Paper Seed

15
""'- 350 Mitchell and Tseng [13] Duncan, J.M., Byrne, P. and Wong, KS., "Strength, Stress-Strain and Bulk Modulus Parameters for Finite Element Analyses of Stresses and Movements in Soil Masses", Geotechnical Engineering Report No. UCBjCTj80-01, University of California, Berkeley, August 1980. De Alba, P., Seed, RB., and Chan, C.K, "Sand Liquefaction in Large-Scale Simple Shear Tests", Journ. Geot. Engrg., ASCE, Vol. 102, No.GT9, Sept. 1976, pp. 909-927. Seed, H.B. and Lee, KL., IIUndrained Characteristics of Cohesionless Soils", Journ. Soil Mech. and Found. Div., ASCE, Vol. 93, No. SM6, Nov. 1967, pp. 117-141. ~y~ Corporation, IIFirst Report on In Situ Site Investigation for Determination of Liquefaction Potential", Prepared for the U.S. National Bureau of Standards, the U.S. Bureau of Rec1amation, and the Japanese Public Works Research Institute, March 1984. [14] [15] [16] '/ 17 SPT-Based Analysis of Cyclic Pore Pressure Generation and Undrained Residual Strength * ** Raymond B. Seed and Leslie F. Harder, fr. INTRODUCTION Prominent among Professor Harry Seed's major contributions to geotechnical en~ineering was the development of methods for evaluation of the seismic stabllity and performance of dams comprised of, or founded on, soils potentially vulnerable to liquefaction. This subject continued to interest bim over the past twenty years, and tbe analytical procedures which he and his colleagues developed continued to evolve tbroughout this periodo This paper, based on recent studies and work in progress at the time of his death, presents a surnmary overview of the most current techniques advocated by DI. Seed and his colleagues for such studies. These will be presented and illustrated by means of application to the re-evaluation of the performance of the Lower San Fernando Dam during the 1971 San Fernando Earthquake, providing a c1ear demonstration of the methods advocated and facilitating assessment of their performance when applied to tbis important case bistory. SUGGESTED APPROACH Numerous engineers and investigators have contributed significantly over the past 20 years to the development of the engineering methods described berein. Many of these individuals worked closely witb Dr. Seed, wbile many others worked independently. Though too numerous to individually cite herein, their contributions are gratefully acknowledged. ~«:;: *Assoc. Prof. of Civil Engineering, University of California, Berkeley, CA 94720. **Supervising Engineer, Calif. Dept. of Water Resources, 1416 Ninth S1.,Sacramento, CA 94236. 351

Transcript of Paper Seed

  • ""'-

    350 Mitchell andTseng

    [13] Duncan,J.M., Byrne,P. andWong,KS., "Strength,Stress-StrainandBulk ModulusParametersfor Finite ElementAnalysesof StressesandMovementsin Soil Masses",GeotechnicalEngineeringReportNo. UCBjCTj80-01, Universityof California,Berkeley,August1980.De Alba, P., Seed,RB., and Chan,C.K, "SandLiquefactioninLarge-ScaleSimpleShearTests",Journ. Geot.Engrg.,ASCE, Vol.102,No.GT9,Sept.1976,pp.909-927.Seed, H.B. and Lee, KL., IIUndrained CharacteristicsofCohesionlessSoils",Journ.Soil Mech.andFound.Div.,ASCE, Vol.93,No.SM6,Nov.1967,pp.117-141.~y~ Corporation,IIFirstReport on In Situ Site InvestigationforDeterminationof LiquefactionPotential",Preparedfor the U.S.NationalBureauof Standards,theU.S.Bureauof Rec1amation,andtheJapanesePublicWorksResearchInstitute,March1984.

    [14]

    [15]

    [16]

    '/

    17SPT-BasedAnalysisofCyclicPorePressure

    GenerationandUndrainedResidualStrength

    * **RaymondB. Seed andLeslieF. Harder,fr.

    INTRODUCTION

    Prominent among Professor Harry Seed's major contributionstogeotechnicalen~ineeringwasthedevelopmentof methodsfor evaluationoftheseismicstabllityandperformanceof damscomprisedof, or foundedon,soils potentiallyvulnerableto liquefaction. This subjectcontinuedtointerestbimoverthepasttwentyyears,andtbeanalyticalprocedureswhichheandhiscolleaguesdevelopedcontinuedtoevolvetbroughoutthisperiodoThis paper,basedonrecentstudiesandworkin progressat thetimeof hisdeath,presentsa surnmaryoverviewof the most currenttechniquesadvocatedby DI. Seedandhis colleaguesfor suchstudies.Thesewill bepresentedandillustratedbymeansof applicationtothere-evaluationof theperformanceof the Lower San FernandoDam during the 1971SanFernandoEarthquake,providinga c1eardemonstrationof the methodsadvocatedandfacilitatingassessmentof theirperformancewhenappliedtotbisimportantcasebistory.

