Combustion of mixtures: A modified IFC approach

While most studies on flares have been done with single hydrocarbon fuels or binary mixtures, the reality of most industrial flares is that they see different mixtures of many gases. Over the past 5 years the International Flare Consortium (IFC) has been conducting research on just this issue. In a recently published literature review, the IFC has explored approaches that have been used to describe the impact of vent gas composition and assist gas ratio on flare combustion efficiency. LeChatelier's Principle has been widely used to estimate the flammability limits (lower and upper) of mixtures of flammable gases. However, LeChatelier's Principle must be slightly modified when one or more of the gas components are inflammable or inert. This paper discusses some minor suggested modification of the reduced steam volume fraction parameter outlined in the IFC Literature Review, then discusses how this approach can be used to describe industrial flare operation, and the impact of assist gas ratio on the observed combustion efficiency. Finally, this paper will present the analysis of combustion efficiency data collected from 7 recent flare test campaigns and illustrate the ability of this modified IFC RSVF method to describe the results obtained. A hypothesis will be offered as to how this approach might be used to predict the performance of other flares, given additional data.

Combustion of Mixtures: A modified IFC Approach Gary R. Mueller, Harshit Agrawal, Chris Caico, and Ronald Olivacce Shell Global Solutions (US) Inc. Abstract While most studies on flares have been done w ith single hydrocarbon fuels or binary mixtures, the reality of most industrial flares is that they see different mixtures o f many gases. Over the past 5 years the International Flare Consortium (IFC) has been conducting research on just this issue. In a recently published literature review, the IFC has explored approaches tha t have been used to describe the impact of vent gas composition and assist gas ratio on flare combustion efficiency. LeChatelier's Principle has been widely used to estimate the flam m ability limits (lower and upper) of mixtures of flammable gases. However, LeChatelier's Principle must be slightly modified when one or more of the gas components are inflammable or inert. This paper discusses some m inor suggested modification of the reduced steam volume fraction parameter outlined in the IFC Literature Review, then discusses how this approach can be used to describe industrial flare operation, and the impact o f assist gas ratio on the observed combustion efficiency. Finally, this paper will present the analysis of combustion efficiency data collected from 7 recent flare test campaigns and illustrate the ability o f this modified IFC RSVF method to describe the results obtained. A hypothesis will be offered as to how this approach might be used to predict the performance of other flares, given additional data. Combustion of Mixtures: A modified IFC Approach Gary R. Mueller, Harshit Agrawal, Chris Caico, and Ronald Olivacce Shell Global Solutions (US) Inc. While most studies on flares have been done w ith single hydrocarbon fuels or binary mixtures, the reality of most industrial flares is that they see different mixtures o f many gases. Over the past 5 years the International Flare Consortium (IFC) has been conducting research on just this issue. In a recently published literature review, the IFC has explored approaches tha t have been used to describe the impact of vent gas composition on flare combustion efficiency.1 LeChatelier's Principle has been widely used to estimate the flam m ability limits (lower and upper) of mixtures of flam m able gases. However, LeChatelier's Principle must be slightly modified when one or more of the gas components are inflammable or inert. Equation (1) has been reproduced from the IFC literature review and displays the form ula th a t can be used to estimate the flam m ability lim it of a mixture of flammable gases with nitrogen, where nitrogen is the nth component of a mixture of flammable gases. Equation (1) is developed from the identity that fo r an inert, F M = <», therefore the xM/F (n) term in the sum goes to zero. F(rnix) Z "1 =1i * ( 0 F ( 0 Equation (1) Other researchers have noted th a t not all inert species exhibit the same combustion behavior as nitrogen when mixed w ith flammable species. Zabetakis2 has produced combustion flam m ability diagrams fo r many hydrocarbons, expressed as percent fuel versus percent inert fo r mixtures w ith air. Zabetakis has developed these diagrams fo r both nitrogen and carbon dioxide as the inert. Figure 1 reproduces the combustion diagram fo r methane developed