EXPERIMENTALsuspended

13thWorldConferenceonEarthquakeEngineeringVancouver,B.
C.
,CanadaAugust1-6,2004PaperNo.
1053SEISMICQUALIFICATIONANDFRAGILITYTESTINGOFSUSPENDEDCEILINGSYSTEMSHiramBADILLO-ALMARAZ1,AndrewS.
WHITTAKER2andAndreiM.
REINHORN3SUMMARYThefailureofsuspendedceilingsystems(SCS)hasbeenoneofthemostwidelyreportedtypesofnonstructuraldamageinpastearthquakes.
FragilitymethodswereusedinthisstudytocharacterizethevulnerabilityofSCS.
SinceSCSarenotamenabletotraditionalstructuralanalysis,full-scaleexperimentaltestingonanearthquakesimulatorwasperformedtoobtainfragilitydata.
Severalceiling-systemconfigurationswerestudied.
Theresultsfromthefull-scaletestingarepresentedinformofseismicfragilitycurves.
Fourlimitstatesofresponsethatcovermostoftheperformancelevelsdescribedinthecodesandguidelinesfortheseismicperformanceofnonstructuralcomponentsweredefinedusingphysicaldefinitionsofdamage.
DatawasobtainedforeverylimitstatetocomparetheeffectofeachconfigurationontheresponseoftheSCS.
Basedontheresultsoftheexperimentaltestingitwasfoundthat(a)theuseofretainerclipsgenerallyimprovedtheperformanceofSCS,(b)undersized(poorlyfitting)tilesaresubstantiallymorevulnerablethanproperlyfittedtiles,(c)includingrecycledcross-teesintheassemblageofthesuspendedgridincreasedthevulnerabilityoftheSCS,and(d)includingcompressionpostsimprovestheseismicperformanceinSCS.
INTRODUCTIONTheresponseofnonstructuralcomponentscansignificantlyaffectthefunctionalityofabuildingafteranearthquake,evenwhenthestructuralcomponentsareundamaged.
Poorperformanceofnonstructuralcomponentsinpastearthquakeshasledtotheevacuationofbuildings,substantialeconomiclossesduetobusinessinterruptionandinextremecasestothelossoflife.
ThefailureofSCShasbeenoneofthemostwidelyreportedtypesofnonstructuraldamageinpastearthquakes.
ReconnaissancehasshownthatfailuresofSCSduringearthquakeshavecausedsignificanteconomiclossesanddisruptioninimportantorcriticalfacilities.
1GraduateStudent,DepartmentofCivil,Structural,andEnvironmentalEngineering,StateUniversityofNewYorkatBuffalo,212KetterHall,Buffalo,NY14260.
Email:hb5@eng.
buffalo.
edu2AssociateProfessor,DepartmentofCivil,Structural,andEnvironmentalEngineering,StateUniversityofNewYorkatBuffalo,230KetterHall,Buffalo,NY14260.
Email:awhittak@eng.
buffalo.
edu3CliffordC.
FurnasProfessor,DepartmentofCivil,Structural,andEnvironmentalEngineering,StateUniversityofNewYorkatBuffalo,231KetterHall,Buffalo,NY14260.
Email:reinhorn@buffalo.
eduEarthquake-historytestinghasbeenusedrecentlyforqualificationandfragilitytestingofstructuralandnonstructuralcomponents.
Seismicqualificationisintendedtodemonstratethroughexperimentationthatacomponentinastructureisabletofunctionduringandafteranearthquake.
Incontrasttoqualificationtesting,theobjectiveoffragilitytestingistoestablisharelationshipbetweenlimitstatesofresponseandarepresentativeexcitationparameterforacomponent.
Thedevelopmentoffragilitycurvesgenerallyinvolvestheuseofbothmathematicalmodelingandphysicalobservations.
