A study of jet fuel sooting tendency using the threshold sooting index (TSI) model

A study of jet fuel sooting tendency using the threshold sooting index (TSI) model
of 15
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Combustion and Flame 149 (2007) 191– A study of jet fuel sooting tendency using the thresholdsooting index (TSI) model Yi Yang, André L. Boehman ∗ , Robert J. Santoro The Energy Institute, The Pennsylvania State University, University Park, PA 16802, USA Received 13 February 2006; received in revised form 16 October 2006; accepted 6 November 2006Available online 26 January 2007 Abstract Fuel composition can have a significant effect on soot formation during gas turbine combustion. Consequently,this paper contains a comprehensive review of the relationship between fuel hydrocarbon composition and sootformation in gas turbine combustors. Two levels of correlation are identified. First, lumped fuel compositionparameters such as hydrogen content and smoke point, which are conventionally used to represent fuel sooting ten-dency, are correlated with soot formation in practical combustors. Second, detailed fuel hydrocarbon compositionis correlated with these lumped parameters. The two-level correlation makes it possible to predict soot formationin practical combustors from basic fuel composition data. Threshold sooting index (TSI), which correlates lin-early with the ratio of fuel molecular weight and smoke point in a diffusion flame, is proposed as a new lumpedparameter for sooting tendency correlation. It is found that the TSI model correlates excellently with hydrocar-bon compositions over a wide range of fuel samples. Also, in predicting soot formation in actual combustors, theTSI model produces the best results overall in comparison with other previously reported correlating parameters,including hydrogen content, smoke point, and composite predictors containing more than one parameter. © 2006 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords:  Jet fuel; Gas turbine combustor; Soot; Fuel composition; Sooting tendency; Hydrogen content; Smoke point 1. Introduction Fuel hydrocarbon composition is one of the mostimportant factors controlling soot formation in gasturbine combustion. Pure hydrocarbons with dissimi-lar molecular structures exhibit dramatically differentsooting performance in laboratory burners and gasturbine combustors. However, it is not yet possible totheoretically derive the propensity to form soot for afuel based on known fuel composition and combus- * Corresponding author. Fax: +1 (814) 863 8892.  E-mail address: Boehman). tion conditions. This is primarily due to the lack of mechanisms of soot formation and oxidation for in-dividual hydrocarbons, as well as to the complexityresulting from the coupling of the turbulent flow andrapid chemical reaction kinetics. Moreover, gas tur-bine fuels contain hundreds of compounds and theircompositions vary from batch to batch, making it im-possible to correlate soot formation to individual hy-drocarbon content.In response to such difficulties, practical investi-gations on the effects of fuel composition on sootformation have largely relied on empirical correla-tions, which can be classified into two categories.One correlates soot formation in actual combustors 0010-2180/$ – see front matter  © 2006 The Combustion Institute. Published by Elsevier Inc. All rights reserved.doi:10.1016/j.combustflame.2006.11.007  192  Y. Yang et al. / Combustion and Flame 149 (2007) 191–205 Nomenclature A  total aromatic content, wt% A/F   air/fuel mass ratioDA dicyclic aromatic content, vol% (or wt%as indicated) f  max  maximum line-of-sight soot volumefraction H   hydrogen content, wt% H/C  hydrogen/carbon molar ratioMA monocyclic aromatic content, vol% (orwt% as indicated)MW molecular weight, g / mol N   naphthene content, wt% P   paraffin content, wt% P  bc  branch and cyclo paraffin content, wt% P  n  normal paraffin content, wt% q ′′ RAD  flame radiation heat flux, W / m 2 SP smoke point, mm x i  molar fraction y i  mass fractionwith certain lumped fuel composition parameters, re-ferred to as  the first-level correlation  in this paper.These lumped parameters, such as hydrogen contentand smoke point, reflect the overall hydrocarbon com-position inthe fuel,butgenerally reveallittle informa-tion on the molecular structure of the hydrocarbons,which is the ultimate fuel factor in soot formation.Another type of correlation is thus necessary to corre-late these lumped parameters to detailed fuel hydro-carbon composition, referred