    SUGGESTED APPROACH

    Numerousengineersandinvestigatorshavecontributedsignificantlyoverthepast20yearsto thedevelopmentof theengineeringmethodsdescribedberein.Manyof theseindividualsworkedcloselywitbDr. Seed,wbilemanyothersworkedindependently.Thoughtoo numerousto individuallyciteherein,theircontributionsaregratefullyacknowledged.

    ~:;:

    *Assoc.Prof. of Civil Engineering,Universityof California,Berkeley,CA94720. **SupervisingEngineer,Calif. Dept.of WaterResources,1416NinthS1.,Sacramento,CA 94236.

    351

  • 352 SeedandHarder

    The followingproceduresfor evaluationof the seismicstabilityandperformanceof damscomprisedof, or foundedon,potentiallyliquefiablesoilsareevolvedfromproceduresoriginallydevelopedbyDr. Seedandhiscolleaguesin the early1970's[1,2]. The mostslgnificantchangessincetheirearlyinceptionare: (a)increasedrelianceonin-situtestsasaprimarybasisfor evaluationofsoilconditions,and(b)morerecentincorporationofresidualor "steadystate"strengthevaluationasa partof overallstabilityandperformanceassessment.Numerousother,moresubtlerefinementsinanalyticalproceduresand recommendedcorrelationswill be presentedwithinthecontextoftheoverallpresentation.

    In simple,conciseterms,the recommendedapproachfor evaluationofliquefactionpotentialand seismicstabilityand performanceof damsinvolvestenbasicstepsasfollow:

    1.2.

    Determinethecross-sectionof thedamtobeusedforanalysis.Determine,withthecooperationof geologistsandseismologists,themaximumor mostseveretimehistoryof baseexcitationtowhichthedamanditsfoundationmightbesubjected.Determine,as accurately~,as possible,the stressesin theembankmentbeforetheearthquake;thisismosteffectivelydoneusingfiniteelementanalysistechniques.Determinethe dynamicpropertiesof the soilscomprisingthedam and foundation,such as shear modulus, dampingcharacteristics,and bulk modulusor Poisson'sratio, whichdeterminedynamicresponsecharacteristics. Since thesematerialcharacteristicsare nonlinear,it is also necessarytodeterminehowthepropertiesvarywithstrain.Compute,usingan appropriatedynamicfiniteelementanalysisprocedure,the acceleratlOnsandjor dynamicshear stressesmducedwithinthe embankmentandfoundation.For narrowcanyonswheretheratioof crestlength:maximumdamheightislessthanabout3:1,three-dimensionaleffectsassociatedwiththecanyonsidewallsmustbe accountedfor in theseanalyses.Two-dimensionalplane strain dynamicresponseanalysesaresufficientfor mostdamgeometrieswhere"crestlen~this longrelativeto embankmentheight. One-dimensional(columnar)analysesarenotgenerallyrecommendedfor dynamicresponseanalysisof dams,butcanprovidesufficientlyaccurateresultsforsomecases.Evaluatetheresistanceofpotentiallyliquefiablesoiltypeswithinthedamandfoundationtoporepressuregenerationundercyclicloadingconditions.This isprobablymosteffectivelydoneatthistimeusingStandardPenetrationTest(SPT) data,thoughrecentimprovementsin correlations between SPT and ConePenetrationTests(CPT) mayrenderCPT a viablealternativewhen soil conditions and geometryare already well-characterizedbymeansofboringsorotherinformation.Basedon the resultsof Steps6 and 7, evaluatelikely porepressuregenerationor cyclicstrainaccumulationwithinthedamandfoundation.This constitutesan evaluationof thepotentialfor "triggering"or initiatingsoilliquefaction.

    3.

    4.

    5.

    6.

    7.

    CyclicPorePressure 353

    If porepressuregenerationin Ster 7 is foundto bepotentiallysignificant,evaluatethe residua undrained("steadystate")strengthsof thedamandfoundationsoils.This is probablymostreliablydoneat presentbasedon SPT data,or on CPT datacorrelatedwith SPT-basedrelationships.Using the resultingresidualstrengths,evaluatetheoverallstabilityof thedamandfoundation. This constitutesa "post-triggering"stabilityevaluation.If Step6 andjorStep8showthedamto besafewithrespecttoeither"triggering"or post-triggeringmanifestationof majorslidemovementsor deformations,evaluatethemagnitudeof (limited)overalldeformationslikelyto resultfrom combinedstaticanddynamicloads,andassesstheirpotentialimpactondamstabilityandperformance.Be sureto incorporatethe reguisiteamountof judgementineachof thesteps(1) through(9),beingguidedby a thoroughknowledgeof typicalsoil characteristics,theessentialsof finiteelementand dynamicresponseanalysisprocedures,and adetailedknowledgeof thepastperformanceof embankmentsinotherearthquakes.

