The Theory of Flammability Limits - Conductive Convective Wall Losses

download The Theory of Flammability Limits - Conductive Convective Wall Losses

of 31

Transcript of The Theory of Flammability Limits - Conductive Convective Wall Losses

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    1/31

    469

    yLimits

    allLosses

    MENTOFTHEINTERIOR

    y

    gDirector

    Publ

    icDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    2/31

    atalogedasfollows:

    ylimits.Conductive-convective

    enching.

    BureauofMines; 8469)

    8469-

    .Title.II.Series:United States.Bu-

    vestigations;8469.

    628.9'22]79-607931

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    3/31

    ATURE

    aofa tube.

    ofa particle.

    reffectiveheatcapacitypergram.

    mof solidparticle.

    onofdustina dust-airmixture.

    ofacoaldustatwhichits maximumSoccurs.

    .

    diameterorquenchingdistance.

    caldustparticle.

    htedaverageparticlediameter.

    ration.

    aparticle.

    freepathfor molecularcollisions.

    berforwall-lossquenching.

    yexpandingflame.

    alparticle.

    aminarburning velocity.

    hevelocityofaflame frontrelativeto

    urningvelocityforquenchingbynatural

    urning velocityforquenchingbyflamestretch

    oyancy(processa).

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    4/31

    gnationlimitburningvelocity.

    ocityforconductive-convectivewall-loss

    ocityforconductive-convective,

    cingbyinertpowders.

    rningvelocityforacoaldust-air flamewhichoccurs

    edgases.

    urnedgases.

    egas withrespecttotheparticle.

    ualparticle.

    n.

    ss.

    effectivediffusivity.

    whichequalsthe numberofflame-zone

    mefrontfromwhichconductivelosses

    tlyinfluencethepropagationprocess

    nessforradialheat conductionlossesto

    reffectivethermalconductivity.

    ofthegas.

    reffectivedensityofreacting,combusti-

    omprisingthedustparticle.

    ngtimeforaparticlein theflamefront.

    ngtimeforthegas intheflamefront.

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    5/31

    rei

    ewall-losslimit:quenchingdiameters3

    efornearlimitflamepropagation6

    sdimensionsandboundaryconstraintsonthe

    ability8

    rtpowders14

    ts;particlelagintheflamefront;the

    ensitiesfor thewall-lossquenchingofa

    e3

    bustionandbuoyancyforcefields7

    zonestructureforhorizontalpropagationina tube8

    tsinmethane-airmixturesforthethree

    ationintubesofvaryingdiameter,

    velocitiesatthoselimit concentrations.9

    gvelocitiesfor(a)buoyant-convective

    ssquenching,asafunctionoftube

    ctionsofflamepropagation12

    oflimit velocities,(S)+(S ),with

    ningvelocity11

    ernalwallquenching"ofmethane-air

    16

    rnalwallquenching"ofcoaldust-air

    ading17

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    6/31PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    7/31

    BILITYLIMITS

    allLossesandThermalQuenching

    gvelocities,formulatedinanearlier Bureau

    appliedtotheproblemofflame propagation

    The limitburningvelocityforconductive-

    ching(processb)is (Su)b=~-,wheretheproper

    nstant,Pe,isdeterminedmainlybytheratioof

    ato flame-zonecrosssectionalarea.Thisratio

    narearelatestotheshapeof theflamefrontand

    dbyboundaryconditions.Thecomparisonof(Su)b

    mitvelocitiesforsystemsmixedbynaturalconvec-

    e influenceoftubedimensionsandboundary

    rearthlylimits forthethreedirectionsof

    sedbyinertpowdersisshownto besimilarly

    mitburningvelocity,(Su)b.Thetube'ssurface-

    simplyreplacedbythepowder'ssurfaceareaper

    blemixture.Thermallossquenchingbythese

    criticalPecletconstantsomewhathigherthan

    dtheproblemis complicatedbythefiniteheat

    andparticlelageffectsinthe flamefront.

    econceptoflimitburningvelocitieswas

    tativetheoryofflammabilitylimits.It wasshown

    ocessescandissipatepowerfromacombustionwave

    ationatsome characteristicallylowlimitvelocity.

    sandonecomplicationwereinvolved:(a)free,

    mist,PittsburghResearchCenter,BureauofMines,

    enthesesrepresentitemsinthelist ofreferences

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    8/31

    onductive-convectivewalllosses;(c)radiation;

    emixing(thecomplication);and(e)flowgradient

    wasshownthatforpremixedgases,the "normal,"

    agation,asconventionallymeasured,involves

    teractionofprocessesaande.

    tion,extinctioniscausedbytheascendanceof

    tingthroughthemechanismofflamestretch.An

    e,emanatingfromaconvectivelyrisingflamekernel,

    aswhosemotionis alsoinfluencedbybuoyancy.

    kernel,thecoldsurroundingsmustmoveoutward,

    athofburned-gaskernel.In theequatorialregions,

    ngsis downward,asrequiredbythenetbuoyancy

    pagationthusoccursintoavelocitygradientinthe

    but finite,propagationvelocitytheflameis

    ncy-inducedflows.Forhorizontalpropagation,

    simply derivableasthebalancebetweenthe

    ndbuoyancyforces,andis(Su )a=[2agpb/pu]1'3.

    twasshownthattheblowofflimit velocitywas

    formulaewereshowntoproperlypredictboth the

    ceofthelean limitforavarietyof fuelsandthe

    elimitwithincreasinggravitationalacceleration.

    veryexistenceof"normal"limitsofflammability

    siscausedbythe competitionbetweencombustion

    thatthiscompetitionresultsinthe presenceofa

    gvelocityatthelimit. Flamepropagationcan

    dealburningvelocityexceedsthelimitvelocity.