InthecaseofSCS,mathematicalanalysisisdifficulttoaccomplishduetouncertaintiesinthephysicalbehaviorofelementsandcomponentsofthesystemoncethattheyareinstalledintheceilingsystem.
Further,thecomplexityofthemathematicalmodelandthehighlynonlinearbehaviorofthecomponentsoncetilesaredislodgedmakerobuststructuralanalysisofSCSvirtuallyimpossible.
SinceanalyticalmethodsaregenerallynotapplicabletothestudyofSCSanddatacollectedfollowingpastearthquakesarenotsuitableforfragilitycharacterization,experimentalmethodsrepresentthebestandmostreliabletechniquetoobtainfragilitycurvesforSCS.
ThemaingoalofthisstudywastodevelopfragilitycurvesofSCSsubjectedtoearthquakeshaking.
FragilitycurveswereobtainedbyexperimentaltestingofSCSonanearthquakesimulator.
Thespecificobjectivesoftheresearchprogramwere:(1)tostudytheperformanceofaSCSthatiscommonlyinstalledintheUnitedStates;(2)toevaluateimprovementsinresponseofferedbytheuseofretainerclipsthatsecuretheceilingpanels(tiles)toasuspensionsystem;(3)toinvestigatetheeffectivenessofincludingaverticalstrut(orcompressionpost)asseismicreinforcementinceilingsystems;and(4)toevaluatetheeffectofdifferentboundaryconditionsontheresponseofaSCS.
SEISMICFRAGILITYANDPREVIOUSSTUDIESONSUSPENDEDCEILINGSYSTEMSSeismicfragilityhasbeendefinedastheconditionalprobabilityoffailureofasystemforagivenintensityofagroundmotion.
Inperformancebasedseismicdesign,failureissaidtohaveoccurredwhenthestructurefailstosatisfytherequirementsofaprescribedperformancelevel.
Iftheintensityofthegroundmotionisexpressedasasinglevariable(e.
g.
,thepeakgroundaccelerationorthemappedmaximumearthquakespectralaccelerationatshortperiods,etc.
),theconditionalprobabilityoffailureexpressedasafunctionofthegroundmotionintensityiscalledaseismicfragilitycurve(SasaniandDerKiureghian[1]).
Fragilitycurvescanbegeneratedviatestingornumericalanalysis.
AlthoughseveralstudieshaveindicatedthatsomeimprovementintheseismiccapacityofSCShasbeenachievedinrecentyears(RihalandGrannneman[2],ANCO[3],andYao[4]),thereexistsnorobustfragilitydataforSCSandnoprovenstrategiestoincreasetheseismicstrengthofSCS.
From2001through2003,ArmstrongWorldIndustriesInc.
undertookanextensiveseriesofearthquakequalificationtestsonSCSattheUniversityatBuffalo(Badillo[5]andBadilloetal.
[6]).
Thefragilitystudiesdescribedbelowbuildonthesequalificationstudies.
EXPERIMENTALFACILITIESFORSEISMICTESTINGANDTESTSPECIMENSEarthquakeSimulatorandTestFrameTheearthquakesimulatorintheStructuralEngineeringandEarthquakeSimulationLaboratory(SEESL)oftheStateUniversityofNewYorkatBuffalowasusedtoevaluateandqualifytheceilingsystems.
Theperformanceenvelopeofthetableis±152mm(6in.
)displacement,±762mm/sec(30in.
/sec)velocityand1.
15gaccelerationatapayloadof197kN(44kips)inthehorizontaldirection,and±76mm(3in.
)displacement,±508mm/sec(20in.
/sec)velocity,and2.
30gaccelerationintheverticaldirection.
A4.
88x4.
88m(16x16ft)squareframeofASTMGrade50steelwasconstructedtotesttheceilingsystems.
Theframewasattachedtothesimulatorplatformusing25mm(1in.
)diameterboltsinthebeamsthatwereorientedintheEast-Westdirection.
Two10.
2x10.
