CFD Prediction of Heat Tranlf.r to a Steel Beam under Ceiling

Transcription

1 CFD Prdictin f Hat Tranlf.r t a Stl Bam undr Ciling TAKASH WAKAMATSU nstitut f Cnstructin Tchnlgy. Kurnag., uum. C Onigakub 43. Tsukuba. barak,. Japan YUJ HASEM Dpartmnt f Architctur and Arch,tllctural Eng'nflr,ng Schl f SCinc and Enginring Wasda Univrsity 3-4- Okub. ShinJuku. Tky. Japan ALEXANDER. V. PTCHELNTSEV VTT Building Tchnlgy Kivimihnti 4. Esp. PO.Bx 803. fin VTT. Finland YOSHHKO HAYASH Building Rsarch nstitut. Ministry f Cnstructin Tathara. Tukuba-City, baraki-prf. ABSTRACT A prdictin f th thrmal rspns f a stl bam installd bnath a ciling and xpsd t a lcalizd fir surc was prfrmd by using CFD mdl. Th validity f stady stat calculatin has bn vrifid by rsults f xprimnts. Cncrning th hat flux, th diffrnc btwn analytical rsults and xprimntal valus incrass with incrasing dimnsinlss hat rlas rat Q* D, Q/ pcptg 2 DHB 3/2 n th cas f th lwr flang, th diffrnc btwn th analytical rsult and xprimntal valu incrass rapidly vr th rang f Q*DHB frm 5 t 0.5, raching th maximum valu f abut 22 (kw/m 2 ). This diffrnc is quivalnt t 56% f th maximum hat flux in th xprimnt. Whil th rrr fr th hat flux scattrd within 20% t 56% f th maximum hat flux masurd in th xprimnt, th maximum rrr fr th tmpratur rmaind 22%. KEYWORDS: lcalizd fir,cfd mdl, cmprssibl,prdictin,thrmal rspns, stl bam NTORODUCTON n Japan,fir rsistanc tsts fr building cmpnnts gnrally valuat with th tmpratur assuming xpsur f th cmpnnt t a fully dvlpd fir by th Building Standards. Givn a sufficintly wid spac in a building r a sufficintly wid pning fr th spac cmparing with th fir lad,fr xampl atriums,airprt, parking buildings tc. any fir wuld b f ful-cntrlld typ and th ffct f hating n structural mmbrs culd b lcalizd. t may b assumd that th ris in tmpratur is smallr than in th vnt f all th Cpyright ntrnatinal Assciatin fr Fir Safty Scinc 553

2 l'llhl'r hing uhj<:t t hating hy fir. Whn a mtal tructural mmhr S hatd nly l<:ally in a fir. th tmpratur f th mmbr will b unifrmd rapidly du t th a<:<:lcratd hat cnductin thrugh th mmbr itslf. This is du t th thrmal cnductivity f mtal bing much highr than that f thr cnstructin matrials. n such circumstancs,th lcalizd tmpratur ris may b rstrictd. n th cas whr a lad baring mmbr is hatd nly lcally in fir,if a nw mthd is stablishd which can accuratly prdict th thrmal rspns f th cmpnnts, th fir safty dsign will bcm mr ratinal. Nw, th authrs hav assumd a typical cas f lcalizd fir, in which th H sctin stl bam installd bnath th ciling and xpsd t lcalizd hating with a hat surc n th flr, and hating charactristics f th mmbr wr masurd thrugh xprimnt l ). Subsquntly, w hav frmulatd hat flux distributin n vry part f th bam as a functin f hat rlas rat and th distanc frm th fir surc t th mmbr. Thn w mad a FEM and F.D.M-basd numrical calculatin f tmpratur rspnss, and ths numrical mdls wr vrifid fr its validity by cmparing th numrical rsults f tmpratur with ths btaind thrugh th xprimnt 2).3). Th