ofpermanent deformations can be obtainedusing the elasto-plastic non-linear codes.Typical of the elasto-plastic non-linear codesused in the analysis of embankments areDIANA (Kawai, 1985), ANSYS (Swanson,1992), FLAC (Cundull, 1993), etc. Theconstitutive models used in these codes varyfrom simple hysteretic non-linear models tomore complex elasto-kinematic hardeningplasticity models. Compared to the EQLcodes, the elasto-plastic non-linear codes aremore complex and put heavier demand oncomputing time. However, they providemore realistic analyses of embankmentsunder earthquake loading, especially understrong shakings. Critical assessments of non-linear elasto-plastic codes can be found inMarcuson et al. (1992) and Finn (1993,2000).

C3.3.5.2 Effective Stress CodesEffective stress codes allow modelling porepressure generation and dissipation inmaterials susceptible to liquefaction and thusto obtain better estimates of permanentdeformations under seismic loading. Theeffective stress codes may be divided intothree main categories: fully coupled, semi-coupled and uncoupled.

(a) Fully coupled codes

In the fully coupled codes, the soil is treatedas a two-phase medium, consisting of soiland water phases. Two types of porepressures are considered, transient andresidual. The transient pore pressures are

related to recoverable (elastic) deformationsand the residual pore pressures are related tonon-recoverable (plastic) deformations. Amajor challenge in fully coupled codes is topredict residual pore pressures. The residualpore pressures, unlike the transient porepressures, are persistent and cumulative andthus exert a major influence on the strengthand stiffness of the soil skeleton. Thetransient pore pressures are cyclic in natureand their net effect within one loading cycleis often equal to zero. An accurate predictionof residual pore pressures requires anaccurate prediction of plastic volumetricdeformations. In the fully coupled codes thisis often achieved by utilizing elasto-plasticmodels based on kinematic hardening theoryof plasticity (utilizing multi-yield surfaces)or boundary surface theory with a hardeninglaw. These models are very complex and puta heavy demand on computing time.

Generally speaking, fully coupled predictionof pore pressures under cyclic loading isvery complex and difficult. The validationstudies performed on a number of thesecodes suggest that the quality of responsepredictions is strongly path dependent.When the loading paths are similar to thestress paths used in calibrating the models,the predictions are good. As the loading pathdeviates from the calibration path, thepredictions become less reliable. Apart fromthe numerical difficulties, part of thisunreliability is also due to the poor or lessthan satisfactory characterization of the soilproperties required in the models. Forinstance, because of sampling problems, it isoften very difficult to accurately determinevolume change characteristics of loose sandsas required by these models. In general, theaccuracy of pore pressure predictions infully coupled modes is highly dependentupon the quality of the input data.

Fully coupled codes include DNAFLOW(Prevost, 1981, 1988), DYNARD (Moriwakiet al., 1988); PLAXIS (2012); ABAQUS(2012) and FLAC (Itasca, 2011).

As described in Perlea and Beaty (2010),FLAC contains a number of general purpose

DRAFT



constitutive models but also provides for theuse of user-defined constitutive models.These include a bounding surface hypo-plasticity model (Wang and Makdesi, 1999)by AMEC Geomatrix Inc.; FLAC-UBCSAND, (Beaty and Byrne, 2008) andPM4Sand (Dafalais and Manzari, 2004;Boulanger, 2010; Boulanger and Ziotpoulou,2012; Ziotopoulou and Boulanger, 2012).PLAXIS and ABAQUS also offer userdefined constitutive models such asUBCSAND.

(b) Semi-coupled codes

Compared to the fully coupled codes, thesemi-coupled codes are more robust and lesssusceptible to numerical difficulties.However they are theoretically less rigorous.In these codes empirical relationships, suchas those proposed by Martin et al. (1975)and Seed (1983), are used to relate cyclicshear strains/stresses to pore pressures. Theempirical nature of the pore pressuregeneration in these codes generally puts lessrestriction on the type of plasticity modelsused in the codes. The semi-coupled codesare in general less complex andcomputationally demanding. The parametersthey require are often routinely obtained inthe laboratory or in the field.

Semi coupled codes include DESRA-2 (Leeand Finn, 1978), DSAGE (Roth, 1985), andTARA-3 (Finn et al., 1986). FLAC (Cundall,1993, Itasca, 2011) can also be used in thismode as in the URS cycle weighted modeldescribed in Perlea and Beaty (2010)

C3.3.5.3 SummaryMore advanced dynamic methods arepotentially expensive and should only bedone by very experienced persons whounderstand the limitations of the analysisand the need to use well consideredproperties. The analysis methods arecontrolled by the quality of data put intothem, the limitations of the methodsthemselves and particularly of those doingthe analysis.

They are in most cases only to becontemplated for dams and foundations

subject to liquefaction, and where the extrarefinement of the analyses are warranted;e.g. where the simplified methods are notreally applicable, and / or are resulting inmarginal assessments on whether remedialworks are required.

Any non-linear deformation analysis needsto be documented in sufficient detail that itcan be reasonably scrutinized. All inputparameters, all numerical details (boundaryconditions, damping parameters, etc), andenough output plots to evaluatereasonableness of initial conditions anddynamic responses. These should bereviewed by person’s expert in the practice.

Perlea and Beaty (2010) give an overview ofUSACE practice and examples of use ofmany of the programs described above.Stark et al (2012), Friesen and Balakrishnan(2012) give useful examples.

C3.4 Assessment of the Effects ofLiquefaction in Embankment Damsand Their Foundations

C3.4.1 Overall ApproachNo commentary.

C3.4.2 Definitions and the Mechanics ofLiquefaction

It should be noted that there are nouniversally accepted definitions forliquefaction. Those provided in theGuideline have fairly wide acceptance.

The definitions for liquefaction, initialliquefaction, flow liquefaction andtemporary liquefaction are based on USNRC(1985), Robertson and Fear (1995),Robertson and Wride (1997), Yamamuroand Lade(1997).

It is important to recognize that thesephenomena apply to monotonic (static) aswell as cyclic loading and are apparent incontractive soils (e.g. in loose fills) , bothcohesionless and those with very lowplasticity.

Robertson and Fear (1995) and Robertsonand Wride (1997) defined the cyclicliquefaction and cyclic mobility terms.

DRAFT



Idriss and Boulanger, (2008) defined cyclicsoftening.

The mechanics of liquefaction are describedin USNRC (1985) and Fell et al (2005,2015). However the description in Idriss andBoulanger (2008) is more complete and upto date.

It is essential that those carrying outliquefaction assessments read Idriss andBoulanger (2008) because it will assist themunderstand the background to the simplifiedprocedure for liquefaction assessment andthe way laboratory performance has beenused to develop the method alongside fielddata.

C3.4.3 Methods for Identifying Soilswhich are Susceptible toLiquefaction

(a) Geology and age of the deposit.

Idriss and Boulanger (2008) refer to Youdand Perkins (1978) and indicate that the soilswhich are most susceptible to liquefactionare non-plastic or very low plasticity fill,alluvial, fluvial, marine and deltaic soils.They indicate soils are most susceptiblewhen they are recently deposited, becomingmore resistant to liquefaction as theybecome older.

The most susceptible soils are Holocene oryounger soils, Pleistocene soils are low orvery low susceptibility except for loess.

Youd et al (2001) discuss the effects of theage of the deposit and conclude that forman-made structures such as fills and

embankment dams, aging effects areminimal, and corrections for age should notbe applied in calculating liquefactionresistance.

Andrus et al (2009) and Hayati and Andrus(2009) present methods for correcting for theage of the soils being assessed forliquefaction. These are discussed further inSection C3.4.4.3. These include datashowing liquefaction has occurred inPleistocene sands.

It is concluded that geological age alonecannot be used as a means of screening forliquefaction susceptibility. It may be takenas additional information to be considered inassessing the likelihood of liquefaction.

(b) Soil gradation, plasticity, moisturecontent

There are a number of methods available toidentify soils which are susceptible toliquefaction based on the particle sizegradation, plasticity and in-situ moisturecontent of the soil. These include:

(i) Seed et al (2003) criteria(ii) Robertson and Wride (1998) soil

behaviour type index Ic for ConePenetration Tests

(iii)Boulanger and Idriss (2006)(iv)Robertson (2010) for Cone

Penetration Tests

It is recommended that at least two andpreferably more of the methods be used toassess the likelihood the soils are potentiallyliquefiable. A suggested approach for doingthis is given in Table C3.1.

Table C3.1 Suggested method for assessing the likelihood that a soil is potentiallyliquefiable

Result of assessment Likelihood this strata is potentiallyliquefiable

Three methods indicate liquefiable Highly likely to certainTwo methods indicate liquefiable Likely

Only Seed et al (2003) indicates liquefiable,others do not

Likely to PossibleDepending if Zone A or Zone B

Only Robertson and Wride (1998) indicatesliquefiable, others do not.

Unlikely

Three methods indicate non liquefiable Liquefaction can be discounted

DRAFT



It should be noted that where soils plot inZone B of the Seed et al (2003) method theyshould be considered as potentiallysusceptible to liquefaction and /or cyclicsoftening. If their behaviour becomes criticalto the assessment of the safety of the dameither the soil should be assumed to beliquefiable or undisturbed samples should besubject to cyclic loading tests rather thanrelying on semi-empirical methods to predictbehaviour.

The following points should be noted:

(a) The Seed and Idriss (1982) ChineseCriteria method has been usedextensively in Australia but often thelast criteria that the natural moisturecontent is ≥ 0.9 times the liquid limithas been ignored which mean thatlower moisture content soils havebeen incorrectly identified aspotentially liquefiable.

(b) Seed et al (2003) and Boulanger andIdriss (2007), Idriss and Boulanger(2008) all indicate that the ChineseCriteria (Seed and Idriss, 1982) areoutdated and should not be used. It isrecommended that this advice beheeded.

(c) Seed (2010) discusses the Boulangerand Idriss (2006) method and plots

field case data which he indicatesshows the Boulanger and Idriss(2006) criteria are non-conservative,with case data showing liquefactionof soils occurring with PI> 7 contraryto the Boulanger and Idriss (2006)criteria. Given these data it wouldappear that caution should be appliedin relying on the Boulanger andIdriss (2006) criteria alone to identifysoils which may liquefy.

(d) Fell (2012) notes that for one damwhere foundation liquefaction wasthoroughly assessed using SPT andCPT data the Robertson and Wride(1998) method identified a number ofstrata as non-liquefiable where theSeed and Idriss (1982) and Seed et al(2003) methods both indicated thesoil was potentially liquefiable.Given this caution should be appliedin relying only on the Robertson andWride (1988) or Robertson (2010)CPT based methods and possiblyother CPT and CPTU based methodsalone.

(e) Particle size distribution alone is notable to identify soils which aresusceptible to liquefaction. Figures11 and 12 in ANCOLD (1998) areoutdated and should not be reliedupon. Figure C3.3 is a better guide.

0

20

40

60

80

100

0.001 0.01 0.1 1 10 100

Particle size (mm)

Per

cent

pass

ing

(%)

SandSilt GravelClay

13

4

5

2

1 - Coarse grained coal mine waste (Dawson et al 1998; Taylor 1984; Bishop et al 1969; Hutchinson 1986)2 - Loose silty sand fills, Hong Kong (upper and lower quartile of pre 1977 fills (HKIE 1998))3 - Hydraulically placed mine tailings and fills in dam embankments (various published sources)4 - Sensitive clays (indicative limits from: Lefebvre 1996; Bentley & Smalley 1984; Mitchell & Markell 1974; Hutchinson 1961, 1965)5 - Sub-aqueous slopes, natural and fill slopes (Koppejan et al 1948; Kramer 1988; Sladen & Hewitt 1989; Cornforth et al 1974)

Approximate bounds of fillssusceptible to static liquefactionand flow sliding

DRAFT



Figure C3.3: Particle size distributions of material types susceptible to liquefaction and flowsliding (Hunter and Fell 2003)

C3.4.4 Methods for Assessing WhetherLiquefaction may occur

C3.4.4.1 IntroductionWhether soils are likely to liquefy underseismic loading is almost always carried outusing the “simplified method” firstdeveloped by Seed and Idriss (1971). In theANCOLD (1998) guideline the methodreferenced was Seed and De Alba (1986)which was a refinement of the Seed andIdriss (1971) method. Since then there havebeen further developments in the details ofthe methods but the basic approach remainsthe same as described in Section C3.4.4.2.

Youd et al (2001) published a summaryreport on the 1996 NCEER and 1998NCEER/NSF Workshops on valuation ofLiquefaction Resistance of Soils. This was aconsensus of the 20 experts who attended theWorkshops and has been widely used forassessing liquefaction since it was published,including by the Australian dam community.

Seed et al (2003) prepared a University ofCalifornia at Berkeley Report whichpresented updated methods based onenlarged databases of cases, and using aprobabilistic approach.

Idriss and Boulanger (2008) from Universityof California at Davis produced MonographMNO-12 for the Earthquake EngineeringResearch Institute (EERI). This reviewed theavailable literature and updated many aspectsof the method.

Seed (2010) issued a University of Californiaat Berkeley Geotechnical Report which wascritical of a number of aspects of Idriss andBoulanger (2008).

Idriss and Boulanger (2010) responded tothis criticism and provided more details onthe basis of their EERI Monograph.

As a result of the differences of opinionEERI set up an ad hoc committee to reviewwhat they termed the “strong differences of

opinion, often personalised and polarized”.This is reported in Finn et al (2010).

Youd (2010) provided some personalinsights to the discussion.

Cox and Griffiths (2011) contributed bycomparing the use of the Youd et al (2001),Seed et al (2003) and Idriss and Boulanger(2008) approaches on three sites.

As a result of these developments it has beennecessary in preparing these guidelines toreview these documents and provideguidance on which method or methodsshould be used, and details which need to beconsidered.

Unlike the ANCOLD (1998) guidelinedetails of the suggested method are notincluded here. Those doing liquefactionassessments should instead work from thepaper, Monograph or Report so they haveavailable the background to the methods.

It has been proposed by Finn et al (2010) thata Workshop similar to the 1996 Workshop beconvened. Practitioners should seek thatpublication if and when it is produced. In themeantime they should keep up to date withthe literature particularly that in high qualityrefereed journals.

Since the drafts of these guidelines wereprepared there have been a number ofpublications by Idriss, Boulanger, and others.These update and refine the methods but donot significantly alter the outcomes so noattempt has been made to review them here.

C3.4.4.2 Outline of the “SimplifiedMethod” for AssessingLiquefaction.

The method requires the calculation orestimation of two variables for evaluation ofliquefaction resistance of soils:

(a) The Cyclic Stress Ratio, CSR, whichis a measure of the cyclic loadapplied to the soil by the earthquake.

DRAFT



(b) The Cycle Resistance Ratio, CRR,which is the capacity of the soil toresist liquefaction. The CRR isestimated from Standard PenetrationTests (SPT), Cone Penetration Tests(CPT) or, less frequently, the shearwave velocity.

(c) If the CSR is greater than the CRR,liquefaction is likely to occur.

The Cyclic Stress Ratio, CSR, is calculatedfrom:

dvovomaxvoav r)/)(g/a(65.0)/(CSR s¢s=s¢t=

Where amax = peak horizontal acceleration atthe ground surface generated by theearthquake; g = acceleration of gravity; svo

and s′vo are total and effective verticaloverburden stresses, respectively and rd =stress reduction coefficient.

The CRR for M7.5 earthquakes can beestimated from the curves in Figure C3.4. Itis usually recommended that the SPT CleanSand Base Curve be used, with correction forfines content to give (N1)60CS. There aresimilar plots for CPT and shear wavevelocity.

In this figure, (N1)60 is the SPT blow countnormalised to an effective overburdenpressure of 100 kPa, a hammer energy of60%, borehole diameter, rod length andsampling method.

(N1)60 is given by:

(N1)60 = Nm CN CE CB CR CS

Where Nm = measured standard penetrationresistance; CN = factor to normalize Nm to acommon reference effective overburdenstress; CE = correction for hammer energyratio (ER); CB =correction factor forborehole diameter; CR = correction factor forrod length; CS = correction for samplers withor without liners.

To allow for fines content the (N1)60 valuesare adjusted to (N1)60cs using:

(N1)60CS = a + b(N1)60

a and b vary with fines content.

DRAFT



Figure C3.4 SPT Clean-sand Cyclic Resistance Ratio (CRR) curve for Magnitude 7.5Earthquakes with data from liquefaction case histories (Youd et al., 2001, modified from Seed etal., 1985).

Magnitude scaling factors MSF

The procedures outlined above give CRR7.5,cyclic resistance ratios for M7.5earthquakes.

Smaller magnitude earthquakes giving thesame peak horizontal acceleration are lesslikely to initiate liquefaction, because theearthquake will have fewer cycles of motion.Larger magnitude earthquakes are morelikely to initiate liquefaction. This is allowedfor by using a magnitude scaling factor(MSF) in the equation for factor of safety(FS) against liquefaction:

FS = (CRR7.5/CSR) MSF

Where CSR = calculated cyclic stress ratiogenerated by the earthquake shaking; andCRR7.5 = cyclic resistance ratio formagnitude 7.5 earthquakes.

Corrections for overburden stresses, staticshear stresses Ks and Ka

The method of assessment of liquefactionpotential described above is for horizontal or

gently sloping ground and for depths lessthan about 15 m.

For assessments of liquefaction forembankment dams, the confining stressesmay be higher than this and the dam willimpose static shear stresses, which may alterthe liquefaction potential.

These are allowed for by using the factorsKs and Ka to allow for the high overburdenstresses and static shear stress respectively.They are applied by extending the equationabove to:

FS = (CRR7.5/CSR).MSF. Ks. Ka

There are three most commonly usedmethods based on SPTs and CPTs:

· Youd et al (2001)· Seed et al (2003), Cetin et al (2004)

and Moss et (2006)· Idriss and Boulanger (2008),

Boulanger and Idriss (2012)

Much of the discussion about the methodrelates to how the stress reduction factor rd,magnitude scaling factor MSF, and

DRAFT



corrections for high overburden stresses andstatic shear stresses Ks and Ka are treated.There are also differences of view regardingthe CRR plot.