2cm(4x4in.
)tubularsectionsconnectedateachcornerservedasmaincolumnsoftheframe.
A3.
8x3.
8cm(1-1/2x1-1/2in.
)anglewasweldedaroundtheperimeterofthetestframe.
A5.
1x15.
2cm(2x6in.
)timberledgerwasattachedtotheangle.
Theperimetertimberledgerservedasastudwallandanchoredtheceilingsystem.
ForadetaileddescriptionofthefeaturesofthetestframerefertoBadillo[5]andBadilloetal.
[6].
Figure1isaphotographofthetestframemountedontheearthquakesimulatorattheUniversityatBuffalo.
Figure1.
TestframemountedonthesimulatorattheUniversityatBuffaloSpecimenDescriptionEachceilingsystemconsistedoftwokeycomponents:asuspensionsystemandtiles.
Insomeconfigurationsretentionclipswereaddedtotheceilingsystems.
Allcomponentsusedinthisstudywereoff-the-shelfitemsusedincommercialceilingconstruction.
Accelerometersanddisplacementtransducerswereusedtomonitortheresponseofthesimulatorplatform,thetestframeandtheceilingsupportgrid.
SuspensionGridTheceilingsystemswereinstalledinagridthatwashungwithsuspensionwiresfromthetopofthetestframe.
Thegridwasconstructedwitha23.
8mm(15/16in.
)exposedteesystem.
A5.
1-cm(2-in.
)wallmoldingwasattachedtotheperimetertimberledger.
ThemainrunnersandcrossrunnerswereattachedtothewallmoldingwithrivetsontheSouthandWestsidesoftheframe,whiletherunnersontheNorthandEastsidesfloatedfree.
ThemainrunnerswereinstalledintheNorth-Southdirectionatspacingof1.
22m(48in.
)oncenter.
The1.
22m(4ft)crossrunnerswereinstalledintheEast-Westdirectionatspacingof61cm(24in.
)oncenter,whereasthe61cm(2ft)crossrunnerswereinstalledintheNorth-Southdirectionsataspacingof1.
22m(48in.
)oncenter.
Acompressionpostwasplaced1.
52m(5ft)fromtheSouthandtheEastsidesoftheframe.
TilesSincetheactualsizesofceilingtilesmaydifferfromthenominalsizedependingonqualitycontrolusedinthemanufacturingprocess,twotypesoftileswereusedforfragilitytestinginthisstudy.
Basedonpersonalcommunicationswithpracticingengineersandmanufacturers,ceilingtileswereconsideredtobeofnormalsizeiftheirplandimensionsarenotsmallerthanthenominaldimensionsbymorethan6.
4mm(1/4in.
)andundersizedotherwise.
OneofthetilestestedwasaFineFissuredHumigardPlustile.
Thistilewassmallerthanthenominalsizebyatleast12.
7mm(1/2in.
)andwasthereforeconsideredtobeanundersizedtile.
TheothertileusedinthisstudywastheHumigardPlustile.
Thistilewasanormalsizedtile.
Table1presentssummaryinformationoneachofthetwotilesusedinthisstudy.
Atotalof49tileswereinstalledintheinnersevenrows(seventilesineachrow).
Cuttileswereusedintheperimeterrowsoftheceilingsystem.
Figure2isaphotographoftheHumigardPlustile.
TABLE1.
SummaryinformationonthetilesusedinthisstudyPaneldimensions[B,D,T]*TileNameDescriptionNominalSize(cm)ActualSize(cm)Weight(kg/tile)FineFissuredHumigardPlusMineralfibertile61x61x1.
659.
7x59.
7x1.
61.
3HumigardPlusMineralfibertile61x61x1.
660.
3x60.
3x1.
61.
7*B,DandT:breadth,depthandthickness,respectivelyFigure2.
HumigardPlusceilingtileClipsClipssimilartothoseshowninFigure3wereinstalledtoinvestigatepossibleimprovementsintheseismicperformanceofSCS.
Theseclipscanbeattachedtomainbeamsorcrossteesbehindlay-inceilingtilesandhelptopreventthepanesfromdislodging.
Inthisstudy,theclipswereinstalledonthe1.
22m(4ft)longcrossteesofthegrid.
Figure3.
RetentionclipsDYNAMICCHARACTERISTICSOFTHETESTFRAMEThetestframewasdesignedtorepresentinanapproximatesensethehorizontalandverticalstiffnessofastoryinabuildingstructure.
Thedynamiccharacteristicsofthetestframewereevaluatedalongthetwoprogrammableaxesoftheearthquake-simulatorplatform,namely,theNorth-Southandverticaldirections.
Threemethodswereusedtoidentifythedynamicpropertiesofthetestframe:freevibration,bymeansofasnap-backtest,andtwoforcedvibrationtests,bymeansofresonancesearchandwhitenoisetests.
DetailsareprovidedinBadillo[5]andBadilloetal.
[6].
Thehorizontalandverticalfrequenciesoftheframewere12and10Hz,respectively.
Thedampingratiosinthefundamentalhorizontalandverticalmodeswereapproximately3%and0.
5%,respectively.
SEISMICQUALIFICATIONANDFRAGILITYTESTINGPROTOCOLTestingProtocolThetestingprotocolforfragilitytestingconsistedofsetsofhorizontalandverticaldynamicexcitations.
Eachsetincludedunidirectionalandbi-directionalresonancesearchtestsusingwhitenoiseexcitationalongeachprogrammableorthogonalaxisofthesimulationplatform(North-Southandvertical).
Eachsetofexcitationsalsoincludedaseriesofunidirectionalandbi-directionalspectrum-compatibleearthquakemotionsthatwereestablishedfordifferentmultiplesofICBO-AC156RequiredResponseSpectrum(Badillo[5],Badilloetal.
[6]andICBO[7]).
Theparameterselectedtocharacterizethegroundmotionforinputtothesimulatorwasthemappedspectralaccelerationatshortperiods,SS(ICC[8]).
ThetargetlevelsforearthquakesimulationrangedfromaSS=0.
25gthroughSS=2.
5g.
InformationonthegenerationoftheearthquakehistoriesfortestingispresentedinBadillo[5]andBadilloetal.
[6].
Figure4presentsthehorizontalandverticalRRSandtheircorrespondingresponsespectracalculatedfromrecordsgeneratedforalevelofshakingcorrespondingtoSS=1.
0g.
0.
00.
20.
40.
60.
81.
01.
21.
40.
1110100Frequency(Hz)Acceleration(g)TargetHorizontalCalculatedHorizontalTargetVerticalCalculatedVerticalFigure4.
Horizontalandverticalresponsespectra(targetandcalculated)foralevelofshakingcorrespondingtoSS=1.
0g.
RESULTSOFSIMULATORTESTINGFourvariablesthataffecttheseismicperformanceofSCSwereinvestigatedinthisstudy:(a)thesizeandweightoftiles,(b)theuseofretainerclips,(c)theuseofcompressionposts,and(d)thephysicalconditionofgridcomponents.
Informationonvariables(a)and(b)arepresentedinthispaper;resultsforvariables(c)and(d)canbefoundinBadillo[5]andBadilloetal.
[6].
Atotalofsixset-upswereconfiguredusingdifferentcombinationsofthesevariables:(1)undersizedtiles(seriesA-D),(2)undersizedtileswithretainerclips(seriesE-G),(3)normalsizedtiles(seriesL-O,Q,RandBB),(4)normalsizedtileswithretainerclips(seriesPandS-U),(5)normalsizedtileswithoutthecompressionpost(series:V-ZandAA)and(6)undersizedtileswithrecycledgridcomponents(seriesH-J).
