n ES L THE AMERICAN SCIETY F MECHANICAL ENGINEERS 345 E. 47 St., Ne York, N.Y. 117 The Society shll not be responsible for sttements or opinions dvnced in ppers or in discussion t meetings of the Society or of its Divisions or Sections, or printed in its publictions. Discussion is printed only if the pper is published in n ASME Journl. Ppers re vilble from ASME for fifteen months fter the meeting. Printed in SA. Copyright 1989 by ASME 89-GT-322 ltr Lo Nx Emissions for Gs nd Liquid Fuels sing Rdil Sirlers H. S. ALKABIE nd G. E. ANDREWS Deprtment of Fuel nd Energy The niversity of Leeds Leeds, LS2 9JT, K ABSTRACT Curved blde rdil sirlers using ll the primry ir ere investigted ith pplictions to len burning gs turbine combustor primry zones ith lo Nx emissions. To modes of fuel injection ere compred, centrl nd rdil sirler pssge injection for gseous nd liquid fuels. Both fuel systems produced lo Nx emissions but the upstrem mixing in the sirler pssges resulted in ultr lo Nx emissions. A 14mm dimeter tmospheric pressure combustor s used ith 43% of the combustor ir flo into the primry zone through the rdil sirler. Rdil gs composition mesurements t vrious xil distnces ere mde nd these shoed tht the flme stbility nd Nx emissions ere controlled by differences in locl mixing t the bse of the sirling sher lyer donstrem of the sirler outlet. For rdil pssge fuel injection it s found tht very high combustion efficiency s obtined for both propne nd liquid fuels t 4K nd 6K inlet tempertures. The flme stbility, lthough orse thn for centrl fuel injection s considerbly better thn for premixed system. The Nx emissions t one br pressure nd 6K inlet temperture, comptible ith high combustion efficiency, for propne nd kerosene ere 3 nd 6 ppm t 15% oxygen. For Gs il the Nx emissions ere higher, but ere still very lo t 12ppm. Assuming squre root dependence of Nx on pressure these results indicte tht Nx emissions of 48ppm for Gs il nd less thn 12ppm for gseous fuels could be chieved t 16 br pressure, hich is typicl of recent industril gs turbines. High ir flo rdil sirlers ith pssge fuel injection hve the potentil for dry solution to the Nx emissions regultions. 1. INTRDCTIN For industril gs turbines, Nx legisltion exists in severl countries tht cn only be met by using ter or stem injection, ith n ssocited performnce penlty. The present ork is concerned ith dry combustor design solution for the reduction in Nx emissions. It is ell knon tht Nx emissions cn be reduced by using len ell mixed primry zones, but the level of reduction is closely relted to the qulity of the improved fuel nd ir mixing. Conventionl combustor designs hve very poor mixing in the primry zone, to give ide stbility rnge, nd this results in very lrge temperture grdients hich generte lrge Nx concentrtions s shon in the internl gs composition trverses of Heitor nd Whitel (1986) nd Bicen et l (1988). Future higher turbine entry temperture gs turbines ill lso require more ir in the primry zone ith better fuel nd ir mixing. Premixed fuel nd ir systems offer the loest Nx emissions but involve mjor prcticl problems (Sothern et l 1984). ne of the mjor problems is tht of flme stbility, hich my be overcome by fuel stging beteen diffusion flme pilot nd premixed min burner (Becker et l 1986, Kurod et l 1987). To stge (rich/len) combustion systems re lso useful for reducing Nx emissions for fuels ith fuel bound nitrogen (Sski nd Itoh, 1987). Hoever, the Nx reduction of these to stge systems is limited due to the Nx produced by the pilot or rich/len mixing stge. Chrles nd Smuelson hve investigted rdilly ir stged sirl combustor ith centrl fuel injector nd xil sirler surrounded by n xil flo of non-sirling ir. Mixing beteen the to strems s poor ith the rich centrl sirling flme producing hot core ith little mixing ith the rdilly stged secondry ir. No Nx mesurements ere mde but the temperture profiles ere poor nd ould result in high Nx emissions. An lterntive technique for overcoming the stbility problem is tht of vrible ir stging, ith ir sitched from the primry zone to the dilution zone t lo poer nd in the reverse direction to give len primry zone t high poer. This involves the complexity of moving prts in high temperture environment nd there hve been fe successful pplictions of this technique in smll gs turbine combustor sizes, but it is more esily chieved in lrge industril gs turbines (Aoym nd Mndi, 1984). Andres et l (1988) hve developed rnge of techniques for lo Nx gs turbine primry zones ith good flme stbility, ll of hich use fuel injection Presented t the Gs Turbine nd Aeroengine Congress nd Exposition June 4-8, 1989 Toronto, ntrio, Cnd Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use