    burningvelocitiesarelessthanthelimit velocity

    yandthushavezerovaluesfor theirrealburning

    ntinuity,orminimum(limit)valueforthe

    wninasubsequentpublicationtohavea profound

    ureofdiffusionflames(11). Theconceptledto

    mestructurewhichis atvariancewiththetradi-

    chmoreconsistentwithrecentdata(15).

    onsiderthesecondcompetitiveprocess:

    walllosses.Anequationisderivedfor thelimit

    ctive-convectivewalllossquenching,(Su)b.

    ttothe conceptofacriticalPecletnumberfor

    htubes,whichconcepthasalreadybeenshowntobe

    valuesforquenchingdiameters.Themagnitudeof

    wsone toassessrealisticallytheinfluenceof

    true"limitmeasurements.Withfurtherdevelopment,

    abletothe problemofthethermalquenchingof

    particles.Suchparticlescanbeconsideredas

    withintheflammablemixture.

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    9/31

    ECTTVEWALL-LOSSLIMIT:QUENCHINGDIAMETERS

    ghtubesoffinitedimensionswilllosecombus-

    dingsby heatconductionthroughthetubewalls

    colderthan theburnedgas).Thisloss process

    turegradientsinthegasnearthe wallandaquenched

    edinfigure1.For large-diametertubes,the

    isfarremovedfromthecentralregionsand doesnot

    actualburningvelocityinthecentralregions.

    ersdiminish,theradialtemperaturegradientsinthe

    convergeinwardandsoonbegintoinfluencethebulk

    allypropagationisquenchedatsomefinitetube

    enchingdiameter.Thesenonadiabaticlossproc-

    vectorsperpendiculartothepropagationdirection.

    lossesnotonlyarein the"rim"regionswhere

    ewallbutalso comefromtheburnedgasregions

    iedbyvon ElbeandLewis(21),Spalding(20),

    erlad(17),andGersteinand Stine(6),andthe

    studieshavebeensummarizedinasurveyby

    equationmaybeobtainedbythe followingelemen-

    l,flat,laminar,flame-frontthatis propagating

    ustionpowerdensityinthepropagationdirectionis

    nsitiesfor thewall-lossquenchingofaflame

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    10/31

    ombustionpowerdensity3is,ineffect,the

    mtheburnedtothe unburnedgasthatmaintainsthe

    rateinacoordinateframemovingattheburning

    powerdensityisapproximatelyequaltotherate

    enthalpy(bythecombustionreactions)perunit

    wnin figure1,thispowerdensityis axialand

    opagationalongthetubeaxis.

    s,processb competeswiththepropagation

    onductive-convectiveheatflowdensitytothe

    (Tb-Tu)= (Tb-Tu)Nu.Thisheatflowis the

    halpyfromtheburnedgasper unitareaoftube

    agationofa planarflamefrontina circular

    gas thatisactivatedbythecombustionpower

    sectionalarea,irr0. Forthecompetingradial

    theareainvolvedisthe flame-zonecontact

    sanadditionalperimeterareain theburnedgases

    Thetotal areainvolvedfortheheatlossflux

    x,whereg isadimensionless,geometricwallloss

    sourcepoweristherefore

    heatlosspoweris

    ssbis obtainedwhentheidealcombustion

    asheatloss power.Equatingsourcepowerand

    = a/(Su).dealandot= ,gives

    gNu)1/2a/r0.(1)

    nchingdiameter,dq,atthelimit offlamepropa-

    creasingdiameterissimply2r0 ,andthelimit

    Su)b, onehas:

    gnizableasacriticalPecletnumber,Pe,and

    ionlessquantitythatcorrelatesthe measured

    nceptsandadefinitionof(Su)idealarepresentedin

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    11/31

    withthecombustionandtransportpropertiesofthe

    owever,itshouldberecognizedthatone isnothere

    berinthe conventionalsense.Theconventional

    f convectivetoconductiveheattransfertoa

    adofasinglesurface,thereare different

    ximatelyorthogonallyorientedrelativetoone

    flow"reflectedinSu [or(Su)ideal]relatesto

    mburnedtounburnedgasthatis necessarytomain-

    agationrateatSu.It istheflow"eigenvalue"

    pagation.Theconductivewalllossreflectedin

    irectionofflamepropagation.Asindicatedin

    of equation2,thePecletconstantistheratio

    heflame-zonesource(requiredtomaintainpropa-

    erateofenergytransferin theorthogonaldirec-

    boundarylayeratthe wall.Themechanismofenergy

    nanddiffusionforboth directions.AvelocitySu

    allynoconvectiveheatbeingtransportedtoany

    hatflow.

    ofequation2, theaxialcombustionsource

    ticflamepropagationwascomparedwiththeradial

    l.Inreality,asoneapproachesthelimit, flame

    gationbecomesnonadiabatic,andaxialheatlosses

    ttothe coolerunburnedgases.Thusinasystem

    quenching,theorthogonalitybetweenpropagation

    odissappear.Someof thesegeometriccomplexi-

    ortly.Nevertheless,theireffectisquantifiable

    3-factor.

    mind,equation2indicatesthat steady-state

    yifthe Pecletnumberexceedssomecriticalvalue;

    stionsourcepowerexceedstheradialpowerloss.

    thecriticalPecletnumberis 60forcircu-

    withtheearliercalculationsofvonElbeand

    easuredPevaluesof14 to18fordownwardpropaga-

    tes,theratio oflossareato flameareais

    ncethecriticalPecletconstantshouldbe0.71

    man'splatevalueswouldcorrespondtotubular Pe

    dberecognizedthatmeasuredvaluesare gener-

    shortchannelsoffinitel ength,whereasthe

    calculationsthatapplytotubesofinfinitelength.

    seffectivequenchersthantubesof infinite

    tofindthemeasuredquenchingdiameterssys-

    hetheoreticalones.Otherfactorsinvolvingthe

    flowconvergenceatthetubeinletshould also

    alueofthecriticalPecletnumberforwall-quenching.

    entialvalidityofequation2by varyingSuanda

    mescontainingvariousinertgases.Potterand

    widerrangeoffuel-oxidizercombinationsand

    uredependenceofdwas accuratelypredictedby

    oftheratioa/Su,as equation2requires.

