MODELING OF wO/e COMBUSTION WITH A ROBUST NAVlER·STOKFS COMBUSTION CODE·

Pak·Yan Liang··. RmaldJ. UngewiUW'*, and Merlin D. Schwnan···RocketdyneDivisioo, Rockwell !ntematiOlial

ABSTRACT SYNOPSIS OF NQMERICAL SCHEME

Delails of the numerical modd aft:' di.cusscd in Ref_ 2. Forcompletcnell, the major governing eqUAtions are listed below:

whenl S is the 4ropkt CCjllpling tenn, G is the body force, andthe vUCOUI stmls tensor i. given by

In an effort co creaIe a ICCO!Id-generation eombuation code wi,thturnaround times fast.enough to enable. incluSion of CPO in thecomboltor ~ligo procell, III efficient chemistry model forHIOIC oombuatiooha.~ developed and inCOlpOtlted into thepl'IlIIIIIe~ Navier·Su:lkes solver of the GALACSY~. '!hI:'model ft:'duce. the HIO/C equilibrium' ~om'lDiQ a ilinSi.ecubic equation fot the oxygen balance, and abo 9<l'I!~"s afinitlNlle relction for the hyd~carboD oxidation.~paria~. wilh .~. shOW, good p~clior;' of a1l~e.proUI."- .:KJIDl; unc;vc:m.,.~ ~i.IO 1M VGI, I.VW V".l. \rUU~IU.W;VI.

level. which aro being investigtted by refinements to 1ho cubiceq~tiOli. .

INTRODUcrlON

The maturation of relatively robust, general-pulpOle Navier·Stoke. lolver ~nol08Y in the past decade hI' led to. thepromise that routine application of Computation FluidDynanics (CPO) CO the design md analySd of QI:lmbulting flowSYltem. ClUl now be a reality. BefO«l that happens, however, afew remaining burdlct need to be sv.nnOlUlled, inetuding:

The generation and validation of a large sel~n ofcbemiltJy model. covering a wide rangll of differentkvell of sophistiCllti<m.

Enhance the confidence level. in mode1l of combullion­rc1aI.ed turbulent Innsport.

SignificlJlt advanccmentJ in the effecliveneu andvcnatility (problem-independcnce) Clf stifflieu·mitigation techniques, elPCcially those dealing withsoun;e-tenn stiffnelJ in spnIy combustion flows. .

MISS:

ap .):a.. +Vo( pu '" Pd

aPu -) V .., 2.-.., S--+Vo(puu:- P·,,(;;-pk3)+"oct+ +pGa.. -

(J ,. f1(Vu + (VII) TJ+ A(Vu) I~ .

aad It il the turbulent kinetic energy.

(I)

(2)

(3)

(4)

Broad-buw anchoring of spray atclmi~tion data fordifferent injector typcl.

whlml the beal flux vector includes boIh conduction and speciesdiffusion iMns: .

The strategy for development of the second-generationGALACSY «leneral .Al,gorithrn fCir Analylel Clf C,ombustionIDtems) family of spray combustion codu at Rwkctdync bbued on the above· observatimu_ Other than SllbslU1~Uy

extending the successful preuured·ba.ed algorithm of theREACT 1 code methodology to handle transonic, muIli-spccicsand multi-phase phenomena, a package of comprebendvcreseucll iniliativcs aimed at each of the' four hurdles mentionedabove u beina unckltaken. This report documents Sa!ic ¥ theprogrIl" being made in the panicular al'Cll of chcmiltJy modeldevelopment for the combustion of methanc, which can beeasily extended to model the combuitloo of othe~ simplehydroc:arbon••

J = KVT - PgDDmVCprr!p v)m

Specie; M:.

(5)

(6)

... Work panially supported by NASA MSFC Contfacl NAS8·40000... Memben Technical Staff, CFD Technology O:nter, MembersAJAA..... Retired Member Technical Slaff, Advanw:! CombustionDevices

1

Turbulent Kinetic Energy:

;l,\k •'f"+Vo(pllk),,-, m

.~VU+(J:VU+Vi~J-Jii+ Ws• CIt droplet

turbulence suppression

The volume fractiao equation (9) arises bei:&UlO GALACSY iIdesigned to handle an additional immi&cible liquid pIwc (which11 incompressible and chcmica1ly inert) ulina the Volume-of­Flllid (VOf) een-partiJionil\s teclmique3. BqlWilllll (l) lhrough(9) were nOl activated for purposes of thiJ Illldy. Note !hat thecClllleJVatica eqllations am aho written for the per unit toIa1mUI (cell aWl1lge ) vamllaes, defined AI

then CHBMEQ executel the chcmi&uy model to solve for: ~molar compolitioo baled on the atomic totals available.Homva-, .. in tho cue of CALCSP, CAlCEQ doct evaluate tbcheat flux. lctm. dua to speciel diffusion for each and everyequilibrium COlnpowld speciei.

