Improving flare design a transition from art-form to engineering science

Many of the day-to-day activities surrounding Flaring, at both the project design and operational stages, are vague and apparently outside the control of the Flare Engineer. Design specifications usually only cover a single case out of a multitude of possible compositions and flows. In service, flow rates, compositions, wind and weather are largely decided by happenstance and the control of utility addition is managed on the basis of visual perception. After the fact, Environmentalists try to estimate the extent of the environmental damage and combustion scientists develop theories and models which may help to predict the extent of this impact. The only point where there exists an opportunity to affect the final outcome is before the cycle even starts, when the main design concepts of the flare tip are developed. However, despite the best efforts of combustion scientists and the best intentions of some manufacturers, the majority of flare designers Worldwide still practice a design method which, in the view of the author, which owes as much to experience and "black art" as it does to scientific knowledge. For the flare designer to achieve a fractional percentage gain in efficiency, or to reduce the chances of generating hazardous pollutants, there is a need for a generalized development, and wider dissemination of parameters which are in control of those issues. This involves a serious effort by the Industry, together with the assistance of academia and the use of advanced technical skills. However, along with this effort comes the responsibility to examine the state of the current baseline. In short, the industry has to move out of the arts into the sciences. This paper uses the three most common process related aspects of Flare engineering, (flame size/shape, minimum tolerable flammability and maximum discharge velocity) to highlight the common use of "rules of thumb", which may represent flawed parametric input and suggests alternative and more scientific approaches which may be beneficial in improving an the global performance of flares, presenting each in a manner which may be recognizable and useful to most engineers.

Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 IM P R O V IN G F L A R E D E S IG N A T R A N S IT IO N F R O M A R T -F O R M T O E N G IN E E R IN G S C IE N C E DAVID SHORE FLAREGAS CORPORATION NANUET Ph: N Y . 1 0 9 5 4 USA 1 -8 4 5 -3 7 1 -2 5 1 9 da vid s@ flaregas.com . Prepared for presentation at AFRC-JFRC 2 0 0 7 Joint meeting Waikoloa, Hawaii October 2 2 - 24, 2 0 0 7 Art_to_Science - AFRC.wpd 1 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 ABSTRACT Many of th e day-to-day activities surrounding Flaring, a t both the project d esign and operational s ta g e s , a re v a g u e an d apparently outside the control of the Flare Engineer. Design specifications usually only cover a single c a s e out of a multitude of possible com positions a n d flows. In service, flow rates, com positions, wind an d w e a th e r a re largely decid ed by h a p p e n s ta n c e an d the control of utility addition is m a n a g e d on th e b a sis of visual perception. After th e fact, Environm entalists try to e stim a te the extent of the environm ental d a m a g e a n d com bustion scientists develop th eo ries a n d m odels which m ay help to predict th e e x te n t of this impact. T he only point w h e re th ere exists a n opportunity to affect the final o u tco m e is before the cycle ev e n starts, w hen the main d esign c o n c e p ts of the flare tip a re developed. However, d e sp ite th e b e s t efforts of com bustion scientists an d the b e s t intentions of s o m e m anufacturers, th e majority of flare d e s ig n e rs W orldwide still practice a d esign m ethod which, in the view of th e author, which o w e s a s m uch to e x p e rie n ce a n d "black art" a s it d o e s to scientific know ledge. For the flare d e sig n e r to ac h ie v e a fractional p e rc e n ta g e gain in efficiency, or to re d u c e the c h a n c e s of gen eratin g h a z a rd o u s pollutants, th e re is a n e e d for a g en eralized developm ent, an d w ider dissem ination of p a ra m e te rs which a re in control of th o se issu e s. This involves a se rio u s effort by the Industry, to g eth e r with the a s s is ta n c e of a c a d e m ia a n d th e u s e of a d v a n c e d technical skills. However, along with this effort c o m e s th e responsibility to e x am in e the s ta te of the current baseline. In short, the industry h a s to m ove out of th e arts into the sc ie n c e s. This p a p e r u s e s the th re e m o st com m on p ro c e s s related a s p e c ts of Flare engineering, (flame s iz e /sh a p e , minimum tolerable flammability a n d m axim um d isc h a rg e velocity) to highlight the com m on u s e of "rules of thum b", which m ay re p re s e n t flawed p aram etric input an d s u g g e s ts alternative an d m ore scientific a p p ro a c h e s which m ay b e beneficial in improving a n th e global perfo rm an ce of flares, presenting e a c h in a m a n n e r which m ay be recognizable an d useful to m ost e n g in eers. Art_to_Science - AFRC.wpd 2 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 IM P R O V IN G F L A R E D E S IG N A T R A N S IT IO N F R O M A R T -F O R M T O E N G IN E E R IN G S C IE N C E DAVID SH O R E © FLAREGAS C O R PO R A TIO N , N AN U ET N.Y. 10954 1-845-371-2519 : davids@ flaregas.com PREFACE The combustion s c i e n c e s e n c o m p a s s a wide range of disciplines including automotive engineers, en gin e designers, rocket engineers, heating engineers, furnace d esig n ers and many more. Within this prestigious field, Flare engineering p resen ts a rather lower profile. A s a practice, g a s flaring h a s apparently little a s so c ia te d com p lex technology and the two primary techniques a s so c ia te d with Flaring involve - being able to ignite an indeterminate stream of g a s of largely unknown composition and - causing the resulting flame to burn without too much, obvious, environmental impact O n ce the flare flame is burning, there are very few a s p e c t s of control and the applied principle is usually to b e well aw ay from the flame and just watch it burn. Most of the better-known d esign functions surrounding flaring are dedicated to allowing the flame to burn safely whilst unattended and deciding just how far aw ay personnel should b e for safety. Throughout the entire cycle of a flare, the overall ability to influence the eventual o u tco m e often s e e m s vague. At the beginning of any flare design p rocess, m ost d esign specifications cover a single c a s e out of a multitude of possible com positions and an almost infinite number of possible flows. B a s e d on this, it falls to an applications en g in ee r to d ecid e which p iec e of equipment will b e the m ost likely to m eet not only the written specification but also the implied utilization, whilst remaining economically attractive to the prospective customer. Later, w h en the flare is in service, hardly anything in the p r o c e s s can b e truly controlled. The flow rate and the g a s composition, the wind and the w eath er are all d ecided by h a p p e n s ta n c e and the control of utility addition (steam or air for exa m p le) is in the h an ds of an operator w h o s e primary motivation is a void an ce of a visible em ission and who, in c o n s e q u e n c e , will probably add sufficient utility to completely hide the flame and significantly impact the concurrent combustion efficiency. Art_to_Science - AFRC.wpd 3 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 Overall, a flare's operational performance is generally evaluated in terms