C3.4.4.3 Discussion of Differences betweenthe available SPT and CPT BasedMethods

Those who wish to understand thebackground to the methods should read thereferences in Section C3.4.4.1.

Cox and Griffiths (2011) used the three SPTbased methods to analyse three case studies.They found that they did not give greatlydiffering factors of safety againstliquefaction in the upper 24m of the soilprofiles. They found that the Idriss andBoulanger (2008) method typically yieldsthe highest factor of safety at depth,generally ranging from 5-15% higher thanthe others, but the differences were muchgreater at raw blow counts greater than 23.This was attributed to Idriss and Boulanger(2008) using higher Ks values but moreimportantly higher CN values. Thesecombined with the shape of the CRR curveat high (N1)60CS to give the largerdifferences.

They found that in the upper soil profile theindividual inputs to the methods varied quitea lot but the effects were often balanced byother factors. In particular they found thatthe quite different rd values did not result ingreatly different factors of safety.

This is consistent with the advice from Idrissand Boulanger (2008) that because the casehistories used in the development of theirmethod were analysed using their methodsfor accounting for depth effects, earthquakemagnitude and fines content, it is importantthat when using their method the samemethods as they used for these effects areadopted.

It is apparent that this would also apply ifusing the Seed et al (2003) method. Inparticular values of rd , CN , and Ks shouldnot be substituted from one method toanother.

Overall it is concluded that:

1. All three SPT based methods may beused and be expected to result insimilar factors of safety againstliquefaction in the upper 15 metres.

2. The Idriss and Boulanger (2008),Boulanger and Idriss (2012) SPTbased method incorporates manyrefinements developed since theYoud et al (2001) method waspublished and its use is preferred.However where Ks values are criticalto the outcome both the Idriss andBoulanger (2008) and Youd et al(2001) / Hynes and Olsen (1999)values should be used, and where theIdriss and Boulanger (2008) valuesare controlling the outcome of theassessment of the likelihood ofliquefaction expert advice should besought.

3. As can be seen in Figures C3.5 andC3.6 there is not a large differencebetween the NCEER/NSFWorkshops (Youd et al (2001) andIdriss and Boulanger (2004, 2008)plots. Within the context of theuncertainty in ground motions thedifferences are small and will makelittle difference in a risk basedanalysis. The Cetin et al (2004) (Seedet al 2003) curve plots to the right ofthe others. This apparent significantdifference between the three authorscan partly be explained by the way inwhich the database has been analysedby Cetin et al (2004). However Idrissand Seed (2010, 2012) present adiscussion of the data points whichcontrol the position of the curve.They cast doubt on the validity of 8key points on the grounds ofincorrect assignation of liquefaction /no liquefaction of four points andsignificant numerical errors betweenthe rd values used to develop theircorrelations and the rd valuescomputed using their applicableequation for four others. In view ofthis the Idriss and Boulanger (2004,

DRAFT



2008), Boulanger and Idriss (2012)method is preferred.

4. There is less information available tomake an assessment of CPT basedmethods. However on what ispresented all three CPT basedmethods may be used for assessingfactors of safety against liquefactionin the upper 15 or 20 metres.

5. It is apparent there is considerableuncertainty in the estimation of Ka. Itis suggested that Idriss andBoulanger (2008) be used but

marginal cases should be assumedliquefiable, or laboratory tests andnumerical analysis of stresses shouldbe carried out to assess the effects.

6. The simplified methods are notconsidered reliable below about 15 or20 metres. Most of the uncertaintydiscussed above is below that depth.For important decisions relating tomarginally liquefiable soils belowthat depth expert advice should beobtained.

Figure C3.5 Curves relating cyclic resistance ratio (CRR) to (N1)60CS values for earthquakemagnitude Mw =7.5. (Idriss and Boulanger (2008).DRAFT



Figure C3.6 Curves relating Cyclic Resistance Ratio (CRR) to (qc1N)cs values for earthquakemagnitude Mw =7.5. (Idriss and Boulanger (2008)).

7. There is considerable uncertainty inmany of the parameters used in thesemethods. Users should not regard theoutcomes as clearly liquefiable or notliquefiable for soils plotting near theboundaries. This can be readilyaccounted for in a risk basedframework as these ANCOLDGuidelines are recommending.

C3.4.4.4 Shear Wave Velocity ( VS) BasedMethods

Youd et al (2001), Seed et al (2003) andIdriss and Boulanger all refer to Andrus andStokoe (2000). Kayen et al (2012) havegathered more case data and have developeda probabilistic approach based on this.

Idriss and Boulanger (2008) point out therelative lack of sensitivity of VS to changesin relative density. A relative density rangeof 30% to 80% results in about 7 x differencein SPT “N” value, 3 x in CPT cone resistanceand only 1.4 x in VS. As a result greaterweight should be placed on SPT and CPTbased methods.

This fundamental constraint means themethod is mostly restricted to use in gravelly

soils where SPT and CPT samplers areaffected by the gravel particles.

C3.4.4.5 Allowance for the Age of the SoilDeposit

There are a number of authors who havepresented data to support the conclusion thatthe age of the soil deposit has an effect onthe resistance of the soil to liquefaction.These include Youd and Perkins (1978),Seed (1979), Arango et al (2000), Leon et al(2006), Andrus et al (2009) and Hayati andAndrus (2009).

Andrus et al (2009) present an alternativemethod based on measured to estimatedshear velocity VS. They include data whichshows how the shear wave velocity VS andCPT and SPT can give quite differentassessments of liquefaction potential inPleistocene and Tertiary sands.

It is suggested that these methods may beconsidered as additional information whereSPT and CPT based methods are showingmarginal potential for liquefaction inPleistocene and Tertiary sands. However themethods are very approximate and not toomuch reliance should be made on them.

DRAFT



They should also be considered forapplication in assessing the liquefactionpotential of recently deposited mine tailingsand dredged fills, as there is evidence theresistance of these to liquefaction may belower than for the soils in the databases uponwhich the CRR are calculated.

C3.4.4.6 Use of Becker Penetration TestIn sandy gravel and gravelly sand soils theSPT sampler may hit gravel size particlesand register high blow counts even if the soilis not dense. In USA and Canada for thesesoils the Becker Penetration Test issometimes used.

The Becker Penetration sampler is 168mmdiameter compared to 35mm diameter in theSPT sampler, and is driven by a down-holediesel hammer. A modern improvement ofthe equipment is described in Ghafghazi et al(2014a, b) and DeJong et al (2014, 2016a)

Some methods for interpreting the results aregiven in Sy and Campanella (1994), andGhafghazi et al (2014a, 2016b).

C3.4.5 Methods for Assessing the Strengthof Liquefied Soils in theEmbankment and Foundation

C3.4.5.1 IntroductionAs discussed in Idriss and Boulanger (2008),Seed et al (2003) and Seed (2010) theresidual shear strength that the liquefied soilmobilizes in the field is affected by voidredistribution, particle intermixing and otherfield mechanisms which are not replicated inthe laboratory.

As a result they conclude that the onlypractical methods for estimating the liquefiedstrength Sr is by back analysis of case studiesand relating these to SPT and CPT data.However there are not a large number ofcases available, and often for these the SPTand CPT data is variable and sometimes oflimited quality, and / or the geometry of theembankment and the stratigraphy of thefoundation is complicated. Olsen and Stark(2002) also describe some of these issues. As

a result considerable judgement is exercisedby the authors in developing the methods,and the results are uncertain, with a widerange of strengths possible.

The issue is further complicated becausethere are two schools of thought:

(a) Those such as Seed et al (2003), Seed(2010) who believe that the liquefiedresidual strength Sr should be relatedto the SPT (N1)60CS value directlywith no allowance for the effectiveoverburden stress. This is termed a“critical state” based method.

(b) Those such as Olsen and Stark (2002)who believe the residual strengthratio Sr /s′vc should be used andrelated to the SPT (N1)60CS value.That is the residual strength iscontrolled by the effectiveoverburden stress. Idriss andBoulanger (2008) and Robertson(2010) have developed such methods.Figure C3.7 shows the Idriss andBoulanger (2008) data andrecommended curves. This is termedthe “normalised strength ratio”approach.

C3.4.5.2 DiscussionThere is no consensus on which of the twoapproaches is best.

Idriss and Boulanger (2008) provide bothmethods without guidance on which to adoptbut say they believe the residual strengthratio method provides a better representationof the potential effects of void redistributionthan the liquefied residual strength, while itis recognised that neither correlation fullyaccounts for the numerous factors thatinfluence void redistribution processes.

DRAFT



Figure C3.7. Liquefied residual strength ratio Sr/s′vc versus SPT (N1)60CS value Idriss andBoulanger, 2008).

Seed (2010) strongly criticises the Idriss andBoulanger (2008) method, in particular the“recommended curve for conditions wherevoid ratio redistribution effects are expectedto be negligible”. He argues that three of thedata points use incorrect and over stated(N1)60CS ratios. He argues that the case datadoes not support this curve.

Youd (2011) discusses this and says thatIdriss and Boulanger unequivocally state thatthe upper curve is based on laboratory dataand was not guided by the points withincorrect strength in any way. He also saysthat a transformation from contractive todilative behaviour occurs at about a correctedSPT blow count of 15, and that should beaccompanied by a major increase in shearstrength. He indicates he believes Idriss andBoulanger are “on the right track”.

The Robertson (2010) CPT based methodhas the limiting value of liquefied undrainedstrength ratio is assumed to be 0.4 at Q1n,cs=70. This is based on the soils being dilativeat Q1n,cs > 70 so not being subject to flowliquefaction.

Baziar and Dobry (1995), Ishihara (1993)and Cubrinovski and Ishihara (2000a, b)have investigated the boundary of flowliquefaction conditions. Figure C3.8summarizes the outcomes. These support theuse of larger liquefied strengths for soils withhigh (N1)60 values at least above (N1)60 = 15.

Hence if the liquefied strength of such soilsis controlling the assessment of the safety ofthe dam, and large costs are potentiallyinvolved in remedial works, it may bewarranted to seek expert advice from one ofthe experts in the field.

DRAFT



Figure C3.8. Comparison of flow liquefaction boundaries in terms of SPT (N1)60 for sands andsilty sands from monotonic laboratory undrained tests and earthquake triggered field case(Hunter and Fell, 2003).

Given the lack of consensus on whichapproach to use (Critical state or normalisedstrength ratio) it is suggested that bothmethods be used for assessing the liquefiedresidual strength, and the results accountedfor assuming they are equally valid.

Either the Idriss or Boulanger (2008), orSeed and Harder (1990) / Seed (2010) orOlsen and Stark (2002) methods may beused. These may be compared to theRobertson (2010) method.

The majority of foundation liquefaction casesin Australia will satisfy the requirements forIdriss and Boulanger (2008) “recommendedcurve for where void redistribution effectscould be significant” because the liquefiablelayer is overlain by a lower permeabilitystrata. In view of this and the reservations ofSeed (2010) it is suggested that only thiscurve (not the upper one) be used unlessexpert advice is sought.

An important issue is what liquefied residualstrength to assume for design of stabilisingberms for existing dams, or for the design ofnew dams.

Olsen and Stark (2002) indicate that in theirview, at least for silty sands with > 12%fines, it is reasonable to allow for theincrease in Sr which would be indicated by

the increase in svc′ from the berm. For highrisk projects they suggest laboratoryconsolidation tests be carried out to confirmthat it is parallel to the steady state linewhich is implicit in the assumption that Sr

/svc′ is constant. NSFW (1998) cautionagainst the use of Sr /s′vc ratios, particularlyfor clean sands. Given this and theuncertainty in which approach is most validit would appear unwise to allow for thestrength increase from the increased effectivestress of the berm and the Sr obtainedwithout the berm should be assumed to applyafter the berm is built, unless expert opinionadvises otherwise.

In all cases it should be recognised that thedata indicates strength lower than thesuggested design values are possible, soeither a significant factor of safety is allowedfor or the uncertainty modelled in risk basedassessments.

C3.4.6 Methods for Assessing the Post-Earthquake Strength of Non-Liquefied Soils in the Embankmentand Foundation

C3.4.6.1 Saturated Liquefiable SoilsSoils which are potentially liquefiable butwhich have a factor of safety > 1.0 for theseismic load being considered will generate

DRAFT



pore pressures under the cyclic loading.These pore pressures should be allowed forin post-earthquake stability analyses. Idrissand Boulanger (2008) refer to Marcuson et al(1990) for level ground conditions but warnthat pore pressures will be higher forsituations where static shear stresses arepresent.

The Marcuson et al (1990) data may be usedto give a guide to what may happen. If thesebecome critical, and the project issufficiently important seek expert advice.

C3.4.6.2 Cyclic softening in Clays andPlastic Silts

Boulanger and Idriss (2006), Idriss andBoulanger (2008) and Bray and Sancio(2006) describe how to estimate the effectsof cyclic softening of clays and plastic silts.The discussion relates to saturated soils withover-consolidation ratios of 1 to 4. Such soilsdo occur in association with liquefiable soils,and may also occur as mine tailings.

The boundary between which this behaviourand liquefaction occurs is discussed inSection C3.4.3.

C3.4.6.3 Well Compacted Plastic and Non-Plastic Soils

Fell et al (2005, 2015) suggest that:

(a) For clay soil zones in theembankment and foundation assignstrength and pore pressures consistentwith the soil’s behaviour in staticloading after being cracked anddisturbed by the earthquake. If theclay is contractive in nature, useundrained strengths. If it is dilative onshearing, use effective stressstrengths c′, f′. Usually there will besome cracking and loosening, and ifso fully softened strengths, wouldapply (e.g. c′ = 0, f′ = f′peak). Someapply an arbitrary 10% or 15% loss ofstrength.

(b) For well compacted free drainingrockfill filters and dense sands andgravels adopt the effective stressstrengths c′, f′, with no change in thepore pressures. If large deformations

are expected in the earthquake thedense granular materials may haveloosened and will have a strengthapproaching the critical state strength,rather than the peak strength. Inpractice this can be accommodated bya small reduction from the expectedpeak strengths.

C3.4.7 Site Investigations Requirementsand Development of GeotechnicalModel of the Foundation

The following are some suggestionsregarding site investigations of foundationsof dams for assessing the likelihood andconsequences of liquefaction.

(a) Review of available data.

Gather together all available data on thefoundations including for existing dams,site investigations and laboratory testingcarried out prior to construction;geological mapping, reports, andphotographs taken during construction;monitoring data; and site investigationsand laboratory testing carried out sincethe dam was constructed.

(b) Preliminary geotechnical model andplanning site investigations.

Draw plans and sections of the availabledata, taking account of the depositionalenvironment of the foundation soils.

From this develop an interpretivegeotechnical model of the strata in thefoundation, and summarize the propertiesin a form useful for liquefactionassessment.

Assess the adequacy of the available dataand plan site investigations to supplementthis data.

(c) Site investigation methods.

Foundations are best investigated by bothboreholes in which SPT tests are carriedout on potentially liquefiable soils, andundisturbed thin wall tube samples takenof other strata; and by Cone PenetrationTests (CPT, CPTU). This is desirable

DRAFT



because CPT are best able to detectlayering in the strata, but SPT andlaboratory tests are required to providedata for some parameters, and to makeavailable two methods for assessing theliquefaction potential and post-earthquake liquefied strengths.

Some of these (SPT and CPT) should bedone adjacent each other for correlationpurposes.

Boreholes should where practicable bedrilled using wash boring methods withcasing and drilling mud support. Toprevent “blow in” at the base of the holewith resulting loosening of the soil inwhich SPT are carried out.

Wash boring should not be used for holesdrilled through the core of a dam becausehydraulic fracture may be induced by thedrill water or mud. If hollow flight augersare used (as will be necessary for holesdrilled through an existing dam) takemeasures to prevent “blowing” of thestrata at the base of the augers withsubsequent reduction in SPT blowcounts.

Sonic drilling may be used but may resultin densification of the soil below thecasing.

Take samples from all SPT tests inpotentially liquefiable strata and test formoisture content and fines content (%passing 0.075mm sieve). Forrepresentative samples test for Atterberglimits.

Record the level of the water table duringdrilling, and preferably installpiezometers to monitor groundwaterpressures at the time of the investigationsand as they fluctuate with river flows.

For SPT tests in gravelly soils take blowcounts for each 75mm as an aid todetecting interference by gravel particlesand allow adjustment to allow for this.For gravelly soils consider the use of

shear wave velocity in addition toboreholes with SPT.

The energy rating of the SPT hammercan be anything from around 55% to 90%to the energy rating of the hammer(s)used for the investigation should bemeasured during the site investigations. Itis not sufficient to use an energy ratingfor the hammer determined elsewherebecause it varies with the site conditionsand the drill equipment.

Past practice has been to not measure theenergy rating and assume 60%. This isnot considered good practice and ispotentially conservative in many cases sothe expenditure to measure the energyrating is warranted.

For CPT and CPTU tests in layered soilsadjust the data for the effects of soft andstiff layers as detailed in Figure 7 ofYoud et al (2001).

It should be noted that it is almostimpossible to take undisturbed samplesof non-plastic soils such as thosepotentially liquefiable soils in thefoundations or mine tailings. Thesampling process disturbs and potentiallycompresses the soil. Even specialisedsamplers are likely to result indisturbance. It is for this reason mostanalyses are based on in-situ testing.

(d) Geotechnical model

Draw plans and sections of the availabledata including the data from the newinvestigations, taking account of thedepositional environment of thefoundation soils.

From this develop an updated interpretivegeotechnical model of the strata in thefoundation, and summarize the propertiesin a form useful for liquefactionassessment.

Take particular attention to the continuityand aerial extent of potentially liquefiablestrata.

DRAFT



(e) Laboratory cyclic shear testing.

For many of the dynamic numericalanalyses methods cyclic simple shear andcyclic triaxial tests will be required toprovide inputs into the models. Expertadvice should be obtained on the natureand detailed requirements for such tests.