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    12/31

    tion,theabsolutevalueofthecritical Peclet

    ratioofthecontactperimeterloss areatothe

    larea.Fortheassumedplanarflamefrontin a

    on,thatratiois2gAx/rQ= 23a/r0Su.Notethat

    urceareavariesinverselywith Su.Anexact

    afactor,g, isdifficulttomake.Its accurate

    etailedsolutiontotheflamepropagationequation

    ncertainfactoris thuslumpedintothePeclet

    ydeterminedempirically.

    ver,thattheaboveratiois necessarilyinflu-

    ezoneshape,and thattheoverallbalancemaybe

    genceandconvectioneffects.Flameshapeiscon-

    ,radialboundaryconstraints(tubeshape),axial

    sedversusopen-endedignition),andselectivedif-

    cellularflamestructures).Theseprocessesmust

    ecritical Pecletnumberatthewallquenching

    cetheprecisedetails oftheflowstructure

    frontandnearthewalls,and theymayaddaflame-

    m.

    HAPEFORNEARLIMITFLAMEPROPAGATION

    gationbehaviorofnear-limitmixturesclearly

    cesinboththeflameand theflowstructurefor

    agation.Levy'sstudies(13),for example,showed

    tureissphericalduringupwardpropagationin tubes

    dsto bemaintainedevenduringflamequenching;but

    ationintubes,thecurvaturediminishes,andatextinc-

    yflat.TherecentobservationsofSapko,Furno,

    uctureofnearlimitflamesin averylargescale

    atthesamebuoyancy-induceddistortionsoccurin

    aryconstraints.Inthecompositionregionbetween

    dlimits,the"normal"sphericalflameisreplaced by

    lshape.

    sphericalflamekernelandassignstoit a

    accelerationvector(8-ll_),theobserveddistortion

    dexpect.Thatis, thebuoyantaccelerationvector

    ntatthetop ofthesphere,andhencethe entire

    atestheupwardpropagationvelocity(neglectingthe

    emoment).Atotherpointsin theupperhemisphere,

    sdiminishedbythe cosineoftheanglebetweenthe

    ward-directed,buoyantaccelerationvector.Asimple

    esultantflame-frontdisplacementshowsthatthe

    maintainsphericalcurvatureintheupperhemi-

    mekernelexpandsand rises.

    n,ontheotherhand,thebuoyantacceleration

    to theoutwardnormalpropagationvectoratthe

    sdeceleratespropagationinthelowerhemisphere

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    13/31

    verybottomofthesphere,theentirevector

    decelerationreducesthedownwardpropagationveloc-

    owerhemispherethedecelerationisdiminished

    eneteffectfordownwardpropagationisto reduce

    aflatteningflame-zone,forasthe flamekernel

    ttendsto becanceledbyitsbuoyantdeceleration.

    tgeneratesapproximatelythecorrectoverallflame

    agation,eventotheextentofreversingthecurva-

    frontto dimpleinwardatthebottomofthe sphere

    ybeginstoexceedthegravity-freeflamespeed(18).

    uchchangesinflameshapemustnecessarily

    etconstant.Considerfirstpropagationfromthe

    anarignition.Neglectbuoyancyforthemoment

    agateat Suintotheinitiallystatic,cold gas.

    low structuregeneratedbythecombustionforce

    itinteractswiththewallboundaryconstraints.

    bedbyJost (12),whoshowedthatpropagationof

    anarwavesoonbecomesimpossible.Thegas

    omtheunburnedside mustadjusttoaPouiseille

    edside,andi tmustalsouniformlyaccelerateto

    thedirectionperpendiculartothe flamefront.

    ossibletosatisfybothconstraintsacrossaflat

    inglytheflamefronttakesonaparaboloidshapein

    ,andvorticesappearinthecoldgaszonejust ahead

    onofflamecurvatureis independentofbuoyancy.

    mplexitiesthatareobservedinthe presenceof

    figures2and 3,whichdepictspropagationina

    esincombustionandbuoyancy

    o

    e.

    y,

    tion

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    14/31

    ndflame-zonestructurefor

    tube.

    flu-

    nd

    .

    rdflowofcoldgas(9-11).Thisconvectivecell motion

    velopingparaboloidflameshape,asdepictedin fig-

    extendsto accumulatehot,burnedgasesnearthe

    nburnedgasesnearthebottom.Thebuoyancyvortex

    ces)propagationnearthetopand diminishestherate

    ttom.

    eshape andwidthoftheflamefront are

    ehorizontallimitisapproached,a conditionis

    onswherethe componentofthebuoyantretardation

    heflamefrontjust balancesthetrueflamespeed.

    einthe lowerregion,andtheflameissaid to

    ube.Atthehorizontallimit, flamepropagation

    arthetop ofthetubeis finallyquenchedby

    alllossesandbyflame-stretchflowgradienteffects.

    e-zonethicknessthatmightbe expectednear

    depictedinfigure3.Theflame-zonethickness

    enedbyconvectiveanddiffusiveflowsnearthe

    eisaflame zonewideningnearthetopofthe tube

    uenchedboundarylayeratthewall.Themaximum

    eshouldbejustbelowthe topofthetube,where

    as narrowest.Thereisnecessarilyaflowgradient

    vectivevortexinallregionsof theflamezone.

    dby balancingthebuoyancyforcecouplethat

    vortexagainstthecombustionforce(8).

    RATUSDIMENSIONSANDBOUNDARYCONSTRAINTS

    SOFFLAMMABILITY

    asurementsformethane-airmixturesintubesof

    ysummarizedinfigure4. Alsoshownaretheburn-

    tsofAndrewsandBradley(1),GunterandJanisch (7),

    ,andEgertonandThabet(4_) .Theleancomposition

    ionareshownasupward-directedarrows.Thearrow

    thediametersof thetubeswithinwhichthelimits

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    15/31

    mits

    asurementsinmethane-air

    ctionsof flame

    yingdiameter,

    velocitiesat

    .

    ws,

    l

    s

    e

    ws

    m-

    al

    e

    u.

    m

    its

    ed

    e

    -

    end.

    b-

    me

    d

    0percent(^2).This ispresumablythelean

    on,intheabsenceof wallcomplications,usinga

    oncriterion.Thesoftcriterionisthe detection

    flamepropagationthatextendsfar beyondthe

    estimatethelimit burningvelocityforwall

    iamtube,usingaconsistentvalue(8) foraof

    e= 25(itsmeasuredvalue)gives(Su)b= 6cm/sec.

    o(Su)a,thehorizontallimit velocityforquench-

    Onewouldthus concludethatwalllossesina

    mpariblesignificancetobuoyancyindetermining

    emaygo abitfurther.ConsiderthemeasuredSu

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    16/31

    (theuppercurveinfigure 4)andthemeasured

    ntmethanein the2.5-cm-diamtube(h2.5in

    mitcompositionupwarduntil itintersectsthe

    ngvelocityof12cm/sec.Is itacoincidence

    ltothe sumof(Su)aand(Su)b?Note alsothat

    ufor theuandh valuesintubesoffixed diameter

    f3cm/secbetween(Su)a>e.fand(Su)a .Oneis

    asimpleadditiveeffect:in thepresenceofboth

    thehorizontallimitswouldcorrespondto the

    )

    ouldreplace (Su)aby(Su)a,e^.

    on3 agreeswiththedatais indicatedin

    nlimitsin tubesofvaryingdiametersaretaken

    .(Su)bis thencalculatedforthevaryingtube

    5, asabove.Themeasuredlimitcompositiongives

    chistakenfromtheburning velocitydataof

    EgertonandThabet(4).The(Su)a valuefor

    akenas6cm/sec,while the(Su)ae/|.valuefor

    enas3cm/sec.Comparisonofthe lasttwocolumns

    ation3isapproximatelycorrect.However,onecan-

    he precisevaluesforeachcasec onsideredinview

    ntaluncertainties.Forexample,themeasuredlimit

    geofwatervaporcontents,varyingfrommethane-

    o mixturesthatwerefullysaturatedatambient

    itcompositionswereforclosed-endignition,

    n.ThemeasuredSuvaluesare takentobeequal

    goodassumptionformixturesabove7 percent

    velocitiesareexceedinglydifficultto measurein

    methane,preciselybecauseofthese competing

    rthatinorder toobtainaccurateleanlimit

    dependentofwalllosseffects,onemustsatisfythe

    (Su)a>e^.If,for example,onesets(Su)b= 0.1

    equirement,thenthe flammabilitytubeforhori-

    sshouldhaveatubediameterin excessof20cm.

    orintroducedbyusinga 10-cmtubeisatmost 0.1

    elativeerror of1to2 partsin50.Clearly,

    ametertubesare "truer"earthlylimitsthanthose

    tertubes.Ultimately,thesizerequiredtoobtainthe

    ythe accuracydesiredbytheinvestigator.

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    17/31

    hesumof limitvelocities,(Su)a+(Su)b,

    burningvelocity

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    18/31

    nt.Thiscouldalterthe proportionalityconstant

    antly.Nevertheless,forsmallerdimension,near

    bsenceofupper boundaryconstraints,equation4

    ardstagnationlimit.Asbefore,thelimitis given

    =(Su )a^+(Su )b.Now,however,thelimitveloci-

    dprocessbcontainanr0-dependence.(Su)bvaries

    (Su)adependsonr0 .Thesum(Su)a^+ (Su)bnow

    pendence.Thisisshowninfigure 5,wherethesum

    wardpropagationiscontrastedwiththesums for

    pagation.Inthelattertwocases,as tubediameters

    a"true"limitcondition,andthelimitvelocity

    ependentoftubedimensionsanddependsonly on

    n,thesituationisclearlymorecomplicated.

    ueforlarge r0,thereisonlya broadminimumin

    .This transitionregionisin facttheregion

    ementsofdownwardlimitsaremade.Forthe curves

    -

    ting

    y

    an

    y

    p-

    ue

    t

    n

    s

    nis

    y

    Su

    n

    mitburningvelocitiesfor

    enching,and(b)wall-

    onoftubediameterfor

    epropagation.