from Zabetakis' work. One can see that carbon dioxide has a more limiting impact on combustion than does nitrogen. The IFC literature review discusses the construction of simplified combustion diagrams fo r vent gases o f varying compositions using Equation 1 to produce the combustion properties of a pseudo-fuel developed from the composition of the vent gas mixture of interest. From Figure 1, one can see th a t the combustion diagram fo r any mixture of flam m able species can be approximated by the trapezoid form ed using the follow ing computed properties of the vent gas mixture: LFLmix, UFLmix, Imix*, LFLmix*, and UFLmix*. For any vent gas composition, Equation (1) can be used to estimate LFLmix, UFLmix, LFLmix*, and UFLmix* . Using the IFC approach, outlined in th e ir Literature Review, the stoichiom etric air needed fo r combustion, A, must also be estimated fo r any mixture to be able to construct the function Cst in Figure 1, which is the fuel concentration in a stoichiometric mixture of air in the pseudo-fuel/inert mixture. Additional mixing rule are needed to compute the values of Amix and Im ix * 1Gogolek, P., et. al. "Emissions from Elevated Flares - A survey of the Literature." submitted to Progress in Energy and Combustion Science, Elsevier Ltd. (expected publication date early 2011). 2 Zabetakis, M.G. "Flammability Characteristics o f Combustible Gases and Vapors," Dept. of Interior Bureau of Mines Bulletin 627, (1965). . Equation (2) can be used to determ ine the stoichiometric quantity of air needed fo r any mixture of pure components. Amix= £ x ( 0 A( 0 Equation (2) While Equation (2) was initially used by the author to also estimate Imix* , it was found to produce spurious results when the pseudo-fuel mixture contained large amounts of a second inert (e.g. nitrogen). Therefore, Equation 3, as recently proposed by Gogolek3 , has been used to estimate Imix* fo r all pseudo-fuel mixtures. I (m ix ) * = F ( m ix ) Equation (3) One should note th a t Equation (3) will simplify to Equation (2) fo r vent gas mixtures that do not contain significant amounts o f other inerts, and therefore these methods give similar results fo r all but a few lim iting cases. In order to construct idealized flam m ability diagrams from the empirical pure component data, a unique value o f I* must be defined fo r each pure component. A way to uniquely define this value was not overtly presented in the IFC Literature Review. For this w ork the value of I* has been uniquely defined as the value of the inert present at the point of tangency drawn from the origin of the flam m ability diagram to the flam m ability curve. Similarly, LFL* and UFL* represent the fuel concentration at the lower and upper intersection o f a vertical line constructed at I* w ith the flam m ability diagram. The utility of this definition will become more obvious later as we discuss flare operation and how flare operating points are represented on these idealized flam m ability diagrams fo r mixtures of vent gases (pseudo-fuels). By constructing these idealized flam m ability diagrams, one can begin to assess the limiting quantity of assist gas th a t can be added to a given vent gas (pseudo fuel ) mixture, such th a t combustion can no longer be supported (e.g. a nonflammable mixture results). Note th at because each inert has a unique impact on flam m ability, these idealized flam m ability plots must be constructed fo r the predom inant inert. For steam-assisted flares this inert is steam. The impact of smaller amounts of other inerts (e.g. nitrogen or carbon dioxide) can be incorporated by using Equation (1) to include them in the flam m ability limits calculated fo r the pseudo-fuel mixture (i.e. the flare vent gas). While Equation (1) holds fo r nitrogen as the inert, each x(i) term in Equation (1) must be modified to incorporate the impact of other non-nitrogen inerts in the vent gas on the flam m ability limit. Equation (4) illustrates the form of Equation (1) needed to account fo r non-nitrogen inerts3. ! ( ! - ) = Sf='i1[^ 0 / ( 1 + I k=inerts(! e - 1 ) X ( k ) ) ] C ^ ) Eq uation (4) For all of the flam m ability diagrams so constructed, the identity, 100= %fuel + %air + %inert, is valid. While some of the vent gases presented in the follow ing data analysis did contain small quantities of carbon dioxide and water, the quantities o f these other inerts were generally less than one volume 3 Personal communication Dr. Peter E. G. Gogolek, Scientist, C A N M ET Energy Technology Centre Ottawa, Energy Technology and Programs Sector, Natural Resources Canada, Government of Canada, June 20, 2011 percent in the vent gas. Therefore, in the subsequent data analysis presented here, Equation (1) was used, not the