of visual perception of the flame regardless of the concurrent combustion efficiency. Only after the fact, and w h en serious pollution con cer n s may have b e c o m e apparent, environmental scientists try to estim ate the true extent of the environmental d am age, and in this spirit, combustion scientists then develop new theories and m od els which may, s o m ed a y , help to predict the extent of this environmental impact. The only point within this entire cycle w here there exists the opportunity to affect the final combustion efficiency is before the cycle e v e n starts, w h en the main d esign c o n c e p ts of the flare tip are d ev elo p e d but, e v e n w h en potential design improvements may b e introduced, the c o s t s of developm ent and testing of potential n ew d e s ig n s tend to im pede w idespread investigation of multiple design candidates. If the flare d esign er is to m a s s a g e n ew or existing d e s ig n s to " sq u eeze" a fractional percentage gain in combustion efficiency, or to reduce the c h a n c e s of generating harmful pollutants, there is a n eed for a generalized d evelop m en t and improved dissemination throughout the industry a s a whole, of the data and parameters which are in control of t h o s e issu es. This involves a serious effort by the Industry, together with the a s s i s t a n c e of a c a d em ia and the u s e of ad v a n ced technical skills. However, along with this effort c o m e s the responsibility to ex a m in e the state of the current "baseline" which, in the view of this author, still follows a practice which o w e s a s much to e x p erien ce and "black art" a s it d o e s to scientific knowledge. In short, the industry h as to m o v e out of the arts into the s c ie n c e s . T H E N E E D F O R E N G IN E E R IN G " Q U A L IT Y " P A R A M E T E R S Any scientific input into flare design n e e d s to b e in a form which can b e easily understood and interpreted by th o s e en g in ee rs responsible for creating or modifying th o s e d esign s. Although ad v a n ced computational a n a ly s e s are available, t h e s e frequently require a d van ced operator skills or a c a s e - b y - c a s e investment, which both mitigate against com m on u sa g e . In the day by day practice of engineering, "real world" d im en sion s and properties are very significant and, although an en orm o u s am ount of excellent scientific thought and combustion research is already published, much g o e s largely unnoticed by the flare industry in d efe r e n c e to a w idespread u s e of engineering "lore" and the "rule-of-thumb" together with a general a c c e p t a n c e of the "statusquo". Art_to_Science - AFRC.wpd 4 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 P o ssib le contributors to this d iscon n ect are - the n e e d within the scientific community to e x p r e s s com p lex scientific logic using d im en sio n le ss terms that accurately describe the various p h e n o m e n a and characteristics. T h ese, although accurate in nature, are s o m e tim e s indeterminate in value; - the n e e d for basic research to simplify research m od els to the least number of variables, often making t h e s e m o d els unrepresentative of "real world" equipment; - the u s e of higher order equations and forms which require computational solutions; - the en g in ee r's day-to-day involvement in project work which precludes the opportunity to s p e n d time in translation of the d im en sio n le ss forms into "real world" numbers; - the commercial requirements of engineering research which frequently condition the results to prove or disprove a pre-defined commercial solution and, consequently, yield limited or slanted results; - the difficulty of obtaining reliable and meaningful large s c a le data. Engineers frequently rely on pre-formulated d esign parameters and treat s u c h a s proven methods. Consequently, useful engineering procedures n eed to be accurate and "solution-oriented". S o m e combustion parameters which apply to flaring are already widely recognized. Most en gin eers will h ave a go o d working understanding of stoichiometry, calorific content and flammable limits and t h e s e features are already u s e d in s o m e c a s e s of formulated design and can b e easily incorporated into others. In this paper, the author d i s c u s s e s a few of the m ost com m on a s p e c t s of flare engineering w here certain existing "rules-of-thumb" n eed to be re-evaluated in order to provide a meaningful and scientifically soun d b aselin e on which to build future data and standards. S o m e alternative and more scientific m eth od s are p roposed and other useful engineering parameters, yet to b e properly evaluated, are identified and rec o m m en d e d for further investigation. T H E F L A R E FL A M E The key to any meaningful design developm ent, must be an understanding of the Flare flame itself. This is, alm ost exclusively, a diffusion flame. Diffusion flare flame characteristics are listed below in terms which might be widely understood within the general engineering community, cross referenced to the e n c lo s e d Figure 1. Art_to_Science - AFRC.wpd 5 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 1. R aw G a s le a v e s the end of the Flare pipe (Flare tip); - Liquids, entrained in the flowing g a s stream are carried into the flame z o n e; 2. Unburned g a s m ixes with air, by diffusion and/or local turbulence; 3. The flame can b e ignited and establish w h en the local mixture c o m e s within a flammable range; 4. The b a s e (ed g e ) of the flame esta b lis h e s a s a stable or unstable interface; 5. The down wind g a s column (flame / plume) d e v e lo p s a c om p le x turbulent form conditioned by the momentum forces of the g a s discharge, the buoyancy forces produced by the heat en ergy r e le a s e and the e d d ie s gen erated in the wind affected "wake" of the flare pipe (Flare Stack); 6. The outer z o n e of the g a s column burns, rather like an en velope, w herever g a s can mix with air in flammable concentrations; 7. The core of the g a s column b e c o m e s hot d u e (mainly) to radiant heat from the burning envelope; - Entrained liquid droplets c o m m e n c e to evap orate from the droplet surface; - G a s e s and vapors in the core of the g a s column d issociate into radicals w h en the local temperature c re a tes molecular instability; - A s a function of the various transport properties of the liquid, molecular dissociation may c o m m e n c e in droplets prior to c om p lete evaporation, fashioning n ew molecular forms with carbon chains resembling polymers; 8. The distribution and concentrations of free, transient and unconverted radicals d ep en d on the local temperature, concentration of the radicals and the resid en ce time s in ce dissociation; 9. With (resid en ce) time, s o m e free carbon radicals (atoms) com b in e to form solid particulate (soot) [ this may also happen with other elem ental solids such a S, Al, Si ]; 10. Solid particulate may not burn (decay) completely during p a s s a g e through the z o n e of o x ygen availability (flame) resulting in solids e m iss io n s (commonly soot); 11. Particulate or u n c o n su m ed liquid droplets, beyond a particular s ize and density, may lose upward momentum and fall from the flame z o n e a s cold d ep o sits or (s o m etim es) a s "burning rain"; 12. Fuel radicals b e c o m e converted to flue g a s e s in the flame zone; 13. U n c o n su m ed fuel pockets migrate to the flame e d g e s and contribute to combustion inefficiency; 14. At the e d g e s of the flame en velope, fuel radicals eventually reach a limiting concentration and/or temperature which inhibits further conversion to flue