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    19/31

    bouttheleanestvalueobservedfor thedownward

    re5predictsthat downwardlimitsshouldshift

    onsincreaseford0 >20cm.Thisis inmarkedcon-

    aviorforthe upwardandhorizontalcurves.

    edatatoindicatethat theformulaforthebuoyant

    ngflamekernel,whichwasusedto deriveequa-

    earevirtuallynodataon thedownwardflame

    ofdiameterssignificantlylargerthan20 cm.

    ardcurveoffigure5 predictthatdownwardlean

    rfortheselargerdimensions.Forexample,the

    50-cm-diamtubeis13.5cm/sec.Fromfigure4,

    wardlimitof6.2 percentmethaneforthe50-cm-diam

    ube,onepredictsadownwardlimitof 6.5pctmeth-

    er,thedownwardlimitshouldreachthe stoi-

    tina tube1,000cmindiameter.

    oclaimthat suchprojectionsareunrealistic.

    sofbuoyancy.Theyresultfromthe operational

    mepropagation.Thetubeisnecessarilyconstrained

    oryreferenceframeasthecenterof flameprop-

    tvelocity.Yet,toan observeratalargedis-

    nition,the effectisrealinthat theflamefront

    e combustiblemediumisofinfinitevolumeandif

    opentothesurroundings.

    atsuggeststhattheeffectis indeedarti-

    ontainerisclosed,thenregardlessof the

    ntrise velocitywillnecessarilyapproachzero

    gwaveapproachestheclosedtopofthe container.

    mncannolongerrise bybuoyancy,downwardpropaga-

    dtheflamefrontcanthen reachanobserversituated

    Suchan effectisindeedobservedforpropaga-

    e.

    hapeofa fireballinanN2-dilutedmethane-air

    dburningvelocityofabout9cm/seccorrespondstothat

    xture.Its motionandshapearedepictedbySapko

    1+Themixturedoesnotpropagateat allinthe

    thefireballdiameterexceeds10to20cm.This

    odistort markedlyandtodimpleinwardatits

    softhe distortedfireballcontinuestoriseupward

    it approachestheclosedtopofthespherical

    tactsthetop ofthesphere,itsbuoyantvelocity

    opagationinthedownwarddirectionbecomespossi-

    servedtoestablishaflatflamefront that

    ternaryonecontaining6.9percentCH^j, 65.8percent

    hichis equivalenttoabinarymixtureof5.8

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    20/31

    wnwardandcompletelyconsumestheremainderofthe

    the12-ft-diamsphere.

    wardpropagation,(Su)a^,+(Su)b^.,infigure5,

    ptionthattheflammabilitytubeis opensothat

    e coldgasflowsrequiredtocompensateforthe

    eburnedgascolumn.Asthe burnedgascolumn

    gasexitsfromthetop andentersfromthebottom.

    datthetop,equation4 soonbecomesinapplica-

    tcanno longerbeastagnationlimit.If thetube

    egasis ignitedatthatclosedend,then thecold

    fthewaveis drivenbytheburnedgasexpansion

    nditionatthetopofthe tube,andtherequired

    cityprofileinthe coldgasventingfromthe

    ownwardlimitwouldpresumablybegovernedby

    amestretch.Thesum(Su)a ^+(Su)bwouldthen

    mptoticlimitinthe6-cm/secrange.

    cyeffectfordownwardflamepropagationappears

    perimentalmethod:themeasureddownwardlimit

    dentonwhetherthesystemisopenorclosedwith

    s.Butitseems hardlycorrecttoconsiderthe

    rtifact,whenitis preciselytheprocessresponsi-

    fthe limitsofflammabilityinthefirst place.

    NERTPOWDERS

    problemofflamepropagationinnarrowtubes,

    tion1,whichdefinesthe limitburningvelocity

    esto thetubewalls.Itis givenby:

    etnumberfor wallquenching.Nowforacircular

    sr, theinternaltubesurfaceareatowhichcom-

    rr.Thesourceof combustionenergyisthegas

    sevolumeisirr02.The surfacetovolumeratiois

    isthisratiothat determinesthecriticaldimen-

    micsourcestrengthisovercomebythe thermallosses

    f thissurfacetovolumeratio,thelimit burning

    boundaryis notanexternalonebutan internal

    s pulverizedintoapowderinsucha waythatthe

    ratioispreservedinthesurfaceareaofpowderper

    gas.Inthis sense,thequenchingeffectcausedby

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    21/31

    wdertoaflammablemixturemaybeviewedasthe

    mbustionwaveby"internalwalls."

    autiousinapplyingthisconcept,forthereis

    tweentheinternalsurfaceareaofthepowderand

    ctwall.The wallboundarycanconductheattoa

    niteheatcapacity;howeverthe"internalwall"

    tcapacity.Wewillevaluatethefiniteheatcapac-

    tionsinthe followingsection.

    d,it isassumedthattheseinertpowderwalls

    onductive-convectivewalllossesin thesamewaythat

    eraninertdustof soliddensityp, consistingof

    eterdp,dispersedin thegasmixtureata mass

    .Thenumberdensityofparticlesisn =6Cm/irdp3pp.

    r unitvolumeofgasis mrdp2,andhencethesur-

    tvolumeof flammablegasis

    (7)

    oequation6and settinga=0.55cm /secgives

    owthe limitburningvelocityforquenching

    nertparticlesintheflammablemixture.

    nmind(to bediscussedinthefollowingsec-

    asthelimitburningvelocityfor inertparticle

    causeoftheadditivityprinciple(Eq.3),(Su)'b

    reductionintheidealburningvelocitycausedby

    owderin theflammablemixture.Asystemiscom-

    ertpowderwhenthatreductionin burningvelocity,

    bybuoyantquenching,isequalto theidealvalue;

    )a =(Su)ideal.However,iftheflameis not

    fthermallossestotheinert additiveisstillfelt

    anditsreal burningvelocityisreducedby(Su) 'b.

    withrespecttoCmgives

    u)'b foragivenCmwiththe reductionin

    yS,andsolvingfor Pegives

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    22/31

    vestudiedthe effectofaddedrockdust(CaC03)

    dersontheburningvelocityofmethaneair mixtures.

