Evaluation of the air-demand, flame height, and radiation from low-profile flare tips using ISIS-3D

Low-profile flare fields pose significant design challenges including elongated flames, adequate air supply to burner tips located on inner rows and high radiation flux from the flame to the surrounding wind fence. Recent work completed by engineers at Alion Science and Technology for Zeeco, Inc. has focused on analyzing the performance of a proprietary burner tip used in large low profile gas flares having upwards of 400 burner tips packed together into a staged piping configuration surrounded by a specially designed wind fence. This paper presents results from the CFD analysis of this gas flare and illustrates the capability of the CFD tool to simulate soot formation, radiant flux, flame shape, and flame height for industrial scale low-profile flare fields. This work was completed in conjunction with flare testing where ethylene was fired through the burner tips. Data collected during these flare tests included video, radiant flux, and sound. Test results were used to calibrate the combustion model and to validate CFD predictions of flame height and air demand. Based on this, predicted flame height and air demand were provided for two full flare field cases. In addition, estimates of radiant flux to the surrounding wind fence were provided.

American - Japanese Flame Research Committees International Symposium Advances in Combustion Technology: Improving the Environment and Energy Efficiency Marriott Waikoloa, Hawaii - Oct. 22 -24, 2007 E v a lu a tio n o f th e A ir-D e m a n d , F la m e H e ig h t, a n d R a d ia tio n f ro m lo w -p ro file fla r e tip s u s in g IS IS -3 D Joseph D. Smith, Ph.D. and Ahti Suo-Ahttila, Ph.D., Alion Science and Technology, Owasso, Oklahoma, USA Scot Smith and Jay Modi, Zeeco, Inc. Zeeco Inc. Broken Arrow, Oklahoma, USA ABSTRACT L o w -p ro file fla re fie ld s pose s ig n ific a n t design challenges in c lu d in g elo n gated flam es, adequate a ir sup p ly to b u rn er tips lo cated on in n e r row s and h ig h ra d ia tio n flu x fro m the fla m e to the surrounding w in d fen c e. R ec en t w o rk c o m p leted b y engineers at A lio n S cien ce and T ec h n o lo g y fo r Z e e c o , In c . has focused on a n a ly zin g the p erfo rm a n ce o f a p ro p rie ta ry b u rn er tip used in la rg e lo w p ro file gas flares h av in g u pw ards o f 4 0 0 b u rn er tips p acked tog eth er in to a staged p ip in g c o n fig u ra tio n surrounded b y a s p e c ia lly designed w in d fence. T h is p aper presents results fro m the C F D analysis o f this gas fla re and illu strates the c a p a b ility o f the C F D to o l to sim u late soot fo rm a tio n , ra d ian t flu x , fla m e shape, and fla m e h e ig h t fo r in d u s tria l scale lo w p ro file fla re fie ld s . T h is w o rk was c o m p leted in con ju n c tio n w ith fla re testing w h e re e th ylen e was fire d throu g h the b u rn er tips. D a ta c o llec ted d u rin g these fla re tests in c lu d ed v id e o , rad ian t flu x , and sound. T es t results w e re used to c alib rate the com bu stio n m o d e l and to v a lid a te C F D p redictions o f fla m e h e ig h t and a ir dem and. B ased on this, p red icted fla m e h e ig h t and a ir d em an d w e re p ro v id e d fo r tw o fu ll fla re fie ld cases. In a d d itio n , estim ates o f ra d ian t flu x to the surrounding w in d fen ce w e re p ro vid ed . IN T R O D U C T IO N A series o f calcu latio n s o f fla re p erfo rm a n ce h av e been m ade. T h e purpose o f the calcu latio n s was to p re d ict a ir d em an d u n d er vario u s co n d itio n s. In a d d itio n , the th e rm a l ra d ia tio n p ro file around the fla re w as also d eterm in ed . T h e p rim a ry c o m p u tatio n a l flu id dynam ics ( C F D ) to o l used in this analysis was IS IS - 3 D [1 , 2, 3 ]. P re v io u s ly , IS IS - 3 D has been used in a v a rie ty o f p o o l fire analyses to p re d ict package th e rm a l p erfo rm a n ce [4 ]. M o re re ce n tly , IS IS -3 D has been a p p lied to fla re analysis. In this a p p lica tio n , n e w com bu stio n m odels h ave been im p le m e n te d fo r h a n d lin g n e w fu e l m ixtu res in c lu d in g propane and e th ylen e. T h e com bu stio n and ra d iatio n m odels h ave been com pared to fla m e size, shape, and ra d iatio n m easurem ents m easured d u rin g s in g le-b u rn er and m u lti-b u rn e r tests u n d er n o -w in d and lo w -w in d a m b ien t con d itio n s. T h e C F D m o d e l in c lu d ed vario u s details depen d in g upon the case that w as run. F o r a single b u rn er case a c o m p u tatio n a l d o m ain o f 6 m x 6 m x 3 0 m was selected. F o r a m u lti-b u rn e r case a d o m ain size o f 35 m x 35 m x 25 m was selected. F o r the fu ll fie ld the c o m p u tatio n a l d o m ain was exten d ed 10m b ey o n d the w in d fen ce surrounding the e n tire fla re fie ld . Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency T h e m a in o b je c tiv e o f this w o rk was to p re d ict the to ta l a ir dem an d and the expected fla m e h e ig h t fo r tw o o p eratin g cases. In a d d itio n , the ra d ia tio n heat flu x p ro file was p red icted fo r the three b u rn er test, and fo r the fu ll fla re fie ld cases. T h e three b u rn er sim u latio n s w e re com pared to e x p e rim e n ta l m easurem ents o f ra d ia tio n in te n s ity at g round le v e l lo cated 15 m and 3 0 m fro m the b u rn er tip . R esults o f these m easurem ents p ro v id e d a p a rtia l v a lid a tio n o f the o v e ra ll c o m p u tatio n a l m o d el. T h is p aper contains descriptions o f the vario u s s im u latio n s, m o d e lin g assum ptions and m eth o d o lo g y, c o m p u tatio n a l results, and com parisons to e x p e rim e n ta l data. C O M B U S T IO N M O D E L T h e com bu stio n m o d e l in IS IS - 3 D is a h y b rid m o d e l c o m b in in g A rrh e n iu s kin etics and tu rb u len t m ix in g . The kin etics c h aracteristic tim e scales. In and turbu len ce m odels are c o m b in ed by sum m in g the a d d itio n to these d y n a m ic m odels, sequences o f irre v e rs ib le c h e m ic al reactions that describe the com bu stio n ch e m is try are req u ired . T o fa c ilita te an e ffic ie n t and p ra ctica l C F D c a lc u la tio n , a m in im u m n u m b er o f c h e m ic al reactions are used that f u lf ill the requirem ents o f to ta l e n erg y y ie ld and species con su m p tion and p ro d u ctio n . F ro m the basis o f heat tran sfer, fla m e size, and a ir dem an d the d etails o f the c h e m ic al reactions are n o t c ritic a l so lo n g as the o xy g en con su m p tion is c o rre c tly b alan ced fo r a g iv e n fu e l type. T o this end, both tw o -step and three-step c h e m ic a l reactio n m odels fo r the d iffe re n t fu e l types h ave been used. A tw o -step c h e m ic al reactio n is used fo r propane. T h e firs t reactio n considered propane plus o xyg en w h ic h reacts to produce w a te r, carbon d io x id e and soot. T h e soot y ie ld fra c tio n was assum ed to be a constant fra c tio n o f fu e l con su m p tion and fix e d at 2 .4 % as rep o rted in the S F P E m a n u a l [5 ]. T h e second reaction consum es soot p roduced in the firs t reaction . T h e tw o -step reaction approach re q u ired a p ilo t b u rn er at the fla re tip to m a in ta in a fla m e . T h is re q u irem e n t was due to the re la tiv e size o f the c o m p u tatio n a l cells com pared to in d iv id u a l je t diam eters and the h ig h flo w v e lo c ities at the tip . W ith o u t a p ilo t fla m e , the h ig h je t v e lo c itie s cause the fla m e to detach and b lo w out. F o r eth y len e and m ix e d gases, a three-step reactio n is used s im ila r to the one used b y G re in e r fo r JP 8 je t fu e l fires [ 6 ]. In a three-step m ech an ism , the firs t reaction burns h a lf the h yd ro g en c o n tain ed in the h yd ro carb o n , and any fre e hyd ro g en . T h e second re ac tio n burns the re m ain in g h yd ro g en in the hydrocarbons and m ost o f the carbon, w ith som e degree o f soot p ro d u ctio n . T h e th ird and fin a l reactio n burns the soot p roduced in the second reactio n . T h is sequence o f reactions is m o re n u m e ric a lly stable since the firs t reaction has a lo w a c tiv a tio n e n erg y ( ~ 20,000 c a l/m o le ) com pared to the second reactio n , w h ic h has a ty p ic a l h yd ro carb o n com bustion a c tiv a tio n en erg y o f 3 0 ,0 0 0 c a l/m o le . T h e lo w a c tiv a tio n en erg y o f the firs t reaction a llo w s p a rtia l com bu stio n at lo w tem peratures, releasin g a p p ro x im a te ly 20 % o f the h yd ro carb o n heat o f com bustion. T h e p a rtia l h eat release keeps the fire b u rn in g w ith o u t using a p ilo t fla m e . F u rth e rm o re , this sequence is closer to that a c tu a lly fo u n d in h yd ro carb o n fire s. A g a in , the s to ic h io m etric details are n o t c ritic a l so lo n g as the fu e l, o xy g en , and soot are consum ed in the correct proportions. P ro p a n e C o m b u s tio n m o d e l F o r propane flam es , a tw o -s te p c h e m ic al reactio n is used that burns propane according to the fo llo w in g fo rm u la : Page 2 of19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency C 3H 8 + 3 .6 O 2 ^ S o o t + 2 .6 6 O 2 ^ 3 C O 2 + 1 .6 H 2O + 0 .0 2 4 S o o t + 4 6 M J /k g p ro p a n e 3 .6 6 C O 2 + 3 2 M J /k g S o o t (1 ) (2 ) T h e c o e ffic ie n ts in the fo rm u la are mass w e ig h ts , n o t m oles. T h e A rrh en iu s kin etic s eq u atio n and param eters fo r these reactions w e re R a te (m o le s /m 3/s ec ) = X C2H8 * X O2 * A E x p ('Ta/T) (3 ) 3 W h e re X is the m o la r co n cen tratio n o f the species (m o le s /m 3), A is the p re -e x p o n e n tia l fa c to r (3 .2 9 E 1 0 - 1st reaction and 8 .0 E 1 1 - 2nd re ac tio n ), Ta is the a c tiv a tio n tem p eratu re ( K ) (1 5 ,9 2 2 and 2 6 ,5 0 0 ), and T is the lo c a l tem p eratu re (K ). T h e ch aracteristic tim e fro m the kin etic s eq u atio n was c o m b in ed w ith the characteristic turbu len ce tim e scale t turb=C A x / £diff (4 ) W h e re A x is the ch aracteristic c e ll size, C is a user in p u t constant (0 .2 E -4 ), £dij f is the edd y d iffu s iv ity fro m the turbu len ce m o d el, and tturb is the turbu len ce tim e scale, i.e . ch aracteristic tim e re q u ired to m ix the contents o f a c o m p u tatio n a l c e ll. T h e re ac tio n rates are c o m b in ed b y s im p le a d d itio n o f the tim e scales E th y le n e C o m b u s tio n M o d e l F o r unsaturated h yd ro carb o n co m bu stio n, the re q u irem e n t o f using e lim in a te d b y im p le m e n tin g a three-step c h e m ic a l reaction . a p ilo t fla m e was U s in g this approach, the eth ylen e c om bustion m o d e l consisted o f the fo llo w in g three step m ech anism : C 2H 4 + 0 .5 7 O 2 ^ 0 .9 3 C 2H 2 + 0 .6 4 H 2O + 9 .4 M J /k g e th y le n e (5 ) C 2H 2 + 2 .5 8 O 2 ^ 2 .7 C O 2 + 0 .7 H 2O + 0 .2 S o o t + 3 4 .1 M J /k g in te r m e d ia te ( 6) S o o t + 2 .6 6 O 2 ^ 3 .6 6 C O 2 + 3 2 M J /k g S o o t (7 ) A s b efo re the c o e ffic ie n ts are mass w eig h ts, n o t m o le w eig h ts. T h e A rrh en iu s kin etic s eq u atio n and param eters fo r these reactions w e re : F ir s t R e a c tio n R a te (m o le s /m 3/sec) = X c 2H4 * X o 2 * 1 . 0 e 1 5 E x p ('10,500/T) ( 8) S e co n d R e a c tio n R a te (m o le s /m 3/s ec ) = X c 2H2 * X 02 * 1 . 0 e 1 1 E x p ('15,500/T) (9 ) T h ir d R e a c tio n R a te (k g /m 3/sec) = Y C * Y 02 *1 .0 e 1 1 E x p ('20,500/T) (1 0 ) W h e re X is a m o le co n cen tratio n (m o le d en sity) and Y is a mass con cen tratio n (p a rtia l mass d en sity). T h e advantage o f the three-step reactio n is that the firs t reactio n has a lo w a c tiv a tio n energ y, w h ic h a llo w s the p a rtia l b u rn in g and heat release o f eth y len e . T h is w ill m a in ta in com bustion since the p a rtia l heat released w ill a llo w the second reactio n , w h ic h produces m ost o f the heat and a ll o f the soot, to occur. A s in the p ropane com bustion m o d e l the e th ylen e A rrh en iu s c om bustion tim e scale is c o m b in ed w ith the turbu len ce tim e scale to y ie ld an o v e ra ll tim e scale fo r the reactio n rate. Page 3 of19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency M ix e d G a s C o m b u s tio n M o d e l A three-step c h e m ic a l reaction fo rm u la tio n was im p le m e n te d fo r a m ix e d gas h av in g an a p p ro x im a te com p o s itio n o f 3 2 % C 2H 4, 20% C 2H 6, and 3 4 % H 2 (m o le p ercen t). A n y re m ain in g gases are ig n o red in the com bu stio n m o d el. T h e s im p lifie d 3-step reactions are 0 .5 7 2 C 2H 4 + 0 .3 8 3 C 2H 6 + 0 .0 4 3 H 2 + 0 .9 8 2 O 2 ^ 0 .5 3 C 2H 2 + 0 .3 4 C 2H 3 + 1 .1 H 2O + 1 4 .2 M J /k g (1 1 ) 0 .6 1 C 2H 2 + 0 .3 9 C 2H 3 + 2 .6 6 O 2 ^ 2 .6 6 C O 2 + 0 .8 1 3 H 2O + 0 .1 8 1 S o o t + 3 4 .4 M J /k g S o o t + 2 .6 6 O 2 ^ 3 .6 6 C O 2 + 3 2 M J /k g (1 2 ) (1 3 ) A s b efo re the c o e ffic ie n ts are mass w eig h ts, n o t m o le w eig h ts. T h e A rrh en iu s kin etics eq u atio n and param eters fo r these reactions w e re F ir s t R e a c tio n R a te (m o le s /m 3/sec) = X fuel * X O2 * 1 e 1 5 E x p (-10500/T) (1 4 ) S e co n d R e a c tio n R a te (m o le s /m 3/s ec ) = Xmix * X o 2 * 1 e 1 2 E x p (-15500/T) (1 5 ) T h ir d R e a c tio n R a te (k g /m 3/sec) = Y C * Y O2 * 1 e 1 1 E x p (-20500/T) (1 6 ) F la r e N o z z le M o d e l T h e fla re burners h ave hundreds s m a ll holes o f vario u s sizes and a lig n ed and d iv id e d on each o f the e ig h t arm s according to the sp ecific tip design. A n S T L file co n tain in g details o f the shape and size o f the fla re arm s and supporting structures was im p o rte d in to IS IS - 3 D to generate an a p p ro x im a te C A D m o d e l o f the burn er. H o w e v e r the thickness o f the arms is on the o rd er o f 1 in ch , w h ic h is b e lo w the reso lu tio n o f the co m p u tatio n a l g rid . A s a resu lt the b u rn er m o d e l is a p p ro x im a te w h ic h is n o t a p ro b le m h o w e v e r because the g rid structure has m in im a l e ffe c t upon the flu id dynam ics around the outside o f the fla re b u rn er tip . T h is m in im a l flo w e ffe c t was established b y e a rlie r calcu latio n s [4 ] that in d ic ated the p rim a ry in flo w o f o xy g en was fro m the sides o f the fla re tip and n o t fro m b e lo w the fla re tip . P o in t sources o f mass, species, and m o m e n tu m w e re used to m o d e l each h o le in each burner. IS IS - 3 D n u m e ric a lly com bines any holes that reside in the same co m p u tatio n a l c e ll in to a single source. E v e ry h o le was in c lu d ed in the sim u latio n s so that the m esh structure co u ld be v arie d w ith o u t re q u irin g a separate b u rn er f ile fo r each m esh structure. M o d e lin g e v e ry h o le in e v e ry b u rn er does cause som e a d d itio n a l C P U o verh ead ; h o w e v e r there is less lik e lih o o d o f an error, since o n ly a single in p u t f ile is created fo r a ll the runs. T o fu rth e r reduce the lik e lih o o d o f C A D e rro r that m ig h t le ad to c o m p u tatio n a l errors, a d d itio n a l F O R T R A N program s w e re used to calc u late the flo w area, 3 -D c oo rd inate lo c atio n , and d ire c tio n cosines fo r each h o le in each b u rn er, as w e ll as the mass flo w rate and v e lo c ity o f the fla re gas (i.e ., propane, e th y len e ) as a fu n c tio n o f d riv in g pressure and tem p eratu re. M O D E L I N G A S S U M P T IO N S A N D A P P R O A C H T h e fo llo w in g assum ptions w e re u tiliz e d in m o d e lin g lo w p ro file flares using IS IS -3 D : Page 4 of19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency 1. T h e to ta l a ir dem an d fo r any case was d e te rm in ed b y m o n ito rin g the flo w across a re ctan g u lar p lan e situated at a sp e cified h e ig h t above the ground, and exten d in g 1-3 m eters b eyo n d the edge o f the outerm o st burn er. F o r cases w ith cross w in d , the flo w across several planes was m o n ito red and com pared. 2. T h e n o zzles w e re represented as p o in t sources fo r m o m e n tu m , mass, and species. T h e m o m e n tu m sources in c lu d ed the d ire c tio n a l o rie n ta tio n and flo w v e lo c ity fro m each h o le. 3. T h e flo w v e lo c ity e x itin g each h o le was assum ed to be p ro p o rtio n a l to the square ro o t o f A P / ( / p C ) w h e re A P is the pressure drop o f the tip , C is a loss c o e ffic ie n t o f 0 .8 5 , and p is the fu e l d en sity e va lu ated at the u pstream tem p eratu re pressure and m o le c u la r w e ig h t. 4. F o r sonic co n d itio n s, the previous fo rm u la fo r flo w rate w h ic h u tiliz e d an o rific e c o e ffic ie n t o f 0 .8 5 and the upstream d en sity and pressure w as n o t used because it is in v a lid w h en the flo w becom es sonic. S onic flo w is ach ieved w h e n e v e r the d riv in g pressure exceeds the pressure w h e re the flo w reaches sonic con d itio n s fo r the specific gas. 5. C o m b u s tio n o f the fla re gas was a p p ro x im a te d b y the app ro p riate 2 o r 3 step irre v e rs ib le c h e m ic al reactio n m ech anism w ith sp e cified k in etics. 6 . T h e rm a l ra d ia tio n was c alc u late d using standard ra d ia tio n m odels. R a d ia tio n shadow ing b y m u ltip le flares was ig n o re d in the C F D c alc u la tio n b u t was accounted fo r in a separate post-processing c a lc u la tio n that accounted fo r shad o w in g and absorption effects. 7. A m b ie n t w in d c o n d itio n , fla re gas in le t tem p eratu re and pressure, and ra d ia tio n effects w e re m easured fo r each test and used in the C F D m o d el. 8 . F la m e len g th w as estim ated fro m the lo c atio n w h e re the co n cen tratio n o f in term ed ia te species goes to zero. IS IS -3 D , the C F D to o l used fo r these analyses, is a p ro p rie ta ry co m p u ter code fo r m o d elin g the d yn a m ic b e h a v io r o f fires in flu e n c e d b y a w id e v a rie ty o f p h ys ic al processes and is based on the con servatio n o f mass, m o m e n tu m , and energ y. IS IS -3 D has been successfully u tiliz e d fo r a w id e v a rie ty o f flo w / h eat tran sfer app licatio ns [1 -4 , 5, 6 ]. T h e c o m p u tatio n a l d o m ain was exten d ed through the en tire fla re fie ld . F o r the fu ll fla re fie ld analysis, the w in d fen ce was sim u la te d as a b a ffle w ith the app ro p riate pressure drop to accu rately m o d e l flo w throu g h the fence. T h e m e th o d o lo g y fo llo w e d to d evelo p and a p p ly the C F D to o l to m o d e l lo w p ro file gas flares in c lu d ed the fo llo w in g steps: 1. C a re fu lly re v ie w a ll fla re d raw in g s p ro v id e d and p repare sketches o f the fla re tip and the associated fla re fie ld . D im e n sio n s n o t p ro v id e d w e re scaled fro m d raw in g s p ro vid ed . 2. Set up the g eo m e tric d o m a in and generate the c o m p u tatio n a l m esh using the IS IS -3 D preprocessor. 3. S elect app ro p riate p h ys ic al and n u m e ric a l sub-m odels (e .g ., turbu len ce, pressure solver, energ y, e tc .). 4. P e rfo rm a sin g le b u rn er s im u la tio n to d ete rm in e the expected flo w and tem p eratu re p ro file s around a single b u rn er and com pared results to m easured e x p e rim e n ta l data. 5. P e rfo rm a th ree -b u rn e r s im u la tio n to evalu ate b u rn er-b u rn er spacing and com pare p red icted fla m e shap e/heig h t to m easured e x p e rim e n ta l data. 6 . B ased on m o d e l v a lid a tio n using fla re test data, p e rfo rm a fu ll fla re fie ld analysis fo r a g iv en set o f o p eratin g and a m b ien t con d itio n s. Page 5 of19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency C o m p u ta tio n a l D o m a in T h e c o m p u tatio n a l d o m ain used fo r these analyses (see F ig u re 1) exten d ed several m eters past the edge o f a ll fla re tips. F o r the ra d ia tio n p re d ictio n , the d o m ain exten d ed to a distance o f 35 m in b o th h o riz o n ta l d irectio n s. T h e h e ig h t o f the d o m ain was n o rm a lly taken as 15 m excep t fo r h ig h pressure cases (2 0 and 3 0 psi) w h e re the h e ig h t was exten d ed to 25 m . C o m p u ta tio n a l M e s h E ach m o d e l u tiliz e d d iffe re n t degrees o f m esh re fin e m e n t, w ith a single b u rn er m o d e l h av in g the m ost re fin e d m esh. A less reso lved m esh was used fo r m u lti-tip sim u latio n s to reduce the o v e ra ll c o m p u tatio n a l dem an d re q u ired to o b ta in a con verg ed so lu tio n in a reasonable am o un t o f tim e . A s seen, a ll meshes used w e re based on re ctan g u la r o rth o g o n al c e ll shapes (as opposed to unstructured c e ll shapes). T h e c o m p u tatio n a l meshes fo r the single b u rn er, the trip le b u rn er, and the fu ll fla re fie ld cases are d ep icted b e lo w (see F ig u re 2 ). T h e dense d a rk to b la c k areas is w h e re the co m p u tatio n a l m esh is b eyo n d the reso lv in g c a p a b ility o f the graphics. A s a resu lt the in d iv id u a l lines congeal in to e ith e r a d a rk mass o r M o ire patterns (an in terfe ren c e p attern created w h en tw o grids are o v e rla id at an ang le, o r w h e n th e y h ave s lig h tly d iffe re n t m esh sizes). T h e b u rn er is o n ly d ep icted in the sin g le b u rn er m esh. In the fu ll fie ld cases the burners co u ld n o t be fu lly resolved. F ig u re 1 - T r ip le B u rn e r C o m p u ta tio n a l D o m a in . T h e d o m ain size fo r a ll analyses was 3 0 m X 35 m X 25 m . T h e tw o green objects show n on the rig h t represent the ra d ia tio n flu x m eters lo cated 15 m and 5 0 m fro m the fla re tip respective Page 6 of19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency S in g le B u r n e r M e s h F ig u re 2 - C o m p u ta tio n a l meshes fo r the sin g le b u rn er case (1 1 0 ,0 0 0 c e lls ), the th ree -b u rn e r case (1 8 8 ,0 0 0 c e lls ), and the fu ll fla re fie ld case (1 .2 m illio n c e lls). E ach m esh shows lo c a lly fin e m esh near the burners Page 7of19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency B o u n d a r y C o n d itio n s T h e B o u n d a ry con d itio n s used w e re h yd ro static pressure on a ll boundaries excep t the ground. N o -flo w con d itio n s w e re selected fo r the g round surface. T h e o n ly exc ep tio n to the h yd ro static b o u n d ary c o n d itio n are those cases w h e re a cross w in d was b lo w in g . W h e n a cross w in d was sim u lated , the upstream boundaries w e re set to the w in d v e lo c ity and a ll o th er boundaries w e re m a in ta in e d as h yd ro static pressure con d itio n s. T h e th e rm a l and species b o u n d ary con d itio n s w e re set to 3 0 0 K (2 7 °C ) and a ir com po sitio n resp ectively. P h y s ic a l a n d N u m e r ic a l S u b -m o d e l S e le c tio n T o sim u la te flu id flo w , the m o m e n tu m so lve r was the IS IS - 3 D L E S turbu len ce m o d el. The turbu len ce plays a ro le in setting the rate fo r com bu stio n and m ix in g o f the h o t plum es w ith air. T h e en erg y equ atio n was u tiliz e d to capture the tem p eratu re changes due to com bu stio n and m ix in g . T h e en erg y eq u atio n also in c lu d ed ra d ia tio n effects. T h e species equations w e re solved to keep tra c k o f the d is trib u tio n and co n cen tratio n o f fu e l, o xyg en , in term ed ia te species, soot, and products o f com bu stio n (C O 2 and H 2O ). T h e com bustion m o d e l was used to p ro vid e the species equations source and s in k term s as a fu n c tio n o f species concentrations, lo c a l gas tem p eratu re, and tu rb u le n t d iffu s iv ity . IS IS - 3 D includes a series o f m odels to p re d ic t fla m e e m is s iv ity as a fu n c tio n o f m o le c u la r gas c o m p o sitio n , soot v o lu m e fra c tio n , fla m e size, shape and tem p eratu re d is trib u tio n . In tu rn these variab les depend upon solutions to the mass, m o m e n tu m , en erg y and species equations. The ra d ia tio n transport m o d e l is used n o t o n ly to p re d ict ra d ia tio n flu x on e xtern a l (an d in te rn a l) surfaces, b u t it also pro vid es source and s in k term s to the e n erg y eq u atio n so that fla m e tem p eratu re d is trib u tio n can be predicted . T r a n s ie n t C a lc u la tio n In each case, the C F D s im u la tio n was started w ith an in itia l tem p eratu re and flo w fie ld and run o v e r s u ffic ie n t tim e to a llo w the flo w to reach steady (o r q u asi-stead y) state con d itio n s. Steady state w as d e te rm in ed b y e x a m in in g flo w and th e rm a l variab les fo r re la tiv e constancy w ith tim e . Since a tran sien t so lve r was used, a ll fie ld variab les flu c tu a te in tim e due to turbu len ce and o th er n o n -lin e a rity ' s in the eq u atio n system . H o w e v e r w h e n e x a m in in g any fie ld v aria b le , no g radu al slope was o bserved - ju s t short term flu ctu atio n s as expected in tu rb u le n t flo w s . T h e con vergence c rite ria chosen fo r the sim u latio n s w e re that the eq u atio n o f state was alw ays satisfied to w ith in 0 .1 % o r less at any lo c atio n in the c o m p u tatio n a l d o m ain . T y p ic a lly the convergence c rite ria was b etter than the m a x im u m a llo w a b le since the tim e step constraint was lim ite d b y C o u ran t co n d itio n s, w h ic h a llo w s the flo w fie ld to be solved to a h ig h e r degree o f accuracy. Page 8 of19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency P o s t P ro c e s s in g C F D R e s u lts A fte r the c a lc u la tio n co n verg ed at steady co n d itio n s, app ro p riate contours m ag n itu d e, tem p eratu re, and v e lo c ity vectors w e re p repared. o f v e lo c ity C ontours and discussion are presented fo r each s im u la tio n b e lo w . F ig u re 3 shows a soot isosurface co lo red b y lo c a l tem p eratu re fo r a 3 -b u rn e r test b u rn in g propane. A ls o d ep icted are g rid lines show ing m esh re fin e m e n t near the burners. A ls o show n is the ra d ia tio n m easuring boxes lo cated 15 m and 3 0 m fro m the fla re tips. T h e ra d ia tio n m easuring boxes w e re objects that w e re p laced in to the m o d e l fro m w h ic h ra d ia tio n flu xe s cou ld be extracted . These ra d ia tio n in te n s ity p redictio n s co u ld then be com pared to actual m easured test data. Isovolume temp -2 e + 0 3 F ig u re 3 - S oot isosurface co lo red b y tem p eratu re in a 3 b u rn er s im u la tio n . A ls o d ep icted is the m esh and ra d ia tio n flu x m onitors Page 9 of19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency ANALYSES S in g le F la r e S im u la tio n s Two sin g le fla re sim u latio n s w e re p e rfo rm e d w ith tw o m esh densities. T h e tw o fla re sim u latio n s w e re o f propane and e th ylen e w ith the sm all fla re tip . T h e fin e r m esh d en sity was used fo r the d e ta ile d s im u la tio n fo r com parison to fla re m easurem ents w h ile the coarse m esh was used to test the m o d e l u n d er fu ll fie ld co n d itio n s. T h e d e ta ile d s im u la tio n u tiliz e d a m esh den sity o f 9 7 ,0 0 0 cells (4 5 x 4 4 x 4 9 ) a p p lied to a p h ysical d o m ain o f 6 m X 6 m X 25 m . B y co m p arin g results fro m the coarse m esh s im u la tio n to results fro m the d e ta ile d m esh s im u la tio n , the so lu tio n was tested fo r g rid d ep endency and o v e ra ll accuracy. T h is p ro v id e d a m eth o d o f c a lib ra tio n w h e n a p p lied to the fu ll fie ld s im u la tio n . T h e fu ll fie ld s im u la tio n used the same coarse m esh fo r each burn er. H e n c e , b y c a lib ra tin g the coarse m esh to g iv e the same result as b o th the d e ta ile d m esh s im u la tio n and the fla re test data, then it can be assum ed that the fu ll fie ld c a lc u la tio n is as accurate as possible. T h r e e F la r e s im u la tio n s S everal three b u rn er fla re sim u latio n s w e re p e rfo rm e d (see F ig u re 5 as an e x a m p le ). These 3 b u rn er sim u latio n s b u rn ed a v e rity o f fuels in c lu d in g propane and eth y len e , each at a v a rie ty o f tip pressures. These sim u latio n s w e re p e rfo rm e d on a m esh w ith 1 8 8 ,0 0 0 cells w h ic h covered a c o m p u tatio n a l d o m ain o f 35 m X 35 m X 3 0 m so the ra d ia tio n com parisons to e x p e rim e n ta l results co u ld be m ade at distances o f 15 m eters and 3 0 m eters. T h e ra d ia tio n com parisons a llo w e d an e v a lu a tio n o f o v e ra ll m o d e l accuracy. T h e ra d ia tio n in te n s ity depends upon m a n y factors such as com bustion c h em istry, soot p ro d u ctio n and b u rn -u p , and flu id dynam ics fo r fla m e size, shape, tem p eratu re and flo w v e lo c ities . O f a ll p red icted fie ld varia b le s, ra d ia tio n in te n s ity h ad the greatest s e n s itiv ity to errors a n d /o r inadequacies in the C F D m o d el. H en c e com pariso n to ra d ia tio n m easurem ents p ro vid ed the best m eth o d o f g lo b a l m o d e l v a lid a tio n . E x p e rim e n ta l results fo r one o f the three b u rn er tests are show n in F ig u re 4. T h is test was conducted b u rn in g p ropane at the sam e flo w rate used in the th ree -b u rn e r fla re s im u la tio n show n in F ig u re 3. T h e m easured fla m e h e ig h t fo r the three b u rn er test was a p p ro x im a te ly 1 1.9 m eters h ig h com pared to the estim ated fla m e h e ig h t o f 12 m eters fro m the sim u la tio n . F u ll F ie ld C a lc u la tio n s F o u r fu ll fie ld calcu latio n s w e re m ade representing a p ea k flo w case and a sustained m ix e d gas case w ith o u t a crossw ind. These calcu latio n s h ad a m esh d en sity o f o v e r 7 0 0 ,0 0 0 cells fo r the m ix e d gas case and 1 ,2 0 0 ,0 0 0 fo r the p eak flo w case. T h e p h ys ic al d o m ain sim u lated was 10 m eters b eyo n d the fen ce p e rim e te r in a ll directio n s w ith an o v e ra ll h e ig h t o f 25 m . R esults fro m these cases w e re used to evalu ate the to ta l a ir dem an d fo r the fla re as w e ll as the ra d ia tio n flu x to tw o m id p o in ts on the w in d fence. Page 10of 19 EvaluationoftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustionTechnology: Improvingthe Environment andEnergyEfficiency F ig u re 4 - E x p e rim e n ta l m easurem ents o f fla m e shape and fla m e h e ig h t fo r single b u rn er test. F la m e lif t o ff represents n o n -lu m in o u s (n o -so o t) com bu stio n near the tip . T o ta l A i r D e m a n d R e s u lts T o ta l a ir dem an d fo r the fu ll fie ld sim u latio n s was e va lu ated b y sum m in g the mass flo w crossing a re ctan g u lar h o riz o n ta l p lan e lo c ated at 20 m e le v a tio n and v e rtic a l re ctan g u lar planes surrounding the fla re b u rn er (s). T h is created a b o x surrounding the fla re b u rn er and a llo w e d d ete rm in atio n o f to ta l a ir flo w . In the absence o f a crossw ind, flo w o u t throu g h the top plan e was s u ffic ie n t to q u a n tify to ta l a ir dem and, since a ll oth er faces had in flo w . T h e to ta l a ir dem an d fo r a ll o f the cases run is presented b e lo w (see T a b le 1). Page 11 of 19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency F ig u re 5 - E x p e rim e n ta l m easurem ents o f fla m e shape, fla m e h eig h t, and ra d ia tio n flu x fro m a th ree -b u rn e r test firin g eth ylen e u n d er "n o -w in d " conditions. T h e r m a l R a d ia tio n E s tim a te T h e rm a l ra d ia tio n w as estim ated fo r th ree -b u rn e r e th y len e fla re and the p red ictio n s w e re com pared w ith e x p erim en t. R a d ia tio n m easurem ents w e re m ad e fo r these tests at distances o f 15m (5 0 ft) and 3 0 m (1 0 0 ft), three d riv in g pressures (2 .8 , 7 .3 , and 11.4 P S IG ), and tw o tip sizes. T h ese ra d ia tio n m easurem ents a llo w a com pariso n o f the o v e ra ll accuracy o f the eth y len e fla re m o d el. T h e c o m p u tatio n a l m o d e l includes a p h ys ic al d o m ain o f 7 0 m in h o riz o n ta l exten t b y 3 0 m h ig h . T h e n u m b er o f c o m p u tatio n a l cells w as 6 2 X 6 2 h o rizo n ta l X 4 9 h ig h (1 8 8 3 5 6 to ta l). T h e b o u n d ary con d itio n s w e re h yd ro static pressure on the top surface and as rep o rted constant average w in d speeds on the h o riz o n ta l boundaries. T h e e th ylen e was in je c te d at 3 d iffe re n t pressures to represent the e x p e rim e n ta l con d itio n s. R a d ia tio n m eters w e re s im u lated as v e rtic a lly a lig n ed surfaces and the in c o m in g ra d ia tiv e flu x w as m o n ito re d on those surfaces. Page 12 of19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency T a b le 1 - P red ic te d R esults fro m Cases C o n sid ered . N o te a ll calcu latio n s w e re m ad e fo r a "n o ­ w in d " c o n d itio n . C ase Evaluation P a n e Total Air Row Fuel Row Air/Fuel Description A rea (m 2) (kg/s) (kg/s) Mass R ato Single Burner Propane 14.63 60 0.46 130 1 Burner (Tip 1) Ethylene 13.26 52 0.94 55 3 Burner (Tip 1 - 7.3PSI) Ethylene 36 134 2.88 47 3 Burner (Tip 1 - 11.4PSI) Ethylene 36 144 3.84 38 3 Burner (Tip 2 - 7.3PSI) Ethylene 36 150 4.26 35 3 Burner (Tip 2 - 11.4PSI) Ethylene 36 160 5.79 28 Full Reid P eak Flow Ethylene 3226 9700 262.3 37 Full Feld Sustained Row Mixed G as 1843 4800 93.6 51.3 O n e c o m p lic a tio n o f the ra d ia tio n m easurem ent th a t needs to be in c lu d e d is the e ffe c t o f the heated g round surrounding the flares and ra d ia tio n m eters. Since the fla re is e m ittin g a s ig n ific a n t am o un t o f ra d ia tio n , the g round surrounding the fla re heats up. T h e heated ground em its ra d ia tio n and that contributes to the o v e ra ll ra d ia tio n flu x sensed b y the m eter. In ad d itio n to e m ittin g ra d ia tio n , the g round w ill also re fle c t any ra d iatio n that is n o t absorbed. T o in clu d e the effects o f e m itte d and re fle c te d ra d ia tio n fro m the g round surrounding the flares , a ground surface w ith an e m is s iv ity and a b s o rp tiv ity o f 1.0 was in c lu d ed in the m o d el. T h is ground surface was a llo w e d to heat to steady state conditions d u rin g the s im u la tio n . U s in g an e m is s iv ity and a b s o rp tiv ity o f 1.0 a llo w s a reasonable a p p ro x im a tio n o f b o th e m itte d and re flec te d ra d iatio n . T h e o v e ra ll e n erg y b alan ce fo r a g re y g ro u n d surface that b o th em its and reflects ra d ia tio n is id e n tic a l to that w h ic h absorbs a ll the in c o m in g ra d ia n t h eat and re-rad iates it at steady state. Thus a ra d ia tio n m eter, w h ic h is assum ed to be a g re y o r b la c k surface, w ill see the same in c o m in g ra d ia tio n flu x in b o th scenarios. I f the g round reflects ra d ia tio n sp ectrally, rath er than d iffu s e ly then som e u n c e rta in ty w o u ld be in tro d u ced in to this assum ption. A second e ffe c t that was in c lu d ed in the calcu latio n s was the atm ospheric tran s m is siv ity. T h e atm ospheric tran s m is siv ity m o d e l o f H a m in s [7 ] was in c lu d ed fo r a ll ra d ia tio n that passed Page 13 of 19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustionTechnology: Improving the Environment andEnergyEfficiency through c lea r air. T h e H a m in s m o d e l depends upon a m b ien t tem p eratu re, source tem p eratu re, and re la tiv e h u m id ity . It accounts fo r ra d ia tio n absorption b y w a te r v ap o r and carbon d io x id e . T a b le 2 presents the p red ictio n s, m easurem ents, and re la tiv e e rro r fo r 12 d iffe re n t cases. T a b le 2 - R a d ia tio n P red iction s and T es t R esults C o m p ariso n Tip Size 3 3 3 3 3 3 4 4 4 4 4 4 Position (m) 15 15 15 30 30 30 15 15 15 30 30 30 Burner Pressure (psi) 2.8 7.3 11.4 2.8 7.3 11.4 2.8 7.3 11.4 2.8 7.3 11.4 Total Predicted Radiation (W/m2) 2700 4750 6150 650 1350 1650 4325 8050 10000 1150 2580 3250 Measured Radiation (W/m2) 3344 4803 6192 671 1184 1532 6371 8192 9536 1513 2464 2747 Difference (%) -20.0 % -1.0 % -0.7 % -3.0 % +14.0 % +8.0 % -32.0 % -2.0 % +5.0 % -23.0 % +5.0 % +18.0 % N o te that there w e re tw o b u rn er sizes operatin g at vario u s tip pressures w ith ra d iatio n data taken at b o th 15 m eters and 3 0 m eters. T h e IS IS - 3 D p redictio n s are q u ite good fo r m ost o f the suite o f m easurem ents w ith the greatest d eviation s at the lo w e st pressures. It is n o t c lea r w h a t causes the d iscrep an cy b etw een p re d ic tio n and m easurem ents b u t the greatest d eviation s appear to be fo r the lo w pressure cases. O n e p o te n tia l cause that is n o t understood m a y be re late d to w in d effects. A s show n in the tab le above, w h e n the ra d ia tiv e effects o f g round em ission and re fle c tio n are accounted fo r, the p re d icted ra d ia tio n flu xe s are v e ry reasonable, and in som e cases q u ite accurate. T h e greatest d e v ia tio n was fo r the 15 m (5 0 fo o t) la rg e tip , w h ere the ra d ia tiv e flu x was u n d er p red icted b y 3 2 % . O th e rw is e , the p red icted flu xe s are w e ll w ith in e x p e rim e n ta l error. T h e e ffe c t o f w in d speed upon ra d ia tio n can b e illu s tra te d w ith the fo llo w in g e xa m p le. T h e larg e n o zz le , at a d riv in g pressure o f 2.8 psig, was chosen as an e x a m p le because it had the greatest d e v ia tio n o f a ll the e x p e rim e n ta l m easurem ents. T h e w in d speed was v a rie d fro m zero to 4 m /s (9 m p h ). T h e w in d d ire c tio n was 4 5 degrees to w a rd the ra d ia tio n m eters b u t n o t v a rie d in the sim u latio n s. W in d causes the fla re h e ig h t to shorten, and m a y cause flares to m erge depending upon w in d d ire ctio n . T h e c alc u late d ra d ia tiv e flu x is show n in F ig u re 6 . Page 14 of 19 EvaluationoftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergyEfficiency Wind Speed (m/s) F ig u re 6 - E ffe c t o f w in d speed upon ra d ia tiv e h eat flu x fro m a trip le b u rn er, e th ylen e fla re at a distance o f 15 m T h e s e n s itiv ity o f ra d ia tiv e heat flu x to w in d show n in the fig u re above illu strates that a m o m e n ta ry red u ctio n in w in d speed d u rin g the ra d ia tio n m easurem ent co u ld b rin g the p re d ictio n and m easurem ent in to m u ch closer ag reem en t. T h e e x p e rim e n ta l data rep orted a w in d speed range, b u t n o t the exact speed d u rin g the m easu rem en t. In the sim u latio n s the average o f the rep orted range was used and h e ld fix e d d u rin g the c a lc u la tio n . T h e ra d ia tio n m easurem ents p ro v id e a v a lid a tio n o f the o v e ra ll m o d e l because the ra d ia tiv e heat loss fro m the fire is the single m ost sen sitive p aram eter to m o d e lin g varia b le s. R a d ia tio n depends upon fla m e tem p eratu re to the fo u rth p o w e r, fla m e e m is s iv ity , and fla m e size. A ll o f these variab les depend u p o n solutions to the g o vern in g equations, m o le c u la r and soot e m is s iv ity m odels, ra d ia tio n heat transport m odels, com bu stio n ch e m is try m odels, and the o v e ra ll setup o f the C F D m o d e l (i.e . M e s h , W in d B o u n d a ry C o n d itio n s , e tc .). I f any o f the in p u t param eters o r m odels w e re in a p p ro p riate, p red icted results w o u ld n o t m atch e x p e rim e n ta l m easurem ents as w e ll as show n in T a b le 2. In con clu sio n, the good com parisons to test data in d ic ate that the m o d e l is v a lid a te d s u ffic ie n tly fo r ra d ia tio n p redictio n s. T h is con clu sio n is based upon the tw e lv e com parisons representing tw o d iffe re n t n o zzles o perated at three d iffe re n t pressures each. T h e ra d ia tio n p re d ic tio n fo r the fu ll fie ld is m o re d iffic u lt to estim ate because ra d ia tio n le a v in g one ro w o f burners m ust pass through o th er row s b efo re it reaches the fence. W h e n ra d ia tio n passes throu g h o th er row s o f flares , o r portions o f ro w s, som e o f the ra d ia tio n is absorbed and re -ra d ia te d in o th er d irectio n s. T h ese effects are n o t considered in the IS IS -3 D ra d ia tio n m o d el. In IS IS -3 D , w h en ra d ia tio n leaves a fla m e no checks are m ade to see i f that ra d ia tio n interacts w ith in te rv e n in g objects o r o th e r flam es . Page 15of 19 T h is keeps the code fast ru n n in g Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency since shadow ing is ig n o red . W h e n such effects do occur, as in the fu ll fie ld s im u la tio n , the user m ust e x p lic itly account fo r th em in the ra d ia tio n p re d ictio n . T o m o d ify the p red icted fu ll fie ld ra d ia tio n flu x to account fo r in te rv e n in g fla m e absorption, a s im p le n u m erica l m o d e l w ith the fo llo w in g assum ptions was m ade. 1. T h e fla m e w id th and h e ig h t are e q u al to the ro w len g th and p red icted fla m e h eig h t. T h a t 2. T h e fla m e em issive p o w e r is equ al to the to ta l ra d ia tiv e heat loss (d e riv e d fro m the IS IS - is a ro w o f flares behaves as a continuous w a ll o f flam es. 3 D c alc u la tio n ) d iv id e d b y the ra d ia tiv e surface area (le n g th *w id th *2 ) o f a ll the row s. 3. T h e flu x fro m any ro w is equ al to the em issive p o w e r, tim es the v ie w fa c to r, tim es an atten u atio n fa c to r fo r any in te rv e n in g row s and the atm osphere. 4. T h e atten u atio n fa c to r fo r any ro w is equ al to 1 -e (-T), w h e re T is the o p tic a l thickness o f the ro w . T h e o p tic a l thickness o f the flam es is sup p lied in the IS IS - 3 D output. 5. T h e e ffe c t o f g ro u n d re fle c tio n and em issio n is calc u late d w ith the sam e assum ptions and m e th o d o lo g y as the 3 b u rn er tests. A n u m e ric a l m o d e l in c lu d in g these features and assum ptions was d ev elo p e d and the results appear in the tab le b e lo w (see T a b le 3 ). R a d ia tio n in te n s ity at the m id p o in t o f the fences p a ra lle l to the b u rn er row s was m o d eled . R a d ia tio n to the m id p o in t on the fen ce p erp e n d icu lar to the row s was n o t m o d eled because the row s o n ly p a rtia lly obscured each other; those cases are la b ele d ( N A ) . P a rtia l o b scuration was b ey o n d the scope o f this s im p le absorption m o d e l since it requires s ig n ific a n tly m o re c o m p lex v ie w fa c to r calcu latio n s. T a b le 3 shows p redictio n s fo r the fu ll fie ld ra d ia tio n flu x to the fences. T h e ra d ia tio n p redictions d ire c tly fro m IS IS - 3 D (i.e ., n o t accounting fo r fla re ro w ab so rp tio n) are show n in the u p p er p art o f the tab le c e ll - la b ele d " IS IS - 3 D O u tp u t" . T h e ra d ia tio n p re d ic tio n m o d ific a tio n that accounts fo r fla m e absorption due to in te rv e n in g row s are show n in the lo w e r p art o f the same c e ll, la b e le d (M o d ific a tio n ). L a b e lin g o f the w a lls around the fu ll fla re fie ld is taken fro m a "p la n -v ie w " persp ective as le ft-w a ll, rig h t-w a ll, and b o tto m -w a ll. T a b le 3 - P red ic te d R esults fro m M o d ific a tio n s In v es tig a ted WALL CASE Case 2 Peak Flow 944 T/hr No Wind Case 3 Sustained 337 T/hr No Wind Left Wall "ISIS-3D Output" (Modification) W/m2 "78,000" Right Wall "ISIS-3D Output" (Modification) W/m2 "63,000" Bottom Wall "ISIS-3D Output" (Modification) W/m2 "108,000" Flame Optical Thickness 0.275 (61,000) " 15,000" (35,000) "15,000" NA "35,000" 0.28 (6,600) (6,600) NA F la m e H e ig h t E s tim a te F la m e h e ig h t is d iffic u lt to q u a n tita tiv e ly assess. T h e reason is that IS IS - 3 D predicts fla m e tem p eratu re and species concentrations b u t does n o t p re d ict v is ib le lig h t in te n s ity w h ic h is w h a t Page 16of 19 EvaluationoftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency an o b server w ill re ly on to m easure fla m e h e ig h t (see F ig u re 4 and F ig u re 5 ). T h e ra d iatio n m odels p re d ict ra d ia tio n th e rm a l flu x , b u t since the p lu m e above the v is ib le fla m e contains C O 2, H 2O , and traces o f soot, it s till radiates a lb e it in the in frared . T h e re fo re , an altern ate procedure was adopted to p re d ict v is ib le fla m e h eig h t. T h e procedure selected was to m o n ito r the c o n cen tratio n o f e th y len e and select the lo c a tio n w h e re the eth y len e co n cen tratio n equals the v alu e corresponding to the m easured fla m e h e ig h t in the trip le b u rn er tests. T h a t eth ylen e c o n cen tratio n turns o u t to be a mass fra c tio n o f 0 .0 3 . A lth o u g h s m all am ounts o f in term ed ia te products and soot m a y s till be present, the d riv e r fo r a ll the c h e m ic al reactions is the ra w fu e l. Photographs and m easurem ents o f the trip le b u rn er e th ylen e tests re v e a l that traces o f soot do e x is t in the p lu m e above the v is ib le fla m e . T h u s, m o n ito rin g soot con cen tratio n overestim ates fla m e h e ig h t since the soot co n cen tratio n at the v is ib le lim it is n o t k n o w n . U s in g e th y len e con cen tratio n as a p re d ic to r o f fla m e h e ig h t, the p eak fu ll fie ld fla m e heights are show n in F ig u re 7. T h e v ie w show n corresponds to a ^ -s y m m e try no w in d conditions s im u la tio n . C o ncentration s show n are iso-surfaces, w h ic h are surfaces o f constant eth ylen e c o n cen tratio n equ al to a 0 .0 3 mass fra c tio n . T h e h e ig h t show n in this im a g e is 25 m . F lam es fro m the le ftm o s t ro w are a p p ro x im a te ly 13 m h ig h w h ile flam es fro m the rig h tm o st ro w are a p p ro x im a te ly 2 m h ig h . Since R o w 1 is a spare, it is n o t show n. A lth o u g h o n ly 4 row s are show n, som e o f the s m a lle r row s h ave been c o m b in ed to fo rm a single ro w to im p ro v e c o m p u tatio n a l e ffic ie n c y . Isosurface temp -l.B e + 0 3 -1.G 5e+03 Z -1.5e+03 1.35e+03 F ig u re 7 - A n isosurface v ie w o f e th ylen e co n cen tratio n fo r a 1 /4 s ym m e try p eak flo w fu ll fie ld case w ith no w in d . T h e fla m e h e ig h t is show n to be b e lo w the fen ce h eig h t. A s show n in F ig u re 7 , the fla m e h e ig h t is estim ated to be w e ll b e lo w the fen ce le v e l. A lth o u g h this is a lA sym m e try im a g e, the re m a in d e r o f the fie ld behaves v e ry s im ila rly . A q u arter s ym m e try p ro b le m was chosen to m a x im iz e the c o m p u tatio n a l c e ll density. Page 17of 19 EvaluationoftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustionTechnology: Improvingthe Environment andEnergyEfficiency RESULTS T h e results o f these sim u latio n s in d ic ate that the to ta l a ir dem an d fo r the various cases is b eyo n d the s to ic h io m etric re q u irem e n t (s to ic h io m e tric re q u irem e n t varies w ith fu e l ty p e ) b u t is o ften in the range 15:1 a ir to fu e l mass ra tio fo r m a n y hydrocarbons. T h e a ir to fu e l ratios fo r the various cases range fro m a h ig h o f 150 to a lo w o f 3 8 (usin g la rg e r fla re tip o perated at h ig h pressure fo r the 3 b u rn er test). F o r the fu ll fie ld analysis, the range w as fro m 4 0 fo r the p eak flo w fie ld to 6 0 fo r the sustained case. T hus the calcu latio n s results support the con clu sio n that s u ffic ie n t a ir w ill be entrain ed and present to b u rn ess en tially a ll fu e l fo r b o th the sustained and the p eak flo w cases. O n e caveat is that these a ir dem an d calcu latio n s represent the "T o ta l A ir D e m a n d " fo r the system as a w h o le and do n o t address the p o s s ib ility th a t lo c a l a ir starvation m a y o ccur som ew h ere in the fla re fie ld . T h e ra d ia tio n p redictio n s in d ic ated that h ig h ra d ia tio n flu xe s in c id e n t on the cen ter p o in t o f the w in d fen ce are possible (as h ig h as 1 0 0 ,0 0 0 W /m 2 fo r p eak flo w case). case, the ra d ia tio n flu x in c id en t on the w in d fen ce ranges fro m 3 5 ,0 0 0 W /m F o r the P e ak F lo w 2 2 to 1 0 8 ,0 0 0 W /m . F o r the Sustained F lo w case the in c id e n t ra d ia tio n to the w in d fen ce ranges fro m 6 ,6 0 0 W /m 2 to 3 5 ,0 0 0 W /m 2. F la m e heights w e re estim ated b y m o n ito rin g eth y len e co n cen tratio n and selecting the lo c atio n w h e re the co n cen tratio n equals the v alu e fo u n d fo r the fla m e h e ig h t m easurem ents in the trip le b u rn er tests. Is o -s u rfa ce contours re v e a l that the fla m e heights fo r the p eak fie ld w ith no w in d are w e ll b e lo w the fen ce h eig ht. C O N C L U S IO N S T h e w o rk presented in this p aper docum ents a tran sien t fla m e analysis fo r the m u lti-tip lo w p ro file fla re . O b je ctiv es fo r this w o rk in c lu d e p re d ictin g the to ta l a ir dem an d and the expected fla m e h e ig h t fo r a sustained flo w case and a p ea k flo w case b u rn in g eth y len e . T h e IS IS - 3 D C F D m o d e l was used to p e rfo rm the co m p u ter sim u latio n s fo r a sin g le b u rn er test and a three b u rn er test to v e rify m o d e l p redictio n s. B ased on m o d e l v e rific a tio n , the fu ll fie ld was sim u lated . F u ll fie ld sim u latio n s in c lu d in g conducted. a ll burners in the fla re fie ld plus the surrounding fen ce w e re IS IS - 3 D p redictio n s in d ic ate that s u ffic ie n t a ir is e n train ed through the fen ce to p reven t fla m e fro m exten d in g b ey o n d the top o f the fen ce and fro m g en eratin g n o ticeab le sm oke fo r the p eak flo w case w h ic h is considered the lim itin g case. R a d ia tio n flu xe s to the w in d fence are p red icted to be up to 1 0 0 ,0 0 0 W /m 2. REFERECNES 1. S u o -A n ttila , A ., W a g n e r, K .C ., and G re in e r, M ., 2004, "A n a lys is o f E n clo su re F ire s U s in g the Is is -3 D T M C F D E n g in e e rin g A n a ly s is C o d e," Proceedings of ICONE12, 12th International Conference on Nuclear Engineering, Arlington, Virginia USA, April 25-29. 2. G re in e r, M ., and S u o -A n ttila , A ., 2004, " V a lid a tio n o f the IS IS C o m p u te r C o d e fo r S im u la tin g L a rg e P o o l F ires U n d e r a V a r ity o f W in d C o n d itio n s ," ASME J. Pressure Vessel Technology, V o l. 1 26 , pp. 3 6 0 -3 6 8 . 3. G re in e r, M ., and S u o -A n ttila , A ., 2003, "Fast R u n n in g P o o l F ire C o m p u te r C o d e fo r R is k A ssessm ent C alc u la tio n s ," presented at the ASME International Mechanical Engineering Congress and Exhibition, N o v e m b e r 1 5 -2 1 , 2 0 0 3 , W a s h in g to n , D C . Page 18of 19 Evaluation oftheAir-Demand, Flame Height, andRadiationfromlow-profileflare tips usingISIS-3D Advances inCombustion Technology: Improving the Environment andEnergy Efficiency 4. G re in e r, M , A re , N ., L o p e z , C ., and S u o -A n ttila , A ., " E ffe c t o f S m a ll L o n g -D u ra tio n F ires on a Spent N u c le a r F u e l T ra n s p o rt Package," Institute o f N u c le a r M a te ria ls M anagem ent 45 th A n n u a l Meeting, O rla n d o , F L , J u ly 1 8 -2 2 , 2 0 0 4 . 5. S o ciety F ire P ro te ctio n E n g in eers, 1 9 9 5 , F ire P ro te ctio n E ngineering, 2nd E d itio n , N a tio n a l F ire P ro te ctio n A ss o cia tio n P u b lica tio n . 6 . G re in e r, M ., and S u o -A n ttila , A ., 2 0 0 6 , "R a d ia tio n H e a t T ra n s fe r and R e a c tio n C h e m is try M o d e ls fo r R is k A ssessm ent C o m p a tib le F ire S im u latio n s ," Jo u rn a l o f F ire P rotection E ngineering, V o l. 16, pp. 7 9 -1 0 3 . 7. Fuss S .P ., A . H a m in s . 2 0 0 2 , "A n estim ate o f the co rrectio n a p p lied to ra d ian t fla m e m easurem ents due to atten u atio n b y atm o sp heric C O 2 and H 2 O " , F ire S afety J o u rn a l , V o l. 3 7 , pp. 1 8 1 -1 9 0 . Page 19of 19