gas; Art_to_Science - AFRC.wpd 6 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 15. With increasing flame length, accumulating flue g a s e s increasingly im pede diffusion of ambient air into the core of the flame; 16. The overall dilution in cr ea ses with downwind travel of the g a s column, a s a function of atmospheric turbulence and wind speed; 17. Combustion c e a s e s at the end of the flame a s fuel radicals finally reach a limiting concentration and/or temperature which inhibits any further conversion, adding to overall inefficiency; 18. Thermal e m is s io n s from unsymmetrical m o le cu les and black-body particulate produce a z o n e of high radiation which d e c r e a s e s with distance from the flame; 19. C om plete and incomplete products of combustion com b in e with more air in a downwind plume of hot flue g a s e s . Each of the foregoing characteristics is a candidate for individual examination by formula, but conventional flare engineering a d d r e s s e s very few of them directly, and t h o s e which are a d d r e s se d may be covered by an unsophisticated or unscientific approach. Computational Fluid Dynamics (CFD) d o e s attempt to include many of t h e s e characteristics within combustion m od els but, d u e to the d e g r e e of complexity n e e d e d for the matrix, most tend to be included a s simplified parameters which require resolution by direct input of a corrective constant and thus fall prey again to a level of subjective interpretation. B e c a u s e of the haphazard nature of flaring, flame m od els for general industrial u s e do not n e e d to be critically accurate a s long a s they truly reflect specific s e le c te d conditions with a good probability. However, within the industry a s a whole, there are at least twelve different and independent, published flame m o d e ls ( 1 )( 2 ). Each is intended to accurately represent a true flare flame for the p u rp o ses of spacial and safety calculations but yet no two m o d els s e e m to produce the s a m e results for overall s ize and s h a p e of the flame and at least four different ap p ro a ch es are u se d for estimation of radiant output. At the time of writing, there s e e m to h ave b een no attempts m a d e to reproduce, investigate or com pare any of t h e s e flame m od els using CFD or by other independent large s c a le study. The API h a s implicitly en d orsed two of the m eth od s by including them in the com m only u se d reference guide, API R P -5 2 1 (3). Neither model g iv e s the s a m e results and the more c om p lex of the two (Brzustowski - BRZ)( 4 ) is fast b ecom ing a de-facto industrial standard, despite having little theoretical underpinning and generating flame s h a p e s which do not reflect true conditions, insofar a s increasing wind s p e e d s gen era te longer flames, contrary to the ob se rved effects of wind turbulence. Art_to_Science - AFRC.wpd 7 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 Although the flame may b e represented pictorially, a s in Figure 1, and b e satisfactory for general d iscu ssions, an alternative scientific co n c e p t(5) can b e represented by the idealized presentation of Figure 2 which show s, mathematically and theoretically, how g a s e s em erging from an orifice (flare pipe) mix into the downwind atm osp h ere and create the flammable e n v e lo p e around the central g a s column. This provides a generalized flame s h a p e which is largely confirmed by full s c a le flame temperature profiles gen erated by a DGMK( 6 } report, although fluid dynamicists frequently n e e d to break down the analysis to a much smaller s c a le w h en using more localized descriptors of turbulent motion and mixing. Only o n e of the various published m od els a d d r e s s e s the stream continuity of the em er g en t flue g a s e s to en ab le con sisten t downwind calculation of flue g a s concentration and temperature. That model is en d o rsed by this author (Buoyant Flame - BUO)( 7} and derives largely from the standard dispersion equations show n in Figure 2. The BUO model, which is also built around the familiar Briggs(8} plume rise model, g e n e r a te s a variety of intermediate v alu es which are u se d cumulatively to model the flame and plume rise. T h e s e include a value for total dwell time to the end of the flame, which may b e useful a s a parameter in determining probable destruction efficiencies, and sep arate c o m p o n e n ts for the three flame descriptors introduced by G o golek ( 9 \ Buoyancy, Momentum and W ake influence. The com p lete BUO procedure, which is reproduced in Figure 3, is simple to follow but lengthy due to the u s e of a number of parameters which are a function of the concurrent atmospheric stability and which gen era te different flame profiles according to the appropriate stability condition. FLA M M A BILITY O F T H E G A S S T R E A M Flammability of the g a s stream d e p e n d s on the intrinsic characteristics of the g a s composition and is an essen tial a s p e c t of g a s e s relieved to a flare. flammable by definition. All g a s e s known a s fuels are intrinsically Many mixtures relieved to flares however, contain non-flammable c o m p o n en ts and the mixtures th e m s e lv e s cannot alw ays b e guaranteed to b e flammable. The most com m only quoted "rules-of-thumb" in flare engineering relate to the flammability of a lean stream and invoke heating value limits which derive directly from EPA Regulation 4 0 C F R 6 0 .1 8 ( 10 } in which the minimum permissible heat content of g a s mixtures is s e t at 2 0 0 Btu/scf for raw g a s e s (u nassisted flares) and 3 0 0 Btu/scf for a s siste d flares. Art_to_Science - AFRC.wpd 8 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 The original Pohl( 11 )( 12) research leading to the regulation simulated lean g a s e s primarily by diluting propane with nitrogen and the valu es s e le c te d for the s u b s e q u e n t regulation result directly from the ch o ice of t h e s e test g a s e s . The published( 13), flammable properties of the tested mixture (propane / nitrogen) are show n in Figure 4. It is clear from this presentation that, although 2 0 0 Btu/scf allows a d eq u a te flammability for a propane/nitrogen mixture, it is quite inadequate for a propane/carbon dioxide mixture, which probably requires a value closer to 3 5 0 Btu/scf. This inadequacy h a s s in ce b e e n recognized, but unresolved by Pohl et al( 14). D espite their view that the p r e s e n c e of Inert c o m p o n en ts h as a w e a k effect on lower flammable limit, the effect is strong w h en consideration is ex ten d ed to the entire mixture, being generally in inverse proportion to the concentration of flammable g a s in the mixture. B e c a u s e the lean limit concentration, C L, is a flame temperature b a s e d limit, the thermal properties of the diluent and other constituents play an important part in determining the low limit for gas/inert mixtures. S h o r e ( 15 ) p rop osed a unifying property (Nitrogen Equivalent [NE]) providing relationships b a s e d on specific heat at representative temperatures for the co m m on diluents carbon dioxide (1.82), water vapor (1.35), sulfur dioxide (2.1) helium (1.07) and argon (0.65). The characteristic is currently well recognized and the s a m e principle is presently em b od ied in the International Standard - ISO 1 0 1 5 6 ( 16) a s a factor K. More recent works by Molnarne et al ( 17 )( 18 ) have ex p a n d ed t h e s e v a lu es for other diluents and mixtures. This approach allows the e a s y construction of a flammable e n v e lo p e for the generalized c a s e of flammable-inert mixtures, a s show n in Figure 5. For a diffusion flame. the interpretations of