C3.4.8 Liquefaction AnalysisThe analysis of liquefaction potential is bestcarried out as follows:

(a) Assemble the SPT and CPT /CPTUdata for potentially liquefiable stratainto spreadsheets. In so doing assessthe data for quality andinconsistencies and remove or “tag”as unreliable poor quality orinconsistent data.

(b) For SPT include all “N” in thespreadsheet. For CPT / CPTU enterdata at 50 mm intervals as recorded.

(c) Analyse the data to determine if thesoils are potentially liquefiable usingthe criteria described above. Excludefrom further considerations soilswhich are not liquefiable, or whichare above the water table, and willnot be saturated under any condition.

(d) For the potentially liquefiable strata,asses the factor of safety againstliquefaction for a number of peakground accelerations within the rangeto be assessed for the site.

(e) From these data, plot plans andsections showing contours of theAnnual Exceedance Probability(AEP) of liquefiable zones.

When assessing whether the soil may besaturated a cautious approach should beadopted taking into account the following:

(i) For some dams the tailwater levelmay vary throughout the year andduring floods. As a result some soilsin the foundation and in theembankment which are potentiallyliquefiable may be below the watertable part of the time. It may beappropriate to assume the soils in thewater table range are sufficiently

close to fully saturated that they mayliquefy.

(ii) Soils in the capillary zone above thephreatic surface may be sufficientlysaturated to liquefy. However theheight of capillary rise is not likely tobe large for sandy soils.

(iii)As noted in Section 3.1, for tailingsdams using upstream or centrelineconstruction, there is a potential forliquefaction of loose to mediumdense partly saturated tailings whereperched water tables or nearlysaturated zones may exist above themeasured phreatic surface. This maylead to a series of persistent saturatedlayers within the tailings.

(iv)Hossain (2010), Yang et al (2004),Nakazawa et al (2004) andTsukamoto et al (2002) presentinformation which shows that thecyclic resistance of partially saturatedsands is higher than saturated sand,and that these can be related to thecompressional (“P” wave) velocity.

C3.4.9 Assessment of the Likelihood andthe Effects of Cracking ofEmbankment Dams Induced bySeismic Ground Motions

Fell et al (2008), Section 5.5 describes amethod for estimating the likelihood oftransverse cracks in the embankment causedby earthquake. It relies on the damageclassification method of Pells and Fell (2002,2003).

Fell et al (2008) uses these data to estimatethe likelihood of failure of the dam given thecracking.

These methods are empirical, and veryapproximate.

C3.5 Methods for UpgradingEmbankment Dams for SeismicGround Motions

C3.5.1 General approach

No commentary

DRAFT



C3.5.2 Embankment Dams not Subject toLiquefaction

No commentary.

C3.5.3 Embankment Dams Subject toLiquefaction

Mitchell (2008), Idriss and Boulanger (2008)and Seed et al (2003) discuss the variousground improvement methods and theirlimitations. These are summarized in TableC3.2.

The most widely used methods in Australiahave been “stone” columns, deep soilmixing, and removal and replacement.

“Stone” columns are the most common butthere are detailed design matters which needto be considered. These include:

(a) If the columns are situated on thedownstream side and there is nocutoff through the alluvium uponwhich the dam is constructed, thecolumns may acts as drains forseepage water. In this case theyshould be backfilled with sandygravel designed to act as a filter to thesurrounding soil, or backward erosionmay initiate into the columns. Thiscan create problems with theconstruction of the columns as thebackfill is low permeability. Forsome projects the backfill has beendesigned as a “some erosion” filter,using the Foster and Fell (2001)method.

(b) The composite strength of thecolumns and the surrounding soil isestimated allowing for the relativeareas of the columns and thesurrounding soil. The strength of the

surrounding soil should be taken asthe liquefied strength unless it can bedemonstrated that the soil betweenthe columns is sufficiently permeableto allow dissipation of pore pressuresbuilt up during the cyclic loading.This can be assessed by the methodof Seed and Booker (1977). Idriss andBoulanger (2008) indicate thatmethod has been refined by Onoue(1988), Iai and Koizumi (1986) andPestana et al (2000), but they cautionthat uncertainties in the hydraulicconductivities of the surrounding soil(particularly as it is disturbed byconstruction of the columns) and thecolumns themselves make it difficultto have confidence of efficientdrainage except in relatively highpermeability soils.

(c) Seed et al (2003) discuss the effect ofthe stiffer stone columns attractingthe cyclic shear stresses andpotentially reducing the build-up ofpore pressure in the liquefiable soil.They caution against relying uponthis because the columns still flex ifthey are longer than three times theirdiameter.

Some examples of seismic upgrades of damsare given in Davidson et al. (2003), Toose etal. (2007), Mejia (2005), Dise et al. (2004)and Luehring et al. (2001)DRAFT



Table C 3.2. Summary of ground improvement methods and their application for remedialworks for liquefiable foundations (Fell et al, 2015)

Method Method ofImprovement

Soils for whichit the method is

applicable

Limitations and Comments

Vibroflotation Densification,increased lateralstresses.

Sand, sand withsome silt,gravelly sand

Ineffective if silt content > 20%approx. May not penetrate gravellystrata.

Vibro-replacement

e.g. stonecolumns

Densification,increased lateralstresses,reinforcement,increaseddrainage.

Sand, silt, clay Ineffective densification if silt content> 20% approx. May not penetrategravelly strata. Columns may beconstructed by driving casing,removing soil from within casing, andcompacting “stone” as casing iswithdrawn. Treatment must penetratethrough liquefiable soil or treatmentwill be ineffective at the base.

Dynamiccompaction

Densification,increased lateralstresses.

Sand, silty sand Ineffective if silt content > 20%approx. Effective only in the upper10m. Quality control difficult.Vibration of adjacent structures.

Compactiongrouting

Densification,increased lateralstresses,reinforcement.

Sand, silt, clay Ineffective at depths < 6m. Augersmust penetrate through liquefiablesoil or treatment will be ineffective atthe base. Quality control may bedifficult.

Deep soilmixing

Reinforcement,reduce earthquakeinduced strainsand porepressures,

Sand, silt, clay Augers must penetrate throughliquefiable soil or treatment will beineffective at the base. Quality controlmay be difficult. Soil-cementelements may be brittle.

Jet grouting Reinforcement,reduce earthquakeinduced strainsand porepressures,

Sand, silt, clay Uses very high pressures so shouldnot be used where hydraulic fracturemay damage the embankment core.Difficult to control diameter oftreatment.

Verticaldrainage

Drainage Sand, sand withsome silt

Ineffective in silty sands and silts

Permeationgrouting

Reinforcement Sand, gravel,sandy gravel,gravelly sand

Particulate grouts e.g. microfinecement will not penetrate fine sandsbecause the sand acts as a filter.Chemical grouts expensive and sometoxic. Quality control difficult.

Removal andreplacement

Removesliquefiable soil

All soils Dewatering, stability duringconstruction may be problems. Giveshigh degree of confidence in finalproduct.

C3.6 Flood Retarding Basins

No commentary.

DRAFT

Commentary – 4.0 Assessment of Concrete Dams for Seismic Ground Motion


C4 ASSESSMENT OF CONCRETEDAMS FOR SEISMIC GROUNDMOTIONS

C4.1 Effect of Earthquakes on ConcreteDams

C4.1.1 GeneralIt was stated in ANCOLD (1998)“Guidelines for Design of Dams forEarthquake” and also ANCOLD (2013)“Guidelines on Design Criteria for ConcreteGravity Dams” but is worth repeating here,that concrete dams do fail. However, thereare no recorded cases where concrete damshave completely failed under earthquakeloading, with the sudden release of thereservoir. However, a number of concretedams have suffered very serious damage,including cracking right through theconcrete. The two best known dams areKoyna Dam (a 103m high gravity dam inIndia) which suffered major cracking throughthe upper part of the dam when subjected to aM6.5 earthquake in 1967 and Sefid Rud Dam(a 106m high massive buttress dam in Iran)which suffered major cracking in thebuttresses when subjected to a M7.3-7.7earthquake in 1990. A number of otherconcrete dams of all types have also suffereddamage under earthquake loading.Generally, the magnitude of the earthquakecausing damage has been M5.3 or greater.Damage has ranged from displaced copingsand minor cracking to major cracks withconsequent leakage from the dam. Hansenand Nuss (2011) and Wieland andChen(2010) give information on theexperiences of concrete dams subjected toearthquake loadings. Arch dams have tendedto perform well under earthquake loading.The 113m high Pacoima Dam (a doublecurvature arch dam) in California wassubjected to a ground shaking ofapproximately 0.5g (horizontal and vertical)during the Northridge Earthquake in 1994.The base accelerations were stronglymagnified to 2g horizontal and 1.4g verticalon the top of the left abutment: the damsuffered only minor damage.

The M7.6 Chi Chi Earthquake in Taiwan(1999) caused severe damage to the ShihKang Dam (a gated concrete gravitystructure) due to a fault rupture under thedam causing a 9m vertical differentialdisplacement of the dam. The rest of the damwhich was subjected to just the earthquakeshaking, performed well. Notwithstanding,the large differential displacement of thedam, there was no uncontrolled loss ofstorage so technically, the dam did not fail.

Major fault displacements directly under aconcrete dam are very difficult to design for.Consequently, it is not included in the scopeof these guidelines. It is covered by ICOLD(1998).

ANCOLD(1998) included a section“Analysis and Design of Concrete Dams”.However, that section was limited in that iteffectively covered only concrete gravitydams. Since that time, there have been anumber of earthquake analyses carried out inAustralia, of other types of concrete dams(e.g. Bendora Dam – a double curvature archdam in the ACT; Carcoar Dam – a doublecurvature arch dam in NSW; and OberonDam – a buttress dam in NSW; MoogerahDam – a double curvature arch dam inQueensland; Mt. Bold Dam-a gravity-archdam in SA, see McKay and Lopez (2015);Clover and Junction Dams-slab and buttressdams in Victoria). As there were no relevantlocal guidelines to cover these dams, resorthad to be made to reference material fromoverseas organisations (e.g. Federal EnergyRegulatory Commission (FERC) and the USArmy Corps of Engineers (USACE) in theUSA).

ANCOLD (1998) also had a number of othershortcomings in today’s (2016) context:

· They were oriented to safety reviews andto standards based acceptance criteria;

· Acceptance criteria only given for gravitydams (see discussion above);

· There was a detailed description givenfor a popular simplified methodology foranalysing 2d gravity dams (fenves &Chopra (1986) method) which was

DRAFT



probably unnecessary as documentationfor this methodology is readily availablevia the internet;

· little information was given regardinganalysis required for risk assessments(i.e. analysis required to help estimateconditional probabilities of failure ofconcrete dams);

· advances in computer software forundertaking sophisticated analysis of alltypes of concrete dams both in 2D and3D space since ANCOLD (1998) waspublished; and,

· recent publication of the updatedANCOLD “Guidelines on DesignCriteria for Concrete Gravity Dams”,(ANCOLD, 2013).

The Chapter of the Guidelines and thisCommentary relating to concrete damsattempts to resolve the above shortcomingsby:

· emphasising the strategy ofincreasing the complexity of analysisonly as the need requires;

· ensuring that the level ofsophistication of the analysis isconsistent with the availableinformation;

· providing more information on theanalysis of arch and buttress dams;

· providing more guidance on non-linear methods in recognition of moresophisticated software available.

C4.1.2 ICOLDIt should be noted that ICOLD has a numberof Guidelines relating to earthquake analysisand design of concrete dams. A number ofrelevant Guidelines published thismillennium are given below:

· Bulletin 62A (ICOLD, 2016b):Inspection of dams followingearthquakes - guidelines

· Bulletin 148 (ICOLD, 2013): Selectingseismic parameters for large dams-guidelines

· Bulletin 120 (ICOLD, 2001): Designfeatures of dams to effectively resistseismic ground motion

· Bulletin 123 (ICOLD, 2002): Earthquakedesign and evaluation of structuresappurtenant to dams

· Bulletin 155 (ICOLD, 2013): Guidelinesfor use of numerical models in damengineering

The above Bulletins should generally beconsidered as background information.

C4.2 Defensive Design Principles forConcrete DamsNo commentary.

C4.3 General Principles of Analysis forSeismic Ground Motions

C4.3.1 Gravity DamsANCOLD (2013) includes a short section onthe design and analysis of gravity dams forearthquake loading. That section is verygeneral in its nature and points to theseGuidelines for information regarding gravitydams subjected to earthquake loading.

Concrete gravity dams including rollercompacted concrete (RCC) gravity dams cangenerally by analysed in the first instance, astwo dimensional structures – see SectionC4.3.4 below for exceptions. This meansthat relatively simplistic analysis methodscan be used to estimate the effects ofearthquake loadings on these dams. Theseinclude the methodology of Fenves andChopra (1986) which will be discussed inSection C.4.4.3.

C4.3.2 Arch DamsArch dams include single curvature archdams (most of which were built in thiscountry, from the end of the 19th century tothe beginning of the 20th century), doublecurvature arch dams (which can range fromthin to thick cross sections) and arch gravitydams. The principles discussed in thischapter also apply to gravity dams which arein relatively narrow, steep sided valleys andare therefore should be considered as threedimensional structures. Due to the relatively

DRAFT



flexible nature of arch dams (especially thindouble curvature arch dams) and theirinherent 3D nature, it can be necessary toconsider factors not applicable to gravitydams or possibly, buttress dams. Theseinclude: water compressibility; dam-waterinteraction; wave absorption at the storageboundary and dam/foundation interaction. Itis usual practice to assume that thefoundation rock is massless andcompressibility of the water and dam-waterinteraction ignored if the fundamentalfrequency of the storage is greater than thatof the dam and foundations alone. Duron etal (1994) give a procedure for calculating thefundamental frequency of the storage.

C4.3.3 Buttress DamsButtress dams can range from mass buttressdams which are akin to concrete gravitydams to thin buttress dams which areessentially reinforced concrete structures.

Slab and buttress type dams can be sensitiveto cross valley shaking. The reinforcedconcrete slabs and the buttress corbelssupporting them have to be able to handleupstream/downstream shaking. However,the relatively thin buttresses and the struttingsystem between the buttresses have to beable to handle the cross valley shaking aswell as upstream/downstream shaking. Theinterface between the slabs and their supportshas to be able to safely transmit the shearforces between them.

Applying concrete buttressing to thedownstream face of an existing concretegravity dam is a common form ofraising/strengthening those dams (e.g.Burrinjuck Dam in NSW, the spillway atHinze Dam in Queensland, Wellington Damin WA). This buttressing can be eithercontinuous (Hinze and Wellington Dams) orintermittent (Burrinjuck Dam). Analysis ofthese dams for earthquake loadings (as wellas static loads applied post buttressing) has toconsider the stress state of the dam prior toapplying the buttressing (i.e. how compositeaction is set up between the original and newconcrete) as well as any other locked in

stresses that could exacerbate cracking whenearthquake stresses are added.

C4.3.4 Other TypesArch gravity dams and gravity dams built innarrow, steep sided valleys will be generallyreviewed according to the concrete gravityguidelines. However, 3D analysis willgenerally be required in addition to anysimplified 2D analysis to ensure thatundesirable stability conditions/stressregimes are not developed in the dam.

C4.4 Seismic Structural Analysis ofConcrete Dams

C4.4.1 Material Properties

C4.4.1.1 ConcreteThe compressive and tensile strengths ofconcrete increase as the rate of loadingincreases. The dynamic compressive andtensile strengths of concrete can therefore beexpected to be greater than the staticstrengths. As dynamic compressive stressesare rarely of concern, the allowablecompressive stress for static loading can beused also for dynamic loading. Raphael(1984) states that the apparent tensilestrength of concrete under seismic loadingwhich should be used with linear finiteelement analyses is given by:

fr= 0.65 fc2

where fc is the concrete compressive strengthin MPa and fr is the apparent seismic tensilestrength in MPa. Values given by thisformula are some 50% greater than theapparent tensile strength for static loading.Raphael suggests that fr is twice the splittingstrength of the concrete under static loading.Clough and Ghanaat (1993) suggest that theapparent dynamic tensile strength is about25% greater than the measured static valuewhich gives apparent tensile strength about20% of the standard compressive strength.They further suggest that there may be a 15to 20% loss of strength across lift joints.These figures may be even lower for poorlyconstructed or defective lift surfaces.However, the peak dynamic tensile stressesonly exist during a fraction of a response

DRAFT



cycle. Even though these peak stresses maygreatly exceed the tensile strength of theconcrete, any cracking that might be initiatedwill not have time to fully develop. It is wellrecognised that a single spike of excessivelocalised tension should not be taken torepresent dam failure.

In consideration of the above however, theseguidelines recommend that for sound liftsurfaces, the apparent tensile strength to beused is 16% of the standard compressivestrength.

For dynamic modulus of elasticity, Cloughand Ghanaat (1993) suggest a value 25%greater than the static value and theseguidelines recommend this be adopted. Forexisting dams, the elastic modulus of theconcrete mass may be determined usinggeophysical means (e.g. derived frommeasured shear wave velocity). Valuesobtained should be compared with static anddynamic small sample laboratory test valuesfor credibility.

C4.4.1.2 FoundationsThe Young’s modulus of deformation for ajointed rock mass can be estimated from theGeological Strength Index (GSI) using themethod proposed by Hoek et al. (1998) andHoek and Brown (1997. Fell et al (2015)describe these and the method by Douglas(2003). The dynamic modulus may be higherthan the static modulus as discussed inClough and Ghanaat (1993) and Scott andVon Thun (1993) and may, as for concrete,be obtained by geophysical means, or byrelation to the static modulus.

A dam's foundations will normally containjoints, shears, and bedding. Consequently, itwill not be possible to transmit tensile stresswithin the foundations and the allowabletensile strength for the foundations willtherefore normally be assumed to be zero.However, if extensive site investigation andstrength testing is able to prove that thefoundations for a particular dam site aremassive rock, e.g. massive sandstone, arecapable of transferring tensile stresses, then

the tensile strength of the rock may beincluded for small concrete structures lessthan about 5m high

C4.4.2 Loads

C4.4.2.1StaticStatic loads applied to a dam for seismic andpost-earthquake analyses for all types ofconcrete dams will generally be the same asthose defined in the ANCOLD (2013)

C4.4.2.2 UpliftNo commentary.

C4.4.2.3 SiltNo commentary.

C4.4.2.4Seismic

GeneralEarthquake ground motion occurs in alldirections. Consequently, it is recorded in 3orthogonal directions: 2 horizontal and onevertical. For 2D linear elastic models,sometimes only a single response spectrum(see below for a discussion on responsespectra) is available. In this case, it isnecessary to consider what verticalearthquake motion might be concurrent withthe horizontal shaking. In the past, thevertical acceleration used in analysis hasbeen taken as a proportion (e.g. half to 2/3)of the peak ground acceleration. USACE(2005) gives the following multipliers forvertical spectral acceleration to horizontalspectral acceleration depending on thedistance of the site from the earthquakegenerating fault:

· <= 10km distance: Sa|V/Sa|H = 1.00· 25 km distance: Sa|V/Sa|H = 0.84· >= 40km distance: Sa|V/Sa|H = 0.67

It is ANCOLDs preferred approach todevelop a vertical to horizontal ratio specificfor the site as described in Section C2.8.

InertiaInertial loads are those loads caused byearthquake accelerations acting on the massof the dam. The acceleration at a particularlevel in the dam will generally be an

DRAFT



amplification of the ground acceleration dueto the flexibility of the dam and its responseto the earthquake shaking. This accelerationis applied to the incremental mass of the damand added water mass (where added mass isused as against coupled hydrodynamicpressures - see Hydrodynamic Loads) toobtain the inertial loads. For superpositionmethods, there will be a set of inertia loadsfor each mode of vibration.

Hydrodynamic

Gravity and Buttress Dams:For gravity and buttress dams which can beconsidered sufficiently rigid or for initialanalysis of more flexible dams, the water canbe assumed to be incompressible. TheWestergaard (1933) pressure distribution istherefore valid. This pressure distribution isconsidered equivalent to a parabolic addedmass of water fixed to the face of the dam.The incremental mass of water ( ) in Kg isgiven by:

=78

( − )A

where:H = depthofwater (m)ρ = densityofwater (Kg/m3)y= heightofincrementaladdedmassofwater,(m) = above the base of the damA =incrementalsurfaceareaonfaceofdam(m2)

The above only applies for a dam having avertical upstream face and a straight axis.For a dam having a sloping or curvedupstream face, the acceleration normal to theface is used when calculating thehydrodynamic force. In this case Zanger(1952) should be used for rigid dams.

The Westergaard pressure distribution hasbeen the standard method for many years.However, consideration should also be givento alternative hydrodynamic pressuredistributions that are more relevant to the

slope of the dam’s upstream face and theconditions at the bottom of the storage. Forexample, the hydrodynamic pressuredistribution of Fenves and Chopra (1986)takes into account the depth of the storagewith respect to the height of the dam and theenergy absorbing characteristics of thestorage bottom.

USBR (2011) presents a discussion on anumber of methods for estimatinghydrodynamic pressures. The reference isspecifically for spillway gates but is relevantfor concrete dams as well.

Arch Dams:For thin arch dams, the assumption ofincompressible water and rigid dam do notnecessarily hold true. If the fundamentalfrequency of the dam (without water) isapproximately equal to the fundamentalresonant frequency of the water in thestorage, then the dam/storage interactionshould be considered in more detail. Thefundamental resonant frequency ( ) in Hzof the storage is given by:

= 4where:

= ℎ ℎ (m)= (m/s)

For the case where the dam is flexible andthe water is compressible, finite elementmodelling should include the storagemodelled using fluid elements, in addition tothe dam. A valuable discussion on this matteris given in FEMA (2014).

C4.4.2.5 Dynamic Earth PressuresNo commentary.

C4.4.2.6 Load CombinationsNo commentary.

C4.4.3 Methods Available and When toUse ThemEarthquake analysis of a concrete dam maybe carried out in the frequency domain or inthe time domain. In the former, a response

DRAFT



spectrum is used which has been derived forthe dam site from available seismic data. Inthe latter, a time-history for groundacceleration or velocity is used that again, issite specific.

A response spectrum is usually prepared by aspecialist Seismologist as described inSection 2.9 . The response spectrum isnormally a plot of spectral accelerationversus natural frequency or natural period ofa single degree of freedom oscillator (e.g. alumped mass on top of a vertical, masslesscantilever). Sometimes however, it ispresented as a tri-partite plot giving spectralacceleration, spectral velocity and spectraldisplacement all plotted against frequency orperiod. The response spectrum gives anindication as to how ground accelerations aremagnified through the height of a dam. For amore detailed description of response spectraand their theoretical background, the readeris referred to Clough and Jenzien (2003).

Time-histories of ground acceleration orground velocity can be either recorded time-

histories or synthetic time-histories.Recorded time-histories have to be derivedfrom earthquakes having a peak groundacceleration (PGA) relevant to the annualexceedance probability (AEP) earthquakeused to analyse the dam. For extremeearthquakes (e.g. the SEE), recorded time-histories will generally not be available.However, synthetic time-histories can begenerated from recorded time-historiesincluding ones from overseas earthquakes.As described in Section 2.9 two methods forgenerating these synthetic time-histories canbe used: amplitude scaling or spectralmatching. Advice should be obtained from aspecialist Seismologist regarding the methodappropriate for a particular dam.

There are a number of methods available toanalyse concrete dams ranging in increasinglevel of sophistication and complexity. Theseare detailed in Table C4-1 of the Guideline,which is taken from ICOLD (2013):

Table C4.1 Methods of analysis of concrete gravity dams for seismic ground motions.(ICOLD, 2013)Initial static conditions and loadsInitial state of the dam (cracking , joints)…Static load conditions including uplift pressureSelection of design earthquake ground motion characteristicsLevel of Seismic Analysis Input OutputLevel 0: preliminaryscreeningRelative evaluation ofseismicvulnerability of a portfolioof dams

PGA or effectiveacceleration,seismic coefficient (maps),smooth design spectra

Damage indicesPGA, C, PSa(ξ,Ti)

Level 1: pseudo-staticanalysis with constantseismic coefficientsEquivalent static forces,rigid body equilibrium - 2Dgravity method

Effective acceleration,seismiccoefficient (maps),Westergaard added mass

Pseudo-static max.stress and velocity

DRAFT



Level of Seismic Analysis Input OutputLevel IIa - Pseudo-dynamicanalysis(Chopra 1988)Standard cross-sectiongravity damsequivalent static forcesrigid body equilibrium2D gravity method

Smooth design spectra,hydrodynamic interaction,foundation interaction

SRSS (CQC) of max.stress and velocity

Level IIb - Linear responsespectra(modal) analysisClassical modal analysis -Arbitrary cross section

Smooth design spectra,reservoiradded mass, Westergaard orfluidelements, Foundation model

SRSS (CQC) of max.stress and velocity

Level IIIa - Linear time-history analysis -frequency domainLinear elastic FE analysisEAGD84, EACD-3D

Spectrum compatibleaccelerograms, analyticalreservoir compressiblemodel, analytical foundationmodel with visco-elastichalf space

Time-historyenvelopes of stress,velocity, acceleration,displacement

Level IIIb - Linear time-history analysis- time domainLinear elastic FE analysis 2-D, 3-D)

Spectrum compatibleaccelerograms, addedWestergaard mass (or addedmass using an alternativemethod e.g. Fenves &Chopra, 1986) for reservoiror fluid elements,foundation model

Time-historyenvelopes of stress,velocity, acceleration,displacement

Level IVa - Non-lineartime-historyanalysis - Time domainFinite element analysisfracture analysis (cracking)

Spectrum compatibleaccelerograms, addedWestergaard mass (or addedmass using an alternativemethod e.g. Fenves &Chopra, 1986) for reservoiror fluid elements,foundation model,fracture material properties(Kic, Kiic, Gf)

Cracking response,stability of crackedcomponents

Level IVb - Non-linear hybridtime-frequencydomain (HFTD methods) -EAGD slideSolution of non-linearfrequencydependent equations ofequilibrium

These ANCOLD Guidelines are generallyconsistent with this table by ICOLD. Furtherdetails on numerical methods appropriate tothe earthquake analysis of concrete dams canbe found in the ICOLD Guidelines as well asUSBR (2006) and FEMA (2014).

C4.4.4 Analysis in the Frequency Domain

Linear Elastic

(a) The simplest method (item (i) above)for undertaking a pseudo-staticstability analysis of a concrete gravitydam is to use the method of Fenves

DRAFT



and Chopra (1986) with a sitespecific response spectrum for thesize (AEP/PGA) earthquake beingexamined. In general terms, thenatural frequency or period of thedam is calculated for a number ofmodes of vibration. The spectralacceleration is then determined forthese frequencies or periods, from thedesign response spectrum for the damsite. Detailed documentation on themethod is readily available via theinternet. The method can be workedthrough on a spreadsheet and assumesa triangular dam on a rigid foundationto calculate the fundamental naturalperiod. This natural period is thenmodified using supplied charts toestimate the effective fundamentalnatural period and damping factoraccounting for the elastic modulus ofthe concrete in the dam relative tothat of the rock foundation, the depthof water against the dam and theamount of seismic wave reflection atthe bottom of the storage. Thegeneralised mass of the dam and theearthquake force coefficient (whichallows the multi degree of freedomdam to equate to a single degree offreedom oscillator) are thencalculated. The hydrodynamic effectof the water is added as a series oflumped masses using Westergaard’sTheory to the lumped masses of thedam (see Sub-section C4.4.2.4).Inertial forces at the levels of thelumped masses are then calculated,with an adjustment that allows forhigher modes of vibration. Theseforces can then be added to the staticforces on the dam, in a normal 2Dcantilever stability analysis of thedam to estimate extent of crackingand sliding stability. A set of inertialforces should be calculated for thedam with the storage at minimumdraw-off level to estimate the extentof cracking from the downstreamtoe/face. This cracking may limit theallowable cracking from the upstream

heel/face for either flood orearthquake cases, especially ifcohesive strength is being reliedupon.

(b) Most commonly available finiteelement analysis (FEA) softwarepackages appropriate for concretedam analysis include the ability todetermine the natural frequencies andmode shapes for a structure and tocarry out a response spectrumanalysis such that stresses induced byearthquake loading can bedetermined.

(c) The most recent review of pseudodynamic analyses using finiteelement methods is given inFERC (2016). This reference interaila, compares Westergaard’s Methodfor hydrodynamic pressures withother researchers.

The extreme response values for the variousmodes of vibration are combined using thesquare root, sum of the squares method(SRSS) or complete quadratic combinationmethod (CQC) combination. In turn, they arefurther combined using the appropriatecombination method to include the effects ofall three components of the earthquakeground motion. The resulting overalldynamic response values obtained will haveno sign and may be considered to be eitherpositive or negative. The SRSS methodusually provides a conservative estimate ofthe maximum response when the vibrationperiods of the dam structure are wellseparated. However, due to the methodignoring the correlation between the adjacentmodes, the total response for the closelyspaced vibration periods is underestimated.In that case, it is better to use the CQCmethod.

Unless the foundations of the dam aresignificantly stiffer than the concrete in thedam (and can therefore be considered asbeing rigid), it is essential that thefoundations are included in any FEM. Usualpractice is to give the foundation rock zerodensity and appropriate stiffness parameters.

DRAFT



C4.4.5 Analysis in the Time Domain

Linear Elastic

Many commonly available finite elementanalysis (FEA) software packages includethe ability to determine the earthquakeresponse of a dam using either directintegration or superposition methods. Theformer method is more accurate but requiresgreater computer resources. The method isless conservative than methods in thefrequency domain. Linear elastic analysis inthe time domain should be undertaken priorto any non-linear analysis. As noted inSection C4.4.4 it is important to include thefoundation rock in any FEM, with time-history analysis, the damping factor has to bebuilt into the analysis. This is different to aresponse spectrum analysis where the inputdata (the response spectrum) accounts for thedamping.

Non-linear

A publication by the USBR (2006) on thestate of practice for the non-linear analysis ofconcrete dams provides up to date advice ondynamic material properties, loads and loadcombinations, damping, fluid elements,appropriate FEA methods, FEM philosophyand ways of dealing with cracking.

A publication by FEMA (2014) providesadvice on analyses that might be carried outas part of a risk assessment for a concretedam.

Consideration should be given to thedirection that a crack might propagate. Forcracking at the base of a dam, the crack islikely to propagate along the foundationinterface. For cracking through the concreteof the dam itself, a crack is likely to behorizontal along a lift surface where thetensile strength across the lift surface is lessthan that of the concrete. However, wherethere is no discernible difference in tensilestrength between the lift surface and theconcrete in general, the crack couldpropagate in a downward direction as well asin the downstream direction (following the

principal stresses). This will affect slidingand overturning stability and shouldtherefore be taken into consideration.Constitutive models in some FEA code existfor doing this. Where horizontal cracking isassumed, gap elements that allow frictionalresistance (but only defined or zero tensilestrength) should be used.

In modelling the dam’s foundations,appropriate consideration should be given tothe geological/geotechnical model, especiallywith regard to continuous defects alongwhich movement might occur.

It is essential that damping factors andstiffness moduli are appropriately modelledsuch that the hysteretic behaviour of the damand foundations is properly captured.

C 4.5 Approach to Analyses andAcceptance Criteria

C4.5.1 During the Earthquake

C4.5.1.1 Gravity DamsFERC (2016) states that stresses inducedduring extreme earthquake loading willinduce stresses exceeding relevant materialstrengths. Consequentially, no acceptancecriteria are given for the behaviour of agravity dam during the course of extremeearthquake shaking. FERC emphasises theneed for post-earthquake static analysis.ANCOLD (2013) and this guideline take theapproach that pseudo-static methods areuseful and in general likely to beconservative, so if factors of safety are ≥minimum required in Tables 6.1 and 6.2 ofthat Guideline the behaviour of the dam forthat seismic ground motion is likely to besatisfactory.

The USACE (2007) requires that duringextreme earthquake shaking, acceptancebased on linear elastic time-history analysisdepends upon the demand/capacity ratio(DCR) for the dam and the cumulative timeexcursions of non-linear behaviour. Thedemand is defined as “the ratio of stressdemands to the static tensile strength of the

DRAFT



concrete”. The DCR is not to exceed 2 for avery limited cumulative non-linear durationand reduces to 1 for a cumulative non-linearduration of 0.75 second. The DCR method isfurther described in USACE (2003b).

C4.5.1.2 Arch Dams:The same principles apply to arch dams asapply to gravity dams. In addition,consideration has to be given to whether ornot to specifically model the storage (asagainst using empirical formulae to estimatehydrodynamic pressure distributions) in anyfinite element, time-history analysis. Thiswill depend on the relative fundamentalfrequencies of the dam and the storage –refer Section 4.4.2.4. If an arch dam hasmassive gravity abutments (thrust blocks)finite element modelling has to be such thatstresses at the interfaces between the ends ofthe arch and the abutment units can be easilyintegrated to produce thrust and shear forcesacting on the abutment units. These forcesare then used to carry out 3D stabilityanalysis of the gravity abutment units. Theacceptance criteria for abutment blocks maybe taken as for concrete gravity dams.

The post-earthquake factors of safety shouldbe as given in ANCOLD (2013) for extremeload cases.

Additional details on the evaluation of archdams can be found in Jonker (2014).

C4.5.1.3 Buttress Dams:Massive buttress dams (e.g. Burrinjuck Damin NSW) should use the acceptance criteriagiven for gravity dams. The followingdiscussion relates to buttress dams that aremore structural in nature (e.g.thin slab andbuttress dams).

While the interaction between the face slabsand the buttresses is generally staticallydeterminate, the interaction between thebuttresses and the struts between thebuttresses is generally staticallyindeterminate. That is, there is a significantdegree of redundancy in that part of thestructure. Consequently, a number of strutscould potentially fail without causing failure

of the structure and leading to anuncontrolled loss of storage. While somestructural elements in the dam may fail,overall failure of the dam (breaching) willnot necessarily occur. It may be necessary toremove structural elements that are likely tofail in order to see how loads re-distribute inthe dam and whether they are likely to leadto further failure of structural elements.Consideration should also be given to theductility of various critically loadedstructural elements. A number of slab andbuttress dams have relatively lightreinforcement in the struts for example andthe cracking moment for the strut is greaterthan the moment capacity for the cracked,reinforced concrete section inferring a non-ductile cross section..

C4.5.2 Post-Earthquake

C4.5.2.1 Gravity Dams:No commentary.

C4.5.2.2 Arch Dams:Due to the usually small distance from theupstream heel to the downstream toe(compared to a gravity dam), the change inuplift force from a fully drained upliftpressure distribution (or linear from theupstream heel to the downstream toe) to acracked uplift pressure distribution (fullheadwater pressure within the cracked zone)will be relatively small. However,consideration should be given to the reducedsliding resistance at the foundation interfacecaused by cracking in the arch. The stabilityof the abutment sections of the arch or of anygravity abutments should be analysed forpost-earthquake loadings taking any crackingoccurring during the earthquake into account.

Linear elastic FEA may be undertaken usingvery low elastic modulus elements along thecrack(s).

The abutments of an arch dam should beanalysed considering any redistribution ofstresses within the arch due to cracking.Consideration should be given to the 3Danalysis of gravity abutment units and of

DRAFT



rock wedges in the abutment foundations thathave viable release surfaces.

C4.5.2.3 Buttress Dams:Carry out a static post-earthquake stabilityand strength analysis of the buttress damconsidering structural cracking that may haveoccurred in face slabs and buttresses andstructural that might have occurred in thestruts between buttresses. Residual shearstrength parameters should be used whereappropriate. Consideration should be givento each individual buttress foundation aslocalised foundation properties could lead todifferential sliding deformations which inturn, could lead to differential structuraleffects within the various parts of the buttressdam. Reference should be made to FERC(1997, “Other Dams”) and to USACE(2007). It should be noted that the upliftpressure under the whole of a buttressfoundation will be determined by thetailwater level.