THE wO/e CHEMISTRY MODEL

Pnlvioui Ilxperienc.; (e.g., Ref. S) has ahown that the che:rnisttyportion of a typical combustion SirnulatiOll can take upwardll of30* of tho overall compulllliOllal tinte ev~ if only a relativelysimple let of eqlillibrium !elmonl w~ involved. and mon: if adetai.Ied let of kinetic reactionl ·wete used. The need 10 invert I

coefficient matrix for eqlillibrium reamonl at eve!)' iteration ortime step makel a general purpose equilibrium chemi.trypackage sucla II CHBMBQ in Rei. 6 inherently time­COIlftllllinll. It is oIxerved th.t fM the. maiorilV of onm......ti"",applitatioo., the core of the chemiltry model caoOstl of thebydto&en-oxygCtl tucUoni which a~ alml»\ always fastenough compared to the fluid dynamic time lcales 10 beCOIlIidered equilibrium reacliOOI. These IUmons can then beangmcnted with rUlite-rate hydtocarbon oxidation or nitrogenkinetic reacri.OOI Ai needed. Fmthennote. it is obIervcd thal theinternal flamo nf\lClUfO is typically dominaled by theavai1abilli:y of oxygen ItomJ at a particular point, and thus atrIIly general eqllillbrillm c:hemimy package is not. neclJslA!)'.The foUowinlit modelthua Wei advantage of mese obsel'YstionslO reduce the hydrogen-oxygen l1lac:tions into a single cubic:equation for oxygen coneenttatlon. whidl can be solved veryefficlCltlLly. All other O/H specieJ can Ihan be algebraicallyevaluated. nWdng the ovcrsll chemiltry taleuJ.atiOllI extremelyfan.

(8)

(9)

(10)

(ll)

ape + vo(~=-~f-c:YPWU + Vi .e.ve] +at '3 r.e

!tcla:Vu.~i..,~.:Jf ..

p.. !fPg + (l.1)p l

Volume Frr.cIion:

which defaulU back to the simple gucoUl variables when noliquid phase is present. Thi. synem of equalioniu then closedwith the multi-Ipecies perfect gas equation and limplesummlllion fUlcs for the gu mixture. Rcal gu property da~

(enthalpy, heat of fOmJ4lion, etc.) in the fonn of lANNAFmbles are aur.omalieally included into the program from libruyfiles.

GALACSY uaea a seq.uenlialsolver bued CII1 ex.tension. of theSIMPLB4 .-pproach. as highliglll4ld in Piglll'C 1. In the I~dy­sta1is vemllll. all lO\Inl8 terml (eval~ with the aid of vanoo.physical submodels) mud be mpplicd to the mQp lUbl'OlUiuclin lerm. of changes per unit time. As iDdioated in the flowchan, up to rOW' major mbro'.ninet C&11 be iDvolved in thacurrent ohemiltry/speeiea diffuJion Iubmodd. CHBMKNcontains a gener&1-pmpole packageS for solving Arrhenius­type kinetic tale cquauOllI. and is invoked by subroutineCALCSP. The lilter i' ca1Ied to I01n the lpCcies conlervationequation, tq. (1), fot species designated IU "kinetic" specie. byme:; user or for all :lpI!ciel if no chemistry is involved. It abocvalu8ta the heat flltt lenni due 10 speciel diffusion for mOle

,-pedes. On the other hlltld. Cor thole speciel that a~ designatedAI "tquihbrium" specie. controlled by faSl reactions, no ulchlhing .. an effective lQCtion rate exists in steady-stale, and thespecies composition aU any point it dictaled by me baWl" ofconvective 4fld diffusive fluxes of the availlible llUlDlic species.Hence, CALCEQ is invoked to solve for lhe conSCl'Vauon of thetotal~ species rather titan the equilibrium compounds, and

- Pg9Ig + Pl(t-nItI=~"""--.o::..--.....

p(12)

Comider the fonowing set of ~Ililibtium !elOtionl:

(Hz<» = <KH2Ol {Hv ~ (OV~

tl'102+H2~ Hz<> (13)

112HZ+ 112 02 ~OH (Oli) = (Koo) ~ (Hz) ~(OV (14)

112H~H (H) = CKw-V(H:z) (15)

112HZ'" Oz¢HOz (HOV == <KHoz)" (H2) (16)

H2 +<>.2~ H:P2 <H2<W <KJIPl){H:z)(OV (17)

l~O (rIzO) '" <Kc}...J<!JV (18)

.....hen: the K', m: the equilibrium eonstants and lIIe chemicalsymbols in pAmlthesel reprelenU molar <:onr;:entrationl. Notethal thll right hand portiona of eql. (13) through (18) applyonly al equilibrium. Naw the hydrogen balance at thebeginning of the equilibrium calculations (designated bysubscript 0). excluding the hydrogen contained in thehydrocalbon fuel is given by

"~ =2 (Hv +Z(H2O) +(011) + (H)

+(HOV+2(Hz02) (19)

2

whete ftl is the nwnher of hydrogen .tonl~ in the hydrocllbonfuel, and He i. Ihc 8cumo hydrocarboQ. (FDf" methane, m :I 4.)This balance doe. nol change II the cOOIpodtionevolveltoward equilibriwn.