into the bse of jet sher lyers. Lo Nx emissions hve been demonstrted for non-sirling jet sher lyer systems for both gs nd liquid fuels (Abdul Hussin et l, 1988 nd Abdul Aziz et l, 1987) nd for sirl stbilised systems (Alkbie et l, 1988). The present ork investigtes techniques to produce further reductions in Nx emissions for rdil sirlers, using the rdil sirler pssge s prtil fuel nd ir premixing zone. In previous investigtions of rdil sirl systems for lo emission gs turbine combustors, the rdil pssges hve been crried through bend into n nnulus round centrl fuel injector, thus becoming form of xil sirler but ith the benefit of greter pssge fuel nd ir mixing time (Smith et l,1987). Ahmd et l (1986) found problems in the chievement of n dequte combustion efficiency nd stbility ith len primry zones using lrge ir flo xil sirlers ith centrl fuel injection. Alkbie et l (1988) shoed tht these problems could be overcome by the use of rdil sirlers, ith their quite different ner sirler erodynmics. Ahmd nd Andres (1984) nd Andres nd Kokbi (1987) shoed tht for fixed xil sirler the position nd direction of fuel injection hd mjor influence on the flme development nd Nx emissions. The objective of the present ork s to investigte the potentil for Nx reduction using better fuel plcement in rdil sirl stbilised combustor primry zones. Axil centrl fuel injection, hich is the most frequently used mode of fuel injection in sirl combustion, resulted for confined sirl flme in very rich core region for ek overll mixture. Rdil centrl fuel injection produced much better mixing, hich s further improved by fuel injection t the sirler periphery. Chrles nd Smuelson (1988) hve lso shon mjor improvement in the flme development of their ir stged sirler flme hen the direction of the centrl fuel injection s chnged from predominntly xil to predominntly rdil. Beltgui nd Mccllum (1988) lso demonstrted the better flme development for xil sirlers ith peripherl fuel injection. Alkbie nd Andres (1989) shoed tht peripherl fuel injection improved on the fuel nd ir mixing compred ith centrl rdil fuel injection for the rdil sirlers. The present investigtion of pssge fuel injection ere undertken to exploit the tin benefits of peripherl fuel injection nd prtil fuel nd ir mixing upstrem of the sirler. For liquid fuels, pssge injection lso hs the dvntge of high irflo ir blst tomistion. 2. RADIAL SWIRLER DESIGN AND TEST CNDITINS The rdil sirlers ere similr in design to those used in the Ruston Torndo industril gs turbine combustor, but ith n incresed flo cpcity. The rdil sirlers used curved blde pssge design in n ttempt to void vne ke effects nd to minimise flo seprtion in the chnnels. The design fetures re shon in Fig.1 together ith the to methods of fuel injection, centrl nd rdil pssge. The vne ngle s 45 deg. nd s the effective vne pssge jet outlet ngle. The sirler outlet dimeter s 76mm nd there ere 8 rdil vne pssges ith depth of 15mm nd minimum pssge idth of 16.3 mm. The minimum flo re of the sirler gve flo blockge of 87% of the 14 mm combustor dimeter into hich the sirler dischrged. The combustor to sirler outlet dimeter rtio, D/d, or expnsion rtio s 1.8, hich gve strong outer recircultion zone s ell s centrl sirl generted recircultion zone. The to zones ere VANE ANGLE /. / PSNAGE JEC AIR SPPLY PILL 163 CAN CMBSTR PRIMARY ZNE (NCLED) d o D _33mm h CENTRAL FEL INJECTIN FIG.1 RADIAL SWIRLER CMBSTR AND FEL INJECTIN CNFIGRATIN. CRNER RECIRCLATI JET BNDARY ZNE - //mm _^ o SPPLY VRTEX PIPE ^l CRE RADIAL SWIRLER..32 x/ - FIG.2 SCHEMATIC DIAGRAM F THE FLW PATTERN ISSED BY RADIAL SWIRLER SING AIR BBBLES R DYE AS TRACER. seprted by strong sirling sher lyer. These erodynmics ere investigted using isotherml ter flo visuliztion, the flo ptterns in the ner sirler region re shon in Fig.2. Nejd et l (1988) reported detiled isotherml erodynmic dt for very similr sirl system to tht of the present ork. An xil sirler s used mounted ell upstrem of its continment pipe dischrge into 152mm dimeter combustor. The sirler expnsion rtio s 1.5 nd the men sirler outlet velocity s 19.4 m/s, compred ith 1.8 nd 33 m/s respectively in the present ork. They found tht the outer recircultion zone extended.63 combustor dimeters donstrem, compred ith.32 dimeters in the present ork s shon in Fig.2. The inner recircultion zone s longer t 8mm or.57d nd there s vortex core tht extended ell beyond these recircultion zone regions. Nejd et l. (1988) lso determined the lrge xil velocity grdients nd high turbulence intensities in the jet boundry sher lyer, equivlent to tht in Fig.2. Alkbie et l (1988) previously investigted this rdil sirler ith the centrl fuel injector shon in Fig.1. This hd eight rdil fuel injection holes nd these injected fuel into the bse of the sirling jet sher lyer. This rich zone ithin the sher lyer gve n excellent flme stbility ith ek extinction t 6K of 25/1.This rich zone t the bse of the sher lyer s rpidly mixed out in the donstrem sirl Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use