C4.5.3 GeneralIn the analysis, uplift pressures are likelyto have been represented in the finiteelement model, by a set of self-equilibrating pressures on a thin layer ofelements at the base of the dam. Itshould be remembered that the stressesobtained from the finite element analysisin this case, will be total stresses. Indetermining the effective stressdistribution through the dam, thedistribution of seepage (pore) pressuresthrough the dam will have to beconsidered.

Post-earthquake analysis would usuallybe required when the dynamic analysisof the dam and its foundations duringearthquake shaking indicate:

· Cracking propagates such that post-earthquake sliding stability will bereduced due to reduced cohesion (ifcohesion is being relied upon).

· Cracking is such that shear strengths willreduce to residual strengths post-earthquake.

DRAFT

Commentary – 5.0 Assessment of Tailings Dams for Seismic Ground Motion


C5 ASSESSMENT OF TAILINGS DAMSFOR SEISMIC GROUND MOTIONS

C5.1 Some General PrinciplesFigures 19.17 to 19.24 of Fell et al (2005,2015) show schematically the methods ofconstruction.

C5.2 The Effects of Earthquakes onTailings DamsSee ANCOLD (2012) for more information.

C5.3 Defensive Design Principles forTailings Dams

These Guidelines caution against the use ofupstream construction methods for High andExtreme Consequence Category tailingsdams and require practitioners taking thisapproach to use conservative assumptionsbased on the following information:

(a) There have been a number of failuresof such dams due to seismic loads.

(b) It is very difficult to ensure thatquality control of placement of thetailings during the operation of theTSF (often many years) will besufficient to give the required lowlikelihood of failure. One loosesaturated layer in otherwise properlydensely placed tailings is sufficient tolead to liquefaction.

(c) There are large uncertainties inpredicting the liquefied strengths anddeformations of liquefied tailings.

(d) It would be contradictory to say it isgood practice to site conventionaldams to avoid liquefaction, and notrecommend against upstreamconstruction for High and ExtremeConsequence Category tailings dams.

This ultimately is a matter for StateRegulators to decide. ANCOLD counsel thatif upstream or centreline construction is usedfor high and extreme consequence categorydams persons expert in liquefaction oftailings be involved; conservative lowerbound strengths be adopted for design; andassumptions made for design confirmed bytesting during operation of the dam.

Small raises of the tailings dam can usuallybe designed so there is sufficient rockfill orother non strain-weakening or crackingmaterial in the starter dams so that adequatepost-earthquake stability can be provided anddeformations limited to tolerable amountswith a high degree of confidence. Essentiallythese become centreline construction raises.

C5.4 Design Seismic Loads and AnalysisMethodThis Guideline, which draws no distinctionbetween conventional and tailings dams,recommends different design earthquakeloadings during operation to what isrecommended in Table 7 of ANCOLD(2012a) . The reasons for this are:

(a) The OBE should be determined bythe Owner in consultation with theConsultant in consideration of damsafety and business risks.

(b) The OBE in Table 7 of ANCOLD(2012a) were not consistent withusual practice in that it is seldom thatan OBE with a frequency as low as 1in 1000 AEP is adopted. OBE of 1 in50 and 1 in 100 are so low a load asto make them not significant.

(c) The SEE (MDE) for TSF should beconsistent with those for conventionaldams. ANCOLD (2012a) has a lowerSEE for Low Consequence Categorydams (1 in 100).

The post closure design earthquake loadingis shown in ANCOLD (2012a) as the MCEfor all Consequence Categories of TSF. Thiswas in recognition of the long life afterclosure of a tailings dam compared to say100 years for a conventional dam.

ANCOLD (2012a) also recommended the 1in 1000 AEP ground motion for OBE tobetter capture the potential for liquefaction oftailings because they may liquefy at 1 in1000 AEP ground motions and not at 1 in500 AEP ground motions.

These issues have been considered for thisGuideline and this Guideline concludes:

DRAFT

Commentary – 5.0 Assessment of Tailings Dams for Seismic Ground Motion


(a) In Australian conditions it is seldompossible to identify the location andactivity of all faults which maycontribute to the seismic hazard sothe design seismic loading should beexpressed in probabilistic terms, notMCE. This is as is done in Table 2.1of this Guideline. For ExtremeConsequence Category dams there isalso a requirement for MCE.

(b) The requirement of ANCOLD(2012a) for MCE post closure forLow and Significant ConsequenceCategory tailings dam could beconsidered onerous. It is inconsistentwith the requirement for conventionaldams.

(c) The argument that tailings dams arerequired to operate post closure for anindefinite period and therefore shouldbe designed for the largest credibleearthquake has some merit but toapply it to all tailings damscontradicts normal risk basedmanagement because theconsequences of failure for Low andSignificant Consequence Categorytailings dams are lower than for Highand Significant ConsequenceCategory tailings dams. Also theprinciples of tolerable risk criteria arethat they are annual, not summedover the life of a structure, so the lifeof the structure is not a factor.

(d) In any case many conventional damsare likely to operate for centuries.

(e) The OBE should not be used as a defacto SEE to pick up liquefactionissues. This Guideline specificallysays to consider liquefaction for theSEE of Significant and LowConsequence Category dams and touse the higher loadings if liquefactionis assessed to be an issue.

(f) The principle that OBE is for Ownersto decide should stand.

Based on these factors it is concluded thatthe post closure design seismic loadingshould be as for the SEE, but that the SEE

should be re-evaluated at the time of closureto allow for any developments inunderstanding of seismic hazard and for anychanges in the consequences of failure, andthat allowance be made for potentialincreases in the PAR post closure.

C5.5 Seismic Deformation Analysis ofTailings Dams not Subject to LiquefactionSection 6.1.5.1 and Figure 6 of ANCOLD(2012a) incorrectly indicate that theSwaisgood (1998) and Pells and Fell (2003)methods can be applied to TSF whichexperience liquefaction. Those methods arebased on databases which excludeliquefaction cases. Swaisgood (1998) doesinclude hydraulic fill dams but his plots donot include those which have liquefied.

C5.6 Assessment of the Effects ofLiquefaction in Tailings Dams andTheir Foundations

As described in ANCOLD (2012) Section6.1.5, tailings are usually stratified, so under-drains may not result in complete drainage ofthe tailings. Perched water tables with layersof saturated tailings should be assumed to bepresent unless extensive investigations proveotherwise.

C5.7 Methods for Upgrading TailingsDams for Seismic Loads

In mining operations there will often be aready and relatively inexpensive supply ofwaste rock to construct berms to improvepost-earthquake stability and reducedeformations to tolerable amounts.

The geochemistry of the rock should beconsidered to ensure that there are no acidmine generation issues.

DRAFT

Commentary – 6.0 Assessment of Appurtenant Structures for Seismic GroundMotion


C6 ASSESSMENT OF APPURTENANTSTRUCTURES FOR SEISMIC GROUNDMOTIONS

C6.1 The Effects of Earthquakes onAppurtenant Structures

It is important that their functional andstructural integrity of appurtenant structuresis retained in the event of a notableearthquake. The importance of the role of theappurtenant structures was particularlyhighlighted in the Wenchuan earthquake inChina in 2008, as discussed in Wieland andChen (2010), which included examples of:

· Outlet works control panels topplingleaving the outlet works inoperable;

· Penstock failure and powerhouseflooding due to rock falls; and

· Large scale rock falls and landslidesleaving dams inaccessible for anumber of months following theearthquake event.

It is recommended that the reader also refersto ICOLD which has a number of Guidelinesrelating to earthquake analysis and design ofappurtenant structures including ICOLD(2002) ‘Bulletin 123: Earthquake design andevaluation of structures appurtenant todams’.

Regardless of the method of analysisselected, the final evaluation of seismicsafety of an appurtenant structure should bebased on engineering judgement andexperience with similar structures, keeping inmind that each structure and its immediateenvironment are unique and may not beduplicated elsewhere.

C6.2 Intake/Outlet Towers

C6.2.1 GeneralThe apparent simple geometry of intaketowers is often complicated by the presenceof various openings and appendages. Thereare also often changes in wall thickness atvarious elevations. Sometimes, the accessbridge that provides access to the tower will

structurally interact with it. There can beextreme variability in the amount ofreinforcement within existing towers notdesigned for seismic conditions with steel toconcrete areas often ranging from 0.2 percentto 2 percent, a variation factor of about 10. Inmany existing towers, minimal or noconfinement (transverse) steel is provided,which is a very important aspect in theearthquake structural engineering. There aremany cases where there will be a strong anda weak axis of the intake tower dependingupon the shape of the tower and the intakeand outlet arrangement. It is recommendedthat the tower is assessed in both directions.All of this variability in the towersunderscores the diversity of designencountered and the need to consider eachintake tower individually, based on the actualconfiguration and applicable seismic loads.

In the assessment it needs to be consideredthat the frequency of the tower will changefollowing the inclusion of added masses foreither or both the external and internal water.

It is to be noted that in general, intake towershave historically performed well duringearthquakes, with minimal reported instancesof tower damage or failure.

C6.2.2 Defensive Design PrinciplesNo commentary.

C6.2.3 Performance RequirementsNo commentary.

C6.2.4 Analysis Procedures

Response SpectrumThis type of analysis assesses the maximumresponse of the structure to earthquakeexcitation by combining the maximumresponses from individual modes and multi-component inputs. This method uses anadded mass representation of hydrodynamiceffects due to surrounding (outside) waterand contained (inside) water (in the case ofwet towers). In addition, it includes theeffects of the tower/foundation interaction.

DRAFT



The determination of the hydrodynamic massof the water should be done using Goyal andChopra (1989a,b,c,d,e). The Square-Root-of-the-Sum-of-Squares (SRSS) method isappropriate to combine the multidirectionalloading components. Although, the CompleteQuadratic Combination (CQC) method canbe used as an alternative to the SRSSmethod. A damping ratio of 5% is typicallyadopted for concrete structures, for steelstructures 2% damping is typically adopted.The accuracy of the results depends on thenumber of vibration modes considered andthe methods of combination used for themodal and multi-component earthquakeresponses. The response spectrum analysisprocedures are detailed in full in USACE(2003a) and USACE (2007).

The response spectrum analysis is adequatefor towers whose responses to earthquakesare within the linear elastic range. All towersshould be designed to remain in the linearelastic range for the OBE. However, it willnot generally be necessary for the tower tobehave elastically during the SEE. Therefore,in order to ensure that the tower can surviveintense ground shaking due to the SEE withlimited damage, it should possess a ductilitycapacity greater than ductility requirementsimposed by the ground motion. This isexpressed as a demand capacity ratio (DCR),which is the ratio of the computed sectionforce demand to the section force capacity.For an SEE the minimum DCR should betaken as equal to 2 for flexure and 1 forshear.

Linear Time-History AnalysisLinear time-history analysis involvescomputation of the complete response historyof the structure to earthquakes. Theprocedure for this analysis is similar to thatdescribed for the response spectrumprocedure, except that earthquake demandsare in the form of acceleration time-histories,rather than response spectra and the resultsare in terms of displacement and stress (orforce) histories. The results of such analysis

serve to demonstrate the general behaviourof the seismic response, and can providesome estimate of the expected inelasticbehaviour or damage when the non-linearityis considered to be slight to moderate. Time-history analysis provides valuable time-dependant information that is not-available inthe response spectrum analysis. Especiallyimportant is the number of excursionsbeyond displacement levels where thestructure might experience strengthdegradation (strain softening).

Linear time-history analysis involvescomputation of the complete response historyof the structure to earthquakes. USACE(2007) provides details for the assessmentand criteria using linear time-historyanalysis. The basic procedure is to perform alinear time-history analysis with appropriateamount of damping to obtain bendingmoment DCR ratios for all finite elements.Initially a damping ratio of 5 percent is usedand then increased to 7 percent if DCR ratiosare approaching 2 and to 10 percent if theyexceed 2. After adjusting the damping thedamage is compared against the criteria.

From the outcomes of the assessment thedamage is considered moderate andacceptable if the following conditions aremet:· Bending moment DCR ratios computed

on the basis of linear time-historyanalysis remain less than 2;

· Cumulative duration of bending momentexcursions above DCR ratios of 1 to 2fall below the acceptance curve providedin USACE (2007); and

· The extent of yielding along the heightof the tower (i.e. plastic hinge length forDCR ratios of 1 to 2) is limited and fallsbelow the acceptance curve provided inUSACE (2007).

If the DCR ratios exceed 2.0 or thecumulative duration and the yield lengthsrise above the acceptance curves, the damageis considered to be severe and would need to

DRAFT



be assessed using non-linear analysisprocedures.

Non-linear Time-History AnalysisThe non-linear analysis of towers is typicallycomplicated and time consuming. Until onlyrecently this type of analysis has beenconsidered non-practical. However theincrease in capabilities of computers and thefurther understanding of the performance ofthese structures in earthquakes is leading tomore emphasis being placed on these typesof analyses. For details on non-linearanalysis of intake tower refer to USACE(2007).

Stability Assessment - Sliding

Response Spectrum MethodThe sliding stability for the responsespectrum method is expressed in terms of afactor of safety. The response spectrumanalysis procedures described for theStrength Assessment above are adopted todetermine the inertial forces on the structure.

Linear Time-History AnalysisThe results of the linear elastic time-historyanalysis can be used to compute time-historyor instantaneous sliding factor of safetyalong a sliding plane. The procedures for thisassessment are provided in USACE (2007).Within this analysis, the stability ismaintained and sliding does not occur if thefactor of safety is greater than 1. However, afactor of safety of less than 1 indicates atransient sliding, which if repeated numeroustimes, could lead to excessive or permanentdisplacement. If this is the case then themagnitude of the sliding displacement mayneed to be evaluated using non-linearanalysis.

Non-linear Time-History AnalysisIn the non-linear time-history analysis anassessment can be made of the totalpermanent sliding displacement of the tower.It then needs to be assessed whether thepermanent displacements meet the criteriafor the SEE.

Stability Assessment - Rotational

A structure will tip about one edge of its basewhen earthquake plus static overturningmoment (Mo) exceed the structure restoringmoment capacity (Mr), or when the resultantof all forces falls outside the base. Rockingresponses to ground motions may include:

· No tipping because Mo > Mr

· Tipping or uplift because Mo > Mr, but norocking due to insufficient ground motionenergy

· Rocking response (Mo > Mr) that willeventually stop due to the energy lossduring impact

· Rocking response that leads to rotationalinstability (considered unlikely)

Tipping Potential EvaluationWhen towers are subjected to large lateralforces produced from an earthquake, thetower may start to tip and start rocking whenthe overturning moment becomes so largethat the structure breaks contact with theground. The assessment of the tippingpotential for the tower, either a rigid orflexible tower, is provided in USACE(2007). If it is determined that no tippingoccurs then the tower is considered stable. Iftipping occurs then the rocking blockanalysis should be conducted.

Rocking Block AnalysisA tower would eventually overturn if themoment Mo > Mr is applied and sustained.However, the overturning moments typicallyonly occur for a fraction of a second in eachcycle in a large earthquake, with intermediateopportunities to unload. Therefore, althoughrocking occurs tower structures are unlikelyto become unstable. USACE (2007) providesthe procedures for assessment of the towerrocking.

C6.3 SpillwaysGated spillway structures may be unstableduring seismic loading due to a combinationof large water loads concentrated near the topof the structure and the relative lack of mass

DRAFT



involved with this type of structure.Increasing the mass at the base of thestructure, anchoring the structure, or othermeasures, may be necessary to stabilise thestructure during seismic loading.

C6.4 Spillway GatesNo commentary.

C6.5 Outlet Works - Water Conduits,Gates and Valves

C6.5.1 GeneralWorld-wide experience with the performanceof tunnels during earthquake has been verygood. Even soft ground tunnels haveperformed well as long as their designincorporated some degree of articulation andflexibility. This has been attributed todecrease of the inertial effects versuskinematic ones in their seismic response andalso due to decreasing rate of seismicintensity in the ground depth. However,particular attention is required at tunnelportals. Since portals are surface features,they are usually situated in weak rock,experience surface wave reflections, and areexposed to blockage by earthquake-inducedrock falls and landslides. Most cases ofhistoric earthquake damage to tunnel systemshave been at tunnel portals. Particularconsideration also needs to be given wheretunnels go from within rock to within anembankment.

A comprehensive study of the effects ofearthquake and tunnels can be found inDowding and Rozen, 1978, and Owen andScholl, 1981. Further information on theassessment procedures for tunnels forearthquakes is provided in Wang (1993).



C6.5.4 Analysis ProceduresGuidance is provided in FEMA (2005)regarding the seismic assessment of outletconduits. This document suggest that unlessthe conduit is founded upon deep soil layers,seismic ground motions for the bedrock aretypically assumed to act on the conduit. For aconduit founded on soil the ground motionshould allow for amplification or de-amplification effects. For a structure with afundamental frequency less than 33Hz, it issuggested that a pseudo-static approach isgenerally suitable. Where structures have ahigher fundamental frequency, a responsespectrum or time-history analysis may berequired.

Information on the assessment andperformance of tunnels for earthquakes isalso provided in Wang (1993). Dynamicloads on the conduit are to include thoseinduced from the earthquake effect on theoverlying dam.

Earthquakes can cause substantialhydrodynamic forces in penstocks andpressure tunnels depending upon thefoundation conditions and length of thepenstock or tunnel. High dynamic pressurescan occur in short sections of tunnel such asthose through the body of a dam. The readeris referred to Weiland (2005) which providesanalysis procedures for the hydrodynamicloads.

C6.6 Retaining Walls

C6.6.1 GeneralNo commentary.



DRAFT



C6.6.4 Analysis ProceduresDepending on the magnitude of wallmovements the backfill material is said to bein yielding, non-yielding, or intermediatestate. Accordingly, the available methods ofdesign and analysis of the backfill soilpressures also fall into similar categories.Further information on the analysisprocedures is provided in USACE (2007).