Summarytestinformationonconfigurations(1)through(4)ispresentedbelow.
Configuration1:UndersizedTilesTheundersizedtilesfailedtypicallybyfirstpoppingupoutofthesuspensiongridandthenfallingthroughthegridtothesimulatorplatformbelow.
Figure5showsatileaninstantbeforeitfelltotheearthquakesimulatorbelow.
Figure5.
Tilerotatingbeforefalling,configuration1Configuration2:UndersizedTileswithRetainerClipsTheretainerclipssubstantiallyimprovedthebehavioroftheSCSintermsoflossoftilesbycomparisonwiththesystemsofconfiguration1.
Byretainingthetiles,theclipsincreasedtheinertialloadsonthegrid,resultingingriddamageatlowerlevelsofshaking.
Figure6showsabuckled1.
22m(4-ft)crossteefollowingsevereearthquakeshaking.
Anothertypeofcommonlyobserveddamagetothegridcomponentswasfailureandfractureofthelatchesofthecrosstees.
Inthesystemsofconfiguration2,tileswerelostprimarilyduetofailureofgridcomponents.
Figure6.
Bucklingin4-ftcrosstees,configuration2Configuration3:NormalSizedTilesThenumberoftilesthatfellduringthesimulatortestsofceilingsystemswithundersizedorpoorlyfittingtileswassubstantiallylargerbycomparisonwiththesystemsequippedwithnormalsized(snug)tiles.
However,ceilingsystemperformanceintermsofdamagetogridcomponentswasbetterinthesystemswithundersizedtilesbecausetheweightofthenormalsizedtileswaslarger(1.
7kg/tile)thantheundersizedtiles(1.
3kg/tile),andthenumberoftilesthatstayedinplaceduringshakingwaslargerforthesystemsofconfiguration3,andthereforetheinertialloadsonthesuspensiongridwerelargerforconfiguration3thaninconfiguration1.
Thebucklinginthewebofthe1.
22m(4-ft)crossteeswassimilartothedamagethatthegridcomponentsexperiencedinconfiguration2duringhigherlevelsofshaking.
Thetilefailurepatterninconfiguration3wassimilartothatofconfiguration1.
Configuration4:NormalSizedTileswithRetainerClipsTheretainerclipssubstantiallyimprovedthebehavioroftheSCSintermsoflossoftilesbycomparisonwiththesystemsofconfiguration3,whereclipswerenotincluded.
Theuseoftheretainerclipsshiftedthedamagefromthetilestothesuspensiongrid.
Thetypeofdamagethatwasobservedinthe1.
22m(4-ft)crossteesofconfiguration2wasalsoobservedinthesystemsofconfiguration4.
Inbothsystems,thelossoftileswasprimarilyduetothefailureofgridcomponents.
FRAGILITYANALYSISANDDATAEVALUATIONOneofthepurposesoffragilityanalysisistoidentifytheseismicvulnerabilityofsystems(orcomponentsofasystem)associatedwithvariousstatesofdamage.
Afragilitycurvedescribestheprobabilityofreachingorexceedingadamage(orlimit)stateataspecifiedlevelofexcitation.
Thus,afragilitycurveforaparticularlimitstateisobtainedbycomputingtheconditionalprobabilitiesofreachingorexceedingthatlimitstateatvariouslevelsofexcitation.
Aplotofthecomputedconditionalprobabilitiesversusthegroundmotionparameterdescribesthefragilitycurveforthatdamagestate(SinghalandKiremidjian[9]).
Theconditionalprobabilityofreachingorexceedingadamagestateis:]|[kyYidDPikP=≥=(4)wherePikistheprobabilityofreachingorexceedingadamagestatedigiventhattheexcitationisyk;DisadamagerandomvariabledefinedondamagestatevectorD={d0,d1,….
,dn};andYisanexcitationrandomvariable.
LimitStatesFourlimitstatesweredefinedinthisstudytocharacterizetheseismicresponseofSCS.