to as  the second-levelcorrelation . Clearly the lumped fuel composition pa-rameters act as the link between the two correlationlevels, which realize the prediction of combustor sootformation from the basic fuel hydrocarbon composi-tion, as shown in Fig. 1.It is of essential importance to select an appropri-ate linking parameter for soot correlations, which rep-resents the inherent sooting tendency of a given fuel.Various fuel composition parameters have been in-vestigated, including hydrogen content, smoke point,aromatics content, and composite parameters contain-ing more than one variable. In this study special at-tention is paid to the threshold sooting index (TSI),which was successfully used to rate the sooting ten-dency of pure hydrocarbons [1,2], for its applicationtocorrelatingfuelhydrocarboncompositionandcom-bustor soot formation.Previous studies on two-level soot correlations, aswell as on the sooting tendency of pure hydrocarbons,are reviewed in the following section. Before doingthat, it may be helpful to briefly mention the lumpedcomposition parameters and the experimental meth-ods to determine them:The smoke point is the greatest flame height withoutsmoke emission under laminar diffusion combus-tion. A higher smoke point means a lower sootingtendency. Laboratory burners were widely used tomeasurethe smokepoints ofpurecompounds (bothgaseous and liquid) and fuel mixtures in the earlydays. For jet fuels, whose sooting tendencies areregulated, the smoke point is measured on a stan-dard smoke point lamp [3].Hydrogen weight content is another commonly stud-ied fuel parameter for sooting tendency, which isdetermined from a fuel CHN test by burning a cer-tain amount of liquid fuel and measuring the CO 2 ,H 2 O, and NO x  produced [4]. The hydrogen con-tent of pure compounds can be determined fromthe molecular formulae of the compounds. Fuelswith higher hydrogen content generally produceless soot in combustion.Fuel aromatics content is known to greatly affectsoot formation during combustion. Higher aromat-ics content is generally associated with higher sootformation. Volumetric aromatics content can bemeasured by fluorescence indicator adsorption [5];mass aromatics content can be measured by high-performance liquid chromatography [6].Composite parameters for soot correlations are usu-ally the combinations of aromatics content andsmoke point or hydrogen content.The threshold sooting index (TSI), which will be de-tailed later, is an artificially defined number represent-ing the inherent sooting tendency of a compound or amixture. Higher TSI means higher sooting tendency. Fig. 1. Prediction of soot formation from detailed and lumped fuel compositions.  Y. Yang et al. / Combustion and Flame 149 (2007) 191–205  193 2. Review of correlations for soot formation 2.1. Correlations of combustor soot formation and lumped fuel composition: the first-level correlation2.1.1. In-combustor soot formations vs exhaust soot emissions Soot formation in gas turbine combustors has beendescribed in two ways, which have distinct relevanceto fuel composition: in-flame soot formation and ex-haust soot emissions. The former is measured withinthe combustor via primary zone radiation, flame sootvolume fraction, and combustor liner temperature,while the latter is directly measured outside the com-bustor. Whether in situ or ex situ measurement of sootformation is more important depends on which aspectof the soot problem is of concern. From the stand-point of environmental pollution and combat stealth,soot emissions exiting the combustor are more criti-cal. But in most cases, the combustor liner durabilityis mostly concerned; therefore the parameters relatedto in-combustor soot formation are investigated morefrequently. Here, fuel sooting tendency, as evidencedby in-combustor soot formation, is of more interestbecause it relates more directly to fuel chemical com-position. Exhaust soot emissions are less relevant inthe present context because exhaust emission is thenet result of soot formation and soot oxidation, wherethe soot oxidation is controlled by the combustionconditions more than the fuel structure.Extensive studies have been carried out to predictsoot formation in actual combustors using lumpedfuel composition. Hydrogen content and smoke pointare the two most widely studied of these correlationsand are generally regarded as the sooting tendency of a fuel. It long has been debated as to which of the twois more important, as shown in the review below. 