Dynamic pressures of yielding backfill.Yielding backfill condition means wallmovements due to earthquake groundmotions are sufficient to fully mobilize shearresistance along the backfill wedge creatinglimit state conditions. The dynamic earthforces will then be proportional to the massin the failure wedge times the groundaccelerations. When designing retainingwalls with yielding backfill conditions forearthquake ground motions, the Mononobe-Okabe (Mononobe and Matuo 1929; Okabe1924) approach and its several variations areoften used. This is the most commonapproach used in the assessment of retainingwalls, refer to USACE (2007).

Dynamic pressures of non-yieldingbackfill. For massive structures with soilbackfill, it is unlikely that movementssufficient to develop backfill yielding willoccur during an earthquake. In this situationthe backfill soil is said to be non-yieldingand is treated as an elastic material. Ifidealized as a semi-infinite uniform soillayer, the dynamic soil pressures andassociated forces for a non-yielding backfillcan be estimated using a constant-parametersingle degree of freedom model or a moreelaborate multiple degree of freedom system.The dynamic soil pressures for a moregeneral non-yielding backfill soil can bedetermined by finite-element procedures.

Intermediate case. The intermediate case inwhich the backfill soil undergoes non-lineardeformations can be represented by finiteelement procedures using a soil-structureinteraction computer program. Thefoundation and backfill soil are represented

using plane-strain 2-D soil elements whoseshear modulus and damping vary with levelof shearing strains, and the non-linearbehaviour is approximated by the equivalentlinear method.

C6.7 Parapet WallsParapet walls should be designed for Staticand dynamic loads to satisfy maximumconcrete and steel reinforcement stresseswhich satisfy AS3600 (Concrete Structures).

Detailing of the wall, and how theembankment is zoned to support the wall iscritical to the performance under seismicground motions.

C6.8 Mechanical and ElectricalEquipment

No commentary.

C6.9 Access Roads and BridgesNo commentary.

C6.10 Reservoir Rim InstabilityThe assessment of landslides is a specialistarea and suitably qualified engineeringgeologists and geotechnical engineers shouldbe consulted.

The following references are relevant to theassessment:

ICOLD Bulletin 124, Reservoir landslides:investigation and Management-Guidelinesand case histories. ICOLD (2002).

Effect of earthquakes on landslides:

· Keefer (1984)

Methods for assessing whether landslideswill travel slowly or rapidly:

· Glastonbury and Fell (2008a, 2008b,2010)

· Fell et al (2007)· Glastonbury et al (2002)

DRAFT



Methods for predicting impulse wavesresulting from landslides into reservoirs:

· Heller et al (2009)· Panizzo et al (2005)

It should be noted that only slides whichmove rapidly after failure, and as they enterthe reservoir will result in waves. Thereferences above give guidance on whether aslide will be rapid.

DRAFT

References


REFERENCES

Abaqus (2012) www.simulia.com.

Abrahamson N.A. and Silva W.J. (2008).Summary of the Abrahamson and Silvaground motion relations, EarthquakeSpectra 24, 67-98.

Abrahamson N.A., G. Atkinson, D.Boore, Y. Bozorgnia, K. Campbell, B.Chiou, I.M. Idriss, W. Silva and R.Youngs (2008). Comparisons of theNGA ground motion relations.Earthquake Spectra 24, 45-66.

Abrahamson. N.A. and K. Shedlock,1997. Overview of ground motionattenuation relations. SeismologicalResearch Letters, 68: 9-23.

Abrahamson, N., (2008), Appendix CProbabilistic Fault Rupture HazardAnalysis: San Francisco PUC, GeneralSeismic Requirements for the Design onNew Facilities and Upgrade of ExistingFacilities, 1–7 p.

Abrahamson, N., W. Silva and R. Kamai(2014). Update of the AS08 Ground-Motion Prediction Equations Based onthe NGA-West2 Data Set. EarthquakeSpectra: August 2014, Vol. 30, No. 3,pp. 1025-1055.

Allen, T. (2011), "The 2012 AustralianSeismic Hazard Map – Catalogue andGround Motion Prediction Equations."Proceedings of the AustralianEarthquake Engineering SocietyConference, 18-20 Nov, Novotel BarossaValley Resort, South Australia.

Allen, T.I. (2012). Stochastic ground-motionprediction equations for south easternAustralian earthquakes using updated sourceand attenuation parameters. GeoscienceAustralia Record 2012/069.

Allen, T.I. and Atkinson, G.M. (2007).Comparison of earthquake source spectraand attenuation in eastern North Americaand south eastern Australia. Bull. Seism.

Soc. Am 97, 1350-1354.

Allen, T.I., Cummins, P.R., Dhu, T. andSchneider, J.F. (2007) Attenuation ofground motion spectral amplitudes insouth eastern Australia. Bull. Seism. Soc.Am 97, 1279-1292.

Allen, T. I., Gibson, G., Brown, A., and Cull,J.P. (2004) Depth variation of seismic sourcescaling relations: implications for earthquakehazard in south eastern Australia,Tectonophysics, 390, 5-24.

ANCOLD (1998) “Guidelines for the Designof Dams for Earthquake” Australian NationalCommittee on Large Dams.

ANCOLD (2003). Guidelines on RiskAssessment. Australian National Committeeon Large Dams, Hobart.

ANCOLD (2012), Guidelines on theConsequence Categories for Dams.Australian National Committee on LargeDams, Hobart.

ANCOLD (2012a). Guidelines on TailingsDams: Planning, Design, Construction,Operation and Closure. Australian NationalCommittee on Large Dams, Hobart.

ANCOLD (2013). Guidelines on DesignCriteria for Concrete Gravity Dams.Australian National Committee on LargeDams, Hobart.

ANCOLD (2016). Guidelines on RiskAssessment. Australian National Committeeon Large Dams, Hobart. In preparation.

Anderson, J. G., and Hough, S.E. (1984).A model for the shape of the Fourieramplitude spectrum of acceleration athigh frequencies, Bull. Seism. Soc. Am.74, 1969-1993.

Andrus, R.D., Hayati, H. and Mohanan, N.P.(2009). Correcting liquefaction resistance foraged sands using measured to estimatedvelocity ratio. J. Geotech. Geoenviron.Engrg. ASCE. Vol.135, 735-744.

DRAFT

References


Andrus, R.D. and Stokoe, K.H., II (2000).‘Liquefaction resistance of soils from shear-wave velocity’. J. Geotech. and Geoenvir.Engrg., ASCE, Vol. 126, No. 11, 1015–1025.

Arango, I., Lewis, M. R., and Kramer,C._(2000)_. “Updated liquefaction potentialanalysis eliminates foundation retrofitting oftwo critical structures.” Soil Dyn. EarthquakeEng., 20, 17–25.

Atkinson, G. M., and Boore, D.M.(2006). Earthquake ground-motionpredictions for eastern North America,Bull. Seism. Soc. Am. 96, 2181-2205.

AS1710 - Structural design actions,Standards Australia.

AS1170.4-2007: Structural designactions, Part 4: Earthquake actions inAustralia. Standards Australia.

AS3600 –Concrete Structure, StandardsAustralia.

AS4100 –Steel Structures, StandardsAustralia.

AS5100 – Bridge Design, StandardsAustralia.

Baker, J.W. (2010). An Introduction tothe Conditional Mean Spectrum.

Baker, J. W. (2011). Conditional MeanSpectrum: Tool for ground motionselection. Journal of StructuralEngineering, 137(3), 322-331.http://dx.doi.org/10.1061/ (ASCE)ST.1943-541X.0000215

Baker, J.W. and Cornell, C.A. (2006).Spectral shape, epsilon and recordselection. Earthquake engineering andstructural dynamics, 35, 1077-1095.

Baziar, M.H. and Dobry, R. (1995). Residualstrength and large scale deformationpotential of loose silty sands. J. GeotechnicalEngineering, ASCE, Vol. 121, No. 12, 896–906.

Beaty, M.H. and P.M. Byrne (2008).“Liquefaction and Deformation AnalysisUsing a Total Stress Approach.” ASCEJournal of Geotechnical andGeoenvironmental Engineering, 134(8),1059-1072.

Boore, D.M., J. P. Stewart, E. Seyhan,and G, M. Atkinson (2014). NGA-West2 Equations for PredictingResponse Spectral Accelerations forShallow Crustal Earthquakes.Earthquake Spectra: August 2014, Vol.30, No. 3, pp. 1057-1085.

Boulanger, R. W. (2010). A sand plasticitymodel for earthquake engineeringapplications. Report No. UCD/CGM-10-01,Center for Geotechnical Modelling,Department of Civil and EnvironmentalEngineering, University of California, Davis,CA.

Boulanger, R.W., and Idriss., I.M. (2006)Liquefaction susceptibility criteria for siltsand clays, J. Geotechnical andGeoenvironmental Engineering, ASCE, Vol132, No 11, 1413-1426.

Boulanger, R.W., and Idriss, I.M. (2007)Evaluation of cyclic softening in silts andclays, Journal of Geotechnical andGeoenvironmental Engineering, ASCE, Vol133, No6, 641-652.

Boulanger, R.W. and Idriss, I.M (2012)Probabilistic Standard Penetration Test –based liquefaction –triggering procedure.Journal of Geotechnical andGeoenvironmental Engineering, ASCE,Vol138, No10, 1185-1195.

Boulanger, R. W. and Ziotopoulou, K.(2012). PM4Sand (Version 2): A sandplasticity model for earthquake engineeringapplications. Center for GeotechnicalModeling, Report No UCD.CGM-12-03.Department of Civil and EnvironmentalEngineering, University of California, Davis,CA.

Bray, J.D. and T. Travasarou (2007).“Simplified procedure for estimating

DRAFT

References


earthquake induced deviatoric slopedisplacements”, Journal of Geotechnical andGeoenvironmental Engineering, ASCE, Vol.133, No. 4, 381-392.

Bray , J.D and Sancio, R.B. (2006)Assessment of the liquefactionsusceptibility of fine grained soils. J.Geotech. And Geoenvion. Eng. 132,1165-1177.

Brown, A. and Gibson, G. (2000).Reassessment of earthquake hazard inAustralia, Proc 12th World Conf onEarthquake Engineering, Auckland,paper 751, 8 pages.

Brown, A. and Gibson, G. (2004). A multi-tiered earthquake hazard model for Australia.Tectonophysics 390, 25-43.

Burbidge, D. (ed.) (2012). The 2012Australian Earthquake Hazard Map.Geoscience Australia Record 2012/71.

Burbidge, D and Leonard, M. (2011),The 2012 Australian Seismic HazardMap – Draft Maps. AustralianEarthquake Engineering Society 2011Conference, South Australia.

Campbell, K.W. and Bozorgnia, Y.(2014). NGA-West2 Campbell-Bozorgnia Ground Motion Model for theHorizontal Components of PGA, PGV,and 5%-Damped Elastic Pseudo-Acceleration Response Spectra forPeriods Ranging from 0.01 to 10 sec.Earthquake Spectra: August 2014, Vol.30, No. 3, pp. 1087-1115.

Campbell, K. W. (2009). Estimates ofshear-wave Q and Kappa forunconsolidated and semiconsolidatedsediments in eastern North America,Bull. Seism. Soc. Am. 99, 2365–2392.

Cetin, K. O., Seed, R. B., Der Kiureghian,A., Tokimatsu, K., Harder,L. F., Kayen, R.E., and Moss, R. E. S., (2004). Standardpenetration test-based probabilistic anddeterministic assessment of seismic soilliquefaction potential, J. Geotechnical and

Geoenvironmental Eng., ASCE Vol 130, No12, 1314–340.

Chiou, B. S.-J., and Youngs, R. R.(2014). Update of the Chiou and YoungsNGA model for the average horizontalcomponent of peak ground motion andresponse spectra, Earthquake Spectra 30,1117–1153.

Clark, D. and Van Dissen, R. (2006)Characterisation of Australian intra-plateactive faulting with special mention of theLake Edgar Fault, Tasmania. GNS ScienceConsultancy Report 2006/099

Clark, D., McPherson, A.and Van Dissen, R.(2012). Long-term behaviour of AustralianStable Continental Region (SCR) faults,Tectonophysics 566-567, 1-30.

Clark D., McPherson, A. and Collins C. D.N. (2011). Australia’s seismogenicneotectonic record: a case for heterogeneousintra-plate deformation. GeoscienceAustralia Record 2011/11.

Clough, R.W. and Ghanaat, Y. (1993)“Concrete Dams:Evaluation for SeismicLoading” Dam Engineering/ICOLDInternational Workshop on Dam SafetyEvaluation Vol. IV

Clough, R.W., Jenzien, J, (2003) “Dynamicsof Structures” 3rd Edition, pub. Computers &Structures Inc.

Cox, B.R. and Griffiths, S.C. (2011).Practical recommendations for evaluationand mitigation of soil liquefaction inArkansas. Project number: MBTC 3017,Mack-Blackwell Transportation Center,University of Arkansas.

Crone, A.J., Machette, M.N., Bowman, J.R.,1997. Episodic nature of earthquake activityin stable continental regions revealed bypalaeoseismicity studies of Australian andNorth American Quaternary faults.Australian Journal of Earth Sciences 44,203–214.

DRAFT

References


Cubrinovski, M. and Ishihara, K. (2000a).Flow potential of sandy soils with differentgrain compositions. Soils and Foundations,40(4): 103–119.

Cubrinovski, M. and Ishihara, K. (2000b).Flow potential of sandy soils. Proc. GeoEng.2000, Technomic Publishing, Lancaster,549–550.

Cundall, P. (1993). FLAC – Fast lagrangiananalysis of continua, ITASCA.

Dafalias, Y. F. and Manzari, M. T. (2004).Simple plasticity sand model accounting forfabric change effects. Journal of EngineeringMechanics, ASCE, 130(6), 622-634.

Davidson, R.R, Fox, S. and Cooper, B.(2003) Seismic upgrade of YarrawongaWeir, ICOLD Congress.. ICOLD, Paris

DeJong, J.T., Ghafghazi, M., Sturm, A.,Armstrong, R., Perez, A., and Davis, C.(2014) “A New Instrumented BeckerPenetration Test (iBPT) for ImprovedCharacterization of Gravelly Deposits Withinand Underlying Dams, 2014, The Journal ofDam Safety, Vol., 12, No. 2, pp. 9-19.

DeJong, J.T., Ghafghazi, M., Sturm, A.P.,Wilson, D.W., den Dulk, J., Armstrong, R.J.,Perez, A., and Davis, C.A.(2016a)“Instrumented Becker PenetrationTest, I: Equipment, Operation, andPerformance”, ASCE J. Geotechnical andGeoenvironmental Engineering, in review.

Dise, K., Stevens, M.,G., and Von Thun, J.L.(1994), USBR Experiences Using DDC at 3Dams for Mitigation of Liquefaction Risk,ASCE GSP No.45, 1-23.

Douglas, K.D. (2003). The shear strength ofrock masses. PhD Thesis. School of Civiland Environmental Engineering, TheUniversity of New South Wales, Sydney.

Dowding, C.H. and Rozen, A. (1978)“Damage to Rock Tunnels from EarthquakeShaking” Journal of the Geotechnical

Engineering Division, Vol. 104, No. 2,February 1978, pp. 175-191.

Duron, Z.H., Ostrom,D.K Aagaard, B (1994)“Measured Steady State, Ambient andTransient Responses of a Small Arch Dam”,Dam Engineering Vol IV, Issue 1

Evernden, J. F., (1975) Seismic Intensities,“size” of earthquakes and related parameters:Seismological Society of America Bulletin,V. 65, p. 1287 – 1313.Fell, R., Glastonbury, J. and Hunter, G.(2007) Rapid landslides: the importance ofunderstanding mechanisms and rupturesurface mechanics. The eighth GlossopLecture. Quarterly Journal of EngineeringGeology and Hydrology, 40, 9-27.

Fell, R., Macgregor, J.P., Stapledon, D.H.and Bell. G. (2005) GeotechnicalEngineering of Dams. Balkema, Leiden.

Fell, R., Macgregor, J.P., Stapledon, D.H.and Bell. G. and Foster, M (2015)Geotechnical Engineering of Dams. SecondEdition. Taylor Francis, London.

FEMA (2000) “Pre-standard andCommentary for the Seismic Rehabilitationof Buildings” Federal EmergencyManagement Agency, – 356, November2000.

FEMA (2005) “Technical Manual: Conduitsthrough Embankment Dams” FederalEmergency Management Agency.

FEMA (2014) “Selecting Analytic Tools forConcrete Dams Address Key Events AlongPotential Failure Mode Paths” FEMA P-1016, July 2014

Fenves, G. and Chopra, A.K., “SimplifiedAnalysis for Earthquake Resistant Design ofConcrete Gravity Dams”, College ofEngineering, University of California,Berkeley, Report No. UCB/EERC – 85/10,June 1986

FERC (2016) “Engineering Guidelines forthe Evaluation of Hydropower Projects –

DRAFT

References


Chapter 3 Gravity Dams”, Federal EnergyRegulatory Commission, March, 2016

FERC (1997) “Engineering Guidelines forthe Evaluation of Hydropower Projects –Chapter 10 Other Dams”, Federal EnergyRegulatory Commission, October, 1997

Finn, W.D. (1988). Dynamic analysis, inGeotechnical Engineering EarthquakeEngineering and Soil Dynamics II. RecentAdvances in Ground Motion Evaluation.Geotech. Sp. Publ. No. 20, ASCE, 523–591.

Finn, W.D.L. (1993). Seismic safetyevaluation of embankment dams, inInternational workshop on dam safetyevaluation, Grundewald, Switzerland, 26–27April, Vol. 4, 91–135.

Finn, W.D.L. (2000). State-of-the art ofgeotechnical earthquake engineeringpractice. Soil Dynamics and EarthquakeEngineering, Vol 20, 1-15.

Finn, W.D.L., Yogendrakumar, M., Yoshida,N. and Yoshida, H. (1986). TARA-3: aprogram to compute the response of 2-Dembankments and soil-structure interactionsystems to seismic loadings. Department ofCivil Engineering, University of BritishColumbia, Canada.