minimum mixture flammability must be m ad e from the curves using the points w here the gas/inert ratio lines cross the theoretical (stoichiometric) line. This e n s u r e s that the mixture is able to com bust completely, with s o m e e x c e s s air, at all le ss e r inert ratios. W h en using published flammability limits and e n v e lo p e s in this w ay it is w is e to also include a minimum Factor of S afety (FOS). The implied FOS a s currently u s e d by the EPA regulation (for propane) is 1.5, b a s e d on the extrem e concentration of flammable g a s in nitrogen, and this would s e e m to be just satisfactory, although a FOS of 1.75 would a g r e e with that u s e d by the U S C G ( 19) for ox ygen concentrations, in regulations governing flares for tanker operations. The s u g g e s t e d procedure to determine the minimum practical heating value for a diluted mixture is show n in Figure 6 and representative valu es for various simple mixtures are given in Figure 7. For mixtures with a number of flammable g a s e s , the lean limit of the flammable mixture co m p o n e n ts may be determined using the com m on, Le Chatelier, approach. Art_to_Science - AFRC.wpd 9 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 Applying the specific heat of water to the flammable e n v e lo p e of propane, in this way, gives a limiting flammable condition (FOS = 1.5) of roughly 2 6 5 Btu/scf for propane in H2O. This g iv e s s o m e cre d e n c e to the u s e of the e n h a n c e d EPA minimum value of 3 0 0 Btu/scf for (steam ) a s siste d flares a s d ev elo p e d from a propane b a s e d flame. Generally, mixtures which are already approaching their limiting heating value gen era te relatively low flame temperatures and are not likely to yield large quantities of solid particulate n eeding additional treatment for s m o k e suppression. N on eth eless, w h en t h e s e are flared in Flare tips which are fitted with s m o k e sup p ression capability, the potential additive flow should be considered a s an appropriate inert diluent within the foregoing calculation. However, there is no recognition of the a s s is t steam rates by the current EPA regulation, and this allows a g a s with the regulatory minimum required heat content for a s siste d flares, of 3 0 0 Btu/scf, to be easily overpow ered by steam e v e n with the concurrent and mandatory pilot flame present at all times. Furthermore, the flare operator's desire to hide a flare flame by over-steam ing may com m only lead to unrecognized and inadvertent flameouts, a condition which is indicated by a visible steam plume from the flare, signaling the probable venting of unburned hydrocarbons. efficiency a s steam rates increase. The Pohl reports clearly s h o w an impact on combustion Figure 9 s h o w s this trend for the specific test condition and, whilst the efficiency value here n ever falls below 99%, the inference of a further increase in steam rate is apparent. A s more com p lex ch em icals enter c om m on u sa g e , the n e e d for flare en g in ee rs to understand the interactions of flammable conditions b e c o m e s more significant and the simple procedure described here can be applied generally and with equal con fid en ce to all relieved materials. F L A M E LIFT, S T A B IL IT Y A N D IN E F F IC IE N C Y The m ain ten an ce of a stable flame e d g e or b a s e is essen tial a s a factor in reducing potential em iss io n s of unconverted materials. A lifted flame is a precursor of flame blow-off and total flame loss from a flare can b e a catastrophic environmental event. There is general a c c e p ta n c e that flame stability is s o m e function of discharge velocity and, within the USA, flare engineering practice a d d r e s s e s flame stability largely by using the velocity-limiting formulae which are to b e applied a s a characteristic of the EPA regulations noted previously and d is c u s s e d in Figure 8. The EPA formulae are only applied to g a s e s with a heating value le s s than 1 0 0 0 Btu/scf, being therefore primarily hydrocarbon mixtures which are m ad e lean by the inclusion Art_to_Science - AFRC.wpd 10 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 of inert content in the manner of the original Pohl research. An upper limit of 4 0 0 fps is arbitrarily applied for all other c a s e s and the minimum value n e e d not b e le s s than 6 0 fps. Under the EPA regulation, three sep arate and distinct rules apply a s a function of the style of flare and hydrogen content of the g a s e s . A s t h e s e formulae w ere derived empirically and independently, they have minimal scientific commonality or basis. Furthermore, the rules technically preclude a w hole c la s s of flares in com m on use, which are specifically d e s ig n e d to operate beyond the normal constraints of the regulations. W h en the EPA regulations are not a controlling factor, a completely different s e t of "rules of thumb" are often applied a c r o s s the industry. Following the recom m en dations of API, allowable exit velocities are related to sonic s p e e d in the gas, on the b asis of much earlier papers by Hottel(20} and Kent( 21 ). T h e s e generalizations implicitly a s s u m e a rich hydrocarbon g a s but h ave no codicils w h a ts o e v e r regarding the en ergy content, flammability or dilution of the g a s in question. This seem in gly haphazard application of wildly different ap p ro a ch es to discharge velocity, understandably c a u s e s confusion a m o n g st en g in ee rs w h o are not intimately a s s o c ia te d with the Flare Industry. A s d escribed previously, a large diffusion flame burns a s a quasi e n v e lo p e of combustion around an unburned core of gas. Much of the air u se d for this combustion is entrained around the root of the flame, w here the jet velocity is the greatest and the thermal turbulence is the least. This leads to a disproportionate ratio of air to g a s in the outer e n v e lo p e of the mixture in this zon e. S in ce the stability of the flame relies on its ability to "anchor" at the rim of the jet, the local mixture conditions are critical to flame stability and, if this mixture is diluted beyond its lean flammable limit [CL], the b a s e of the flame will lift aw ay from the jet and attempt to re-establish at s o m e distance a b o v e the jet. This is a condition known a s flame-lift. O n ce the flame is lifted, its stability is questionable and it can, eventually, be sufficiently disrupted that the flame can extinguish in a condition termed blow-off. Localized conditions of flame stability have b e e n exam ined recently by many researchers, most notably by Buckmaster( 22 \ but there h ave b een several other significant stu d ies( 23 )( 24 )( 25 } relating to instability characteristics of the flame e d g e and co-flowing air and g a s streams. There s e e m s to be a g ree m en t that the velocity of flame-lift and blow-off of such stream s can b e described using formulae which typically incorporate discharge velocity, jet diameter and flame sp e e d . Karbasi and W ierzba( 26 } s h o w e d a general pattern of stability for co-flowing g a s and air stream s which clearly indicates the influence of increasing air dilution on flame stability. The pattern of this Art_to_Science - AFRC.wpd 11 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 instability is reproduced in Figure 10. From this representation can be s e e n a relatively low d e p e n d e n c e of the unstable flame conditions on co-flowing air stream velocity, which s u g g e s t s that the governing conditions are related very locally to the flame e d g e and are dominated by the g a s flow itself. A stronger d e p e n d e n c e of the blow-off limit on co-flowing air velocity may b e an