"<C>o '" (CO) + (COV (28)

(28a)

1 "• 2' f(H)o - (Om - (KO».2 (H202)J (20)

where q is the nwnber of caJb6n atoms in the lIydll>CllrbOll. (Formelhane. q:l 1.)

Combtning eql. (rT) and (ZS) and ~ging lead. to

Sl,IbItiluting equation (20) into cquaUQII (19) yield_

01

(21)

(22)

(29)

(30)

<H2)=

Similarly, the oxygen balance is B,iVat by

(23) N«e that if (H:z) '" 0 and (H~) '" 0, then COz sbou1d be zero and"(CO) setcqual to (C>o.

Substitullnll eq. (29), (30) and (13) back into (26) yield.

"(0)0=2 (0)> + (H2O) + (OH)+ 2(tt0» + 2 (HzOV

+ (0)+ (CO) + 2 (COz) (24)

(24&)(31)

where n is the number of oxygen atoms in the hydroeartxln fuel(for methane. n :I 0).

Nowdcfine

Again. let(32)

"eta .. (0)0. (OR) ~ 2 (H0:z). 2 (H202) • (0)

Subltiluling equaUOll (25) into equalion (24) y:iddJ

(25)where p. should be set to I if (H:z):I 0 and (H20) =o.

iilld mlmituting it mlo eq. (31) giVei

(26) (33)

whete (<::0) md (COil will ha'110 nonzero valllCl if a hydrocarbonfuel ill involved. In~ case, the CO/CO2 shift equilibriummwtbe~,which is

Fmally, combining the hydrogen balance equation (23) with(33) yield•• after some manipulation. the cubic equation foroxygen

The calbon balance. again excluding the unburned fuel, is

3

(34)

11uu the WO/e equilibrium caJculation. ltart with cvalualion ofall the K valdlSl .It tbe oe1llenlpeta1llre, which am UIUlIIed to beof the fonn

in the data Iibnuy, and whem TA = T/lOOO. The variabler aH,

ao and Paftl. lilen calculated and the cubic root for the oxygenequ-rioo souJltr. If ttIQre lhan OI'IC~ poWve root whicll allO

IIlCeb the criterlOil '1./ (0,)2:$(~)o axi*U, then~ smaIlClt root

'"if picked if tho ccJl is in I bydrogllD-rich zane, i.ll., if (H)o>1\

(0).... 0IherwlIa the largelt root iI lJie:teod. OnOll the oxV!len

COIl~tiOll i. known, the hydrogen concenttatiOil follow.from eq. (23) and all oIhet equilibrium rpecier can ba~from the eqWli.briuJn COII.ltlnb,

Up to thi. pOint me finite lItC oxidation .tep of thebydrocvbon fuel hu been let OOL TIw. rate, in tennl of mollllof CH4 depicted per unit tUne per unitv~ ill expreIsed .. aone-WlY, lin$le.Aep reactiQn

SAMPLE TFSr CASE

To validate tb.e currmt HIOIC chenlistry model as well II theoveTaI1 SalClOUl eomoolElOll eapabililier of the GALACSY code,the diffu.ion flame experiment by Mitchell' et. aI. hu beenchosen AI a benchmark telt case. In this cooIute4. <»-flawinsrnethane-eiI' experimenl the flame wu probed extensively toobtain detailed species concenll'ation proftles. Furthermore, thelaminar nalUre of the flow removes ooe of the biggest SOIlfCeS ofuncertainty in an CFD simulation. and especially incomblution caces - th.u due: 10 bltbulence modeling. Figure 2contain. a .ch\!m.uU: of the IabontoJY burner taken from Ref. 7.A 41 x 101 (2.54 em x 30 em) uilymmetric grid wu used in thepteIMl CClIDpOtltlaoL Table 1 fIlDImarizes tho peniucnt flowparametm. In 1M numerical model, a total of 12 species (0, c,II. N2•CH4' ~, "Z' ~, "z0, CO; H~, H20 2) arc ttKkedindlldinll \he intrt N." Speclfll. and atomic C S!Jilciel even

. lhovgb it is not involved with any reactiOllI. Far boundarycondition., aU q1WltilillI are specified al the inflow plane, IC(III1tlIn! pre&1lIre of 1.01296 Jl 106 dynes/cm2 ~ lpecified a1the outflow pi-, symmeuy axil ill used .It llie centerline, anda.cJiabItic JIOoWp ooaditioal Ire used It the outer cylinder wall.Gravity is iDc:luded because the problem is largely one of naturalcoovectiQtl. The compuutionlll domain i. initialized 10 beprefilled with aU (76.1%NZ by weight).