Finn, W.D.L., Kramer, S.L., O’Rourke, T.D.,and Youd, T.L. (2010). Ad Hoc Committeeon soil liquefaction during earthquakes.Earthquake Engineering Research Institute

Friesen, S. and Balakrishnan, A (2012).General approach used for the seismicremediation of Perris dam. ProceedingsUSSD Conference, New Orleans, 1549-1564

Foster, M and Fell, R. (2001) “AssessingEmbankment Dam Filters That Do NotSatisfy Design Criteria”, Journal ofGeotechnical and GeoenvironmentalEngineering, Vol . 127, No. 5, May, 2001.ASCE.

Ghafghazi, M., DeJong, J.T., Sturm, A.P.,Wilson, D.W., Davis, C., Perez, A., andArmstrong, R. (2014a) “Characterization of

Gravelly Soils for Liquefaction PotentialAssessment at Dam Sites Using the iBPT”,USSD 2014 Conference, San Francisco, CA,2014, pp. 285-297

Ghafghazi, M., Thurairajah, A., DeJong, J.T.,Wilson, D.W., Armstrong, R. (2014b)“Instrumented Becker Penetration Test forImproved Characterization of GravellyDeposits” GeoCongress 2014, GSP 234,2014, pp. 37-46.

Ghafghazi M, DeJong JT, Sturm AP, TempleCE. (2016)“Instrumented Becker PenetrationTest, II: iBPT- SPT correlation forliquefaction assessment in gravelly soils”,ASCE J. Geotechnical andGeoenvironmental Engineering, in review.

Gibson, G. (1994) Earthqauke Hazard inAustrali. Seminar, Acceptable Risks forExtreme Events in the Planning and Designof Major Infrastructure. ANCOLD-MunroCentre, 26-27 April 1994, Sydney.

Glastonbury, J. and Fell, R. 2008a. Adecision analysis framework for theassessment of likely post failure velocity oftranslational and compound natural rockslope landslides. Canadian GeotechnicalJournal, Vol 45, No 3, 329-350.

Glastonbury, J.P. and Fell, R. 2008b. Thegeotechnical characteristics of large slow,very slow and extremely slow landslides.Canadian Geotechnical Journal, Vol 45, No7, 984-1005.

Glastonbury, J.P. and Fell, R. 2010Geotechnical characteristics of large rapidlandslides, Canadian Geotechnical Journal,Vol. 47, No 1, 116-132.

Glastonbury, J., Fell, R. and Mostyn, G.(2002). Report on the Post CollapseBehaviour of Debris From Rock SlopeFailures. UNICIV Report R-406, ISBN85841 373 6, School of Civil andEnvironmental Engineering, the Universityof New South Wales.

DRAFT

References


Gomberg, J.S. (1992). Tectonic deformationin the New Madrid seismic zone: inferencesfrom boundary element modelling.Seismological Research Letters, 63, 3, 107-425.

Goyal, A. and Chopra, A.K. (1989a)“Earthquake Analysis and Response ofIntake-Outlet Towers”, EarthquakeEngineering Research Centre, Report No.UCB/EERC-89/04, University of California,Berkeley.

Goyal, A. and Chopra, A.K. (1989b)“Earthquake Analysis of Intake-OutletTowers Including Tower-Water-Foundation-Soil Interaction”, Earthquake Engineeringand Structural Dynamics, 18(3), 325-344.

Goyal, A. and Chopra, A.K. (1989c)“Earthquake Response Spectrum Analysis ofIntake-Outlet Towers”, Proc. ASCE, Journalof Engineering Mechanics, Vol. 115, No. 7.

Goyal, A. and Chopra, A.K. (1989d)“Hydrodynamic and Foundation InteractionEffects in Dynamics of Intake Towers:Frequency Response Functions”, Proc.ASCE, Journal of Structural Engineering,Vol. 115, No. 6.

Goyal, A. and Chopra, A.K. (1989e)“Simplified evaluation of AddedHydrodynamic Mass for Intake Towers”,Proc. ASCE, Journal of EngineeringMechanics, Vol. 115, No. 7.

Gregor, N., Abrahamson, N.A.,Atkinson, G.M., Boore, D.M.,Bozorgnia, Y., Campbell, K.W., Chiou,B.S-J. I. M. Idriss, R. K.,Seyhan,E.,Silva, W., Stewart. J.P. and Youngs,R. (2013) Comparison of NGA-West2GMPEs. Earthquake Spectra 30:3, 1179-1197.

Gulerce, Z. and Abrahamson, N. (2011).Site-Specific Design Spectra for VerticalGround Motion. Earthquake Spectra:November 2011, Vol. 27, No. 4, pp. 1023-1047.

Hall, L., Dimer, F and Somerville, P. (2007).A Spatially Distributed Earthquake SourceModel for Australia. Proceedings of the 2007Annual Meeting of the AustralianEarthquake Engineering Society.

Hansen, K.D. and Nuss, L.K. (2011)“Lessons Learned from the EarthquakePerformance of Concrete Dams” WaterPower and Dam Construction, July 2011

Hayati, H., and Andrus, R.D.,(2009)Updated liquefaction resistance correctionfactors for aged sands, Journal ofGeotechnical and GeoenvironmentalEngineering, American Society of CivilEngineers, 135(11), p. 1683-1692.

Heller, V., Hager, W.H. and Minor, H.E.2009. Landslide generated impulse waves inreservoirs: Basics and computation. VAWMitteilungen 211, 156pp.

Hoek, E. and Brown, E.T. (1997). Practicalestimates of rock mass strength. InternationalJournal of Rock Mechanics and MiningSciences, 34(8), 1165–1186.

Hoek, E. Marinos, P. and Benissi, M. (1998)Applicability of the geological strength index(GSI) classification for very weak andsheared rock masses. The case of the AthensSchist Formation. Bulletin of EngineeringGeology and the Environment, 57, 151–160.

Hossain, M.D. (2010) Correctingliquefaction resistance of partially saturatedsoil using compressional wave velocity. M.Sc. Thesis, Graduate School of ClemsonUniversity.

Hough, S.E., Armbruster, J.G., Seeber, L.,and J.F. Hough (2000). On the ModifiedMercalli intensities and magnitudes of the1811-1812 New Madrid earthquakes, J.Geophys. Res., 105-B10,23839-23864.

Hudson, M.B., Martin, B.,Beikae, M.andIdriss, I.M. (1994). “QUAD4M, a ComputerProgram to Evaluate the Seismic Responseof Soil Structures Using Finite ElementProcedures and Incorporating a Compliant

DRAFT

References


Base”, NISEE, University of California,Berkeley, California.

Hunter, G. and Fell, R. (2003). Traveldistance for rapid landslides in constructedand natural slopes. Canadian GeotechnicalJournal, Vol. 40, No. 6, 1123–1141.

Hynes-Griffin, M.E. and A.G. Franklin(1984). “Rationalizing the SeismicCoefficient Method”, Miscellaneous PaperGL-84-13, US Army Engineer WaterwaysExperiment Station, Vicksburg, MS.

Hynes, M.E., and Olsen, R.S. (1999).Influence of confining stress on liquefactionresistance. Proc., Int. Workshop on Phys.And Mech. Of Soil Liquefaction, Balkema,Rotterdam, The Netherlands, 145–152.

Iai, S., and Koizumi, K., (1986). Estimationof earthquake induced excess pore waterpressure for gravel drains, in Proceedings, 7th

Japan Earthquake Engineering Symposium,679–84.

ICOLD (1983). Seismicity of Dams andDesign. Bulletin 46, InternationalCommission on Large Dams. Paris.

ICOLD (1986). Earthquake analysisprocedures for dams – State of the art.Bulletin 52, International Commission onLarge Dams. Paris.

ICOLD (1998) Neotectonics and DamsRecommendations and Case Histories.Bulletin 112, International Commission onLarge Dams. Paris.

ICOLD (1999a). Design Features of dams toresist seismic ground motion. Bulletin 120,International Commission on Large Dams,Paris.

ICOLD (1999b) Seismic Observations ofDams. Bulletin 113, InternationalCommission on Large Dams. Paris.

ICOLD (2002) Seismic Design andEvaluation of Structures Appurtenant toDams, Bulletin 123, International Committeeon Large Dams.

ICOLD 2002. Reservoir landslides –guidelines for investigation andmanagement, International Commission onLarge Dams, Paris. Bulletin 124.

ICOLD (2012) Selecting seismic parametersfor large dams - Guidelines (revision ofBulletin 72).Bulletin 148, InternationalCommission on Large Dams, Paris.

ICOLD (2013) Guidelines for Use ofNumerical Models in Dam Engineering.Bulletin 155, International Commission onLarge Dams, Paris.

ICOLD (2016a) Selecting SeismicParameters for Large Dams - Guidelines,Bulletin 148, International Committee onLarge Dams, Paris.

ICOLD (2016b) Inspection of DamsFollowing Earthquakes, Bulletin 62A,International Committee on Large Dams,Paris.

Idriss 2014, reference to be added.

Idriss, I.M. and Boulanger, R.W, (2008) Soilliquefaction during earthquakes. MonographMNO-12, Earthquake Engineering ResearchInstitute, Oakland, Ca.

Idriss, I.M. and Boulanger, R.W, (2010) SPT–Based Liquefaction Triggering Procedures.Report No UCD/CGM-10/02, Center forGeotechnical Modelling. University ofCalifornia at Davis.

Idriss, I.M., Lysmer, J., Hwang, R. and Seed,H.B. (1973). QUAD-4: A computer programfor evaluating the seismic response of soilstructures by variable damping finite elementprocedures. Report No. EERC73-16,University of California, Berkeley.

Idriss and Seed (2010,2012) , reference to beadded.

Ishihara, K. (1993). Liquefaction and flowfailures during earthquakes. Geotechnique,Vol. 33, No. 3, 351–415.

DRAFT

References


Itasca (2011) UBCSAND ConstitutiveModel on Itasca UDM Web Site by M. H.Beaty and P. M. Byrne.

Johnson, A.C., and Shedlock, K.M. (1992).Overview of Research in the New Madridseismic zine: inferences from boundaryelement modelling. Seismology ResearchLetters, 63, 3 193-208.

Jonker, M and Espander, R. (2014).Evaluation of Existing Arch Dam DesignCriteria in Lieu of ANCOLD Guidelines.

Kawai, T. (1985). Summary report on thedevelopment of the computer programDIANA – Dynamic interaction approach andnon-linear analysis. Science University ofTokyo.

Kayen, R., Moss, R. E. S., Thompson, E., M,Seed, R. B., Cetin, K. O., Der Kiureghian,A., Tanaka Y. and Tokimatsu, K. (2012)Shear-Wave Velocity–Based Probabilisticand Deterministic Assessment of SeismicSoil Liquefaction Potential. J. ofGeotechnical and GeoenvironmentalEngineering, Vol. 139, No. 3, 407–419

Keefer, D.K. (1984) Landslides caused byearthquakes. Geological Society of AmericaBulletin, Vol 95, 406-421.Kempton, J.J. and J.P. Stewart [2006].Prediction Equations for Significant Durationof Earthquake Ground Motions ConsideringSite and Near-Source Effects,” EarthquakeSpectra, 22(4), pp. 985–1013.

Khalili, N., Fell, R. and Tai, K.S. (1996). Asimplified method for estimating failureinduced deformation. Proc. Seventh Int.Symp. On Landslides. Editor Senesett, K.,Balkema, Rotterdam, 1263-1268.

Yamamuro, J.A. and Lade, P.V. (1997).Static liquefaction of very loose sands.Canadian Geotech. J. 34, 905-917.

Lee, M.K. and Finn, W.D.L. (1978).DESRA-2, Dynamic effective stressresponse: Analysis of soil deposits withenergy transmitting boundary including

assessment of liquefaction potential. SoilMechanics Series No.38, Department of CivilEngineering, University of British Columbia,Vancouver, Canada.

Leon. E., Gassman. S.L. and Talwani. P.(2006). Accounting for soil aging whenassessing liquefaction potential. J. ofGeotechnical and GeoenvironmentalEngineering, Vol. 132, No. 3, 363-377.

Leonard M. (2010a). Earthquake FaultScaling: Relating Rupture Length, Width,Average Displacement, and MomentRelease. Bulletin of the SeismologicalSociety of America 100, 1971-1988.

Leonard, M., T.I. Allen, D.R. Burbidge, D.J.Clark, A. A. McPherson and C.D.N. Collins(2012). The 2012 Australian EarthquakeHazard Map. Geoscience Australia Record2012/Xx

Liang, J. Z., Hao,H., Gaull, B.A.andSinadinovski, C. (2008). Estimation ofstrong ground motions in southwestWestern Australia with a combinedGreen's function and stochasticapproach, J. Earthq. Eng. 12, 382–405.

Lysmer, J., Udaka, T., Tsai, C.F. and Seed,H.B. (1975). FLUSH – A computer programfor approximate 3-D analysis of soil-structure interaction problems. Report No.EERC75-30, University of California,Berkeley.

Luehring R, N Snorteland, L Mejia, and M.Stevens (2001): Liquefaction mitigation of asilty dam foundation using vibro-stonecolumns and drainage wicks: a case historyat salmon lake dam. Bureau of ReclamationWater Operation and Maintenance Bulletin,No. 198, 1-15.

Makdisi, F.I. and Seed, H.B. (1978). Asimplified procedure for estimating dam andearthquake induced deformations. JASCEGeotechnical Eng., Vol. 1, 105, No. GT7,849–867.

DRAFT

References


Marcuson, W.F., Hynes, M.E. and Franklin,A.G. (1990). Evaluation and use of residualstrength in seismic safety analysis ofembankments. Earthquake Spectra, Vol. 6,No. 3, 529–572.

Marcuson, W.F., Hynes, M.E. and Franklin,A.G. (1992). Seismic stability and permanentdeformation analysis: the last twenty fiveyears. Proc. ASCE Specialty Conference onStability and Performance of Slopes andEmbankments – 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 undercyclic loading. Journal of the GeotechnicalEngineering Division, ASCE, Vol. 101, 423–438.

McCue, K. and M. Michael-Leiba (1993).Australia’s deepest known earthquake.Seismological Research Letters 64 (3-4).

Mcpherson, A.A. and Hall, L.S. (2007)Development of the Australian RegolithSite Classification Map. GeoscienceAustralia Record 20067/07

Mejia, L. (2005). Seismic Retrofit ofEmbankment Dams to MitigateLiquefaction: Seleted Case Histories.Earthquake Engineering and SoilDynamics: 1-14.

McKay, M. and Lopez, F. (2015)“Seismic Analysis of a Concrete Damwith a Complex Configuration:Nonlinear, Time-History of Mt BoldDam” ANCOLD Proceedings 2015

Mitchell, J. K., (2008). Mitigation ofliquefaction potential of silty sands, in FromResearch to Practice in GeotechnicalEngineering, Laier, J.E., Crapps, D.K.,Hussein, M.H., (eds.), Geotechnical SpecialPublication 180, ASCE, Reston, VA.

Mononobe, N., and Matsuo, H. (1929), Onthe Determination of Earth Pressures During

Earthquakes, Proceedings, WorldEngineering Congress, 9.

Moriwaki, Y., Beikae, M. and Idriss, I.M.(1988). Non-linear seismic analysis of theUpper San Fernando Dam under the 1971San Fernando earthquake. Proc. 9th WorldConference on Earthquake Engng., Tokyoand Kyoto, Japan, Vol. III, 237–241.

Moss, R. E. S., Seed, R. B., Kayen, R. E.,Stewart, J. P., Der Kiureghian, A., and Cetin,K. O., (2006). CPT-based probabilistic anddeterministic assessment of in situ seismicsoil liquefaction potential, J. Geotechnicaland Geoenvironmental Eng., ASCE 132(8),1032–051.

Motazedian, D., and Atkinson, G.M.(2005). Stochastic finite-fault modellingbased on a dynamic corner frequency,Bull. Seism. Soc. Am. 95, 995–1010.

Newmark, N.M. (1965). Effects ofearthquakes on dams and embankments. 5thRankine Lecture, Geotechnique, Vol. 15, No.2, 139–160.

Nakazawa, H., Ishihara, K, Tsukamoto, Y,and Kamata, T. (2004) Case studies onevaluation of liquefaction resistance ofimperfectly saturated soil deposits. Proc. Int.Conf. on Cyclic Behavior of Soils andLiquefaction Phenomena. Bochum,Germany, Taylor and Francis Group, 295-304.NSFW (1998). Shear strength ofliquefied soils. Proceedings of NationalScience Foundation Workshop, Editors T.D.Stark, S. Molson, S.L. Kramer and T.L.Youd. National Science Foundation,Earthquake Hazard Mitigation Program

Okabe, S. (1924) General theory on earthpressure and seismic stability of retainingwall and Dam, Journal of the JapaneseSociety of Civil Engineering, Vol. 10, No. 6pp1277-1323, 1924.

Olsen, S.M. and Stark, T.D. (2002).Liquefied strength ratio from liquefactionflow failure case histories. Canadian

DRAFT

References


Geotechnical Journal, Vol. 39, No. 3, 629–647.

Onoue, A., 1988. Diagrams considering wellresistance for designing spacing ratio ofgravel drains, Soils and Foundations 28(3),160–68.Panizzo, A., De Girolamo, P, andPetaccia, A. 2005. Forecasting impulsewaves generated by subaerial landslides.Journal of Geophysical research 110C12025: 1-23.

Pells, S. and Fell, R. (2002). Damage andcracking of embankment dams byearthquakes, and the implications for internalerosion and piping. UNICIV Report R 406,The University of New South Wales, ISBN85841 3752.

Pells, S. and Fell, R. (2003). Damage andcracking of embankment dams by earthquakeand the implications for internal erosion andpiping. Proceedings 21st InternationalCongress on Large Dams, Montreal. ICOLD,Paris Q83–R17, International Commissionon Large Dams, Paris.

Perlea, V.G. and Beaty, M.H. (2010) Corpsof Engineers practice in the evaluation ofseismic deformation of embankment dams.In Fifth International Conference on RecentAdvances in Geotechnical EarthquakeEngineering and Soil Dynamics, San Diego.