    e2,andtheresultantvaluesfor Pecalculated

    on10.The additionofthepowderedinertdusts

    lquenchingbyheatconductiontotheadditive,but

    seinemissivityofthe burnedgascausedbythe

    ncreaseresultsinradiativelosses,whichare

    diativeloss limitvelocity(Su)c.Themagnitude

    8cm/secdependingonthedustconcentrationand

    bothlosseffectsarepresentsimultaneously,it

    ssmallvalue of(Su)cfromthemeasuredASU

    heeffectthatisattributableto (Su)balone.

    ersforinertquenchingaveragetoaboutPe =100.

    somewhathigherthanthe theoreticalwallloss

    one cannotbetooimpressedwiththeresult that

    wasnoless effectivethanthe9umparticle.

    sfor"internalwallquenching"

    yinertpowders

    10.

    fectofinternalquenchingbyinertparticles

    SmootandHorton (19)dataontheburning velocity

    eirdatashowthat burningvelocitiespeakat

    hatdependmarkedlyonparticlesize.Forfine

    ngvelocity,(Su)(max),occursatrelativelylow

    reasforcoarserduststhemaximumappearsatamuch

    .(Su)(max)issignificantlysmallerforthe

    rehigherdustloadings.

    orrelatesto thefactthattheoverallrate

    esemaximainS ,islimitedby therateofpyrol-

    ecoalparticles.All peakvaluescorrespond

    a stoichiometricconcentrationofcombustible

    ecoarserthedust,themoredifficultit isfor

    atilizeduringits passagethroughtheflame

    cle,thesmalleris thefractionofitscom-

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    23/31

    strequiredtogenerateastoichiometricconcentration

    orthecoarsecoals,onlytheshallowshellnear

    pestcornerscandevolatilizeintimetocontribute

    hefinercoals,mostof thecoalparticledevolati-

    .

    massof coal(ortheremainingcharresidue)

    powdertowhichcombustionenergyislost.The

    adingis seeninthedata asareductionof

    thePecletconstants,calculatedfromequation10,

    tioninSu(max)by theexcessinertdust.Forthe

    for"internalwallquenching"of

    xcesscoalloading

    10.

    hequenchingofexcesscoaldustin the

    whichisquiteclosetothe valuescalculated

    ngofmethane-airmixturesbyrockdustandalumina

    FECTS;PARTICLELAGINTHEFLAMEFRONT;

    LAYER

    thederivationof equation8,thatrelatedto

    naddedpowder.Equation6, fortube-wall

    ndertheassumptionthatheatwascontinuouslybeing

    undarylayerandthroughtheinteriorsurfaceto a

    pacity.Ineffect,itwasassumedthat thecold

    edcoldat Tuduringandafterthe passageof

    sassumptionwasincludedinthesubsequentderiva-

    ever,thewallboundaryhasafiniteheatcapacity

    volumeit contains),thenthetemperaturegradient

    edheatlossto thewallwouldnotbe maintained.

    ternalwalls"ofinertparticles,wheretheir

    ausetheirwalltemperaturetorise significantly

    eof theflamefront.Iftheparticlemass isso

    risesas fastasthegastemperaturerises within

    mallossesare notmaintained,andthegradient

    stanceitis notthesurface-to-volume

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    24/31

    portant,butonlyits heatcapacityperunitvol-

    tothemassconcentrationCm,andindependentof

    a"smallparticle"is thecaseofaninert gas

    etoconsiderthisextremecasein ordertobe

    equation8. TheN2particlesaddedtoa flammable

    s rapidlyasthecombustionproducts,andsincethey

    sing equation8tocalculateinertingrequirements

    esult.Thiscanbe demonstratedasfollows:

    etheinert gasismixedonthe molecularscale,

    ldbe themoleculardiameterofthenitrogenmol-

    olid) Ndensityforpand typicalvaluesfor

    dto thepredictionthataPstoichiometricmethane-

    dat aCmvalueofonly 1mg/1.Thepredictionis

    eN2moleculesarenotcoldwallsthat remaincool

    e,butrathertheir heatcapacityissosmallthat

    creasinggastemperature.Datashowthatthemix-

    aCmof500mg/1,whichcorrespondstosome

    heabsurdresult of1mg/1wasobtainedbecause

    ndits rangeofvalidity.The1mg/1valuecouldbe

    culehadaninfiniteheatcapacityandcouldthus

    t wasimbeddedinreactinggaswhosetemperature

    s physicallyimpossible,sinceeachmoleculehas

    .

    volumeof1 mg/1ofN2istrivial compared

    unitvolumeof thereactingmixturewhosetotal

    00mg/1.Thus,theadded1mg/1of N2hasatrivial

    etemperatureortherate offlamepropagation.

    500mg/1ofaddedN2to reducetheflametempera-

    eof2,200 Ktoits limitvaluenear1,500K (8).

    nalinsightintothefallacyjust consideredby

    ofCm- 500mg/1andsubstitutingthetypicalvalues

    the"effectiveparticlediameter"atwhichquench-

    ogenwere tobeconsideredasaninert powder.

    value isclearlymuchlargerthanthe molecular

    nificantlylargerthanthemeanfreepathfor

    1ym.Thislatter conditionisnecessaryforthe

    fdensity,heatcapacity,andthermalconductivity

    ngsin thefluidcontinuum.Clearlycontinuumequa-

    quations1 and8,andhencetheirvalidityis

    L.Thisconditionisviolatedbychoosing

    theN2moleculeasthe particlediameter.

    erthis particleheatcapacityandheating

    To whatextentcaninertparticlesaddedtoa

    wtheincreasinggastemperaturesintheflamefront,

    agthegastemperatures?