indicator of a critical total air volum e and could e v e n possibly be related to the dimensional constraints of the test equipment. A similar pattern is discernible in papers by other resea r ch ers( 27 ). Although the published formulae for blow-out h av e a g ood correlation with test v a lu es in the small scale, the u s e of flame s p e e d a s a parameter, which is not com m only available a s a single value in engineering tables, and an un-m easurable flame diameter factor both com bine to m ake them u nusable for flare design engineering. However, the published results for lift-off of various mixtures by several authors provide a useful cro s s reference to facilitate further investigation. Pohl et al. also considered flame lift in the empirical EPA studies of flame efficiency. That report concluded that the velocity n e e d e d for satisfactory efficiency, together with no lift-off, varied only with g a s en ergy content. However, the Pohl results appear to contain in con sisten cies in that they yield different patterns for the two primary flammable g a s e s in the study, propane and hydrogen sulfide. S o m e of t h e s e results are p resen ted in Figure 11, together with other results and the relevant EPA parameters. There s e e m s to b e no clearly apparent relationship b etw een the various results t h e m s e lv e s or the EPA regulations which are intended to reflect them. A s with the matter of minimum heating value noted previously, t h e s e d iscrepan cies are recognized but unresolved. A careful re-examination of the Pohl data, available in published EPA reports, s h o w s that the variations in flame stability with g a s composition do not fall into the pattern described only by the heating value (as em b od ied in 4 0 CFR 60 .1 8 ) but can b e more easily explained by derivation from the minimum en ergy transfer or temperature function for the flame. This analysis s u g g e s t s that a re-statem ent of stability using the characteristics of the lower flammable limit for the total mixture [ CLM ], and including any diluents in the m anner previously described, may provide a suitably universal parameter to permit prediction of unstable discharge velocity for any g a s or mixture in a flare tip. The analysis in this paper u s e s the m a s s ratio of air to g a s [RAM] at the lower flammable limit of the mixture a s o n e parameter. A sep arate analysis of this parameter against data( 28 ) from hydrocarbon/CO2 mixtures finds a go o d correlation b etw een this and flame s p e e d providing a con fid en ce level for c o n s e q u e n t sup p ression of the flam e s p e e d a s a required input in the manner Art_to_Science - AFRC.wpd 12 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 of the Wierzba formulae. A very similar parameter w a s u se d by J oh n son and Kostiuk(29} in unifying a number of other results for wind-blown inefficiency of various lean mixtures. This analysis also introduces the parameter UO/p/DO (velocity/jet periphery) a s a rearrangement of the Wierzba formula. This d ev e lo p s a relationship with d im en sion s of Hertz and r e s e m b le s the acou stic Ring Frequency (Mach1/p/DO) which defines a critical vertical shedding frequency for stream ed d ie s at the point of disch arge and may, therefore, h ave a bearing on the vertical stability of the flame at the s a m e point. W h en the interpolated Wierzba results for lift-off are plotted together with the modified Pohl results on the s a m e basis, they s h o w a remarkably c l o s e correlation, d espite being performed on different g a s e s and at vastly differing s c a le s. Additional work on propane dilution of hydrogen fla m es by Wu et al( 30 ) displays a similar correlation. All t h e s e results are plotted a s Figure 12. The resulting formula of the com m on line b e c o m e s U o/ p /D o = w here 5 - Hz [Ram / 6 .8 5 ] Uc stable discharge peak velocity feet per s e c o n d Do discharge diameter feet RAM air/gas m a s s ratio at lean limit wt/wt ((100 p OM Pa C LM C l m ) / C LM ) ( p OM / p A ) density of discharge mixture lb/ft3 density of air lb/ft3 lean flammable limit of mixture % The empirical constant [ 6 .8 5 ] may contain e le m e n ts of viscosity and other transport properties of the g a s and air, however, it is clear from the c lo s e correlation of the curve, that variations in t h e s e v a lu es may b e relatively insignificant in the larger s c a le of the problem. A s the b asis of this relationship is the Lower Limit of Flammability [ CL ] a s u se d previously in determination of the minimum rec o m m en d e d heating value, u s e of the s a m e safety factor [ FOS ] s e e m s to b e appropriate. For FOS = 1.5, this would result in a diameter limitation, for an unstabilized flame, b a s e d on the formula w h e re Art_to_Science - AFRC.wpd Do = [ 8 0 5 * qo / (R am)5 ](1'31 qO = volumetric flow - cfs 13 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 The involvement of volumetric flow in this manner raises a question of volumetric normalization. As the conditions of flammability which control this feature will b e dominated by ambient conditions, u s e of published valu es of CLat the conventional temperature of 7 7 O F [25O C] is s u g g e s t e d a s a b asis for calculation of RAM. However, the initially cautious approach to sizing, in the a b s e n c e of contrary indications, would u s e the actual design g a s temperature w hen determining g a s density [ pO ] and sub seq u en tly impacting the diameter in proportion to (TO R)(4/3). The included figure also displays a marker showing a representative refinery fuel g a s and a typical line indicating the condition of Mach 0.5 in a 3 0 inch flare tip, for the various mixtures represented by the points on the graph. It is interesting to note that this marker and the intersection of the two lines ap p ear in the s a m e general stability region, providing a justification for t h o s e prior results which indicate a satisfactory relationship b etw een Mach number and stability for com m on undiluted hydrocarbons. Clearly, a s this result is derived from interpolated results, additional independent verification by other research is desirable and is strongly encouraged. There are several a s so c ia te d conditions surrounding this specific result which should b e reproduced in further work - the condition represented is flame lift/instability, not blow off; - the flame is not stabilized by active or p a ss iv e devices; - no pilot or continuous re-ignition is present; - the downward propagation limit should be used; - mixtures with inert g a s content should b e corrected using the Nitrogen Equivalent (NE) to calculate the true value of CLM_DOWN for the entire mixture. The potential predictive ramifications of this relationship are significant: - large flares may be operated at higher velocity than small flares on the s a m e gas; - many large flares already in the field can be c ross referenced for suitability, overcoming the n eed to re-permit under restrictive regulations; - a pattern of air entrainment at the root of the flam e may e m e r g e to permit formulation of the performance of so n ic or other high pressure flares; - g a s e s rich in hydrogen may b e flared at higher velocities without requiring oversized discharge; - by relating ed d y s ize or shedding frequency to stability, a pattern and additional formulae may e m e r g e for other typ es of flame stabilization such a s su d d en exp an sion and bluff body; Art_to_Science - AFRC.wpd 14 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 - there may b e a sep a ra te relationship (additional formula) b a s e d on the differences b etw een flammable limits for upward and downward propagation which d escrib