THE APPLICATION OF FUEL STAGING ON REDUCING NO x EMISSIONS FROM GAS BURNER

Transcription

1 Jurnal Mekanikal, Jilid Il, 1996 THE APPLICATION OF FUEL STAGING ON REDUCING NO x EMISSIONS FROM GAS BURNER Mhammad Nari Mhd. Jaafar Department faernautics Faculty fmechanical Engineering Universiti Teknlgi Malaysia ABSTRACT This paper present the methd f burning gaseus fuel at tw different places, i. e. in the swirler vane passages and at the exit plane fthe first cmbustrfr a tw stage cmbustin system. The present wrk nly used natural gas. The effect f injecting the secndaryfuel has a vast influence n reducing emissins fxides fnitrgen. The reductin was fund t be greater than 50% even at near stichimetric cnditins. The secndary fuel acted as a reducing elementfr the thrughly frmed xides fnitrgen frm the first stage cmbustin. 1.0 INTRODUCTION The effects fincreased levels fno, in the atmsphere are wide-reaching. In the atmsphere NO is rapidly xidised t NO and this frm plays an essential rle in the frmatin f trpspheric ne and phtchemical smg, and is xidised t frm nitric acid that may then be depsited as acid rain (Harrisn, 1990). At grund 1

2 Jurnal Mekanikal, Jilid JJ, 1996 level, increased cncentratins (abve 0.06 ppm) f NO can cause respiratry prblem (Wrld Health Organisatin, 1987). The legislatin f NO x emissin limits ill many, parts f the wrld has substantially cmplicated the prcess f burner design. Attempts at lwering NO x emissins by reducing the flame temperature will lead t reduce flame stability r increased CO emissins. The lwest NO x emissin btained in a given cnfiguratin is always limited by unacceptable stability prblems r CO emissins. Thus the burner design has becme a trial-and-errr, multi-parameter ptimisatin prcess (Van Der Meij et ai., 1994). Basically there are tw techniques f cntrlling NO x : thse which prevent the frmatin f nitric xide (NO) and thse which destry NO frm the prducts f cmbustin. In the present wrk the latter methd is emplyed: lean cmbustin fr lw thermal NO x fllwed by secndary fuel injectin fr cmbustin reactin in the secnd stage ne, which culd destry first stage NO x thrugh a reburn mechanism. The methds that prevent the frmatin fno invlved mdificatins fthe cnventinal burner designs r perating cnditins, such as lean primary ne, rich primary ne, rich/lean, r reduced residence time, since the main factrs gverning frmatin f NO is temperature and xygen availability. Hwever, the rich/lean methd tends t increase CO and UEe. Advanced cmbustr designs are needed fr reducing all fur majr pllutants simultaneusly ver a range f thermal r engine pwer utputs. This gives rise t the use f variable gemetry cmbustr and staged cmbustin t cpe with the demands f burner turndwn and pwer variatins in gas turbines, when the verall A/F is increased as pwer is reduced. Fr ultra lw NO x emissins, lean premixed-prevaprised cmbustrs and catalytic cmbustrs are being develped. In staged cmbustin, the cmbustin prcess is arranged t ccur 10 a number f discrete stages. In thery, either circumferential, radial r axial 2

3 Jurnal Mekanikal, Jilid II, 1996 staging may be emplyed. Hwever, in practice circumferential fuel staging mcreases NO x - instead f the fuel being distributed unifrmly arund the liner, it is injected at a small number f pints, where it prduces regins f high temperature (Lefebvre, 1975). The elabratin's fr the abve mentined three types f fuel staging are as fllws: a) Circumferential. Usually this entails discnnecting alternately lcated nles frm the fuel supply. It is ideally suited t tubannular systems but n annular chambers its advantages are largely ffset by the quenching effects f the surrunding cld air n the lcalised burning nes. b) Radial. The simplest applicatin f this technique is t duble-banked annular cmbustrs where, at lw fuel flws, it is a relatively simple matter t inject all the fuel int the inner r uter cmbustin ne. c) Axial. By designing the primary ne fr ptimum perfrmance at lw pwer settings, and then injecting the extra fuel needed at higher pwer levels at ne r mre lcatins dwnstream. In the present wrk, axial fuel staging was emplyed that cnsisted f leanlean staged cmbustin. The first cmbustr was perated very lean with all the air needed fr cmbustin intrduced in this ne and the peratin was set clse t the lean stability limit. Fuel, withut any air, was then injected int the cmpletely burnt prducts f this lean primary cmbustin ne t bring the burner t the desired verall excess air. Typically the lean ne may have an equivalent rati f 0.6 with fuel injected in the secndary ne t bring the verall equivalent rati t 0.9. Thus, it is a lean/lean staged system. 3

4 Jurnal Mekanikal, Jilid II, EXPERIMENTAL SET-UP This system cmprised ftw different sies cmbustrs. The smaller ne f 76 mm inside diameter was attached t the plenum chamber and acted as the first stage. The radial swirler f 40m:rp utlet diameter and 21.5 mm depth was used as a flame stabiliser. The first cmbustr was fuelled via the radial vane passage injectin mde. The air and fuel were mixed thrughly prir t ignitin. At the ex.it plane f the first cmbustr a wall fuel injectr f 76 mm diameter with eight fuel nles f 3.5 mm diameter was attached. This is the injectr fr the secnd stage secndary fuel. The mixtures f flue gas frm the first cmbustr and the secnd fuel were allwed t expand freely int a larger cmbustr f 140 mm internal diameter. The wall injectr and the secnd cmbustr were attached t the first cmbustr by the used fflanges. The schematic diagram fset-up f fuel staging test rig is shwn in Fig mm DIAMETER PLENUM RADIAL SWIRLER ORIFICE PLATE WALL INJECTOR (SECONDARY f'uel) PRIMARY COMBUSTOR T 140mm --l\------_' IGNITOR 320mm )rm 330mm Fig. 1 Schematic Diagram ffuel Staging Test Set-up Rig 4

5 Jurnal Mekanikal, Jilid II, J996 The staged cmbustin tests were run at 40 mmwg pressure lss t simulate burner perating cnditins. The radial swirler in the primary ne in Fig. 1 has a swirler utlet f40 mm. This culd be fitted with different rifice plates t reduce the swirler utlet areas. The rle f the rifice plate was t enhance flame stabilisatin and prvide a better mixing f the air and fuel prir t ignitin. It als created the pressure lss at the utlet rather than in the vane passage which generated maximum turbulence in the swirl shear layer. The rifice plate als "helped t prevent fuel frm entraining int the cmer dumped expansin uter recirculatin ne and thus create a rich lcal ne which wuld lead t higher NO, emissins frm this area. The rifice plate was munted at the exit plane f the radial swirler as shwn in Fig. 1. All the tests were carried ut at atmspheric pressure with an air inlet temperature f 400 K, with burner upstream pressure f 40 mmwg. The rifice plate sies invlved were 25.4 mm and withut any rifice plate (40mm); designated as 25 mm and withut (40 mm), respectively in all the figures. The 25.4 mm rifice plate was chsen t shw the effect fturbulent enhancement due t area restrictin which helps in reducing emissins frm exhaust gas. 3.0 MEAN EXHAUST EMISSION RESULTS 3.1 Fuel Staging Test Results fr N Orifice Plate Case Figures 2-7 shw the fuel staging test results as cmpared t the baseline results with all the fuel supplied t the primary swirler fr n swirler utlet rifice plate. As can be seen generally, a marked reductin in NO, emissins can be achieved. A reductin f greater than 50% was achieved even at near stichimetric cnditin where the equivalent rati was Figure 2 shws an perating equivalent rati fr fuel staged case f The rich cnditin was limited by a drastic increase in carbn mnxide 5

6 Jurnal Mekanikal, Jilid Il, 1996 (CO) emissin t an unacceptable value. At an equivalent rati f 0.913, a crrected NO x emissin t 0% xygen n a dry basis f 29.3 ppm were frmed representing a 50% reductin. Figure 3 shws that a cmbustin inefficiency f 0.01% was btained fr a wide range f equivalent ratis up t when the cmbustin inefficiency started t rise. This means that a cmbustin efficiency f 99.99% was achieved fr a wide range f equivalent ratis. Figure 4 shws CO emissins f less than 10ppm were btained ver a wide range f equivalent ratis up t Even at an equivalent rati f 0.8 the CO emissin were nly 15 ppm. The lwest CO emissin was 6ppm at an equivalent rati f Hwever, fuel staging increases the CO t 100ppm fr equivalent rati f 0.9 but were acceptable at 15 ppm fr an equivalent rati f This CO increases due t the lwer residence time with fuel staging and the lwer xygen availability. Unburned hydrcarbn (URC) emissins f less than 5 ppm were achieved ft the entire range f perating equivalent ratis. Figure 6 shws the NO x results as a functin f the adiabatic flame temperature, which shws that even at high flame temperatures the increase in NO x was nt very steep fr fuel staging. This was due t higher temperature which increased the rate f frmatin fno, when single stage cmbustin ccurred. Patry and Engel (1950) studied the reactin between nitric xide and methane at temperatures up t 11OODC and fund that 70 percent f the nitric xide was cnverted in 3 secnds at this temperature, t frm hydrgen cyanide, water and hydrgen via the fllwing reactin: (1) exhaust. Thus, HCN can be an undesirable emissin generated at lw temperature in the 6

7 Jurnal Mekanikal, Jilid If Fuel Staging Test Results fr 25mm Orifice Plate Case Figures 8-13 shw the fuel staging test results as cmpared t the baseline results fr 25 mm swirler utlet rifice plate. It can be seen that thee marked reductin in NO, emissins can be achieved. The reductin f crrected NO, is even greater than the case f n rifice plate cnditin (Fig. 8). A reductin f greater than 70% was achieved at an equivalent rati f (near stichimetric cnditin). Figure 8 shws an perating equivalent rati fr fuel staging case f The rich cnditin was nce again limited by a drastic increase in carbn mnxide (CO) emissin t an unacceptable value. At an equivalent rati f 0.875, a crrected NO, emissin t 0% xygen n a dry basis f 15ppm were frmed representing a 72% reductin. Figure 9 shws that a cmbustin inefficiency f 0.01% was btained fr a wide range f equivalent ratis up t 0.83 when the cmbustin inefficiency started t rise. This means that a cmbustin efficiency f 99.99% was achieved fr wide range f equivalent ratis. 80!-'-''-'-'...a...L.L..L-...Jl...L..I...L.L..L-L-.L..1...L...I...J..."J...L..L-...,.,..'-'-'...L-L >' 40 a: C\l 20 ':i Q. Q Wl WIOUT STAGING -.- WITH STAGING / /... w 4 8 2l L...L.1..J...&-..L...I...l...JI..-L...J...L..L...L...I...JL L...L..L..J..."...L-O Fig. 2 METERED EQUIVALENCE RATIO (<1» Crrected NO, vs Metered Equivalence Rati fr N Orifice f1p = 40 mmh 2 0 ; Tin = 400 K; <1>primary =

8 Jumal Mekanikal, Jilid II, 1996 > ZW (3 i! It 0.01 i= tn ::) CD -.- W/OUT STAGING O.1 l-l-...fi-l...l..l'-&-l.l...l..l..l.l...l...r...j...l..lc...l...l..l..1...l...l..l /.-.-. p. / METERED EQUIVALENCE RATIO (<1>), Fig. 3 Cmbustin Inefficiency vs Metered Equivalence Rati fr N Orifice P = 40 mmh 2 0 ; Tin = 400 K; <1>primary = W/OUT STAGING c wc X s II: « METERED EQUIVALENCE RATIO (<1» Fig. 4 Carbn Mnxide vs Metered Equivalence Rati fr N Orifice = 40 mmh 20; Tin = 400 K; <1>primary =

9 Jurnal Mekanikal, Jilid II, a CD 10 « >- I 4 0 LU Z ::> CD 2 ::> \ -.- W/OUT STAGING \ e--e METERED EQUIVALENCE RATIO (<1» s;:- C\I <t- 20 ':f. a.. )( 8 I- 6 O' () 4 Fig. 5 Unburned Hydrcarbn vs Metered Equivalence Rati fr N Orifice P = 40 mmli.o; Tin = K; ci>primary = W/OUT STAGING FLAME TEMPERATURE (K) Fig. 6 Crrected NO x vs Flame Temperature Rati fr N Orifice P = 40 mmll.o; Tin = 400 K; <Dprimary =

10 Jurn al Mekanika l, Jilid II, / _...I L..L...-.l L '---, -. - WIOUT STAGING -<0- "'",STAG,"G1 > U ZUJ U u:: U UJ Z Z 0.01 CO ei e /e'fi e-e --.-/ CORRECTED NOx (PPM, 0% 02 DRY) Fig. 7 Cmbustin Inefficienc y vs Crrected NO, fr N Orifice ilp = 40 mmh 2 0 ; Tin = 400 K; <!>primary = c:: 0 N 0 20 ifl WIO STAGING - e- WITH STAGING e--e ':i e-- e, e e----- e::. e >< e 0 UJ 6 e 0 8 -: I- U UJ c:: 4 c:: 0»-: METERED EQUIVALENCE RATIO (<1» Fig. 8 Crrected NO, vs Metered Equivalence Rati fr 25 mm Orifice ilp = 40 mmh 2 0 ; Tin= 400 K; <Dpri mary =

11 Jurnal Mekanikal. Jilid ll, WIO STAGING l 0.06 >- Z 0.04 w 0 u:: u, w 0.02 Z 0 t= (/) ::J CD J \/0",,/.--. e METERED EQUIVALENCE RATIO (4)) Fig. 9 Cmbustin Inefficiency vs Metered Equivalence Rati fr 25 mm Orifice L1P = 40 mrnl-l.o ; Tin = 400 K; <1>primary = e Figure 10 shws CO emissins f less than 10ppm were btained ver a wide range f equivalence ratis up t This is als a better result than the n rifice plate case. Even at an equivalent rati f 0.84 the CO emissin were nly 20ppm. The lwest CO emissin was 7ppm at an equivalent rati f Hwever, fuel staging increases the CO t mre than 100 ppm fr equivalent rati greater than This CO increases was due t the lwer residence time with fuel staging and the lwer xygen availability. Unburned hydrcarbn (UHC) emissins fless than 5 ppm were achieved fr mst f the range f perating equivalent ratis as shwn in Fig Figure 12 shws the NO, results as a functin f the adiabatic flame temperature, which shws that even at high flame temperatures the increase in NO, was nt very steep fr fuel staging. Figure 13 shws that lwest NO, emissin f 6ppm was btained at an cmbustin inefficiency f0.01%. 11

12 Jurnal Mekanikal, Jilid II, CONCLUSION A NO x reductin f at least 50% culd be achieved fr bth rifice plate sies using fuel staging methd with a very lean primary ne. The lwer NO x at an equivalent rati f0.78 was btained fr the 25mm rifice plate. NO x with staged fuel injectin had a much lwer dependence n flame temperature than a single stage cmbustin system. Single digit NO x crrected t 0% 0 dry culd be btained at equivalent rati f 0.78 (30% excess air) fr the 25mm rifice plate case with the lwest value f 8.75 ppm btained with the 25 mm rifice plate _- WIO STAGING UJ Cl X ::2: ld a: «0.001 )" e-e... e. ee _ - e --_ METERED EQUIVALENCE RATIO (<1» Fig. 10 Carbn Mnxide vs Metered Equivalence Rati fr 25 mm Orifice f1p = 40 mmh 2 0 ; Tin = 400 K; <I>primary =

13 Jurnal Mekanikal, Jilid ll, WIO STAGING, e: « Cl 6 >- J: Cl 4 w Z ::> 10 2 ::> 1 Cl c:! 20?ft, e: Cl 6 w I- li! 4 e e \/\ METERED EQUIVALENCE RATIO (/Il) Fig. 11 Unburned Hydrcarbn vs Metered Equivalence Rati fr 25 mm Orifice L1P = 40 mmll.o; Tin = 400 K; <1>primary = I--'----l_-'-----I._...L.-...J._.L---'-_'----'-----JL...--'-----L_...L.----L_.L-...L.--t WIO STAGING 21:...--'-----JL--'----l_-'----l_-'----L_'----'-----JL L----L_...l----L_..L..-...L.----L FLAME TEMPERATURE (K) Fig. 12 Crrected NO x vs Flame temperature fr 25 mm Orifice L1P = 40 mmll.o; Tin = 400 K; <1>primary =

14 Jurnal Mekanikal, Jilid II, > w u: u, w 0.02 i= CJ) :::!; WIO STAGING -.- WITH STAGING -. I \/\/ '-- ---'_...=..._-----J._--L-...L- ----=:..._----! CORRECTED NOx (PPM, 0% 02 DRY) Fig. 13 Cmbustin Inefficiency vs Crrected NO x fr 25 mm Orifice P = 40 mmh 2 0 ; Tin = 400 K; <l>primary = REFERENCES 1. Harrisn, R.M., Imprtant Air Pllutants and Their Chemical Analysis. Pllutin: Causes, Effects and Cntrl, Ryal Sciety f Chemistry, Lndn, 1990, Chapter 8, pp Lefebvre, A.H.; Pllutin Cntrl in Cntinuus Cmbustin Engines, Fifteenth Sympsium (Internatinal) n Cmbustin, The Cmbustin Institute, 1975, pp Patry, M and Engel, G.; C.R. Acad. Sci., Paris 1950, Vl. 231, pp

15 lurna/ Mekanika/, lilid II, / Van Der Meij, C.E., et al.; On the Effects f Fuel Leakage n CO Prductin frm Husehld Burners as Revealed by LIF and CARS, Twenty Fifth Sympsium (Internatinal) n Cmbustin, The Cmbustin Institute, 1994, pp Wrld Health Organisatin, Air Quality Guidelines fr Eurpe, WHO Reginal Publicatins, Eurpean Series N. 23, Cpenhagen,