(36)Luninu mcosiey throughout i. e.valuated ulinS Sutherland'sfonnula

whm ~11Jl1 .. 1.4S7 " 10-5 -r312'(T+ ltD.) poise (38)

(31)

and a, b specify the effective order of the reactiOll. For themethane test CUe balow, the valUCI used am

wh_ the empirical eoefficienu are lhOle for air. As il turnedout. final ipecies and tempcrfiUi'a promo prediction. are quite_Iitive to the mspective diffurioo c:oeffiQertU, and hence totho effective Schmidt and Prandl1 numbars. Uling constants faraU points of different compositions was clCllrly inadequate, AIII first ordc:r n:finement, the fonowing fannula is used:

AHC = III 1013• ~ = 0, BHe '" 1.S78 x 104•• ::: 0.2$and b = 1.5 (m eg. units).

(39)

The uverall pfOI:edure fer solving the speci~ composition cannow be NIIIDlaQt.ed II followl:

~ Jnll1uding cq. (36) &I • sourea term, solve thehydrocarbon specie. COIUlervatioo eqUl!tioo inCALCSP.

~ SolvG thc al.<linic ip6cl.;;i eooiCfVatioo. eqWOili

for (PC>tQtal' (PH)lOIal and (PO>to«al in CALCEQ.

1\ 1\ A

SeeD C Evaluate the (C)o, (H)o, (0)0 sublQl.l!!s at end of

~

-.u Go on wilh the rest of H/O/C equilibriumca1culatiOll. in CHEMEQ,

4

whl:nl R i. the local gd conilailt and Cp. b the local specificheat It CQIl.l4lU pre31URI.

In the coul'lO of the computatioos, it wu also discovered thaircci.."C'..;1ating l=Jd1ow CiU OOOtlt ~ij ihe ~. plane Ulwarm!he 0lJ1Cf walL This revl!IW flow CUI make convergence to theateady-llat4 sohnfon impossible. ThUI, it wu found neces.uy\0 elimlMte Ibi! batldlow by gt"'..dulilly inareallmg Ihe visco.Utyvaluo toward the exit.. This wu dOl'K\ for oe1ls of. I > 60 (X > 12ern). Sinca all now futurel of intel'Clt are located upstream ofX "" 7 em, lhU treallllCllt should not significantly impact theaccuncy of the rerulli.

Figures 3 through 7 highlighl the computed JesuIls III well astheir comparilOll with dala. As can be SC'ClI in Fig. 3b, the npidheat-up of the central jet remltl in II large acceleration of thecenter stream IIId II compensating revel'$c flow of air ncar the

. w.n, although this recirculation zone is lcrminated by theartificial increase of viscosity tawllrd. the exit. The speciesconlours all show a well-defined OQnical fl8l1le shape with a

InflowT~

Table 1SUlilmary of Flow Parameten tor Methane Buraer

Ibner Tube Outer TubeRadiul ro " 0.635 Clll Ro =2S4cm

Velocity UM=4.50 Uair ,,9.88

c-mJ. cm/I

Vol Flow VW =5.7 Vair .. 1PrT.7

RIlet;C/. ceI.

Imlow 1.013 It 106 1.013 x 106""""...... dvnel/cm2 dyncl/cm2

298 des. K 298 dca. K ,.

'--~-----'----

slight ~ial cx1»n.ion at the bale of tb Olllle. CH4COIIJ1IIIlPtiIJli is virtually eomp1ete by X =6 ern.

Quntitauvc comparbon of l'*dial profilc' at Ume will.loe&tion. 8M pmsented ill Fig,. 5, 6, and 7. The followingobIcrvatiQnf can be made:

I. The major specieI of 02. H20 and CO2 arc gcnuaUyq1.lite well predicted,e~y on the air .ide of thcfluno. A dip in thc prcdi=ed CO2 profile in lbe mid·radilu 1IIl1e imide of lbe flame envelope c:orrespondf tou. <MIl' predicdOil of tho minor .peelcl HZ and. ro in!be sane ",gim. ThiI may be becaute of~asmthe H2 e".l~lion bll'cd on eq. (2) whtn OltYl1:11

concenmlicm is low. (KHlO being • vary large

number), which in tum lcacb to inaccuracie. in CO andCO2 by way of the shi.ft equilibrium eqs. (Z7) and (28).H20 prediction is not affected that much by thi.probkm becaUle. U can be Seal in cq. (13), (H20) will

be wdl pmiided u looS u (Hz) ...; (OJ) .. «a!(KHz<>} is

properly ~allla&ed.