2.1.2. Correlations of combustor soot formationwith fuel hydrogen content  Schirmer [7] was among the first to correlate fuelhydrogen content with the soot production in gas tur-bine combustors. The results were obtained from aPhillips 2-in.-diameter model combustor operated un-der different combustor pressures (7.5–15 atm) andinlet air temperatures (400–1000 ◦ F). Four jet fuelswith hydrogen content from 13.8 to 15.0 wt% weretested. Soot formation was detected via measurementof the flame radiation, and of the optical density of the exhaust gas. Flame radiation and exhaust sootwere correlated with fuel hydrogen content, inlet airtemperature, and combustor pressure in a polynomialform. It was found that the influence of hydrogen con-tent on soot formation declines as the engine inlettemperature and pressure increase.Friswell [8] measured the smoke emissions froma 3-in.-diameter laboratory-scale gas turbine combus-tor under pressure from 2 to 22 atm, inlet temperaturefrom ambient to 572 ◦ F (300 ◦ C), and air/fuel ratioof 60/1. Fifteen fuels from various sources, including jet fuels, gas oils, and solvents, were tested, wherehydrogen content varied from 10.88 to 15.22 wt%and aromatics content varied from 2 to 81 vol%. Theauthor found that fuel hydrogen content produced abetter linear correlation with exhaust soot emissionsthan fuel aromatic content did. Furthermore, Friswellfound that as the combustion pressure increased thefuel composition effect on soot formation decreased,which confirmed Schirmer’s observation.Clark  [9] studied the primary combustion zone ra-diation as a function of fuel hydrogen content (11.76to 13.60 wt%) and aromatic content (0.06 to 43.4wt%) in a 6-in.-diameter combustor. Eight fuels, in-cluding gasoline, jet fuels, diesel fuels, and fuel oils,were tested under pressures from 3 to 12 atm, equiv-alence ratio from 0.15 to 0.30, and inlet tempera-ture from 620 ◦ F (600 K) to 800 ◦ F (700 K). The“effective hydrogen content,” defined in the form [ H  - m( DA ) ]  or  [ H  - ( DA ) n ]  ( H   is the hydrogen weightpercent, DA is the dicyclic aromatic weight percent,and  m  or  n  is constant for a specific fuel and operat-ing condition) correlated much better than hydrogencontent alone with the primary zone radiation.Naegeli and Moses performed a series of studieson the effect of fuel hydrogen content on soot for-mation in gas turbine engines [10–13]. They foundthat hydrogen content correlates most effectively withcombustor soot formation, but that the molecularstructure does play a significant role. For example, inRef. [12], six fuels with equal hydrogen content (12.8wt%) were tested in a Phillips 2-in. combustor undervarious pressures, air/fuel ratios, and inlet tempera-tures. The flame radiation, averaged from differenttest conditions and measured in the form of com-bustor liner temperature increase, varied significantlywith fuel structure despite the constant hydrogen con-tent. Fuels containing methylnaphthalene and tetralinproduced more radiation than those containing xyleneand decalin. The authors concluded that “[W]hile theresults do not dispute the importance of the hydro-gen content correlation, there is strong evidence thatmolecular structure can play a significant role in sootformation.” It is notable that smoke point correlatedwell with the flame radiation in this case.In a following paper, Naegeli et al. [13] studiedsoot formation as a function of fuel hydrogen con-tent (11.5–14.5 wt%) and flame temperature (2190–2230 K) in the same combustor under 0.93 atm,700 ◦ F inlet, and  A/F   =  102. Flame temperature wasobserved to increase with hydrogen content, but thedependence was quite weak. Relative soot formation,  194  Y. Yang et al. / Combustion and Flame 149 (2007) 191–205 measured from flame opacity, generally increasedwith decreasing hydrogen content. But four fuels withsimilarly low hydrogen content (11.5 wt%), but dif-ferent naphthalene content, again exhibited signifi-cant variations in soot formation, indicating the im-portance of dicyclic aromatic content in this process.Bowden et al. studied the roles of fuel hydrogencontent and molecular structure in soot formation ina Thornton 1.6-in. (40 mm)-diameter model combus-tor [14,15]. In Ref. [14], Jet A-1 was tested in blends with up to 10 vol% of tetralin and methylnaphtha-lene under 9.9 atm, 752 ◦ F (400 ◦ C) inlet, and  A/F  ratio from 40 to 50. Hydrogen content, which variedslightly from 13.06 to 13.77 wt%, correlated satisfac-torily with soot formation in terms of exhaust sootconcentrations, flame tube temperatures, and flameradiation.In Ref. [15], two