L flo nd lo Nx emissions (1 ppm t 15% oxygen) ere obtined ith Nturl Gs. Hoever, the Nx emissions for propne ere nerly tice tht for nturl gs nd ere only just inside the EPA limits scled to the one tmosphere operting pressure. Alkbie et l (1988) lso investigted smller outlet dimeter (4mm) rdil sirler ith the sme sirler flo re s the 76mm sirler. The much lrger vne pssge depth of 3mm gve some premixing of the fuel nd ir inside the sirler nd very lo Nx emissions ere chieved but ith flme stbility little better thn fully premixed system. In the present ork the use of pssge fuel injection s used, this gve prtil fuel nd ir mixing in the short pssges. The objective of the present ork s to quntify the expected reduction in stbility nd Nx emissions tht this improved mixing ould entil. Gseous nd liquid fuels ere investigted, to ccess hether the reltively high Nx emissions for the higher hydrocrbon fuels ith centrl injection could be reduced ith pssge injection. Alkbie et l (1988) investigted for centrl injection the influence of the primry zone Mch number on the combustion performnce nd emissions of rdil sirler. The primry zone Mch number s bsed on the combustor cross sectionl re Al, the totl primry zone ir mss flo nd the upstrem temperture. Alkbie et l (1988) shoed tht provided the pressure loss s the sme then the Mch number hd little influence on the optimum Nx emissions. Hoever, if the Mch number s reduced t constnt blockge then the pressure loss nd hence turbulent mixing ould decrese ith the Mch number. This then resulted in incresed Nx emissions t the loer Mch number. Alkbie et l (1988) used Mch numbers of.14 nd.2 for the 14mm combustor nd.3 for 76mm combustor. Most of the dt s reported t 2.5% pressure loss for the to combustor sizes nd this required.14 Mch number in the 14mm combustor nd.3 in the 76mm combustor. In the present ork.2 Mch number s used, for both centrl nd pssge fuel injection, s the higher pressure loss ould give greter turbulent mixing nd loer Nx emissions s ell s better liquid fuel tomistion in the rdil vne pssges. The rtio of the primry zone Mch number,.2, to the reference Mch number,.47, gve the proportion of the totl combustor ir tht s pssed through the sirler s 43%. The rdil sirler used hd pressure loss of 4.2% t the.2 test Mch number, hich is typicl conventionl combustor pressure loss. 3. EXPERIMENTAL EQIPMENT The tmospheric pressure test fcility consisted of n ir supply fn, venturi flo metering, electricl ir preheter, 1m long 17mm dimeter pproch plenum chmber, rdil sirler, 33mm long 14mm dimeter uncooled combustor, men gs smple probe, ter cooled 14mm pipe nd bend in the ter cooled exhust pipe ith n observtion indo on the combustor centre line. The ir nd fuel flo rte metering ccurcy s +/- 2% for the ir flo nd 5% for the fuel flo. The 14mm combustor s prcticl ero nd industril cn combustor size, no scling of the results to prcticl combustor size s necessry. Hoever, provisded the D/d sirler expnsion rtio of 1.8 s mintined these results could be scled to lrger dimeter combustors, s shon by Alkbie et l (1988) for 76 nd 14mm combustors. The combustor s instrumented ith ll sttic pressure tppings nd thermocouples so tht xil profiles of temperture nd sttic pressure could be mesured. These ere used to determine the recircultion zone size nd the xil loction of the min het relese zone. The tmospheric pressure testing used in the present ork is rpid reltively chep technique for the evlution of combustor performnce. It is prticulrly suitble for the evlution of lo Nx combustor designs, s systems hich hve high Nx emissions t one br must hve even higher Nx emissions t higher pressures. Thus lo Nx design t one br pressure, ith dequte ir prehet, is firm bsis for testing t full pressure providing there is high combustion efficiency ith the lo Nx emissions t one br. If 99.9 % combustion efficiency cn be chieved t one br, then pressure cnnot significntly improve on this. The Nx emissions cn then be resonble relibly scled to higher pressures using the conventionl squre root dependence on pressure. Men exhust plne gs smples ere obtined using 2 hole ter cooled 'X' configurtion probe. The smple gses ere pssed into heted smple line nd on through heted filter nd pump to heted gs nlysis system. The gs nlysis results ere computer processed to provide ir to fuel rtio, combustion efficiency, men dibtic flme temperture nd vrious pollution prmeters. For the sme clibrtion gses the test to test repetbility s better thn 5% on ll the gses nlysed, including Nx. All clibrtion gses for Nx ere clibrted using the sme 1% ccurte reference Nx stndrd in BC Spectrosel gs cylinder. To inlet tempertures ere used: 4K to simulte lo poer nd 6K to simulte high poer opertion. s supplied from bnk of six liquified propne cylinders. The composition of the propne s nlysed by gs chromtogrphy nd this composition s used in the gs nlysis computtions. For liquid fuels, kerosene nd gs oil ere used ith hydrogen contents of 13.7 nd 13.1% respectively. These ere tomised in the rdil pssges by the ction of the high velocity ir flo reltive to the lo velocity liquid fuel