e s piloted flam es (upward propagation) v e rsu s un-piloted flam es (downward propagation) - by extension, the latter may define a further formula to determine the minimum number of external flame s o u r c e s or pilots to properly stabilize an intrinsically unstable flame. Many of the a b o v e points would also s e e m to b e intuitively correct and figure 13 p resen ts many of t h e s e characteristics pictorially. If the approach can b e e xten d ed to piloted and stabilized flames, then calculation of a working value of Ram, for stability purposes, m ay involve F la m m a b le Limits for D o w n w a r d v s Upw ard P r o p a g a tio n . T h e s e may not alw ays b e published, requiring an estim ate to b e m ad e by the engineer. In t h e s e c a s e s the practical approach would be to consider the given value to b e CLUP. The relativity of CLUP and CLDOWN is a function of a g a s e s propensity to re le a s e radiant transient solids into the flame and is, therefore, intimately c o n n ected to the chemistry of the compound. B a s e d on a very m o d est regression analysis of relatively limited published data for directional limits of com m on lean g a s e s , it would s e e m that a rough approximation of the ratio may b e obtained from the formula - CL_UP / C L_DOWN w here = ZUD = 1.2 - C/5 + H/30 + S/5 + N/10 + O/4 >= 1 C, H, S, N and O are the number of atom s of carbon, hydrogen, sulfur, nitrogen and oxygen respectively in a representative m olecule of the original g a s mixture of flammable com p on en ts. Empirical results for flammable limits s e e m to be apparatus d e p en d en t and can vary by a s much a s +/- 20% or 30%. A more carefully studied review of the true stability properties of g a s e s is rec om m en d e d for definitive or critical results. Complex d isch arges may require special analytical treatment w h en using the RAM approach. For exam ple, m ost A ir A s s i s t e d F l a r e s may be generally described a s an exterior duct carrying a flow of fan-blown air, with an interior g a s distribution nozzle which r e l e a s e s the g a s to b e flared into the air stream in the vicinity of the b a s e of the flame. A s there is no standardization of g a s nozzle d esign or flare exit conditions a c r o s s the industry, the flame properties and stability will undoubtedly vary a c r o s s the available commercial m od els and relative duct size. To coincide with the argum ents of this model, for a practical design, the blown air n e e d s to be considered a s a variable feature of the design procedure a s it is often in the direct and independent Art_to_Science - AFRC.wpd 15 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 control of the plant operator. Two different operating s ta te s may b e relevant for individual consideration; - the blown air is greater than the required ratio to produce a CL (lower limit) concentration of g a s in air, resulting in flame co llap se and stabilization on the inner nozzle; - the blown air is le s s than the required ratio to produce a CU (upper limit) concentration of g a s in air, such that it acts a s a diluent in the g a s stream, with stabilization on the outer diameter. A s the mixture transitions b etw een t h e s e two sta te s there will a lso b e two points of discontinuity, o n e of which may possibly relate to the blow-off condition of Figure 10. C A R B O N A N D S O O T FO R M A T IO N , A N D R A D IA N T L O A D A great deal of the d esign effort a s s o c ia t e d with flares involves attempting to s u p p r e s s particulate carbon formation in the flame. Initially, this is for the obvious visual effect and a s so c ia te d community relations, but is also to facilitate com pliance with Environmental standards which correctly u s e s m o k e production a s a marker for other inefficiencies. Previous papers presented to AFRC h av e indicated the occurrence of h ea vy aromatic pollutants from smoking flames. Simplistically, carbon forms initially in a hydrocarbon flame b e c a u s e of the thermal breakdown of the hydrocarbon m olecule (pyrolysis) into com p on en t radicals. W hen this occurs in the core of the flame, and in the a b s e n c e of oxygen, s o m e of the carbon atom s agglom erate into larger crystalline forms such a s chains or sp h er es. T h e s e forms are solid in nature and must d e c a y physically, from the outer surface toward the center, w h en the solid form m e e t s an oxygen-rich interface which permits the eventual conversion to CO2. During this time, the particulate acts rather like a thermal black body at constant temperature, and provides a highly radiant contribution of thermal en ergy within and from the flame zon e. A much smaller radiant contribution, will c o m e from the unsymmetrical m olecu les in the flue gas, primarily CO2 and H2O, and probably le s s than 10% of the thermal load of the flame will be radiated in this manner. At this time there is no com m on a g ree m en t about the m eth od s n e e d e d to calculate solid carbon fraction within the flame, nor its radiant effect. Schwartz( 31 } a ck n o w le d g e s this in a 1 9 9 6 review through a recomm endation against using similar radiative fractions for alternative flame models. Adding to the accumulation of "flare lore", radiant heat (em issive) fraction is treated more like a marketing tool than a scientific condition, leading to a significant amount of inconsistency and confusion a c r o s s the Industry. Art_to_Science - AFRC.wpd 16 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 The conditions which g en era te solid carbon fraction within the flam e z o n e are not d is c u s s e d in detail here but must finally involve - a time b a s e covering the oxygen-deficient particulate growth; - a time b a s e covering the d e c a y in the z o n e of o x ygen availability; - the potential carbon loading attributed to a single representative g a s molecule; - the intrinsic thermal stability of the original molecular compound; - the free en ergy potential of the m olecule w h en com pared with that of molecular carbon. T h e s e id eas remain to b e d ev elo p e d in a future paper. C O N C L U S IO N This paper primarily d i s c u s s e s just three major a s p e c t s of flare and flame modeling and presents e a c h in a manner which m ay b e recognizable and useful to m ost engineers, rather than a s d im en sio n le ss an alyses. The flammability and stability m o d els which have b e e n proposed in this paper are more co m p reh en siv e and involved than the basic m od els currently published by the API and EPA (as part of 40C F R 60.18). It is believed that t h e s e n ew m od els more accurately describe the minimum requirements for a satisfactory and stable flame which m eet the current e m iss io n s standard for flares s e t by the EPA. T h e s e n ew m o d els do not a d d ress whether a standard of 98% conversion is appropriate for the maximum design c a s e , which is an ongoing d eb a te to b e d is c u s s e d further in a s u b s e q u e n t paper. On the o n e hand, there s e e m s to b e e v id e n c e within the public domain which could support a standard of 99% conversion, for a variety of operating conditions, particularly in concert with increased con fid en ce in a stability model. On the other hand, s o m e resear ch ( 32 } ap p ea rs to s h o w reduced efficiency for wind blown flam es w h en acco m p a n ie d by a w a k e influenced condition (downwash). Further examination of t h e s e i s s u e s will n e e d to build on the s a m e flammability co n cer n s which form the b asis of this paper. The author a ck n o w le d g e s that the developm ent of s o m e a s p e c t s of t h e s e current m od els are b a se d on limited research