series of fuels were tested in thesame combustor at 3.5 atm, 494 ◦ F inlet, and  A/F  ratio from 30 to 60. One was the Shellsol T (a sol-vent containing 99% isoparaffins) blended with oneof the cyclic compounds methylnaphthalene, tetralin,dicyclopentadiene, decalin, or toluene at 2:1 ratio byweight. The other was Shellsol T blended with eachof these compounds to obtain a fixed hydrogen con-tent of 13.8 wt%. In both cases, samples containingmethylnaphthalene or tetralin produced much moresootthanthosecontainingdicyclopentadiene,decalin,or toluene, indicating the importance of multi-ringaromatics in soot formation. For fuels with fixedhydrogen content, the methylnaphthalene-containingsample produced the highest exhaust soot concentra-tion, while the tetralin-containing sample generatedthe most flame radiation, indicating that exhaust sootemission is not necessarily consistent with the actualsoot formation during combustion.Inalaterpublication,Bowdenetal.[16]correlatedboth flame radiation and exhaust sootconcentration ina full size Rolls Royce Tyne combustor with those inThornton and Phillips model combustors. Very goodlinearity was observed, which means that the resultsobtained from model combustors can be applied con-fidently to soot prediction in full-size combustors.Sampath et al. [17] developed a model for predict-ing exhaust smoke emissions from a full-scale Pratt &Whitney PT6A-65 engine. The exhaust emission in-dex (EI), in grams carbon per kilogram fuel, was cor-related with fuel hydrogen content ( H  , wt%), com-bustorinlettemperature( T  3 ,K)andpressure( p 3 ,Pa),and air/fuel ratio ( A/F  , mass), as in the equation(1)EI  = k  p 3 ( 1 − H/ 100 ) 3 (A/F)(H/ 100 ) 2  2 . 7 T  − 8 . 663  , where  k  is a combustor-dependent constant. 2.1.3. Correlations of combustor soot formationwith fuel smoke point  On the other hand, many found that smoke pointis more related to soot formation than hydrogen con-tent in gas turbine combustion. Numerous correla-tions have been proposed to predict soot formationfrom the smoke point of a fuel.In a buoyant turbulent diffusion flame, Markstein[18] found that the radiative fraction of total heatrelease rate ( x RAD ), which is closely related to theamount of soot formed, is linearly correlated with thefuel smoke point (SP, mm), as in the equation(2) x RAD  = 0 . 43 − 9 . 1 × 10 − 4 SP . This observation was also found to be applicable tonitrogen diluted flames.Rosfjord [19] studied the variation of primaryzone radiation using fuels from various srcins andcontaining a wide range of hydrogen, aromatic, andnaphthalene content. The tests were conducted on alaboratory-designed burner with diameter 5 in. andoperated under 13 atm and 800 ◦ F inlet. The effectsof fuel physical properties were de-emphasized by se-lecting injectors that produced highly atomized andhence rapidly vaporizing sprays. It was found thatsmoke point (SP − 0.6 ) produced a better linear re-lationship with the primary zone radiation than hy-drogen content ( H  − 1 . 65 ), but the best correlation re-sulted from a parameter combining dicyclic aromaticcontent (DA, vol%) and hydrogen content (wt%), asin the following equation:(3) q ′′ RAD  ∝  H  − 1 . 2 ( 100 − DA ) − 0 . 4  . Chin and Lefebvre [20,21] reviewed previous cor-relations and found that a parameter composed of smoke point and dicyclic aromatic content correlatedvery well over a wide range of flame radiation dataavailable in the literature. Their correlation took theform of Rosfjord’s but replaced hydrogen contentwith smoke point:(4) q ′′ RAD  = f   SP − 0 . 92 ( 100 − DA ) − 0 . 4  . The relationship between flame radiation and the pre-dictor is nonlinear for most cases investigated, whichis different from other correlations of the same type.They also reported that for a single-parameter corre-lation, smoke point is better than hydrogen content.Gülder et al. [22,23] studied fuels with the samehydrogen/carbon ratio in flames under smoke pointconditions and found that the smoke emission andthe maximum in-flame soot volume fraction were notconstant, but correlated well with smoke point. Theyconcluded that the smoke point and hydrogen/carbonratio (or hydrogen content) are two complementary,instead of alternative, parameters in soot formation.  Y. Yang et al. / Combustion and Flame 149 (2007) 191–205  195 This conclusion was based on the argument that fuelstructure and flame temperature were the two majorfactors in soot formation [22–24], and smoke pointrepresented the effect of fuel structure and hydro-gen content represented the effect of flame temper-ature. They also proposed a semiempirical model inthe form of the Arrhenius equation, which correlatedthe measured maximum soot volume fraction with aterm containing smoke point and hydrogen-to-carbonatomic ratio  (H/C) , as shown in the equation(5) f  max  ∝  SP 0 . 