2. CH4 CUlCC11truiOIi is ovcfllil=dided eipeeUlIy towai'dthe centerline. though ~ment iI better t!lfIl obWnedwith the Inodcl by Mitchell ¢C. ;1. in Ref. 7. (SeediSGlUtion on Fi,. 8 billow). Thi., tog~r with thehigh valuei of (CO) and (H2). wii'6iiJOiids 'iiolh a rowvl1ue of (Nz) in the. flam;; zone. A pOiSible explanationmay bll lnlllfficient diffulioo rateI for Ditrogen, or foroxygen, or bQtb. A larger Ill~ dift'udOti nile CiIi

presumably Rducc (Cl4) &110, but the temperature plotlshow that the "flame front." II' indicated by thetemperatnta maximum, if aheady to the outUde of themeasured locations.

:3. The peak tcmpcraturc as well as the centerlinetemperaturel are genetally ovcrprcdicted by about 10%,which caUIeS over predictions on the velocity profllesII well. The SQIlfCe of this discrepancy can be two foleLFint, for the par1ial species intemli energy in theenergy conloN.tion equation, the sensible energywhich includes the heat of formation is cunently IiSedinstead of the absolul.C energy (refetenced 10 absohilezero) which does n~ include the heat of fonnauOII. The

5

advantage of this fonnuh\ic)Q is that no energy IOllICC

tesm WQV1d .rilll due to chemical rm::ti0l1J; the changeilltan~would bll backed out frem the change inCl;IInpocition~ The potc.ntilll. djJ&dv~ge is: thatbecaUIC the diffemnt speciec have VlIstIy different healsof fonnatioo, nUtlllte m.ccuncie. in IQIJIc speciec ClD1lead to lIIrae inacc:utae!ea ill temperature. Seoond, iiDee-sf b the COI1Ie1'Yation vatiab~e in cq. (4), lha heat

::~kr::iD~ihlt-ffi::.':ti:":~::::;~:program to difference mtl diffusion term implicitly.'1'biI leads to additional c:orrcction terms involving.gradlaIu of the· spcGific huIt which .gaiD can ba very-aap and ina<:curate to compute in multi-apcciet flows.Bodi of thwlil source; of =or will be systematicallyinvestigated in Ibc future.

4. The tanperltlu'c, <COV and (H20) all show a "kink" inthm profilei _ the peaks. The cause of this problemis unclear, though lhe kink is leIS severe than wheneqUilibrium chemistry was alm used for (04) in someprelimhwy calculations not n:ported here. &cause italocation i, where oXYllen concenU'lltion practically

gael to zero, inaccutacy in the '" (0)> cubic root at suchpoint may .,Wi be t1i~ ca1pri1.

S. Examination of the 02 and CH4 prois1es shows that!here is a region wh~re utlooml;>oated Oz and CIt4~ whidlmay be interpreted al the fhme woe.The thicknell of thiJ region is small, however, (leiSdwl 0.2 <:!Ii}, which 'STee, with the c;tperimcnullyobserved thil:knen of the blue reaction zone. ThisllXplainIlhe tdaUve IUecca. of earlicr compurations allusing Bllrke and Scbumlll1ln'sS llirnple flame-sheetCODCepl where the reaction between fUd and air isinf'anitely fan and one-way. The mutua! fuel/oxidizerpenetration would nOI be negligible at elevatedpressures Of if fuels lba1 oxidize more slOWly were used,or if turbulence disrupts the diffusion flamc front. Insuch cales thc need for a realistic set of chemicalreactions funy coupled with fluid dynamics, asattanpted ~, will be mon: pronCltltlOed.

Figure 8 repeats some of !he cummtly computed resultsIIJld data IIJld COUIpares them with the theoretical values

complUcd by Mitdlell7 et. al. Except for tempcratuRl,the cunmt reaulb ate aenen1ty superior. CO and HZWeN nol included at ill in Miu:hcJl'. model, and hia(CH4) ".In~ at tbCl centerline &fll much too high. Hismodified flame-sbeClt model allowl zero mutualpenetralion of tbCl C1f4 and~ speci.ClI.

6. Fin.Ily, in Fig. 9, sevClw CDIltoun: Q( thCl effedivenoichiDlDCltric oxygeJI coefficient IIRI plotted. The~IallyobIerved lmninou. flame~ closed 00.

the am at S.8 em dlo\'e the bumcr plate. According toRB!.7, Ws cons repreRDIIthe imlor edgCl (cUclea) ofthe blue reacUOI1 ZOI1tI UIOciI~ with bumoIlt of CO,wbola lll11al' edge it ~tedby Iqu&Ia. Tha but fitfor the citdoI il the COOIOllt with I value « 2.2, and tbcsquaIM~r 1.3. 'IbiJ coalrllt wil:h 1M p~etcd

valuel of 1.76 and 2.0 ill Ref. 7. SiIIc:e ZOO it theIDeoretie4i nina Jor eompieus IWiCIUClltDtluicCOI:IIbtquon it indeed should corrtlq)OIld to the bumOlltof CO nil the O\11llf wae of the blue zoue.Conobontioo. with JIii. 7 showl !hal the cunent highervalue it an inunediau: reault of averprediClCd ruua ofco and H20, both of which in nun can be uaeed to anover pmiidiOl1 of H2 ill the low (Oz) I'Ilgioo. ThUI,impmvemenu in mu maune. of the oxygen cubictqIJaliQn is c&Iled for.

CONCLUSIONS

The benebmaJk test _ bal COilfinned lhat Ille iimplifiedHIO/C ccmbtlltion model i. indeed capable of good .pede.conamtralioa predictions. For furtbet tdinement, t1tn:c .RIIIprelenl themselves for invesliptlDrt:

1. M~.ophinicated fonnul.. fot the eficeUVCl Sdunidtand Prllu!il num~ to caplnRl the V~I di.ft'lUiOl1~ for diffcmnt specie..

2. A swiwh of van.blCl for the energy conlervationeqaatiOD, from sCilsible energy to lIbwolute illteDlalenergy to inc:MUe aceutacy of temperature prediction.,lind po"ibly alternative tlUtment of thll heatoonduclion term.

3. Refine tha root selecbOl1 procell for the oxygen cubictqUlltioo II low (02) leveb to enJUte .. mOQlhIll' (Oz)profile.

10 aMilion, foluI'Il .,;;:lj,vi.lieJ will' abo inch1de application of themodel to tuJbuleot comblution flow.,

ACKNOWLEDGEMENT

1'hll ~tbcn wiIh II) Ihfilk Dr. P. hJI.cCom.au&bcY and Mr. K.Tuclter at MSFC for their continued mpport md direcdoo of theGALACSY codtl deve1o(llnenl effQft. Memben of R.ockeulynet'Advanced Canburtioo. DevialI am also acknowledged for theirsupport and tcclmiea1 input.

REFERENCES

I. Chan. D.C., Hadid, A.H., and. Sindir, M.M., ~On theDeveklpmenl of • ~olds.Averaged Navicr.oStokCll Solverfor Turbomachinery," Proc. 2nd. UII. Symp. on TransportPhenomena, Dynamics and Design of Rolatillg Machinery,Apr. (1988).

6

2, Liang, P.Y, and. Chan, D.C., "Deve1op11lent of a RobustPressure-Bued Numerical Scheme for Spray CombustionApptieuilllll," AIAA·93..0902, (1993).

3. Hi.rt, CW. and Nichols. B.D., "Volume of Fluid (VOP)Method for the Dynlllllics of Free BQW1dariCll," 1. Comp.PIty., vol. 39. no. I, (198J), pp. 201-225.

4. Patankar, S.V. and Spalding, D.B., "A Calculation ProcedUI'llfor Hellt, Mil. and Momentum Tl'lUufer in ThreCl­Dimendoa.al Parabolic Flows," Int. I, Heat Mus Transfer,voL IS. (1912), Po 1787.

5. IJ.aaS, P. and UngewilUlt, Ro, "MuIti·Phue Simulationl ofCoWal IDjector CoolbustiOll," AIAA.92-0345, Jan. (1992).

6. Cloutmm, 1..0., Dutowicz, L.K.. Ramlhllw, 1.0., andAmsdml. A.A~ ·C'ONCHAS-sPRAY: A Computer Code (orRe&dive Flow. wUb Pocl Spray: LA-9294-MS Loa A1ll11lOlNit. Lab., May (1982).

7, Mitebc11, R.n., S.rofim, A.F., and Clomburg, L.A.•"Experimental and Numerical Invesuguion of ComlIledLanUnar Diffusion Flamel: CQIll!>ustion and Flame,Vol, 37, pp. 227-244, (1980),

8, Bulte. S.P. an6 SchumllDIl, T.IlW., Indust. EnS. Chern .•voL 29, (1928), p. 998.

