Page 84 - konishipaper

andthetemperatureresolutiontobe:!:0.1.C[lto，Masuda，andSaito， 1991]. UsingtheHIsystem，weobservedalargeliquidconvection bothintheuniformandpulsatingspreadregions. Theliquidconvec- tionintheuniformspreadregionwasasurprise，becausetheheadof theliquidconvectionprecededtheflame'sleadingedgebyaproxi- mately1cm[Ito，Masuda，andSaito，191]，whichisentirelydiferent fromAkita'sresult[1972]. Tomakesureoftheac curacyofthedata， werepeatedtheexperimentsseveraltimesandfoundthereproducibil・ itytobewithin土 90%. Thus，weproposedanewmechanism:the flamespreadintheuniformspreadregionisgovernedbytheliquid convection[Ito，Masuda，andSaito，191]. However，atthattimewe didnotrealyunderstandwhyAkita'sshadowgraftresultsweresodif- ferentfromourHIresuIts.
Later，HowardRos'sgroupatNASALewisResearchCenterap- pliedRainbowSchlierenDeflectometry(RSD)tomeasureliquidcon- vectiongeneratedbythespreadingflame[MilerandRoss，192].They foundlitIeliquidconvectionaheadoftheflame'sleadingedgeinthe uniformspreadregion. TheirRSDdataweresimilartoAkita's shadowgraphsanddisagredwithourHIresults. Overthepastthre years，bothNASAandourgrouphavetriedtounderstandthereasons forthedisagrement，butfaHedtodosountilaseriesofparametric experimentswascompleted.Inthoseexperiments，theefectsofsix diferentparameterswereinvestigated:sensitivityofbothRSDandHI systems，impurityofpropanol(ifany)，therelativehumidityoftheair， theambientairtempera印 re，severaldiferenttypesofignitionmethod， andfinalydiferenttraywidths. Wefoundthatthefirstfiveparam- etershadlit1einf1uenceonthedisagrement. However，whenwe checkedtheefectoftraywidth(0.5，1，and2cm)ontheliquidconvec- tion，theleastsustectedparameter，asurprisingresultemerged:the0.5 cmwidetrayhadalargeliquidconvectionintheuniformspreadre- gion，Whilethetraysthatwere1cmand2cmwidehadathinlayerof liquid convection. 1t was so thin and smal l that only a very sensitive measurementtechniquecoulddetecti.t.Theresultforthe2-cmtray wasverysimilartotheNASA'sresultforthe2四 cmtray. A closerlo ok attheirRSDresultalsorevealedasmalandthinlayerofliquidcon- vectionthatwasalwaysaheadoftheflame'sleadingedge. Compari- sonofthesefind
TheCurrentProblem NASA'smicrogravitytestresultsshowedthatpulsatingspreaddid
notocurineithershalowordeppolexperiments[Ros，194];in- steadtheflamewasextinguished，indicatingtheimportanceofthebuoy- ancyefectinthegasphaseonthemechanismofpulsation. Schiler andSirignano，usingnumericalcalculations[192]，predictedtheex- istenceofaverysmalcirculationinthegasphaseaheadoftheflame's leadingedge. Theysugestedthatthegω-phasecirculation，whichis uniqueinthepulsatingspreadandapearswiththeliquid-phaseCIrcu-
lation，mayplayanimportantroleinflamepulsation[192，196]. ThereareLDVflowmeasurementdataobtainedwithLaser-Dopler Velocimetry(LDV)bySantoroeta.1[1978]showingagas-phasecir- culationofaproximatelyl-cmdiameterinthepulsatingspreadre- gion. However，theirLDVresultisratheraqualitativeindicationof theformationofcirculation.
Toconfirmtheprediction[SchilerandSirignano，192;and SirignanoandSchiler，196]andunderstandtheroleofthesmalgas- phasecirculationintheflamepulsation，transientvelocityprofilesin thegasphasejustabovetheliquidsurfaceandjustaheadoftheflame's leadingedgemustbemeasured. Thesemeasurementsarenoteasy， becausetherateofflamespreadvariesfromafewcentimeterstoafew tensofcentimeterspersecondandthetargetregion，wheredetailed velocityprofilesareneded，isanorderofseveralmilimetersindiam- eter.
PreviouslyweleamedthatLDVwasnotsuficientlyaClratefor ourmeasurements[Venkatesh，Ito，Saito，andWichman，196]. In- steadweusedPTLStomeasuref10wprofilesin2-Dwithaspatial resolutionontheorderofamilimeterintheprimaryanchoringregion ofasmalpolfire[Venkatesh，Ito，Saito，andWichman，196].Based onthatexperience，weapliedthesametechniqletomeasureatime seriesofdetailedvelocityprofilesinthepulsatingspreadregion. Asa result，weconfirmedtheformationofasmalcirculationaproximately inthesamelocationaspredictedbytheUCInumericalmodel[Schiler andSirignano，.192].Wealsoconfirmed，inagrementwiththemodel prediction，thattheformationofasmalcirculationapearedonlyin thepulsatingspreadregion.
Tofurtheradvanceourunderstandingoftheflamespreadover liquids，thedirectmeasurementofpropanolconcentrationovertheliq- uidsurfacejustaheadoftheflame'sleadingedgewasconductedusing D W H I. O u r D W H I d a t a a r e a c c u r a t e w h e n t h e f1 0 w o f t h e g a s p h a s e hasa2-Dprofile. Whenthef10wprofileshiftsfromthe2-Dto3-D， ourcurentDWHIsystemwi1l10seitsaclracy. Toovercomethis limitation，wemustworkonthedevelopmentofanadvanced3-D-DWP.J system，whichisbeyondthescopeofthispaper.
EXPERIMENTALMETHODS Flame-SDreadTest-ADoaratus
TheFlame-spreadaparatls(Fig.1)usedforthisstldyisesen- tialythesameasthatusedinapreviousflame目 spreadstudy[Ito， Maslda，andSaito，191;Ito，Saito，andCremers，195;andTashtoush， Narumi，Ito，Saito，andCremers，196]. APyrextrayof2cmwideX 2.5cmde ep X 25cmlongwasused. Thetwolongsidesofthefl lel trayweremadeofPyrexofO.2-cmthicknes. Theentireliquidtray wasenclosedinaglascel14cmhighX 15cmwideX 35cmlongin ordertorninimizelaboratorydraftandprovidearepeatableflame-spread condition.百1et怠mperatureofflelwasmeasuredbyplacingachromel- alumelthermocouplewith50μmwirediameterwhichwasplacedat thecenterofthetray2mmbelowthefl lelsl lrface. The臼 mperatureof ambientairwasmeasuredneartheglassidewalbythesametypeof thethermocolpleusedintheabove. Anejector-nozlewasaddedto thepreviolslydevelopedparticlefeder[Venkatesh，Ito，Saito，and Wichman，196]sothattalcpartic1escoulduniformlydisperseover theliqlidsurface.Theuniformityofp紅 tic1edistributionwaschecked vislaly.Aninfrared(IR)cameraandthermalimagingsystemthat weredevelopedduringourpreviouslpwardflamespreadexpeliments [Qian，Ishida，andSaito，194]wereapliedtomeasurethe2-Dtem- poralmapofliquidsurfacetemperatlre. Anormalvideocamerawas