    The importanceof engineeringjudgmentat each step cannot beoveremphasized.A goodexampleis Step5: seismicresponseanalyses.Atthis stagethereis alwaysa needto weighand balancethe meritsofsimplificationandassociatedeaseofparameterdeterrninationandanalysisagamstthepotentiallossofaccuracyassociatedwithoversimplification.Forexample,talldamsin narrowcanyonsaresubjectto significant3-D effectsassoclatedwith the canyonsidewalls,andconventional2-D planestraindynarnicresponseanalysescanbernisleading.On theotherhand,fully3-Dresponseanalysesarerenderedextremelydifficultby theoverallsizeandcomplexityof theproblem,andtheinabilityto generateanddevelopfulIy3-D solutionswitha sufficiently"fine"meshcanalsoadverselyimpacttheresults.An optimumsolution(or somecasescanbe theuseof 2-D,planestrainanalysesbutwithfictitiouslyhighdynarnicshearmoduliusedtomodelthe additionalrestraintprovidedby the canyonsidewalls.The resultingcalculateddynamicshearstrainswithintheembankmentandfoundationcan then be "post-processed",using the actual (not fictitiouslyhigh)dynarnicshearmoduli,to generategoodestimatesof theactual,3-D cyclicshearstresses.This, of course,requiresconsiderablejudgment,andit isdifficulttoreducesuchaproceduretoasimplesetof "roles".

    8.

    9.di''i,;i~1

    10.

    Similarly,at the other extreme,simple one-dimensional(columnar)dynamicresponseanalysescanprovidesufficientlyaccuratecyclicshearstressesasto representa reliablebasisfor liquefactionanalysesfor somecases.~.houghthesesimplisticanalyses"tendto providenotoriouslypoorpredictionsof accelerations,strainsandcyclicstressesnearthecrestandslopingfacesof dams(dueto theirinabilityto correctlymodelthe localproblemgeometryandjor "topographicamplification"),theycanprovidereasonablya9curateestimatesof cyclicshearstressesdeepwithin theinteriorof anembankmentor itsfoundation,especiallyonewithrelativelyflatfaceslopesandahighratioofcrestlengthtodamheight.

    .",.dl

  • 354 SeedandHarder

    THE LOWER SAN FERNANDO DAM SLIDE OF FEBRUARY 9,1971

    TheLowerSanFernandoDamin Californiadevelopedamajorslidein theupstreamslopeanderestasa resultof the1971SanFernandoearthquake(ML::::6.6).An investigationof theslide,includingtrenchesandborings,insitudensitytests,undisturbedsampling,indextesting,staticandcyclieloadtesting,andanalyseswasperformedandreportedbySeedetal.[1,3,4]andLeeetal. [5].The fieldinvestigationshowedthattheslideoccurredduetoliquefactionof azoneof sandyandsiltyhydraulicfill nearthebaseof theupstreamshell.

    Two erossseetionsof theLowerSanFernandoDamarepresentedin Fig.1,oneshowingtheobservationsmadein a trenchexeavatedthroughtheslideareaandtheothershowinga reconstructedcrosssectionof thedam,illustratingthe zone in which liquefactionoceurred. Large blocksofessentiallyintaetsoilfromtheupstreamsectionof thedammovedintothereservoir,ridingoveror "floating"on theliquefiedsoil. After movementsstopped,theliquefiedsoilwasfoundtohaveextIlldedoutbelowthetoeofthedamandup betweentheintactblocks,withmaximummovementsasmuehas200ft (61m)beyondthetoeof thedam.Theblockof soilwhicheontainedthetoeof thedammovedabout150ft (46m)intothereservoir.

    Datafromseismoscopesloedted'on theabutmentandon theerestof theembankmentindieatedpeakaeeelerationsof about0.55gatbothloeations,andananalysisof theseismoscoperecordon thedamerestindicatedthattheslideoeeurredabout20to30seeondsaftertheearthquakeshakinghadstopped. Thus the largeslide movementsapparentlydevelopedin theabseneeof earthquake-inducedstressesandwereeausedonlybythestaticstressesdueto theweightof thematerialsin theembankment.It canthusbe inferredthattheearthquakeshakingtriggereda lossof strengthin thesoilscomprisingtheembankmentandthatitwasthislossof strength,ratherthanthemertiaforcesinducedbytheearthquakeshaking,whichledto theslidingoftheupstreamslope.. -"Whileit is readilyapparentthatslidingdueto liquefaciionoecurredin theupstreamshellof theembankment,performancedatafrom'thefilesof theCityof Los AngelesDepartmentof WaterandPowershowthatthewaterlevelsmeasuredin wells installedin the downstreamshellshowedonlysmallchangesin elevationasa resultof theearthquakeshaking[6]. Thesewellsreadingswereobtainedonefull dayaftertheearthquake,andso didnot providea reliableindicationof peak pore pressuredevelopment.Nonetheless,the well readings,alongwith the abseneeof significantdeformationsof thedownstreamshellof thedam,appearto indicatethatwhile the earthquakemayhavecausedat leastsomeincreasein porepressureratio.in the downstreamshell andits foundation,therewas nosignificantextentof soil liquefaetionin the downstreamportionof thehydraulicfill.