Limitstates1through3accountforthenumber(orpercentage)oftilesthatfellfromthesuspensiongrid.
Thefourthlimitstateisassociatedwithstructuraldamagetothesuspensiongrid.
Thefourlimitsstateswere:(1)minordamage(lossof1%ofthetilesfromthegrid),(2)moderatedamage(lossof10%ofthetilesfromthegrid),(3)majordamage(lossof33%ofthetilesfromthegrid),and(4)gridfailure.
DetaileddescriptionsoftheselimitstatesareprovidedinBadillo[5]andBadilloetal.
[6].
GroundMotionIntensityParametersSeveralintensityparametershavebeenusedinpreviousstudiestocreatefragilitycurves,namelypeakgroundacceleration,peakgroundvelocity,spectralaccelerationatspecificperiods,andspectralaccelerationoverafrequencyrangethatwouldbracketthein-servicedynamicpropertiesofaspecificsystem.
Thereisnouniformlyacceptedintensitymeasureforauseintheconstructionoffragilitycurves.
Inthisstudy,twoexcitationparameterswereusedtoconstructthefragilitycurvespresentedbelowandinBadillo[5]andBadilloetal.
[6]:(1)peakgroundacceleration,and(2)averagehorizontalspectralaccelerationsatselectedperiods.
Theselectedperiodsrepresentabroadrangethatshouldincludemostin-serviceconditionsforSCSinbuildings:0.
2,0.
5,1.
0,1.
5and2.
0seconds.
Thespectralaccelerationordinateswereobtainedbycalculatingthemeanspectralaccelerationforeachceilingsystemconfigurationtestedateachlevelofearthquakeshaking.
EvaluationofFragilityDataThefourlimitstatesusedtocharacterizetheseismicperformanceofSCSwereselectedwiththeintentofcoveringmostoftheperformancelevelsdescribedincurrentseismiccodesandguidelinesforseismicperformanceofnonstructuralcomponents.
TheproceduretodevelopthefragilitycurvesforeachconfigurationisillustratedinFigure7.
Thedatapresentedintheillustrationoftheprocedureisfromthe6systemsthatwerepartofconfiguration3:SystemsL,M,N,O,RandBB.
Theprocedurewasasfollows:(1)obtainthemeanspectralaccelerationresponseforeachshakinglevelwiththeaccelerometermountedonthesimulatorplatform(seetheheavysolidlineinFigure7),(2)computethespectralaccelerationsatselectedperiods(0.
2,0.
5,1.
0,1.
5and2.
0seconds)fromthemeanspectralaccelerations(seethearrowsinFigure7forthe1-secondcalculation,S1.
0=2.
36g),(3)countthenumberoftilesthatfellfromthegridforeachsystem(6systemsinthisexample)ateachshakinglevelasapercentageofthetotalnumberoftilesintheceilingsystem,(4)comparethepercenttilefailurewitheachlimitstateforeachsystem,and(5)calculatetheprobabilityofreachingorexceedingthelimitstateas:NNPff=(5)whereNfisthenumberofsystems(trials)wherethelimitstatewasreachedorexceededandNisthetotalnumberofsystemsinconfiguration.
AsNapproachesinfinity,Pfapproachesthetrueprobabilityofreachingorexceedingalimitstate.
ThefragilitycurveswereobtainedbyplottingPfforeachshakinglevelversusthecorrespondingmeanspectralacceleration.
Theprocesswasrepeatedforeachofthesixconfigurationstestedinthisstudy.
0.
01.
02.
03.
04.
05.
06.
00.
010.
1110Period(seconds)Acceleration(g)SeriesLSeriesMSeriesNSeriesOSeriesRSeriesBBMean2.
36gFigure7.