jet injection, no pressure tomistion s used. Internl gs composition rdil trverses inside the 14mm dimeter combustor ere mde using 1mm dimeter ter cooled gs smple probe. This hd 1mm smple hole on the side of the probe 1mm from the end nd the stepping moter trverse controller hd positioning ccurcy of better thn.1mm. The influence of the size of the probe nd the position of the smple point hs been investigted in 76mm dimeter xil sirler combustor nd found to hve only smll influence donstrem of the sirler, but significnt differences in the ner sirler region (Andres nd Kokbi, 1987). In the present much lrger combustor the probe interference effects ould be smller thn in the 76mm combustor. The sme heted smple line, pump nd gs nlysis system s used s for the men exhust plne smple. 4. WEAK EXTINCTIN RESLTS At constnt men velocity nd inlet temperture the fuel flo s grdully reduced until visul observtion, through the 1mm indo in the exhust, shoed the flme to go out. The extinction process s lso detectble from the gs nlysis by sudden increse in HC emissions. The ek extinction results ere repetble to +/-.2 equivlence rtio nd re summrised in Tble 1 for rdil vne pssge fuel injection of propne, kerosene nd gs oil. The simulted overll ek extinction s determined by multiplying the mesured ek extinction A/F by the rtio of the reference Mch number to test Mch number (.47/.2). This ssumed tht the ddition of the Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use
Tble 1. Wek Extinction Equivlence Rtio for Mch Number of.2 Fuel Inlet Temp. Eq.R A/F P/P% Simulted verll W.E. Eq.R A/F Rdil Pssge Fuel 4K.4 39 5.8.17 92 6K.29 53.12 125 4K.34 43 5.1.14 11 6K.25 59.11 139 Gs il 4K.43 34 5.1.18 8 6K.19 77.81 181 Centrl Rdil Fuel 6.8 195 4.2.34 458 Premixed nd Air 6.47 34 4.4.2 8 remining ir ould hve no influence on the primry zone sirler ek extinction. This is vlid for dilution ir injected ell donstrem of the primry zone. Comprtive results re lso included in Tble 1 for centrl injection of propne using n eight hole rdil fuel injector. Wek extinction dt for centrl injection hs lso previously been obtined for Mch number of.14 (Alkbie et l 1988). This ork lso included dt for nturl gs hich hd similr ek extinction to propne. Hoever, the rdil pssge gs injector fuel feed system s complex nd third one s not vilble for the lrger hole sizes needed for nturl gs. The ek extinction results shoed tht for propne the rdil pssge injection resulted in mrked deteriortion in the ek extinction compred ith centrl injection. Hoever, it s still 56% improvement on the premixed ek extinction nd the simulted overll flme stbility s ell outside the opertionl requirements. This indicted tht, s expected, there s better fuel nd ir mixing ith fuel injection into the rdil pssges. Prt of the improved mixing s due to the increse in pressure loss from 4.2 to 5.8% cused by the erodynmic blockge of the fuel in the rdil vne pssges. This 1.6% increse in pressure loss s hlved hen using liquid fuel, indicting tht full vporistion did not occur in the rdil vne pssges. The kerosene ek extinction ere very similr to those for propne, but slightly ider. This indicted substntilly similr combustion performnce to propne, hich the gs nlysis results support. The significntly ider stbility limits for gs oil t 6K indicted less vporistion in the rdil vne pssges thn for kerosene, nd richer regions t the bse of the stbilising sirling jet sher lyer, s shon by the rdil trverses discussed lter. At 4K gs oil hd orse stbility thn for kerosene nd s no better thn the premixed sitution. The poor vporistion t 4K ould produce insufficient fuel vpour to crete rich zone t the bse of the sher lyer. The min burning s then donstrem of the sher lyer in the ll region here the fuel s more completely vporised nd mixed ith the ir. The combustion results support this ith Gs il combustion efficiency superior to tht of kerosene nd similr to tht of premixed combustion t 4K. l I 6 J of Lu Lu cc m 8 4 i Centrl _ y, 6 K Injectin Premixed) 1 Pssges 5 1 15 2 25 3 35 DISTANCE FRM SWIRLER. L. mm FIG.3 INFLENCE F FEL INJECTIN METHD N THE F- 81 L 6 J Lu Lu co E 4 CMBSTR AXIAL WALL TEMPERATRE PRFILE. DISTANCE FRM SWIRLER. L. mm FIG.4 INFLENCE F FEL TYPE N CMBSTR AXIAL WALL TEMPERATRE PRFILE. 5. WALL TEMPERATRE PRFILES The xil ll temperture profiles t 6K inlet temperture re shon in Figs. 3 nd 4 t n equivlence rtio of.44, hich s close to tht for the minimum combustion inefficiency. Fig.3 shos the comprison beteen premixed, centrl nd the rdil pssge modes of fuel injection nd Fig.4 compres the three fuels for the rdil pssge injection. The results sho tht both the methods of fuel injection nd the type of fuel hd significnt influence on the xil temperture profiles. The internl gs composition results shoed tht pssge injection crried fuel to the centre of the combustor creting richer core thn for centrl injection. This creted lener ll region nd sloer flme development, s shon by the ll temperture results in Fig.3. behved in similr y to propne, but the gs oil temperture profiles ere substntilly different to kerosene ith more rpid development of 4 Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use