5 (H/C) a exp  − b( SP )/(H/C)  , where  a  and  b  are constants depending on test con-ditions. This model was tested against three sets of flame radiation data from gas turbine combustors, andgood predictions were observed.Pande and Hardy [25] did an in-depth evalua-tion of previously reported soot formation predictorsin a Rolls Royce Tyne combustor and an AllisonModel T56 combustor. They compared parametersincluding smoke point, hydrogen content, aromaticcontent, composite parameters from Rosfjord, Chinand Lefebvre, and Gülder, and three new compos-ite predictors. Their results indicated that two newpredictors, one combining smoke point, monocyclicaromatic content (vol%), and dicyclic aromatic con-tent (vol%), and one combining hydrogen content,monocyclic aromatic content, and dicyclic aromaticcontent, fit best over a broad matrix of fuel samplesand operating conditions. Single parameters such assmoke point and hydrogen content performed muchless satisfactorily than composite predictors in theircorrelations. 2.1.4. Summary of the first-level soot correlations Based on the above discussion, the followingstatements regarding the effect of lumped fuel com-position on soot formation (i.e., the first-level corre-lation) can be made:1. Hydrogen content and smoke point are the twomost effective parameters.2. Dicyclic aromatic content is more important thanmonocyclic aromatic content and is a good com-plementary factor for hydrogen content or smokepoint in the correlation of soot formation.3. Single parameters are not sufficient to describethe role of fuel composition in soot formation.At least two parameters are required for a goodcorrelation.4. In-combustor soot formation and exhaust smokeemissionwerebothusedtocorrelatewithlumpedfuel composition, but little attention was paid todifferentiating the two. 2.2. Correlations of lumped composition parameters with detailed hydrocarbon composition:the second-level correlation Hydrogen content and smoke point, which foundsignificant use in the first-level correlation, do not ac-tually provide explicit information about the type orthe amount of hydrocarbons contained in a fuel. Nu-merous hydrocarbon combinations can be formulatedfor a given smoke point or hydrogen content, and thisimpedes an understanding of the fundamental effectsof individual hydrocarbons or hydrocarbon classes onsoot formation. The second-level correlation, in thissense, connects the fuel hydrocarbon composition tothe lumped parameters and thus makes it possible topredict smoke emissions or flame radiation from thehydrocarbon composition of a fuel. Most of these cor-relations are developed for smoke point, because hy-drogen content and aromatic content can be directlyor indirectly derived from the known composition.Smoke point measurement is performed in a lam-inar diffusion flame. The simplified flow field makesit possible to develop theoretical models to describethe flame structure. For example, Roper’s model pre-dicted that the flame height is proportional to the fuelflow rate, which for a circular port burner is expressedas [26](6) h/  ˙ V   = (T  0 /T  f  ) 0 . 67 4 πD 0 ln ( 1 + 1 /S), where  h  is the flame height,  ˙ V   is the volumetric fuelflow rate,  T  0  and  T  f   are the ambient and flame tem-peratures,  D 0  is the diffusion coefficient at ambienttemperature, and  S   is the stoichiometric air/fuel vol-ume ratio.Markstein [18] studied flame radiation in smokepoint flames of ethylene, propylene, isobutene, and1,3-butadiene. He found that the radiation heat releaserate (  ˙ Q RAD , SP ) is about 30% of the total heat releaserate (  ˙ Q TOT , SP ) at the smoke point condition(7) ˙ Q RAD , SP  = 0 . 30  ˙ Q TOT , SP  − 2 . 9 (W) . The correction factor of   − 2.9 W was attributed to theheat loss to the burner tip.Kent [27] used Markstein’s result and developed amodel based on the energy balance between the radi-ation heat loss caused by soot formation and the char-acteristic sooting temperature, 1300 K [28], in flamesunder smoke point conditions. Kent’s result showedthat the maximum soot volume fraction in a laminardiffusion flame is determined by the equation(8) f  max  × 10 6 = 0 . 1 (  ˙ m SP H  c  − 3 )(  ˙ m SP S/ MW ) 1 . 5  ,
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks