| OCR Text |
Show F u e l C o m p o s i t i o n E f f e c t o n F la r e F l a m e I n e f f i c i e n c y J. Pohl V irginia Tech, Alexandria, V A P. G ogolek CETC-Ottawa, Ottawa, Ontario, Canada K 1A 1M1 pgogolek@ nrcan.gc.ca J. Seebold ChevronTexaco (retired), Atherton, C A 94027 R. Schwartz John Zink LLC, Tulsa, OK 74116 In tro d u c tio n The effects o f fuel com position on the performance o f flare flam es have long been known. P ossibly the first performance indicator w as the tendency o f a fuel to sm oke. Various fuel characteristics such as average m olecular w eigh t and carbon-to-hydrogen ratio were advanced to guide the design o f early sm okeless flares (Baukal, 2001 ). The U S EPA controls the em ission o f flare flam es based on 40C F D 60.18. This regulation is based on the landmark studies performed by Pohl and co-workers (19 8 3 -1 9 8 6 ). This work m easured the experimental flam e stability as functions o f heating value and flare exit velocity. The em issions from flare flam es were then sim ply correlated w ith the flam e stability lim it (see Fig. 1). A b ove the flam e stability the com bustion efficien cy o f flam es o f all com pounds tested w as greater than 98%, near the flam e stability lim it the com bustion efficien cy decreased rapidly to 85 -90% . This program did not thoroughly investigate the effect o f w ind and w ind structure, chem ical com position. In particular, W alsh et al. (2002) show ed mixtures o f hydrogen were m uch more stable than m ixtures o f propane-nitrogen. G ogolek and H ayden (2002) show ed that increased ratio o f the w ind energy to the flare je t energy also resulted in com bustion inefficiency. N ob el, et al. (1984) made an attempt to correlate flam e stability w ith a flam e temperature and the ratio o f the low er and upper flam m ability limits. Pohl, et al. (1984) used his data to confirm that this relationship also held for com bustion efficien cy (see Fig. 2). H ow ever, the proposed relationship is based on lim ited data, has considerable scatter, and sm all differences in the temperature ratio o f flam m ability lim its predict large changes in flam e stability. A brief, up-todate review o f the flare efficien cy literature is found in Seebold et al., (2003). Com bustion inefficiency o f a flare flam e results from extinguished portions o f the flam e. These portions becom e extinguished w hen the pocket is diluted by m ixing o f air or steam through m olecular diffusion, flam e stretch, and turbulent m ixing. W hen a pocket is diluted sufficiently it does not generate heat faster than it loses it, the flam e pocket extinguishes itself. The flam e can also be extinguished through dilution to the point w here the flam e speed is slow er than the im posed v elocity and the flam e blow s off. Finally, the flam e can be extinguished by dilution beyond the low er flam m ability limit. The purpose o f this paper is to investigate the relationship betw een chem ical properties o f the com pound flared and the com bustion efficien cy o f the flare. W e are particularly interested whether a relationship can be developed betw een the chem ical nature o f the flared com pound and the com bustion efficien cy that w ill allow us to generalize the relationship to the point where the performance o f untested com pounds can be predicted. F lo w P a r a m e t e r s a n d M ixing R e g i m e s W hen attempting to correlate flare efficien cy with fuel properties, it is important to not confuse m ixing regim es. The m ixing in a flare flam e, w ith the possible presence o f a crossw ind, w ill depend on the air and fuel properties, average speeds and pipe size. From these flo w variables one can form tw o dim ensionless parameters representing the relative strength o f the m om entum fluxes. One is the Froude N um ber F r = ----------------------(Pa - Pp ) gD p ( 1) the ratio o f the je t strength to the buoyancy o f the burned gases. The other is the m om entum flux ratio p fU 2 R = t- U r P aU a ( 2) w hich g iv es the relative strength o f the fuel je t to the crosswind. A rough regim e map based on these tw o parameters is given in Fig. 3. The regim es identified are the jetting regim e, where fuel inertia dom inates both buoyancy and crossw ind; buoyant flam es where there is little initial fuel m om entum and little or no crossw ind; wake stabilized where the crossw ind m om entum dom inates both fuel m om entum and buoyancy. Each o f these extreme cases has distinct m ixing m echanism s. W ithin a flo w regim e, the rate o f com bustion w ill be a com bination o f m ixing rate and therm ochem ical properties (reaction rate). T h erm o ch e m ic a l P ro p e rtie s A s noted earlier, operating regulations have used the heat content to summarize the therm ochem ical properties o f the fuel mixture. More precisely, it is the lo w heating value, LHV, on a volum e basis o f the gas mixture. The LH V can be easily calculated for a mixture given the L H V s o f the individual species. The L H V has several desirable features: □ Clearly and precisely defined for a pure species. □ Tabulated values for a w ide range o f species. □ Accurate rule for calculating values for mixture given the pure species data. □ U nam biguous m ethod o f handling inert species. These are the requirements w e have for any criterion accounting for the effect o f fuel com position on flare flam e efficiency. Flam m ability lim its are determined experim entally. The upper flam m ability limit, U FL, is the m axim um concentration as a volum e % o f fuel that can sustain a flam e. The low er flam m ability limit, LFL, is the m inim um concentration as a volum e % o f fuel that can sustain a flam e. There is som e am biguity in the values obtained, since they can depend on the configuration o f the experimental equipment. T w o extensive collections o f flam m ability data are Coward and Jones (1952) and Zabetakis (1965). The flam m ability lim its for a mixture o f flam m able gases is estim ated using Le Chatelier's Rule 100 % = y _ x ^ F L ma ( 3) i FLi and similarly for the UFL. This rule works w e ll for mixtures o f hydrocarbons. Som e significant deviations can be found, for exam ple, CH 4 w ith H 2S. Data for dilution o f a single flam m able gas w ith a single inert gas is available. It is generally true that inert diluents have a strong effect on the UFL but a relatively w eak effect on the LFL. Zabetakis (1965) g iv es a m odification o f Le Chatelier's Rule to incorporate inert gases. The flam e speed is another experim entally determined value. It is the speed at w hich a flame propagates in a fuel-oxidant mixture. There is no com m only accepted m ethod for estim ating the flam e speed o f mixtures. The adiabatic flam e temperature is readily calculated for any flam m able mixture, as w ell as being w id ely tabulated. The ignition temperature is the m inim um temperature at w hich a mixture o f fuel and oxidant ignites. It is an experimental quantity. There is no rule for estim ating the ignition temperature for m ixtures, or for the presence o f inert species. Table 1 has values for LH V on a m ass basis, ignition temperature, adiabatic flam e temperature and m axim um flame speed for selected pure com pounds. Flame speed and ignition temperature do not have the four required features for a fuel com position parameter. The candidates as sim ple fuel parameters are the LH V (either volum e or m ass basis), adiabatic flam e temperature, and the flam m ability limits. S c re e n in g th e C a n d id a te s The data available in the open literature on jetting flares and w ake-stabilized flares in C A N M E T 's Flare T est Facility are useful to screen the candidate parameters. N on e o f the sim ple parameters are adequate, but the various deficiencies m ay guide the developm ent o f a useful com pound parameter. The adiabatic flam e temperature clearly does not correlate w ith experience in fuel reactivity in flare flam es. H ydrogen is clearly the m ost reactive fuel, but has adiabatic flam e temperature b elo w those o f ethylene and propylene. LH V on a m ass basis has similar problems: methane has a higher value than m any m uch more reactive fuels (acetylene, ethylene, propylene in particular). LH V on a volum e basis works w ell for sim ple pure hydrocarbons but has revealed deficiencies for hydrogen rich flares (W alsh et al. 2002). It also fails to account for the effect o f different inert diluents. Figure 4 show s the com bustion efficien cy for flare flame in the w ake-stabilized regime firing natural gas diluted w ith nitrogen or carbon dioxide. The fuel gas has similar LH V (in fact, the nitrogen diluted gas has slightly low er L H V ) but the inefficien cy o f the carbon dioxide diluted natural gas is dramatically higher, and blow s out at the highest w ind speed o f 42 km/h. The low er flam m ability lim it is not able to explain these results because the LFL is largely unchanged during dilution. H ow ever, the ratio UFL:LFL appears to have som e correlative strength as show n in Fig. 5. This strength does not appear to transfer to the jetting regime. Figure 6 show s the b lo w -o ff data o f W alsh et al. (2002) for hydrogen flares and the EER screen tests reported in Pohl et al. (1984). A t roughly the same value o f UFL:LFL, there is a fourfold variation in the b lo w -o ff velocity. The com pounds in this group include propylene, propane, butane, carbon m onoxide and ammonia. C o n clu sio n There is a gap in the know ledge about h ow fuel com position affects the performance o f flare flam es. Four criteria are set forward that a parameter m ust satisfy to be useful for predicting the fuel effect. These are: □ Clearly and precisely defined for a pure species. □ Tabulated values for a w ide range o f species. □ Accurate rule for calculating values for mixture given the pure species data. □ U nam biguous m ethod o f handling inert species. Several sim ple therm ochem ical properties o f fuel gases have been scrutinized against these criteria and the available experimental data as possible candidates to summarize the effect o f fuel properties. N on e has proven satisfactory for all flo w regim es. H ow ever, several com pound parameters m ay be devised that work for specific flow regim es. N o m en c latu re Dp Fr FLt Flare pipe diameter, m F L mix Calculated flam m ability lim it for a mixture, % g LFL LHV R Ua A cceleration due to gravity, m /s 2 L ow er flam m ability limit, % L ow er heating value (m ass or volum e basis) Froude number for flare flam e, Flam m ability lim it (upper or low er) for com pound i, % M om entum flu x ratio for fuel to w ind, W ind speed, m /s Uf UFL Fuel exit speed, m /s Xi V olum e fraction o f com pound i, % Pa D ensity o f air in wind, kg/m 3 Pf Pp Upper flam m ability limit, % D ensity o f fuel exiting the flare, kg/m 3 D ensity o f hot com bustion products in flare plum e, kg/m 3 R eferen ces Baukal, C.E., and Schwartz, R.E. (Editors), John Zink Combustion Handbook, CRC Press, 2001. Coward, H.F. and G.W . Jones, Lim its o f Flam m ability o f G ases and Vapors, Bulletin 503, Bureau o f M ines, U nited States G overnm ent Printing O ffice, W ashington, 1952. Fristrom, R.M ., Flame Structure and Processes, Oxford U niversity Press, 1995. G ogolek, P.E.G. and A .C .S. H ayden, E fficiency o f Flare Flam es in Turbulent Cross W inds, A m erican Flame Research Com m ittee, 2002. Joseph, D ., J. Lee, C. M cK innion, R. Payne, and J. Pohl, Evaluation o f the E fficiency o f Industrial Flares: Background-Experim ental D esign-F acility, EPA Report N o . 600/2-8 3-0 7 0 , 1983. N ob el, R.K., M .R. Keller, and R.E. Schwartz, A n Experimental A nalysis o f Flame Stability o f Open A ir D iffusion Flam es, A m erican Flame Research Com m ittee International Sym posium on Alternate Fuels and Hazardous W astes, Tulsa, OK, 1984. Pohl, J.H., J. Lee, R. Payne, B. Tichenor, The Com bustion E fficien cy o f Flare Flam es, 77th Annual M eeting and Exhibition o f A ir Pollution Control A gency, San Francisco, CA, June 1984. Pohl, J. H ., R. Payne, and J. Lee, Evaluation o f the efficien cy o f Industrial Flares: Test Results, E P A -6 0 0 /2 -8 4 -0 9 5 , 1984. Pohl, J. H. and N . R. Soelberg, Evaluation o f the efficien cy o f Industrial Flares: Flare Head D esign and Gas C om position, E P A -6 00/2-85-106, 1985. Pohl, J. H. and N . R. Soelberg, Evaluation o f the efficien cy o f Industrial Flares: H 2S Gas M ixtures and Pilot A ssisted Flares, E P A -6 00/2-86-080, 1986. Seebold, J.G., B. C. D avis, P. E. G. G ogolek, L. W . Kostiuk, J. H. Pohl, R. E. Schwartz, N . R. Soelberg, M. Strosher and P.M . W alsh, Reaction E fficiency o f Industrial Flares - The Perspective o f the Past, Com bustion Canada '03, Vancouver, Canada, 2003. W alsh, P., D .K . M oyeda, W .S. Lanier, C.M. B ooth, E.E. Folk, J. M axw ell, W .K . W hitcraft, R.K. N ob le, J.M. Clopton, T. Rogers, C. N icely , R.E. R osensteel, and C. A . M iler, Flame Stability Lim its and Hydrocarbon Destruction o f Flares Burning Streams Containing Hydrocarbon and Inert G ases, A m erican Flame Research Com m ittee Fall M eeting, N ovem ber 2002. Zabetakis, M .G ., Flam m ability Characteristics o f Com bustible G ases and Vapors, Bulletin 627, Bureau o f M ines, U nited States Governm ent Printing O ffice, W ashington, 1965. Tables Table 1 - Therm ochem ical data for a selection o f gases from Fristrom (1995) and Baukal (2001). LH V Ignition Temperature A diabatic Flame Temperature M axim um Flame Speed MJ/kg K K m /s H2 120.0 844 2318 3.06 C 2H 2 48.3 578 2598 1.63 CH 4 50.0 905 2148 0.39 C 2H 6 47.5 745 2248 0.46 C bH 8 46.3 777 2198 0.45 C 2H 4 47.1 763 2375 0.79 C 3H 6 45.7 831 2367 0.5 H 2S 15.2 563 2091 NH3 18.5 965 2074 Compound 9 9 .9 9 0.01 99.9 0.1 99.0 1.0 90.0 Unburned Combustible (percent of initial combustible) Destruction Efficiency (Percent) F ig u re s 10.0 C3 A m m o n i a » 1 . 5 - 4 . 5 O 1»3-8utadiene C 7 E thylene |7 ^ Propane Propane Percent in O xide in in P ropane-N itrogen N itrogen N itrogen i n A m mo ni a T e s t s in Hydrogen S u l f i d e T ests _______________I_____________________ I_________________ I[ 1 0 0 . 0 X.O 2.0 H eating Value/M inim um H e a tin g V alue for S tab ility Figure 1 - D ecrease o f flare efficien cy near the stability lim it for jetting regim e, from Seebold et al. (2003). Nozzle Exit Velocity/Sonic Velocity Figure 2 - The N o b el correlation for stability o f jetting flare and com parison against EER data, from Seebold et al. (2003). ln (F r) ln (R ) Figure 3 - Schem atic regim e map for flares including wind. 40 □ N2 ■ ■ CO2 Inefficiency, % 30 25 20 15 10 □ ■ 5 ■ ■ 0 5 10 15 20 25 □ □ □ □ 0 30 35 40 45 Uw, km/h Figure 4 - Com bustion in efficien cy for natural gas diluted w ith 60% o f nitrogen or carbon dioxide (G ogolek and Hayden, 2003). 40 -| ■ CO2 Dilution 35 O N2 Dilution ANG In efficien cy , % 30 25 - 20 15 - - 10 5 O O 0 1.50 1.70 1.90 2.10 2.30 2.50 2.70 2.90 UFL/LFL, - Figure 5 - Relationship betw een in efficien cy and the ratio o f flam m ability lim its for natural gas w ith inerts at fixed flo w conditions in the wake stabilized regime. Cross w ind speed is 10 m /s. (G ogolek and H ayden, 2003) 900 O EER screening data ■ Walsh H2 data 800 Velocity at stability limit, ft/s 700 600 500 400 300 200 O O 100 <>o 0 1 10 UFL/LFL, - Figure 6 - The dependence o f b lo w -o ff v elocity on ratio o f flam m ability limits for jetting flares. The collection o f points for the EER data includes propylene, propane, butane, carbon m onoxide and am monia. Data from W alsh et al. (2002) and Pohl et al. (1984). 100 |