Pestana, J. M., Hunt, C. E. and Goughnour,R. R., (1997). FEQDrain: a Finite ElementComputer Program for the Analysis of theEarthquake Generation and Dissipation ofPore Water Pressure in Layered SandDeposits with Vertical Drains, Report No.EERC 97-17, Earthquake Engineering Res.Ctr., University of California at Berkeley

PLAXIS (2012) Delft, The Netherlands E-mail: [email protected]

Prevost, J.H. (1981). DYNAFLOW: A non-linear transient finite element analysisprogram. Princeton University, Departmentof Civil Engineering, Princeton, N.J.

Prevost, J.H. (1998). “DYNAFLOW, Version98, Release 02.A”, Princeton University,Department of Civil Engineering & Op. Res.,Princeton, New Jersey.

Quigley , M.C., Cupper, M.L., andSandiford, M. (2006) Quaternary faults ofsouth eastern Australia: palaeoseismicity,slip rates and origin. Australian Journal ofEarth Sciences 53, 285-301.

Rafael, J.M. (1984) “Tensile Strength ofConcrete” ACI Journal March-April 1984

Rezaeian, S., Bozorgnia, Y., Idriss, I. M.,Campbell, K., Abrahamson, N., and Silva,W., 2014a. Damping scaling factors forelastic response spectra for shallow crustalearthquakes in active tectonic regions:“Average” horizontal component,Earthquake Spectra 30, 939–963.

Rezaeian, S., Bozorgnia, Y., Idriss, I. M.,Abrahamson, N., Campbell, K., and Silva,W., 2014b. Damping scaling factors forvertical elastic response spectra for shallowcrustal earthquakes in active tectonic regions.Earthquake Spectra 30, 1335–1358.

Richter C.F., 1935, An instrumentalearthquake magnitude scale, Bull.Seismological Society Am., 25: 1-32.

Robertson, P.K. (2010). Evaluation of flowliquefaction and liquefied strength using theCone Penetration Test. ASCE J. Geotech.Geoenviron. Eng. Vol. 136, No 6, 842-853

Robertson, P.K. and Fear, C.E. (1995).Liquefaction of sands and its evaluation.Proc., 1st Int. Conf. On Earthquake Geotech.Engrg.

Robertson, P.K. and Wride, C.E. (1997).Cyclic liquefaction and its evaluation basedon the SPT and CPT. Proc. NCEERWorkshop on Evaluation of LiquefactionResistance of Soils. Editors T.L. Youd andI.M. Idriss, NCEER Technical Report 97-0022, 41–88.

Robertson, P.K. and Wride, C.E. (1998).Evaluating cyclic liquefaction potential using

DRAFT

References


the cone penetration test. Can. Geotech. J.,Ottawa, 35(3), 442–459.

Robertson, P.K. 2010, Evaluation of flowliquefaction and liquefied strength usingthe Cone Penetration Test. ASCE J.Geotech. Geoenviron. Eng. Vol. 136, No.6, 842-853.

Roth, W.H. (1985). Evaluation of earthquakeinduced deformations of Pleasant ValleyDam. Report for the City of Los Angeles.Dames and Moore, Los Angeles.

Sadigh, K., Chang, C.-Y., Egan, J.A.,Makdisi, F. and Young, R.R. 1997.Attenuation relationships for shallow crustalearthquakes based on California strongmotion data. Seismological Research Letters68, 180-189.

Sandiford, M., Wallace,N. and Coblentz, N.(2004). Origin of the in situ stress field insoutheastern Australia. Basin Research 16,325-338.

Schnabel, P.L., Lysmer, J. and Seed, H.B.(1972). SHAKE: A computer program forearthquake response analysis of horizontallylayered sites. Report No. EERC72-12,University of California, Berkeley.

Schwartz, D.P. and Coppersmith, K.J.(1984). Fault behavior and characteristicearthquakes: Examples from the Wasatchand San Andreas Fault Zones. Journal ofGeophysical Research, 89, Pages 5681–5698

Scott, G.A. and Van Thun, J.L. (1993)“Interim Guidelines-GeotechnicalStudies for Concrete Dams” USBRJanuary 1993

Seed, H.B. (1979). Soil liquefaction andcyclic mobility evaluation for level groundduring earthquakes. J. Geotech. Engrg. Div.,ASCE, 105(2), 201–255.

Seed, H.B. (1983). Earthquake-resistantdesign of earth dams. Proc., Symp. SeismicDesign. Of Earth Dams and Caverns, ASCE,New York, 41–64.

Seed, H.B. and De Alba, P. (1986). Use ofthe SPT and CPT tests for evaluating theliquefaction resistance of sands. In-Situ 86,Conference on use of in-situ tests ingeotechnical engineering. Clemence, S.P.(ed.). ASCE Geotechnical SpecialPublication, No. 6, 281–302.

Seed, H.B. and Idriss, I.M. (1982). GroundMotions and Soil Liquefaction DuringEarthquakes, Monograph Series, EarthquakeEngineering Research Institute. Berkeley,California.

Seed, H.B. and Idriss, I.M. (1971).Simplified procedure for evaluating soilsliquefaction potential. J. Geotech. Engrg.Div., ASCE, 97(9), 1249–1273.

Seed, H. B. and Booker, J. R., (1977).Stabilization of potentially liquefiable sanddeposits using gravel drains, J. GeotechnicalEng. Div., ASCE 103 (GT7), 757–68.

Seed, H.B., Lee, K.L., Idriss, I.M. andMakdisi, F.I. (1973). Analysis of slides in theSan Fernando Dams during the earthquake ofFebruary 1971. Report No. EERC/73-2,University of California, Berkeley.

Seed, H.B., Tokimatsu, K., Harder, L.F. andChung, R.M. (1985). The influence of SPTprocedures in soil liquefaction resistanceevaluations. J. Geotech. Engrg., ASCE,111(12), 1425–1445.

Seed, R.B. (2010) Technical review andcomments: 2008 EERI monograph titled Soilliquefaction during earthquakes,Geotechnical Report No, UCB/GT-2010/01,University of California, Berkeley.

Seed, R.B., Cetin, K.O., Moss, R.E.S.,Kammerer, A., Wu, J., Pestana, J., Riemer,M., Sancio, R.B., Bray, J.D., Kayen, R., andFaris, A. (2003) Recent advances in soilliquefaction engineering: A unified andconsistent framework, EarthquakeEngineering Research Center Report NoEERC 2003-06.

DRAFT

References


Seed, R. B, and Harder, L. F.(1990). SPT-based analysis of cyclic pore pressuregeneration and undrained residual strength,in Proceedings, Seed Memorial Symposium,J. M. Duncan, ed., BiTech Publishers,Vancouver, British Columbia, 351–76.

Sherard, J.L. (1967). Earthquakeconsiderations in earth dam design. Journalof the Soil Mechanics and FoundationsDivision, American Society of CivilEngineers, Vol. 93, No. SM4, 377–401.

Somerville, P.G., and G. Gibson (2008).Seismic hazard at Hinze Dam.Proceedings of the ANCOLD 2008Conference, Gold Coast, Queensland,November 12-15.

Somerville, P.G., R.W. Graves, N.F. Collins,S.G. Song, S. Ni and P. Cummins (2009).Source and ground motion models ofAustralian earthquakes. Proceedings of the2009 Annual Conference of the AustralianEarthquake Engineering Society, Newcastle,December 11-13.

Somerville, P.G. and H.K. Thio (2011).Epistemic Uncertainty in Seismic HazardAnalysis. Proceedings of the 2011ANCOLD Conference, Melbourne, October27-28.

Somerville, P.G. (2009). Probabilisticseismic hazard analysis. Proceedings of theAustralian Geomechanics Society 2009Seminar, Seismology and EarthquakeEngineering, Adelaide, September 24.

Somerville, P., P. Quijada, H.Kie Thio, M.Sandiford, and M. Quigley (2008).Contribution of Identified Active Faults toNear Fault Seismic Hazard in the FlindersRanges. Proceedings of the 2008 AnnualConference of the Australian EarthquakeEngineering Society, Adelaide, November11-13.

Somerville, P.G. and Thio, H.K. (2011).Development of ground motion time-histories for seismic design. Proceedingsof the Ninth Pacific Conference on

Earthquake Engineering Building anEarthquake-Resilient Society, 14-16April, 2011, Auckland, New Zealand

Stark, T.D., Beaty, M.H., Byrne, P.M.Castro, G., Walberg, F.C., G. Perlea, V.G.,Axtell, P.J., Dillon, J.C., Empson, W.B. andDavid L. Mathews D.L.(2012). Seismicdeformation analysis of Tuttle Creek Dam.Can. Geotech. J. 49: 323–343 (2012).

Swaisgood, J.R. (1998). Seismically-induceddeformation of embankment dams. 6th USNational Conference on EarthquakeEngineering, Seattle, Washington.

Swanson (1992). A computer programdesigned by Swanson Analysis Systems, Inc.

Sy A. and Campanella R.G. (1994).Becker and standard penetration tests(BPTSPT) correlations withconsideration of casing friction.Canadian Geotechnical Journal, 31(3):343-356.

Toose, J., Hunter, G. and Watkins, G(2007). Design and Construction of theLance Creek Reservoir Seismic Upgrade,NZSOLD/ANCOLD Conference onDams.

Travasarou, T., Bray, J.D., and Abrahamson,N.A. [2003]. “Empirical attenuationrelationship for Arias Intensity,” EarthquakeEngineering and Structural Dynamics; 32,1133–1155.

Tsukamoto, Y. Ishihara, K., Nakazawa, H.,Kamada, K., and Huang, Y. (2002)Resistance of partially saturated sand toliquefaction with reference to longitudinaland shear wave velocities. Soils andFoundations, 42,(6), 93-104.

USACE (1999) EM 1110-2-6050 ResponseSpectra and Seismic Analysis of HydraulicStructures United States Army Corps ofEngineers.

USACE (2003a) EM 1110-2-2400 Design ofOutlet Works. United States Army Corps ofEngineers.

DRAFT

References


USACE (2003b) EM 1110-2-6051 Time-History Dynamic Analysis of HydraulicStructures. United States Army Corps ofEngineers.

USACE (2005) Determination of StandardResponse Spectra and Effective Peak GroundAccelerations for Seismic Design andEvaluation USACE, ERDC/CHL CHETN-VI-41, December

USACE (2007) EM 1110-2-6053Earthquake Design and Evaluation ofConcrete Hydraulic Structures. United StatesArmy Corps of Engineers.

USBR (2011) “Seismic Induced Loads onSpillway Gates: Phase 1 Literature Review”USBR Report DSO-11-06, September, 2011.

USBR (2006) State-of-Practice for the Non-linear Analysis of Concrete Dams at theBureau of Reclamation US Bureau ofReclamation, Denver, Co.

USBR (1989). Design Standards.Embankment dams, No. 13, Chapter 13.Seismic design and analysis. US Bureau ofReclamation, Denver, Co.

USNRC (1985). Liquefaction of soils duringearthquakes. National Academy Press,Washington DC.

Van Houtte, C., Drouet, S., and Cotton,F. (2011). Analysis of the origins of k(kappa) to compute hard rock to rockadjustment factors for GMPEs, Bull.Seism. Soc. Am. 101, 2926–2941.

Wang, J.N., (1993) Seismic Design ofTunnels – A simple State-of-the-Art DesignApproach. Parsons Brinkerhoff Quade andDouglas Monograph 7.

Wang, Z.L. and Makdisi, F.I. (1999).Implementing a bounding surface hypoplasticity model for sand into the FLACprogram, FLAC and Numerical Modelling inGeomechanics, A.A. Balkema, Netherlands,483-490.

Watson-Lamprey, J. and Abrahamson, N.[2006]. “Selection of ground motion timeseries and limits on scaling.” Soil Dynamicsand Earthquake Engineering, Vol. 26, 477–482.

Wesnousky, S.G., C. Scholz,C.H.,Shimazaki, N. andMatsuda,T.(1983). Earthquake frequencydistribution and the mechanics offaulting. Journal of GeophysicalResearch 88, 9331-9340.

Westergaard, H.M. (1933) Water Pressureon Dams during Earthquakes, ASCETransactions, Vol 98, pp 418-472.

Wieland, M. (2005). Hydrodynamicpressures, Int. Journal Water Power and DamConstruction, October, pp 18-24.

Wieland, M. and Chen, H. [2010] Lessonslearnt from observed damage of dams andpower plants caused by the May 12, 2008Wenchuan Earthquake, The DamsNewsletter, February 2010, No. 9.

Yang, J., Savidis, S., and Roemer, M. (2004)Evaluating liquefaction potential of partiallysaturated sand. J.Geotech and Geoenvir.Eng. 130 (9), 975-979

Youd, T.L. (2011) A look inside the debateover EERI Monograph MNO 12, BrighamYoung University. Available from:http://www.cgea.org/downloads/calgeo-ac-2011.pdf

Youd, T.L., and Perkins, D.M. (1978)Mapping of liquefaction-induced groundfailure potential, Journal of GeotechnicalEngineering, ASCE Vol104, No4, 433-446.

Youd, T.L., Idriss, I.M., Andrus, R.D.,Arango, I., Castro, G., Christian J.T., Dobry,R., Finn, W.D.L., Harder, L.F., Hynes, M.E.,Ishihara, K., Koester, J.P., Liao, S.S.C.,Marcuson, W.F., Martin, G.R., Mitchell,J.K., Moriwaki, Y., Power, MS., Robertson,P.K., Seed, R.B. and Stokoe, K.H. (2001).Liquefaction resistance of soils: summaryreport from the 1996 NCEER and 1998NCEER/NSF Workshops on evaluation of

DRAFT

References


liquefaction resistance of soils. J.Geotechnical and GeoenvironmentalEngineering, ASCE, Vol. 127, No. 10, 817–834.

Zangar, C.N. (1952) “HydrodynamicPressures on Dams Due To HorizontalEarthquake Effects” EngineeringMonograph No. 11 USBR 1952

Zienkiewicz, O.C., Valliappan, S. and King,I.P. (1986). Stress analysis of rock as a non-tension material. Geotechnique, 18, 55–56.

Ziotopoulou, K, and Boulanger, R.W.(2012). Constitutive modeling of durationand overburden effects in liquefactionevaluations. Proc. Second Int. Conf. onPerformance-Based Design in EarthquakeGeotechnical Engineering. Taormina, Italy.

DRAFT

Appendix A


APPENDIX A MODIFIED MERCALLI SCALE 1956 VERSION (Richter, 1958; Hunt, 1984).

Intensity EffectsI Not felt. Marginal and long period effects of large earthquakes.II Felt by persons at rest, on upper floors, or favourably placed.III Felt indoors. Hanging objects swing. Vibration like passing of light trucks. Duration

estimated. May not be recognised as an earthquake.IV Hanging objects swing. Vibration like passing of heavy trucks or sensation of a jolt

like a heavy ball striking the walls. Standing motor cars rock. Windows, dishes,doors rattle. Glasses clink. Crockery clashes. In the upper range of IV wood wallsand frames creak.

V Felt outdoors, duration estimated. Sleepers wakened. Liquids disturbed, some spilled.Small unstable objects displaced or upset. Doors swings, close, open. Shutters,pictures move. Pendulum clocks stop, start, change rate.

VI Felt by all. Many frightened and run outdoors. Persons walk unsteadily. Windows,dishes, glassware broken. Knickknacks, books, etc. off shelves. Pictures off walls.Furniture moved or overturned. Weak plaster and masonry D cracked. Small bellsring (church, school). Trees, bushes shaken (visibly, or heard to rustle – CFR).

VII Difficult to stand. Noticed by drivers of motor cars. Hanging objects quiver.Furniture broken. Damage to masonry D, including cracks. Weak chimneys broken atroof line. Fall of plaster, loose bricks, stones, tiles, cornices (also unbraced parapetsand architectural ornaments – CFR). Some cracks in masonry C. Waves on ponds;water turbid with mud. Small slides and caving in along sand or gravel banks. Largebells ring. Concrete irrigation ditches damaged.

VIII Steering of motor cars affected. Damage to masonry C, partial collapse. Somedamage to masonry B, none to masonry A. Fall of stucco and some masonry walls.Twisting, fall of chimneys, factory stacks, monuments, towers, elevated tanks. Framehouses moved on foundations if not bolted down; panel walls thrown out. Decayedpiling broken off. Branches broken from trees. Changes in flow or temperature ofsprings and wells. Cracks in wet ground and on steep slopes.

IX General panic. Masonry D destroyed; masonry C heavily damaged, sometimes withcomplete collapse; masonry B seriously damaged. (General damage to foundations –CFR). Frame structures, if not bolted, shifted off foundations. Frames cracked.Serious damage to reservoirs. Underground pipes broken. Conspicuous cracks inground. In alluviated areas sand and mud ejected, earthquake fountains, sand craters.

X Most masonry and frame structures destroyed with their foundations. Some well-builtwooden structures and bridges destroyed. Serious damage to dams, dikes,embankments. Large landslides. Water thrown on banks of canals, rivers, lakes, etc.Sand and mud shifted horizontally on beachheads and flat land. Rails bent slightly.

XI Rails bent greatly. Underground pipelines completely out of service.XII Damage nearly total. Large rock masses displaced. Lines of sight and level distorted.

Objects thrown into the air.Note: Masonry A, B, C, D. To avoid ambiguity of language, the quality of masonry, brick or otherwise, is specifiedby the following lettering (which has no connection with the conventional Class A, B, C construction).– Masonry A: Good workmanship, mortar, and design; reinforced, especially laterally, and bound together by using

steel, concrete, etc.; designed to resist lateral forces;– Masonry B: Good workmanship and mortar; reinforced, but not designed to resist lateral forces;– Masonry C: Ordinary workmanship and mortar; no extreme weaknesses such as non-tied-in corners, but masonry

is neither reinforced nor designed against horizontal forces;– Masonry D: Weak materials, such as adobe; poor mortar; low standards of workmanship; weak horizontally;– CFR indicates additions to classification system by Richter (1958).

DRAFT

ANCOLD Guidelines for Design of Dams and … · ANCOLD Guidelines for Design of Dams and...

Documents

Transcript of ANCOLD Guidelines for Design of Dams and … · ANCOLD Guidelines for Design of Dams and...