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    25/31

    particleoffinitemasssubjectedto alami-

    emperatureriseisgiven by:

    ume,citsheat capacitypergram.andAis its

    minarheatflux totheparticleis equatedtothe

    rdensity,Scp (T-T ).Thisgives:

    -T).

    mof particleandgasareapproximatelyequal

    echaracteristic timeconstantforparticle

    ontgives:

    aracteristicheatingtime isT= a/Su2.

    heatingtimetoflamegasheatingtime is

    clearlytheparticletemperatureshould

    perature.AT-ratioof lessthanunityis aphysi-

    wouldviolateour continuumconstraintthat

    wouldmeanthattheparticletemperatureshould

    ggastemperatureofthe combustingsystem.In

    oor lessthanunity),theparticlec annotbe

    all,andonlyitsmassor totalheatcapacityis

    effect.If, ontheotherhand,theT-ratio is

    nequation8 isapplicable,andtheparticleshould

    ngflamefrontpassage.Settinga=0.55 cm2/sec,

    articleandgasdensities,and usingthelimit

    cm/sec,oneobtainsr- 0.5ymfora T-ratioof

    smaller)particleshould closelyfollowthe

    rewithintheflamefront,andequation8 wouldnot

    hand, foralOym-diamparticletheT-ratiois 10

    peratureriseoftheparticlewill beanorderof

    atofthegas. ThusalOym-diam(orlarger)particle

    allduringflamefrontpassage,andequation8should

    sconsideredintables 2and3werei ntheappli-

    sthekinematicsofinert particlemotionin

    citly assumedinderivingequations8and11

    he flamefrontwouldinstantaneouslyfollowthe

    attheir velocitieswerealwaysequaltothegas

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    26/31

    rthevalidityofthatassumption.Thequestionis

    First,thereis thequestionoftheheattransfer

    theparticle.Theconductive-convectiveheat

    oasolidsurface(or thereverse)isproportional

    whichis proportionaltotheratioof particle

    yerthickness.Forthedimensionsandvelocitiesof

    tothe squarerootoftheReynoldsnumber,/Re.

    relativevelocities,Reis smallandNu->1. This

    sumedinthederivationofequation11,and only

    boundarylayerwhosedimensionswerecomparableto

    consideredtobeimportantinthe heattransfer

    articleshouldlagthegas flow,Nucanexceed

    rheattransfertotheparticle.If thereis

    ofgaswithrespecttothe particle,theboundary

    smallerthantheparticledimensions.Actually,for

    tiesofinteresthere,Nuis neversignificantly

    firstproblemisofminor importance.

    er,isthesecondreasonforconsideringparti-

    elags thegasflowsothat itstimeoftravel

    gnificantlylongerthan thatofthegas,then its

    heflame frontwillbehigherthan itsinitial

    staticmixture.Ifparticleslagin theaccelerating

    eflamefrontandtheir effectismagnified.

    icsofparticlemotioninorderto estimatethis

    ceonasphere ofradiusr, whichisimbeddedin

    ngwithauniformrelativevelocityuwithrespect

    Stokes'law:

    ectionoftherelativevelocityu.A moreaccu-

    or(1 +3/16Re),whichwouldincreasethedrag

    hemostextremecase.Wehereignorethesecond

    rmula.Nowin acoordinatesystemmovingwiththe

    elocity,gasand particlesareinitiallyatrest

    andbothenter theflamefrontatthe velocityS.

    aversesthroughthe flamefrontandreactstogive

    ityisSb=Supu/pb.The detailsoftheacceleration

    ombustionforcedependonthe detailsoftheflame

    andprobablyreasonablyaccurate,toassumea

    svelocityas ittraversesacrosstheflamefront.

    x= 0),(vg)0=Suwhile att= Tgas=a/Su2

    = Supu/pb.Thisgives

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    27/31

    ocityuponentering theflamefrontisalso

    u)0 =(vg)0-(vp)0=0.At anytimet,while

    amefront(thatis,t

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    28/31

    kedly.Itsexitvelocityfromtheflamefrontwould

    gasvelocity.Thus,althoughit andthegasenter

    sec,thegasexitsatS -180cm/secwhereasthe

    60 cm/sec.Theparticlevelocityhasthusonly

    cityhasincreasedbya factorofsix.Thissame

    wouldbepresentevenat thelimitvelocityfora

    ameter.Theeffectofthisparticlelag isasig-

    ectiveconcentrationofthedustinthe flame

    ethatboththefiniteheatcapacityeffect

    particledrageffectconsideredabove,tendto

    e-sizedependencepredictedbyequation8.Thefinite

    thequenchingeffectivenessofverysmallparti-

    ticalsize inthe1-10ymrange,they cannolonger

    cledrag effectcausesthecoarserparticlesto

    flowof theflamefront,andasthey "pileup"

    ionincreases.Thisincreasedeffectiveconcentra-

    orthecoarserparticles'lowersurfacearea.Both

    particle-sizedependenceofthequenchingbehavior

    sizedthattheseconsiderationsapplyonlyto

    -> 1.Forsuchinertparticlestheboundarylayer

    od,and heattransportisbypure conduction.If,

    tinertand cangeneratevolatilesduringitspass-

    t,thenits effectivenessmaybemagnifiedintwo

    ationis endothermic,thenclearlyitseffec-

    gnified,andit wouldtendtoremaincolderfor

    ation11 wouldthenunderestimateitst-ratio,

    inasacoldwalldownto smallerparticlesizes.