    / .

    ..J ~ z::! ..J

  • 356 SeedandHarder

    This sliderepresentsa uniquelyimportantcasehistoryfor calibrationandverificationof liquefactionandseismicstabilityanalysismethodologiesforslopesand embankments.Historical constructionrecordsas well asextensivepost-earthquakegeotechnicalinvestigationsprovideunusuallygoodinformationregardingembankmentandfoundationgeometryandsoilconditions.Instrumentalrecordingsprovidea goodbasisfor developmentof inputaccelerations,aswell asfor evaluationof theresultsof dynamicresponseanalyses.Most importantof all, however,is the fact thatthelargelysimilarupstreamanddownstreamzonesof the dambehavedinhighlydissimilarfashion,withtheupstreamsectionslidingmorethan100feetintothereservoir, whilethedownstreamsectionremainedstableandexperiencedonlyminordeformationsof less thana footatmostlocations.Accordingly,usingwell-definedconditions,a reliableanalysismethodologymustbeabletousetherelativelyminordifferencesbetweentheupstreamanddownstreamportionsof thedamas a basisfor accuratelypredictingthesesignificantlydifferentobservedmodesofperformance.

    THELOWERSANFERNANDODAM

    The internalgeometryandsP!1conditionswithinthedamandfoundationarewell-establishedasa resultof extensivegeotechnicalstudiesperformedin theearly1970's,andasecondsetof studiesperformedoverthepastfiveyears[1,5,6,7,8]. Figurel(b) showsa typicalcross-sectionthroughtheLower San FernandoDam as it existedimmediatelyprior to the 1971earthquake.Embankmentconstructionbeganin 1912.The embankmentwasfoundedon an alluvialfoundationconsistingprimarilyof stiff clayeysoilswithlayersandlensesofsandandgravel.

    The majorityof theembankmentconsistsof hydraulicfill placedbetween1912and 1915.This materialwassluicedfromthefloorof thereservoiranddischargedfromstarterdikeson.theupstreamanddownstrea~edgesof the embankment.The actualdlmenslOnsof thesestarterdlkesareunknown.The hydraulicfill processresultedin upstream,anddownstreamshellsconsistingprimarilyof sandsandsiltsanda centrarcoreconsistingprimarilyof clayeysoils. Constructionphotos of the hydraulicfillplacementandpastreportsindicatethattheupstreamand downstreamsectionswereraisedsymmetricallyand constructedin a similarmanner.Therefore,it isreasonabletoassumethatthegeneralcharacteristicsof theupstreamanddownstreamhydraulicfill shellsaresimilar.

    A 10-to._15-foot-thickhydraulicfill layerconsistingof "ground-up"shalefromthe left abutmentwasplacedin 1916overthe initial hydraulicfilldescribedabove.Umitedsamplingof thegroundshalein 1985disclosedawidelygradedsandandsiltysand,andconstructionrecordsindicate thatthemaximumparticlesizeof thegroundshalewasabout3 inches.

    Theembankmentwasraisedanumberof timesbetween1916and1930byplacementof rolledfills.Themaximumheightoftheembankmentofabout140feetwasreachedin 1930.A thinblanketwasplacedonthelowerpart

    ---: ,

    CyclicPorePressure 357

    of thedownstreamslopein 1929and1930,apparentlyfor seepagecontroland to provideadditionalstabilitydueto the raisingofthe crest. Thecomposltionof theblanketwasdescribedin apost-constructionreportasamixtureof shale and gravellymaterialplacedin 12-inchlayersandcompactedbytrucks.The finaladditionto thedamwasa 4.5H:1V(rolledfill) bermplacedonthedownstreamslopein 1940.

    EVALUATION OF STATIC AND CYCLIC STRESSES

    ~

    An earIystageof theanalysisprocessinvolvesevaluationof thestaticstressconditionswithintheembankmentandfoundation.Theseareimportantintwowayswithintheoverallanalyticalscheme:(a)staticstresses,especiallyeffectiveconfiningstresses,can significantlyinfluencedynamicresponsecharacteristics(e.g.dynamicshearmoduli),and so can influencecyclicstresscalculations,and(b) staticstresses,principlytheeffective(vertical)overburdenstress(a'o)andthestatic"drivmg"shearstressona horizontalplane (rhv)can exerta significantinfluenceon the resistanceto porepressuregenerationor "triggering"of liquefactionat anypointwithintheembankment.

    Seedetal. [1]presentedresultsof staticfiniteelementanalysesof LowerSanFernandoDamperformedusingthecomputerprogramISBILD [9].Asecond,morerecentanalysisperformedin 1987is presentedbySeedetal.[6,10],usingthefiniteelementmeshshownin Figure2. Both analysesusedsimilarmodellingprocedures,andbothyieldedsimilarresults.Thistypeof analysisis, in general,relativelyinsensitiveto choiceof computercode,solutionmethodology,andeventypeof constitutivemodel. Thereare,however,severalvitalcomponentsto suchananalysis:(1)nonlinear,stress/and stress-Ieveldependentsoil behaviormust be adequatelymodelled,(2) soilmodelparametersmustprovideadequatemodellingofthe in situ soils, and (3) hydraulicforces(e.g./ bouyancyandseepagegradients)mustbeincludedin theanalyses.

    The 1987analysesusedtheprogramFEADAM84 [11],a 2-D planestrainfinite elementcodeo Nonlinearsoil behaviorwas modelledwith the"hyperbolic"modelproposedbyDuncanetal. [12],asmodifiedbySeedandDuncan[11,13].Theanalysismodelledconstructionof theembankmentinaseriesof "steps",placingnewelementsin theactualconstructionsequence.

    i1Fig.2: FiniteElementMesh-LowerSanFernandoDam

    ,,..

    ./- ,.,- - .L:.---- '--I-r---.-/f' - 1.-1.>---

    K.: f--r.-. .L'"I \--I I I r ---I I

  • 358 SeedandHarder

    Soilsbelowthewatertableweremodelledusingeffectiveor "bouyant"unitweightstoaccountforbouyancyin evaluatingeffectivestresses.Theeffectsof seepageforceswereevaluatedbasedonaninitialflownetanalysisfromwhichseepageforceswerederived;thesewerethenappliedasequivalentnodalforces. The final calculatedeffectivestresseswlthineachelementprovidedthe key staticstresses(ar';)and 1hv)necessaryfor subsequentstagesof theoverallanalyticalprocess.

    Seedet al. [1,4] presenteddynamicresponseanalysesof theLowerSanFernandoDam performedin 1972using the programQUAD4 [14].Similaranalyseswereperformedmorerecentlyusingthe codeFLUSH[15],and the meshillustratedin Figure 2. Both analysesusedstaticstresses,calculatedas describedabove,as a basisfor modellingdynamicshearmoduliof cohesionlesszones,thoughslightlydifferentrelationshipswereusedto modelthenonlinearrelationshipsbetweenshearstrainanddynamicshearmodulusanddamping:[16]fortheearlieranalyses,and[17]for themorerecentanalyses.Similarly,therelationshipsproposedin [16]wereusedtomodelstrain-dependentmodulianddampingin thecohesivezonesin theearlieranalyses,and[18]in themorerecentanalyses.

    The earlieranalysesusedtheinputmotionsdescribedbySeedetal. [1,4]:(a) an interpretationof the abutmentrecordby Scott [19],and (b) amodifiedversionof thetimehistoryrecordedatthePacoimastationduringthe 1971San Fernando!,/earthquake. The modificationsconsistedoftrimrningof accelerationpulsesof greaterthan0.9g, thenscalingtherecordtoamaximumhorizontalaccelerationof 0.6g,providingamotioningoodagreementwith.Scott's[20]interpretationof the 1971seismoscoperecordfromtheabutmentof theLowerSanFernandoDam,butwithouttheunusuallowfrequencycomponentsof theinterpretedabutmentrecord.The morerecentanalysesemployedthemodifiedPacoimainputmotionscaledto0.55g.

    The resultsof the1973and1987analyseswerein closeagreement:bothproducedmaximumhorizontalcrestaccelerationson theorderof 0.5to0.55g,in goodagreementwiththeactualrecorded-peakcrestaccelerations.Bothanalysesalsocalculatedsimilarpeakcyclichorizontalshearstresses(1hv,cycliJwithinthehydraulicfill zonesof theembankine~:

    \\

    It is interestingto note that Jong [20] performedone-dimensional,columnar analysesof individual vertical soil columns through theembankmentusingtheprogramSHAKE [21]. Theseanalyses,modellin~verticalpropagationof shearwavesand usingthe samenonlinears011modelsand soil parametersas the 2-D FLUSH analyses,significantlyunderestimatedbothaccelerationsandcyclicshearstressesnearthecrestand upper facesof the embankment.These analysesalso, however,providedrelativelygood agreementwith the 2-D dynamicresponseanalyseswithregardto cyclicshearstresses(11vcy}:lidwithinthehydraulicfill zonesnearthebaseof theembankment,typlcallycalculatingpeakCYclicshearstressesonly5% to 15%lower thanthosecalculatedby the 2-Danalysesin thesezones.

    CyclicPorePressure 359

    EVALUATION OF LIQUEFACTION RESISTANCE

    Havingcalculatedthecyclicshearstressesresultingfromtheearthquakeloadingat eachpointwlthinthehydraulicfill, thenextstepis to evaluatethe resistanceof this materialto cyclicpore pressuregenerationoraccumulationof cyclicshearstrain. This constitutesevaluationof theresistanceto "triggering"or initiationof potentialliquefactionfailure,definedas sufficientpore pressureor strainaccumulationto bringthematerialto a conditionat whichundrainedresidual(or "steadystate")strengthwill controlfurtherbehavior.

    Figure 3 showsa recommendedrelationshipbetween"corrected"SPTpenetrationresistanceand the equivalentuniform cyclic stressratiorequiredto "trig~er"liquefactiondunnganearthquakewitha duration(ornumberof loadmgcycles)representativeof a typicalearthquakewith amagnitude of M ::::7~,as suggestedby Seedet al. [22,23]. In thisrelationship,cyclicstressratio (CSR) is definedas theratioof theCYclicshearstressactingon a horizontalplane (1hvc) to the initial (pre-earthquake)effectivevertical or overburdenstress(a'o), as CSR =(1hvc)ja'O'The relationshipspresentedin Figure3 representasignificantimprovementoverearlier,slIllilarrelationshipsdevelopedbyDr. SeedandhiscoUeagues,as(a)theydirectlyaccountfortheinfluenceof finescontenton the relationshipbetweenpenetrationresistanceand li~uefactionresistance,and(b) theyarebasedon a "corrected"or "standardlzed"SPTpenetrationresistance.

    The standardizedpenetrationresistance(N1)60is a new"standard"SPTblowcount,basedonstandardizedequipmentandproceduresaspresented(inpart)in Table1[22,23]. Theuseof othertypesof harnmer(e.g.donutharnmers),or othertypesof mechanismsto ralseanddrop the harnmer(e.g.automaticmechanical"trip"harnmers,"freefaU"harnmers,ropeandcatheadwiththreeturnsof therapeaboutthecathead,etc.),canimpartdifferentlevelsof energyto thetopof thedrill stem.Thesenon-standardpraceduresandequipmentrequirecorrectionof theblowcountsin ordertodevelopthestandardizedblowcount.The (N1)60"standardized"systemandprocedures,combinedwitha "typical"ropeandcatheadsystem(withtwoturnsof therafe aboutthecathead)typicaUydeliverap'proximately60%ofthe theoretica"freefaU"harnmerenergyto the dnU stem. For othersystems,the measuredpenetrationresistances(N, blowsjft) shouldbecorrectedas

    ERN60 =Nx 60%

    '~

    >'!Q

    whereER or energyratiois the"efficiency"or percentof theoreticalfreefaUenergydeliveredbytheharnmersystemactuaUyusedto thetopof thedrill stem.This canbemeasureddirectly,usinga pileanalyzer,or canbeestimated(for the mostcornmonalternatesystemsin wldespreaduse)basedoncorrelationsanddatasummarizedbySeedetal. [22,23].

  • 360 SeedandHarder

    0.6.Z9 ..o

    15II

    ~5I

    0.51

    04_o~g'

    ..:.

    IIIIIII

    "'

    ."

    ,,I,,,I

    I

    Noto:' Datashownonlylonlt.. with>5" tinos.oon~dorabl.datalor

  • 362 Seedand Harder

    00

    CN0.8

    '00

    OA 1.2 1.6

    80

    g- 2ea..a.:;

    iO

    .~/!.

    ee"

    ~6~ . D.''' """'O,'21~o/;:;oJ

    O om, ' by..Id , ,e>~J! 8;;

    . o , " -o'n

    oO 00 60,o 20 30 .0

    IN,'.o10

    Fig.4: ChartforValuesofCN Fig.5: ApproximateRelationshipBetweenDR and(Nl)60 [25]

    ,'/

    Table2: RelationshipBetweenMagnit~de,NumberofEquivalentUniformLoadCycles,andLiquefactionResistanceFactorCM

    loadingcyclesrepresentingthe earth'l.uakein question(afterSeedet al.,[27]),andthenusethesecondandthlrdcolumnsof Table2 to selectanappropriatevalueofCM, Eitherof theseproceduresresultin relationshipsbetween(Nl)60andCSRforearthquakesofothermagnitudesthanM"",7~thatarein goodagreementwithavailablefielddata[22,23],thoughthefield (casehistory)databaseis less extensivefor eventsof otherrangesofmagnitude.

    Virtuallyall of thefield (casehistory)datareflectedin Figure3 (andinsimilarcollectionsof datafor othermagnituderanges)arefor levelgroundconditionsandrelativelyshallowsoilswithrelativelysmallinitialeffective

    10

    CyclicPorePressure 363

    overburdenstresses.At highereffectiveoverburdenstresses,a givenCSRandnumberof loadingcycleswillbemoredamaging.This isbecausewhilesoils generallydevelophigher cyclic load resistancewith. increasingconfinement,the normalizedresistanceas expressedin termsof cyclicstressratiousuallydecreaseswith increasingconfinement.Accordingly,valuesof CSRfromFigure3 canbeusedforin situconditionswherea'o::;1 tonjft2 (1 kgjcm2),but mustbe correctedfor conditionswith initialeffectiveoverburdenstressesgreaterthan1tonjft2as

    CSR(a'o=a~)=CSR(a=1tsf). Ku

    A recommendedrelationshipbetweenKu anda'ois presentedin Figure6,basedondatasummarizedbyHarder[28].

    Finally,all of theabovehasbeenbasedon "levelgroundconditions",orconditionsin whichthereis no static"driving"shearstressactingon ahorizontal plane in the soil. Generationof pore pressuresandaccumulationof shearstrainsundercyclicloadingcan be significantlyaffectedbythepresenceof astatic(non-cyclic)drivingshearstress,andthistoomustbeaccountedfor in analyslsof liquefactionresistancewithindamsandembankments.

    1.2 o

    1.0

    0.4

    0.2

    00 1.0 2.0 3.0 4.0 5.0 6.0EFFECTIVE CONFININGPRESSURE (1st)or (ksc)

    7.0 8.0

    Fig.6: RelationshipBetweenEffectiveVerticalStress(a'o)andKa

    ,Jj

    MagnitudeorDurationEarthquake No.of representative CorrectionFactor:Magnitude,M cyclesat0.65r cyclic,max CM 0.8

    81/2 26 .---. 0.8971/2 15

    " 1.0" 0.663/4 10 1.13.6 5-6 .3251/4 2-3 1.5

    - ..- -.- -.- ..- -. -.- ,,- -'-..,-

    .0"o ." o "'-" " ,.'

    oo .: oI ',' '- ) o. f.,'"ON'OA" -... LAKE ARROWHEAOOA" El -..

  • 364 SeedandHarder

    EarlyrelationshipssuggestedbyDr. Seedtoaccountforthissuggestedthatthe presenceof a staticdrivingshearstresson a horizontalplanewasstronglybeneficial,andthatit significantlyincreasedthesoil'sresistancetoliquefaction.This remainstruefor relativelydensesoils,or soilswhichwouldtendto dilateunderuni-directionalshearing.More recentstudies,however,haveshownthatfor veryloosesoils(soilswhicharelessdilatantor morecontractiveunderuni-directionalshearing),thepresenceof astaticdrivin~shearstresscanactuallydecreasetheresistanceof thesoil to theinitiatlOnof liquefaction.Good data regardingthis remainssomewhatlimited,andfurtherinvestigationsof thiswouldbeofsignificantvalue.TheoriginalmethodproposedbySeed[29]toaccountfor theeffectsof drivingstaticshearstressesemployedthefollowingequation

    CSR(a=a) =CSRl(a=O). Ka(wherea is definedastheratioofstaticdrivingshearstressonahorizontalplaneto the initial effectiveoverburdenstressasa =r hv/a'o)Basedondataavailableat this time(datasummarizedby Harder [28],as well asresearchin progress),it is suggestedthatther~lationshipsbetweena and~ proposedby Seed[29]be replacedby the relationshipspresente~inFigure7. Thesearebasedondatafor conditionswherea' ::;3 tons/ft (3kgjcm2),andareappropriateonlyfor theseconditions.At higherinitialeffectiveoverburdenstresse~,soilswill belessdilatantor morecontractive,andKavalueswilldecrease.'./ .

    2.0

    1.5

    CTo'

    I . rv

    -~......Koe

    ""'.,.oe ;I

    ~I, 0'0

    0.5

    Dr~ 35%

    I(Yo ~3tsf

    oO 0.1 0.2 0.3

    o(0.4 0.5

    Fig.7: RelationshipBetweena andKa

    "'-

    CyclicPorePressure 365

    1140LOWER SAN FERNANDO DAM

    DDwNSTREAM S"ELL ZONESSPT EXPLDRmONS IN MUO-FlLLED

    ROTAR' BOREHOLES.

    {8 o-, . F-, 8G-I

    .." ..,., . F., ~ G"...,-,

    {Osoo, o' '."" 0"' 7'''' oO

    1120

    1100 OO

    ~t...~

    1080

    ELEVATION(feen 7

    MEAN~ 19d:

    1060

    MEAN~ 14.5

    1040

    7

    1020

    MEAN '"14.57 8

    1000~. 77

    . .OENO'"""

  • 366 SeedandHarder

    As shownin Figure8,theresulting(Nl)6Qvaluesfrom1971and1985showsimilartrends,andthehydraulicfill canbesubdividedintofourelevationzones of apparentlysimilar properties. Table 3 summarizesthecorresponding(N1)60valueswithineachof thesezones.As shownin thistable,two elevationzonesappearto havesignificantlylowerpenetrationresistancesthantheotherzones. It shouldbe notedthat,although"bestestimates"or meanvaluesof (N1)60areappropriatefor theseanalysesof acasestudyinvolvingcomparisonwith observedfield performance,moreconservative(lower thanmean)valueswouldnormallybe selectedforanalysesperformedin orderto predictthe likelyfuturebehaviorof anygivendamor embankment.

    The (N1)6Ovaluessummarizedin thesecondcolurnnof Table3 arebasedonpost-earthquakeblowcounts,andsomecorrectionis requiredtoaccountfor densificationlikelyto haveoccurreddueto theearthquakeshaking.Seedet al. [6]haveestimatedthatthepre-earthquakeblowcountsin thedownstreamhydraulicfill wouldhavebeenlower,ontheaverage,byabout2blows/ft,andthisis reflectedin thethirdcolurnnofTable3whichwill betakenasthevaluesfor thedownstreamzonesin theseanalyses.

    There is some minar questionas to how thesevalues should beextrapolatedto theupstreamportionof thehydraulicfill withinwhichtheslide occurred. Sirnilarityand uniforrnityof constructionproceduressupporttheuseof thesame'(N1)60valuesfortheupstreamsection.On theotherhand,it hasbeensuggestedthattheupstreamsectionmayhavebeensomewhatless denselyconsolidateddue to reservoir bouyancyand theabsenceof thedownstreamberms. If reservoir fillinghadbeeninitiatedsufficientlyquicklyaftercompletionof constructionthatfull consolidationof the embankmentunderits ownweightwasnot completed,thenthemaximumdifferencein densitybetweenthe upstreamand downstreamhydraulicfill sectionswouldhavebeensuchthattheupstream(N)60valueswouldneedto be reducedby about1 blow/ft. As shownin Table 3, areductionof 1 blow/ft will be usedfor theanalysesdescribedherein. Itshouldbenotedthatthisrelativelyminordecreasein theupstreamsection's(N1)6Ovaluesdoesnot explainthesignificantdifferencesin performance

    Table3: Representative(N1)ti

  • 368 SeedandHarder

    'f;]

    o FSL > 1.4!ZJ FSL '" 1.1to 1.3. FSL < 1.1

    CLAYEYCOREZONE

    Fig. 9: Results of Analy'sesof Resistanceto lnitiation or "Triggering"ofLiquefaction Fmlure -Lower SanFernando Dam

    Comparing the equivalent,uniform, earthquake-inducedcyclic stressratio(CSR q) with the uniform cyclic stress ratio necessaryto fully triggerliquera.ction,thefactorof safetyagainst"triggering"canbeca1culatedas

    FSl =CSRt.field = 0.288 = 1.27CSReq 0.227-r --r /

    Thisfactorof safetycanbeinterpretedin a numberof ways. Figure 10showsaplotof residualexcessporepressureratio(ru = !:J.uja'o)basedonlaboratorytestdatafor levelgroundconditions(a =O)assurnmarizedbyTokimatsuandYoshimi,[30]forsandysoils,andEvans[31]andHynes[32]forgravellysoils,assurnmar~zedbyMarcusonandHynes[33].For non-level groundconditions(a *" O), the effectiveverticalstressneed notnecessarilybe fullyeliminatedbyporepressureincreasesto initiate"large"deformationsin thepresenceof combinedcyclicand"driving"staticshearstresses. Accordingly,whena *"O,theresidualexcessporepressureratioisbestdefinedastheratioof cyclically-generatedporepressures(!:J.u) to thevalueof!:J.unecessarytoinitiatelargedeformations(!:J.Ul~) as

    r _!:J.uup*"O- !:J.ulim

    Lessdatais availableregardingtherelationshipsbetweenru.andFS fornon-Ievelgroundconditions,anddifferencesof opinionexistastohowthisdatacanbestbeinterpreted.Muchof thisisduetouncertaintyregardingthenatureofthetransitionfroma conditionwhereincyclicJ?orepressuregeneration(or cyclicstrainaccumulation)behaviorexertspnmarycontrolonperformance,to "post-triggering"conditionswhereinundrainedresidual(or IIsteadystate")strengthscontrolbehavior.

    Nonetheless,at this time it appearsthat a reliable analysiscan beperformedbyconsideringthat:

    1.,

    Soil elementswith low factorsof safety (FSe ~ 1.1) would achieveconditions wherein soil liquefactionfailureshouldbe consideredto

    CyclicParePressure 369

    1.0

    wa;::>

    :z 0.6wa;Q.Wa;oQ.

    ~ 0.4w(j)(w....e 0.2UJwa;

    ,""'

    LEGEND

    777/ GRAVEL (EVANS 1987.:L.::L.: ANO HYNES 1988)

    "a;O 0.8 ,.i=