I

'-----""

KEY SUBBQ!J11NES

MODINP

CALCUV

CALCP

OUTBC

CALeF

CALCENL

ATOMIZPTRACK

EVAPPFlND

CALCSPCHEMKNCALCEQCHEMEQOUTBC

CALCSC {ITE}CALCSC (lED)

MODVIS

CALCSC (tEN)TEMPER

MODPRO

PRINT

Fig 1. Flow chart ofGALACSY algoritlun.

'\

Exit ImperviouSBoun~ary_ B.OUnda!y)

_-.'V ... ..... __

THEORETICAL BURNER

~ Iso- "-Thermal:BurnerPlate"

't.l tAIl-A!1 I. l I \ 1Pdf Fuel I Pdr

rL! Ro

l

A

Stainless _SteelWooI ---...;,,;,,;,;;;;;

Sil'iCOl1e ORliquid S9aJ

Perforated .~'~~~~ffim...-;;tm7;-;~-­BrassBumerPlal.e

40Mesh

CylindricalPyrex Shield

A1tInlet

Fuelfnklt .. '

Fig. 2. Schematic ofmethane burner.

""-..-/'

( ( (

t ...........I ... _......t· •.{ ....I. _ ..I.I •.., ,, .,...,.,.

I:': :........::Hl..... ... ... .

I:: ::J~~~1~:L

,,,:I,~,••;:j~~ :

... ,.f

..... +1••• ·t• .. • ~ I· .,· .,•• • 01:.. ,· .,

:-~.:

.. , - .. 1

I q • "lutHUHlt,u tlk t f- I

•• "'IIIHtllllll'u" •.." "'\llIlItUlIlJlto, ••... .,,"ll tUIII •••.••• ,111\ UIt, •••.. • '111 tltr ..... -n' II" ... • HI I" ... ..., " ... . ~ ••It .... ' '. ', ...... , I, ..... .• i I .

:::::[1 /1:::00~ ...~ .. t .t,. • • "... w.1 f ••• _... _.11 { ••••....... 'II , ...

t ••••

••

Temperature. ,r.. 300B 450C 600D 750IE 900r 1050G 12001M 1350I 1500

J 1650K 1800L ~950M 2100N 2250o 2400

(a) Temperature contouf$ (OK) (b) Ve.locity ...ector plots

Fig. 3. Computational results of temperature and velocity.

• I •

CH4 Molal' Concentration

~ 0.00S 0.05C 0.10o O.ISE IUO

.F 0.25G 0.30f.l 0.351 0.40J 0.45K o.SaL 0.55Id 0.5QN 0.65o 0.70p 0.75Q O.BOR OJ!5S 0.90r 0.95I) 1.00

fa)

COZ Molar Concent.ration

A 0.00e 0.01C; 0.02o 0.03E 0.04F 0.05C 0.06I'i 0.01I 0.08J 0.09

I

~

H20 Molar COIlI:~ntraLlon

1\ 0.008 O.O~C O.()4o o.OlSe; O.OBF 0.10G 0.12H 0.14I 0.18

J O.IBK 0.20t. 0.:22

(b)

.. __.-_.--._------

CO MohH' Coni:l:nlrelion

~ 0,00El 0.01C 0.02o O.OJf.: 0.04'" 0.05C 0,06H 0,01I 0.08

J 0.09

(

(c) (d)

Fig.4 Computational results of species mo~ar concentra,tV-" contours of (a) CH4, (b) H2O, (I;) C02 and (d) CO.

\... (

(

SOLID EXPEFlJM

UNE N\,jMEFl.lCAL

lamJ6ralUrfl

0.2 0.4 0.6 0.8 •'1.0 1.2 1.4 1.6 1.8OlstaflCQ from SymOle1ricfA:xls {em)

200

(

~ 180g.~ 160B~ 140~ .~~ 120t­o 100~E 80

Sit' 60g,~ 40>

.20

00.0

..

SCUD EXPERIMENT

UN!: NUMERICAL1

SO ' , I " I I,! lb.' 10.0 0.2 0.4 0.6 0.8 1.0 U! 1.4 1.6 1.8

Dlstanca from Symmetric A:ds {em}

80

'[ 70Ma..(IJ

15~6Qco

~C~ 50c::o

Q

2'g 400­

W

"

SOUD EXPERIllv\ENT

Uto!E NUMER1GJIL

1.8.

SOUD EXPERIMeNT

LINE NUMERICAl..

0.2 0.4 0.6 O.B 1.0 1.2 1.4' 1.6Dlslance Iram Synmetrle Axis (em)

11

10

~ 9<l:le<l:l 8a....9!a 7~c 60',g

r! 51:IIIg 400

'" 3-88. 2en

1"

30 .. i C[ " J j r I , i i & j ,

28:::;- 26 ~ .. -\c~ 24£ 22

~20

e18l5 16

i 14c8 12g 10 ~gld"'"o~ 8'~ 6

~ 4r ;;:;~::::·d~:~0.0 0.2 ~0.6 0.81.0 1.2 1.4 1.6 1.8

Djstanca from Symmetric ,Axis (em)

Fig. 5. Comparison of numerical and experimental concentrations, temperature and velocity profiles at 1.2 em above the burner plate:(solid: ex.perirnental~ open: numerical).

SOLIO EllPERJMENT

lINE NUMERICAl. \,

SOUo EXPERIMENT

U«l: t-<l,lME.RLCJiJ..

TiIi1lJ*a\ln

0./1 1.0 1.2 1.4 1.6 .1.8

co

o.e

20

Q{I,a 0.2 0.4

11

10

I 9

l. 8.!!!c 7ec 6

L f • •~ 48 • •;I 3

1 2 1 •1

240 i • I220 i' i ,. I , I ' ,

]200

f~:F!. ·'iI~ 14€l •.i 12D A~ ..

~ 100 \::: z -~.dIi 80

~ 60E

{!!. 40

1.8

•

SOUD EXPERIMENt'

UHE N1UMERICItl.

$QUD EXPEFIIMEttT

LINE NUI.lS'!lCAL

{ 70l:!3!.'"o~60c

~~ 50<>t:

<3~

140en

2

80 I I • i • I -'~ 0'00 b 01

30 L..--.<-"--o.._

0.0 0.2 0.4 O.a a.sl 1.0 1.2 1.4 1.6

: f ii' r i ·.~6:6 '8 ti ~

~ 16~&t 16qsoaJ::jRo~~!14g 12 r •~ 10 I:. •c'"g 8!- •8.~6t* •! 4

~MI~~efo-~e '0 tJ 0---' 0 I , " , I • I ct813G 0 0 00 0 a o ...a--a--e--a---J" .4 -0.6- O.B 1.0 1.2 UA. 1.6 1.8 0.0 0.2 0.4 o.e O.S 1.0 1.2 1.4 1.6 1.8

Distance from Symmetnc A;ills (em) . DIstance from Bymmelrlc Axis tern),Fig. 6. Comparison of numerical and experimental concentration, temperature and velocity profiles at 2.4 em above the burner plate:

(solid: experimental; open: numerical).

( ("-

(

( ( (

240

II .. l!20

g 70 I 200

~ -- 1S0

~ f 160o ~

6 60 ::. 140 t '" ~ ~Lll) EXPERIMENT65·1! 6" 120 A. lItle-NUMERICAL

] 50 ~ 100~ ,~~~ §~

, 1ii I ~ ,. ' t· •• ,•~ UNE UUMSFlICAL i eoi 4(1 ~ 40m ~

2DI'-/ ..,30 L .t. J ....... ...t. _' "" I • 1 • I 0 ............... __ , :......

{I.D 0.2 OA 0.6 0.8 1.0 1.2 1.4 1.S 1.S 0.0 0.2 0.4 O.S o.a 1.0 1.2: 1.4" 1.5 1.3

·22 \ • , 11

20 1~

~ 18 --I 9

~ 15 II "'-"0. a~ 14 ~ 7e 6.§ 12 sotJDEXPERlMEm" 8 EI

g I r \ \ SOLID EXPERlMEtITc 10 UNE NUMEflICI\L 5g 8 4 ~ UNENU~EflIC,IJ.8 8

! ~ 8 • 3

14 12

Gw E 0

2 1 •••°0.0 0.4 0.6 G.a 1.0 1.2 1.4 1.6 1.8 °0.0 0.2 oX 0.6--0:8--1:0--1-:2- -1:4 - 1.6- -1.8

Ol:;tanc9 from 8ymmafrlc Axis (em) Dlstance1rom Symmelrlc Aids ~cml

Fig. 7. Comparison ofnumerical and e:xperimental concentration. temperatur~ anq vel~ity pro:ijles at 5.0 em above the burner plate:(solid: experimental; open: numerical).

CH4

...-'

•---

•

SYMBOl.. EXPI:RIMENTDASHED UNe·MITCHaLSCUD UNE-GALACSY .

02

•

.Fig. 8. Selected reactant !lmd prodUict species profile comparisons at 1.2 em between GALACSY1

mood of Mitchell et aI, and data.

( ('

",

(,

"(

(em)LO r

NumericalStoichIometricCoefficienll.s

6.0 I 0 Edlge of Exp.Luminou:s FlameSurface

5.0

o Edge of Exp.Blue ReactionZone

4.0

3.0

-2.0 ~1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 (em)0.0 l ' J 11.:.[\~~~'~~~I-:_..t..l!.r;Ll-_--l.-• LLL:.!! I

_'__ J!

Fig. 9. Laminar flame shape co~parisons.