more generally applicable Equation (4). While considerable empirical data exists fo r nitrogen and carbon dioxide as the inert gas2,4 much less data exists fo r steam as the inert. While previous works are sufficient to demonstrate that the impact of steam as the inert will be somewhere between nitrogen and carbon dioxide, a unique expression for estimating where steam falls on this continuum has not been definitively defined. Based upon discussions w ith IFC principal investigator Dr. Peter Gogolek, the impact o f steam on flam m ability has been estimated here by using an Isteam* value th a t was estimated by giving the IN2* a 30% w eight and ICO2* value a 70% w eight (e.g. the Isteam* value is closer to th a t o f CO2 than it is to N2). 4 Molnarne,M., P. Mizsey, and V. Schroder, "Flammability o f gas mixtures Part 2: Influence of inert gases." Journal of Hazardous Materials, A121, pp. 45-49. ADDED INERT, v o lu rn a-p ercen t Figure 1. Estimated Flammability Diagram for Methane with Nitrogen and Carbon Dioxide1 Values fo r LFLi, UFLi, Ii*, LFLi*, and UFLi* fo r pure component hydrocarbons were estimated from the work of Zabetakis2 and have been tabulated in Table 1. Using these concepts, flam m ability diagrams fo r tw o lim iting vent gas compositions encountered during recent flare testing were developed. Figure 2 represents a high hydrogen content vent gas (pseudo-fuel); while Figure 3 represents a high nitrogen vent gas. One can also see from Figures 2 and 3 th a t the different vent gases represented have considerably different critical steam quantities. T a b le 1. V a lu e s o f I * , LFL * , an d UFL * e s t i m a t e d f r o m Z a b e t a k i s 2 Component 1 N2 1 1 CO2* Hydrogen Carbon monoxide 69.5 52.8 55.0 Methane * Est. * Est. * LFL co2 LFLH2O UFL n2* UFL co2 57.8 4.1 5.1 4.8 7.3 12.5 11.0 36.9 42.4 13.4 17.6 16.3 18.1 28.3 25.2 35.0 22.0 25.9 6.2 6.7 6.6 7.6 8.2 8.0 Ethane 42.1 30.1 33.7 3.3 4.1 3.9 4.1 4.6 4.5 Propane 39.4 26.7 30.5 2.7 3.5 3.3 3.5 4.4 4.1 Butanes 36.5 25.7 28.9 2.2 2.7 2.6 3.4 3.9 3.7 Pentanes 40.5 24.8 29.5 1.9 2.1 2.1 2.7 4.0 3.6 Hexanes 40.0 26.4 30.5 1.6 1.9 1.8 2.7 2.9 2.8 Hexane+ 36.8 25.7 29.0 0.7 0.8 0.8 1.3 1.4 1.3 Ethylene 46.9 31.0 35.8 3.4 2.8 3.0 5.9 8.4 7.7 Propylene 39.6 26.0 30.1 2.5 3.3 3.1 3.9 4.7 4.5 Butenes 41.0 28.4 32.1 2.2 2.6 2.5 3.5 4.0 3.8 Pentenes 42.4 28.7 32.8 2.0 2.3 2.2 2.9 3.3 3.1 Acetylene 64.7 48.9 53.6 2.6 2.6 2.6 4.0 14.7 11.4 Benzene Hydrogen sulfide 39.1 24.6 29.0 1.5 2.0 1.9 3.1 4.0 3.8 25.3 28.9 5.3 5.1 I LFL N2* * 2 O t. s E UFL h2o 12.9 12.0 The IFC Literature Review defines the critical steam volume fraction, X*, as the intersection of the stoichiometric air line w ith the vertical Imix* section of the flam m ability diagram. This critical steam volume fraction was defined as the steam at this point divided by the sum o f the steam plus fuel at this point, which is represented by the intersection of the light green line and the purple line in Figure 2 . Mathematically, X* can be expressed by Equation (5) below: X* = Equation (5) However, as seen in Figures 2 and 3, there exist steam volume fractions greater than this X* point that will still pass through the flam m ability diagram and therefore still support combustion. While this was not immediately apparent in the flare test data obtained from our initial test results, discussion w ith others w ho were also doing testing indicated th a t this conservative estimate of the critical steam volume fraction could indeed represent a reasonable num ber of operating conditions were high combustion efficiencies might exist. As a result, a M odified IFC steam volume fraction was proposed fo r this work. Percent Steam, v% * This new lim iting steam value is not represented by the intersection of the vertical Imix section w ith the stiochiometric air line, but rather the tangent to the LFLmix* point on the flam m ability diagram drawn from the origin. Recall that Imix* was uniquely defined in this M odified IFC theory by the tangent so drawn. This critical steam volume fraction is also represented on Figure 2 as the dark green line. One can see th a t this form ulation more closely defines the true lim iting quantity of inert th a t will still support combustion. This modified critical steam volume fraction has been defined as X'* and can be mathematically described by Equation (6) below: X' * = - ----- I(m ix )*------ (I(m ix )*+ F(m ix)* ) Equation (6) In discussing flare operation, it is useful to expand these flam m ability diagrams to show the entire range of pseudo-fuel, steam, and air mixtures. This has been done fo r a given vent gas composition in Figure 4. One should note th a t the line from 0% steam, 100% fuel to 100% steam, 0% fuel represents the line