Proceduretodevelopfragilitycurves,configuration3:normalsizedtilesFigure8apresentsthefragilitycurveforpeakgroundacceleration(0secondperiodspectralacceleration)andFigure8bpresentsthefragilitycurveforthespectralperiodof1.
5seconds,forconfiguration1foreachlimitstatedefinedearlier.
Similarfigureswereobtainedinthisstudyforeachofthespectralaccelerationperiodsselectedandforeachofthesixconfigurations.
0.
00.
10.
20.
30.
40.
50.
60.
70.
80.
91.
00.
00.
51.
01.
52.
02.
5PGA(g)ProbabilityofexceedanceMinor(1%fell)Moderate(10%fell)Major(33%fell)Gridfailurea)Fragilitycurvesforpeakgroundacceleration0.
00.
10.
20.
30.
40.
50.
60.
70.
80.
91.
00.
00.
10.
20.
30.
40.
50.
6S1.
5(g)ProbabilityofexceedanceMinor(1%fell)Moderate(10%fell)Major(33%fell)Gridfailureb)Fragilitycurvesforspectralaccelerationat1.
5secondsFigure8.
Fragilitycurvesforconfiguration1:undersizedtilesFigure9presentsthesameinformationpresentedinFigure8butforthefirstfourofthesixconfigurationstestedinthisstudyforthecaseofminordamage.
Similarfigureswereobtainedinthisstudyforeachofthespectralaccelerationperiodsselectedandforeachlimitstatesdefined.
Someofthefragilitycurveswereincompletebecausethemaximumacceleration,velocity,anddisplacementofthesimulatorarelimitedto1.
5g,94cm/sec(37in/sec)and14cm(5.
5in.
),respectively.
Differentscaleswereusedinplottingthefragilitycurvesbecausethemagnitudeofthespectralaccelerationchangedsubstantiallyasafunctionofperiod.
0.
00.
10.
20.
30.
40.
50.
60.
70.
80.
91.
00.
00.
51.
01.
52.
02.
53.
0PGA(g)ProbabilityofexceedanceNormalNormalw/clipsUndersizedUndersizedw/clipsa)Fragilitycurvesforpeakgroundacceleration0.
00.
10.
20.
30.
40.
50.
60.
70.
80.
91.
00.
00.
10.
20.
30.
40.
5S1.
5(g)ProbabilityofexceedanceNormalNormalw/clipsUndersizedUndersizedw/clipsb)Fragilitycurvesforspectralaccelerationat1.
5secondsFigure9.
Fragilitycurvesforlimitstate1:minordamageCONCLUSIONS1.
Themostcommonfailuremodeoftileswhenretentionclipswerenotusedwastilespoppingoutofthegrid.
Ifthetilesdidnotreturntotheoriginalpositiononthesuspensionsystem,itwasverylikelyforthetilestorotateandfalltothesimulatorplatformbelow.
2.
TheuseofretainerclipssubstantiallyimprovedthebehavioroftheSCSintermsoflossoftiles.
However,byretainingthetiles,theuseofclipsincreasedtheinertialloadsonthegrid,resultingingriddamageatlowerlevelsofshaking.
Thelossoftilesinsystemswithretentionclipswasdueprimarilytothefailureofgridcomponents.
3.
TheeffectofasmallvariationintilesizeontheperformanceoftheSCSwasconsiderableintermsoflossoftiles.
Evenwhentheweightofnormalsizedtileswaslargerthantheweightoftheundersizedtilesbya30%approximately,thenumberoftilesthatfellduringtheshakingtestsofceilingsystemswithundersizedtileswassubstantiallylargerbycomparisonwiththesystemsequippedwithnormalsizedtiles.
However,ceilingsystemperformanceintermsofdamagetogridcomponentswasbetterinthesystemswithundersizedtilesbecausetheinertialloadsonthesuspensiongridweresmallerthaninthosesystemswithnormalsizedtiles.
4.
TherivetsthatattachedthemainrunnersandcrossteestothewallmoldingplayedaveryimportantroleintheseismicperformanceoftheSCS.
Damageintheceilingsystemsintermsoflossoftileswasmuchlargerwhenarivetfailedthanwhenalloftherivetswereundamagedandthecrossteesremainedfirmlyattachedtothewallmolding.
5.
PresentinginformationintheformoffragilitycurvesappearstobeaconvenientwaytorepresenttheseismicbehaviorofSCS.
FragilitycurveshelptoidentifyregionsofundesirableandunsafeperformanceofSCS,suchasthecasewhentwofragilitycurvesintersect.
Forexample,theregionbeyondtheintersectionoffragilitycurvesforlimitstate3(majortilefailure)andlimitstate4(gridfailure)shouldbeavoidedbecausefailureoflargesectionsoftilesandgridcouldcausealife-safetyhazard.
ACKNOWLEDGMENTSArmstrongWorldIndustriesInc.
providedalloftheceilingsystemcomponentsforthefragilitytestingprogram.
Thissupportisgratefullyacknowledged.
SpecialthanksareduetoMessrsPaulHoughandThomasFritzofArmstrongWorldIndustriesandMessrsMarkPitman,ScotWeinreberandDwayneKowalskioftheDepartmentofCivil,StructuralandEnvironmentalEngineeringatUniversityatBuffalo.
ThefirstauthorwouldliketothanktotheNationalCouncilofScienceandTechnologyofMexico(CONACYT)fortheirfinancialsupportduringhisstayattheStateUniversityofNewYorkatBuffalo.
PartialsupportfortheworkdescribedinthispaperwasprovidedbytheMultidisciplinaryCenterforEarthquakeEngineeringResearchthroughgrantsfromtheEarthquakeEngineeringCentersProgramoftheNationalScienceFoundation(AwardNumberEEC-9701471)andtheStateofNewYork.
REFERENCES1.
Sasani,M.
andDerKiureghian,A.
"SeismicFragilityofRCStructuralWalls:DisplacementApproach",JournalofStructuralEngineering2001;127(2):219-228.
2.
Rihal,S.
,andGranneman,G.
"ExperimentalInvestigationoftheDynamicBehaviorofBuildingPartitionsandSuspendedCeilingsDuringEarthquakes",Rep.
No.
ARCER84-1,CaliforniaPolytechnicStateUniversity,Pomona,California,1984.
3.
ANCO.
"EarthquakeTestingofaSuspendedCeilingSystem",ANCOInc.
,CulverCity,Cal.
,1993.
4.
Yao,G.
C.
"SeismicPerformanceofDirectHungSuspendedCeilingSystems",JournalofArchitecturalEngineering2000;6(1):6-11.
5.
Badillo,H.
"SeismicFragilityTestingofSuspendedCeilingSystems",M.
S.
Thesis,SchoolofEngineering,StateUniversityofNewYorkatBuffalo,Buffalo,NewYork,2003.
6.
Badillo,H.
,Whittaker,A.
S.
,andReinhorn,A.
M.
"SeismicFragilityofSuspendedCeilingSystems",Papertobesubmittedforpossiblepublication,EarthquakeSpectra,March2004.
7.
ICBO"ICBOAC156AcceptanceCriteriafortheSeismicQualificationofNonstructuralComponents",ICBOEvaluationService,Whittier,California,2000.
90601-2298.
8.
ICC.
"InternationalBuildingCode",InternationalCodeCouncil,FallsChurch,Virginia,2000.
9.
Singhal,A.
andKiremidjian,A.
S.
"MethodforProbabilisticEvaluationofSeismicStructuralDamage",JournalofStructuralEngineering1996;122(12):1459-1467.
10.
Shinozuka,M.
,Feng,M.
Q.
,Lee,J.
,andNaganuma,T.
"StatisticalAnalysisofFragilityCurves",JournalofEngineeringMechanics2000;126(12):1224–1231.