data and, in many w ays, represent h y p o th e s e s rather than results. In proposing the m od els here, the primary intentions are - to highlight confusing d iscrepan cies in the various published documentation and regulations and, hopefully, correct th o s e inaccuracies (perceived by this author) which give rise to Global m iscon cep tions of the true requirements for satisfactory flaring of g a s mixtures; Art_to_Science - AFRC.wpd 17 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 - to en c o u r a g e the Flare Industry a s a w hole to take a more scientific position on the technical is s u e s surrounding the flare business; - to provide ideas which may stimulate additional research into flame size, stability, lift off and a s so c ia te d combustion inefficiency by other researchers. Such research is strongly en co u raged by the author, in the interests of improving a general level of technical know ledge and building a stronger technical foundation for future d evelopm ents. Art_to_Science - AFRC.wpd 18 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 T he A u th o r David S hore h as b e e n designing and working with Flares and Combustion sin ce 1 965 and is presently Chief Engineer of Flaregas Corporation in N.Y. USA. Mr. S hore served a s a m em b er of the original steering com mittee responsible for the research behind the current EPA regulation 4 0 CFR 6 0 .1 8 and h a s also worked in Europe a s a Chartered Engineer (Eur-Ing) w here he w a s also a s so c ia te d with Flare research by the DGMK. Mr. S h o r e 's previous papers include presentations to A.I.Ch.E and Noise.Con. A c k n o w le d g m e n ts T h e author ack n ow led ges, with thanks and appreciation, the contribution and a s s i s t a n c e of family and the M an agem en t of Flaregas Corporation in facilitating the preparation and presentation of this paper. Art_to_Science - AFRC.wpd 19 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 Art_to_Science - AFRC.wpd 20 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 How a Flame Envelope establishes b y a tm o s p h e r ic d iffu s io n / d is p e r s io n R 0 = 2 0 V [ -2 * ( c r 2) * l n { x * 2 * 7 i * ( a 2 )} ] 4 0 6 0 L E L 8 0 1 0 0 U E L F ig u re 2 - A th eo retical F la m e E n v e lo p e Art_to_Science - AFRC.wpd 21 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 BUO - BUOYANT FLAME PROCEDURE CD 3 hfi i-j- ] e * 5. 1 1 [* = At the end-of-flame location XF; for conventional verticalflares; units to be consistent Btu; Ibm;ft; OR sec a) determine theflow characteristicsfor the discharged gas qF = heatflux CV = calorific value Po = density Cl = leanflammability limit =flame emissivity e b) determine the physical characteristics oftheflare Hs = stack height Ds = stack top outer diameter rO = inner radius c) determine the ambient conditions Ua = wind speed Ta = temperature Pa = density cpa = specific heat d) determine the atmospheric conditions Stability category varies A / B / C / D default = D S = varies 0.039 / 0.0459 / 0.0318 / 0.0142 N = varies 2.0016/1.8580 /1.7973 /1.7727 e) determine the terrain conditions Ho = wind speed reference height Y = wind height exponent (varies; 1/5 to 1/11.5 refrelevant wind code) default =1/ 9.5 d) estimate or calculate corrected values of Xc = down wind travel to midflame Hc = midflame height = wind speed at stack height for Zf = Hs/ Ho Uas = ZfY* Ua = wind speed at midflame Uac = ZcY* Ua for zc = Hc/ Ho = S * X f 2 = stability constant at midflame Sx = 32.174 = gravitational constant g = buoyancy constant Kb = 1.6 = momentum constant Km = 2.3 = stability constant Ks = [ 1/ (2 * p *Sx) ] =flame constant Kf = CV *Po * Cl =flame dwell time tF = [ ( qF *Ks) / ( Kf * Uac3) ]™ = downwindflame length Xf = [ Uac *tF] = true discharge velocity Uo = [ ( qF/ ( CV *p o * P *ro2) ] = downwash velocity modifier = [ ( 2 * p a / p o W * U a s ) ] Ud = modified discharge velocity U01 = [ Uo- Ud] = density buoyancyfactor Fb = [ 1- ( Po/ Pa) ] * Uo1 *ro2 = thermal buoyancyfactor = [ g / ( P * Cpo * T a * P a ) ] Fh =flame buoyancyfactor Ff = momentum factor Fm = [ ( PO/ PA) * U012 *ro2] = downwash estimate = [ 5 * D s * U q 1 / U d ) ] <= 0 DHd = buoyant rise offlame DHF1 = { 0.5 *Kb *[ ( Ff(1/3) * qJ1/3)) / Uaf] *Xj2/3)} = momentum rise DHm = [ Km *[ F jI/3)/ Ua/ 2/3)] *X j1/3)] = total rise DHf = h + dHm+ dHd Down wind plume rise in any condition is given by dhp = { DHP1 or AHP2 or DHP3} + DHm + DHd AHP1 = { Kb *[ (Fj1/3> / Uas] *X/2/3)} = buoyant rise; raw gas ; density base = buoyant rise; hot gas; thermal base DHp2 = { Kb *[ ( Fh(1/3) *q(1/3)) / Uaf] *Xj2/3)} = buoyant rise; flue gas beyondflame DHf3 = [ DHp2 - H X f} ] = height correction atflame end DHeX f} = DHp2{Xf} - DHf{Xf} Art_to_Science - AFRC.wpd 22 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 F l a m m a b i l i t y E n v e l o p e s f o r P ro p a n e / I n e r t M i x t u r e s in r e f : F ig 3 2 : B O M B u lle tin 5 0 3 0 10 20 30 40 50 A\r 60 % I n e r t in O r ig in a l M i x t u r e F ig u re 4 - P rop an e F la m m a b ility Art_to_Science - AFRC.wpd 23 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 Figure 5 - Generalized Flammability Art_to_Science - AFRC.wpd 24 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 FLAM M A BILITY AND NITRO G EN EQUIVALENCE R s R ll R il = Lean Flammable Limit fo r Flammable component in Air = Air / Gas Stoichiometric Ratio = Air / Gas ratio at Lean Limit = Limit Ratio o f Inert/Flammable in N2 (or w ' xs Air) N e = Combined Nitrogen Equivalent o f the Inert content Cl = (100 - C l ) / C l =R ll - R s = S { NEi * I%I} =fra ctio n a l quantity o f ith In ert com ponent % N eI = nitrogen equivalent o f ith In ert com ponent F OS R % ix C V VF = Factor o f Safety = Limit Ratio o f Inert/Flammable in true mixture = Combined Calorific value (vol) o f Flammable components C V vm = m in flam m able m ixture heating value (vol) = R il / N e / F OS = C V vf / ( R ix + 1 ) F ig u re 6 - N itro g e n E q u iv a le n c e REPRESENTATIVE LIMITINGHEATING VALUE FOR SOME GAS/ INERTMIXTURES CALCULATED USING STANDARD FLAMMABLE LIMITSAND NITROGENEQUIVALENCE OFINERT propane / nitrogen propane / water propane / carbon dioxide propane / sulfur dioxide propane / helium propane/ argon 200 Btu/scf 265 Btu/scf 340 Btu/scf 385 Btu/scf 135 Btu/scf 215 Btu/scf methane / nitrogen ethylene / water ammonia / CO2 hydrogen sulfide / SO2 hydrogen / helium carbon monoxide/ argon 140 Btu/scf 160 Btu/scf 250 Btu/scf 105 Btu/scf 25 Btu/scf 115 Btu/scf Uses FOS =1.5; LELfor upwardpropagation onlyfrom BOM503; ambient temperature properties Figure 7 - M inimum Flammability o f M ixtures Art_to_Science - AFRC.wpd 25 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 C O M M E N T A R Y O N T H E E P A R E G U L A T IO N - 40 C F R 60.18 The well-known and widely discussed EPA Regulation, 40 CFR 60.18, (and 63.11) is one example o f a generalized parametric input intended to assist in improving the global performance o f Flares by setting limits on volumetric heating value o f gases which may be flared and on gas discharge velocity, thereby controlling flare size (in the U.S.A.). Insofar as the E P A 's purpose is to regulate and monitor emissions, the associated control o f flare tip diameter suitably focuses the flare user's attention on flare design. However, a major concern with the EPA regulation is that the formulation o f the rules fo r heating value and velocity give an impression on a Worldwide scale that, because the formula is implicitly endorsed by the U.S. Government, it properly describes the flame functions. In consequence, the various limits which are set by the regulation on heating value and velocity tend to be universally applied as a "rule-of-thumb" and with a high level o f confidence w