of all possible mixtures of steam and fuel w ith no air mixed in. In essence, fo r flares tha t have center steam addition, the am ount o f center steam addition represents a unique point on this line and is indicative of the conditions th at exist inside the flare stack just prior to exiting the flare tip. If a line is drawn from any point on the fuel/steam line to the origin, this line will represent all possible mixtures of air once the vent gas leaves the flare stack. The point on the fuel/steam line physically represent the point in space where the vent gas leaves the flare stack. If one then draws a line from this point to the origin, this line now represent all possible combination of the combustion zone gas w ith air. The point on the fuel /steam line representing the composition just prior to exiting the tip and the origin representing the point of infinite dilution w ith air. By definition, if this line crosses the flam m ability diagram, then there should be a point in space where combustion should take place if an ignition source exists. The existence of a point in space where combustion can occur does not tell us w hether the flame is stable, blown off the stack, or smokes, just th a t combustion can occur. Obviously, the physical design of a flare will have an impact on the point in space where a combustible mixture is achieved and ultim ately the stability o f tha t flame. Figure 4 can also be used to illustrate that not only is the am ount of total steam added im portant, but also both the ratio o f center steam to tip steam and the design of the tip steam injection ring can have an influence upon the limiting steam value where the flame is quenched. In Figure 4, the light blue line represents the potential dilution w ith air if only center steam is injected. Now let us assume th a t tip steam is also added. The yellow line in Figure 4 represents an am ount o f total steam added to the flare above the critical steam value. It can be seen tha t this line does not cross the vent gas flam m ability diagram, and therefore combustion should not occur. However, if the am ount of steam added in the yellow line case was not all added as center steam, but rather was split between center steam and tip steam, as different conclusion can be deduced. If one assumes th a t the center steam added is equivalent to the original light blue case, and tip steam is added to reach the total am ount of steam added in the yellow case using an ideal 10:1 air to steam educator, then one obtains potential vent gas/steam/air mixtures represented by the medium blue line, which ultim ately term inates at the yellow steam line. Clearly, this alternate method o f adding the same quantity o f steam to the flare leads to a different path where combustion could be supported even w ith a total steam addition rate beyond the critical value. The dark blue case presents the same center steam addition, but represent a less efficient 3:1 tip steam educator design. As can be noted, again the same total quantity of steam has been added, but the path taken by this design does not result in a vent gas/steam/air mixture th at will support combustion. Figure 4. Flare Operating Conditions Expressed on Flammability Diagram The IFC literature review fu rth e r defined the dimensionless reduced steam volume fraction (RSVF) as the actual steam volume fraction, X= volume steam added to flare/(volum e of fuel + volum e of steam to flare) divided by the critical steam volume fraction, represented by Equation (5) in the IFC Literature Review, but by Equation (6) in this M odified IFC Approach. For a flare w ith center steam only combustion should not occur at an RSFV of 1.0. Figure 5 displays combustion efficiency data obtained on 7 full scale flares over the past 2 years. The critical steam volume fractions have been determined according to Equation (6), and the observed combustion efficiency data have been plotted against the reduced steam volume fraction (RSVF) . Figure 5. Modified IFC RSVF' vs. Combustion Efficiency 1 00 .0% ■-------- * * ** O - o w w - Shell S erie s A Shell Serie s B 9 0.0% Shell S erie s C X Shell Serie s D MCT 7 6 5 cm -1 Shell S erie s D MCT 2 0 8 0 cm -1 to TCEQ 6 0 0 BTU/SCF 7 0.0% TCEQ 3 5 0 BTU/SCF O M a ra th o n TXC A8 M a ra th o n TXC A 1 1 o o Oo TCEQ 2 1 4 5 BTU/SCF O 0 < o «o ° :° 8 0.0% M a ra th o n TXC A 19 X + U 6 0.0% :s M a ra th o n TXC S erie s B M a ra th o n TXC S erie s C M a ra th o n TXC S erie s E X M a ra th o n TXC S erie s G □ 5 0 .0 % \ ►"< M a ra th o n TXC LTS 4 0.0% 0 >k M a ra th o n DET S erie s A = O □ M a ra th o n TXC S erie s D A r? A M a ra th o n DET S erie s B M a ra th o n DET S erie s C M a ra th o n DET S erie s D \ 3 0.0% a M a ra th o n DET S erie