cncpt f applying CFD mdl dvlpd in th fild f fluid dynamics t study f fir safty scinc has xistd sinc arly tims. Hwvr, it facs many prblms t nsur practical accuracy fr th analysis. Mst f ths prblms ar attributd t th fact that fir is a cmplicatd physical phnmnn invlving cmbustin. A calculatin mdl which can trat prprly th ffcts f cmbustin and radiatin, if availabl, wuld hlp dvlpmnt prmising mthds. Rcntly, Yshi t al cnductd a fundamntal study, basd n th fild mdl, t cmpar analytical rsults and an xprimntal frmula cncrning th spd and tmpratur f thrmal plum frm th hat surc 4).5).6). Th rsults f this study shwd that th mdl was abl t prdict th flw fild f this thrmal plum with practical accuracy. Tmpratur prdictin f mmbrs accrding t FE.M. and FD.M dscribd abv rquird th hat flux as a bundary cnditin 2 ),J).Th hat flux data wr btaind thrugh tim-cnsuming, xpnsiv xprimnl. At prsnt, facilitis and prfssinals fr such xprimnts ar limitd. Accrdingly, if th hat flux itslf can b calculatd nt thrugh xpr;mnts, but thrugh simulatins, it will nt b ncssary t prfnn xprimnts ach tim th shaps, tc., fmmbrs chang. n viw f th abv mntind pint, this papr studid th apprpriatnss f th analytical mthd f hating cnditins (hat flux and tcmpratur) f mmbrs xpsd t lcalizd fir using th fild mdl. Th final purps f this study is t prdict th hating cnditin f th mmbrs by mans f calculatin nly, withut xprimnts. fild S ld SOFE. and SO'lE hll th rullljwln'l"ll\ ''. llllllft"s: Eddy brak-up <:mbustin mdl Thr-dimnsinal finit vlum mlhd,,- turbulnc mdl wilh buyllncy mdl"luna n additin t th abv. this pmgnunlll'llhlnnll""rbllcs0' svl'ml rprsntativ malrials strd as data fils and can us lhm cnllidrln,lmpr.lurc-dpcndncy. NUMERCAL SMULATON Alysis is mad f a hal transfr fild in which n hhllpcd sll ham is lcatd undr a clhg, and t.h cntr f. th bam is hllld hy a nl\',un:t' n h nr. POf this analysis, cnsldnng ts symmtflcal frm f lh spcimn, w rrcpllrd /4 thr dimnsinal thrmal analysis mdl rprsnting th ciling and ham. Pnt RE. shws an analytical mdl. An actual fir surc usd in th xprimnl was ().(m)dllllntr rund prus burnr. Smc a rctangular crdinat systm was usd in this calculalin, lh prtin shwn m th figur was usd as a mflw bundary surfac s thallh lla WLS ljuivalnt t /4 f th rund burnr ara. Fr calculatin, a slutin dmain f 2. (x) X 3.0 (y) X 2.3 (.)(m) was dividd int a ttal f 4535 grid clls, ach bing 9 X 7 X 45. As bundary wndilins. th whl f th tp surfac at a hight f (m)in th y dirctin frm th ciling slab was dlincd as a statc-prssur bundary. Th vrtical sid-wall which runs paralll t th bam axis was als dfind as a static-prssur bundary, whilst th axial nd wall was dfind as a slid. Th prssur bundaris/walls wr lcatd at a (hrizntal) distanc f n (m) frm th dg f th prlt clmg slab. Th grid clls shwn in FGURE. ar prtin usd t dfin th wall as a bundary cnditin. As shwn in FGURE. 2, th vrtical sctin f th H sh.apd bam mmbr was rprsntd using six clls vrtically and tw hrizntally. Sinc a mllllmum f tw clls must b usd in any dirctin whn a slid is dfind, th wb and flang wr dividd int tw clls in a vrtical dirctin. Sinc a unifrm grid spacing was assumd, ths gtvs a msh whch S a pr match t th tru gmtry, but this was invitabl du t th rlativly narrw stl thicknss cmpard t th vrall dimnsins f th tst rig. OUTLNE OF SOFE SOFE (Simulatin Of Firs n Enclsurs) is usd as a calculatin cd 7 ). SOFE is a fild mdl fr th prdictin f firs in cmpartmnts. Th SOFE cd aims t rprduc th cr faturs f currnt cmmrcially-availabl, gnral-purps CFD cds. SOFE dvlpd mainly by Cranfild Univrsity f England, Lund Univrsity f Swdn, VTT f Finland, and CSTB f Franc. Th rsults f xprimnt which assumd rm fir and analytical rsults using this mdl agrd wll 8). Th gvrning quatins f apprximativ cmprssibl flw which assum stady prssur FGURE. Cmputatinal msh

3 OlT.NJ: Oi ('AU'tll.ATON Fr this study, SOFlE(vr-s2070l,standard k- Symmtrical bundary surfac turhulnc mdl ) was mplyd with buyancy mdificatins, and with ptinal Rdi cntr-lin crrctins 9 ). Th ddy brak-up sub-mdl was applid fr cmbustin, with prpan bing takn as th ful. Th prssur crrctin algrithm SMPLEC schm is usd tgthr with mmntum intrplatin fr prssur smthing in th nnstaggrd grid. Th nn-staggrd grid dfins varius quantitis at th sam pint in th cntrl vlum. FGURE. 2 Divisin f bam mmbr int Thrmal radiatin is cmputd using a dtrministic ray-tracing apprach DTRMlOlbasd upn th discrt transfr algrithm. 'Wightd sum f gry gas' WSGG slutins") t th radiatin transfr quatin ar availabl, tgthr with a fully-cupld st-radiatin mdl. Th CGSTAB cnjugat-gradint slvr was usd fr th prssur crrctin quatin in all cass. Th HYBRD intrplatin schm was usd fr all slvd variabls xcpt vlcitis, fr which th SOUP (Scnd-Ordr UPwind) schm was chsnl 2l. Th rdinary stl was dfind fr bam, prlit fr th ciling slab, and th high-dnsity cncrt fr th flr. Th prprtis fr th stl and prlit wr drivd frm xprimntal masurmnts and xprssd as 4 th -rdr plynmials f tmpratur as fllws: Cp,sll = '-486t +297' Cp,prlil= ' ' "stl = '+47.22'-.68' "prlit=3w.8834t+.93u-.9«i!.+f..l.722f.' whr; t = TlOOO... () Th matrial dnsitis wr assumd t b as fllws: 78 kg/m 3 cnstant fr stl, 789 kg/m 3 cnstant fr prlit, 2800 kg/m' cnstant fr high-dnsity cncrt. Th missivity f all slid surfacs was assumd t b 0.9. Th inflw bundary cnditins wr st as K =0.005, =0.00 m 2 /s' rfrrd t S.Wlch and A. Ptchlintsv's rsarch 3) in which th thrmal rspns f a stl bam xpsd t lcalizd fir was analyzd. Minimum rsidual (mass rrr)was st t ', and th xtrnal air tmpratur was st t 290 (k). Th tim fr th calculatin was abut 900 scnds pr 0 itratins with EWS f Dc a 0 MHz CPU, SENSTVTY STUDY n th invstigatins fr simulating th thrmal bhavir f a bam xpsd t a lcalizd fir,s. Wlch and A. Ptchlintsv has xamind th influnc f diffrnt grid rslutins, us f th WSGG radiatin mdl and diffrnt numbrs frays in th discrt transfr mdl. (2) (3) (4) RESULTS AND DSCUSSON SENSTVTY STUDY,, d -. _.._-u_.-t---_..-.n"... '- FGURES.47shw cmpari;n btwn th prdictd and xprimntal valus f hat flux and tmpratur distributin n th lwr flang FGURE. 3 Numbr fray vclrs dfind by and wb with Q=lOO(kW),HB=(m). n th cas and f32 divisins,th prdictd hat flux and tmpratur distributin ar cnsidrably lwr than th xprimntal valus. Th maximum rrr f hat flux t th lwr flang at th stagnatin pint is 6 (kw/m 2 ). n th cas f 64 rays, th maximum rrrs f hat flux t th lwr flang is abut 7 (kw/m 2 ) at a pint O.3(m) frm th stagnatin pint. Th analytical rsults f th tmpratur distributin n th lwr flang and wb surfac agrd wll with th xprimntal valus. FGURES.8 shws th analytical rsults and xprimntal valus cncrning th hat flux and tmpratur distributin n th lwr flang and wb with Q= (kw),hb=(m). n ths FGURES, th black-triangl pints rprsnt th calculatin rsults f 6 rays and th whit-triangl pints shw 32 rays, Fr bth cass, th prdictd rsult f th hat flux and tmpratur f th lwr flang was highr than th xprimntal valus at nar frm th stagnatin pint, with th diffrnc bing 7 (kw/m ) and ROt: fr 6 divisins and 4 (kw/m 2 ) and 70"( fr 32 rays. As fr th hat flux and tmpratur distrihutin f th wb surfac, th analytical rsults rughly agrd with th xprimntal valus. Th calculatin rsults and masurd valus f hat flux distributin t th lwr flang with Q=200 (kw) and HB=(m) ar shwn in FGURE 2. n this figur, th numbr f ray vctrs ar changd 6, 32, and 64. Frm this graph, w can knw that th analytical rsult with Q=200 (kw) and <>,.9,.. fllfi

4 _ \( m) did shw suhstantial dtriuratiun f analytical accuracy. accrding lu th numbr uf rays, and il agrd fairly wll with th xprimntal valus. t is diffrnt frm th cas with Q=OO (kw) and HB=(m). As a whl, th calculatd hat flux indicatd lwr than th xprimntal valus fr all pints xcpt fr th stagnatin pint. Whn cmpard in trms f th numbr f rays, th rrrs f hat flux t nar frm th stagnatin pint is largr with dcrasing numbr f ray vctrs. FGURE.3 shws a cmparisn in trms f th distributin f hat flux t th wb surfac whn Q =200 (kw) and HB =(m). Th analytical rsults fr hat flux agrs rughly wll with th xprimntal valus and thy ar lss affctd by th numbr f ray vctrs. As dscribd abv, it shuld b ntd that analytical accuracy may dtrirat gratly unlss a larg numbr f rays ar st undr analytical cnditins with th hat gnratin rat as small as Q = (kw). Sinc th diffrnc btwn cass with 32 and 64 rays is small, hwvr, th analytical rsult with a ttal f 32 rays( B=4 and =8)was usd subsquntly fr cmparisn btwn analytical rsults and xprimntal valus in rdr t shrtn th calculatin tim. COMPARSON OF CALCULATONS AND EXPERMENTAL RESULTS n this study,data f xprimnts and prdictins ar cmpard at stady stat. FGURES. 4 and 5 shw tmpratur distributin alng th bam whn Q =200 (kw) and HB=(m). Th analytical rsults and xprimntal valus agr cmparativly wll in bth th lwr flang and wb. As fr th lwr flang, xcpt th stagnatin pint, th rgin nar frm th stagnatin pint (r=0.9m), th tmpratur f th lwr flang wr highr than thr rgin, with th diffrnc bing a maximum f looc). FGURES.l6 thrugh 9 shw th tmpratur and hat flux distributin with Q=95 (kw). FGURES. 20 thrugh 23 shw th sam cmparisn with Q=30 (kw), whil FGURES.24 thrugh 27 shw th sam cmparisn with Q = (kw). n all cass, HB was 0.6 (m). During analysis, calculatin was mad undr cnditins cmpltly similar t th cas in which HB was.0 (m). Whn cmpard with th analytical rsult with HB=.0 (m), that with HB = 0.6(m) shwd larg diffrnc frm th xprimntal valus. With HB= 0.6 m, th calculatd rsults f hat flux wr lwr than xprimntal valus rgardlss f th hat rlas rat. A larg diffrnc was bsrvd within a distanc f 0.6(m)frm th stagnatin pint. Undr xprimntal cnditin f Q= 30 (kw), th diffrnc fr th lwr flang was a maximum f 8 (kw/m 2 ) at a pint O.l5(m) frm th stagnatin pint. n all xprimntal cnditin f HB=O.6 (m), th tmpratur dcras alng th axial dirctin is mr slwly in th analytical rsults than in xprimntal valus. Th xprimnt hav shwn that th hat flux t th dwnward surfac f th lwr flang was cntrlld primarily by th flam lngth which flws alng th lwr surfac f th bam ). Th flam lngth can b rprsntd as a functin f Q*OHB. Q * DHB Q / pcptg l / 2 DHB 3 / 2... (5) As is knwn frm th xprimnt that th incras f th flam lngth bcms insignificant as (Q*OHB) is incrasd in th dmain f (Q*OHB»5, and th incras fflam lngth with th incras in th hat rlas rat Q was nt rmarkabl in th cas f HB=O.6(m). This was diffrnt frm th cas with HB=.0(m). Th diffrnc f th analytical accuracy is dpnding n th hight btwn th burnr surfac and bttm f th bam(hb). t may b attributd mainly ht' "ulurt',hw lh ahv actual phnmnn in h analyticalllldcl. rgures.28 shws th rlatinship btwn analytical rrr f hat flux and tht' dimnsinlss hat rlas rat (Q*OHB), assuming a charactristic lngth-scal (HB). Als FGURE.29 shws th rlatinship btwn analytical rrr f th tmpratur and Q*OHB. n FGURE.28, valus n th main axis indicat th maximum diffrnc btwn analytical and xprimntal valus.that n th scnd axis (a right vrtical axis) indicat th rati f rrr fr th maximum hat flux masurd in ach xprimnt. Cmparisn f valus n th main axis f th figur shws that th rrr fr hat flux incrass with incrasing Q*OHB fr bth lwr flang and wb. n th cas f th lwr flang, th rrr fr hat flux is largr undr thr xprimntal cnditins (Q*OHB ranging frm 5 t 0.5), with HB bing 0.6m than undr cnditins with HB bing m. n this cas,th maximum diffrnc is 22 (kw/m 2 ). CONCLUSONS W hav cnductd th prdictin f th thrmal rspns f a stl bam installd bnath a ciling and xpsd t a lcalizd fir surc by using CFD mdl. Frm th rsults f th calculatin. fllwing cnclusins can b drawn. () Th numbr f ray vctrs was changd t 6, 32, and 64 and th ffcts n analytical rsults wr cmpard. Th analytical rsults fr hat flux with 32 ray vctrs undr xprimntal cnditins with Q = (kw) and HB =(m) was cnsidrably lwr, with th diffrnc frm th xprimntal valus at th stagnatin pint bing a maximum f 80%. Undr cnditins with 64 ray vctrs r a highr hat rlas rat, th analytical rsults agrd wll with th xprimntal valus cncrning hat flux distributin. t is knwn frm this fact that, in th cas f th lwr hat rlas rat,th analytical accuracy may dtrirat substantially unlss th numbr f ray vctrs is incrasd. (2) Undr any cnditins f HB=0.6(m), th analytical rsults fr hat flux was lwr than th xprimntal valus. Undr xprimntal cnditins with a hat rlas rat f l(kw), th diffrnc was 27 (kw/m\which is quivalnt t abut 47% fth maximum hat flux in th xprimnt. (3) Cncrning th hat flux, th rrr incrasd with incrasing Q*DHB' n th cas f th lwr flang, th rrr fr hat flux is largr undr thr xprimntal cnditins with HB=0.6m (Q*DHB ranging frm 5 t 0.5)than undr cnditins with HB=lm. n this cas,th maximum diffrnc is 22 (kw/m\ Whil th rrr fr th hat flux scattrd within 20% t 56% f th maximum hat flux masurd in th xprimnt, th maximum rrr fr th tmpratur rmaind 22%. Th analytical accuracy diffrd cnsidrably whn xprimntal cnditin f HB=O.6 (m) wr calculatd using sam analytical cnditins tund fr HB=.0(m). n th cas f HB=.0(m), prdictin is pssibl with a practical accuracy. Whn th distanc btwn th burnr surfac and th bttm f th bam(hb) is varid, hwvr, it will b ncssary t tun calculatin cnditins again. Systmatic studis will b mad n hw t st th calculatin cnditins fr any cmbinatin f distanc(hb) and th hat rlas rats(q). t is ncssary t idntify th caus fr analytical rrr ncuntrd in this study. ACKNOWLEDGMENTS Th authrs wish t thank assistant Prf., Sinsuk Kat ftky Univ. and Dr.Ryuichir Yshi f Mada Crpratin fr th advic f numrical simulatin using CFD mdl. Th authrs ar als indbtd t Mr. Y.Ykbayashi, Skisui Hus Ltd., fr th assistanc in th xprimnts

5 ).Uwl'r FhlnK!' ExpC'rimcnt -a-- Lwr Flang SOFE 0.6,m) 0.9 ray=3... ''''''l'......;;l;,w...,...f,;;la...n::.g.;.;s...o;,;f.;e;;..,;;<a;;;y_=6...4;.. FGURE.4 Effcts f th Numbr f Ray Vctr Q=lOO(kW), H B =.0(m).2 \0 Wt'h Exp('rimf'nt -.-Wb SOF}; 0.9 rny:()... Wb SOFE,"y=:l2 "E 30 f--+-+-l--+-+-l--l-..a :: 20 FGURE.5 Effcts f th Numbr f Ray Vctr Q=(kW), H B =.O(m).2.::; ' 400 l OL--'---'--...L-.L---L_'---..._'_-.J 0.6 lu.2 FGURE. 0 Effcts f th Numbr f Ray VC(;lr Q=(kW), H B =.O(m) -_ :... *' loll M. n n.., flu rm} T " \\,pl.-:""!} Wc b SOFE n)'-::l... W SUF-; r.y-w FGURE. %... " f h Nllllhcr f Ray Vctr Q=()(kW).. -.0(ll) 0 B 400 :: <>- Lwr Flang Exprim ;l... Lwr Flang SOFE ray=32... Lwr Flang SOFE ray=64 >-... l::::::.-, _ FGURE.6 Effcts f th Numbr f Ray Vctr Q=OO(kW), H B =.0(m) -0- Wb Exprimnt L. T l... Wb SOFE,"y=32,... Wb SOFE ray=64 0 B 400.'l Q) : Q) E'< 200 h --.,... rm) FGURE.? Effcts f th Numbr f Ray Vctr Q=OO(kW), H B =.O(m). \... Lwr Flang Exprimnt... Lwr Flang SOFE ray=6... Lwr Flang SOFE my=32... Lwr Flang SOFE ray=64 \\ \ "" _.-- '-... """ >---c '-- FGURE.2 Effcts f th Numbr f Ray Vctr Q=200(kW), H B =.O(m) 0.: ,(m) FGURE.3 Effcts f th Numbr f Ray Vctr Q=200(kW), H B =l.o(m) 40..:l 30 r. 20 :r: Lwr Flangt Exprimnt t--... Lwr Flang SOFE ray=6... Lwr Flang SOFE ray=32 '\ ",-'.. --._. - \ "' >--., FGURE.8 Effcts f th Numbr f Ray Vctr Q=(kW), H B =.O(m) _.,-,--,-,==========;t: -<>-Wb Exprimnt 'jl f--< Wb SOFE ray=6 :: Wb SOFE ray=32 30 f =::::t=:=::t====t=t=:=t=:=t:::= 20 f j Q) :r: O--+","".!.:::-+-+--f FGURE.9 Effcts f th Numbr f Ray V Q=(kW), H B =.O(m) E '0 j.j. L l -<>- Lwr Flang Exprimnt... Lwr Flang SOFE "'-.. '-." "' k k - FGURE.4 Tmpratur distributin (lwr flang) Q=200(kW), H B =l.o(m) S 200 -_. -<>- Wb Exprimnt... Wb SOFE "- '.:: "-.. "' j:", r--....'-. FGURE.l5 Tmpratur distributin (wb) Q=200(kW), H B =.O(m) 5 56

6 _ f " i "" ii L... L Lwr Flang Exprimnt... Lwr Flang SOFE " r-. hd FGURE.6 Tmpratur distributin (lwr flang) Q=95(kW), H B =O.6(m) -r-r-=-.2 0' & mnt J : :::::: r----..,, "" j' FGURE.7 Tmpratur distributin (wb) Q=95(kW), H B =O.6(m).2 -S E;: 20 :r: f\ fj.l_l.l.. -<>- Lwr Fianc Expnmant Lwr Fianc SOFE \ - \ \... \ \ " '-...,...._ FGURE.22 Hat flux distributin (lwr flang) Q=30(kW), H B =O.6(m) j J i Wb ft:xprlmnt -W.bSOFE Q3 QU Q9.2 rm) FGURE.2 lal flux tlislributin (wb) Q= (kw), " =O.6(m) > ; \ U --0- i"\ \... i'l>. \-.,. Lwr Flang Exprimnt... Lwr Flang SOFE "" FGURE.8 Hat flux distributin (lwr flang) Q=95(kW), H B =O.6(m) , 30 r:; 20 :J r- r--... "'-. "- -<>- Wb EXprimnt -Wb SOFE FGURE.9 Hat flux distributin (wb) Q=95(kW), H B =O.6(m) 0 G ' " 400 " Lwr Flang Exprimnt... Lwr Flang SOFE "- i","... t-- h t :>--- FGURE.24 Tmpratur distributin (lwrflang)q=(kw), H B =O.6(m) 0 0' r... WbSOFE r-..., , t-- -:';;:'Wb Exprimnt] FlGURE.25 Tmpratur distributin (wb) Q=(kW), H B =O.6(m) t t"-. r.. L...L. L.. L...L Lwr Flang Exprimnt..., _ Lwr Flang SOFE t j-- r-...._- r- r r--p.. >-- t- FGURE.20 Tmpratur distributin (lwr flang) Q=30(kW), H B =O.6(m),--r-r-r=========r 400 f--f t-+---l--t--r--j E 300 f---+-t'l;--+--t-+---l--t--r--j ±"'''''''=+--j-t--l-" """"",,,k::-+-=,,+ L..!--L-l..---l_l--...l----L----L----'----J FGURE.2 Tmpratur distributin (wb) Q=30(kW), H B =O.6(m) = r;: i 20 :: _J. L -0- J "\ -<>- Lwr Flang SOFE \ 0-.. Lwr Flang "xprimnt \ \ \ \.- ---_. \._--. f'..-- > FGURE.26 Hat flux distributin (lwr flang) Q=(kW), H B =O.6(m). -;::::r=====::;-] -<>- Wb Exprimnt..- 40, 30 t LJ--.l---.l.-l-C:±:!: 0.: 0.0 rm) FGURE.27 Hat flux distributin (wb) Q= (kw), H" =O.()(m)

7 Firs in Tunnls: Thr-Dimnsinal Numrical Simulatin and Cmparl.n with th Exprimnt ALEXEY V. KARPOV, DMTRY V. MAKAROV, VLADMR V MOLKOV and ALEXEY M. RYZHOV Ail-RUSSian Rsarch nstitut fr Fir PrtOCln, VNPO 2. Balashlkha-3, Mscw rgin. 4J900 R.' 0.0 FGURE Q'DHB Rlatin btwn Q" DHB and rrr fhat flux. _ QH b' (-) FGURE.29 Rlatin btwn QHB* and rrr f tmpratur. REFERENCE ) Ykbayashi,Y.,Hasmi,Y.,Wakamatsu, T., Wakamatsu, T., "Exprimntal Study n th Hating Mchanism and Thrmal Rspns f a Stl Bam Undr Ciling Expsd t a Lcalizd Firs," Jurnal f Structural and Cnstructin Enginring, Transactins f AJ, N A98 (in Japans) 2) Wakamatsu,T.,Hasmi,Y., Ykbayashi, Y.,A.V.Ptchlintsv,"Hating Mchanism f Building Cmpnnts Expsd t a Lcalizd Fir - FEM Thrmal and Structural Analysis f a Stl Bam Undr Ciling," Prcdings, OMAE '97 Ykhama 997 3) Wakamatsu, T.,Hasmi, Y., "Hating Mchanism f Building Cmpnnts Expsd t a Lcalizd Fir- FDM Thrmal Analysis f a Stl Bam Undr Ciling -," Fir Scinc and Tchnlgy - Prcdings f th Third Asia-Ocania Sympsium, Singapr pp , ) Hara, T., Yki, M., Mrikawa, Y., " Analysis f Thrmal Plums with Numrical Simulatin and Exprimntal Rsults Basic Plum Analyss fr th CFD Prfrmanc f th Standard k- mdl. Annual Mting," Architctural nstitut f Japan, 998(in Japans) 5) Yshi, R.,WE Ran, " CFD Analysis f Fir Plum - (Partl) nflunc f Cmputatinal Dmain and Msh Divisin n Numrical slutin," Annual Mting, Architctural nstitut f Japan, 998(in Japans) 6) Yshi,R.,and WE Ran "CFD Analysis f Fir Plum - (Part2) nflunc f Numrical schm fr cnvctin trms and mdl fr fir surc," Annual Mting, Architctural nstitut f Japan, 998(in Japans) 7) Wlch, S.,P.Rubini "SOFE Usrs Manual" Schl f Mchanical Enginring, Cranfild Univrsity, 996 8) A.V.Ptchlintsv, M.Niklank, Hayashi, Y., Wakamatsu, T.,"Sfi Cmputatin f th BR Rm Cmr Fir ntrim Rprt", SOFE mting, 998 9) Rdi, W. "Turbulnt buyant jts and plums", Prgamn, 982 ) Brsslff, N.W., Mss, J.B., Rubini, P.A."Assssmnt f a diffrntial ttal absrptivity slutin t th radiativ transfr quatin as applid in th discrt transfr radiatin mdl", Numrical Hat Transfr, Part B: Fundamntals, vl. 29, part 3, pp , May, 996 ) Brsslff, N.W., Mss, J.B., Rubini, P.A. "CFD prdictin f cupld radiatin hat transfr and st prductin in turbulnt flams," Twnty-sixth Sympsium n Cmbustin, Th Cmbustin nstitut, pp , 996 2) "Fir safty dsign mthd, cmprhnsiv fir prvnting prcdurs, d. by Japan nstitut f Cnstructin Enginring," vla, April, 988 (in Japans) 3) S.Wch,A.Ptchlintsv., "Numrical Prdictin f Hat Transfr t a Stl Bam in a Lcalizd Fir," Prcdings f th Scnd ntrnatinal SminarFir-and Explsin Hazard f Substancs and Vnting f Dflagratins,-5 August, 997, Mscw, Russia, pp ABSTRACT Th cmputatinal fluid dynamics cd SOFE has bn usd fr thr-dimnsinal unstady simulatin f xprimntal fir in th tunnl with sizs 2xl.5xl.6 m prfrmd prviusly by Japans rsarchrs. Th stady and unstady burning rat apprachs wr usd fr numrical mdling. t has bn shwn that th cmputd flw vlcitis and tmpratur distributin fr unstady, grwing in tim, burning rat ar in bttr crrspndnc t th xprimntal data. t is shwn that an apprach t th dscriptin fbuming rat influncs significantly n th magnitud and th distributin f hat fluxs in th nclsur, that is imprtant in particular fr slving th fir rsistanc prblm. KEYWORDS Fir, tunnl, unstady CFD mdlling, variabl burning rat, tmpratur, vlsity, hat flux NTRODUCTON Th cmputatinal fluid dynamics (CFD) is mplyd mr and mr ftn fr fir phnmna invstigatin during last dcads. CFD mdling is th ssntial tl fr prfrmanc basd apprach t fir safty nginring. t nabls t raliz rliabl cst-ffctiv fir safty dsign f th prmiss, minimiz xpnss n tsting, study rgularitis f fir phnmna in numrical xprimnts. Th CFD tchniqu f fir mdling is th mst infrmativ bcaus it calculats tim and spac distributin f thrm- and hydrdynamic paramtrs. Advancd CFD cds bcm mr cmplicatd and tak int accunt nw physical aspcts ffir phnmnn. On th n hand it givs nw paramtrs which bing varid can influnc n numrical slutin On th