hich m ay be unw arranted. The underlying concept o f this regulation considers that the combustion efficiency o f any stable, active flame will be more than 98% and that the intrinsic heating value o f gases and mixtures can describe both a lowest value to support any flame and the upper velocity able to avoid flame lift-off. Compliance with the regulation is deemed to imply a conversion efficiency o f 98% or better which then allows flare operators to evaluate their probable output o f emissions with some reasonable accuracy. Although subsequent investigations lead to a conclusion that the base value o f 98% may need review, particularly fo r wind blownflames, the regulation stands, and some o f its basic characteristics are included in the recently issued API Recommended Practice 537(333 and are likely to be included in the proposed new standard ISO 25457(34) The EPA formulae resolve to fo r non-assisted and steam assisted flares, UO1 = 26.6 * 1 0 (8507CV) (min = 60 fps; max = 400fps) except fo r gases with a Hydrogen content greater than 8%, which may be calculated from UO2 = 12.8 * (H2 vol% - 6) (max = greater o f 122 fp s or UO1) fo r air assisted flares UO3 = 28.6 + 0.0867 * CV (min = 60 fps; max = 400fps) where UO# =free discharge velocity - fe et per second CV = volumetric heating value - Btu/scf (non asisted flares min = 200 Btu/scf) (assisted flares min = 300 Btu/scf) F ig u re 8 - 4 0 C F R 6 0 .1 8 Art_to_Science - AFRC.wpd 26 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 Efficicncy INFLUENCE OF 5TEAM ON GLOBAL EFFICIENCY Mass ratio of Steam / Gas Figure 9 - Steam and Combustion Efficiency Art_to_Science - AFRC.wpd 27 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 S T A B IL IT Y P A T T E R N F O R A D IF F U S IO N F U M E JET D IS C H A R G E V E L O C IT Y (A F T E R K A R B A S I A N D W IE R Z B A ) C O - F L O W I N G A I R S T R E A M V E L O C IT Y AIR AXIS INTERVALS MAY BE APPROX 0.01 * JET VELOCITY INTERVALS Figure 10 - Flame Stability Art_to_Science - AFRC.wpd 28 of 33 C a lo r if ic V a lu e - B tu /sc f Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 Figure 11 - EPA Flare Stability Curves Art_to_Science - AFRC.wpd 29 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 C o n d itio n s fo r S ta b le D iffu sio n F la m e Uo / ( Pi * Do ) - H z E P A -6 0 0 / 2 -8 4 -0 9 5 :2 -3 & 2 -8 5 -1 2 3 :2 -4 A i r / G a s R a tio a t L e a n Limit - w t/w t eb C 3 /N 2 - Typ M ach 0.5 a H 2 S /N 2 * W ie rzb a C 1 /H 2 Typ mid range H /C ■ Kalghatgi C 3 /H 2 F ig u re 12 - A ir M as s ra tio relatio n sh ip Art_to_Science - AFRC.wpd 30 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 F ig u re 13 - L e a n L im it F la m e S ta b ility Art_to_Science - AFRC.wpd 31 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 B IB L IO G R A P H Y 1 Guigard, S.E., Kinderzierski, W.B., and Harper, N.; "Heat Radiation from Flares"; Alberta Environment, May 2000. 2 Institut National de I'Environment Industriel et des Risques, "Formalisation du savoir et des outils dans le domaine des risques accidentels (DRA-35) - Feu Torche", June 2003 3 Recommended Practice RP-521; "Guide for Pressure-Relieving and Depressuring Systems"; American Petroleum Institute, Washington. D.C. 4 Brzustowski, T.A., & Sommer, E.C. Jr.; "Predicting Radiant Heating from Flares"; Proceedings - Division of Refining, Volume 53, pp 865-893; American Petroleum Institute, Washington. D.C; 1973. 5 Pasquill, F.; "The Estimation of the Dispersion of Wind borne Material"; Meteorological Magazine. 90, 1063, 33-49, 1961. 6 Siegel, Dr.-Ing. K.D.; "Entwicklung schadstoffarmer Industiefackeln"; Deutsche Gesellschaft fur Mineralolwissenschaft und Kohlenchemie E.V.; Teilvorhaben 135-02; 1980 7 Shore, D; "A Proposed Comprehensive Model for Flare Flames and Plumes"; A.I.Ch.E, Proceedings o f 40th Loss Prevention Symposium, 2006. 8 Briggs. G.A.; "Plume Rise"; U.S. Atomic Energy Commission, 1969 9 Gogolek, P.E.G. & Hayden, A.C.S.; "Wind Turbulence and Elevated Flare Flames", Presentation to the American Flame Research Committee, 2004. 10 Code of Federal Regulations; U.S. Environmental Protection Agency; Section 40 CFR 60.18 11 Pohl. J.H., Payne. R., and Lee.J; "Evaluation of the Efficiency of Industrial Flares: Test Results"; U.S. Environmental Protection Agency; EPA-600 / 2-84-095 12 Pohl. J.H., and Soelberg. N.R.; "Evaluation of the Efficiency of Industrial Flares: H2S Gas Mixtures and Pilot Assisted Flares"; U.S. Environmental Protection Agency EPA-600 / 2-86-080 13 Coward, H.F., and Jones, G.W.; "Limits of Flammability of gases and Vapors"; U.S Bureau o f Mines Bulletin 503, 1952. 14 Pohl, J., Gogolek, P., Seebold, J., and Schwarz, R.; "Fuel Composition Effect on Flare Flame Efficiency", Presentation to the American Flame Research Committee, 2004. 15 Shore. D; "Making the Flare Safe"; A.I.Ch.E, Proceedings o f 30fh Loss Prevention Symposium, Paper 12d. 1996. 16 ISO 10156; "Determination of Oxidizing ability of Toxic and Corrosive Gases and Gas Mixtures"; International Organization for Standardization; 2005. 17 Schroeder, V., and Molnarne, M.; 'Flammability of Gas Mixtures: Fire Potential"; Journal o f Hazardous Materials A121; 37-44; 2005. Art_to_Science - AFRC.wpd 32 of 33 Improving Flare design - From Art to Science David Shore / AFRC-JFRC 2007 18 Molnarne, M., Mizsey, P., and Schroeder, V.; 'Flammability of Gas Mixtures: Influence of Inert Gases"; Journal o f Hazardous Materials A121; 45-49; 2005. 19 Code of Federal Regulations; U.S. Coast Guard; Section 33 CFR 154. 20 Hottel. V.O., & Luce, R.G.; "Burning in Laminar and Turbulent Fuel Jets"; Fourth Symp (intl) on Combustion; Combustion Institute; 1953. 21 Kent, G.R.; "Practical design of Flare Stacks"; Hydrocarbon Processing and Petroleum Refiner, Aug 1964. 22 Buckmaster, J.; "Edge Flames"; Progress in Energy and Combustion Science 28, 435-437; 2002. 23 Wu, Y., Al-Rhabi, I.S., Lu, Y., and Kalghatgi, G.T.: "The Stability of Turbulent Hydrogen Jet Flames with Carbon Dioxide and Propane addition"; Fuel (2007) 24 Lyons, K.M.; "Towards and understanding of the Stabilization Mechanisms of Lifted Turbulent Jet Flames"; Progress in Energy and Combustion Science, 33, pp 211-231, 2006. 25 Watson, K. A.; "Experimental Studies on the Leading Edge and Local Extinction in Lifted-Jet Diffusion Flames"; Doctorate Thesis, North Carolina State University, Mech Eng. May 2002. 26 Karbasi, M., and Wierzba, I.; "The Effects of Hydrogen Addition on the Stability Limits of Methane Jet Diffusion Flames"; Int J. Hydrogen Energy; 23-2, pp 123-129; 1998. 27 Cha, M.S., and Chung, S.H.; "Characteristics of Lifted Flames in Non-Premixed Turbulent confined Jets", 26th Combustion Symposium. The Combustion Institute, pp 121-128, 1996. 28 Jabbour, T.;"Classification de l'inflammabilite des fluides frigorigenes basee sur la vitesse fondamentale de flamme"; L'Ecole des Mines de Paris ; 2004. 29 Johnson, M.R.; and Kostiuk, L.W.; "Effects of a Fuel Diluent on the Efficiencies of Jet Diffusion Flames in a Crosswind"; Combustion Institute (Canada), Spring meeting, 1999 30 Wu.Y., Lu, Y., Al-Rhabi, I.S., and Kalghatgi, G.T.; "The Stability of Hydrogen and Hydrocarbon Fuel Jet Flames", Proc. European Combustion meeting, 2007 31 Schwartz, R.E., and White, J.W.; "Flare Radiation Prediction: A Critical review"; A.I.Ch.E, Proceedings o f 30fh Loss Prevention Symposium; 1996. 32 Kostiuk, L.K., Johnson, M.R., and Thomas, G.; "Flare Research Project, Final Report"; University o f Alberta, 2004 33 Recommended Practice RP-537; "Flare Details for Refinery and Petrochemical Service"; American Petroleum Institute, Washington. D.C. 34 ISO 25457; "Flare Details for Refinery and Petrochemical Service"; International Organization for Standardization; Art_to_Science - AFRC.wpd 33 of 33