s LTS 0A o Flint Hills LOU S erie s A Flint Hills LOU S erie s B 20.0% Flint Hills LOU S erie s C o o Flint Hills AU S erie s A Flint Hills AU Serie s B Flint Hills AU S erie s C 10.0% * Flint Hills AU Serie s D 9 8% C o m b u s tio n E fficien cy 9 8% CE RSVF 0.0% ■J---------------1 ---------------1---------------1----------0 .0 0 0 .20 0.4 0 0 .6 0 ---- B5B3D----- ■---------------1 0 .8 0 1.00 1 .2 0 IFC M o d ifie d RSVF' Theoretically, the M odified IFC approach would say th a t above a RSVF of 1 no combustion should be observed. Clearly, while Figure 5 indicates th a t combustion efficiency is indeed trending to zero as the RSVF approaches one, some measurable combustion efficiencies have been observed above a RSVF of 1. In this type of pooled, full- scale data there are many potential explanations fo r variation from the expected result, which include but are not limited to the following: • Differences in the physical design of the flares (ground flare versus elevated, steam /air tubes versus tip steam, different ratios o f center steam to tip steam) • Different impacts of center steam versus tip steam, as previously illustrated in Figure 4 • Measurement uncertainty both in CE measurements at unstable combustion conditions, and steam and vent gas flo w measurement at low flow conditions • Uncertainty in pure component flam m ability data • Temporal fluctuation in vent gas composition not picked up by online analyzers While the ability of this Modified IFC approach to describe these full scale data is far from perfect, this approach does show some promise as a technique that can be used to estimate the critical RSVF (X'*) and critical combustion zone net heating value (NHVCZG ) where a vent gas of a given composition will be extinguished. Being able to determ ine the X'* and NHVCZGvalue where a given vent gas composition will be extinguished is valuable, of greater interest is a means to estimate the RSVF and NHVCZG below which and above which, respectively, a combustion efficiency of 98% is maintained. From the combustion efficiency data obtained from the testing of any flare, an ample margin o f safety parameter can be developed tha t when applied to the RSVF parameter would yield a lim iting RSVF and NHVCZGthat ensures a combustion efficiency of greater than 98%. This factor can be estimated by plotting observed combustion efficiency data from a flare versus the M odified IFC reduced steam volume fraction (RSVF' = X/X'* ), as illustrated in Figure 5. The RSVF98% can be estimated by drawing a line from the theoretical point of flame extinguishment, (1.0, 0) on Figure 5 to the last point on the test data th at produced a reliable combustion efficiency of 98%. From the intersection of this line w ith the observed data, a vertical line can be dropped down to the RSVF axis. The intersection of this vertical line w ith the x-axis represents the margin o f safety factor applied to the point of flame extinguishment th a t insures the desired combustion efficiency will is be maintained. For the data shown in Figure 5 it is clear that this ample margin of safety factor is not necessarily the same value fo r each flare. However, as shown in Figure 5, an RSVF of 0.8 appears to be a reasonable approximation covering the bulk of the data presented in this paper. While beyond the scope of this paper, it is hypothesized that fo r flares of similar design this empirically derived maximum RSVF could be the same or similar. Summary Both vent gas composition and assist gas ratio can have an impact on the observed combustion efficiencies of flares. A modified International Flare Consortium (IFC) approach has been used to assess the critical lim iting steam value that will support combustion fo r any given vent gas/assist gas mixture. This analysis approach has been used to analyze data from 7 full-scale flare tests. While combustion efficiency trends to zero at the critical reduced steam volume fraction (RSVF) of 1.0 calculated from this M odified IFC approach fo r a variety of flares and vent gas compositions, some deviation from the ideal performance was note fo r a limited num ber of compositions and flare configurations. While several potential reasons fo r such deviations have been suggested, full rationalization of these discrepancies was beyond the scope of this work. An approach has also been outlined whereby the critical RSVF calculated from this approach can be used to define the maximum RSVF th a t can be employed to achieve a specified combustion efficiency. W ith the exception o f one flare in this study, an RSVF of 0.8 gave reasonable assurance of achieving a combustion efficiency o f 98%. Further data is required to assess how generally applicable this value is fo r other flares.