    neratesvolatileduringitspassagethroughthe

    ferbetweenthegasand theparticleisnolonger

    theboundarylayer.Instead,thedevolatilizing

    rroundingflamegasesonadimensionalscalethat

    tudelargerthantheboundarylayerthicknessfor

    e.Assumingthatthegasesare inertorinhibit-

    convectivecooling(orchemicalinhibition)that

    pureconductionprocessusedto deriveequations2

    ogytothis argumentfortheadditionofgases:

    moleculeswithmanydegreesofmolecularmotion

    ertingagentsthansimpleratomicordiatomic

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    29/31

    gvelocities,whichwaspreviouslyappliedto

    chingbybuoyancy(processa),appearsto beapplica-

    hermalquenchingofflamesbyheatlosses toexter-

    nallyadded inertpowders.Forwalllosses

    tyis(Su)b =aPe/2r0,andinthe presenceof

    sses,asimpleadditivityprincipleappearsto apply.

    th lossprocesses,thelimitofflamepropagation

    f limitvelocities,(Su)a+(Su)b,is justequalto

    ofthemixture,(Su)ideal(whichis measuredinthe

    ).ThevalueofthecriticalPecletconstant,Pe,

    pearstobe sensitivetoflameshapeandboundary

    he rangePe-25 to60appearstoapply tomost

    roblemof downwardflamepropagationinopen

    dependenceontubediameter.Thisunusualdepend-

    odefinea downwardflamepropagationlimitthat

    toftubesize. Downwardflamepropagationlimits

    sualdependenceonboundaryconstraints,namely,

    openorclosed.Thesespecialcomplicationsare

    rhorizontalflamepropagation.Forthosetwo

    ation,"true"limitsareobservablesolongas the

    largeenoughthat(S) (S) .

    owderaddition,alimitvelocityisobtainable

    83PeCm/ppdp.MeasuredcriticalPecletconstants

    addedinert powdersareintherange Pe-75-175.

    acityeffects,theequationfor(Su)'bis applicable

    10ymin diameter.Ananalysisofthekinematics

    eleratingflamefrontwasalso presented.It

    articlesabove50umin diameter,asignificant

    hisparticlelageffectincreasestheparticle's

    heflame front.Boththefiniteheatcapacity

    effectlimitthe applicabilityoftheabove

    dtoreduceormoderatethe predicted

    nce.

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    30/31

    radley.TheBurningVelocityofMethane-Air

    dFlame,v.19,1972,pp.275-288.

    Jones.Limits ofFlammabilityofGasesand

    ,1952,155pp.

    eap,andR.Pritchard.AmbientAtmosphereEffects

    ntsofBurningVelocity.CombustionandFlame,

    K.Thabet.FlamePropagation:TheMeasure-

    sofSlowFlamesandtheDeterminationof

    c.Roy.Soc.(London),Ser.A,v.211,1952,

    chingofLaminarOxyhydrogenFlamesbySolid

    stionSymposium,Madison,Wisc.,Sept.7-11,

    kinsCo.,Baltimore,Md.,1949,pp.110-120.

    tine.AnalyticalCriteriafor Flammability

    nternat.)onCombustion,Pennsylvania

    k,Pa.,Aug.20-25,1972.TheCombustion

    973,pp.1109-1118.

    h.MeasurementsofBurningVelocityinaFlat

    andFlame,v.19,1972,p.49.

    yofFlammabilityLimits.NaturalConvection.

    pp.

    ymp.(Internat.)onCombustion,

    nol.,Cambridge,Mass.,Aug.15-20,1976.

    Pittsburgh,Pa.,1977,p.1404.

    ashdollar.TheFlammabilityLimitsofCoal

    978FallTech.MeetingTheCombustion

    Nov.29-Dec.1,1978.TheCombustion

    pp.35-1to35-4.

    dollar,C.Litton,andD.Burgess.TheDiffusion

    Buoyancy-InducedFlows,Oscillations,

    rge-ScaleLimitingRates.BuMinesRI8263,

    CombustionProcessesinGases.McGraw-Hill

    ,93pp.

    dyofFlammabilityLimits.Proc.Roy.Soc.

    965,p.134.

    PublicDomain,

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle

  • 8/9/2019 The Theory of Flammability Limits - Conductive Convective Wall Losses

    31/31

    S.M.Simpson.LimitsofFlammability.

    )onCombustion,YaleUniversity,NewHaven,

    ReinholdPub.Co.,NewYork,1957,pp.20-27.

    BuoyantDiffusionFlames:SomeExperimental

    ch.Meeting,TheCombustionInstitute,

    Dec.1,1978.TheCombustionInstitute,

    4.

    sin CombustionScienceandTechnology.

    k,v.1,1960,145pp.

    L.Berlad.TheEffectofFuelTypeand

    hing.Proc.6thSymp.(Internat.)onCombustion,

    en,Conn.,Aug.19-24,1956.ReinholdPub.Co.,

    .

    o,andJ. M.Kutcha.FlameandPressureDevelop-

    Air-N2Explosions.BuoyancyEffectsandVent-

    esRI8176,1976,32pp.

    Horton.ExploratoryStudiesofFlameandExplo-

    ort.V.1,BuMinesGrantNo.G0177034,

    ilableatPittsburgh MiningandSafety

    yof InflammabilityLimitsandFlame-Quenching.

    Ser.A,v.240,1957,p.83.

    wis.Theoryof Ignition,QuenchingandStabiliza-

    entGasMixtures.Proc.3dCombustion

    ept.7-11,1948.TheWilliamsandWilkins

    ,pp.68-79.

    mabilityCharacteristicsofCombustionGasesand

    ,1965,121pp.

    TINGOFFICE:1980-603-102/85INT.-BU.OFMINES,PGH.,PA.24751

    Google-d

    igitized

    /http://www.h

    athitrust.org/access

    _use#pd-g

    oogle