H the flme. The internl trverses shoed the propne nd kerosene flmes developed similr rdil temperture profiles in the initil region ner the ll, but gs oil burned hotter ner the ll. This s becuse t 6K the delyed vporistion of gs oil resulted in richer sher lyer nd outer recircultion zone thn for kerosene. 6. INTERNAL RADIAL GAS CMPSITIN PRFILES The rdil gs composition trverses ere crried out t 6K inlet temperture t n equivlence rtio of.43. This s the equivlence rtio t hich the men exhust gs nlysis shoed the loest Nx emissions ith t lest.1% inefficiency. Detiled gs composition dt s obtined t six xil positions, but spce does not permit the full presenttion of ll this dt here. Dt for gs oil s not obtined in the ner burner region, but s determined for propne nd kerosene. These results for equivlence rtio, temperture, combustion inefficiency nd Nx re shon t the 1 nd 3mm ner sirler xil positions in Fig.5. At the 1mm position the propne fuel distribution s much more uniform thn for kerosene, hich hd richer core region, probbly due to the trnsport of lrger droplets into the core due to the rdil sirler pssge ir momentum. s better mixed inside the pssge giving the more uniform distribution. Hoever, by 3mm Fig.5 shos tht the of C. 1. 4., T-- 4 1. _. PRPME + KERSENE. PRPANE + KERSENE z PRPANE + KERSENE 1..5 PRPANE + KERSENE1 kerosene s more uniformly distributed in the core region ith only the outer recircultion zone being reltively ek. Fig. 6 shos tht the kerosene distribution s lmost completely uniform by the 11 mm xil position. Fig.6 lso shos the gs oil equivlence rtio distribution, ith richer sher lyer zone clerly shon t the 5mm position. Gs oil s uniformly distributed by the 11mm position. The ner sirler flme development is shon in Fig.5. The C nd HC emissions ere lo in the centrl core region indicting tht this consisted of fully burnt combustion products. The inefficiency in this region for kerosene s less thn 1% t 1mm nd less thn 2% t 3mm, nd better results ere found for propne. The min flme development s from the sher lyer into the outer recircultion region. At the 1mm position this outer zone s very high in HC nd hence high inefficiency for both propne nd kerosene. inefficiencies ere higher thn for kerosene in the ll region. This s lso the pttern t 3mm s shon in Fig.5, ith inefficiencies in the ll region of 6 nd 3% for propne nd kerosene respectively. This loer inefficiency for kerosene gve slightly higher tempertures in the ll region thn for propne, in spite of the much lener locl composition s shon in Fig.5. The gs nlysis bsed temperture profiles in Fig.5 sho tht the centrl core region s t high temperture due to the high combustion efficiency. The ll region s t reltively lo temperture for both fuels up to 3mm. nd propne hd very similr temperture profiles ith the min difference being the hotter core temperture for kerosene t the 1mm position, due to the loclly richer mixture. Fig. 6 shos tht by the 11 mm position the temperture profile s close to uniform for ll three fuels nd flme propgtion s predominntly complete. The differences in temperture ere due to smll differences in the men equivlence rtio. x 9ASIL.5 2.5 x SASIL r Y i1 i 1 r/r (X-fmm) r/r (X-3mm) FIG.5 VARIATIN F LCAL TEMPERATRE AND SPECIES CNCENTRATIN AT PLANE 1 & 3mm AWAY FRM THE SWIRLER WITH PASSAGE INJECTIN AND AIR INLET TEMPERATRE 6 K. E X 1 r/r (X-5.m) r/r (X-iioem) FIG.6 VARIATIN F LCAL TEMPERATRE AND SPECIES AT PLANE 5 6 11mm AWAY FRM THE SWIRLER WITH PASSAGE INJECTIN AND AIR INLET 6 K. loo 5 Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use
L z - L z z.1 N m E n 1 1 11 X 1 pek in the gs oil Nx profile in the sher lyer region, hich s the richest zone t the 5mm xil position. This sher lyer zone my hvehd higher Nx levels due to richer locl mixtures upstrem of 35mm, s indicted by the lener ek extinction results for gs oil t 6K in Tble 1. There my hve been fuel Nx contribution for gs oil (Willims et l., 1986) nd this ould lso give rise to the erly production of Nx shon in Fig.6. The trverse results close to the combustor exit plne re shon in Fig.7. The equivlence rtio nd tempertures ere very uniform for ll three fuels, s they ere t the 11mm position in Fig.6. The combustion inefficiencies ere belo.3% t ll rdil positions for ll three fuels. There s still evidence of the rdil propgtion of the flme ith the outer regions ith the orst inefficiency. The Nx emissions ere very uniform, reflecting the uniformity of the equivlence rtio nd temperture profiles. These uniform Nx emissions ere identicl to the men Nx emissions t.43 equivlence rtio for ll three fuels, s discussed belo. The origin of the reltively lrge differences in Nx emissions beteen the three fuels s shon by these trverse results to be in the ner burner sirling sher lyer region nd s not due to differences in therml Nx genertion. r/r (x-39) FIG.7 VARIATIN F LCAL CNCENTRATIN F Nx & CMBSTIN INEFFICIENCY AT PLANE 3mm AWAY FRM SWIRLER WITH PASSAGE INJECTIN, 6 K. The Nx levels t the ner sirler positions ere reltively high, s shon in Fig.5. At the 3mm position the higher Nx emissions for kerosene ere ell estblished t ll rdil positions nd this difference s mintined through to the 3mm xil plne s shon in Fig.7. The origin of the higher Nx emissions for kerosene cn clerly be seen to be the higher temperture richer core region t the 1mm position. The Nx levels for kerosene here ere pproximtely 15ppm compred ith 9 ppm in the centre t the 3mm position, s shon in Fig.7. Even in the ll region the 3ppm Nx levels t 1mm only increse to 8 ppm t the 3mm position, s shon in Fig.7. The dominnt mechnism for Nx genertion thus ppers to be by prompt Nx mechnism ith only smll therml Nx contribution. A strong therml Nx contribution should hve resulted in strong xil vrition of Nx. The good fuel nd ir mixing minimised the therml Nx production s there ere fe loctions here the flme tempertures ere sufficient to generte therml Nx rpidly. Fig.5 shos tht 18K s not exceeded for propne t ny rdil position nd this s true t ll xil positions. This is the temperture beyond hich therml Nx genertion becomes rpid. Fig.6 shos tht t the 5mm position the gs oil Nx emissions ere considerbly higher thn for kerosene, but ith similr rdil profile shpe. Fig.7 shos tht t the 3mm position the gs oil Nx emissions ere less thn t 5mm in the centrl region nd less thn t most rdil positions t the 11mm position. Thus therml Nx s not the cuse of the higher Nx emissions for gs oil. The cuse of the higher Nx s likley to be due to the prompt Nx mechnism, but in the bsence of dt upstrem of 5mm the presence of loclly richer zones thn for kerosene cnnot be excluded. Fig. 6 shos tht there s 35ppm 7. MEAN CMBSTR EXIT EMISSINS 7.1 Influence of the Method of Fuel. The men combustor exit plne emissions ere compred t 6K nd M=.2 for premixed propne/ir, centrl propne nd rdil vne pssge propne injection. The unburned hydrocrbon (HC) emissions ere negligible t nerly ll equivlence rtios tested for ll methods of fuelling. The C emissions, the resultnt combustion inefficiencies nd the Nx emissions re shon s function of the metered equivlence rtio in Figs. 8,9 nd 1 respectively. The Nx emissions corrected to 15% oxygen nd reference humidity re shon s function of the combustion inefficiency in Fig.11. These results sho tht t 6K there s little difference due to the method of fuel injection in the combustion inefficiency or C emissions for mixtures richer thn.45 equivlence rtio. For eker mixtures x z.1.1 1 11 Centrl Pssge 6 K Premixed.5 1. EQIVALENCE RATI FIG.8 INFLENCE F FEL INJECTIN METHD N THE CARBN MNXIDE% EMISSIN. 6 Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use
L) - ln] 1 6 K Centrl Pssge W C-),i 1 z Centrl 6 K Premixed F- V) Premixed F-.1 Pssge..5 1. EQIVALENCE RATI. FIG.9 INFLENCE F FEL INJECTIN METHD N 1 CMBSTIN INEFFICIENCY. 6 K Centrl I V / J /, Premixed 1 F- Pssge ^.5 1. EQIVALENCE RATI FIG.1 INFLENCE F FEL INJECTIN METHD N TTAL Nx EMISSIN. the greter stbility of centrl injection gve ider rnge of equivlence rtios over hich good efficiency could be chieved. Hoever, the min influence of the method of fuel injection s on the Nx emissions. s shon in Fig.1. The rdil pssge injection gve substntilly loer Nx emissions thn for centrl injection nd ere reltively close to those for premixed combustion. This s due to the improved mixing in the rdil pssges nd reduction in the richness of the stbilising sirling jet sher lyer. The corrected Nx emissions in Fig.11 ere extremely lo for rdil pssge injection. The ider stbility compred to premixed fuel/ir lloed slightly eker mixture to be used ith n dequte combustion inefficiency. The optimum Nx emissions for.1% or better inefficiency ere 11, 2.5 nd 1.5 ppm for centrl, rdil pssge nd premixed combustion respectively. The 2.5 ppm for the rdil pssge injection is n extremely lo Nx emission ithout ny conventionl premixing. Previous ork by Alkbie et l 1 1 1 1 Nx CRRECTED T 15% XYGEN,ppm. FIG.11 INFLENCE F FEL INJECTIN METHD N VARIATIN F CMBSTIN INEFFICIENCY% WITH Nx CRRECTED T 15% XYGEN. (1988) for centrl injection of nturl gs shoed tht Nx emissions ere pproximtely hlf of those for propne nd similr reduction could be resonbly expected for pssge injection. Thus pssge injection offers good potentil s dry solution to the reduction of Nx emissions, ith no combustion efficiency problems nd ith n dequte flme stbility. The present centrl injection results re pproximtely 5% loer thn those reported by Alkbie et l (1988). For Mch number of.14 nd 2.5% pressure loss they found tht the optimum Nx emissions ere 17 ppm ith.2% inefficiency. The reduction in the Nx nd inefficiency in the present ork s due to the higher Mch number nd ssocited incresed pressure loss. This improved the turbulent mixing nd reduced the burnt gs residence time, but the improved mixing s the dominnt effect in the reduction of the Nx emissions, s shon by Alkbie et l (1988). 7.2 Comprison beteen Gs nd Liquid Fuels. The unburned hydrocrbon (HC) emissions t 4 nd 6K re shon s function of the equivlence rtio in Figs. 12 nd 13 respectively. At 4K both propne nd gs oil hd very lo HC emissions for equivlence rtio s ek s.55. Hoever, the kerosene emissions ere much higher nd ere coupled ith high C emissions s shon in Fig.14. These results indicte tht the kerosene flmes hd chieved sufficient vporistion to stbilise the flme ith rich zone t the bse of the sher lyers. For gs oil the vporistion s sloer nd flme stbilistion occured further donstrem here more mixing hd occured. The ek extinction results lso indicted better mixing nd orse stbility for gs oil t 4K. At 6K ll three fuels gve lo HC emissions, but gs oil gve the highest nd these ere higher thn t 4K. The gs oil flme s stbilised in the sher lyer, s shon by the ll temperture profiles nd the better stbility. The sloer vporistion thn kerosene ould produce richer sher lyer regions nd hence higher HC nd C emissions, s shon in Figs. 13 nd 15 for mixtures richer thn.48. 7 Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use
1111 K 1 4 K e1 N M < 1.1 r z 1 L L Gsoil Gs 1 1 1 1.4.6.8 EQIVALENCE RATI.4.6.8 EQIVALENCE RATI FIG.12 INFLENCE F FEL TYPE N THE EMISSIN F FIG.14 INFLENCE F FEL TYPE N THE EMISSIN F 1111.1 E n co Q 1 CJ or } NBRNED HYDRCARBNS FR 4 K INLET. CARBN MNXIDE PERCENTAGE FR 4 K INLET. 1 Gsoi 1 6 K x.1 Gsoil 6 K 1 L L L L.5 1. EQIVALENCE RATI FIG.13 INFLENCE F FEL TYPE N THE EMISSIN F NBRNED HYDRCARBNS FR 6 K INLET. The combustion inefficiencies re shon s function of equivlence rtio in Figs. 16 nd 17 for 4 nd 6K respectively. Aprt from kerosene t 4K n inefficiency of pproximtely.1% s chieved by ll the fuels in the ek region. The 98% efficiency for kerosene t 4K nd one tmosphere pressure s quite resonble performnce for conventionl combustor design. These results sho tht pssge injection of liquid fuels did not produce ny mjor problems of poor tomistion nd vporistion hich ould result in very high HC emissions. The results lso men tht there could be no mjor problem of rdil pssge ll etting nd fuel dribble from the pssge exits becuse this ould lso cuse high HC emissions. The Nx emissions t 6K re shon s function of equivlence rtio in Fig.18. nd propne hd similr lo Nx emissions, except close to ek extinction here the kerosene Nx emissions ere 1 11 n 5 1 EQIVALENCE RATI FIG.15 INFLENCE F FEL TYPE N THE EMISSIN F CARBN MNXIDE PERCENTAGE FR 6 K INLET. slightly higher. The internl trverse results shoed tht these higher Nx levels ere generted by locl differences in mixing in the ner burner sher lyer region. The gs oil Nx emissions ere higher t ll equivlence rtios tested, but still reltively lo. The higher emissions ere cused by the richer burning in the sher lyer, s discussed bove. They my lso hve been incresed by the lo levels of fuel nitrogen tht occur in gs oil but not in kerosene (Willims et l 1986). The Nx emissions t 6K corrected to 15% oxygen nd the stndrd dy humidity re shon s function of the inefficiency in Fig.19. The minimum Nx emissions comptble ith lo inefficiency my be obtined from Fig.19. For n inefficiency of pproximtely.1% or loer the minimum corrected Nx emissions ere 2.5, 6 nd 13ppm for propne, kerosene nd gs oil respectively. These re ultr lo Nx emissions for both propne nd kerosene. The gs oil results re the loest ttined by ny combustor design 8 Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use
II 4 K 111 6 K >- C) I^ 11.1 Gsoil i 1 Gsoil L 1 L- L co m.1.4.6.8 EQIVALENCE RATI FIG.16 INFLENCE F FEL TYPE N THE CMBSTIN INEFFICIENCY PERCENTAGE FR 4 K INLET. 1 Gsoil 6 K 1.5 1. EQVALENCE RATI FIG.18 INFLENCE F FEL TYPE N TTAL Nx EMISSINS. WIN 1 1 1.5 1. Nx CRRECTED T 15% XYGEN,ppm. EQIVALENCE RATI M _ z 1 L L W 2.1 L co m 6K Gsoil FIG.17 INFLENCE F FEL TYPE N THE CMBSTIN FIG.19 INFLENCE F FEL TYPE N THE VARIATIN F INEFFICIENCY PERCENTAGE FR 6 K INLET. CMBSTIN INEFFICIENCY% WITH THE EMISSIN F Nx CRRECTED T 15% XYGEN. for this fuel nd re s lo s for centrl injection limits, ith propne in Fig.11. The EPA 75ppm Nx regultion lthough inferior to those for centrl converts, using the therml Nx squre root pressure injection. reltionship, to 2-24ppm t tmospheric pressure for the 15-2 br operting pressure rnge of mny 8.2 For propne t 6K inlet temperture the rdil industril gs turbine combustors. The present results pssge injection reduced the minimum corrected Nx indicte tht for gs oil pproximtely hlf the EPA emissions, comptible ith.1% inefficiency, from 13 Nx limit my be chieved ith fctor of ten to 2.5 ppm. Nx emissions very close to fully reduction possible for propne nd nturl gs. Len premixed system ere thus obtined. primry zones using rdil sirlers ith pssge fuel injection thus offer one of the best potentil 8.3 For kerosene nd gs oil the combustion development solutions vilble for lo Nx emissions, ithout the s very similr to propne. The minimum corrected Nx problems of fully premixed system. emissions ere 6 nd 13 ppm for kerosene nd gs oil respectively. Thus the rdil sirler pssge fuel 8. CNCLSINS injection chieved lo Nx emissions for liquid fuels. 8.1 Rdil vne sirlers ith fuel injection in the vne pssges improve the fuel nd ir mixing compred to centrl fuel injection. But there s sufficient unmixedness in the stbilising sirling sher lyer to give considerble extension of the premixed stbility 8.4 The internl gs composition mesurements shoed tht the fuel nd ir mixing s good quite close to the sirler nd tht this s the key to the lo Nx emissions. The mximum locl equivlence rtio s.75 for men of.43 nd this occurred in the sirler 9 Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use
sher lyer nd s responsible for the enhnced stbility. The differences beteen the Nx emissions for the three fuels originted in the ner sirler region. Therml Nx s not mjor contributor to the overll Nx emissions or to the differences beteen the three fuels. ACKNWLEDGEMENTS We ould like to thnk the Science nd Engineering Reserch Council for series of grnts to develop the gs turbine test fcility t Leeds. R.A. Borehm operted the test fcility nd commissioned the equipment. H.S. Alkbie thnks the Irqui government for three yer reserch scholrship. We thnk Ruston Gs Turbines for providing design dt on the Torndo rdil sirler. REFERENCES Abdul Aziz, M.M., Abdul Hussin,.S., Al Dbbgh, N.A., Andres, G.E. nd Shhbdi, A.R., 1987, 'Len Primry Zones: Pressure Loss nd Residence Time Influences on Combustion Performnce nd Nx Emissions', 1987 Tokyo Interntionl Gs Turbine Congress, Vol.III pp.89-96. Abdul Hussin,.S., Andres, G.E., Cheung, W.G. nd Shhbdi, A.R., 1988, 'Lo Nx Primry Zones sing Jet Mixing Sher Lyer Combustion', ASME Pper 88-GT-38. Ahmd,N.T., Andres,G.E., Kokbi,M. nd Shrif,S.F., 1984,'Centrifugl Mixing Forces in Enclosed Sirl Flmes', 2th Symposium (Interntionl) on Combustion, The Combustion Institute, Pittsburgh, pp.259-267. Ahmd,N.T., Andres, G.E., Kokbi M nd Shrif, S.F., 1986, 'Centrifugl Mixing in Gs nd Liquid Fuelled Len Sirl Stbilised Primry Zones', ASME Pper 85-IGT-13, Int.J.Turbo Jet Engines, Vol.3, pp.319-329. Alkbie, H.S., Andres, G.E. nd Ahmd, N.T.,1988, 'Len Lo Nx Primry Zones sing Rdil Sirlers', ASME Pper 88-GT-245. Becker,B., Berenbrink,P. nd Brndner,H. (1986), 'Premixing Gs nd Air to Reduce Nx Emissions ith Existing Proven Gs Turbine Combustion Chmbers', ASME Pper, 86-GT-157. Beltgui,S.A., Mccllum,N.R.L. nd Rlston,T., 1988, 'The Effect fo Fuel Modes on Combustion of Sirling Flos', Session 1 Pper 5, British Flme Dys 1988, British Flme Reserch Committee, Institute of Energy. Chrles,R.E. nd Smuelson,G.S., 1988, 'An Experimentl Dt Bse for the Computtionl Fluid Dynmics of Combustion', ASME Pper 88-GT-25. Kurod,1987,'Development of Dry To-stge Lo Nx Combustor for Gs Turbine', ASME Pper 87-GT-64. Nejd,A.S., Fvloro,S.C., Vnk,S.P. nd Lngenfeld,M.S.C., 1988, 'Appliction of Lser Velocimentry for Chrcteristion of Confined Sirling Flo', ASME Pper 88-GT-159. Sski, M nd Itch, T., 1987, 'Nx reduction sing Fuel Bonded Nitrogen in To Stge Combustor for Regenertive Automotive Gs Turbine', The 1987 Tokyo Interntionl Gs Turbine Congress, Vol.III pp.83-87. Sothern, A, Perce, D.E. nd verton, D.L. (1984), 'Some Prcticl Aspects of Stged Premixed Lo Emissions Combustors', ASME Pper 84-GT-88. Smith, K.., Kurzynke, F.R. nd Angello, L.C., 1987, 'Experimentl Evolution of Fuel Configurtions for Len Premixed Lo Nx Gs Turbine Combustor', ASME Pper 87-GT-141. Willims, P.T., Brtle, K.D. nd Andres, G.E., (1986), 'The Reltion beteen polycyclic romtic compounds in diesel fuels nd exhust prticultes', Fuel Vol.65, pp.115-1158. Alkbie, H.S. nd Andres, G.E., 1989, 'The Influence of Fuel Plcement on Nx Emissions from Flmes Stbilised by Rdil Sirlers', Joint meeting of the British nd French Sections of the Combustion Institute, Rouen. Andres, G.E., Abdul Aziz, M.M., Abdul Hussin,.S., Al Dbbgh, N.A., Ahmd, N.A., Ali, A.F., Ali Al Sikly, A.F., Alkbie, H.S., Kokbi, M. nd Shhbdi, A.R., 1988, 'High Intensity Burners ith Lo Nx Emissions', British Flme Dys 1988, Furnce Combustion Reserch nd its Applictions, The Institute of Energy. Andres, G.E. nd Kokbi, M., 1987, 'Gs Smpling Probe Influences on Composition Trverses through Sirling Flo Simulted Gs Turbine Primry Zone', NAT Advnced Study Institute, Instrumenttion for Combustion nd Flo in Engines, Pper 9, 16pp, Vimeiro, Portugl, Ed. D.F.G. Duro, Instituto Superior Tecnico, Lisbon. Aoym,K. nd Mndi,S., 1984, 'Development of Dry Lo Nx Combustor for 12MW Gs Turbine', ASME Pper 84-GT-44 m Donloded From: https://proceedings.smedigitlcollection.sme.org/ on 2/24/218 Terms of se: http://.sme.org/bout-sme/terms-of-use